https://cpw.cvlcollections.org/files/original/533817fc841eb5a19568f211166ad87f.pdf a8c08ba2542cc82e80a2f9b6deec5031 PDF Text Text Colorado Division of Parks and Wildlife September 2013-September 2014 WILDLIFE RESEARCH REPORT State of: Cost Center: Work Package: Task No.: Colorado 3420 1663 N/A Federal Aid Project No. N/A : : : : Division of Parks and Wildlife Avian Research Terrestrial Species Conservation Development of distribution models for management of greater sage-grouse in North Park Colorado Period Covered: September 1, 2013 – August 31, 2014 Author: M. Rice Personnel: T. Apa, L. Rossi All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT Rangewide declines of greater sage-grouse and recent energy development within sagebrush habitat has led to concern for conservation of greater sage-grouse (Centrocercus urophasianus) (GRSG) populations across Colorado, including in North Park, which supports approximately 20% of the state’s GRSG. Seasonal variations in habitat use by GRSG can provide important information for biologists and managers on the ground. These habitats have been mapped at the statewide level at a large scale, but have not been completed specifically for the North Park population. Investigating the smaller scale seasonal habitat selection of GRSG in North Park is important as no data from North Park was used in the statewide analysis. Since GRSG habitat use is known to be influenced by both landscape-scale and localscale factors, data specific to North Park can be used to refine the statewide models at a more local scale. Almost 4,000 locations of 117 radio-marked female GRSG were collected from April 2010February 2012 in North Park. These locations were used to map breeding, winter, and summer habitat using a logistic regression in program R. Variables were chosen based on vegetation, topography, and oil/road development across North Park. All three seasons indicate a high probability of GRSG in areas where they currently reside. The breeding and winter model tend to be more similar, predicting high probability of use in large expanses of sagebrush and little to no probability in riparian areas. The summer model predicts greater use along the edge of riparian areas and a more disjunct high probability surface. Overall, oil and gas development had no effect in the winter or summer seasons, but a significant negative effect during the breeding season. The North Park population has a relatively low level of oil and gas development currently and these prediction surfaces provide managers a good baseline for seasonal habitat management of GRSG. As the possibility of oil and gas development and other landscape changes occur in North Park, it will become more critical to know how the seasonal habitat selection of GRSG in North Park may change. These seasonal models provide data-driven, defensible distribution maps that managers and biologists can use for identification and exploration when investigating GRSG issues specific to North Park. �WILDLIFE RESEARCH REPORT DEVELOPMENT OF DISTRIBUTION MODELS FOR MANAGEMENT OF GREATER SAGEGROUSE IN COLORADO MINDY B. RICE PROJECT OBJECTIVES The goal of this study is to obtain detailed, current information on GRSG seasonal habitat use in North Park, Colorado based on new telemetry data, and compare to the current statewide modeling previous completed (Rice et al. 2012). SEGMENT OBJECTIVES 1. Develop seasonal habitat use models using the telemetry database of GRSG locations in North Park to predict the probability of sage grouse locations in North Park, Colorado. 2. Compare the new North Park model to the recently completed northwest Colorado GRSG range map and determine the ability to update this population on the map. INTRODUCTION The greater sage-grouse (Centrocercus urophasianus) (GRSG) is a species of conservation concern due to historical population declines and range contraction (Schroeder et al. 2004). In 2010, the Fish and Wildlife Service (USFWS) determined that greater sage grouse were warranted but precluded from being listed as an endangered species (US Fish and Wildlife Service 2010). In Colorado, there are 6 distinct populations of GRSG in Northwestern Colorado including the North Park, Middle Park, Meeker/White River, North Eagle/South Routt, Northwest, and Parachute/Piceance/Roan populations (Fig. 1). Of these populations, the North Park and Northwest populations are the largest and most stable; the North Park population supports approximately 20% of the statewide population (Colorado Division of Wildlife 2008). For species that use large areas including GRSG (Connelly et al. 2011), there is value in evaluating habitat selection at multiple spatial scales (Shirk et al. 2014) as species may respond differently at larger or smaller scales (Cunningham et al. 2014). Consideration of scale is necessary for deciding how habitat data should be applied in resource management (Boyce et al. 2003). Often data is collected in response to a prior analysis on habitat selection which may have been too coarse in scale or limiting to the population of concern. However, it remains unclear how to incorporate multi-scale results in applied management where legal constructs such as critical habitat lack a defined scalar context (De Cesare et al. 2012). Drawing conclusions about habitat selection based on observations at any one scale may misconstrue the importance of variables driving system behavior (Doherty et al. 2010). Finer scale models based on detailed species and landscape information have shown great potential to detect crucial habitat not obvious at broader scales (Klar et al. 2008). Colorado Parks and Wildlife, the Bureau of Land Management, and the United States Forest Service are currently using a preliminary priority habitat (PPH) and preliminary general habitat (PGH) map to make decisions on GRSG management across all 6 populations of GRSG in Colorado (hereafter called the PPH/PGH map). The seasonal habitat selection models used as a part of the statewide PPH/PGH map development used only vegetation data to describe habitat and over 20,000 GRSG telemetry locations, none of which were specific to the North Park population (Rice et al. 2012). Although these statewide seasonal models correctly predicted and validated GRSG habitat in the North Park population, the scale of the analysis was large and fine-scale habitat selection patterns within populations were not detected by the model (Rice et al. 2012, Fig. 2). Although GRSG require sagebrush throughout the year, specific habitat requirements may differ among seasons (Connelly et al. 2000, �Colorado Division of Wildlife 2008) and current patterns of use specific to North Park are not welldocumented. Fedy et al. (2012) found wide variation in inter-seasonal movement distances between specific populations which indicates that a model specific to the scale of the North Park population may be warranted for refined management actions. Energy development has also emerged as a major issue in GRSG conservation because areas currently under development contain some of the highest densities of GRSG (Naugle et al. 2011). Research examining the effects of energy development suggests that GRSG populations are negatively affected by energy development activities, especially those that degrade important sagebrush habitat (Naugle et al. 2011, US Fish and Wildlife Service 2010). The rangewide rate of producing wells has tripled from the 1980s to 2007 within GRSG habitat and the impacts at conventional well densities (8 pads per 2.6 km2) exceeded the species threshold of tolerance (Naugle et al. 2011). North Park has a relatively low level of developed oil/gas fields (Copeland et al. 2009, Fig. 1), with an estimate of 6.2% of the area covered by oil fields (COGCC) and most is concentrated within the McCallum Field (Braun et al. 2002). The McCallum Field is a relatively small moderately developed oil production area and suitable sagebrush (Artemisia spp.) habitats are still available (Braun et al. 2002). The current anthropogenic disturbance as estimated in the DRAFT Northwest Colorado GRSG Resource Management Plan (RMP) and Environmental Impact Statement (EIS) is 1.7% and overall disturbance