Printer Friendly

Winter habitat associations for spotted skunks (Spilogale spp) in South-Central Wyoming.

Introduction

Two members of the Spilogale genus occur within the United States, the eastern spotted skunk (S. putorius) and western spotted skunk (S. gracilis). Both S. putorius and S. gracilis have been described as small omnivorous mephitids that are difficult to distinguish from each other based solely on morphology (Kinlaw, 1995; Verts et al., 2001; Gompper and Hackett, 2005). Whereas the status of S. gracilis is unknown, population trends for S. putorius suggest this species has experienced a significant range-wide decline (Gompper and Hackett, 2005). During the early 20th century, S. putorius was thought to relatively abundant across a wide distribution from the southern Appalachian Mountains to Great Plains regions east of the Rocky Mountains (Choate et al., 1974; Lee et al., 1982; Kinlaw, 1995; Gompper and Hackett, 2005). However, in the 1940s, records from throughout this range indicated harvest rates were reduced, and several reports over subsequent decades suggested that S. putorius was locally uncommon or declining (Conaway and Howell, 1953; Barbour and Davis, 1974; Choate et al., 1974; Merritt, 1987; Wires and Baker, 1994; Reed and Kennedy, 2000; Gompper and Hackett, 2005). Although the cause of this decline remains undetermined, S. putorius has been classified as a species of conservation concern in most states throughout the Great Plains and east coast regions (Gompper and Hackett, 2005). In 2012 the US Fish and Wildlife Sendee found petitions to list the Great Plains subspecies of S. putorius (S. p. interrupta) as a Federally Threatened species under the Endangered Species Act throughout the range of the subspecies may be warranted (USFWS, 2012).

Although this heightened conservation status dictates the development of range-wide management and conservation strategies particularly for S. putorius, information on the natural history associated with both spotted skunk species in North America is lacking. Before the 21st century, information on the ecology of Spilogale was based largely on a few localized studies and anecdotal observations, leaving large portions of the distribution of this genus largely unstudied (see Crabb, 1941, 1948; Mead, 1968; Geluso, 1972; Baker and Baker, 1975; McCullough and Fritzell, 1984; Crooks, 1994; Crooks and Van Vuren, 1995; Reed and Kennedy, 2000). More recently, habitat selection, home range, den site selection, and survey techniques have been evaluated for S. putorius in the Ouachita Mountains of Arkansas and Missouri (Hackctt et al., 2007; Lesmeister et al., 2008, 2009). Likewise, habitat use, denning, and activity patterns of S. gracilis have been examined in Texas (Doty and Dowler, 2006; Nieswenter and Dowler, 2007; Nieswenter et al., 2010). However, throughout much of the continental-wide distribution of this genus throughout North America, these basic natural history traits remains poorly understood. Of particular importance to conservation efforts, this information gap includes regions where the northern distributions between S. p. interrupta and S. gracilis converge, which is thought to occur along the eastern third of the Rocky Mountain States of Wyoming and Colorado (Van Gelder, 1959; Dragoo and Honeycut, 1999).

The handful of published reports related to the ecology of spotted skunks has demonstrated the importance of vegetative cover in selection of habitat for both species. For example habitat use in Washington and den site selection in Texas and California for S. gracilis has been found to be positively associated with dense shrub cover (Carey and Kershner, 1996; Carroll, 2000; Doty and Dowler, 2006; Neiswenter and Dowler, 2007). Likewise, S. putorius was found to be associated with thick understory in Tennessee and displayed second order selection for young dense forest stands with complex understory structure and closed canopy in the Ouachita Mountains (Reed and Kennedy, 2000, Lesmeister et al., 2009). Similar to other small mesocarnivores, spotted skunks presumably select for cover as a means of reducing predation, concealing den sites, and increasing availability or vulnerability of prey (Lesmeister et al., 2009, 2010; Reed and Kennedy, 2000).

Few studies have examined habitat use and the importance of cover to spotted skunks within the foothill transitional habitat that defines the shift from lowland grasslands to mountain ranges that are abundant across the Rocky Mountain States, which comprise the predicted northern contact zone between S. putorius and S. gracilis. Within these foothills vegetative cover is commonly limited to sparse disjoined patches of short trees and shrubs, whereas the complex forest structure more typical of the eastern and western United States are restricted to elevations higher than 2000 m (Knight et al., 2014). Among a greater expanse of open sagebrush and grasslands, foothills are often littered with rocky outcrops and escarpments formed primarily from glacial moraines and erosion of sedimentary rocks (Knight et al., 2014). How spotted skunks utilized the limited cover available or these rocky habitats within this landscape remains unclear even as conservation and management of this genus is an increasing priority for Rocky Mountain States along the predicted boundary.

