1 Running header: Myotis septentrionalis roost selection 2 Roost selection by male northern long-eared bats (Myotis 3 septentrionalis) in a managed fire-adapted forest 4 5 Jesse M. Alstona,b*, Ian M. Abernethyb, Douglas A. Keinathc, and Jacob R. 6 Goheend 7 8 a. Program in Ecology, Department of Zoology and Physiology, University of Wyoming, 9 Laramie, WY 82071, USA 10 b. Wyoming Natural Diversity Database, University of Wyoming, Laramie, WY 82071, USA 11 c. Wyoming Ecological Services Field Office, U. S. Fish and Wildlife Service, Cheyenne, WY 12 82009, USA 13 d. Department of Zoology and Physiology, University of Wyoming, Laramie, WY 82071, USA 14 15 *Corresponding author: jalston@uwyo.edu 16 1 17 Abstract: Wildlife conservation in multi-use landscapes requires identifying and conserving 18 critical resources that may otherwise be destroyed or degraded by human activity. Summer day- 19 roost sites are critical resources for bats, so conserving roost sites is a focus of many bat 20 conservation plans. Studies quantifying day-roost characteristics typically focus on female bats 21 due to their much stronger influence on reproductive success, but large areas of species’ ranges 22 can be occupied predominantly by male bats due to sexual segregation. We used VHF telemetry 23 to identify and characterize summer day-roost selection by male northern long-eared bats (Myotis 24 septentrionalis) in a ponderosa pine (Pinus ponderosa) forest in South Dakota, USA. We tracked 25 18 bats to 43 tree roosts and used an information-theoretic approach to determine the relative 26 importance of tree- and plot-level characteristics on roost site selection. Bats selected roost trees 27 that were larger in diameter, more decayed, and under denser canopy than other trees available 28 on the landscape. Much like studies of female northern long-eared bats have shown, protecting 29 large-diameter snags within intact forest is important for the conservation of male northern long- 30 eared bats. Unlike female-specific studies, however, many roosts in our study (39.5%) were 31 located in short (≤ 3 m) snags. Protecting short snags may be a low-risk, high-reward strategy for 32 conservation of resources important to male northern long-eared bats. Other tree-roosting bat 33 species in fire-prone forests may benefit from forest management practices that promote these 34 tree characteristics, particularly in high-elevation areas where populations largely consist of 35 males. 36 37 Key words: Black Hills, Chiroptera, forest management, habitat use, prescribed fire, ponderosa 38 pine (Pinus ponderosa), radiotelemetry 39 2 40 1. Introduction 41 Habitat degradation by humans is a leading cause of extinction and population declines of 42 species globally (Dobson et al., 1997; Halpern et al., 2008; Hansen et al., 2013). Less than 15% 43 of Earth’s land surface falls within a protected area, and less than half of that area is free from 44 human development, agriculture, livestock grazing, light pollution, and transportation 45 infrastructure (Jones et al., 2018). Even in relatively intact ecosystems, land uses other than 46 conservation of nature—such as wildfire prevention, livestock grazing, recreation, and extraction 47 of timber and other forest products—are the norm rather than the exception. Conservation 48 measures targeting these multi-use landscapes are thus vital for conserving species (Kremen and 49 Merenlender, 2018). 50 In multi-use landscapes, successful conservation often requires the identification of 51 critical resources for species of conservation concern so that the supply of those critical resources 52 can be maintained or increased. Day-roosts appear to be critical resources for many bats, 53 providing shelter from predators and environmental stressors (Fenton et al., 1994; Solick and 54 Barclay, 2006), communal sites for social interactions (Willis and Brigham, 2004), and secure 55 places to raise young (Kunz, 1982). Bats spend most of their time in day-roosts, alone or in 56 groups of up to millions of individuals, depending on sex, species, and reproductive status. 57 Patterns of bat abundance and distribution are correlated with roost availability (Humphrey, 58 1975), and declines in reproductive success have been documented when pregnant or lactating 59 bats are experimentally excluded from preferred roosts (Brigham and Fenton, 1986). Because 60 day-roosts are so important for bats, measures to conserve roosts feature prominently in bat 61 conservation plans. Resource managers seeking to conserve bats while managing landscapes for 62 multiple uses benefit from knowledge that promotes bat roost conservation. 