is 6.2% (Bureau of Land Management 2013). Given the relatively low level of existing development in North Park, current models of seasonal habitat selection may provide a useful baseline with which to examine the effects future development may have on GRSG distribution and habitat use. With over 81% of federal lands with the potential for oil and gas development having already leased their rights (Copeland et al. 2009) and 70% of remaining sagebrush habitat being publicly owned, none of which is protected within a federal reserve system, there is opportunity for widespread expansion of energy development throughout GRSG habitat in North Park (Naugle et al. 2011). We had three main objectives for this study. First, we developed seasonal habitat selection maps using data collected in the North Park population. We predicted that the addition of specific data collected for this purpose in North Park would refine the results from the statewide seasonal habitat selection models and provide managers with more detailed habitat surfaces. Second, we evaluated how using used the new seasonal models to inform andwould modify the PPH/PGH map surface for North Park to compare, compared to the current PPH/PGH map based on statewide seasonal models. We predicted that the refined models would significantly change the PPH/PGH surface in North Park using seasonal habitat selection models developed with more variables that were specifically evaluated for North Park. Lastly, we investigated if the addition of an oil/gas development variable changes or better informs the top models in each season. We predicted that current development (wells, roads) is at low enough levels that there would be little to no effect of adding the oil/gas development variable to the models. STUDY AREA The study area is located within the North Park population of greater sage-grouse in Jackson County in Northwestern Colorado (Fig. 1). Jackson County is bounded on the north by Wyoming, the west by the Park Range, the south by the Rabbit Ears Range, and the east by the Medicine Bow Range. The semi-arid basin in the center is commonly referred to as North Park. GRSG occupied range in North Park covers approximately 170 hectares and includes the majority of the North Park basin. The North Park basin is largely comprised of a mix of sagebrush uplands interspersed with riparian and wetland communities located along the multiple waterways throughout the Park. The human population in Jackson County is small with less than one person per square mile and a total population of 1,370 reported in 2011 (U. S. Census Bureau). �METHODS Data Layers GRSG were captured and radio-marked during April 2010 and April 2011 using spot-lighting techniques (Giesen et al. 1982, Wakkinen et al. 1994) and a 4-wheel drive ATV. We collected telemetry locations from 117 female GRSG over the study period for this analysis. All GRSG captured were weighed (±1 g) using an electronic scale and marked with uniquely numbered aluminum leg bands. The age and gender of each GRSG was determined using wing (Dalke et al. 1963) and other plumage or morphological characteristics. VHF transmitters applied were 17-g necklace-mounted radio transmitters with a 30 cm antenna lying between the wings and down the back of the grouse. Transmitters have a minimum battery life of 18 months and a 4-hour mortality circuit. The radio transmitter package was 1.0% and 1.2% of the body weight for adult and yearling females, respectively. Following release, the movements and survival of all radio-marked GRSG were monitored approximately once per week. Incubating females were monitored more frequently (> 1 time per week) to determine nest fate. Females with broods and unsuccessful females were located approximately once per week. Radio-marked birds were also located approximately once per week during the winter. Handheld Yagi antennas attached to a receiver/scanner were used to locate radio-marked grouse. The loudestsignal method was used to locate grouse/transmitters (Springer 1979). Monitoring efforts were distributed equally among 3 diurnal periods; morning (< 4 hours following sunrise), midday (> 4 hours after sunrise) and evening (< 4 hours before sunset). All grouse were circled at a 50 – 100 m radius (Apa 1998) to determine habitat type while at the same time avoiding flushing the bird. This radius was larger during the winter months when birds were grouped in flocks and flushed more easily. A precise Universal Transverse Mercator (UTM) location was not possible at the time of location (the birds were not intentionally flushed). To obtain more precise use locations, the observer selected a location approximately 50 m in one of the 4 cardinal directions from the estimated location of the bird. The observer took a Global Positioning System (GPS) location, and then manually corrected the UTM location. At each location, date, time, UTM coordinates, slope and aspect were recorded. A fixed-wing aircraft assisted to locate any grouse not located or lost during ground monitoring efforts. Average movements for each individual were calculated and averaged over all the birds in each season (Table 1). The average daily movements for each season were then used to buffer all presence and available locations within that season. We used this buffer to account for possible error in the telemetry locations as well as to summarize the environmental covariates at a biologically relevant scale. We randomly generated a sample of 10,000 “available” locations, which were not allowed to overlap the presence locations, within the same geographical extent as the presence values (Stokland et al. 2011). We conducted a sensitivity analysis of the available sample size for each season using the method outlined in Northrup et al. (2013) and ran the analysis at 1,000 intervals up to 10,000. Each of the 10,000 available locations were specific to each season. We found the ideal number of available locations where coefficients converged at 7,000 in the breeding and winter seasons and 6,000 in the summer season. We applied the same daily movement buffer specific to each season to the available locations. All data was summarized within these buffered locations. Vegetation classification was obtained from the basinwide vegetation layer and was classified according to biologically meaningful groupings resulting in 14 vegetation groups (Appendix A). This land cover layer was constructed with 25-m resolution lands at imagery as part of the Colorado Vegetation Classification Project administered by the Colorado Division of Wildlife in collaboration with the Bureau of Land Management and the United States Forest Service and was completed in 2005. From those 14 groups, we determined that there were values less than 0.1% for residential, bare, greasewood/herbaceous, bitterbrush (Purshia tridentata), talus, alpine, and water within our buffers; thus we excluded those categories from the analysis. In addition, we removed aspen (Populus tremuloides) and forest as models did not converge due to the lack of the variable in the presence buffers in all seasons. Thus, the vegetation categories we used were irrigated agriculture, sagebrush, grassland, �sagebrush/grassland and riparian, which all have support from previous GRSG studies (Smith et al. 2014, Rice et al. 2012, Doherty et al. 2010, Aldridge and Boyce, 2007). We obtained elevation data from the USGS digital elevation model. We used the national hydrography dataset to measure the average distance to perennial water sources as well as the density of water within each buffer. Hydrology has been included as a variable in previous studies as well (Dzialak et al. 2012). We also included the Normalized Difference Vegetation Index (NDVI), a measure of plant health within the study area (Aldridge and Boyce 2007, Blomberg et al. 2012). A plant that is actively photosynthesizing will absorb most of the visible light resulting in a higher NDVI value and higher plant productivity (Tedrow and Weber 2011). Links between NDVI and bird populations can be found with higher NDVI variances being linked to higher habitat variability (Pettorelli et al. 2011). The NDVI values for sagebrush are mainly