Unsuspected retreats: autumn transitional roosts and presumed winter hibernacula of little brown myotis in Colorado

Abstract

Documentation of autumn and winter roosts of many species of hibernating bats are lacking from western North America. However, recent evidence suggests that rather than using caves and mines, many individuals and some species of bats may roost in inconspicuous rock crevices at these times of year. I investigated autumn use of rock crevices and other roosts by the little brown myotis (Myotis lucifugus) in the Rocky Mountains of western Colorado through radiotelemetry (n = 38). Objectives were to determine the types and characteristics of roosts, describe patterns of movements to these roosts from summer colonies, and contrast findings with results of surveys of bats in caves and abandoned mines in Colorado during autumn and winter. Forty-four autumn transitional roosts and presumed hibernacula were located in buildings, trees, and rock crevices. Bats used short-distance movements changing in elevation to autumn transitional roosts and presumed hibernacula rather than major latitudinal migrations. Roost type and distance from capture site to roosts had the highest variable importance at the landscape scale. Microclimate comparisons showed that buildings provided warmer minimum average temperatures, which may benefit juvenile bats early in the transition season. Tree roost temperatures during autumn would allow bats to conserve energy by using daily torpor and passive rewarming to assist with afternoon arousals. Rock crevice roosts in talus were found to be suitable for hibernation by exhibiting the coolest average temperatures and maintaining the highest relative humidity levels. Autumn access and spring egress to high-elevation talus sites used by these bats were not obstructed by winter snow pack. These rock crevices also provided temperature and humidity levels that would support the persistence and growth of Pseudogymnoascus destructans (Pd), the causal agent of white-nose syndrome. However, bats in this study appeared to roost alone, which could inhibit the bat-to-bat spread of Pd. Surveys of caves and mines within and surrounding the study area revealed few hibernating little brown myotis, suggesting that most individuals in Colorado may instead utilize rock crevices as roosts during winter.

caves, hibernacula, microclimate, Myotis lucifugus, rock crevices, roost selection, talus, transitional roosts

Conservation of hibernating bats in North America has become more complex with the onset and subsequent spread of the fungal disease white-nose syndrome (WNS—Langwig et al. 2015; Hammerson et al. 2017). WNS has impacted little brown myotis (Myotis lucifugus) more than any other species in the eastern United States (Langwig et al. 2015), leading to a proposal for federal listing under the Endangered Species Act (ESA, 16 U.S.C. §§ 1531–1544—Kunz and Reichard 2010) in the United States, and an endangered listing in Canada under the federal Species at Risk Act (SC 2002, c. 29). Concerns related to possible population declines of this species from WNS have arisen in western North America as the epizootic progresses westward, making identification of autumn transitional roosts and winter hibernacula a priority. Efforts to implement proactive management actions for WNS in western North America have been hindered by a lack of basic knowledge about the natural history of western populations of little brown myotis, particularly where they roost in autumn and winter. Differences in topography, resource availability, and how landscape components affect where bats hibernate become apparent when comparing studies conducted in eastern and western portions of the continent. For example, use of caves and mines as hibernacula by little brown myotis in eastern North America has been well described (e.g., Hitchcock 1949; Fenton and Barclay 1980), with bats migrating several hundred kilometers between summer and winter roosts (Griffin 1945; Norquay et al. 2013). In this part of their range, little brown myotis show high site fidelity and form winter colonies that often number over 1,000 individuals (Davis and Hitchcock 1965; Humphrey and Cope 1976; Thomas et al. 1979).

Efforts to identify use of caves and mines as bat hibernacula in western North America have long yielded different findings, with little evidence of large overwintering aggregations of little brown myotis despite surveys of thousands of such sites (e.g., Twente 1960; Perkins et al. 1990; Nagorsen et al. 1993; Priday and Luce 1997; Hendricks 2012). Twente (1960:70) attributed the seeming inadequacy of these sites for bat hibernation to microclimates being too warm or cold and postulated that bats “may hibernate underground or in deep crevices in cliffs which remain cold but above freezing.” Perkins et al. (1990:62) reiterated this conclusion decades later stating that bats were more likely to “hibernate in sites not readily accessible to examination by humans.” When bats were noted at caves and mines, they were predominantly accounted for by 1 species, the Townsend’s big-eared bat (Corynorhinus townsendii). The absence of most western bat species from these sites has led management efforts, such as those for WNS, to reference eastern studies where use of caves and mines by bats in winter is more common.

Although many caves and mines are available for use by bats in western North America, the prevalence of other possible roosts, such as rock crevices, may explain the difference in abundance of bats at these sites in comparison to eastern North America. A growing body of literature indicates that some vespertilionid bats use rock crevices as autumn transitional and winter roosts in western North America, including big brown bats on the eastern slope of the Rocky Mountains in Colorado (Neubaum et al. 2006) and southeastern Alberta (Lausen and Barclay 2006; Klüg-Baerwald et al. 2017); western small-footed myotis (Myotis ciliolabrum) in the badlands of North Dakota (Barnhart and Gillam 2017); little brown myotis, western long-eared myotis (Myotis evotis), western small-footed myotis, and big brown bats in Yellowstone National Park (Johnson et al. 2017); and northern long-eared myotis (Myotis septentrionalis), tricolored bats (Perimyotis subflavus), and big brown bats in Nebraska (Lemen et al. 2016). Additionally, in eastern North America, little brown myotis and northern long-eared myotis have been found using rock crevices, trees, and buildings as roosts in autumn (Lowe 2012) and the eastern small-footed myotis (Myotis leibii) and big brown bats were noted using talus slopes during spring, autumn, and winter (Moosman et al. 2015, 2017). Use of rock crevices by bats during winter in Norway was also noted by Michaelsen et al. (2013), who found 3 species roosting in crevices within rock walls and scree.

The primary objectives of this study were to determine seasonal movement patterns of little brown myotis in Colorado during the autumn transition to the winter hibernation period, and to characterize roost use in terms of microclimate and landscape context using models. I also examined records from winter bat surveys of caves and mines in Colorado to determine the extent of use of these sites by little brown myotis. Colorado provides a good proxy for western North America given its diverse range of ecosystems that offer many roosting opportunities for bats. I predicted that little brown myotis would use rock crevices as transitional roosts and potential hibernacula in the study area, which would explain the absence of large aggregations of bats typically noted in caves and mines here during these periods. Further, roost microclimate information from these sites is particularly pertinent to determining the potential for survival and persistence of Pseudogymnoascus destructans (Pd), the causal agent of WNS (Verant et al. 2012) and cause of large-scale declines in eastern bat populations. Where bats roost in autumn and winter, and whether they aggregate or roost alone, may influence the potential establishment of WNS, as well as its rate and extent of spread in western populations (Neubaum et al. 2017).

