https://cpw.cvlcollections.org/files/original/95709d1f90929625d2b37564121fc9b3.pdf 200fea96bc706189146eef0d15568d3a PDF Text Text The research in this publication was partially or fully funded by Colorado Parks and Wildlife. Dan Prenzlow, Director, Colorado Parks and Wildlife • Parks and Wildlife Commission: Marvin McDaniel, Chair • Carrie Besnette Hauser, Vice-Chair Marie Haskett, Secretary • Taishya Adams • Betsy Blecha • Charles Garcia • Dallas May • Duke Phillips, IV • Luke B. Schafer • James Jay Tutchton • Eden Vardy �Mammalian responses to changed forest conditions resulting from bark beetle outbreaks in the southern Rocky Mountains JACOB S. IVAN,1, AMY E. SEGLUND,2 RICHARD L. TRUEX,3 AND ERIC S. NEWKIRK1 1 Colorado Parks and Wildlife, 317 West Prospect Road, Fort Collins, Colorado 80526 USA Colorado Parks and Wildlife, 2300 South Townsend Avenue, Montrose, Colorado 81401 USA 3 U.S. Department of Agriculture Forest Service, Rocky Mountain Region, 1617 Cole Boulevard, Building 17, Lakewood, Colorado 80401 USA 2 Citation: Ivan, J. S., A. E. Seglund, R. L. Truex, and E. S. Newkirk. 2018. Mammalian responses to changed forest conditions resulting from bark beetle outbreaks in the southern Rocky Mountains. Ecosphere 9(8):e02369. 10.1002/ ecs2.2369 Abstract. Spruce beetle (Dendroctonus rufipennis) and mountain pine beetle (Dendroctonus ponderosae) outbreaks have impacted millions of acres of conifer forest from Alaska to northern Mexico. These species are native to North America, and periodic outbreaks have shaped the structure and composition of conifer forests for millennia. However, the extent and severity of current outbreaks, fueled by favorable climatic conditions and increased susceptibility of forests, are unmatched in recorded history. To characterize the response of a suite of mammalian species to beetle-induced changes in vegetation in the southern Rocky Mountains, we deployed cameras at 300 randomly selected sites during summer 2013–2014. Selected sites spanned gradients of years elapsed since bark beetle outbreaks (YSO) and severity. We fit single-season occupancy models to detection/non-detection data collected for each species to examine a variety of plausible relationships between use of a given stand and YSO, severity, or both. Ungulates exhibited a positive association with bark beetle activity, although the nature of these associations varied by species. Elk (Cervus canadensis) were positively associated with severity, but not YSO; mule deer (Odocoileus hemionus) exhibited the opposite relationship. Moose (Alces alces) responded in a quadratic fashion; use of forest stands adjacent to preferred willow habitat peaked 3–7 yr after an outbreak commenced, but only at high severity. Similarly, yellow-bellied marmot use of impacted stands adjacent to rock outcroppings followed a quadratic trend, but only at high severity. Red squirrel (Tamiasciurus hudsonicus) use declined in severely impacted stands, likely as a response to diminished cone crops. Golden-mantled ground squirrels (Callospermophilus lateralis) and chipmunks (Neotamias spp.) exhibited a shallow negative relationship with YSO, as did coyotes (Canis latrans). Contrary to our hypotheses, black bears (Ursus americanus), American marten (Martes americana), snowshoe hares (Lepus americanus), and porcupines (Erethizon dorsatum) did not appear to be substantially influenced by beetle activity. Red fox (Vulpes vulpes) use was positively associated with YSO, but overall use declined as severity increased. Note that changes in probability of use described here could reflect changes in abundance, home range size, habitat use, or some combination, and in several cases, there was considerable uncertainty across competing models. Key words: bark beetle outbreak; camera trap; climate change; Colorado; Dendroctonus ponderosae; Dendroctonus rufipennis; mammals; mountain pine beetle; spruce beetle. Received 29 May 2018; accepted 7 June 2018. Corresponding Editor: James W. Cain III. Copyright: © 2018 The Authors. This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. E-mail: Jake.Ivan@state.co.us ❖ www.esajournals.org 1 August 2018 ❖ Volume 9(8) ❖ Article e02369 �IVAN ET AL. INTRODUCTION types have been heavily and extensively impacted by two species of bark beetle: mountain pine beetle (Dendroctonus ponderosae) and spruce beetle (Dendroctonus rufipennis), respectively. Two primary factors have propelled current outbreaks of these bark beetle species to epidemic levels. First, historical events have primed the susceptibility of these host forests to bark beetle outbreaks. That is, in many places forests are characterized by relatively dense, uniform stands of mature trees that are favored for successful completion of the bark beetle life cycle (Fettig et al. 2007, Raffa et al. 2008, Bentz et al. 2009). This has resulted largely from stand-replacing forest fires and extensive timber harvest near the turn of the 20th century (Fettig et al. 2007, Bentz et al. 2009). Second, rising global temperatures have increased the vulnerability of host trees to infestation, while simultaneously enhancing the growth and development of individual beetles and beetle populations (Bentz et al. 2009, Sambaraju et al. 2012). Severe drought caused by warm regional temperatures creates water stress in trees, which reduces their ability to take in carbon and decreases resources available for growth, tissue repair, and construction of biochemical defense systems. Thus, under prolonged drought conditions, trees are more likely to succumb to beetles and the threshold number of beetles necessary for a successful attack is reduced (Bentz et al. 2009). Rising temperatures associated with climate change also reduce constraints on growth of beetle populations, further facilitating eruptions. For instance, eggs and larvae of mountain pine beetles are susceptible to cold-induced mortality during fall and spring when their cryogenic defense mechanisms are minimal (Bentz et al. 1991, 2009). A reduction in cold snaps during these periods directly improves survival rate of eggs, larvae, and brood adults, and thus improves recruitment. Likewise, the spruce beetle life cycle includes pre-pupal diapause during which larval development is paused in late summer and resumes again the following summer after temperatures reach a critical threshold (Schebeck et al. 2017). However, warming temperatures permit diapause omission, resulting in completion of the life cycle in a single year, which can quickly lead to exponential population growth (Hansen and Bentz 2003). Native bark beetles are important disturbance agents in conifer forests of western North America. Along with fire, they have shaped forest composition and structure for millennia (Veblen 2000). While insect activity is a naturally occurring phenomenon in this region, the simultaneous eruptions of several species of bark beetle, and the extent and severity of these impacts (e.g., billions of trees situated on hundreds of millions of acres), are unmatched in recorded history (Bentz et al. 2009). Generally, the life cycle of bark beetles proceeds as follows: Adults attack a specific species of live conifer tree, usually of larger diameter (e.g., >10 cm diameter at breast height), by burrowing through the bark and into the phloem. There, they copulate, excavate egg galleries, and deposit their eggs. When larvae hatch, they produce more galleries as they feed on phloem tissue. Eventually, larvae pupate and adults emerge from the tree, ready to attack again. The excavation of the egg galleries by the adult and feeding galleries by the larvae serves to interrupt the flow of water and nutrients within the tree resulting in its death (Raffa et al. 2008, Bentz et al. 2009, 2010). This phenomenon has ecological consequences at multiple spatio-temporal scales. At the forest stand scale, beetle outbreaks and ensuing death of individual trees alter the age, size, and species composition of the stand. The death of individual trees also results in decreased canopy cover, changes in the understory due to increased sunlight, and accumulation of coarse wood. (Raffa et al. 2008, Bentz et al. 2009). The effects of these relatively sudden changes propagate through several decades as the stand is reverted to an earlier successional sere. At a larger scale, impacts caused by beetles at the stand level manifest as changed mosaics of stand types across the landscape, the signature of which may last a century or more (Raffa et al. 2008). Lodgepole pine (Pinus contorta) and Engelmann spruce (Picea engelmannii)–subalpine fir (Abies lasiocarpa) forests are among the most common and expansive cover types in mountainous regions of western North America (Alexander and Shepperd 1990, Alexander et al. 1990, Lotan and Chritchfield 1990). In recent decades, these forest ❖ www.esajournals.org 2 August 2018 ❖ Volume 9(8) ❖ Article e02369 �IVAN ET AL. survival, reproduction, competition, predation, and other axes encompassed in a species’ niche (Morrison et al. 1992). Thus, it is an important and logical starting point to examine response to significant environmental changes induced by bark beetles. Based on previous work and the natural history and ecology of the species sampled, we expected a variety of direct responses to changes in vegetation (Table 1). However, we also anticipated complex, indirect responses that potentially cascade across trophic levels. For example, we might easily predict a decline in red squirrel use of conifer stands after bark beetle outbreaks due to the loss of cone crops resulting from the death of mature trees (Koprowski 2005, Johnson et al. 2015). In contrast, the response by American marten is more difficult to predict. Martens prey on red squirrels and use their middens as resting and denning habitat (Buskirk and Ruggiero 1994). Thus, like red squirrels, we might expect use