on the lower end between 0 and 0.45 (Tedrow and Weber 2011, Baghzous et al. 2010). We also measured the distance to sagebrush for those locations that did not contain sagebrush within their buffers. GRSG may use irrigated agriculture, but were found to stay within 45 meters of intact sagebrush stands (Bureau of Land Management 2013, NTT 2011). Although most of our presence locations included sagebrush within their buffers, those not located in sagebrush were most likely close to patches of sagebrush. A list of the variables used at the start of each seasonal model are in Table 2. Model Building by Season We first calculated Pearson correlation coefficient(r) for all the variables. In all three seasons, irrigated agriculture was the only variable that was highly correlated (r>0.65) with other variables and was thus removed as a variable used in model development. We standardized variables in each season to have a mean of 0 and a SD of 1 (McAlpine et al. 2008). This allowed us to directly compare our coefficients for variables measured at different scales. We used a logistic generalized linear model with a logit link using the lme4 package in program R (Development Core Team 2014). We used a random intercept to control for the unbalanced sampling among individuals within each season (Gillies et al. 2006). Individual variability is thus identified explicitly and the scope of inference can be extended to the entire population (Gillies et al. 2006). We constructed a set of alternative models from all linear combinations of the informative explanatory variables in each season (McAlpine et al. 2008). Lukacs et al. (2010) determined that model averaging can protect against spurious results and is appropriate for developing a prediction tool (Burnham and Anderson 2002). Therefore, rather than selecting a single best model, we used model weights up to 95% to average into a final model for each season (Burnham and Anderson 2002). We assessed variable importance by summing the Akaike model weights across models that included the variable of interest (Smith et al. 2014, Burnham and Anderson 2002, Arnold 2010). The coefficients are equivalent to selection ratios (Manly et al. 2002) and exp(βi) can be interpreted directly as the odds ratios. When larger than 0, use is occurring more than expected and less than expected if less than 0 (Boyce et al. 2003). In order to have the most predictive potential for population level inferences, we removed noninformative variables in each season based on 85% confidence intervals around parameter estimates that included 0 (Smith et al. 2014, Arnold 2010). Poor power and inconclusive statistical inference is expected from covariates with confidence intervals that approach or overlap 0 (Johnson et al. 2004). The averaged model for each season was used to create a prediction surface in ArcMap 10.1(ArcGIS 10.1; Environmental Research Systems Institute, Redlands, CA). We applied the logistic equation in the following form to create the probability of GRSG presence across North Park: w * ( xi ) = exp( B0 + B1 x1 + ... + Bn xi ) 1 + exp( B0 + B1 x1 + ... + Bn xi ) �Where observations i=1…η, β0 is the mean intercept and βη are the estimates for covariates χi. The logistic function was used to create a probability of presence surface with values between 0 and 1 across the study area for each season (1=high, 0=low). Development impact analysis Due to the restricted amount of oil and gas development within North Park, we first created a layer of the oil/gas fields from the Colorado Oil and Gas Conservation Commission (COGCC) currently within North Park (Fig. 3). The wells in North Park are concentrated in a much smaller area than the entire North Park study area, so we did not want to diffuse the influence across the entire North Park, but rather restrict the dataset to only those presence and available buffers located near the oil/gas fields. Based on the oil and gas field classification in the COGCC, we applied a 350m buffer around the fields based on the buffer used in Walker et al. (2007) and then selected within each season those presence and available buffers that overlapped this area (Fig. 3). A shapefile was also obtained from the COGCC website which outlined locations for wells, ownership, and well ID information (Table 3). The next step was to determine historical dates for all 677 wells from the COGCC website including well status, spud date, completion date, first production date, last production date, and expiration date. We only included those wells that were active and the disturbed well pad could be seen visually on satellite and/or NAIP imagery (Harju et al. 2010; Table 3). Due to the lack of digital transportation data in Jackson County, Colorado, we created a database of historical road networks versus roads created for energy development purposes. From the Colorado Department of Transportation (CDOT), we were able to obtain shapefiles with existing highways, major county roads, and local roads, however, oil and gas roads had to be digitized by hand with the assistance of hard copy maps and digital topographic maps (USGS, ESRI, National Geographic Society 2011, Google Earth 1999-2013). We used our well pad coordinates and identified which roads were leading towards oil/gas fields and/or oil/gas wells or were within a oil/gas field and classified these as roads created for the purpose of oil and gas production (unless they were classified as a county/local road by CDOT). We were interested in the overall activity levels within the oil and gas fields so we created an index that combined both the density of oil/gas roads leading to active wells and the density of active wells within 1 km2. This was done by first creating a density of active oil well pads layer in ArcMap and then creating an oil road density map. Combining these two layers created an index of oil/gas development surface layer which was used to assess the effect of development within each season. We then re-ran the best model from each season with the additional oil/gas development variable to see if the model improved (reduced AIC) or if development was an informative predictor of GRSG habitat in North Park (Doherty et al. 2008). We used the average models from each season as a starting point rather than starting the model with all variables as we were interested in the influence of development variables on our “best” prediction surface. If the seasonal model fit improved with the development variable, we applied that model to the raster surfaces to look at the change in the prediction surface in ArcMap. Updating Effect of New Models on Colorado GRSG priority habitat maps We updated incorporated the new seasonal habitat models into the statewide GRSG habitat priority map for the North Park population based on the newly collected data specific to this populationfor comparison to the current map. In order to do this, we used the same process as was used to develop the statewide PPH/PGH map. The PPH is defined as habitat with a high probability of use during the summer, winter, or breeding habitat models developed in a draft report by Rice et al. (2012) within a 6.4 km buffer around any lek active in the last 10 years. Areas outside of these criteria, but still within occupied range are considered PGH. This map was developed for the BLM’s environmental impact statement for the Northwest GRSG draft land use plan (BLM 2013). These criteria were applied to the North Park seasonal habitat maps in order to have a direct comparison between PPH/ PGH map predictions for the statewide analysis versus the North Park specific analysis. �Model Validation Common methods of evaluating model performance including Kappa and receiver operating curves are inappropriate for presence/available data as the distribution of used sites is drawn directly from the distribution of available sites (Boyce et al. 2002). Therefore, we performed a k-fold cross validation 5 times withholding 20% of data randomly for each iteration in program R (Johnson et al. 2004, Boyce et al. 2002). The model evaluation was assessed for presences only as available locations were randomly chosen and not true absences (Hirzel et al. 2006). For