Materials and Methods

Study area

I investigated characteristics of autumn transitional roosts and presumed hibernacula along the Crystal River Valley of west-central Colorado from 2011 to 2014 (Fig. 1). The region is characterized by high peaks and deep drainages cut by rivers and streams. Elevations range from 1,890 m along the Crystal River to 3,950 m near the summit of Mount Sopris directly adjacent to the river valley. The weather station located in the Crystal River Valley, at 2,038 m elevation, recorded a mean annual temperature of 8°C (ranging from −26°C to 36°C) from 2009 to 2017. Mean annual precipitation in the study areas totals 233 cm with accumulations occurring primarily in the form of snow (195 cm) from October through May (Weather Underground, https://www.wunderground.com/personal-weather-station/dashboard?ID=KCOCARBO1, accessed 7 March 2018). Much like other valleys in the Rocky Mountain Region, the weather patterns for the Crystal River Valley consist of severe winters from October until April and months free from hard frosts between May and September. The valley floor, which includes the riparian corridor along the Crystal River, has largely been converted for agricultural purposes. The landscape quickly transitions into the foothills before rising steeply up the flanks of large peaks. Trees include cottonwood (Populus sp.), pinyon pine (Pinus edulis), and juniper (Juniperus sp.) at lower elevations (1,890–2,600 m). Douglas-fir (Pseudotsuga menziesii) and aspen (Populus tremuloides) are common at elevations above 2,400 m before transitioning into Engelmann’s spruce (Picea engelmannii) and subalpine fir (Abies lasiocarpa) above 2,600 m.

Fig. 1.

Locations of autumn roosts (buildings, rock crevices, trees) and capture locations used by little brown myotis (Myotis lucifugus) in the Crystal River watershed, west-central Colorado, 2012–2013. Due to the close proximity of some roosts, location markers for roost types may overlap.

Autumn capture and radiotelemetry

Exit counts and internal surveys conducted at a building in 2011 were used to document the first week of August as the time of initial disbanding for a little brown myotis maternity colony. I captured little brown myotis at this maternity roost, and later sampled transitional and night roosts, ranging in elevations from 1,950 to 2,100 m, along the Crystal River Valley during late summer and early autumn. Bats were captured from buildings as they emerged in the evening or over open water by using a variety of standard techniques, including mist nets, H-nets, harp traps, and insect hoop nets (Waldien and Hayes 1999; Kunz and Parsons 2009). Sex, reproductive status, body mass (g), and forearm measurements (mm) were recorded for each bat. Body mass was determined using portable scales (Ohaus HH120D Portable Hand-held Pocket Portable Scale, Ohaus Corporation, Parsippany, New Jersey). Bats were classified as adults based on epiphyseal fusion in the phalanges (Kunz and Parsons 2009). Reproductive status of female bats was categorized as pregnant, lactating, or postlactating by examination of nipple morphology (Kunz and Parsons 2009).

Study methods conformed to the Colorado Parks and Wildlife (CPW) Capture and Handling Guidelines for Bats (Neubaum and Jackson 2015, ACUC # 04-2016), which were based on guidelines of the American Society of Mammalogists (Sikes et al. 2016) and American Veterinary Medical Association (AVMA 2013). Capture and handling work was conducted under CPW scientific collection license CPW004. Radiotransmitters (0.3–0.5 g, model A2414 or A2415, Advanced Telemetry Systems, Isanti, Minnesota, and 0.5–0.6 g, model LB-2X, Holohil Systems Ltd, Carp, Ontario, Canada) were attached to bats on the dorsal skin between the scapulae using surgical cement (Perma-Type Company, Inc., Plainville, Connecticut). As suggested by Aldridge and Brigham (1988), the relationship between transmitter mass and body mass followed a “5% rule,” which has been used previously on another Colorado bat species with no major adverse long-term effects (Neubaum et al. 2005). I tagged female and male bats to investigate if variation of use in seasonal resources occurs so that conservation needs of both sexes could be addressed appropriately (Weller et al. 2009). Radiotagged bats were tracked to determine locations of autumn transitional roosts, movement patterns during the transition season, and characteristics of individual roost sites. I initially searched for radio signals using a roof-mounted 5-element yagi or whip antenna on vehicles. Once a signal was detected from the road, the exact location was determined on foot using a handheld 3-element antenna. In 2012 and 2013, 5 flights in fixed-wing aircraft were taken over the Roaring Fork and Crystal River Valleys in an attempt to locate signals that had not been detected from the ground. Pulses per minute were collected for temperature sensitive ATS transmitters in 2012 at the time the bat was located and surface body temperatures derived from calibration curves. These surface body temperatures helped determine if bats were torpid, a condition, when coupled with date-of-the-year, that is indicative of transitional roosts and presumed hibernacula.

Roost characterization

Once roosts were identified, locations were georeferenced (NAD 83, Zone 13) using a handheld GPS and imported to a geographic information system (ArcMAP 10.3, Esri, Redlands, California). A landscape analysis was used to determine if resources used by little brown myotis were in proportion to their availability on the landscape and how their proximity to summer roosts compares to those of random locations within the study area. For comparison at the landscape scale, only used sites with unique values were considered. If more than 1 individual used any part of the same roost, all values except distance from capture location to roost were the same. Consequently, data for only 1 individual was used in the analysis to avoid pseudoreplication, but with the distance from capture location to roost value averaged across all individuals that used the site for the 1 data entry. A conservative ratio of 5 available locations (i.e., potential roost sites selected at random) to 1 use location (i.e., known roost sites) was implemented. Random locations were generated in ArcGIS using the Create Random Points algorithm within the full extent of known roosts buffered by an additional 4 km. Other studies found little brown myotis averaged 2.5–5.9 km between capture sites and autumn transitional roosts, and ≤ 4 km between the roosts themselves (Lowe 2012; Johnson et al. 2017).

Landscape variables measured for known roosts and randomly selected sites included roost type (building, tree, rock crevice), elevation (m), aspect of the hillside (converted to sine and cosine radians), shade levels, distance from capture location to first roost (m), distance to other similarly categorized roosts (m; i.e., used to used or random to random) to provide a measure of spatial aggregation, distance to perennial rivers and streams (m), and a solar radiation measure that incorporates elevation, aspect of the hillside, and shade levels in 1 value to examine the landscapes influence on microclimate (Watt hours/m² summed). These variables or components of them were previously found to be ecologically important in determining selection of autumn transitional roosts previously characterized at the landscape scale for bats in North America (Neubaum et al. 2006; Lowe 2012; Johnson et al. 2017; Klug-Baerwald et al. 2017). High-resolution imagery (0.3 m; World Imagery, Esri, Redlands, California) was used to determine roost type of random locations. Distances were measured using the Point Distance tool, shade using the Hillshade tool, and solar radiation values generated for each location during the autumn window when bats were being tracked using the Solar Radiation tool in ArcMap.