3 63 We evaluated day-roost selection by male northern long-eared bats (Myotis 64 septentrionalis) in a ponderosa pine (Pinus ponderosa) forest in the Black Hills of South Dakota, 65 USA. Our study population inhabits a managed fire-adapted forest at the western edge of this 66 species’ range. Northern long-eared bats inhabit much of the eastern United States and southern 67 Canada (Caceres and Barclay, 2000), but are increasingly threatened by white nose syndrome 68 and have been protected in the United States under the Endangered Species Act since 2015 and 69 in Canada under the Species at Risk Act since 2014. Throughout their range, northern long-eared 70 bats roost almost exclusively in tree cavities and under sloughing bark within intact forest (Lacki 71 et al., 2009), and forage within forests or at forest edges (Henderson and Broders, 2008; Owen et 72 al., 2003; Patriquin and Barclay, 2003). 73 At our study site and other high-elevation areas in the Black Hills, male bats are much 74 more common than females (Choate and Anderson, 1997; Cryan et al., 2000). Sexual segregation 75 driven by elevation or temperature is widespread among bats, and is believed to be driven by 76 differences in energy requirements that allow males to inhabit areas that are colder or have less 77 prey (Barclay, 1991; Ford et al., 2002; Senior et al., 2005). Male northern long-eared bats are 78 therefore likely to occupy substantially different habitat than females, but range-wide 79 conservation for the species is informed predominantly by studies focusing on female bats (J. 80 Alston, unpublished data). Forest managers in male-dominated areas may therefore rely on 81 incomplete information to conserve the majority of bats within their jurisdictions. Our study 82 provides managers in such areas with information to appropriately guide management in male- 83 dominated areas and supplement the existing wealth of information on female habitat use. 84 To evaluate factors driving roost selection, we tracked adult male northern long-eared 85 bats to day-roosts and quantified characteristics of both used and available roost trees using 4 86 variables easily measured by forest and wildlife managers. We evaluated these data using an 87 information-theoretic approach to select the best models from a suite of candidate models. We 88 hypothesized that in this managed forest, bats primarily select roost trees with characteristics that 89 promote cavity formation (e.g., tree size and amount of decay), the number of nearby roosts (e.g., 90 plot-level tree and snag density), and thermal characteristics suitable for behavioral 91 thermoregulation (e.g., canopy cover and orientation in relation to sunlight). 92 93 2. Methods 94 2.1. Study Area 95 We conducted our study during the summers of 2017 and 2018 on Jewel Cave National 96 Monument (43˚ 45’ N, 103˚ 45’ W) and surrounding areas of Black Hills National Forest, 16 km 97 west of Custer, South Dakota, USA. In this area, mean monthly summer high temperatures range 98 between 22 – 27˚C and mean monthly summer precipitation ranges between 60 – 80 mm 99 (Western Regional Climate Center, 2018). Open ponderosa pine forests dominate our study site, 100 with Rocky Mountain juniper (Juniperus scopulorum) and quaking aspen (Populus tremuloides) 101 occurring locally. In our local study area, forests form a heterogenous mosaic with northern 102 mixed-grass prairie where a large stand-replacing fire occurred in 2000. A large cave system and 103 several smaller caves lie underground at our study site, and there is substantial topographic relief 104 on the landscape in the form of intersecting canyon systems and rock outcrops. 105 Forests in this landscape are intensively managed. Black Hills National Forest typically 106 uses even-aged management techniques other than clear-cutting (e.g., two-step shelterwood 107 harvest). Stand harvest rotations are 120 years on average, but selective cutting occurs at 10- to 108 20-year intervals to harvest mature trees and thin the understory. Aside from large severe 5 109 wildfires, the forest self-regenerates and does not require planting. Forest management on private 110 lands generally also follow this formula but thinning intervals vary (B. Phillips, personal 111 communication). Forests on Jewel Cave National Monument are managed for resource 112 preservation, primarily using prescribed fire. 113 114 2.2. Capture and VHF Telemetry 115 We used mist nets to capture bats over permanent and semi-permanent water sources 116 (e.g., springs, stock tanks, and stock ponds). In summer (Jun–Aug) 2017 and 2018, we netted 20 117 and 49 nights at 15 water sources. Mist netting sites were distributed throughout our study area, 118 and all were in or near large burned areas and harvested areas. We opened mist nets at civil 119 sunset and closed them after five hours and during inclement weather. We affixed VHF 120 transmitters (0.28 g LB-2X model – Holohil Systems Ltd., Carp, ON, Canada; 0.25 g model – 121 Blackburn, Nacogdoches, TX, USA) between the scapulae of adult male northern long-eared 122 bats with latex surgical adhesive (Osto-Bond, Montreal Ostomy, Montreal, QC, Canada). In our 123 study area and others in the region (Cryan et al. 2000), sex ratios are overwhelmingly male. 124 Because patterns of roost selection can differ between male and female bats (Boland et al., 2009; 125 Elmore et al., 2004; Hein et al., 2008; Perry and Thill, 2007), we targeted males specifically. 