to decline in post-beetle stands. However, use of such stands by marten might also increase with expected increases in red-back vole populations (Saab et al. 2014) and availability of coarse wood and large snags, which are important components of marten diet and habitat (Buskirk and Ruggiero 1994). Other important habitat components such as snowpack and subnivean spaces could be impacted by loss of canopy cover; all of these changes could unpredictably impact the competitive abilities of martens compared to other carnivores that use these forests. Thus, our current, broadscale effort is limited to descriptions of observed changes in use associated with beetle outbreaks. We offer potential mechanisms for observed changes where possible, but these should largely serve as hypotheses for future research. Lodgepole pine and spruce-fir forests in western North America provide habitat for a wide array of mammalian wildlife including game species such as elk (Cervus canadensis) and mule deer (Odocoileus hemionus), sensitive species such as Canada lynx (Lynx Canadensis) and American marten (Martes americana), and foundational members of the subalpine community such as red squirrels (Tamiasciurus hudsonicus). Despite the varied and high interest in these species, and the massive changes in their habitat across North America, little has been done to examine the consequences of bark beetle outbreaks on these and other members of the mammalian community. Saab et al. (2014) reviewed literature on response of mammalian species to mountain pine beetle outbreaks in lodgepole forests and found only a handful of works, none of which were peerreviewed. Most existing literature focused on response of red squirrels, which was generally negative (Stone 1995, Drever and Martin 2007, Mosher 2011). Novel results published by Saab et al. (2014) in addition to their review, and later work by Johnson et al. (2015), confirmed this result. Saab et al. (2014) and Stone (1995) found that red-backed vole (Myodes gapperi) response is flat or mixed, and tied mostly to accumulation of coarse woody debris. Stone (1995) also suggested that ungulate species responded positively to mountain pine beetle outbreaks in northern Utah, as did golden-mantled ground squirrels (Callospermophilus lateralis), but responses of snowshoe hares (Lepus americanus), chipmunk (Neotamias spp.), and red-backed voles were muted or mixed. Still, a paucity of published information exists regarding mammalian response to bark beetle outbreaks, especially for species other than red squirrels in systems other than lodgepole pine. Here, we present an occupancy analysis designed to assess mammalian response to mountain pine beetle and spruce beetle outbreaks in lodgepole pine and spruce-fir forests of the southern Rocky Mountains where over 4 million acres have been impacted since 1994 (USFS Rocky Mountain Region 2018). Specifically, we estimate probability of use of these two forest types by various mammalian species and describe changes in this probability as a function of time elapsed since beetle outbreaks and severity of those outbreaks. Habitat use is fundamental to wildlife ecology as it sets the stage for ❖ www.esajournals.org METHODS Study area The study area included all subalpine forests (lodgepole pine and spruce-fir) in the state of Colorado (Fig. 1). These forests occurred between 2590 and 3660 m elevation (Benedict 2008:497, 538) in steep and varied topography that generated drastic changes in vegetation over short distances depending on slope, aspect, and elevation. Generally, at the lower end of this 3 August 2018 ❖ Volume 9(8) ❖ Article e02369 �IVAN ET AL. Table 1. Summary of data collected (number of photographs), analysis status (whether data were sufficient to complete the three-step analysis), predicted response to bark beetle outbreaks (generally, without differentiating response through time from response to severity), and rationale for predicted response for species detected via camera traps across a gradient of bark beetle-impacted areas in Colorado, USA, summers 2013–2014. Photographs Analysis completed Predicted response 1901 Yes + 13,124 Yes + Chipmunk (Neotamias spp.) 2284 Yes + Coyote (Canis latrans) 1869 Yes None Elk (Cervus canadensis) 132,953 Yes + Golden-mantled ground squirrel (Callospermophilus lateralis) Moose (Alces alces) 1780 Yes + 2652 Yes + Mule deer (Odocoileus hemionus) 55,220 Yes + Porcupine (Erethizon dorsatum) 1246 Yes — Red fox (Vulpes vulpes) 2421 Yes None Red squirrel (Tamiasciurus hudsonicus) 17,018 Yes — Snowshoe hare (Lepus americanus) 17,484 Yes + Yellow-bellied marmot (Marmota flaviventris) Bobcat (Lynx rufus) Bushy-tailed woodrat (Neotoma cinerea) Canada lynx (Lynx canadensis) Mountain cottontail (Sylvilagus nuttallii) Mountain lion (Puma concolor) Mouse (Peromyscus spp.) Pika (Ochotona princeps) Raccoon (Procyon lotor) Striped skunk (Mephitis mephitis) Vole (Myodes, Microtus spp.) Weasel (Mustela spp.) Western spotted skunk (Spilogale gracilis) Wyoming ground squirrel (Urocitellus elegans) 626 Yes None 243 479 No No N/A N/A Increasing coarse woody debris favors primary prey (small mammals) and provides marten habitat; understory development favors secondary prey (snowshoe hares) New understory vegetation and berry crops provide abundant food New vegetation provides abundant food resources Plasticity in behavior, diet, and habitat use allows exploitation of variety of conditions Abundance of grasses, forbs, shrubs, and saplings provides abundant food and cover New vegetation provides abundant food resources. Open canopy is preferred habitat Abundance of grasses, forbs, shrubs, and saplings provides abundant food and cover Abundance of grasses, forbs, shrubs, and saplings provides abundant food and cover Diminished supply of live bark and foliage for foraging Plasticity in behavior, diet, and habitat use allows exploitation of a variety of conditions Diminished cone crop causes food scarcity. Several previous works found negative response Increasing understory density provides abundant food and cover Primary habitat is alpine tundra, talus slopes, and rock outcroppings; forest is secondary N/A N/A 26 301 No No N/A N/A N/A N/A 48 336 446 20 132 132 27 22 No No No No No No No No N/A N/A N/A N/A N/A N/A N/A N/A N/A N/A N/A N/A N/A N/A N/A N/A 8 No N/A N/A Species American marten (Martes americana) Black bear (Ursus americanus) elevation band, or at higher elevations that were drier and/or south-facing, lodgepole pine was the dominant forest cover. At the higher end of this elevation band, and at cool, north-facing slopes at lower elevation, forests were generally ❖ www.esajournals.org Rationale for predicted response dominated by spruce-fir. Note, however, that lodgepole pine does not occur in the southern third of the state (range limit), and thus, sprucefir forests were dominant at all elevations and aspects in this portion of the study area. Large 4 August 2018 ❖ Volume 9(8) ❖ Article e02369 �IVAN ET AL. Fig. 1. Randomly selected sampling sites (gray circles) where passive infrared game cameras were deployed in spruce-fir (green) and lodgepole pine (yellow) forests in Colorado, USA, 2013–2014. Brown and orange are the approximate extents of spruce beetle and mountain pine beetle impacts in spruce-fir and lodgepole pine forests, respectively, as of 2014. such, juxtaposition with human development was minimal outside of scattered seasonal residences, small developments, and occasionally ski runs. However, many subalpine areas were heavily recreated by both motorized and nonmotorized users, especially during summer months. Also, many study sites fell within cattle or sheep allotments on public lands. The climate was typical of continental mid-latitude regions at high elevation (Benedict 2008:149–150). Mean July temperature on the study area was 14.2°C; mean January temperature was 6.1°C. More than half of the annual precipitation (37.7 cm) fell as snow between late October and April, and snow cover often persisted through early June, especially on northfacing slopes. Mean March snow depth across all snotel sites in the study area was 1.3 m. Most of aspen (Populus tremuloides) stands were interspersed throughout these subalpine forests, depending on disturbance history, soil characteristics, and elevation. Douglas-fir (Pseudotsuga menziesii), bristlecone pine (Pinus aristata), limber pine (Pinus flexilis), and blue spruce (Picea pungens) also occurred sporadically where conditions were appropriate. The vegetation band of interest was bordered by alpine tundra above and montane forest consisting largely of Douglas-fir and ponderosa pine (Pinus ponderosa) below. Even at these subalpine elevations, forested stands were often dissected by high elevation meadows and open valleys. We restricted our sampling to public lands managed by the United States Forest Service, National Park Service, Bureau of Land Management, and Colorado State Forest Service. As ❖ www.esajournals.org 5 August 2018 ❖ Volume 9(8) ❖ Article e02369 �IVAN ET AL. manner. Severity of outbreaks was largely a function of size and species composition such that stands dominated by large lodgepole or large Engelmann spruce trees often experienced high severity when outbreaks occurred, whereas stands with a significant subalpine fir component (or other species) experienced lower severity. We had no way of assessing severity a priori. Our sample ultimately spanned the entire gradient of severity (0–99% of overstory trees impacted) but was skewed toward the lower end of this gradient (mean overstory mortality = 20%). To assess occupancy of the mammalian community, we deployed a passive infrared camera trap (Reconyx PC800, Holmen, Wisconsin, USA) near the center of each selected 1 km 9 1 km cell. Cameras were strapped to a large tree approximately 0.5 m above the ground and pointed toward a lure tree 4–5 