To address this lack of information for the Spilogale genus in the Rocky Mountain region, we evaluated basic habitat associations for spotted skunks within the foothill transitional habitat in south-central Wyoming. Our objective was to measure the influence of environmental attributes on occupancy of spotted skunks in habitats common to the Rocky Mountain region to determine which habitat attributes may be important to these species. Given the importance of cover documented in other regions, we focused our study on areas where some level of cover would be available. Specifically, these areas consisted of the bases of small mountain ranges and rocky escarpments distributed throughout the open grasslands and sagebrush of the region. We hypothesized, as in other regions, cover would an important habitat component to both spotted skunk species at a home range scale such that (1) occupancy would be positively associated with areas of dense shrub and tree cover, or (2) occupancy would be positively associated with areas comprised of rocky outcroppings that could provide similar benefits to skunks as vegetative cover.

STUDY AREA

We conducted surveys for spotted skunks throughout south-central Wyoming in the vicinity of the Laramie (42.206000, -105.435000), Pedro (42.170000, -106.652000), and Shirley (42.317000, -106.866000) Mountains, as well as the Sentinel and Sweetwater Rocks areas (42.496000, -107.450000; Fig. 1). This region is dominated by sagebrush shrubland and grasslands broken by foothills, escarpments, and small mountain ranges. Throughout our study area, sagebrush shrubland typically consisted of sages (i.e., Artemisia tridentata, A. ridida, A. arbusclua), rabbitbrushes (e.g., Ciysothamus viscidijlorus, Ericameria nauseosa), bitterroots (Purhsia tridenatata), and service-berry (Amelanchier alnifolia) (Knight et al., 2014). Foothills and mountainous areas were comprised largely of rocky outcrops interspersed with xeric and montane pines (Pinus albicaulis, P. contorta, P. flexilis, P. ponderosa), firs (Picea engelmannii, Abies lasiocarpa), mahogany (Cercocarpus montanus, C. ledifolius), and juniper (Juniperus scopulorum-, Knight et al., 2014). Regional elevation ranged from approximately 1500-3000 m. Average temperature and precipitation for the region during October through December from 1980-2015 was -0.6C (SD [+ or -] 1.2) and 7.4cm (SD [+ or -] 2.0), respectively (NOAA, 2015).

PRESENCE ABSENCE SURVEYS FOR SPOTTED SKUNK

We focused our survey efforts for spotted skunks primarily in the bases of mountainous regions and rocky escarpments for two reasons. First, the grasslands and sagebrush habitats that compose much of these regions lack the complex understory and overhead cover found to be an important habitat component in other parts of the ranges of both spotted skunk species. Second, an unrelated study exploring occupancy rates of swift foxes (Vulpes velox) failed to detect spotted skunks at 100 locations throughout grassland and sagebrush habitat throughout the eastern two thirds of Wyoming using as similar survey protocol used here (Cudworth et al., 2011; Cudworth and Grenier 2013). Therefore, we felt these habitats were poorly suited for spotted skunks, especially during the winter months when ground snow cover would further reduce the ability of skunks to conceal themselves.

For each surveyed region, we overlaid a grid of 500 tn radius circular cells. Size of cells was selected to roughly equate the home range size of S. putorius during fall (Lesmeister el al., 2009). We generated random cells throughout each surveyed region using ArcGIS 10.1 and selected a subset of those cells for survey that we could reliably access given weather conditions and terrain. We used baited camera stations to record the presence of spotted skunks from October through December 2014. We installed a single camera (Reconyx Hyperfire hc600, Reconyx Inc, Holmen, WI) near the center of each cell. We baited each camera station with a can of wet cat food (Friskies salmon or tuna flavored. Nestle Purina PetCare Company, St. Louis, MI) and 15 ml of scent lure (Caven's Gusto Lure or Tinctured Skunk Essence, F&T Fur Harvesters Alpena, MI). We attached bait combinations at 0.5 m up the trunk of a tree approximately 4 m in front of the camera. Cameras were positioned to capture the image contained within 2 m of the base of the bait tree to 1 m above the bait. We programmed cameras to record a series of three photographs each time the internal motion sensors were activated for a period of approximately 500 consecutive hours (21 d). After retrieving cameras, we reviewed photos to generate detection histories for all spotted skunks detected during surveys. Due to the morphological similarity between S. putorius and S. gracilis, we did not attempt to distinguish between spotted skunks species in our photographs.