126 Additionally, the roosting habits of male bats are less studied than those of females—only 2 of 127 the 14 peer-reviewed studies on roost selection of northern long-eared bats provide data on 128 males, and 11 out of 111 peer-reviewed studies on roost selection of cavity-roosting bats in 129 general provide data on males (J. Alston, unpublished data). All transmitters weighed <5% of the 130 mass of the bat (Aldridge and Brigham, 1988). We tracked bats to roosts each day transmitters 131 were active using handheld VHF receivers (R-1000 model, Communication Specialists Inc., 6 132 Orange, CA, USA) equipped with flexible H antennae (RA-23K model, Telonics Inc., Mesa, AZ, 133 USA). All tracking was conducted on foot. All protocols were approved by the University of 134 Wyoming and National Park Service Animal Care and Use Committees and met guidelines 135 approved by the American Society of Mammalogists (Sikes et al., 2016). 136 137 2.3. Roost Characterization 138 To characterize roosts, we collected data for each roost and randomly sampled available 139 roost trees in our study area. We identified available roost trees by generating a sample of 200 140 random points within 2.53 km (the farthest distance we located a bat roosting from its capture 141 site during our study) of sites where we captured northern long-eared bats and selecting the 142 nearest available roost tree at a random bearing from each point. We therefore compared used 143 roosts to 200 available roosts. We defined available roost trees as live trees >20 cm in diameter 144 or any dead tree with a visible defect (e.g., sloughing bark or cavities) sufficiently large for a bat 145 to roost within. For each tree and plot, we measured characteristics that may influence roost 146 suitability (Table 1; Table A.1). We measured vegetation characteristics at two spatial scales: 1) 147 individual trees, and 2) a 706.86-m2 (15-m radius) plot around the tree. We also measured 148 topographic variables at the plot scale. 149 150 2.4. Statistical Analysis 151 To quantify differences between roost trees used by northern long-eared bats and the 200 152 randomly sampled available roost trees, we used the R statistical software environment (R Core 153 Team, 2018) to build binomial-family generalized linear models. Because we were unable to 154 confirm that available roost trees were never used by bats, our analyses should be interpreted 7 155 within the context of the use-availability resource selection framework (Beyer et al., 2010; 156 Johnson et al., 2006; Manly et al., 2007). We employed an information-theoretic approach using 157 Akaike’s Information Criterion adjusted for small sample sizes (AICc) to compare competing 158 models (Burnham and Anderson, 2002) using the ‘MuMIn’ package (Barton, 2018). We 159 calculated AICc values and model weights (wi) for all possible combinations of a maximum of 8 160 predictors (one variable for each 5 observations) in our set of candidate models to prevent biased 161 coefficient estimates and unreliable confidence interval coverage (Vittinghoff and McCulloch, 162 2007). Predictors with variance inflation factors (VIFs) > 10 were removed from consideration in 163 our global model to reduce problems associated with multicollinearity (Kutner, 2005). Because 164 no model had a wi > 0.90, we averaged model coefficients for all models with cumulative wi > 165 0.95 using the full-averaging method to obtain a final averaged model (Burnham and Anderson, 166 2002). Finally, we validated our averaged model using area under the receiver operating 167 characteristic curve (AUC; Manel et al., 2001; Swets, 1988). 168 169 3. Results 170 We located 2.4 ± 0.3 (range: 1-5) roost trees per bat during our study, for a total of 44 171 roosts used on 59 days by 18 bats. Aside from one roost in a rock crevice, bats roosted 172 exclusively in ponderosa pines, either in cavities or under loose bark. Thirty-six out of 43 tree 173 roosts (83.7%) occurred in dead trees (hereafter termed “snags”). Seventeen of 43 (39.5%) roosts 174 that we located occurred in broken-off snags ≤ 3 m in height. Bats typically roosted in the same 175 patch of contiguous forest for the active life of the transmitter. Bats roosted 790 ± 90 m (range: 176 55 – 2,530 m) from the sites at which they were captured. 