Winter habitats of bats in Texas

Few studies have described winter microclimate selection by bats in the southern United States. This is of particular importance as the cold-adapted fungus, Pseudogymnoascus destructans, which causes the fatal bat disease white-nose syndrome (WNS), continues to spread into southern United States. To better understand the suitability of winter bat habitats for the growth of P. destructans in this region, we collected roost temperature and vapor pressure deficit from 97 hibernacula in six ecoregions in Texas during winter 2016–17 and 2017–18. We also measured skin temperature of Rafinesque’s big-eared bats (Corynorhinus townsendii), Townsend’s big-eared bats (C. townsendii), big-brown bats (Eptesicus fuscus), southeastern myotis (Myotis austroriparius), cave myotis (M. velifer), tri-colored bats (Perimyotis subflavus), and Mexican free-tailed bats (Tadarida brasiliensis) during hibernation to study their use of torpor in these habitats. We found that temperatures within hibernacula were strongly correlated with external air temperatures and were often within the optimal range of temperatures for P. destructans growth. Hibernacula and skin temperatures differed among species, with Rafinesque’s big-eared bats, southeastern myotis, and Mexican free-tailed bats occupying warmer microclimates and having higher torpid skin temperatures. For species that were broadly distributed throughout Texas, hibernacula and skin temperatures differed within species by ecoregion; Tri-colored bats and cave myotis in colder, northern regions occupied colder microclimates within hibernacula and exhibited colder skin temperatures, than individuals of the same species in warmer, southern regions. These data illustrate the variability in microclimates used as hibernacula by bats in Texas and suggest similar variation in susceptibility to WNS in the state. Thus, monitoring microclimates at winter roosts may help predict where WNS may develop, and where management efforts would be most effective.


Introduction
The choice of overwintering habitat has numerous impacts on mammalian hibernators. Suitable hibernacula must provide an appropriate thermal environment. Defining appropriate thermal habitats has been the focus of research for decades, as the temperatures inside PLOS  temperatures that one might expect in the subtropics may result in more favorable conditions for P. destructans to thrive compared to those farther north [11,20], potentially resulting in more severe infections. Thus, WNS pathology in areas such as Texas is difficult to predict in the absence of data on winter habitat selection and behavior in the region. Despite this need, few researchers have studied hibernation in southern populations of bat species with broad distributional ranges [11,[37][38][39].
To improve our overall understanding of the suitability of bat habitats for P. destructansand potentially WNS-in Texas, and to better inform management practices, we described the microclimates used by seven bat species as hibernacula across the state. We predicted that temperatures inside hibernacula would strongly correlate with external temperatures, and would be suitable for P. destructans growth. Based on species distribution and previous research on winter activity [40-42], we predicted that Rafinesque's big-eared bats (Corynorhinus rafinesquii), southeastern myotis (Myotis austroriparius), and Mexican free-tailed bats (Tadarida brasiliensis) would select microclimate within hibernacula having higher ambient temperatures and lower vapor pressure deficit (VPD) than all other species, and would subsequently have warmer torpid skin temperatures. We hypothesized that two species commonly found throughout Texas, the cave myotis (Myotis velifer) and tri-colored bat (Perimyotis subflavus), would exhibit variation in their microclimate selection across the state. Because bats hibernating in cold climates would be exposed to lower ambient roost temperatures, we predicted that they would therefore have lower skin temperatures than bats in warm climates. Finally, we predicted that tri-colored bats hibernating in culverts would have higher skin temperatures than those roosting within caves due to the more exposed and variable conditions present within culverts.

Materials and methods
All methods followed ASM guidelines [43] and were approved by the Texas A&M Institutional Animal Care and Use Committee (IACUC 2015-0296).