COVARIATES FOR OCCUPANCY AND DETECTION

We evaluated the influence of a variety of environmental variables on occupancy and detection of spotted skunks. Specifically, we considered the influence of habitat attributes on occupancy of spotted skunks at a landscape scale defined by each 500 m radius cell. We elected not to evaluate the influence of microscale habitat attributes because presence of a spotted skunk at small scales may be biased by the baits and lures used at each camera station. Within each cell we calculated percent vegetative cover, percent rocky outcropping, elevation, slope, and presence of a stream based on appropriate layers or imagery in ArcGIS 10.1. We calculated percent vegetative cover (VegCov) by averaging all vegetative cover values assigned to either the tree or shrub layers contain with each cell based on the 2012 Existing Vegetative Cover layer (LANDFIRE, 2012). We calculated the percent of rocky outcropping (Rock) within each cell by digitizing the extent of rocky outcropping contained within each cell using visual analysis of aerial imagery since no appropriate layers were available for our entire study site. We measured elevation (Elev) and slope (Slope) of each cell by averaging values contained within each cell based on the USGS Digital Elevation Models (USGS, 2015). We evaluated whether a stream or river classified as perennial (stream) was contained within a cell based on the stream layer within the National Hydrology Database (USGS, 2013). To measure detection we considered each day as a unique sampling occasion for a total of 21 sampling occasions per cell. For each sampling occasion, we collected daily weather variables including temperature, wind speed, and precipitation from the nearest NOAA weather station to each respective cell. We also included polynomial terms for temperature and precipitation. In addition we measured the influence of seasonality on detection by including a variable for the day of the year for each sampling occasion. We evaluated correlation between covariates for occupancy or detection using Pearson's correlation test. We standardized all covariates prior to modeling to facilitate comparison, interpretation, and model convergence (Grueber et al., 2011). We used the equation [[x.sub.i standardized] = ([x.sub.i] - [mu]) / [sigma] to standardize all covariates.

OCCUPANCY MODELING

We conducted single-species, single-season occupancy analysis to estimate the influence of the covariates on occupancy and detection of spotted skunks (MacKenzie et al., 2005). We developed a global model that contained all detection and occupancy covariates. We assessed goodness-of-fit of the global model by calculating the overdispersion parameter (c) based on paramedic bootstrapping with 1000 permutations (MacKenzie and Bailey, 2004). We conducted a two-staged modeling approach in order to reduce the number of models being considered for analysis. First, we found the best set of covariates for modeling detections by generating a model set that contained all combinations of our detection covariates while occupancy was held constant (intercept only). We determined the best model for detection under this framework calculated from Akaike Information Criterion adjusted for small sample size (AICc; Burnham and Anderson, 2002). We then generated our model set for occupancy which contained all combinations of our occupancy covariates with the best detection model found from the previous step. We then used weighted model averaging based on model weights (w) calculated based on AICc to estimate the effect size and unconditional standard error of each covariate on occupancy (Burnham and Anderson, 2002; Doherty et al, 2012). We calculated the cumulative AICc weight ([summation] w) of all covariates and denoted covariates with a [summation] w [greater than or equal to] 0.5 to have an important influence on occupancy or detection (Barbeiri and Berger, 2004). All analyses were conducted in program R using packages unmarked (Fiske and Chandler, 2011), MuMIn (Barton, 2009), and AlCcmodavg (Mazerolle, 2015).

Results

We detected spotted skunks at 16 of 72 survey sites (22%) during the months of October to December 2014 comprised of detections at 10 out of 46 sites the Laramie Mountains, four out of seven sites in the Sentinel and Sweetwater Rocks, one out of three sites in the Pedro Mountains, and one out of 16 sites in the Shirley Mountains (Fig. 1). All skunks were detected between the hours of 1735 and 0830 and between the elevations of 1829 m and 2535 m (average = 2138, SD = 229).

We found moderate positive correlation between the occupancy covariates of VegCov and Elev (r = 0.62, P-value < 0.001). We therefore removed elevation from further analysis because this covariate was felt to be redundant given that areas of dense vegetative cover are typically restricted to higher elevations throughout our study area.

We found the model which included covariates Precip, Temp, and the polynomial term for Temp to be the best model for detection (AICc weight = 0.12; Table 1). Using this detection model, we found our model which included only the covariate of Rock to be the best model for occupancy (AICc weight = 0.38; Table 1). Based on model averaging, the average probability of occupancy was 0.24 (SE = 0.09, 95% ci = 0.10-0.47) and average probability of detection was 0.07 (SE = 0.02, 95% CI =0.04-0.13). Cumulative AICc weight was [greater than or equal to] 0.5 only for the covariate for Rock ([summation] w = 0.99), which was also found to have a significant positive effect on occupancy ([beta] = 2.51, SE = 1.11, 95% CI =0.34-4.68, P-value = 0.02; Table 2). Conversely, cumulative AICc weight for VegCov was 0.32 and was found to not have a Significant effect on occupancy ([beta] = -0.19, SE = 0.44, 95% CI = -1.77-0.56, P-value = 0.31, Table 2).