8 177 Our global model distinguishing used roost trees from available roost trees incorporated 178 DBH, tree height, decay class, slope, aspect (split into two components—eastness and 179 southness), percent bark remaining, plot tree density, plot snag density, plot canopy cover, and 180 interaction terms between slope and eastness and slope and southness. The snag variable was 181 removed from consideration so that no variable in the global model had a VIF >10. The global 182 model provided an adequate fit to the data (le Cessie-van Houwelingen-Copas-Hosmer global 183 goodness of fit test; z = 0.805, p = 0.421). Our averaged model (incorporating 104 models in our 184 confidence set; Table A.2) indicated that DBH, decay class, and canopy cover were important 185 variables (Table 2). Significant (p < 0.05) averaged model coefficients, confidence intervals, and 186 scaled and unscaled odds ratios are reported in Table 3. Mean differences between used and 187 available roost trees among our variables of interest are reported in Table 4. Predictive 188 performance of the averaged model was very high (AUC = 0.924). 189 Three variables (DBH, decay class, and canopy cover) were positively related to roost 190 selection (Fig. 1; Table 2). For each 5 cm increase in DBH, odds of selection increased by 61% 191 (95% CI: 21-113%). Use was greater than availability at all diameters >37 cm. For each 1 unit 192 increase in decay class, odds of selection increased by 111% (95% CI: 47-203%). Use was 193 generally greater than availability for decay classes >2. For each additional 10% increase in 194 canopy cover, the odds of selection increased by 126% (95% CI: 55-230%). Use was greater 195 than availability at all canopy cover levels >19%. 196 197 4. Discussion 198 Male northern long-eared bats primarily selected roosts in trees with characteristics that 199 promote cavity formation. At the level of individual trees, bats selected for large-diameter trees 9 200 with substantial decay. This corroborates previous work on northern long-eared bats (Jung et al., 201 2004; Rojas et al., 2017) and is intuitive because large trees with more decay have more roost 202 structures (i.e., cavities and loose bark) for bats to use (Reynolds et al., 1985). This is particularly 203 true of ponderosa pines, which can produce large amounts of resin to defend against physical 204 injury (Kane and Kolb, 2010; Lewinsohn et al., 1991) and therefore tend to develop cavities only 205 when they are scarred or dead. In intensively managed landscapes like the Black Hills, cavities 206 are found overwhelmingly in snags because most trees are harvested before they reach ages at 207 which cavities typically form. 208 Conservation actions targeting male northern long-eared bats should include preservation 209 of large snags whenever possible. Our study demonstrated that male northern long-eared bats 210 select large-diameter snags (>37 cm), and large-diameter snags also tend to remain standing 211 longer than thinner snags (Bull, 1983; Chambers and Mast, 2014). These large-diameter snags 212 need not be tall—short (≤ 3 m) snags are important resources for male northern long-eared bats 213 as well. Seventeen of 43 (39.5%) roosts that we located occurred in broken-off snags ≤ 3 m in 214 height. These are important resources and are likely more vulnerable to loss during forest 215 management activities (particularly prescribed fire) than other potential roost trees. Snags are 216 often intentionally removed during forest management activities because of hazards posed to 217 forest management personnel (e.g., loggers and firefighters) and the general public. However, 218 these short snags pose less danger to forest management personnel and the public than taller 219 snags, and their preservation is therefore a realistic and actionable step toward bat conservation. 220 Of the variables we considered that may influence thermal characteristics of roosts, only 221 canopy cover influenced roost selection significantly. Trees were more likely to be used as roosts 222 as surrounding canopy cover increased, and use was greater than availability at all canopy cover 10 223 levels >19%. Although many snags were available within our study area in open areas burned by 224 a severe wildfire in 2000, bats in our study rarely used those snags, instead selecting snags in the 225 interior of forest stands with live canopy. Forty out of 43 (93.0%) roosts were within intact forest 226 stands with live canopy, and all roosts were within 50 m of intact forest stands. Bats may prefer 227 these areas because canopy cover creates cooler environments, but they may also simply prefer 228 to be immediately near forested areas where they forage (Henderson and Broders, 2008; Owen et 229 al., 2003; Patriquin and Barclay, 2003). Either way, stand-replacing fire likely poses risks to 230 local populations of northern long-eared bats at the western edge of its range, where severe 231 wildfire is increasingly prevalent due to climate change (Westerling et al., 2006). Clearcutting 232 also poses risks to local populations of northern long-eared bats in these areas, even if snags are 233 retained. Selective logging that leaves some level of canopy cover remaining would ensure that 234 snag retention is effective for bat roost conservation. 