m away. We soaked one 1inch2 piece of wool in 15 mL of peanut butter and another in 1 mL of rabbit lure (Pro-Pest Professional Lures, Yazoo City, Missouri, USA), then affixed each to the lure tree with natural twine at 0.5 and 0.25 m above ground, respectively. These small amounts of lure were used to enhance our probability of detecting individuals living locally while minimizing the probability of attracting individuals residing far away. We based the camera configuration and lure selection on pilot work and that of Blecha (2015). Our aim was to maximize our ability to photograph a diversity of species ranging from small-bodied rodents to largebodied ungulates. We recorded camera trap details as suggested by Meek et al. (2014; Appendix S1: Table S1). Cameras were deployed in half of the cells in each stratum between 28 May and 24 July 2013 and in the other half of cells in each stratum between 30 May and 21 July 2014. Cameras remained deployed for >4 weeks and were not re-visited until they were retrieved. All photographs were examined by at least two observers (a third observer served as a referee in cases where the first two observers disagreed) using the CPW Photo Warehouse software (CPW; Ivan and Newkirk 2016). After assigning a species to each photograph, we used the software to create encounter histories by binning photographs into discrete occasions beginning the day after the camera was deployed at either noon (nocturnal species) or midnight (diurnal species). We the annual precipitation that did not fall as snow during winter came as regular afternoon monsoons during mid–late summer (NOAA 2017). Sampling design Because this project occurred in tandem with sampling to determine avian response to bark beetle outbreaks, we deployed a 1 km 9 1 km grid across the study area to serve as our preliminary sampling frame, as per the Bird Conservancy of the Rockies’ protocol for Integrated Monitoring of Bird Conservation Regions (Hanni et al. 2012). We then overlaid a digital elevation model, ownership data, and vegetation data from the Colorado Vegetation Classification Project (CVCP; CPW 2017). From these overlays, we extracted grid cells that occurred on public land, were above 2590 m across ≥75% of the cell, and were comprised of ≥75% spruce-fir or lodgepole pine forest cover. We defined this collection of cells (n = 15,113) as subalpine forest in Colorado, and it served as the final sampling frame, or population of cells, to which we sought to make inference (Fig. 1). From this population, we selected a spatially balanced random sample (Stevens and Olsen 2004, Theobald et al. 2007) of n = 150 spruce-fir cells and n = 150 lodgepole pine cells. Sampling effort was based on a priori power analyses performed on pilot data, which indicated that sampling 125–175 cells would result in an 80% chance of detecting a linear trend in occupancy of 0.03 per year over 15 yr (e.g., if occupancy were to increase steadily from 0.1 to 0.5 or decrease from 0.5 to 0.1). We constructed histograms of our selection for each stratum, binned by “Years since initial bark beetle outbreak” (YSO), which we estimated from aerial survey data (USFS Rocky Mountain Region 2018). Our initial sample provided a balanced cross section of sites across the YSO gradient for the lodgepole stratum, but overloaded unimpacted sites for the spruce-fir stratum as the spruce beetle epidemic had not yet reached the aerial extent of the mountain pine beetle epidemic. Therefore, to obtain a more balanced representation along the YSO axis, we re-selected grid cells from the spruce-fir stratum as before, but dropped the inclusion probability for unimpacted sites such that the resulting histogram was more evenly distributed along YSO, but still selected randomly and in a spatially balanced ❖ www.esajournals.org 6 August 2018 ❖ Volume 9(8) ❖ Article e02369 �IVAN ET AL. assigned a “1” to the encounter history string if the species of interest was detected during an occasion or “0” if it was not detected. There are numerous options for binning continuous data into discrete occasions. Generally, if occasions are too short, estimates of detection probability approach zero, which creates estimation issues, especially for uncommon species. If the overall length of a survey is too long, occupancy estimates approach 1.0, especially for common species, which is an unhelpful outcome when the inferential goal is to parse differential responses of several species to a perturbation. We conducted preliminary analyses to test several binning strategies on both rare and common species in our data set. We settled on seven 4-d occasions as this configuration optimally maintained mean occupancy estimates (i.e., intercept only models) away from boundaries for most species, thus preserving the opportunity to observe changes in both directions (negative or positive) along the YSO or severity axes. This configuration simultaneously maintained estimates of detection probability away from boundaries as well (Appendix S1: Table S2), even for uncommon species. It also aligns with recommendations of (Hamel et al. 2013) who suggested 20–30 problem-free days of camera operation to produce stable and precise estimates of occupancy and detection. cover (all material <0.25 m tall) by deadwood >15 cm in diameter, and (8) percent bare ground of all material <0.25 m in height. Using GIS, we extracted raster (30 m resolution) or polygon data at each of the 16 points in a cell to quantify (1) YSO, based on data collected from United States Forest Service Forest Aerial Detection Surveys (USFS Rocky Mountain Region 2018), (2) topographic wetness index plus (an index of soil moisture based on slope, basin characteristics, and solar radiation; Theobald 2007), (3) topographic position index (an index of concavity or convexity to indicate position along a slope from valley to ridge top; Weiss 2001), (4) heat loading (a measure of the total lumens accumulated at a given pixel over a year, taking into account slope, aspect, and shadows from adjacent landforms; Theobald et al. 2015), (5) elevation, and (6) a binary indicator of whether the cell was in federally designated wilderness or not. For both groups of variables, those collected in the field and those extracted via GIS, we computed the mean across all 16 sample points and assumed this mean adequately represented average forest conditions within the 1-km2 cell. We used these means as covariates in an occupancy analysis to determine whether any were related to use of the portion of the forest sampled by the camera trap. Note that home range size matched or exceeded the 1-km2 cell size for ungulate and carnivore species in our data set (Armstrong et al. 2011). Thus, use of the area in which the camera was placed might be expected to be influenced by forest conditions at the 1-km2 (16-pt) scale. However, rodents and lagomorphs had home ranges <0.2 km2 (Armstrong et al. 2011), meaning that many of the 16 points would be outside of the home range of individuals captured by the camera trap. To better scale covariates to space use of these smaller-bodied species, we computed means across the all points within 300 m (generally the 4–5 closest points) of the camera location. Such a radius would conservatively capture a 0.2 km2 circle centered on the camera. Covariates In addition to deployment of cameras to detect mammals, avian point counts were conducted at 16 equally spaced (250 m) point locations within each 1 km 9 1 km cell. Prior to conducting point counts, observers visually estimated canopy cover by species within 50 m of each point. These field data allowed us to groundtruth strata membership of each cell (i.e., post-stratification), resulting in a spruce-fir:lodgepole pine ratio of 186:114 rather than the 150:150 originally selected based on CVCP data. In addition to obtaining (1) updated stratum data, crews also visually estimated within 50 m of each point: (2) percent of canopy trees that were dead (i.e., severity of outbreak), (3) percent canopy cover of aspen, (4) mean shrub (woody species 0.25–3 m tall, including saplings) height, (5) percent shrub cover by deciduous species, (6) percent shrub cover by coniferous species, (7) percent ground ❖ www.esajournals.org Analysis We fit single-season occupancy models to the encounter data to determine how mammalian species altered their use of subalpine forest stands relative to vegetation changes stemming from bark 7 August 2018 ❖ Volume 9(8) ❖ Article e02369 �IVAN ET AL. beetle activity. As such, the two parameters that required estimation were Ψ, the probability that a sample unit was occupied, and p, the probability that a species was detected in a unit, given that the unit was occupied. In our case, the units were technically the detection zones of each camera (thus we know each cell containing a camera with a detection was “used” to some degree, but we make no assertion that our camera sampled the entire 1-km2 cell for each species), and because animals were able to cross into and out of the detection zone of cameras between occasions, our sampling scheme violated the closure assumption of this class of models. Therefore, we interpreted Ψ as the probability that the detection zone of the camera trap was used by the species of interest during the course of the survey period (~1 month during summer), as per MacKenzie et al. (2006: page 105). We treated YSO as a group variable, and thus binned encounter data into 12 groups; stands that were still green when sampling occurred were considered “0 YSO,” those impacted by beetles the year prior to our sampling were binned as “1 YSO,” etc. To efficiently identify the structures and covariate combinations that produced the best fit for each species, we employed a sequential process in the vein of Lebreton et al. (1992). First, we held the Ψ parameter constant at a general structure (i.e., additive model including all covariates) and fit three structures for the detection parameter (p): (1) detection constant across occasions, (2) detection as a linear trend across occasions, and (3) detection as a quadratic trend. The latter models allowed detection probability to decline as lure was washed away by weather or was removed by animals through time. We selected the best-fitting structure for detection using Akaike’s Information Criterion adjusted for small sample size (AICc; Burnham and Anderson 2002). Second, after selecting the best-fitting model for p, we then fixed that structure and modeled Ψ by fitting all combinations of four variables from among the 14 listed above (omitting YSO and mortality of canopy trees; see below). This step served to identify a relatively simple model to account for background variability in how species organize themselves across the landscape based on vegetation and topographic variables. Finally, we fit models to assess our true question of interest: whether covariates related to the ❖ www.esajournals.org bark beetle epidemic could explain further variation in Ψ over basic models representing Ψ as a function of topographic and vegetation variables. We fixed the best-fitting structures for p and Ψ from steps 1–2, and then added additional structure to Ψ reflecting potential responses to beetleinduced changes in vegetation. Specifically, we hypothesized that species may respond to beetle outbreaks through time (i.e., YSO) by exhibiting (1) no change in their use of forest stands after impact by beetles, (2) a linear increase or decrease in use of stands in the years following impact by beetles, (3) a quadratic increase or decrease in use of stands after impact by beetles, or (4) a response that followed a third-order polynomial form (Fig. 2). We also hypothesized that some species would not begin responding until overstory needles dropped and understory vegetation was released. Therefore, we fit models representing the same general shape as 2–4, but with a delay in the onset of the response until Fig. 2. Example curves fit to occupancy data depicting immediate (upper panel) and delayed (lower panel) response of mammals to bark beetle outbreaks. Actual slopes and inflections were dictated by the specific data for a given species. 8 August 2018 ❖ Volume 9(8) ❖ Article e02369 �IVAN ET AL. 3 yr post-beetle impact, which in both forest types was approximately when needles began to drop, canopies opened, and understory changes began (Fig. 2). We assumed that delays in species responses that started >3 YSO would be adequately captured by either quadratic or polynomial models. Additionally, we considered severity of the outbreak as an additive effect that could shift any response up or down, or as an interaction such that the shape of the response itself could change at varying severities. Step 3 of our modeling procedure resulted in 19 models (six plausible response curves, additive or interactive severity combined with each, plus an intercept only model). We report coefficient estimates for YSO and severity when these occurred as additive effects in best-fitting linear or quadratic models from step 3 (we did not attempt to interpret coefficient estimates from cubic polynomials). However, in the figures we show predicted responses for low (10% tree mortality), medium (50% tree mortality), and high (90% tree mortality) severity averaged across all models in the set to fully capture both uncertainty within modeled predictions and that arising from the model-selection process. Our choice of levels for predictions was arbitrary, but covers the full range of severity we observed in the field. We depicted estimates separately for spruce-fir and lodgepole pine stands if the top model from step 2 included a strata effect; otherwise, results were pooled across forest types. passing over those units should not have impacted sample integrity. Thus, we feel that the realized sample represented the population of interest as only 1% of the initial selection was omitted due to logistical concerns. We collected 336,665 photographs of 26 terrestrial mammal species from our sampling effort (Appendix S1: Table S1, Table 1). Of these 26 species, we obtained sufficient data on 13 to complete the analysis specified above (Table 1). Note that for mule deer, coyote (Canis latrans), and golden-mantled ground squirrel (Callospermophilus lateralis), 1–4 models of the 19 we fit included an interaction between YSO and severity that was not well estimated (i.e., point estimates were unreasonable and/or SEs were several orders of magnitude larger than all other SEs). In each case, we removed these models from the set and made inference from the remaining models. As per our predictions, elk, mule deer, and moose (Alces alces) exhibited a positive association with bark beetle activity, although the nature of these associations varied by species (Fig. 3). Elk use was positively associated with the severity of the outbreak in a given location (bSeverity = 3.37, 95% CI = [1.38, 5.35]), but there was relatively little association with YSO (bYSO = 0.05, 95% CI = [ 0.15, 0.05]). Conversely, mule deer were more strongly associated with YSO (use followed a cubic trend that began increasing at about 7 YSO), but this relationship did not change appreciably with increasing severity (Fig. 3). Use of subalpine forests by moose was best modeled by an interaction between YSO and severity. At low severity, mean use was low and nearly flat across the range of YSO (Fig. 3). However, as severity increased, mean use increased in a quadratic fashion and peaked 3–7 YSO (bSeverity = 5.29, 95% CI = [1.34, 9.34]; bYSO = 0.66, 95% CI = [ 0.10, 1.42]), though there was considerable uncertainty around this relationship, the majority of which was model-selection uncertainty. Red squirrel, golden-mantled ground squirrel, chipmunk, and coyote were among the species that exhibited a negative association with beetle activity, but again, responses varied (Fig. 4). The top model for red squirrels was invariant across YSO at low severity, but as severity increased, use declined sharply after 3 YSO (bSeverity = 1.69, 95% CI = [ 3.17, 0.23]), approximately when trees no longer had foliage and cone crops RESULTS Crews successfully deployed cameras at 273 of the 300 cells originally selected for sampling. Of the 27 cells that were not sampled, 15 occurred in recently burned areas, nine were located predominantly on private inholdings, and three were located on sites that were too dangerous to survey (e.g., too steep, cliff bands). In all cases, replacement cells from the spatially balanced list were substituted for those omitted in order to maintain a sample of 300, which is a strength of the spatially balanced approach, provided that substitution does not happen so often as to impact representativeness of the sample (Stevens and Olsen 2004). Given that we did not intend to make inference to burned areas or private lands, ❖ www.esajournals.org 9 August 2018 ❖ Volume 9(8) ❖ Article e02369 �IVAN ET AL. presumably declined. Red squirrel use was estimated to be about 0.1 (~15%) higher in spruce-fir stands compared to lodgepole stands. As with moose, much of the uncertainty was modelselection uncertainty. Similarly, the top model for golden-mantled ground squirrels included a cubic relationship with YSO and additive impacts of severity (bSeverity = 2.64, 95% CI = [ 5.89, 0.60]), although neither effect was particularly evident in the model-averaged output (Fig. 4). Conversely, the top models and model-averaged output for both chipmunk (bYSO = 0.08, 95% CI = [ 0.16, 0.00]) and coyote (bYSO = 0.14, 95% CI = [ 0.29, 0.01]) indicated that use was negatively associated with YSO, but severity had little impact on occupancy. Use of subalpine forests by American marten, black bears (Ursus americanus), snowshoe hares, and porcupines (Erethizon dorsatum) was not substantially influenced by beetle activity (Fig. 5). The top model for each species did not include effects of either YSO or severity. They only included the habitat covariates from the previous modeling step, which resulted in largely “flat” mean responses. However, for snowshoe hares, the second best model had a nearly identical AICc score to the top model and included a quadratic relationship with YSO (bYSO = 0.29, 95% CI = [0.00, 0.58]) such Fig. 3. Mammalian species that exhibited a positive association between use of forested stands and beetle activity (either years since the outbreak occurred, severity, or both). Curves and 95% confidence intervals represent model-averaged responses across all fitted curves described in Fig. 2. From left to right, panels indicate predicted model-averaged responses for cases where 10%, 50%, and 90% of the overstory in a stand is killed by beetle activity. From top to bottom, panels show predicted responses for elk, mule deer, and moose. Probability of use was estimated to vary little between the spruce-fir and lodgepole pine stands, so responses were pooled across habitat types. ❖ www.esajournals.org 10 August 2018 ❖ Volume 9(8) ❖ Article e02369 �IVAN ET AL. 95% CI = [ 2.02, 10.28]), which is why mean snowshoe hare association with YSO was predicted to flatten out at high severity. Snowshoe hare use was estimated to be 0.3 (110%) higher in spruce-fir stands than in lodgepole pine stands. that hare use reached a shallow peak approximately 5 yr after beetles first impacted the stand. Some snowshoe hare models that included an interaction with severity also held weight and indicated a weak positive relationship (bSeverity = 4.13, Fig. 4. Mammalian species that exhibited a negative association between use of forested stands and beetle activity (either years since the outbreak occurred, severity, or both). Curves and 95% confidence intervals represent model-averaged responses across all fitted curves