each data fold, the withheld data can be assessed against the model predictions of the training data using correlations between bin rank of the RSF values and the frequency of independent, withheld observations in the same bin rank standardized for area (Johnson et al. 2006). We assessed the relationship between predicted occurrence for withheld animal locations and their frequency within 10 incrementally larger bins of equal size adjusted for area (Wiens et al. 2008, Johnson and Gillingham 2008). RESULTS Bird Capture and Movements We collected a total of 3,985 locations from GRSG in North Park from April 2010 until February 2012, from 95 birds radio-marked during the first year and 22 additional birds in the second year. Locations were assigned to one of three seasons based on the following: breeding season (April – July 15), summer season (July 16-September), and winter season (October-March). There were a total of 1,480 locations in the breeding season, 874 locations in the summer season, and 1,631 locations in the winter season. Birds in the North Park population tended to move more, regardless of the metric, during the winter season and less during the summer season (Table 1). On average we were able to collect a telemetry location on each bird every 12.5 days. Patterns in the overall data for all 3 seasons include presence buffers being closer to sagebrush than available buffers and lower NDVI values (Table 2). The summer season exhibited the most differences from the breeding and winter seasons based on mean values of the presence and available buffers. The summer season had lower sagebrush in the presence buffers than the available buffers, higher proportion of grassland, higher proportion of riparian, and a higher proportion of irrigated agriculture (Table 2). In the winter and summer seasons, the presence buffers were closer to water and in higher densities of water than the available buffers (Table 2). Seasonal Models The model-averaged breeding season model included elevation, grassland, NDVI, riparian, sagebrush, sagebrush/grassland, distance to sagebrush, and distance to water (Table 4). The most important predictor was sagebrush as birds were three times more likely to be found in sagebrush than not. There was an overall negative relationship with elevation, NDVI, distance to sagebrush (indicating closer or within sagebrush), and distance to water whereas there is a positive relationship with sagebrush, grassland, riparian, and sagebrush/grassland (Table 4). The resulting prediction surface indicates large areas of good habitat in non-riparian areas (Fig. 4). The model-averaged winter model included negative relationships with elevation, grassland, NDVI, riparian, distance to sagebrush, and distance to water. There were positive relationships with sagebrush, sagebrush/grassland, and water density (Table 5). Sagebrush was also the most important variable with the probability of a bird location in sagebrush 2.3 times that of a non-use location. The resulting prediction surface was very similar to that from the breeding season with more variations within the large swaths of sagebrush (Fig. 5). The model-averaged summer model included negative relationships with elevation, distance to sagebrush, and distance to water (Table 6). Grassland, NDVI, riparian, Sagebrush/grassland, and water density were positively related. Birds were 3 times as likely to be located in riparian areas, 2.3 times likely to be in sagebrush/grassland, and 2 times as likely to be using vegetation with a higher NDVI value. The resulting prediction surface indicated areas along riparian habitat were more important to the �birds in the summer time (Fig. 6). All 3 of the model-averaged seasonal models validated well (Table 7) including the breeding seasonal model with development, although the average R2 for this model was lower than the other seasonal models. These results emphasize the robustness of the data for all of the seasonal models. Development Models The number of locations, both presence and available, were restricted within 350m of an oil field due to lack of development across the entire study area (Fig. 3). We had reduced presence and absence values for each season for the breeding season (185 presence and 537 absence buffers), winter season (130 presence and 632 absence buffers), and the summer season (44 presence and 532 absence buffers locations). Due to the sparse data in the summer season for presence, we did not analyze the addition of the oil/gas development variable to the models due to possible spurious or false results. We ran the breeding and winter seasonal models with the added development variable, but found that although there was a negative effect of development in the winter, it was considered an uninformative parameter for improving model fit. Therefore, we did not continue with a winter prediction surface for the winter development model. Our breeding season model was improved with the development variable , although the reduced dataset also caused some of the original variables from the breeding model uninformative in the final breeding oil/gas development model (Table 8). The result was a model that included a negative effect of elevation, water density, distance to water, and development and a positive relationship with sagebrush. The prediction surface from the breeding development model indicated a loss of habitat especially in the McCallum field area in the northeast section of North Park (Fig. 7). The development model reduced the amount of highly probable sage grouse presence from 68% to 58% based on a cutoff value of 0.50. Updating Effect of New Models on Colorado statewide GRSG habitat map There is was a distinct difference between the statewide based priority habitat map in the North Park area and the North Park specific analysis (Fig. 9). In the statewide analysis, that did not include any North Park specific data, 147km2 of the North Park population was classified as PGH and 1508km2 as PPH. In our current map developed with North Park specific data, the PGH increased to 792 km2 and the PPH decreased to 864 km2. This The local-scale models provided a much more specific habitat selection analysis rather than the large expanse of priority habitat predicted by the statewide model. We also included comparisons between the raw seasonal models for both the PPH/PGH and the North Park specific models in Appendices B and C. DISCUSSION GRSG selected similar habitat variables during the winter and breeding seasons. The odds of a GRSG being located within sagebrush was 2.3 times as likely in the winter and almost 3 times as likely in the breeding season. This pattern shows in the probability prediction surfaces for both seasons in which the patterns across the landscape are similar. The other variables important during both the winter and breeding season included distance to sagebrush and the sagebrush/grassland vegetation. This result highlights the importance of sagebrush and sagebrush/grassland habitats to GRSG in North Park. GRSG are sagebrush obligates during both the breeding and winter periods and although lekking occurs in open areas, a short distance to sagebrush is critical for escape cover and feeding (Colorado Division of Wildlife 2008, Connelly et al. 2000, Connelly et al. 2011). The summer season probability surface is quite different from the other two seasons, especially since sagebrush was not informative to the model. GRSG used a greater variety of vegetation during the summer season, and riparian and sagebrush/grassland were considered to be the more important variables. Also, birds tended to be much closer to sagebrush when they were not directly in the sagebrush vegetation type. As sagebrush communities dry out, GRSG typically respond by moving to a greater variety of habitats and generally more mesic habitats. Riparian habitats devoid of woody vegetation provide forbs �and insects in the summer (Doherty et al. 2010; Connelly et al. 2000, Connelly et al. 2011). GRSG were also two times as likely to be in areas with a high NDVI value during summer, suggestingthey follow the healthy vegetation more closely in the summer as other plants start to dry out. For