Given the relatively weak AICc weight of our top detection model (AICc weight = 0.12), we conducted a post-hoc analysis to evaluate how selection of this model for use in modeling occupancy may influence the effect sizes of our occupancy covariates, specifically the covariate Rock (see Appendix I). We evaluated a second model set in which all combinations of detection and occupancy covariates were modeled simultaneously and found the estimated effect size of Rock to show only slight variation regardless of which detection model was used. The top 20 models based on AICc weight and associated estimates for effect size of Rock can be viewed in Appendix I.

Discussion

The importance of vegetative cover in selection of habitat has consistently been illustrated for both spotted skunk species for regions outside of the foothill transitional habitat that characterizes south-central Wyoming. Cover likely provides several benefits to Spilogale, including protection from predators and increased availability or vulnerability of prey (Lesmeister et al., 2009; Reed and Kennedy, 2000). However, throughout south-central Wyoming, we found vegetative cover, measured by canopy and shrub cover at a landscape scale, to be of little importance to habitat use for this genus. Instead we found a strong positive relationship between the amount of rocky outcrops and occupancy such that spotted skunks were nearly 100 times more likely to occupy an area entirely comprised of rocky outcropping compared to an area devoid of rocky outcropping (Fig. 2). The strength of this relationship suggests that throughout our study area, rocky outcroppings represent an important habitat component to Spilogale during the late-fall and winter months.

Crevices, overhangs, and other features within rocky outcrops likely provide many similar benefits to skunks as vegetative cover. For example these rocky habitats likely provides for concealed denning opportunities and protection from primary predators of both Spilogale species, including coyotes (Canis latrans), bobcats (Lynx rufus), and birds of prey (Kinlaw, 1995; Verts et al., 2001; Lesmeister et al, 2010). Rocky outcrops additionally serve as barriers to high winds and blowing snow, leaving windward slopes relatively snow-free (Knight et al., 2014). In areas where snow does accumulate, the structural complexity of rocky outcrops may also allow for the formation of subnivean spaces under winter snowpacks. We suspect these subnivean shelters could be valuable to spotted skunks for thermoregulation and by providing secure resting sites when snow conditions would otherwise make travel energetically inefficient, as has been documented for other mesocarnivore species (Taylor et al, 1996; Wilbert et al, 2000). In addition although we did not assess the availability of common prey species such as eastern cottontails (Sylvilagus floridanus), voles (Microtus spp), and other small mammals (Kinlaw 1995), the shelter provided by rocky outcroppings during winter may also similarly benefit these prey species thereby increasing the abundance or diversity of food resources to Spilogale.

The relationships with rocky outcrops and vegetative cover we observed in Wyoming may be due in part to regional habitat availability. The dense vegetative cover associated with forested habitat types is typically restricted to mid to high elevations of mountainous regions throughout south-central Wyoming (Knight et al., 2014). These areas may be less accessible to Spilogale especially during winter months due to the depth and persistence of snow. Rocky outcroppings found along the bases of mountain ranges or escarpments therefore may not only provide many of the same habitat needs as the dense vegetation cover spotted skunks are more commonly associated with throughout other parts of their distribution but may also be more accessible and available to spotted skunks. Although our study is limited in ability to measure seasonal differences in habitat associations, we suspect rocky outcrops remain important to spotted skunks throughout the year in the foothills of Wyoming and other Rocky Mountain regions as few nearby habitat features are likely to offer the same structural complexity necessary for avoiding predators, concealing den sites, and minimizing exposure to harsh weather conditions. Consideration of these regional differences in habitat availability will be essential to obtaining a comprehensive understanding of the ecology of Spilogale throughout the diversity of habitat types over which these mephitids occur.

Although the survey technique we used here (i.e., photographs) was inadequate in conclusively determining which of the two species of Spilogak we detected at each location, we note an important characteristic that could be useful to species-level interpretation of our results. Some authors have suggested tail coloration is a distinguishing feature between 5. p. interrupta and S. gracilis, such that the tail of the former is nearly entirely black in color while the latter is marked along the top third with white coloration (Van Gelder, 1959). Although tail coloration is thought to vary geographically throughout the range of S. p. interrupta (Van Gelder, 1959), this morphologic feature has been cited as a species identifier by a recent petition to list this subspecies as Federally Threatened (USFWS, 2012). We felt the variation within this characteristic would make species identification subjective and inconclusive especially along the potential contact zone between species. However, in the photographs we obtained as part of the study, roughly the top third of the tails of all Spilogale detections were clearly white in coloration. If tail coloration is unique to species, then these skunks we detected could all be assigned to S. gracilis. Given the distribution of our observations, this would also indicate the distribution of the species extents eastward at least to the Laramie Mountains (the eastern extent of our skunk detections). This distribution would roughly align with the predicted boundary for S. gracilis previously proposed (Kinlaw, 1995; Van Gelder, 1959). In addition our results would suggest that rocky outcroppings are an important habitat feature for S. gracilis in south-central Wyoming, an attribute that could be used to possibly differentiate between habitat associations of S. gracilis and S. p. interrupta. However, we caution that until a genetic-based approach is advanced in order to determine if tail coloration alone can distinguish between S. p. interrupta and S. gracilis, especially along a possible contact zone between the two species, our results may be more appropriately assigned to the genus-level.