Effort needed to track bats during this season and over difficult terrain along with logistical restraints related to staff time and equipment prohibited a full microhabitat-scale analysis. However, microclimate readings and snow depth measurements were collected at a subset of transitional roosts and presumed hibernacula to determine their thermal suitability for use by bats (Humphries et al. 2002) and to examine potential for persistence of Pd at these sites (Verant et al. 2012). Transitional roosts were categorized as sites used after the maternity colony had disbanded in August and temperatures allowed for torpor early in the day as well as passive warming later in the day. Transitional roosts in trees were often shallow enough that tagged bats could be seen while roosting. In contrast, presumed hibernacula were categorized as sites where internal ambient temperatures remained cool and constant to allow for extended torpor bouts (Humphries et al. 2002). Hibernacula temperatures generally did not allow for passive warming or freezing and were deeper, with bats not visible (Neubaum et al. 2006; Klug-Baerwald et al. 2017).

Internal ambient temperature and relative humidity were collected for a subsample of known roosts representing all roost types with recordings made every 3 h using Hygrochron iButtons (Hygrochron Temperature-Humidity Logger iButton, model DS1923, Maxim Integrated, San Jose, California) housed in plastic tubing to inhibit moisture and ultrasound emissions (Willis et al. 2009). Internal microclimate data were used to characterize conditions near the areas where bats were assumed to roost and to assess the roost’s potential to harbor WNS (Verant et al. 2012). Microclimate has consistently been shown to be one of the most influential variables in determining roost use by bats at the microhabitat level (Neubaum et al. 2006; Klug-Baerwald et al. 2017).

Snow depth was recorded at a subset of high-elevation rock crevice roosts located in talus slopes categorized as presumed hibernacula to investigate if access to and egress from the sites by bats is limited by seasonal snow cover and to examine its influence on roost microclimate. Access and egress dates from hibernacula can influence survival and reproductive success of bats (Norquay and Willis 2014; Meyer et al. 2016). In addition, snow depth at high-elevation talus has been shown to insulate underground refugia for insects using talus (Millar et al. 2015; Schoville et al. 2015) and may influence the microclimate within these presumed hibernacula as well. Three remote trail cameras (Hyperfire HC600, RECONYX, Holmen, Wisconsin) using time lapse settings took daily photographs, starting in October 2013 and ending a full calendar year later in 2014, to record approximate snow depth from 3-m poles marked with 30-cm increments placed at the roost locations. The latest autumn date when rocks were still visible in the photographs was recorded as an estimate of roost accessibility. The greatest snow depth was recorded as the highest point snow reached on the marker pole. Egress from the roost was noted as the earliest date that rocks reappeared from beneath snow cover. As with roost access, egress from the roost site may occur slightly earlier or later than this date but I believe these photos provide a good estimate of these dates.

Cave and mine surveys

Between 2010 and 2017 CPW, the Colorado Natural Heritage Program, the Colorado River Field Office of the Bureau of Land Management, and the White River National Forest conducted winter season internal cave and mine surveys in and surrounding the study area. In addition, the CPW Bat and Abandoned Mine databases, scientific collection reports, and museum records were searched for accounts of winter bat use within 90 km of the study area centroid. Movements of radiotracked little brown myotis during autumn migrations in Wyoming and Alaska (Johnson et al. 2017; Shively and Barboza 2017) and big brown bats (Eptesicus fuscus) in Colorado (Neubaum et al. 2006) were < 100 km. I used this buffer to evaluate all cave and mine records from the center of the study area to examine winter use of these sites over time. For this study, the winter season was defined as 1 November to 31 March to be consistent with other surveys and tabulations of winter records across western North America (Perkins et al. 1990; Priday and Luce 1997; Hendricks 2012). All bats observed during internal surveys were documented and species identified without handling to minimize disturbance. When identification could not be determined, Myotis species were only assigned to genus. Winter bat use of buildings in the Crystal River Valley was assessed by searching 2 maternity colonies located in buildings and reviewing winter occurrence records from multiple bat databases that CPW maintains (Whitaker and Gummer 1992).

Data analysis

Data from 2012 to 2013 were pooled to compare autumn roosts used by little brown myotis with available randomly selected locations at the landscape scale and characteristics of roosts summarized (mean, SE, and confidence intervals [CIs]) in R version 3.4.4 (R Development Core Team 2018). Logistic regression (GLM function) was used to compare categorical data related to sex and age proportions, across elevation zones and between months, and an information-theoretic approach used to select between competing explanatory models (Burnham and Anderson 2002). The proportion of individuals in 1 category (P₁) was compared to the proportion in a second category (P₂) and modeled under the constraint P₁ = P₂ (all data pooled), and compared with a general model where P₁ ≠ P₂ (allowing for a group effect). Akaike’s Information Criterion corrected for sample size (AIC_c—Burnham and Anderson 2002) was used to rank models. AIC_c differences (ΔAIC_c; difference in AIC_c score between ith and top-ranked model) and Akaike weights (w_i; probability that the ith model is the best approximating model among candidate models—Hosmer and Lemeshow 2000) were also calculated. The best-fitting model was assumed to have the lowest AIC_c score, with a competing model occurring only if ΔAIC_c < 2 (Burnham and Anderson 2002).

Modeling roost-site selection at the landscape scale was performed with a binomial error distribution and a logit link function using logistic regression models (Hosmer and Lemeshow 2000) with site type (autumn roost versus randomly selected locations) as the response variable and the site characteristics as independent variables. Sample sizes were too small to examine roost use by sex as a fixed effect (Weller et al. 2009). I selected 5 independent variables to develop a global model (Table 1) with variables that were highly correlated (r > 0.70) eliminated prior to analysis (e.g., aspect, shade, sum of distances from roost to roost). I ranked models with all possible combinations of the 5 independent variables using AIC_c. Akaike weights were calculated (w_i) and relative importance (w₊) for each predictor variable attained by summing these weights across all models in the set (Burnham and Anderson 2002). I also report the parameter estimates (β), associated SEs, and 95% CIs for each variable included in the top or competing models from the set (Burnham and Anderson 2002).

Table 1.

Measures of continuous habitat variables at autumn roosts used by little brown myotis (Myotis lucifugus) and randomly available sites, shown as the mean ( $\overline{X}$⁠), SE, and associated 95% CI. Bat roosts were found in the Crystal River watershed, Colorado, by radiotracking little brown myotis during autumns of 2012–2013. The number of buildings, trees, and rock crevices examined for used and random sites is presented for the categorical variable roost type (n) and all continuous variables.