235 Dynamics of regional disturbance may be important when evaluating local-scale factors 236 that influence roost selection (O’Keefe and Loeb, 2017). The ponderosa-dominated landscape 237 where we conducted our research is substantially different than other landscapes (i.e., deciduous 238 and mixed forests in eastern North America) where roost selection by northern long-eared bats 239 has been studied. Although many of the factors driving roost selection appear to be similar 240 among areas, the processes that create roosts may be fundamentally different in different areas. 241 Snags in ponderosa pine forests are often generated in large pulses by severe wildfire and 242 mountain pine beetles (Dendroctonus ponderosae), but the long-term ramifications of these 243 resource pulses for bats are not well understood. Severe wildfire appears to create snags that are 244 largely unused by bats. Mountain pine beetle outbreaks may do the same if beetle-induced 245 mortality reduces or eliminates canopy cover over large areas, or if outbreaks lead to more severe 11 246 fires. Bats may instead depend on snag-generating processes that operate at more local scales and 247 over longer intervals to create suitable roosts. 248 Roost selection by bats varies by sex, age class, and reproductive condition (Elmore et 249 al., 2004; Hein et al., 2008). Studies on roost selection generally focus on females because they 250 tend to drive reproduction, which is required to sustain populations. However, targeting roost 251 conservation toward females exclusively may neglect resources that are important for males. 252 Because sex ratios can be heavily biased in some areas (Cryan et al., 2000), ignoring the needs of 253 males could leave resources that are important for most individuals inhabiting these areas 254 unprotected. On the other hand, designing roost conservation measures on studies of males alone 255 will leave resources that are important for females unprotected. For example, short (≤ 3 m) snags 256 are important resources for males, but they may not be for females, which aggregate in maternity 257 colonies that may contain over one hundred individuals and require larger cavities than largely 258 solitary males (Perry and Thill, 2007). Resource managers seeking to conserve bats should take 259 these sex differences into account when developing conservation plans and designing studies to 260 inform those plans. In high-elevation areas, males may be more important than females for 261 sustaining local populations because there are few females in those areas. 262 263 5. Conclusions 264 Forest managers require actionable knowledge to guide conservation, and our results 265 indicate that conserving large-diameter snags within intact forest stands is one such action that 266 can be taken to conserve male northern long-eared bats in wildfire-prone coniferous forests. 267 Short (≤ 3 m) snags in particular represent a low-risk, high-reward resource to target for 268 preservation in male-biased, high-elevation populations of this species. For federally threatened 12 269 northern long-eared bats, conserving these snags at the western edge of their range may prevent 270 range contraction and local extinction. Similar patterns may hold true for other cavity-roosting 271 bat species in wildfire-prone coniferous forests, like those found throughout western North 272 America. Further study on roost selection by male bats represents an underappreciated 273 conservation research opportunity that may be particularly valuable for high-elevation bat 274 populations. Although bats face danger from many threats unrelated to roosts (e.g., white nose 275 syndrome, wind energy development, etc.), roost conservation remains an important tool for bat 276 conservation in the face of such threats. 277 278 Acknowledgements 279 Many thanks to L. Boodoo, C. McFarland, E. Greene, B. Tabor, and B. Phillips for help with 280 fieldwork; J. Rick for helpful comments on pre-submission versions of this manuscript; R. 281 Anderson-Sprecher for helpful comments concerning statistical analyses; and P. Ortegon, D. 282 Licht, M. Wiles, D. Austin, B. Phillips, and E. Thomas for their logistical support of this project. 283 Research funding was provided by the National Park Service, the Department of Zoology and 284 Physiology at the University of Wyoming, the Berry Ecology Center, the American Society of 285 Mammalogists, Prairie Biotic Research, Inc., and the Wyoming Chapter of The Wildlife Society. 