example, over the summer period, the NDVI of sagebrush was 0.27-0.29 which is a relatively low value on the NDVI scale (Baghzouz et al. 2010). GRSG may seek out more valuable plant species along the riparian corridors, based on their summer distribution in North Park. The addition of an oil/gas development variable to the seasonal models produced mixed results. First, our summer model did not have sufficient data to provide any useful information about oil/gas development on GRSG distribution. Second, although there were enough data during the winter season and the result was a slightly negative response to the oil/gas development, the variable itself was not considered informative in this analysis. During winter not all roads are plowed or maintained which may result in less traffic in the oil/gas fields and less impact on GRSG. Third, the oil/gas development variable in the breeding season was informative to the model and did improve model fit. The oil/gas development variable negatively influenced GRSG locations. The main drivers of the model were still a positive association with sagebrush and being in areas with lower water density, but we can conclude that the birds avoided oil/gas development during the breeding season. Overall the impact of oil/gas development in the North Park population may not be at a level that could impact GRSG habitat selection other than during the breeding season. This result does not stand when we use all locations across the study area due to the diffusion of the effect as only 6% of the study area has oil/gas development. If we investigate how much habitat is above the 0.5 threshold of probability of GRSG use, the reduction of habitat when the oil/gas development variable was added to the breeding model should cause concern for wildlife and habitat managers. Energy development is a negative impact to GRSG populations even with mitigation impacts (BLM 2013), so any increased development may start affecting the North Park sage-grouse population not only during breeding, but possibly in the other seasons eventually. Copeland et al. (2009) found that we may expect a 7-19% population decline in GRSG across their range from future oil/gas development and the impacts will be greatest in sagebrush habitats. Over 68% of the vegetation within the North Park population boundaries is sagebrush or sagebrush/grassland mix so the impact in this population is a concern for managers. Approximately 52% of the North Park surface is privately owned and with 81% of federal lands possibly having sold their rights (Copeland et al. 2009), oil/gas development potential is potentially problematic for North Park GRSG management. There are a range of proposed anthropogenic disturbance caps being proposed in the DRAFT Northwest Colorado GRSG RMP and EIS (BLM 2013) and these seasonal maps or the PPH/PGH may be used to define the management for these caps. This analysis provides an initial estimate of seasonal habitat selection for GRSG in North Park at what we consider a relatively low oil/gas development level, and may be used to examine potential impacts of future oil/gas development scenarios. Our other objective was to evaluate how incorporating these new seasonal habitat models would modify the existing PPH/PGH map for North Park, update the statewide priority and general habitat mapswhich was created using GRSG location data from across portions of the range in northwestern Colorado with no data coming from theoutside North Park population. The PPH/PGH map is being used to make decisions regarding the development of management alternatives for maintaining and increasing GRSG habitat in Colorado. It is important to make use of the best available data and to update the map when better data become available. Following the same process used for the statewide PPH/PGH map, we found a significant change in the distribution of the priority habitat in the North Park population. The amount of PPH in North Park went from 91% of occupied range when mapped at the statewide level to 52% using our North Park specific models. There is value to both scales of analysis and to both maps for the management of the North Park population. On one hand, our North Park specific analysis does not evaluate any indirect effects that variables such as oil/gas development noise may have on the population. The value of maintaining the courser scale of analysis is in the fact that it provides a conservative estimate of habitat in North Park and would protect more of North Park if we used the PPH/PGH map. This doesn’t directly account for indirect effects of any alterations on the ground, but it provides a more broad �assessment of habitat that may be important to GRSG. On the other hand, the PPH/PGH map indicates that river habitat is important to sage-grouse which may be too general compared to using the North Park specific map that delineates that GRSG are using the edge of riparian habitats rather than the rivers themselves. On the ground management can better be achieved when more details are provided in specific seasons for North Park. Each of these analyses may provide useful information for both policy and on the ground management in different situations, but the use of more local, fine-resolution models are preferable and should be used when available. Often times though, the resources, access, technology, or time are difficult to obtain these local scale models in which case the coarse scale models still provide useful information for management. Use of habitat models specific to North Park and using field data collected for this purpose allowed us to delineate GRSG habitat selection patterns at finer scales than we could with the statewide map. At this smaller scale and with data specific to this population, there is also more detail provided across the North Park landscape which could not be detected at the statewide level. Many previous studies have shown that scale can have a profound effect on the outcome of a resource selection model (DeCesare et al. 2012, McAlpine et al. 2008, Johnson et al. 2004). This analysis also supports the idea that the scale and data input for any analysis can have a major effect on how a population is managed at the local scale. Our ability to add additional variables important to the North Park population also provided us with the ability to be more specific at this smaller scale. We emphasize that although we present a PPH/PGH map for North Park using the new seasonal models, this was for exploratory purposes only and CPW has not officially made any changes to the existing statewide PPH/PGH map. There are numerous policy decisions involved in making changes to the statewide PPH/PGH map, and new local-scale habitat selection models, along with other new biological information, are being considered for other GRSG populations in Colorado as well. Although CPW is committed to using the best information available to guide GRSG conservation, a policy evaluation of alternatives for how to best incorporate new seasonal models e.g., (using quartiles of the distribution surface or some other method), and the timing and frequency of such changes to statewide conservation priority maps, has not been completed. Our local seasonal models in North Park can assist land and wildlife managers to identify areas most important to this population for any mitigation or development proposals that may occur in the future. In the breeding season, the addition of the oil/gas development variable also allows managers and biologists to look at areas that might be impacted most if development were to increase in certain areas. There is also the possibility of conducting “what if” oil/gas development scenarios for the future based on expected changes. Finally, we found that using data specific to the North Park population changed what we considered PPH by a 39% reduction. PPH/PGH mapping could have many effects in an area that is privately owned and may affect landowner and wildlife management practices if the species were to be listed as endangered. The importance of scale in wildlife management should always be dependent on the quality and quantity of data specific to the problem. However, while you always want to get the most detailed data possible for habitat selection models, there are limitations on the ability to design and implement a study to get more exact data as we did in this study. Getting money, field staff, and time to design a study specific to a small population is not always a feasible option for managing a declining species. The goal of using multiscale approaches to habitat selection is to utilize the information from all scales of analysis to develop an effective and comprehensive monitoring and management program for a species. This additional analysis in North Park provides a multiscale picture of seasonal habitat mapping in this population and provides informative products for effective management. LITERATURE CITED Aldridge, C. L. and M. S. Boyce. 