			Autumn roost			Random site
Variable	Abbreviation	n	$\overline{X}$ (SE)	95% CI	n	$\overline{X}$ (SE)	95% CI
Elevation (m)	Elevation	44	2,200 (42)	2,116 to 2,284	220	2,615 (22)	2,571 to 2,659
Capture to roost distance (m)	CapToRDists	44	3,014 (435)	2,145 to 3,883	220	11,829 (347)	11,146 to 12,512
Distance to perennial rivers and streams (m)	StreamDist	44	185 (26)	134 to 236	220	514 (2)	472 to 556
Solar radiation (Watt hours/m2)	Solar	44	459,206 (12,296)	434,634 to 483,778	220	479,986 (9,360)	461,570 to 498,401
Roost type	RoostType
Building		18			3
Tree		16			195
Rock crevice		10			22

I used a balanced model set derived from a priori independent variables as all were judged to be biologically meaningful based on my current knowledge of little brown myotis ecology in autumn and winter settings. Developing importance values requires that a balanced model set (i.e., all variables appear in an equal number of models) be considered (Burnham and Anderson 2002), and is justified in situations where use of a limited number of a priori predictions are difficult to state due to a lack of background information (Doherty et al. 2012). Data on temperature and relative humidity collected during the autumn transition season of 1 August to 15 October 2014 were compared by roost type using pairwise t-tests with a Šidák correction for alpha level (< 0.0028 significance) as control sites were not available (Šidák 1967).

Results

Capture and radiotelemetry

During 22 visits in August and September 2012–2013, I netted at 6 sites and captured 184 bats of 6 species including 171 little brown myotis. At lower elevations (< 2,000 m) within the Crystal River Valley where little brown myotis maternity colonies occurred, captures were biased toward adult females and juveniles (adults = 17 females, 2 males; juveniles = 27 females, 18 males). Higher-elevation sites above the river valley floor and agricultural development produced more balanced sex and age ratios of bats (adults = 30 females, 19 males; juveniles = 30 females, 27 males; Table 2). Differences in proportions of captures by month were supported for adult bats, with more females recorded in August (39 females, 12 males) but similar sex ratios in September (8 females, 9 males; Table 2). Conversely, the proportion of juvenile bats was roughly equal across months (August = 12 females, 20 males; September = 45 females, 25 males) but values of ΔAIC_c < 2 suggest competing models or a weak correlation (Table 2).

Table 2.

Rankings by Akaike’s Information Criterion adjusted for small sample size (AIC_c) of top logistic regression models comparing sex and age ratios of bats caught in the Crystal River watershed in 2012–2013 at lower versus upper elevations and for August versus September. Symbols: ΔAIC_c is the difference in AIC_c value between the focal and top-ranked model; w_i is the Akaike weight (probability that the ith model is the best approximating model among the candidate models). The proportion of individuals of each trait (i.e., P₁, P₂, 1 − P₁, 1 − P₂), was modeled under the constraint P₁ = P₂ (i.e., all data pooled), and compared with a general model where P₁ ≠ P₂ (i.e., allowing for a group effect).

Analysis and model	K	ΔAICc	wi
Proportion of males captured at valley versus foothill sites
General	2	0.00	0.87
Constrained	1	3.72	0.13
Proportion of males captured in August versus September
General	2	0.00	0.80
Constrained	1	2.79	0.20
Proportion of juveniles captured at valley versus foothill sites
General	2	0.00	0.78
Constrained	1	2.58	0.22
Proportion of juveniles captured in August versus September
Constrained	1	0.00	0.65
General	2	1.23	0.35

I radiotagged 38 little brown myotis (37 adults, 1 juvenile; 13 males, 25 females) captured at buildings and over open water. Average body mass at time of capture was 8.4 ± 0.8 g for males and 8.4 ± 0.7 g for females. Thirty-six of the 38 tagged bats (35 adults, 1 juvenile; 12 males, 24 females) were tracked to 44 roosts spanning 1,252 m in elevation (Fig. 1). Nineteen of the roosts were in buildings, predominantly at lower elevations in the Crystal River Valley. Fifteen trees ranging from cottonwoods along the river to aspens as high as 2,734 m, 1 crevice in a rock fin at 2,172 m, and 9 rock crevices in talus on the flanks of Mount Sopris between 2,705 and 3,170 m were used as autumn transitional roosts and presumed hibernacula. Radiotagged bats used an average of 2 roosts (range 1–4) while being tracked during the life of the transmitter. Four building roosts were used by more than 1 radiotagged bat during the autumn of 2013 and 3 building roosts where bats were located in 2012 were used again in 2013. In addition, all 4 roost sites where bats were captured in 2013 were buildings used by tagged bats in 2012 and shared by multiple radiotagged individuals for at least 1 day (Fig. 1). One tree was used by 2 bats at different times. Marked bats were tracked an average of 14 days before loss of signal (range 0–39 days) and moved an average of 2 times (range 1–12).

Movements of bats between capture sites (which generally occurred in their summer range) and autumn roosts averaged 3 km straight-line distance with elevations of these roosts ranging from 1,918 to 3,170 m (Table 1). The maximum change in elevation between summer and autumn roost sites was 1,251 m. Flights in fixed-wing aircraft covering a 100-km radius from the center of the study area were taken in an attempt to locate missing signals in 2012 (n = 3) and 2013 (n = 2) with minimal success. One missing radiotagged adult female found during a flight moved 20 km from a building in the valley to an aspen tree in the mountains, a gain of 780 m in elevation. The bat stayed at the tree for 12 nights then returned to the original building (Fig. 1). Two bats were never detected after being marked.