We acknowledge a limitation to our study related to our relatively low sample size as well as our constrained ability to measure covariates that strongly influence the detection of Spilogale. These factors likely combined to produce the relatively weak AICc weight of our top detection model. We suspect that with an increased number of observations and the capability to collect covariate data which may more accurately contribute to detection such as proximity to den site, age, and sex of individuals, etc., models that better describe detection of spotted skunks could be formulated. However, our post-hoc analysis to evaluate the impact of the detection model used to model occupancy suggested while the effect size of the Rock covariate did vary slightly depending on which detection model was used, the overall trend of a positive association between occupancy and rocky outcroppings remained consistent. Therefore, while additional research to understand factors contributing to detection of Spilogale would be useful to future efforts employing an occupancy-based framework, the association with rocky outcrops suggested by our results is unlikely to be artifact of our modeling approach.

Our knowledge of spotted skunks, especially within the region thought to represent the transition line between the eastern and western species in the United States, remains limited even as conservation concerns rise. The results we present here help elucidate habitat associations of spotted skunks in this region by suggesting that rocky outcrops may provide many of the habitat requirements for the Spilogale genus skunks that are satisfied by dense vegetative cover elsewhere. Our finding will therefore be valuable to subsequent research addressing spotted skunks ecology in the foothill transitional habitat of the Rocky Mountains regions. Understanding regional differences in habitat associations will be critical to future management and conservation efforts for spotted skunks.

Acknowledgments.--This project was funded by U. S. Fish and Wildlife Semce State Wildlife Grants and Wyoming State Legislature General Fund Appropriations. Thank you to Z. Walker, N. Bjornlie, M. BenDavid, and D. Keinath for reviewing earlier versions of this manuscript. Special thanks to P. Mahonev for his assistance and advice. Additionally, we thank M. Grenier for his contributions to this project.

Literature Cited

Armstrong, D. M., J. P. Fitzgerald, and C. A. Meaney. 2011. Mammals of Colorado, second Edition. Denver Museum of Nature and Science and University Press of Colorado. Boulder, USA 704 p.

Baker, R. H. and M. W. Baker. 1975. Montane habitat used by the spotted skunk (Spilogale putorius) in Mexico. J. Mammal. 56:671-673.

Barbour, R. W. and W. 11. Davis. 1974. Mammals of Kentucky. Univ. Press of Kentucky, Lexington. 321. p. Barton, K. 2009. MuMIn: multi-model inference. R package, version 0.12.2.

Burnham, K. P. and D. R. Anderson. 2002. Model selection and multimodel inference. A practical information-theoretical approach, 2nd ed. Springer, New York, New York. 488 p.

Carey, A. B. and J. E. Kershner. 1996. Spilogale gracilis in upland forests of western Washington and Oregon. Northwest. Nat., 77:29-34.

Carroll K. N. 2000. Macro- and microhabitat characteristics of the western spoiled skunk, Spilogale gracilis, in the Sierra Nevada of northern California. M.S. diesis, California State University, Sacramento.

Choate, J. R., E. D. Fleharty, and R. J. Little. 1973. Status of the spotted skunk, Spilogale putorius, in Kansas. Trans. Kans. Acad. Sri., 76:226-233.

Conaway, C. H. and J. C. Howell. 1953. Observations on the mammals of Johnson and Carter counties, Tennessee, and Avery County, North Carolina. J. Tenn. Acad. Sri., 17:53-61.

Crabb, W. D. 1941. Food habits of the prairie spotted skunk in southeastern Iowa J. Mammal.. 22:349-364.

--. 1948. The ecology and management of the prairie skunk in Iowa. Ecol. Monograph., 18:201-232.

Crooks, R.R. 1994. Demography and status of the island fox and the island spotted skunk on Santa Cruz Island, California. Southwest. Sal.. 39:257-262.

--, M. Grigione, A. Scoville, and G. Scoville. 2008. Exploratory use of track and camera surveys of mammalian carnivores in the Peloncillo and Chiricahua mountains of southeastern Arizona. Southwest. Sat., 53:510-517.