described in Fig. 2. From left to right, panels indicate predicted responses for cases where 10%, 50%, and 90% of the overstory in a stand is killed by beetle activity. From top to bottom, panels show predicted responses for red squirrel, golden-mantled ground squirrel, chipmunk spp., and coyote. Red squirrel use was estimated to vary between the spruce-fir (blue) and lodgepole pine (gray) stands. For other species, habitat stratum was less important and responses were pooled across habitat types. ❖ www.esajournals.org 11 August 2018 ❖ Volume 9(8) ❖ Article e02369 �IVAN ET AL. Fig. 5. Mammalian species that exhibited little association between use of forested stands and beetle activity (either years since the outbreak occurred, severity, or both). Curves and 95% confidence intervals represent modelaveraged responses across all fitted curves described in Fig. 2. From left to right, panels indicate predicted responses for cases where 10%, 50%, and 90% of the overstory in a stand is killed by beetle activity. From top to bottom, panels show predicted responses for American marten, black bear, snowshoe hare, and porcupine. For snowshoe hares and porcupine, use was estimated to vary between the spruce-fir (blue) and lodgepole pine (gray) stands. For other species, habitat stratum was less important and responses were pooled across habitat types. Overall porcupine use of subalpine stands in Colorado was the lowest of any species analyzed and was 0.06 (4.5%) higher in spruce-fir than in lodgepole pine stands. Only two species showed a mixed response to bark beetles along the YSO and severity gradients ❖ www.esajournals.org (Fig. 6). Red fox (Vulpes vulpes) use was positively associated with YSO (bYSO = 0.08, 95% CI = [ 0.05, 0.21]) but negatively associated with severity (bSeverity = 6.24, 95% CI = [ 12.7, 0.23]). In contrast, yellow-bellied marmots (Marmota flaviventris) were negatively associated with YSO 12 August 2018 ❖ Volume 9(8) ❖ Article e02369 �IVAN ET AL. Fig. 6. Mammalian species that exhibited mixed associations between use of forested stands and beetle activity (either positive association with years since the outbreak but negative association with severity, or vice versa). Curves and 95% confidence intervals represent model-averaged responses across all fitted curves described in Fig. 2. From left to right, panels indicate predicted responses for cases where 10%, 50%, and 90% of the overstory in a stand is killed by beetle activity. From top to bottom, panels show predicted responses for red fox and yellow-bellied marmot. Use was estimated to vary little between the spruce-fir and lodgepole pine stands, so responses were pooled across habitat types for these species. (bYSO = 0.18, 95% CI = [ 0.43, 0.07]) but positively associated with severity (bSeverity = 3.65, 95% CI = [0.52, 6.77]). In both cases, however, small changes in mean response were swamped by considerable uncertainty across models. overlapping the cameras. Thus, it is impossible to determine whether changes in use presented here reflected changes in density, home range size, habitat use, movements, or some combination thereof. Despite this, we suggest that our coarse approach is a valuable and informative initial assessment of mammalian response to this large-scale perturbation. Indeed, quantifying intensity of use has long been a staple of ecological investigation (e.g., Manly et al. 2002). Furthermore, sampling at a large scale across a broad array of wide-ranging species leaves one with few choices other than to engage a coarse metric such as “use.” Also, in many cases coarse inference can be honed by invoking species’ ecology. For instance, apparent declines in red squirrel use following a bark beetle outbreak could be due to a reduction in home range size, movement out of impacted stands to unimpacted stands, or reductions in density. However, because this species tends to rely heavily on cone crops as a dietary staple (Armstrong et al. 2011), loss of this food source is unlikely DISCUSSION Our sampling entailed the deployment of point-based detectors (cameras sample a relatively small area) in a landscape of continuous space use by the species of interest. That is, our design violated the assumption of closure during the survey period because individuals likely moved into and out of the area sampled by cameras, and therefore sample units toggled between “occupied” and “unoccupied” during each 28-d survey period. This reality necessitated an altered interpretation of the estimates of Ψ as described earlier. Efford and Dawson (2012) noted that under these sampling conditions, Ψ reflects the product of home range area and density, or the “combined ‘footprint’ of the population formed from the union of the home ranges” ❖ www.esajournals.org 13 August 2018 ❖ Volume 9(8) ❖ Article e02369 �IVAN ET AL. Use of beetle-impacted subalpine stands by moose was negligible in low-severity conditions, but in mid- to high-severity conditions, mean use increased after a bark beetle outbreak, followed by a sharp decline 7–8 yr later. We observed significant variation about this mean response, which was due mostly to model-selection uncertainty (i.e., the top models included a quadratic effect that was significantly different from zero, but competing models without this effect also had weight). In Colorado, subalpine forest stands provide secondary habitat for moose, with primary habitat being forest edge and adjacent wet meadows and willow (Salix spp.) carrs (Armstrong et al. 2011). Thus, the observed pattern could reflect moose taking advantage of the adjacent flush of new growth when convenient, but electing their preferred habitat once navigation in forest habitats becomes difficult due to dense vegetation and/or down wood, or to decline in nutritional value of regenerating understory. Use of subalpine stands was negatively associated with shrub height based on the top model from step 2 (Appendix S1: Table S3), possibly lending some support to either mechanism. Regardless of mechanisms driving ungulate responses, it appears that ecologists and managers can expect an increase in ungulate use of beetle-killed stands, especially those that were severely impacted. Such changes, whether due to an increase in abundance and/or changes in habitat use, could influence a variety of processes such as regeneration rates of forests and understory, species composition of regenerating forests and understory, space use and density of predators, and dispersion of the hunting public on the landscape during early seasons (e.g., August, early September). As noted earlier, red squirrels exhibited the strongest negative response to bark beetle outbreaks of any of the species sampled, especially in high-severity locations. We suggest the observed decline in use in highly impacted areas reflects a decline in squirrel density related to diminished cone crops, which are a staple in the red squirrel diet (Armstrong et al. 2011). Goldenmantled ground squirrels and chipmunk spp. were also negatively impacted, but not to the same degree. Like red squirrels, their diets include seeds, but also a variety of other items including forbs, buds, and insects (Armstrong to lead to a reduction in home range size (if anything, the opposite would occur so that the species could find adequate food resources). Furthermore, the statewide scale of the current outbreaks leaves few opportunities for red squirrels to escape to un-impacted stands. Thus, the most likely explanation is that lower use reflects a reduction in density. Similar understandings of species biology could be used to augment occupancy results, or at least to posit mechanisms for future research. We note that the estimates presented here pertain only to use of these forested areas over a 28-d period during summer. Patterns of use may have differed during other seasons. Generally, ungulate species exhibited a positive association with bark beetle outbreaks, as we hypothesized, although the shape and nature of their responses were variable. The positive association is likely due to the growth of a diverse and abundant assemblage of grasses, forbs, shrubs, and tree saplings that follows the beetleinduced removal of overstory canopy. Such conditions should be a dietary boon to ungulate species (Wallmo 1981, Toweill et al. 2002, Franzmann and Schwartz 2007). In addition to ample forage, all three species rely on well-distributed cover for escape and loafing behavior during midday. The thick understory regeneration present in the most severely impacted forests could provide this needed cover near an abundant food source, increasing use of these stands (Wallmo 1981, Toweill et al. 2002, Franzmann and Schwartz 2007). These results are consistent with Stone (1995) who noted a positive linear relationship between fecal pellet groups of ungulates and outbreak severity following a mountain pine beetle epidemic in Utah. Notably, we found that mule deer responded to YSO, but not severity, which was a strong, positive predictor of elk and moose responses. Severity was included in each of the four models that were discarded from the mule deer analysis because parameters were not well estimated. In each model, evidence suggested that the direction of the severity effect was positive, same as for elk and moose. Thus, it is likely that mule deer use of beetleimpacted stands was positively associated with severity, but our particular data set prevented this association from surfacing in a well-estimated manner. ❖ www.esajournals.org 14 August 2018 ❖ Volume 9(8) ❖ Article e02369 �IVAN ET AL. martens to be positively associated with beetleimpacted stands based on expected increases in small mammal prey (i.e., red-backed voles; Stone 1995, Saab et al. 2014) and coarse woody debris (Armstrong et al. 2011), but did not observe this. Perhaps mixed (snowshoe hare) to negative responses (red squirrels) of other prey items mitigated use of impacted stands or more complex relationships with competitors or altered habitat features muted their response. We expected a negative response from porcupines, but instead observed a flat response, and maybe more importantly, almost no porcupine observations in the study area despite a relatively high probability of detection (Appendix S1: Table S2). The small number of detections in subalpine forests in Colorado matches recent observations in Montana (Mally 2008), Arizona (Brown and Babb 2009), and potentially California (Allen and Casady 2012, Appel et al. 2017). Given the plasticity in behavior, diet, and habitat exploitation displayed by coyotes and red foxes (Armstrong et al. 2011), we expected no response to bark beetle outbreaks. We also did not expect to detect yellow-bellied marmots at our camera stations, let alone observe a response to beetles, given that they typically reside in tundra and talus habitats (Armstrong et al. 2011). However, coyotes and red foxes did show at least some association with either YSO, severity, or both. Coyotes were negatively associated with shrub height (Appendix S1: Table S3), so perhaps changes in vegetative structure drove their declining use of beetle-impacted forests. We cannot posit a mechanism for the mixed response exhibited by red fox. We do, however, suggest that the peaked use of severely impacted forest stands by marmots may reflect foraging forays from preferred talus slopes and rock outcroppings into adjacent forest to exploit new vegetative growth, similar to the pattern and mechanism we suggested for moose. Our work is among the first to quantify the response of a suite of mammalian species to the extensive bark beetle outbreaks that have impacted much of western North America. We demonstrated that the magnitude and direction of responses, as well as the role of outbreak severity and elapsed time since inception of the outbreak, vary widely among species. This result was expected given the diversity of taxa examined, et al. 2011) which may buffer their response compared to red squirrels. Also, top models for both golden-mantled ground squirrels and red squirrels included a positive relationship with bare ground, suggesting that part of the negative response to beetle impacts may simply be related to an aversion to dense vegetation at ground level. Johnson et al. (2015) summarized the foundational roles that red squirrels fulfill in the subalpine forest community including impacts to small mammal and carnivore communities via their midden-building activities (Ruggiero et al. 1998, Pearson and Ruggiero 2001), avian communities via nest predation (Siepielski 2006), the vegetation community via their influence on cone serotiny (Benkman and Siepielski 2004), and predator communities via their role as prey for several species including northern goshawks (Accipiter gentilis), American marten, and others (Buskirk and MacDonald 1984, Squires 2000). Thus, declining red squirrel use of beetleimpacted subalpine forests could result in cascading changes throughout the system, at least until cone crops recover. Observed responses of several other species did not track our predictions. We expected black bears to respond positively to the flush of new grasses, forbs, shrubs, and berries available in beetle-impacted forests, but their response did not vary with YSO or severity. Perhaps their high mobility and propensity to exploit a variety of habitats precluded a strong response to changes in subalpine forest. We also expected snowshoe hares to respond strongly to understory release, but their response was muted. Yeager and Riordan (1953) and Stone (1995) likewise predicted a positive response for snowshoe hares, yet observed an equivocal one. We note that nearly all of the top models from step 2 of our analysis of snowshoe hares included a positive effect of “bare ground.” Perhaps increasing foraging opportunities for hares associated with increased density of conifer saplings in beetle-impacted stands were offset by a lack of bare ground in these stands, which may have impeded locomotion. Alternatively, positive responses to increased conifer density in some stands could have been offset by negative response to tall, dense grass and forb regeneration in others (understory regeneration can be highly variable from site to site depending on conditions). We expected ❖ www.esajournals.org 15 August 2018 ❖ Volume 9(8) ❖ Article e02369 �IVAN ET AL. although many of our species-specific findings did not track our a priori predictions. We discussed possible mechanisms for these observed responses, further evaluation of these hypotheses is required. Our inferences were based on a relatively coarse metric, and we suggest that future work include intensive efforts to determine impacts to density, survival, and recruitment of key species, impacts to species richness and community assemblages, and direct tests of intermediate disturbance hypotheses. Benkman, C. W., and A. M. Siepielski. 2004. A keystone selective agent? Pine squirrels and the frequency of serotiny in lodgepole pine. Ecology 85:2082–2087. Bentz, B. J., J. A. Logan, and G. D. Amman. 1991. Temperature-dependent development of the mountain pinebeetle (Coleoptera, Scolytidae) and simulation of its phenology. Canadian Entomologist 123:1083–1094. Bentz, B. J., J. R�egni�ere, C. J. Fettig, E. M. Hansen, J. L. �n, and Hayes, J. A. Hicke, R. G. Kelsey, J. F. Negro S. J. Seybold. 2010. Climate change and bark beetles of the western United States and Canada: direct and indirect effects. BioScience 60:602–613. Bentz, B. J., et al. 2009. Bark beetle outbreaks in western North America: causes and consequences. USDA Forest Service, Salt Lake City, Utah, USA. Blecha, K. A. 2015. Risk-reward tradeoffs in the foraging strategy of cougar (Puma concolor): prey distribution, anthropogenic development, and patch selection. Thesis. Colorado State University, Fort Collins, Colorado, USA. Brown, D. E., and R. D. Babb. 2009. Status of the Porcupine (Erithizon dorsatum) in Arizona, 2000–2007. Journal of the Arizona-Nevada Academy of Science 41:36–41. Burnham, K. P., and D. R. Anderson. 2002. Model selection and multimodel inference: a practical information-theoretic approach. Second edition. Springer, New York, New York, USA. Buskirk, S. W., and S. O. MacDonald. 1984. Seasonal food habits of marten in south-central Alaska. Canadian Journal of Zoology 62:944–950. Buskirk, S. W., and L. F. Ruggiero. 1994. American marten. Pages 7–37 in L. F. Ruggiero, K. B. Aubry, S. W. Buskirk, L. J. Lyon, and W. J. Zielinski, editors. The scientific basis for conserving forest carnivores: American marten, fisher, lynx, and wolverine in the western United States. GTR RM-254. USDA Forest Service, Rocky Mountain Research Station, Fort Collins, Colorado, USA. CPW [Colorado Parks and Wildlife]. 2017. Colorado Vegetation Classification Project. https://www.arc gis.com/home/item.html?id=893739745fcd4e05af816 8b7448cda0c Drever, M. C., and K. Martin. 2007. Spending time in the forest: responses of cavity-nesters to temporal changes in forest health in interior British Columbia. Pages 236–251 in J. A. Bissonette and I. Storch, editors. Temporal dimensions of landscape ecology: wildlife responses to variable resources. Springer, New York, New York, USA. Efford, M. G., and D. K. Dawson. 2012. Occupancy in continuous habitat. Ecosphere 3:1–15. Fettig, C. J., K. D. Klepzig, R. F. Billings, A. S. Munson, �n, and J. T. Nowak. 2007. T. E. Nebeker, J. F. Negro ACKNOWLEDGMENTS We thank Kevin Aagaard, Dan Tripp, and two anonymous reviewers for helpful comments on early drafts. Tim Hanks, Britta Schielke, Nick Meyer, Mandi Leigh, Serena Rocksund, Kathryn Bernier, Emily Latta, Blake Bartz, Erin Nigon, Jackie Johnson, Tyler Stratman, Kyle Bond, Joe Seufert, Brent Pease, Erin Sawa, Jake Schas, Bob Taylor, Carla Hanson, and Jonathan Lewis completed the hard work of sampling hundreds of random locations throughout Colorado, as well as pouring through hundreds of thousands of photographs that resulted from the field effort. Funding was provided by the Species Conservation Trust Fund via Colorado Parks and Wildlife. We thank numerous field personnel in local Colorado Parks and Wildlife and U.S. Forest Service Offices for logistical support. LITERATURE CITED Alexander, R. R., R. C. Shearer, and W. D. Shepperd. 1990. Abies lasiocarpa. Pages 60–70 in R. M. Burns and B. H. Honkala, editors. Silvics of North America. USDA Forest Service, Washington, D.C., USA. Alexander, R. R., and W. D. Shepperd. 1990. Picea engelmannii. Pages 187–203 in R. M. Burns and B. H. Honkala, editors. Silvics of North America. USDA Forest Service, Washington, D.C., USA. Allen, M. L., and D. Casady. 2012. Recent observations of porcupines in El Dorado County, California. California Fish and Game 98:175–177. Appel, C. L., W. J. Zielinski, F. V. Schlexer, R. L. Callas, and W. T. Bean. 2017. Distribution of the North American porcupine (Erethizon dorsatum) in Northern California. Western Wildlife 4:17–28. Armstrong, D. M., J. P. Fitzgerald, and A. M. Carron. 2011. Mammals of Colorado. Second edition. University Press of Colorado, Boulder, Colorado, USA. Benedict, A. D. 2008. The naturalist’s guide to the Southern Rockies. Fulcrum, Golden, Colorado, USA. ❖ www.esajournals.org 16 August 2018 ❖ Volume 9(8) ❖ Article e02369 �IVAN ET AL. Mally, K. A. 2008. Hierarchical summer habitat selection by the North American porcupine in western Montana. Thesis. University of Montana, Missoula, Montana, USA. 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. Second edition. Kluwer Academic, Norwell, Massachusetts, USA. Meek, P. D., et al. 2014. Recommended guiding principles for reporting on camera trapping research. Biodiversity and Conservation 23:2321–2343. Morrison, M. L., B. G. Marcot, and R. W. Mannan. 