Roost characterization

Landscape variables were measured at 44 autumn transitional roosts and potential hibernacula, and 220 randomly selected available locations. The 5 independent variables (Table 1) resulted in 32 models (including the intercept-only model). The highest-ranked model and 1 competing model (ΔAIC_c < 2) accounted for w_i of 0.52 and 0.28, respectively (Table 3). All remaining models were considered noncompeting. The variables CapToRDist and RoostType appeared in the top 8 models and were assigned the highest importance value, w₊ = 1.00 (Table 4). Parameter estimates and odds ratios from the top model for CapToRDist show that for every additional 10,000 m from summer capture sites, there would be 299 fewer occupied autumn roosts (SE = 1.26). Little brown myotis moved on average 3.0 km (range 0–21.4 km) between summer and autumn roosts, which was shorter than average distances to random sites (Table 1; $\overline{X}$ = 11.9 km, CI = 2.15 to 3.88 km). Little brown myotis used buildings, trees, and rock crevices as roosts, compared to random sites, which were predominantly trees (Table 1). Roost types used by bats included 28 buildings (SE = 0.96) and 7 rock crevices (SE = 0.68) for every 1 roost in a tree at the landscape scale (Table 5). The odds of bats roosting increased by 14:1 (SE = 1.05) with each 1,000 m increase in elevation, with the caveat that at some unmeasured elevation, odds would start to decrease. Roosts were also closer to rivers and streams, with 10 fewer random sites to every 1 used roost as distances decreased (SE = 1.07; Table 5). The second-highest-ranked model added solar radiation (Solar). Solar had the lowest relative importance value across all models, accounting for less than one-third that of CapToRDist and RoostType (Tables 3 and 4). A positive relationship (β = 0.19, CI = −0.22 to 0.60) implied known roosts had slightly higher solar radiation accumulations, but the CI overlapped 0, suggesting its effect was not detectable (Table 5).

Table 3.

The top-ranked logistic regression models comparing autumn roosts and presumed hibernacula of little brown myotis (Myotis lucifugus) to randomly selected sites in the Crystal River watershed, Colorado, 2012–2013. Models were ranked using Akaike’s Information Criterion corrected for small sample size (AIC_c) with corresponding Akaike weight (w_i), and the number of estimable parameters (K) reported. Competing models are considered as those with ΔAIC_c < 2. Variables are defined in Table 1.

Model	K	ΔAICc	wi
RoostType + Elevation + CapToRDists + StreamDist	7	0.00	0.52
RoostType + Elevation + CapToRDists + StreamDist + Solar	8	1.27	0.28
RoostType + Elevation + CapToRDists	6	3.62	0.09

Table 4.

Relative variable importance (w₊) of variables in predicting autumn roost and presumed hibernacula use by little brown myotis (Myotis lucifugus) at the landscape scale in the Crystal River watershed, Colorado, 2012–2013 based on a competing models analysis (see Table 1 for descriptions of variables).

Variable	w+
CapToRDist	1.00
RoostType	1.00
Elevation	0.93
StreamDist	0.85
Solar	0.34

Table 5.

The parameters of the variables included in the top 2 models for autumn roosts and presumed hibernacula used by little brown myotis (Myotis lucifugus) compared to randomly selected sites selected in the Crystal River watershed, Colorado, 2012–2013. Associated odds ratios and 95% CI are presented. Descriptions of variables are in Table 1.

Model	Variables	Estimate	SE	CI	Odds ratios	CI
RoostType + Elevation + CapToRDists + StreamDist	Roost type—Building	3.33	0.96	1.44 to 5.21	27.92	4.24 to 183.89
	Roost type—Rock crevice	1.98	0.68	0.64 to 3.32	7.25	1.91 to 27.61
	Elevation	2.67	1.05	0.61 to 4.73	14.45	1.84 to 113.53
	CapToRDist	−5.70	1.26	−8.17 to −3.23	< 0.01	< 0.01 to 0.04
	StreamDist	−2.32	1.07	−4.42 to −0.23	0.10	0.01 to 0.79
RoostType + Elevation + CapToRDists + StreamDist + Solar	Roost type—Building	3.24	0.96	1.35 to 5.12	25.45	3.85 to 168.14
	Roost type—Rock crevice	2.10	0.70	0.73 to 3.46	8.14	2.08 to 31.89
	Elevation	2.73	1.04	0.69 to 4.76	15.28	2.00 to 116.93
	CapToRDist	−5.90	1.29	−8.44 to −3.37	< 0.01	< 0.01 to 0.03
	StreamDist	−2.40	1.09	−4.52 to −0.26	0.09	0.01 to 0.77
	Solar	0.19	0.21	−0.22 to 0.60	1.21	0.80 to 1.82

I analyzed the microclimate of 23 autumn roosts in 5 buildings, 8 rock crevices in talus, 1 crevice in a rock fin, and 11 trees identified during the autumns of 2012 and 2013 (Table 6). Minimum average temperatures were significantly different between roosts in buildings and trees (t₇ = 5.0, P = 0.0008) using the conservative P < 0.0028 Šidák correction (Table 7). Rock crevices tended to remain at or above freezing on average ( $\overline{X}$ = 1.5°C), whereas roosts in trees regularly dropped well below freezing (Fig. 2; Table 6). No differences in average maximum temperatures were detected among roost types (Table 7), as 95% CI of all 3 roost types overlapped widely (Table 6). Differences in average temperatures between buildings and rock crevices (t₁₁ = 5.40, P < 0.0001), and rock crevices and tree roosts (t₁₂ = −2.96, P = 0.0027), were noted. Buildings averaged the highest temperatures of the 3 roost types, with tree roosts averaging several degrees cooler and rock crevices recording the coolest averages (Fig. 2; Table 6). Although SEs of average temperatures varied across sites for all roost types throughout autumn (Table 6), the pattern of daily temperature fluctuations within a given site differed by roost type: temperatures in buildings and trees fluctuated considerably across 24 h but those in rock crevices exhibiting much more stable regimes (Fig. 3). Average minimum relative humidity levels differed between rock crevices and tree roosts (t₉ = 3.73, P = 0.0007), whereas 95% CI of buildings overlapped widely with those of rock crevices or tree roosts (Table 6). Average maximum relative humidity levels of rock crevices differed from those in both buildings (t₉ = −5.85, P < 0.0001) and tree roosts (t₁₁ = 6.95, P < 0.0001). Average relative humidity levels at buildings and tree roosts overlapped widely, as noted by 95% CIs, and tended to be lower than those of rock crevices, where humidity was often so high that datalogger readings were saturated (Fig. 2; Table 6).

Table 6.

Summary of microclimate data collected as temperatures (°C) and relative humidity levels (% RH) at autumn roosts of little brown myotis (Myotis lucifugus) radiotracked in the Crystal River watershed, Colorado, during autumns of 2012–2013. The coldest (⁠${\overline{X}}_{\min }$⁠), warmest (⁠${\overline{X}}_{\max }$⁠), and average (⁠$\overline{X}$⁠) autumn temperatures at each roost were averaged over all roosts within a roost type. Relative humidity measures over 100% indicate the datalogger experienced periods of saturation.