--and D. Van Vuren. 1995. Resource utilization by two insular endemic mammalian carnivores, the island fox and island spotted skunk. Oecologia 104:301-307.

Cudworth, N., L. Van Fleet, and M. Crenier. 2011. Monitoring trends and documenting distribution of swift fox in Wyoming. In: M. Crenier (ed.) Threatened, Endangered, and nongame birds and mammal investigations. Annual completion report. Wyoming Game and Fish Department.

--And M. Grenier. 2013. Habitat characteristics influencing swift fox (Vulpes velox) occupancy in eastern Wyoming. In: A. Orabona (ed.) Threatened, Endangered, and nongame birds and mammal investigations. Annual completion report. Wyoming Game and Fish Department.

Desanty, J. 2001. A review of the status of plains spotted skunk (Spilogale putorius interrupta) throughout its range in North America. Missouri Department of Conservation, Columbia.

Doherty, P. F., G. C. White, and K. P. Burnham. 2012. Comparison of model building and selection strategies. J. Ornithol., 152:S317-S323.

Don. J. B. and R. C. Dowler. 2006. Denning ecology in sympatric populations of skunks (Spilogale gracilis and Mephitis mephitis) in west-central Texas. J. Mammal., 87:131-138.

Drygoo, J. W. and R. L. Honeycutt. 1999. Eastern spotted skunk Spilogale putorius. p. 185-186. In: D. E. Wilson and S. Ruff (eds.). The Smithsonian book of North American mammals. Smithsonian Institution Press, Washington, DC. 816 p.

Fiske, I. and R. Chandler. 2011. Unmarked: an R package for fitting hierarchical models of wildlife occurrence and abundance. J. Slat. Softw., 43:1-23.

Foreman, K. R. and D. E. Pearson. 1998. Comparison of proposed survey procedures for detection of forest carnivores. J Wildl. Manag., 62:1217-1226.

Geluso, K. N. 1972. Western spotted skunk in Oklahoma. Southwest. Sal.. 16:457-458.

Gompper, M. E. and II. M. Hackett. 2005. The long-term, range-wide decline of a once common carnivore: the eastern spotted skunk (Spilogale putorius). Anim. Conserv., 8:195-201.

Grueber, C. E., S. Nakagawa, R. J. Laws. and I. G. Jamif.son. 2011. Multimodel inference in ecology and evolution: challenges and solutions. J. Evol. Biol., 24:699-711.

Hackett, H. M., D. B. Lesmeister, J. Desanty-Combes, W. G. Montague, J.J. Millspaugh, and M. E. Gompper. 2007. Detection rates of eastern spotted skunks (Spilogale putorius) in Missouri and Arkansas using live-capture and non-invasive techniques. Am. Midl. Nat., 158:123-131.

Kinlaw, A. 1995. Spilogale putorius. Mamm. Species., 511:1-7.

Knight, D. H., G. P. Jones, W. A. Reiners, and W. H. Romme. 2014. Mountains and plains: The ecology of Wyoming landscapes, second edition. Vale University Press, New Haven, Connecticut. 404 p.

Landfire, 2014. Existing Vegetation Type Layer, LANDFIRE 1.2.0, US Department of the Interior, Geological Survey. Accessed 15 June 2014.

Lee, D. S., L. B. Funderburg Jr., and M. K. Clark. 1982. A distributional survey of North Carolina mammals. Occasional Papers of the North Carolina Biological Survey, No. 1982-10. North Carolina State. Mus. Nat. Hist., Raleigh. 72 pp.

Lesmiester, D. B., M. E. Gompper, and J. J. Millspaugh. 2008. Summer resting and den site selection by eastern spotted skunks (Spilogale putorius) in Arkansas. J. Mammal.. 89:1512-1520.

--, --, And --. 2009. Habitat selection and home range dynamics of eastern spotted skunk in the Ouachita Mountains, Arkansas, USA. J Wildl. Manag., 73:18-25.

--. J. J. Millspaugh. M. E. Gompper, and T.W. Mong. 2010. Eastern spotted skunks (Spilogale putorius) survival and cause-specific mortality in the Ouachita Mountains, Arkansas. Am. Midl. Nat., 164:52-60.

Mackenzie, D. I. and L. L. Bailey. 2004. Assessing the fit of Site-Occupancy Models. J. Agric. Biol. Environ. Stat., 9:300-318.

--,--, J. A. Royle, K. H. Pollock, L. L. Bailey, and J. E. Hines. 2005. Occupancy estimation and modeling--inferring patterns and dynamics of species occurrence. Elsevier Publishing, Burlington, Massachusetts. 324 p.