286 The findings and conclusions in this article are those of the authors and do not necessarily 287 represent the views of the U.S. Fish and Wildlife Service or the National Park Service. 288 289 Appendix A: Supplementary Data 290 Table A.1. A priori rationales for including variables of interest in the global model. 291 13 292 Table A.2. Candidate models, ΔAIC values, and model weights (wi) used to determine model- 293 averaged coefficients. 294 295 Fig. A.1. Density plots of significant variables in the averaged model. Use was generally great 296 than availability at all decay classes > 2, and greater than availability for all DBHs >37 and all 297 canopy cover levels >19%. 298 299 Data Availability 300 Data and R code used in analysis have been archived on Zenodo. They can be located using the 301 following link: https://zenodo.org/record/2727206#.XNY-iKR7k2w. 302 303 Literature Cited 304 Aldridge, H.D.J.N., Brigham, R.M., 1988. Load carrying and maneuverability in an 305 insectivorous bat: a test of the 5% “rule” of radio-telemetry. Journal of Mammalogy 69, 306 379–382. https://doi.org/10.2307/1381393 307 Barclay, R.M.R., 1991. Population structure of temperate zone insectivorous bats in relation to 308 foraging behaviour and energy demand. Journal of Animal Ecology 60, 165–178. 309 https://doi.org/10.2307/5452 310 Barton, K., 2018. MuMIn: multi-model inference. 311 Beyer, H.L., Haydon Daniel T., Morales Juan M., Frair Jacqueline L., Hebblewhite Mark, 312 Mitchell Michael, Matthiopoulos Jason, 2010. The interpretation of habitat preference 313 metrics under use–availability designs. Philosophical Transactions of the Royal Society 314 B: Biological Sciences 365, 2245–2254. https://doi.org/10.1098/rstb.2010.0083 14 315 Boland, J.L., Hayes, J.P., Smith, W.P., Huso, M.M., 2009. Selection of day-roosts by Keen’s 316 myotis (Myotis keenii) at multiple spatial scales. Journal of Mammalogy 90, 222–234. 317 https://doi.org/10.1644/07-MAMM-A-369.1 318 Brigham, R.M., Fenton, M.B., 1986. The influence of roost closure on the roosting and foraging 319 behaviour of Eptesicus fuscus (Chiroptera: Vespertilionidae). Canadian Journal of 320 Zoology 64, 1128–1133. https://doi.org/10.1139/z86-169 321 Bull, E.L., 1983. Longevity of snags and their use by woodpeckers (General Technical Report 322 No. GTR-RM-99), Proceedings of the Symposium: Snag Habitat Management. Rocky 323 Mountain Research Station, Forest Service, US Department of Agriculture, Fort Collins, 324 CO. 325 Burnham, K.P., Anderson, D.R., 2002. Model selection and multimodel inference: a practical 326 information-theoretic approach, 2nd ed. ed. Springer, New York. 327 Caceres, M.C., Barclay, R.M.R., 2000. Myotis septentrionalis. Mammalian Species 1–4. 328 https://doi.org/10.1644/1545-1410(2000)634<0001:MS>2.0.CO;2 329 Chambers, C.L., Mast, J.N., 2014. Snag dynamics and cavity excavation after bark beetle 330 outbreaks in southwestern ponderosa pine forests. Forest Science 60, 713–723. 331 https://doi.org/10.5849/forsci.13-018 332 Choate, J.R., Anderson, J.M., 1997. The bats of Jewel Cave National Monument, South Dakota. 333 The Prairie Naturalist 29, 38–47. 334 Cryan, P.M., Bogan, M.A., Altenbach, J.S., 2000. Effect of elevation on distribution of female 335 bats in the Black Hills, South Dakota. Journal of Mammalogy 81, 719–725. 336 https://doi.org/10.1644/1545-1542(2000)081<0719:EOEODO>2.3.CO;2 15 337 Dobson, A.P., Bradshaw, A.D., Baker, A.J.M., 1997. Hopes for the future: restoration ecology 338 and conservation biology. Science 277, 515–522. 339 https://doi.org/10.1126/science.277.5325.515 340 Elmore, L.W., Miller, D.A., Vilella, F.J., 2004. Selection of diurnal roosts by red bats (Lasiurus 341 borealis) in an intensively managed pine forest in Mississippi. Forest Ecology and 342 Management 199, 11–20. https://doi.org/10.1016/j.foreco.2004.03.045 343 Fenton, M.B., Rautenbach, I.L., Smith, S.E., Swanepoel, C.M., Grosell, J., van Jaarsveld, J., 344 1994. Raptors and bats: threats and opportunities. Animal Behaviour 48, 9–18. 345 https://doi.org/10.1006/anbe.1994.1207 346 Ford, W.M., Menzel, M.A., Menzel, J.M., Welch, D.J., 2002. Influence of summer temperature 347 on sex ratios in eastern red bats (Lasiurus borealis). The American Midland Naturalist 348 147, 179–184. 349 Halpern, B.S., Walbridge, S., Selkoe, K.A., Kappel, C.V., Micheli, F., D’Agrosa, C., Bruno, J.F., 350 Casey, K.S., Ebert, C., Fox, H.E., Fujita, R., Heinemann, D., Lenihan, H.S., Madin, 351 E.M.P., Perry, M.T., Selig, E.R., Spalding, M., Steneck, R., Watson, R., 2008. A global 352 map of human impact on marine ecosystems. Science 319, 948–952. 