Measure	Roost type	n	${\overline{X}}_{\min }$ (SE)	95% CI	${\overline{X}}_{\max }$ (SE)	95% CI	$\overline{X}$ (SE)	95% CI
Temperature (°C)
	Building	5	4.7 (1.0)	1.8 to 7.5	32.8 (4.3)	21.0 to 44.6	17.3 (0.6)	15.6 to 19.0
	Rock crevice	9	1.5 (1.3)	−1.6 to 4.5	26.3 (0.8)	24.4 to 28.1	10.3 (1.1)	7.7 to 12.9
	Tree	11	−2.1 (0.4)	−3.0 to −1.2	29.4 (1.2)	26.8 to 32.0	13.6 (0.4)	12.7 to 14.6
Relative humidity (% RH)
	Building	5	26.0 (4.0)	14.9 to 37.1	86.5 (6.9)	67.3 to 105.6	57.7 (4.3)	45.9 to 69.5
	Rock crevice	9	32.8 (3.6)	24.6 to 41.0	105.5 (1.0)	103.3 to 107.7	90.2 (3.6)	81.9 to 98.4
	Tree	11	19.4 (1.0)	17.1 to 21.6	97.5 (1.9)	93.1 to 101.9	64.2 (1.2)	61.5 to 66.9

Table 7.

Comparison of average minimum (⁠${\overline{X}}_{\min }$⁠), maximum (⁠${\overline{X}}_{\max }$⁠), and mean (⁠$\overline{X}$⁠) temperatures (°C) and relative humidity levels (% RH) by roost type collected at autumn roosts of little brown myotis (Myotis lucifugus) radiotracked in the Crystal River watershed, Colorado, during autumns of 2012–2013 using pairwise t-tests with a Šidák adjustment.

	${\overline{X}}_{\min }$			${\overline{X}}_{\max }$			$\overline{X}$
Measure or comparison	t	d.f.	Pa	t	d.f.	Pa	t	d.f.	Pa
Temperature (°C)
Building—Rock crevice	1.94	12	0.0630	1.50	4	0.0260	5.40	11	< 0.0001
Building—Tree	5.03	7	0.0008	0.76	4	0.2220	4.00	10	0.0233
Rock crevice—Tree	2.02	12	0.0369	−2.40	18	0.1630	−2.96	12	0.0027
Relative humidity (% RH)
Building—Rock crevice	−1.26	10	0.1328	−2.74	4	0.0003	−5.85	9	< 0.0001
Building—Tree	1.70	4	0.1105	−1.57	5	0.0148	−1.18	5	0.2500
Rock crevice—Tree	3.73	9	0.0007	3.92	15	0.0369	6.95	11	< 0.0001

^aŠidák adjustment indicates that differences are statistically significant at P < 0.0028.

Fig. 2.

Boxplot of A) minimum and B) average temperature (°C), and C) average relative humidity (% RH) measured inside 23 autumn roosts by roost type (building, rock crevice, and tree) used by little brown myotis (Myotis lucifugus) during the autumns of 2012–2013 in the Crystal River watershed, Colorado. Median (bold bars), 75th and 25th quartiles (upper and lower box limits, respectively), minimum and maximum values of each stage (whiskers), and possible outliers (circles) are presented.

Fig. 3.

Temperatures (°C) collected from 21 to 23 October 2013 at roosts used by little brown myotis (Myotis lucifugus) at an unheated building (square dot), a tree (dash), and a rock crevice in talus (solid) in the Crystal River watershed, Colorado. Temperatures were collected every 3 h at 17, 33, and 37 min past the hour, respectively.

Dates of latest access to presumed hibernacula, prior to complete snow coverage of the talus, occurred as early as 22 November and as late as 30 January (Supplementary Data SD1 and SD2). Maximum depth for 2 cameras situated on the same talus field was 143 cm, recorded on 2 February. The third camera became covered by snow on 8 March, preventing collection of the highest snow depth measurement at that site. Rocks surrounding the roosts became uncovered from the snow as early as 5 April and as late as 17 May (Supplementary Data SD1 and SD2).

Cave and mine surveys

Forty-eight winter surveys of 22 caves and 2 abandoned mines were conducted by the author and others or identified from the CPW Bat and Scientific Collections records within 90 km of the study area centroid between 1958 and 2017 to examine winter use of these features. Elevations of caves and mines surveyed averaged 2,462 m (range 1,763–3,173 m). A total of 4,204 bats were found during internal surveys, with the Townsend’s big-eared bat accounting for 97%, Myotis species 2%, and other species < 1% of the records. Reports of Townsend’s big-eared bat from 7 different surveys of 1 large hibernacula account for 91% of the total records. If this 1 cave is removed, a total of 373 bats were noted during 41 surveys, with Townsend’s big-eared bats accounting for 75% of these records. Myotis species were noted on 17 of the 48 surveys (35%) of caves and mines and averaged 5 bats per survey when found (range 1–23).

Discussion

As early as 1945, Griffin (1945:22) suggested that in areas where caves are less prevalent, bats could be hibernating in other features on the landscape such as small, deep rock crevices that provide suitable microclimates, and stated “The habits of bats are too little known to dismiss the possibility that in caveless areas they may habitually hibernate in unsuspected retreats.” Tracked bats in this study used short-distance migrations in elevation rather than latitude. Little brown myotis used several roost types including buildings, trees, and rock crevices as they transitioned away from summer maternity roosts in autumn. Each roost type conveys unique microclimates that would benefit bats in different transitional autumn stages. The likelihood of finding little brown myotis roosting in buildings and rock crevices on the landscape was higher than in randomly available structures, which were predominantly trees. Rock crevice roosts in talus selected by bats appear to facilitate microclimates with temperature, humidity, and dates of snow pack all supporting their use as presumed hibernacula; notably, these conditions would also support persistence of Pd. Myotis sp., and thus little brown myotis, seem to be largely absent from caves and mines that are available for use during the same winter window.

Segregation of the sexes during summer and use of elevational migration were documented through low-elevation captures in August that were biased toward adult females and juveniles, followed by higher-elevation captures in September that were more evenly distributed across sex and age (Table 2). Movements of radiotagged bats from lower to higher elevations in autumn also support this premise. Such differential use of elevational zones by season likely affords temperate bat species a number of benefits, including increased access or availability of suitable autumn transitional roosts and hibernacula, longer autumn foraging windows or better exploitation of food resources to build up energy reserves, participation in swarming (related to mating and orientation to hibernacula), use of daily torpor to curb physiological and energetic costs, and improved responses to changes in climatic factors (McGuire and Boyle 2013). Short migrations would allow for a greater portion of newly acquired fat reserves to be maintained by bats heading into hibernation. Barclay (1991) noted sexual segregation in little brown myotis from the Rocky Mountains in Canada and suggested that females migrate down in elevation during late spring to capitalize seasonally on food resources and roost temperatures tied to their reproductive process. Similar findings were documented for female big brown bats making seasonal movements in elevation on the Front Range of Colorado (Neubaum et al. 2006).