Mazerolle, M. J. 2015. Model selection and multimodel inference based on (Q)AIC(c). R package. Version 2.0-3.

McCullough, C. R. and E. K. Fritzell. 1984. Ecological observations of eastern spotted skunks on the Ozark Plateau. Trans. Missouri Acad. Sri. 18:25-32.

Mead, R. A. 1968. Reproduction in western forms of the spotted skunk (Genus Spilogale). J. Mammal., 49:373-390.

Merritt, J. E. 19S7. Guide to the mammals of Pennsylvania. Univ. of Pittsburgh Press. Pittsburgh. 408 p.

Neiswenter, S. A. and R. C. Dowler. 2007. Habitat use of western spotted skunks and striped skunks in Texas. J. Wildl. Manag., 71:583-586.

--, --, and J. H. Young. 2010. Activity patterns of two sympatric species of skunks (Mephitis mephitis and Spilogale gracilis) in Texas. Southwest Nat, 55:16-21.

NOAA. 2015. National oceanic and atmospheric administration. United States Department of Commerce, <http://www.noaa.gov/> Accessed 10 Jan. 2015.

Reed, A. W. and M. L. Kennedy. 2000. Conservation status of the eastern spotted skunk Spilogaleputorius in the Appalachian Mountains of Tennessee. Am. Midl. Nat., 144:133-138.

Taylor, S. L, and S. W. Buskirk. 1994. Forest microenvironments and resting energetic of the American marten Martes americana. Urography, 17:249-256.

USGS. 2013. National Hydrography Geodatabase (NHD). United Stales Geological Survey. <http:// viewer.nationalmap.gov/viewer/nhd.html> Accessed 1 Dec 2016.

USGS. 2015. National Elevation Dataset (NED). United States Geological Survey. <http://viewer. nationalmap.gov/> Accessed I Dec 2016.

[USFWS] United States Fish and Wildlife Service. 2012. Endangered and Threatened Wildlife and Plants; 90-day finding on a petition to list the prairie gray fox, the plains spotted skunk, and a distinct population segment of the Mearn's eastern cottontail in eastern central Illinois and western Indiana as Endangered or Threatened species. Fed. Reg.. 77:71759-71771.

Van Gelder, R. G. 1959. A taxonomic revision of the spotted skunk (genus Spilogale). Bull. Am. Mus. Nat. Hist., 117:229-392.

Verts, B. J., L. N. Carraway, and A. L. Kinlaw. 2001. Spilogale gracilis. Mamm. Species., 674:1-10.

Wilbert, C.J., S.W. Buskirk, and K. G. Gerow. 2000. Effects of weather and snow on habitat selection by American martens Martes americana. Can. J. Zool., 78:1691-1696.

Wires. L. R. and R. J. Baker. 1994. Distribution of the spotted skunk (Spilogale putorius) in Minnesota. Minnesota Department of Natural Resources, St. Paul. 16 p.

Submitted 9 March 2016

Accepted 31 March 2017
Appendix I.--Top 20 models based on AICc weight from post-hoc
analyses to evaluate the influence of the detection model on
the estimated effect size of the covariate Rock on occupancy.
For this analysis, we evaluated a model set in which all
combinations of detection and occupancy covariates were modeled
simultaneously. We found the estimated effect size of Rock to
show only slight variation regardless of which detection model
was used

Model                                    Effect size     SE
                                            (rock)

p(Precip+Temp+[Temp.sup.2]),                 2.51       1.05
psi(Rock)
p(Temp+[Temp.sup.2]), psi (Rock)             2.51       1.08
p(Precip+[Temp.sup.2]), psi(Rock)            2.55       1.23
p(Precip+Temp+[Temp.sup.2]),                 2.45       1.04
psi(Rock+VegCov)
p(Precip), psi(Rock)                         2.73       1.19
p(Temp+[Temp.sup.2]),                        2.44       1.07
psi(Rock+VegCov)
p(Precip+Temp+[Temp.sup.2]),                 2.62       1.28
psi(Stream+Rock)
p(Precip+Temp+[Temp.sup.2]),                 2.50       1.27
psi(Rock)
p(Temp+[Temp.sup.2]), psi(Rock)              2.62       1.34
p(Temp), psi(Rock)                           2.57       1.11
p(Temp+[Temp.sup.2]),                        2.51       1.06
psi(Stream+Rock)
p(Precip+Temp+[Temp.sup.2]+Wind),            2.43       1.21
psi (Rock)
p(Precip+[Precip.sup.2]+Temp+                2.50       1.04
[Temp.sup.2]), psi(Rock)
p(Precip+Temp), psi (Rock+VegCov)            2.51       1.06
p(Date+Precip+Temp+[Temp.sup.2]),            2.58       1.16
psi (Rock)
p(Precip+Temp+[Temp.sup.2]),                 2.48       1.05
psi(Rock+Slope)
p(Temp+[Temp.sup.2]+Wind), psi (Rock)        2.51       1.09
p(Date+Temp+[Temp.sup.2]), psi (Rock)        2.80       1.83
p(Temp+[Temp.sup.2]), psi(Rock+Slope)        2.65       1.18
p(Precip+Temp), psi (Stream + Rock)          2.72       1.35