2007. Linking occurrence and fitness to persistence: habitat based approach for endangered greater sage-grouse. Ecological applications 17:508-526. Apa, A. D. 1998. Habitat use and movement of sympatric sage and Columbian sharp-tailed grouse in southeastern Idaho. Dissertation, University of Idaho, Moscow, Idaho, USA. �Arnold, T. W. 2010. Uninformative parameters and model selection using Akaike’s Information Criterion. Journal of Wildlife Management 74:1175-1178. Baghzouz, M. D. A. Devitt, L. F. Fenstermaker, and M. H. Young. 2010. Monitoring vegetation phonological cycles in two different semi-arid environmental settings using a ground-based NDVI system: a potential approach to improve satellite data interpretation. Remote sensing 2: 990-1013. Bureau of Land Management (BLM) 2013. Northwest Colorado Greater sage-grouse draft land use plan amendment and environmental impact statement. August 2013. Blomberg, E. J., J. S. Sedinger, M. T. Atamian, and D. V. Nonne. 2012. Characteristics of climate and landscape disturbance influence the dynamics of greater sage-grouse populations. Ecosphere 3: 1 20. Boyce, M. S., M. G. Turner, J. Fryxell, and P. Turchin. 2003. Scale and heterogeneity in habitat selection by elk in Yellowstone National Park. Ecoscience 10: 421-431. Boyce, M. S., P. R. Vernier, S. E. Nielsen, and F. K. A. Schmiegelow. 2002. Evaluating resource selection functions. Ecological Modelling 157:281-300. Braun, C. E., O. O. Oedekoven, and C. L. Aldridge, 2002. Oil and gas development in western North America: effects on sagebrush steppe avifauna with particular emphasis on sage grouse. Transactions of the North American Wildlife and Natural Resources Conference 67:337-349. Burnham, K. P. and D. R. Anderson. 2002. Model selection and multimodel inference: a practical information-theoretic approach. Springer-Verlag, New York, New York, USA. Colorado Division of Wildlife. Colorado Greater sage-grouse steering Committee. 2008. Colorado greater sage-grouse conservation plan. Colorado Division of Wildlife, Denver, USA. Connelly, J. W., M. A. Schroeder, A. R. Sands, and C. E. Braun. 2000. Guidelines to manage sage-grouse populations and their habitats. Wildlife Society Bulletin 28:967-985. Connelly et al. 2011 in SAB…I can get if you need it. It is the habitat use review chapter. Copeland, H. E., K. E. Doherty, D. E. Naugle, A. Pocewicz, and J. M. Kiesecker. 2009. Mapping oil and gas development potential in the US intermountain West and estimating impacts to species. PlosOne 4:e7400. Cunningham, R., D. Lindenmayer, P. Barton, K. Ikin, M. Crane, D. Michael, S. Okada, Ph. Gibbons, and J. Stein. 2014. Cross-sectional and temporal relationships between bird occupancy and vegetation cover at multiple spatial scales. Ecological applications 24: 1275-1288. Dalke, P. D., D. B. Pyrah, D. C. Stanton, J. E. Crawford, and E. F. Schlatterer. 1963. Ecology, productivity and management of sage grouse in Idaho. Journal of Wildlife Management 27:811841. De Cesare, N. J., M. Hebblewhite, F. Schmiegelow, D. Hervieux, G. J. McDermid, L. Neufeld, M. Bradley, J Whittington, K. G. Smith, L. E. Morgantini, M. Wheatley, and M. Musiani. 2012. Transcending scale dependence in identifying habitat with resource selection functions. Ecological applications 22: 1068-1083. Doherty, K. E., D. E. Naugle, and B. L. Walker. 2010. Greater sage-grouse nesting habitat: the importance of managing at multiple scales. Journal of Wildlife Management 74:1544-1553. Doherty, K. E., D. E. Naugle, B. L. Walker, and J. M. Graham. 2008. Greater sage-grouse winter habitat selection and energy development. Journal of Wildlife Management 72:187-195. Dzialak, M. R., C. V. Olson, S. M. Harju, S. L. Webb, and J. B. Winstead. 2012. Temporal and hierarchical spatial components of animal occurrence: conserving seasonal habitat for greater sage-grouse. Ecosphere 3: 1-17. Fedy, B. C., C. L. Aldridge, K. E. Doherty, M. O’donnell, J. L. Beck, B. Bedrosian, M. J. Holloran, G. D. Johnson, N. W. Kaczor, C. P. Kirol, C. A. Mandich, D. Marshall, G. McKee, C. Olson, C. C. Swanson, and B. L. Walker. 2012. Interseasonal movements of greater sage-grouse, migratory behavior, and an assessment of the core regions concept in Wyoming. The Journal of Wildlife Management 76:1062-1071. �Fischer, R. A. 1994. The effects of prescribed fire on the ecology of migratory sage grouse in southeastern Idaho. Dissertation, University of Idaho, Moscow, Idaho, USA. Fischer, R. A., K. P. Reese, and J. W. Connelly. 1996a. An investigation on fire effects within xeric sage grouse brood habitat. Journal of Range Management 49: 194-198. Fischer, R. A., K. P. Reese, and J. W. Connelly. 1996b. Influence of vegetal moisture content and nest fate on timing of female sage grouse migration. Condor 98: 868-872. Giesen, K. M., T. J. Schoenberg, and C. E. Braun. 1982. Methods for trapping sage grouse in Colorado. Wildlife Society Bulletin 10:224-231. Gillies, C. S., M. Hebblewhite, S. E. Nielsen, M. A. Krawchuk, C. L. Aldridge, J. L. Frair, D. J. Saher, C. E. Stevens, and C. L. Jerde. 2006. Application of random effects to the study of resource selection by animals. Journal of Animal Ecology 75:887–898. Harju, S. M., M. R. Dzialak, R. C. Taylor, L. D. Hayden-Wing, and J. B. Winstead. 2010. Thresholds and time lags in effects of energy development on greater sage-grouse populations. Journal of Wildlife Management 74:437–448. Hirzel, A. H., G. Le Lay, V. Helfer, C. Randin, and A. Guisan. 2006. Evaluating the ability of habitat suitability models to predict species presences. Ecological modeling 199: 142-152. Johnson, C. J., and M. P. Gillingham. 2008. Sensitivity of species-distribution models to error, bias, and model design: an application to resource selection functions for woodland caribou. Ecological Modelling 213:143–155. Johnson, C. J., S. E. Niesen, E. H. Merrill, T. L. McDonald, and M. S. Boyce. 2006. Resource selection functions based on use-availability data: theoretical motivation and evaluation methods. Journal of Wildlife Management 70:347–357. Johnson, C. J., D. R. Seip, and M. S. Boyce. 2004. A quantitative approach to conservation planning: using resource selection functions to map the distribution of mountain caribou at multiple spatial scales. Journal of Applied Ecology 41:238–251. Klar, N., N. Fernández, S. Kramer-Schadt, M. Herrmann, M. Trinzen, I. Büttner, and C. Niemitz. 2008. Habitat selection models for the European wildcat conservation. Biological Conservation 141:308–319. Lukacs, P. M., K. P. Burnham, and D. R. Anderson. 2010. Model selection bias and Freedman’s paradox. Annals of the Institute of Statistical Mathematics: Special Issue of AISM in Honor of Dr. Hirotugu Akaike 62:117–125. Manly, B. F. J., L. L. McDonald, D. L. Thomas, T. L. McDonald, and W. P. Erickson. 2002. Resource selection by animals: statistical design and analysis for field studies. Kluwer Academic Publishers, Dordrecht, The Netherlands. McAlpine, C. A., J. R. Rhodes, M. E. Bowen, D. Lunney, J. G. Callaghan, D. L. Mitchell, and H. P. Possingham. 2008. Can multiscale models of species’ distribution be generalized from region to region? A case study of koala. Journal of Applied Ecology 45:558–567. Naugle, D. E., K. E. Doherty, B. L. Walker, M. J. Holloran, and H. E. Copeland. 