Much of the habitat used by bats across western North America encompasses strong elevational gradients (Adams 2003). The diverse topography of western North America contains an abundance of rock resources available for use at the surface (Theobald et al. 2015), and could eliminate the need for bats to undertake extensive migrations to less-common subterranean refuges, such as caves and mines, for use as hibernacula. In contrast, larger portions of the landscape in eastern North America, including parts of older mountain ranges such as the Appalachians, and most of the Midwest are either covered with vegetation or lack rock crevices at the surface (Theobald et al. 2015). Little brown myotis and other species are therefore required to make longer movements, sometimes > 300 km, to areas where caves are available (Griffin 1945; Norquay et al. 2013).

Landscape models suggested that roost type, capture to roost distance, elevation, and distances to perennial rivers and streams influenced use of autumn roosts by little brown myotis in Colorado. Bergeson et al. (2015) suggest that the adaptive roosting habits of little brown myotis may give them an advantage over other myotis species that are more restricted in roost selection. Randall et al. (2014) found little brown myotis used similar roosts (i.e., buildings, rock crevices, trees) during summer in southwestern Yukon, but that roost type varied by sex. Although sample size limited my ability to examine sex as a covariate, both male and female little brown myotis used rock crevices and buildings more than trees when compared to random sites during autumn in my study (Table 5). Similar findings in Yellowstone (Johnson et al. 2017) suggest that seasonal differences in roost type may become less sexually biased in autumn and winter as bats are preparing to hibernate. In eastern North America, where caves and mines are common winter roosts for little brown myotis, Lowe (2012) documented use of buildings, rock crevices, trees, and tree stumps during autumn. The broad array of features used by little brown myotis as roosts across their range suggests that they may capitalize on many roost types depending on the availability and suitability of these resources. In addition, priorities for individual bats heading into the autumn transition season vary (e.g., males attempting to swarm and mate, females still adding fat reserves, juveniles learning where to roost) and may determine which roost type is selected based on the distinct advantages each offers (Veith et al. 2004; Burns and Broders 2015).

Little brown myotis made smaller movements between summer and autumn roosts than distances between summer roosts and random sites (Table 1). Little brown myotis in my study moved on average 3.0 km (range 0–21.4 km) between summer (capture site) and autumn roosts. Autumn movements noted for this species in Wyoming and Nova Scotia were similar, averaging 5.9 km (range 1.2–19.6) and 2.5 km (range 0.2–13.1), respectively (Lowe 2012; Johnson et al. 2017). Despite bat movements in my study being small, they often rose steeply in elevation. In addition to providing short-distance migrations, elevation plays a role in determining microclimates at roosts (Neubaum et al. 2006, 2007; Wilcox and Willis 2015; Klug-Baerwald et al. 2017). Accordingly, elevation may influence when during the autumn season a roost type is selected for use by bats (Fig. 3). Elevation was shown to be associated with use of autumn roosts in rock crevices by big brown bats in Colorado (Neubaum et al. 2006). The final variable of relative importance, perennial rivers and streams, may provide natural movement corridors that migrating bats follow during autumn transition, which may explain why roosts were located closer to these features (Table 1). Rock crevice roosts used by little brown myotis in autumn in Yellowstone appeared to be highly aggregated along such landscape features (Johnson et al. 2017). Available food resources may be another reason that autumn roosts were concentrated near perennial streams and rivers, particularly at higher-elevation sites. Tracking signals for some radiotagged bats at night confirmed erratic movements, presumably indicating foraging. Johnson et al. (2017) showed that some insects remain active into early autumn periods, which coincides with use of rock crevices by bats in autumn in Yellowstone. Grizzly bears (Ursus arctos) in that same area have been documented using talus slopes in early autumn to feed on invertebrates such as miller moths (Euxoa auxiliaris—Mattson et al. 1991). These moths have extremely high body fat (up to 72%—French et al. 1994), making them an ideal food to consume in preparation for hibernation. Valdez and O’Shea (2015) found big brown bats in Colorado capitalizing on miller moths in early summer while the moths were making their annual plains-to-mountains migration.

Microclimates of autumn roosts suggested each roost type (i.e., building, rock crevice, tree) has distinct advantages that bats can utilize as autumn progresses into winter. Autumn transitional roosts in buildings were generally situated closer to original maternity colonies at lower elevations, and remained warmer on average longer into the autumn, than other roosts used during this season (Figs. 1–3). A warmer autumn roost could benefit bats born later in the summer and mothers who provided extended care of young by extending their autumn foraging activity. Shively and Barboza (2017) tracked little brown myotis to buildings in interior Alaska during early autumn and suggested these structures may buffer colder temperatures and facilitate foraging activity. This benefit would be important not only to female bats who must balance their use of torpor while rearing pups (Dzal and Brigham 2013) but also for male bats who may expend higher amounts of energy during autumn swarming activities (Ingersoll et al. 2010).

Some autumn transitional roosts used by little brown myotis in Colorado included cottonwood trees at lower elevations, and aspen trees at middle elevations. Tree roosts may facilitate use of daily torpor with average temperatures being cooler than those of buildings (Fig. 2; Table 5) and offered passive rewarming each afternoon (Fig. 3). Bats were found behind sloughing bark in all but 1 tree roost, which allowed for increased solar radiation in the afternoon. Passive rewarming would allow bats to maximize the use of torpor for much of the day while reducing the energy needed to arouse in the afternoon, a desirable situation for bats that are still foraging nightly in preparation for hibernation. These benefits continued until colder temperatures at some threshold below 0°C set in, at which time trees became unsuitable for use by bats due to the risk of freezing (Figs. 2 and 3; Table 6).

Rock crevices in my study were on average cooler than buildings but warmer than trees (Fig. 2; Table 6). In addition, minimum temperatures of rock crevice roosts tended to remain above freezing in late autumn while those at trees did not (Fig. 2; Table 6). Talus slopes are utilized as subterranean winter retreats for some insects such as ice crawlers (Grylloblattodea) since temperatures are buffered from freezing by snow cover (Kamp 1979; Schoville et al. 2015; Wipfler et al. 2014). Rock crevices used by bats in talus slopes in this study similarly provide suitable microclimates for hibernating because they maintain relatively constant, cool temperatures above freezing that are buffered by snow cover (Figs. 2 and 3).

Humidity also affects the suitability of hibernacula used by little brown myotis (Thomas and Clouthier 1992; Thomas and Geiser 1997). Rock crevices used by little brown myotis during autumn were more humid on average than roosts in buildings and trees (Fig. 2; Tables 6 and 7). Little brown myotis appear less capable of withstanding evaporative water loss during hibernation than larger species like big brown bats (Thomas and Clouthier 1992; Thomas and Geiser 1997) and have been documented gleaning water condensation from the fur during arousals (e.g., Davis 1970). Humidity levels recorded in rock crevices in my study reached complete saturation as indicated by relative humidity readings of over 100% on some dataloggers (Table 6).