JESSE T. BOULERICE (1) and BRIAN M. ZINKE

Wyoming Game and Fish Department, Lander 82520

(1) Corresponding author: e-mail: Jesse.Boulerice@wyo.gov

Caption: Fig. 1.--Map of areas surveyed for spotted skunks in south-central Wyoming

Caption: Fig. 2.--Model averaged effect sizes (dark line) and 95% confidence intervals (shaded area) for the occupancy covariates of rocky outcrop. Percentage represent the amount of each 500 m radius cell (0.78 [km.sup.2]) determined to be comprised of rocky outcropping during surveys conducted during October-December 2014, south-central Wyoming
Table. 1.--Model ranks, likelihoods, number of parameters, and weights
for both the detection and occupancy models calculated from Akaike
Information Criterion adjusted for small sample size ([AIC.sub.c])
used to model occupancy and detection of spotted skunks from October
to December 2014, south-central, Wyoming. Modeling was conducted
using a two-staged approach. First, we found the best detection model
by evaluating a model set in which all combinations of detection
covariates were analyzed while occupancy was held constant
(intercept only). Second, we generated our model set for occupancy
which contained all combinations of our occupancy covariates with the
best detection model found from the previous step Detection Models

Model                      [AIC.sub.c]     Delta       Likelihood
                                         [AIC.sub.c]

Detection Models
  Precip+Temp+               281.28           0          135.19
 [Temp.sup.2]
 Temp+[Temp.sup.2]           281.49         0.20         136.46
 Precip+Temp                 281.59         0.30         136.50
 Temp                        282.39         1.10         138.02
 Precip                      283.12         1.83         138.38
 Precip+Temp+                283.36         2.08         135.04
 [Temp.sup.2]+Wind
Occupancy Models
 Rock                        263.49           0          125.10
 Rock-t-VegCov               264.93         1.44         124.60
 Rock+Stream                 265.36         1.87         124.80
 Rock+VegCov+Stream          265.95         2.46         125.10
 Rock+VegCov+Slope           266.97         3.49         124.34
 Rock+Stream-(-Slope         267.47         3.98         124.59

Model                      k   [AIC.sub.c]
                                 weight

Detection Models
  Precip+Temp+             5      0.12
 [Temp.sup.2]
 Temp+[Temp.sup.2]         4      0.10
 Precip+Temp               4      0.10
 Temp                      3      0.07
 Precip                    3      0.05
 Precip+Temp+              6      0.04
 [Temp.sup.2]+Wind
Occupancy Models
 Rock                      6      0.38
 Rock+VegCov               7      0.18
 Rock+Stream               7      0.15
 Rock+VegCov+Stream        7      0.11
 Rock+VegCov+Slope         8      0.07
 Rock+Stream+Slope         8      0.05

Table 2.--Cumulative AICc weight, effect sizes, standard error,
95% confidence intervals, and P-values for all occupancy covariates
considered in our model set. Effect sizes and standard errors were
calculated based on weighted model averaging of [AIC.sub.c]
weights of all models where standardized values were used
for all covariates

            Cumulative
Covariate   Aide weight   P-value   Effect size     SE    Lower 95%
                                                              CI

Rock           0.99        0.02         2.51       1.11      0.34
VegCov         0.32        0.66        -0.19       0.44     -1.77
Slope          0.22        0.99        <-0.01      0.27     -1.15
Stream         0.30        0.75         0.09       0.30     -0.61

Covariate   Upper 95%
               CI

Rock          4.68
VegCov        0.58
Slope         1.13
Stream        1.32
COPYRIGHT 2017 University of Notre Dame, Department of Biological Sciences
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2017 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Boulerice, Jesse T.; Zinke, Brian M.
Publication:The American Midland Naturalist
Article Type:Report
Geographic Code:1U8WY
Date:Jul 1, 2017
Words:5966
Previous Article:Initial movements of re-introduced elk in the Missouri Ozarks.
Next Article:Use of buildings by Indiana bats (Myotis sodalis) and other bats in South-central Iowa.
Topics:

Terms of use | Privacy policy | Copyright © 2019 Farlex, Inc. | Feedback | For webmasters