353 https://doi.org/10.1126/science.1149345 354 Hansen, M.C., Potapov, P.V., Moore, R., Hancher, M., Turubanova, S.A., Tyukavina, A., Thau, 355 D., Stehman, S.V., Goetz, S.J., Loveland, T.R., Kommareddy, A., Egorov, A., Chini, L., 356 Justice, C.O., Townshend, J.R.G., 2013. High-resolution global maps of 21st-century 357 forest cover change. Science 342, 850–853. https://doi.org/10.1126/science.1244693 16 358 Hein, C.D., Castleberry, S.B., Miller, K.V., 2008. Sex-specific summer roost-site selection by 359 Seminole bats in response to landscape-level forest management. Journal of Mammalogy 360 89, 964–972. https://doi.org/10.1644/07-MAMM-A-335.1 361 Henderson, L.E., Broders, H.G., 2008. Movements and resource selection of the northern long- 362 eared myotis (Myotis septentrionalis) in a forest—agriculture landscape. Journal of 363 Mammalogy 89, 952–963. https://doi.org/10.1644/07-MAMM-A-214.1 364 Humphrey, S.R., 1975. Nursery roosts and community diversity of Nearctic bats. Journal of 365 Mammalogy 56, 321–346. https://doi.org/10.2307/1379364 366 Johnson, C.J., Nielsen, S.E., Merrill, E.H., McDonald, T.L., Boyce, M.S., 2006. Resource 367 selection functions based on use-availability data: theoretical motivation and evaluation 368 methods. The Journal of Wildlife Management 70, 347–357. 369 https://doi.org/10.2193/0022-541X(2006)70[347:RSFBOU]2.0.CO;2 370 Jones, K.R., Venter, O., Fuller, R.A., Allan, J.R., Maxwell, S.L., Negret, P.J., Watson, J.E.M., 371 2018. One-third of global protected land is under intense human pressure. Science 360, 372 788–791. https://doi.org/10.1126/science.aap9565 373 Jung, T.S., Thompson, I.D., Titman, R.D., 2004. Roost site selection by forest-dwelling male 374 Myotis in central Ontario, Canada. Forest Ecology and Management 202, 325–335. 375 https://doi.org/10.1016/j.foreco.2004.07.043 376 Kane, J.M., Kolb, T.E., 2010. Importance of resin ducts in reducing ponderosa pine mortality 377 from bark beetle attack. Oecologia 164, 601–609. https://doi.org/10.1007/s00442-010- 378 1683-4 379 Kremen, C., Merenlender, A.M., 2018. Landscapes that work for biodiversity and people. 380 Science 362, eaau6020. https://doi.org/10.1126/science.aau6020 17 381 Kunz, T.H., 1982. Roosting ecology of bats, in: Kunz, T.H. (Ed.), Ecology of Bats. Springer US, 382 Boston, MA, pp. 1–55. https://doi.org/10.1007/978-1-4613-3421-7_1 383 Kutner, M.H., Nachtsheim, C.J., Neter, J., Li, W., 2005. Applied linear statistical models, 5th ed. 384 ed, The McGraw-Hill/Irwin series operations and decision sciences. McGraw-Hill Irwin, 385 Boston. 386 Lacki, M.J., Cox, D.R., Dickinson, M.B., 2009. Meta-analysis of summer roosting characteristics 387 of two species of Myotis bats. The American Midland Naturalist 162, 318–326. 388 Lewinsohn, E., Gijzen, M., Croteau, R., 1991. Defense mechanisms of conifers: differences in 389 constitutive and wound-induced monoterpene biosynthesis among species. Plant 390 Physiology 96, 44–49. https://doi.org/10.1104/pp.96.1.44 391 Manel, S., Williams, H.C., Ormerod, S.J., 2001. Evaluating presence–absence models in 392 ecology: the need to account for prevalence. Journal of Applied Ecology 38, 921–931. 393 https://doi.org/10.1046/j.1365-2664.2001.00647.x 394 Manly, B.F.L., McDonald, L., Thomas, D.L., McDonald, T.L., Erickson, W.P., 2007. Resource 395 selection by animals: statistical design and analysis for field studies. Springer Science & 396 Business Media. 397 O’Keefe, J.M., Loeb, S.C., 2017. Indiana bats roost in ephemeral, fire-dependent pine snags in 398 the southern Appalachian Mountains, USA. Forest Ecology and Management 391, 264– 399 274. https://doi.org/10.1016/j.foreco.2017.01.036 400 Owen, S.F., Menzel, M.A., Ford, W.M., Chapman, B.R., Miller, K.V., Edwards, J.W., Wood, 401 P.B., 2003. Home-range size and habitat used by the northern myotis (Myotis 402 septentrionalis). The American Midland Naturalist 150, 352–359. 403 https://doi.org/10.1674/0003-0031(2003)150[0352:HSAHUB]2.0.CO;2 18 404 Patriquin, K.J., Barclay, R.M.R., 2003. Foraging by bats in cleared, thinned and unharvested 405 boreal forest. Journal of Applied Ecology 40, 646–657. https://doi.org/10.1046/j.1365- 406 2664.2003.00831.x 407 Perry, R.W., Thill, R.E., 2007. Roost selection by male and female northern long-eared bats in a 408 pine-dominated landscape. Forest Ecology and Management 247, 220–226. 409 https://doi.org/10.1016/j.foreco.2007.04.041 410 R Core Team, 2018. R: A language and environment for statistical computing. R Foundation for 411 Statistical Computing, Vienna, Austria. 412 Reynolds, R.T., Linkhart, B.D., Jeanson, J., 1985. Characteristics of snags and trees containing 413 cavities in a Colorado conifer forest (USDA Forest Service Research Note No. RM-455). 414 Rocky Mountain Forest and Range Experiment Station, Fort Collins, CO. 415 Rojas, V.G., O’Keefe, J.M., Loeb, S.C., 2017. Baseline capture rates and roosting habits of 416 Myotis septentrionalis (northern long-eared bat) prior to White-Nose Syndrome detection 417 in the Southern Appalachians. Southeastern Naturalist 16, 140–148. 