Snow accumulation fully covered roosts in talus used by little brown myotis by late November at the highest-elevation site and by late January at the lower-elevation talus field (Supplementary Data SD1 and SD2). Based on the dates little brown myotis arrive at hibernacula in other regions of North America (Fenton and Barclay 1980), access to roosts in talus during autumn should not be blocked for this species in Colorado. Earliest dates of possible egress by bats from these sites ranged from early April to mid-May (Supplementary Data SD1). While mid-May seems late for bats to depart hibernacula, these dates correspond with winter roost studies for this species in areas with pronounced winter conditions (Czenze and Willis 2015; Meyer et al. 2016). Little brown myotis have been found using rock crevices and tree root wads in southeastern Alaska during autumn. Some roosts were snowbound for most of the winter, but others, particularly the lower-elevation root wads, were only intermittently covered by snow (K. Blejwas, USFS, Juneau, Alaska, pers. comm.). Others have suggested that winter activity by bats may be driven by the need for water (Thomas and Cloutier 1992; Ben-Hamo et al. 2013), which could be problematic for bats using talus slopes in winter where snow cover precludes egress. Beyond the potential for water condensation on the fur, I found that the talus fields within my study area have subterranean streams flowing under them due to the large snow fields and subsurface glacial activity. The presence of high humidity, streams, and melting snowpack should maintain free water for bats using talus slopes in winter. Myotis sp. have been documented drinking from free water in cave hibernacula (Twente 1955). Presence of potential drinking resources during hibernation may lead bats to undertake fewer arousals since each occurrence is more productive (Klug-Baerwald et al. 2017). Although radiotransmitters died before hibernation in bats could be confirmed, the combination of adequate temperatures, suitable humidity levels, and access to free water would suggest that talus slopes provide microclimate conditions appropriate for use as hibernacula.

Surveys of caves and mines within 90 km of the study area centroid revealed that Myotis sp. were rarely detected, with records collected from only 6 caves, averaging 5 (range 1–23) bats per survey. Previous efforts to survey abandoned mines in Colorado during winter found similar results (Ingersoll et al. 2010; Hayes et al. 2011). Notable cave and mine hibernacula for little brown myotis have been identified for some western states and provinces but are limited in number (e.g., Hendricks et al. 2000; Reimer et al. 2014). These results suggest that Myotis sp. are not using caves and mines as hibernacula at levels comparable to eastern portions of their range. Buildings also do not appear to be used as winter retreats by little brown myotis in my study area, unlike in areas of the midwestern United States (Whitaker and Gummer 1992).

If, as this study and others suggest, little brown myotis use rock crevices in talus as presumed hibernacula, there could be significant implications for cave management related to bats and WNS in areas where these resources coincide. The U.S. Forest Service and U.S. Bureau of Land Management have developed triggers for cave closures based on detection of WNS within certain buffer distances of their management boundaries (U.S. Forest Service 2013; U.S. Bureau of Land Management 2014). The expected spread of WNS to Colorado in the coming years will only intensify concerns over the status of little brown myotis and other bat species due to the potential impacts of the disease on native populations. Persistence and growth of Pd is highly tied to temperatures with optimal conditions occurring between 12.5°C and 15.8°C (Verant et al. 2012). At rock crevices used by little brown myotis in late autumn, and thus presumed hibernacula, the average overwinter temperature was 10.3°C with a range from 7.7°C to 12.9°C. While Pd can continue to persist and grow at such temperatures that are below optimal, it does so at a moderated pace (Verant et al. 2012). In addition, although the sample of marked bats I tracked is small relative to the local population, I did not confirm more than 1 bat at any of the rock crevices, a behavior which would inhibit bat-to-bat spread of WNS.

The presence of high, stable relative humidity levels in talus roosts also has important implications for presence and persistence of WNS. Marroquin et al. (2017) showed that mycelia growth of Pd at 13°C was most evident at relative humidity above 81.5% and impeded but not restricted below 70%. In my study, talus roosts averaged 90% relative humidity, which would facilitate growth of the fungus if spores arrive at these sites. An improved understanding of the lack of use of caves and mines as hibernacula by Myotis sp., the most vulnerable bats to WNS, will help inform recreational cave closure decisions in Colorado and western North America. Because caves and mines in the western United States generally do not appear to be used by as large of numbers of individuals or as high a diversity of bat species as many eastern hibernacula, I encourage those who are planning future WNS efforts that focus on caves and mines to consider the growing body of evidence supporting use of rock crevices as hibernacula.

Acknowledgments

I am grateful for property access and field support provided by I. Carney and the Wexner Ranch. I am indebted for the rigorous autumn fieldwork and data management provided by K. Ehler, K. Keisling, C. Jandreau, J. Pelham, and A. Tobin. J. Runge contributed invaluable assistance with statistical analysis. Funding and logistical support were provided by T. Jackson and B. Petch of Colorado Parks and Wildlife, P. Nyland of the White River National Forest, and H. Boyd, B. Hopkins, D. Long, and K. Leitzinger of the Bureau of Land Management. I am grateful to all of the private landowners who agreed to facilitate this study by allowing access to their property. Input from K. Aagaard, M. Alldredge, and J. Runge improved the content of earlier drafts. T. O’Shea, the associate editor, and an anonymous reviewer provided meaningful reviews of the final manuscript.

Supplementary Data

Supplementary data are available at Journal of Mammalogy online.

Supplementary Data SD1.—Access to rock crevice roosts in talus used by 3 little brown myotis (Myotis lucifugus) at A) the latest date bats could access the talus before coverage by snow in late autumn, B) date when maximum snow depth during winter was reached, and C) date of earliest egress from the talus field when snow has melted enough to reveal rocks in the Crystal River watershed, Colorado, 2013–2014.

Supplementary Data SD2.—Measures of snow depth collected during the winter of 2013–2014 at 3 autumn roosts of little brown myotis (Myotis lucifugus) tracked to high-elevation (m) talus slopes in the Crystal River watershed, Colorado, during autumn of 2012. The latest date of possible access was considered the last date before all access to the talus by bats was blocked due to snow completely covering all rocks in the immediate vicinity of the roost site. Maximum snow depth is provided as the date in which the snow reached the highest measure (cm) on the marker pole placed at the roost. The earliest date of possible egress was considered the first date when snow melted enough for rocks in the immediate vicinity of the roost site to be uncovered.