2011. Energy development and Greater sage-grouse. Pp. 489-503 in S. T. Knick and J. W. Connelly (editors). Greater Sage-Grouse: ecology and conservation of a landscape species and its habitats. Studies in Avian Biology (vol. 38), University of California Press, Berkeley, CA. Northrup, J. M., M. B. Hooten, C. R. Anderson, Jr., and G. Wittemyer. 2013. Practical guidance on characterizing availability in resource selection functions under a use-availability design. Ecology 94: 1456-1463. NTT (sage-grouse national technical team). 2011. A report on national greater sage-grouse conservation measures. December 2011. Pettorelli, N., S. Ryan, T. Mueller, N. Bunnefeld, B. Jedrzejewska, M. Lima, and K. Kausrud. 2011. The normalized difference vegetation index (NDVI): unforeseen successes in animal ecology. Climate research 46:15-27. �Rice, M. B., T. D. Apa, M. L. Phillips, J. H. Gammonly, B. F. Petch, and K. Eichhoff. 2013. Analysis of regional species models based on radio-telemetry datasets from multiple small-scale studies. Journal of Wildlife management 77:821-831. Schroeder, M. A., C. L. Aldridge, A. D. Apa, J. R. Bohne, C. E. Braun, S. D. Bunnell, J. W. Connelly, P. A. Deibert, S. C. Gardner, M. A. Hilliard, G. D. Kobriger, and C. W. McCarthy. 2004. Distribution of sage-grouse in North America. Condor 106:363–376. Shirk, A. J., M. G. Raphael, and S. A. Cushman. 2014. Spatiotemporal variation in resource selection: insights from the American marten (Martes Americana). Ecological applications: 1434-1444. Smith, K. T., C. P. Kirol, J. L. Beck, and F. C. Blomquist. 2014. Prioritizing winter habitat quality for greater sage-grouse in a landscape influenced by energy development. Ecosphere 5:1-20. Springer, J. T. 1979. Some sources of bias and sampling error in radio triangulation. Journal of Wildlife Management 43:926–935. Stokland, J. N., R. Halvorsen, and B. Stoa. 2011. Species distribution modeling- Effect of design and sample size of pseudo-absence observations. Ecological Modelling 222:1800-1809. Tedrow, L. and K. T. Weber. 2011. NDVI changes over a calendar year in the rangelands of southeast Idaho. Pages 105-116 in K. T. Weber and K. Davis (Eds.), Final report: assessing post-fire recovery of sagebrush-steppe rangelands in Southweastern Idaho (NNX08AO90G). 252 pp. U. S. Fish and Wildlife Service, 2010, Endangered and threatened wildlife and plants; 12-month findings for petitions to list the greater sage-grouse (Centrocercus urophasianus) as threatened or endangered; proposed rule: Federal Register, v. 75, p. 13,910-14,014. Wakkinen, W. L., K. P. Reese, J. W. Connelly, and R. A. Fischer. 1992. An improved spotlighting technique for capturing sage-grouse. Wildlife Society Bulletin 20:425-426. Walker, B. L., D. E. Naugle, and K. E. Doherty. 2007. Greater Sage-grouse population response to energy development and habitat loss. Journal of Wildlife Management 71:2644-2654. Wiens, T. S., B. C. Dale, M. S. Boyce, and G. P. Kershaw. 2008. Three way k-fold cross-validation of resource selection functions. Ecological Modelling 212:244–255. �Table 1. Movement statistics (movement between consecutive locations for each individual) for North Park greater sage-grouse telemetry data from April 2010 to February 2012. Overall Breeding Winter Summer Average movement (m) 2170.42 1934.87 3234.99 883.95 Average days between locations 12.55 12.83 15.89 10.64 Movement per day (m) 172.94 150.81 203.59 83.08 Movement per week (m) 1210.58 1055.67 1425.13 581.56 ��Table 2. Variables used in greater sage-grouse habitat models with the mean value for presence and absence buffers in the breeding, winter, and summer seasons. Breeding Variable presence mean (range) Winter absence mean (range) presence mean (range) Summer absence mean (range) presence mean (range) absence mean (range) Sagebrush 0.803 (0.00-1.00) 0.482 (0.00-1.00) 0.783 (0.00-1.00) 0.487 (0.00-1.00) 0.477 (0.00-1.00) 0.504 (0.00-1.00) Grassland 0.036 (0.00-0.728) 0.115 (0.00-0.983) 0.038 (0.00-0.603) 0.116 (0.00-0.981) 0.127 (0.00-1.00) 0.107 (0.00-1.00) Sagebrush/grassland 0.110 (0.00-0.938) 0.105 (0.00-1.00) 0.101 (0.00-1.00) 0.104 (0.00-1.00) 0.140 (0.00-1.00) 0.106 (0.00-1.00) Riparian 0.006 (0.00- 0.357) 0.024 (0.00-0.965) 0.004 (0.00-0.414) 0.024 (0.00-0.843) 0.031 (0.00-0.657) 0.021 (0.00-0.946) Irrigated agriculture 0.027 (0.00-0.921) 0.150 (0.00-1.00) 0.026 (0.00-0.822) 0.150 (0.00-1.00) 0.167 (0.00-1.00) 0.136 (0.00-1.00) Elevation 8250.9 (7880-8808) 8330.5 (7845-9715) 8204.7 (7892-8786) 8329.7 (7849.9-9768) 8200.9 (7834-8831) 8331.2 (7839-9876) NDVI 0.277 (0.189-0.509) 0.333 (-0.004-0.602) 0.012 (-0.016-0.089) 0.022 (-0.032-0.289) 0.451 (0.227-0.708) 0.456 (-0.009-0.823) Water density 1.905 (0.350-5.238) 2.060 (0.00-6.258) 2.266 (0.322-5.894) 2.063 (0.00-6.283) 2.634 (0.569-5.270) 2.018 (0.00-6.335) Distance to water 0.256 (0.001-0.994) 0.243 (0.00-2.252) 0.205 (0.00-1.192) 0.243 (0.00-1.870) 0.156 (0.00-1.081) 0.244 (0.00-2.165) Distance to sagebrush 0.0008 (0.00-0.352) 0.042 (0.00-1.447) 0.0004 (0.00-0.237) 0.034 (0.00-1.812) 0.019 (0.00-0.508) 0.058 (0.00-1.817) �Table 3. Status of wells from the Colorado Oil and Gas Commission list of oil wells located in the North Park Study area. Status *PR DA PA AL *IJ XX SI *DG WO Name Activity Producing Dry and Abandoned Permanently abandoned Abandoned Location Injection Well Proposed location Shut-in well Drilling Waiting on completion # of wells Definition 119 232 146 82 39 30 25 3 1 currently producing oil, gas, and/or water well is no longer productive and was abandoned well was permanently plugged and abandoned well was abandoned a well used for pumping water or gas into reservoir a new or proposed location a well capable of producing, but not currently a well being used with rigs and crews status undetermined and waiting *These categories were deemed “active” and used in the development layers as long as there was visual confirmation active inactive inactive inactive active on hold on hold active on hold �Table 4. Coefficients for the average model set for greater sage-grouse in North Park in the breeding season including 85% confidence intervals and odds ratios. Variable estimate SE LCI UCI Intercept -2.286 Elevation -0.108 0.048 -0.178 -0.040 Grassland 0.166 0.077 0.055 NDVI -0.353 0.064 Riparian 0.149 Sagebrush Odds ratio LCI UCI 0.898 0.838 0.961 0.277 1.180 1.057 1.320 -0.446 -0.260 0.702 0.640 0.770 0.056 0.068 0.230 1.161 1.071 1.259 1.097 0.110 0.930 1.251 2.994 2.556 3.508 sagebrush/grassland 0.382 0.050 0.307 0.453 1.465 1.363 1.575 distance to sagebrush -0.781 0.279 -1.186 -0.381 0.458 0.306 0.685 distance to water -0.082 0.037 -0.135 -0.030 0.922 0.874 0.971 �Table 5. Coefficients for the average model set for greater sage-grouse in North Park in the winter season including 85% confidence intervals and odds ratios. Variable estimate SE LCI UCI Intercept -2.483 Elevation -0.411 0.052 -0.485 -0.336 Grassland -0.318 0.078 -0.430 NDVI -0.439 0.061 Riparian -0.437 Sagebrush Odds ratio LCI UCI 0.663 0.616 0.714 -0.206 0.728 0.651 0.814 -0.527 -0.352 0.644 0.591 0.703 0.104 -0.587 -0.288 0.646 0.556 0.750 0.835 0.083 0.716 0.953 2.305 2.047 2.595 sagebrush/grassland 0.266 0.041 0.207 0.324 1.304 1.230 1.383 distance to sagebrush -1.490 0.435 -2.117 -0.864 0.226 0.121 0.422 water density 0.379 0.038 0.324 0.434 1.461 1.382 1.543 distance to water -0.289 0.038 -0.345 -0.233 0.749 0.708 0.792 �Table 6. Coefficients for the average model set for greater sage-grouse in North Park in the summer season including 85% confidence intervals and odds ratios. Variable estimate Intercept -2.369 Elevation -0.002 Grassland SE LCI UCI Odds ratio LCI UCI 0.0002 -0.002 -0.001 0.998 0.998 0.999 0.660 0.239 0.315 1.004 1.934 1.370 2.730 NDVI 0.786 0.406 0.202 1.371 2.120 1.224 3.938 Riparian 1.128 0.496 0.414 1.843 3.091 1.512 6.136 sagebrush/grassland 0.833 0.198 0.548 1.118 2.300 1.730 3.058 distance to sagebrush -6.611 0.780 -7.734 -5.489 0.001 0.000 0.004 water density 0.376 0.045 0.311 0.441 1.456 1.365 1.554 distance to water -0.999 0.253 -1.364 -0.634 0.368 0.256 0.530 �Table 7. Cross validated spearman-rank correlations (rs) between RSF bin ranks and area-adjusted frequencies for individual and average model sets. Included are the Area Under the Curve estimates for each model set to compare with the original model. Set Breed AUC Winter rs