418 https://doi.org/10.1656/058.016.0202 419 Senior, P., Butlin, R.K., Altringham, J.D., 2005. Sex and segregation in temperate bats. 420 Proceedings of the Royal Society B: Biological Sciences 272, 2467–2473. 421 https://doi.org/10.1098/rspb.2005.3237 422 Sikes, R.S., and the Animal Care and Use Committee of the American Society of Mammalogists, 423 2016. 2016 guidelines of the American Society of Mammalogists for the use of wild 424 mammals in research and education. Journal of Mammalogy 97, 663–688. 425 https://doi.org/10.1093/jmammal/gyw078 19 426 Solick, D.I., Barclay, R.M.R., 2006. Thermoregulation and roosting behaviour of reproductive 427 and nonreproductive female western long-eared bats (Myotis evotis) in the Rocky 428 Mountains of Alberta. Canadian Journal of Zoology 84, 589–599. 429 https://doi.org/10.1139/z06-028 430 Swets, J.A., 1988. Measuring the accuracy of diagnostic systems. Science 240, 1285–1293. 431 https://doi.org/10.1126/science.3287615 432 Vittinghoff, E., McCulloch, C.E., 2007. Relaxing the rule of ten events per variable in logistic 433 and Cox regression. American Journal of Epidemiology 165, 710–718. 434 https://doi.org/10.1093/aje/kwk052 435 Westerling, A.L., Hidalgo, H.G., Cayan, D.R., Swetnam, T.W., 2006. Warming and earlier 436 spring increase western U.S. forest wildfire activity. Science 313, 940–943. 437 https://doi.org/10.1126/science.1128834 438 Western Regional Climate Center, 2018. NCDC 1981-2010 Monthly Normals: Custer, SD. URL 439 https://wrcc.dri.edu/cgi-bin/cliMAIN.pl?sd2087 440 Willis, C.K.R., Brigham, R.M., 2004. Roost switching, roost sharing and social cohesion: forest- 441 dwelling big brown bats, Eptesicus fuscus, conform to the fission–fusion model. Animal 442 Behaviour 68, 495–505. https://doi.org/10.1016/j.anbehav.2003.08.028 443 20 444 Figure Legends 445 Fig. 1. Unscaled odds ratios associated with each variable in the averaged roost selection model. 446 Error bars represent 95% confidence intervals. 447 21 448 Table 1. Variables measured at used and available summer day-roosts of male northern long-eared bats (Myotis septentrionalis) in the 449 Black Hills of South Dakota, 2017–2018. Variable Definition DBH Tree diameter at breast height (cm); measured with a diameter tape Height Tree height (m); measured with an electronic clinometer Snag Tree status (live/dead) Decay Class Stage of tree decay on ordinal scale from 1-9; higher values denote more decay (sensu Maser et al., 1979) Bark Remaining Bark remaining on tree trunk (%); estimated visually Canopy Cover Average of 4 canopy cover measurements (N/E/S/W) taken 5 m from tree (%); measured with a convex spherical densiometer Slope Slope of 706.9-m2 (15-m radius) plot centered at tree (%); measured with an electronic clinometer Tree Density Number of live trees in 706.9-m2 plot centered at tree Snag Density Number of snags in 706.9-m2 plot centered at tree Eastness Difference between aspect of 706.9-m2 plot centered at tree and 90 degrees (˚); measured with a compass Southness Difference between aspect of 706.9-m2 plot centered at tree and 180 degrees (˚); measured with a compass Slope*Eastness Interaction term between slope and eastness Slope*Southness Interaction term between slope and southness 450 22 451 Table 2. Coefficient estimates in the averaged model and 95% confidence intervals. Bold variables denote significance at α = 0.05. Variable Estimate LCL (95%) UCL (95%) Height 0.0133 -0.0767 0.1033 DBH 0.0948 0.0382 0.1514 Decay Class 0.7465 0.3835 1.1094 Bark Remaining 0.0033 -0.0113 0.0180 Snag Density 0.1010 -0.0039 0.2059 Tree Density -0.0182 -0.0653 0.0289 Canopy Cover 0.0816 0.0438 0.1195 Slope 0.0323 -0.0354 0.0999 Eastness -0.0069 -0.0207 0.0068 Southness 0.0004 -0.0041 0.0050 Slope*Eastness 0.0001 -0.0004 0.0005 Slope*Southness 0.0000 -0.0002 0.0002 452 453 23 454 Table 3. Averaged model coefficients, scaled and unscaled odds ratios (OR), and scaled lower and upper confidence limits 455 (UCL/LCL) for significant variables. Variable Coefficient Unscaled OR Scaled OR Units Scaled OR LCL (95%) Scaled OR UCL (95%) DBH 0.0948 1.0995 1.6065 5 cm 1.2105 2.1321 Decay Class 0.7465 2.1095 2.1095 1 unit 1.4674 3.0327 Canopy Cover 0.0816 1.0850 2.2619 10% 1.5491 3.3025 456 24 457 Table 4. Means and standard errors for variables of interest among used and available trees. 458 Bold font denotes statistically significant variables in the final averaged model. Roost Available Variable Mean SE Mean SE Height (m) 8.53 1.11 9.01 0.43 DBH (cm) 35.69 1.57 30.33 0.69 Decay Class 4.95 0.33 3.72 0.18 Bark Remaining (%) 74.19 4.22 69.73 2.49 Snag Density 4.74 1.03 2.12 0.23 Tree Density 19.84 2.15 10.76 1.12 Canopy Cover (%) 36.83 3.02 14.96 1.39 Slope (%) 16.87 1.62 11.66 0.64 Eastness (˚) 76.36 8.21 93.35 3.81 Southness (˚) 109.48 11.14 96.58 5.48 459 25