1992. Wildlife-habitat relationships: concepts and applications. University of Wisconsin Press, Madison, Wisconsin, USA. Mosher, B. A. 2011. Avian community response to a mountain pine beetle epidemic. Thesis. Montana State University, Bozeman, Montana, USA. NOAA [National Oceanic and Atmospheric Administration]. 2017. Climate data online: mapping tool. https://gis.ncdc.noaa.gov/maps/ncei/summaries/mo nthly Pearson, D. E., and L. F. Ruggiero. 2001. Test of the prey-base hypothesis to explain use of red squirrel midden sites by American martens. Canadian Journal of Zoology 79:1372–1379. Raffa, K. F., B. H. Aukema, B. J. Bentz, A. L. Carroll, J. A. Hicke, M. G. Turner, and W. H. Romme. 2008. Cross-scale drivers of natural disturbances prone to anthropogenic amplification: the dynamics of bark beetle eruptions. BioScience 58:501–517. Ruggiero, L. F., E. Pearson, and S. E. Henry. 1998. Characteristics of American marten den sites in Wyoming. Journal of Wildlife Management 62:663– 673. Saab, V. A., Q. S. Latif, M. M. Rowland, T. N. Johnson, A. D. Chalfoun, S. W. Buskirk, J. E. Heyward, and M. A. Dresser. 2014. Ecological consequences of mountain pine beetle outbreaks for wildlife in western North American forests. Forest Science 60:539–559. Sambaraju, K. R., A. L. Carroll, J. Zhu, K. Stahl, R. D. Moore, and B. H. Aukema. 2012. Climate change could alter the distribution of mountain pine beetle outbreaks in western Canada. Ecography 35:211– 223. Schebeck, M., E. M. Hansen, A. Schopf, G. J. Ragland, C. Stauffer, and B. J. Bentz. 2017. Diapause and overwintering of two spruce bark beetle species. Physiological Entomology 42:200–210. Siepielski, A. M. 2006. A possible role for red squirrels in structuring breeding bird communities in lodgepole pine forests. Condor 108:232–238. The effectiveness of vegetation management practices for prevention and control of bark beetle infestations in coniferous forests of the western and southern United States. Forest Ecology and Management 238:24–53. Franzmann, A. W., and C. C. Schwartz. 2007. Ecology and management of the North American moose. Second edition. University Press of Colorado, Boulder, Colorado, USA. Hamel, S., S. T. Killengreen, J.-A. Henden, N. E. Eide, L. Roed-Eriksen, R. A. Ims, N. G. Yoccoz, and R. B. O’Hara. 2013. Towards good practice guidance in using camera-traps in ecology: influence of sampling design on validity of ecological inferences. Methods in Ecology and Evolution 4:105–113. Hanni, D. J., C. M. White, R. A. Sparks, J. A. Blakesley, J. J. Birek, N. J. Van Lanen, J. A. Fogg, J. M. Berven, and M. A. McLaren. 2012. Integrated Monitoring of Bird Conservation Regions (IMBCR): field protocol for spatially balanced sampling of landbird populations. Bird Conservancy of the Rockies, Brighton, Colorado, USA. Hansen, E. M., and B. J. Bentz. 2003. Comparison of reproductive capacity among univoltine, semivoltine, and re-emerged parent spruce beetles (Coleoptera: Scolytidae). Canadian Entomologist 135:697–712. Ivan, J. S., and E. S. Newkirk. 2016. CPW Photo Warehouse: a custom database to facilitate archiving, identifying, summarizing, and managing photo data collected from camera traps. Methods in Ecology and Evolution 7:499–504. Johnson, T. N., S. W. Buskirk, G. D. Hayward, and M. G. Raphael. 2015. Timber harvest interacts with broad-scale forest mortality to affect site occupancy dynamics of a vertebrate seed predator. Forest Ecology and Management 340:95–102. Koprowski, J. L. 2005. Pine squirrel (Tamiasciurus hudsonicus): a technical conservation assessment. http://www.fs.fed.us/r2/projects/scp/assessments/pine squirrel.pdf Lebreton, J.-D., K. P. Burnham, J. Clobert, and D. R. Anderson. 1992. Modeling survival and testing biological hypotheses using marked animals: a unified approach with case studies. Ecological Monographs 62:67–118. Lotan, J. E., and W. B. Chritchfield. 1990. Pinus contorta. Pages 302–315 in R. M. Burns and B. H. Honkala, editors. Silvics of North America. USDA Forest Service, Washington, D.C., USA. MacKenzie, D. I., J. D. Nichols, J. A. Royle, K. H. Pollock, L. L. Bailey, and J. E. Hines. 2006. Occupancy estimation and modeling: inferring patterns and dynamics of species occurrence. Academic Press, Oxford, UK. ❖ www.esajournals.org 17 August 2018 ❖ Volume 9(8) ❖ Article e02369 �IVAN ET AL. Toweill, D. E., J. W. Thomas, and D. P. Metz. 2002. North American elk: ecology and management. Smithsonian Institution Press, Washington, D.C., USA. USFS Rocky Mountain Region [United States Forest Service Rocky Mountain Region]. 2018. Forest and grassland health aerial detection survey data. https:// www.fs.usda.gov/main/r2/forest-grasslandhealth Veblen, T. V. 2000. Disturbance patterns in southern Rocky Mountain forests. Pages 31–54 in R. L. Knight, F. W. Smith, S. W. Buskirk, W. H. Romme, and W. L. Baker, editors. Forest fragmentation in the southern Rocky Mountains. University Press of Colorado, Boulder, Colorado, USA. Wallmo, O. C. 1981. Mule and black-tailed deer of North America. University of Nebraska Press, Lincoln, Nebraska, USA. Weiss, A. D. 2001. Topographic position and landforms analysis. http://www.jennessent.com/down loads/tpi-poster-tnc_18x22.pdf Yeager, L. E., and L. E. Riordan. 1953. Effects of beetlekilled timber on range and wildlife in Colorado. Transactions of the North American Wildlife Conference 18:596–616. Squires, J. R. 2000. Food habits of Northern Goshawks nesting in south central Wyoming. Wilson Bulletin 112:536–539. Stevens, D. L., and A. R. Olsen. 2004. Spatially balanced sampling of natural resources. Journal of the American Statistical Association 99:262–278. Stone, W. E. 1995. The impact of mountain pine beetle epidemic on wildlife habitat and communities in post-epidemic stand of a lodgepole pine forest in northern Utah. Dissertation. Utah State University, Logan, Utah, USA. Theobald, D. M. 2007. LCaP v1.0: landscape connectivity and pattern tools for ArcGIS. Colorado State University, Fort Collins, Colorado, USA. Theobald, D. M., D. Harrison-Atlas, W. B. Monahan, and C. M. Albano. 2015. Ecologically-relevant maps of landforms and physiographic diversity for climate adaptation planning. PLoS ONE 10:e0143619. Theobald, D. M., D. L. J. Stevens, D. White, N. S. Urquhart, A. R. Olsen, and J. B. Norman. 2007. Using GIS to generate spatially balanced random survey designs for natural resource applications. Environmental Management 40:134–146. SUPPORTING INFORMATION Additional Supporting Information may be found online at: http://onlinelibrary.wiley.com/doi/10.1002/ecs2. 2369/full ❖ www.esajournals.org 18 August 2018 ❖ Volume 9(8) ❖ Article e02369 � https://cpw.cvlcollections.org/files/original/ebfcca2c38531da0c76072be4535e71c.pdf 2453fbae6a7d64b937e26ca84becd669 PDF Text Text Ecosphere Mammalian responses to changed forest conditions resulting from bark beetle outbreaks in the southern Rocky Mountains Jacob S. Ivan, Amy E. Seglund, Richard L. Truex, and Eric S. Newkirk Appendix S1 �Table S1. Standardized reporting of camera trap research details following Meek et al. 2014. Category Guiding Principle Habitat Ecosystem Type Subalpine Forest, southern Rocky Mountains, Colorado, USA Vegetation Associations Engelmann spruce (Picea engelmannii) - subalpine fir (Abies lasiocarpa), and lodgepole pine (Pinus contorta); Quaking aspen (Populus tremuloides) mixed with both primary forest types. Climate High elevation, continental mid-latitude (Benedict 2008, p. 149-150) Elevation 2559–3660 m Temperature Weather Station Data (N = 150 stations in study area): absolute min = -4.86 C, mean min = -1.00 C; absolute max = 30.9 C, mean max = 22.12 C. Rainfall 37.7 cm annual precipitation (more than half as snow from October– April) Species targeted Mammalian Community: Rodents, lagomorphs, carnivores, and ungulates. Small-bodied chipmunks (~50 g) to large-bodied moose (454 kg). Layout Spatially balanced random sample of N = 300 1-km2 cells from a population of 15,113 potential cells that contained majority subalpine forest on public land. Within each selected cell, camera trap was placed within 50-m of one of the 4 central avian sampling points (see Hanni et al. 2012). Final placement based on finding 2 trees 4-5 m apart with little vegetation between them. Number of camera trap locations N = 300 (single camera traps) Minimum Distance between camera locations Min = 1.38 km, Mean = 7.25 km, Max = 63.00 km Camera Type & Model Reconyx PC800 Camera Height/Distance to Lure Height = 0.5 m, Distance to lure = 4–5 m Camera Orientation Horizontal, lens pointed parallel to ground. Placement relative to animal passage Intentionally avoided game trails, human trails, and other obvious pathways. Habitat Modification Minimal. Allowed some branch removal or pulling clumps of grass. Ideally we avoided sites needing modification. Lure type, distance, refresh interval Peanut butter (15 ml) + commercial rabbit lure (1 ml) 4–5 m from camera, soaked in wool, tied to tree at 0.5 and 0.25 m above ground, respectively. Lure was not refreshed during the 28-day sampling period. Video, still, time lapse Still photos only (default settings) Flash Type Infrared flash (up to 21 m) Images per trigger 3 Survey Design Camera Settings Details �Analysis Results Capture Delay 1 second between each burst of 3 photos after a trigger; 1 second delay until next possible trigger. Trigger Speed 1/5 second Data Coding Format Each photo identified to species including "unknown" or "none". Number of individuals was recorded but not used in this analysis. Definition of Event For occupancy analysis, photo data were grouped into 7 4-day occasions starting 1 day after the camera was deployed at either 12:00 am (diurnal species) or 12:00 pm (nocturnal species). A detection of a particular species was defined as 1 or more photos of that species captured at a given location during a given period (occasion). Data Management & Software CPW Photo Warehouse (http://cpw.state.co.us/learn/Pages/ResearchMammalsSoftware.aspx) Duration of Study 5/28/2013 to 8/21/2013 (85 days) + 5/30/2014 to 8/18/2014 (80 days) = 165 days (total) Number of Images 155,733 (2013) + 180,932 (2014) = 336,665 (total) Blank Images 52,461 or 33% (2013) + 57,632 or 32% (2014) = 110,093 or 33% (total) False Negative and False Positives All photos were reviewed by at least 2 independent observers. Any disagreements were forwarded to a referee to decide on the correct identification. Camera Trap Days 12,750 (2013) + 12,000 (2014) = 24,750 (total) �Table S2. Real parameter estimates (SE) for detection probability, p, from the best (minimum AICc) model, Summer 2013−2014, Colorado, USA. Each occasion comprised 4 days. Camera traps were baited with 15 ml of peanut butter and 1 ml of rabbit lure, each soaked into a 1 inch x 1 inch piece of wool and tied to a tree 4–5m in front of the camera. Lure was not replenished after initial camera deployment. Detection Probability (SE) Occasion 1 Occasion 2 Occasion 3 Occasion 4 Occasion 5 Occasion 6 Occasion 7 American marten 0.29 (0.05) 0.21 (0.03) 0.16 (0.02) 0.14 (0.02) 0.12 (0.02) 0.12 (0.02) 0.13 (0.03) Black bear 0.23 (0.04) 0.19 (0.03) 0.16 (0.02) 0.14 (0.02) 0.11 (0.02) 0.09 (0.02) 0.08 (0.02) Chipmunk spp. 0.45 (0.05) 0.41 (0.04) 0.36 (0.03) 0.32 (0.02) 0.28 (0.03) 0.24 (0.03) 0.21 (0.03) Coyote 0.09 (0.02) 0.09 (0.02) 0.09 (0.02) 0.09 (0.02) 0.09 (0.02) 0.09 (0.02) 0.09 (0.02) Elk 0.39 (0.03) 0.40 (0.02) 0.40 (0.02) 0.38 (0.02) 0.35 (0.02) 0.31 (0.02) 0.26 (0.03) Golden-mantled ground squirrel 0.43 (0.06) 0.39 (0.05) 0.35 (0.04) 0.31 (0.03) 0.27 (0.04) 0.24 (0.04) 0.21 (0.05) Moose 0.12 (0.05) 0.09 (0.03) 0.07 (0.03) 0.08 (0.03) 0.10 (0.03) 0.14 (0.04) 0.25 (0.08) Mule deer 0.40 (0.01) 0.40 (0.01) 0.40 (0.01) 0.40 (0.01) 0.40 (0.01) 0.40 (0.01) 0.40 (0.01) Porcupine 0.22 (0.07) 0.2 (0.05) 0.17 (0.04) 0.15 (0.04) 0.13 (0.04) 0.12 (0.04) 0.1 (0.04) Red fox 0.33 (0.05) 0.27 (0.04) 0.21 (0.03) 0.16 (0.02) 0.12 (0.02) 0.09 (0.02) 0.07 (0.02) Red squirrel 0.52 (0.02) 0.51 (0.02) 0.49 (0.02) 0.48 (0.01) 0.47 (0.02) 0.46 (0.02) 0.45 (0.02) Snowshoe hare 0.52 (0.02) 0.52 (0.02) 0.52 (0.02) 0.52 (0.02) 0.52 (0.02) 0.52 (0.02) 0.52 (0.02) Yellow-bellied marmot 0.17 (0.05) 0.17 (0.05) 0.17 (0.05) 0.17 (0.05) 0.17 (0.05) 0.17 (0.05) 0.17 (0.05) �Table S3. Beta parameter estimates (SE) for the best (minimum AICc) model for each species analyzed to determine impacts of bark beetle outbreaks on mammals in Colorado, USA, Summers 2013−2014. Parameters for modeling detection probability (p) included an intercept (Int), linear trend in detection (T), and quadratic trend (T2) in detection. Parameters for modeling occupancy (Ψ) included an intercept (Int), habitat stratum (spruce-fir or lodgepole pine), shrub height (ShrubHt), shrub cover due to deciduous species (DCover), shrub cover due to conifer species (CCover), cover of coarse woody debris (DeadDown), cover of bare ground (Bare), canopy cover of aspen (Aspen), topographic wetness index plus (TWIP), topographic position index (TPI), heating loading index (Heat), elevation (Elev), wilderness cell or not (Wild), linear trend through time after beetle impacts (T), quadratic trend through time after beetle impacts (T2), cubic trend through time after beetles (T3), severity of the outbreak expressed as percent of overstory cover that was dead (Severity), and a delayed linear trend through time that began 4 years after beetle impacts (DT). p Ψ Species Int T T2 Int American marten -0.37 (0.38 ) -0.57 (0.22 ) 0.05 (0.03 ) 1.11 (0.64 ) Black bear -1.07 (0.26 ) -0.21 (0.06 ) 11.32 (3.81 ) Chipmunk 0.00 (0.22 ) -0.19 (0.05 ) 2.85 (0.93 ) Coyote -2.35 (0.25 ) Elk -0.51 (0.25 ) 0.11 (0.15 ) Golden-mantled ground squirrel -0.10 (0.31 ) -0.18 (0.07 ) Moose -1.28 (0.83 ) -0.72 (0.46 ) Mule deer -0.40 (0.06 ) Porcupine -1.06 (0.46 ) -0.15 (0.10 ) -0.10 (1.13 ) Red Fox -0.38 (0.29 ) -0.32 (0.07 ) -0.26 (0.53 ) Red Squirrel 0.11 (0.12 ) -0.04 (0.03 ) -0.82 (0.63 ) Snowshoe Hare 0.06 (0.07 ) -2.63 (0.58 ) Yellow-bellied marmot -1.69 (0.43 ) -3.56 (0.67 ) Habitat 1.75 (3.49 ) -0.03 (0.02 ) ShrubHt DCover -2.63 (1.27 ) -1.42 (1.27 ) -3.06 (0.88 ) 0.11 (0.05 ) 8.50 (6.44 ) -1.79 (1.26 ) 1.81 (1.30 ) -6.26 (4.08 ) 1.71 (0.80 ) 0.53 (0.31 ) 1.25 (0.29 ) �Table S3. Continued. Ψ Species CCover DeadDown Bare American marten -2.43 (1.02 ) Black bear -3.23 (1.26 ) Aspen Elk -17.17 (5.27 ) Golden-mantled ground squirrel -0.01 (0.00 ) -5.34 (2.07 ) 0.70 (0.49 ) 0.38 (0.25 ) -2.38 (1.03 ) 1.16 (0.32 ) 0.01 (0.00 ) 2.02 (1.19 ) Porcupine Red Fox 1.55 (0.62 ) -0.002 (0.001) -2.34 (1.20 ) -5.91 (1.72 ) -9.29 (4.14 ) -4.17 (2.46 ) Red Squirrel 2.92 (0.88 ) Snowshoe Hare 2.28 (0.77 ) Yellow-bellied marmot 2.00 (1.39 ) -7.48 (5.99 ) -7.71 (3.29 ) Elev -0.001 (0.0003) -0.78 (0.24 ) Moose Mule deer TPI -0.74 (0.32 ) Chipmunk Coyote TWIP 5.21 (1.45 ) 0.01 (0.00 ) �Table S3. Continued. Ψ Species Wild American marten -0.53 (0.36 ) Black bear 0.79 (0.42 ) T T2 Chipmunk -0.08 (0.04 ) Coyote -0.14 (0.08 ) Elk -0.05 (0.05 ) Golden-mantled ground squirrel -0.78 (0.66 ) 0.29 (0.20 ) Moose 0.66 (0.39 ) -0.08 (0.04 ) Mule deer 1.29 (0.68 ) -0.29 (0.15 ) Porcupine -2.02 (1.07 ) Red Fox -0.71 (0.43 ) Red Squirrel -0.40 (0.33 ) 0.08 (0.07 ) T3 Severity 1.28 (0.71 ) -0.02 (0.02 ) -2.65 (1.66 ) 5.29 (2.01 ) 0.02 (0.01 ) -6.24 (3.30 ) -1.70 (0.75 ) -0.18 (0.13 ) DT 3.37 (1.01 ) Snowshoe Hare Yellow-bellied marmot T*Severity 3.65 (1.60 ) 0.57 (0.38 ) 0.08 (0.07 ) � 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 Journal Articles Description An account of the resource CPW peer-reviewed journal publications 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 Mammalian responses to changed forest conditions resulting from bark beetle outbreaks in the southern Rocky Mountains Description An account of the resource <span>Spruce beetle (</span><i>Dendroctonus rufipennis</i><span>) and mountain pine beetle (</span><i>Dendroctonus ponderosae</i><span>) outbreaks have impacted millions of acres of conifer forest from Alaska to northern Mexico. These species are native to North America, and periodic outbreaks have shaped the structure and composition of conifer forests for millennia. However, the extent and severity of current outbreaks, fueled by favorable climatic conditions and increased susceptibility of forests, are unmatched in recorded history. To characterize the response of a suite of mammalian species to beetle-induced changes in vegetation in the southern Rocky Mountains, we deployed cameras at 300 randomly selected sites during summer 2013–2014. Selected sites spanned gradients of years elapsed since bark beetle outbreaks (YSO) and severity. We fit single-season occupancy models to detection/non-detection data collected for each species to examine a variety of plausible relationships between use of a given stand and YSO, severity, or both. Ungulates exhibited a positive association with bark beetle activity, although the nature of these associations varied by species. Elk (</span><i>Cervus canadensis</i><span>) were positively associated with severity, but not YSO; mule deer (</span><i>Odocoileus hemionus</i><span>) exhibited the opposite relationship. Moose (</span><i>Alces alces</i><span>) responded in a quadratic fashion; use of forest stands adjacent to preferred willow habitat peaked 3–7 yr after an outbreak commenced, but only at high severity. Similarly, yellow-bellied marmot use of impacted stands adjacent to rock outcroppings followed a quadratic trend, but only at high severity. Red squirrel (</span><i>Tamiasciurus hudsonicus</i><span>) use declined in severely impacted stands, likely as a response to diminished cone crops. Golden-mantled ground squirrels (</span><i>Callospermophilus lateralis</i><span>) and chipmunks (</span><i>Neotamias</i><span> spp.) exhibited a shallow negative relationship with YSO, as did coyotes (</span><i>Canis latrans</i><span>). Contrary to our hypotheses, black bears (</span><i>Ursus americanus</i><span>), American marten (</span><i>Martes americana</i><span>), snowshoe hares (</span><i>Lepus americanus</i><span>), and porcupines (</span><i>Erethizon dorsatum</i><span>) did not appear to be substantially influenced by beetle activity. Red fox (</span><i>Vulpes vulpes</i><span>) use was positively associated with YSO, but overall use declined as severity increased. Note that changes in probability of use described here could reflect changes in abundance, home range size, habitat use, or some combination, and in several cases, there was considerable uncertainty across competing models.</span> Bibliographic Citation A bibliographic reference for the resource. Recommended practice is to include sufficient bibliographic detail to identify the resource as unambiguously as possible. Ivan, J. S., A. E. Seglund, R. L. Truex, and E. S. Newkirk. 2018. Mammalian responses to changed forest conditions resulting from bark beetle outbreaks in the southern Rocky Mountains. Ecosphere 9:e02369. <a href="https://doi.org/10.1002/ecs2.2369" target="_blank" rel="noreferrer noopener">https://doi.org/10.1002/ecs2.2369</a> Creator An entity primarily responsible for making the resource Ivan, Jacob S. Seglund, Amy E. Truex, Richard L. Newkirk, Eric S. Subject The topic of the resource Bark beetle outbreak Camera trap Climate change Colorado <em>Dendroctonus ponderosae</em> <em>Dendroctonus rufipennis</em> Mammals Mountain pine beetle Spruce beetle Extent The size or duration of the resource. 18 pages Date Created Date of creation of the resource. 2018-08-16 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> <a href="https://creativecommons.org/licenses/by/3.0/" target="_blank" rel="noreferrer noopener">Attribution 3.0 Unported (CC BY 3.0)</a> 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. Ecosphere Type The nature or genre of the resource Article