AUC Summer rs 0.775 AUC rs 0.731 Breed developed AUC rs Original model 0.745 0.669 1 0.750 0.988 0.768 0.997 0.734 0.954 0.742 0.708 2 0.756 0.976 0.801 1.000 0.723 0.939 0.865 0.880 3 0.757 0.967 0.768 0.988 0.744 0.976 0.775 0.766 4 0.720 0.976 0.646 0.902 0.744 0.964 0.799 0.903 5 0.735 0.903 0.663 0.854 0.690 0.964 0.826 0.803 Average 0.744 0.962 0.729 0.948 0.727 0.959 0.801 0.812 �Table 8. Coefficients for the average model set for greater sage-grouse in North Park in the breeding season with the inclusion of oil and gas development in the model along with the 85% confidence intervals and odds ratios. Variable estimate SE LCI Intercept -2.052 Elevation sagebrush UCI Odds ratio LCI UCI -0.627 0.167 -0.868 -0.387 0.543 0.427 0.690 0.806 0.154 0.584 1.027 2.285 1.831 2.850 development -0.412 0.167 -0.652 -0.171 0.655 0.515 0.833 water density -1.237 0.193 -1.515 -0.959 0.283 0.214 0.373 distance to water -0.205 0.118 -0.374 -0.033 0.815 0.687 0.965 �Figure 1. The six Greater sage-grouse populations in Colorado. �Figure 2. Greater sage-grouse preliminary priority habitat (PPH) and preliminary general habitat (PGH) used for the Bureau of Land Management’s Northwest Colorado greater sage-grouse draft land use plan amendment and environmental impact statement (BLM 2013). Priority habitat was determined by combining areas with high probability of presence from models for breeding, winter, and summer seasons (Rice et al. 2013) along with 4 mile buffers around active leks. General habitat is that which birds have occupied outside of priority habitat. North Park population prediction �Figure 3. Oil fields as outlined by the Colorado Oil and Gas Conservation Commission (COGCC) as defined by producing and/or plugged and abandoned wells and provide approximate boundaries (valid as of January 2011). �Figure 4. Probability of greater sage-grouse presence in North Park during the breeding season. �Figure 5. Probability of greater sage-grouse presence in North Park during the winter season. �Figure 6. Probability of greater sage-grouse presence in North Park during the summer season. �Figure 7. Probability of greater sage-grouse presence in North Park during the breeding season when an oil and gas development variable is included in the model. �Figure 8. Comparison of the breeding season probability surface and the probability surface for the breeding season which included oil/gas development in the model. Breeding model with oil/gas development variable Oil/gas fields Baseline breeding model (no oil/gas development) Oil/gas fields �Figure 9. Comparison of the Preliminary Priority Habitat (PPH) and Preliminary General Habitat (PGH) map created by Colorado Parks and Wildlife based on Rice et al. 2012 models (no North Park data collected), and the map created using the North Park models developed in this paper. Statewide PPH/PGH map North Park Specific PPH/PGH map �Appendix A. Vegetation classifications based on the basinwide vegetation layer Basinwide category High density residential areas, lawns, planted trees. Irrigated crops and fields. Rangeland dominated by annual and perennial grasses. Rangeland codominated by grasses and forbs. Disturbed or overgrazed rangeland. Sparsely vegetated grasslands, 10-40% vegetation. Sagebrush with rabbitbrush, bitterbrush. Low elevation shrubland dominated by greasewood. Shrubland dominated by PUTR2 (Bitterbrush). Codominate sagebrush shrubland and perennial grassland. Codominate sagebrush/Mesic Mtn shrub mixed with grass/forb. Deciduous forest dominated by Aspen. Codominate Aspen and Gambel oak deciduous woodland. Coniferous forest dominated by PSME. Coniferous forest dominated by PICO. Coniferous forest dominated by PIFL. Coniferous forest co-dominated by PICO, PIEN, and ABCO. Mixed forest codominated by Aspen and PICO. Talus and scree slopes, nearly 100% rock. Bare soil and fallow agriculture fields. High elevation meadows co-dominated by grass and forbs Shrub riparian areas consisting primarily of shrub willows. Shrub riparian areas dominated by shrub willow species. Non-woody riparian areas consisting primarily of sedges. Lakes, reservoirs, rivers, streams. class used for North Park models residential irrigated agriculture grassland grassland grassland bare ground sagebrush greasewood bitterbrush sagebrush/grassland sagebrush aspen aspen forest forest forest forest forest talus bare alpine riparian riparian herbaceous riparian water �Appendix B. Comparison between the percent of habitat predicted in each quartile for the PPH/PGH seasonal models and the North Park specific models based on the fourth quartile cutoff values from the PPH/PGH models for the A) breeding season, B) summer season, and C) winter seasons. A) Breeding season Cutoff values PPH/PGH percentages North Park percentages Quartile 1 0.04 0.00% 9.70% Quartile 2 0.53 10.1% 26.6% Quartile 3 0.76 31.8% 31.4% Quartile 4 1.00 58.1% 32.3% Quartile 1 0.23 0.00% 21.4% Quartile 2 0.46 24.9% 5.6% Quartile 3 0.55 33.5% 7.4% Quartile 4 1.00 41.4% 65.6% Quartile 1 0.09 8.3% 18.4% Quartile 2 0.15 16.1% 3.1% Quartile 3 0.23 60.0% 3.2% Quartile 4 1.0 15.6% 75.4% B) Summer season Cutoff values PPH/PGH percentages North Park percentages C) Winter season Cutoff values PPH/PGH percentages North Park percentages �Appendix C. Comparison between the prediction surfaces for the PPH/PGH seasonal models and the North Park specific models based on the fourth quartile cutoff values from the PPH/PGH models for the A) breeding season, B) summer season, C) winter seasons, and D) combined seasons. A) Breeding season B) Summer season �C) Winter season D) Combined seasons �� Dublin Core The Dublin Core metadata element set is common to all Omeka records, including items, files, and collections. For more information see, http://dublincore.org/documents/dces/. Title A name given to the resource Reports Text A resource consisting primarily of words for reading. Examples include books, letters, dissertations, poems, newspapers, articles, archives of mailing lists. Note that facsimiles or images of texts are still of the genre Text. Dublin Core The Dublin Core metadata element set is common to all Omeka records, including items, files, and collections. For more information see, http://dublincore.org/documents/dces/. Title A name given to the resource Development of distribution models for management of greater sage-grouse in North Park Colorado Description An account of the resource Rangewide declines of greater sage-grouse and recent energy development within sagebrush habitat has led to concern for conservation of greater sage-grouse (<em>Centrocercus urophasianus</em>) (GRSG) populations across Colorado, including in North Park, which supports approximately 20% of the state’s GRSG. Seasonal variations in habitat use by GRSG can provide important information for biologists and managers on the ground. These habitats have been mapped at the statewide level at a large scale, but have not been completed specifically for the North Park population. Investigating the smaller scale seasonal habitat selection of GRSG in North Park is important as no data from North Park was used in the statewide analysis. Since GRSG habitat use is known to be influenced by both landscape-scale and local-scale factors, data specific to North Park can be used to refine the statewide models at a more local scale. Creator An entity primarily responsible for making the resource Rice, Mindy B. Subject The topic of the resource Greater sage-grouse <em>Centrocercus urophasianus</em> Wildlife management Extent The size or duration of the resource. 36 pages Date Created Date of creation of the resource. 2014 Rights Information about rights held in and over the resource <a href="http://rightsstatements.org/vocab/InC-NC/1.0/" target="_blank" rel="noreferrer noopener">In Copyright - Non-Commercial Use Permitted</a> Type The nature or genre of the resource Text Format The file format, physical medium, or dimensions of the resource application/pdf Language A language of the resource English Is Part Of A related resource in which the described resource is physically or logically included. Cost Center 3420 Avian Research Work Package 1663 Terrestrial species conservation