Study area
We conducted surveys for overwintering bats throughout Texas. Texas exhibits substantial variation in landscape, environment, latitude and climate, and is characterized by 12 level III ecoregions [44][45][46]. Thirty-three bat species have been documented across these ecoregions, the richest diversity of bat species found in any state in the United States [47]. We conducted surveys between November and March of 2016-2017 and 2017-2018 at 97 hibernacula, including 44 caves, 45 culverts, 3 buildings, 3 bat towers, 1 tree, and 1 bat box. We obtained access to caves through information provided by the Texas Speleological Society (TSS), Texas Cave Management Association (TCMA), Texas Grottos, Texas Parks and Wildlife biologists, and private landowners, and obtained access to other sites (i.e., buildings, bat towers, tree, and bat box) from landowners and biologists. We obtained information on historic culvert bat colonies from previous literature [48][49], from the Texas Department of Transportation (TxDOT), and from biologists. We randomly selected 77 10 x 10 km grid cells across the state using the Generalized Random Tessellation Stratified (GRTS) design [50] of the North American Bat Monitoring Program (NABat). Within those grid cells we surveyed an additional 158 culverts, of which eight contained bats. All sites included in our study were located across the following six level III ecoregions: Chihuahuan Deserts, East Central Texas Plains, Edwards Plateau, South Central Plains, Southwestern Tablelands, and Texas Blackland Prairies (Fig 1).

Data collection
We conducted hibernacula surveys between 08:00-17:00 hrs. To reduce disturbance, we collected all data without any animal handling. Upon entrance into the hibernacula, we identified species present based on visual characteristics and counted the number of bats belonging to each species. We collected data on all bat species observed at each site.
To assess the relationship between external and hibernacula air temperatures, and to understand how representative our single visits to hibernacula were of the entire winter season, we deployed EL-USB-2 temperature and humidity data loggers (EasyLog, Lascar Electronics, PA, USA) programmed to record internal temperature (site temperature) every hour from October-March at two caves and two culverts occupied by bats. We retrieved average external daily To assess the microclimates within each site used by hibernating bats, we collected roost temperature and relative humidity (RH) within 5 cm of each hibernating bat or cluster using a temperature and humidity pen (Extech 445580). In order to take into account the relationship between temperature and RH, we calculated vapor pressure deficit (VPD) which describes the difference between the amount of moisture in the air and how much moisture the air can hold when saturated [53][54]. We converted RH to VPD by first calculating the saturation water vapor pressure (SWVP) for every roost temperature recorded and then multiplying SWVP by RH to produce the actual water vapor pressure (AWVP) [55]. VPD was calculated by subtracting AWVP from SWVP. We also measured skin temperatures of bats from a maximum distance of 0.3 m using a digital infrared thermometer (Extech; Waltham, MA) to gather baseline data on torpor behaviors in each species. We collected these skin temperature readings from the back of each bat. Although infrared thermometers are not the preferred method to measure bat skin temperature, they are a non-invasive and quick method of assessing temperature, and have been used in other research with success [56]. In order to ensure that we recorded reliable skin temperatures from torpid bats, we did not collect data from any bats that showed visible signs of arousal (e.g., vibrating, movement). To determine if bats were hibernating at the temperature within the hibernacula, we also determined substrate temperature adjacent to roosting bats.

Data analysis
To test the hypothesis that site temperature is related to external temperature, we used a Pearson product-moment correlation. To test the hypothesis that Rafinesque's big-eared bats, southeastern myotis, and Mexican free-tailed bats would roost in areas of hibernacula with warmer roost temperatures and higher AWVP, and would therefore have higher torpid skin temperatures than all other species, we used Kruskal-Wallis H-tests (comparisons of roost temperatures and skin temperatures among species were made using separate tests). We used non-parametric statistics to analyze these data because Shapiro-Wilks tests indicated that both roost temperatures and skin temperatures for each species were not normally distributed [57]. Prior to analysis, we averaged skin temperatures, roost temperatures, substrate temperatures and VPD per site for each species to avoid pseudoreplication. We also used Pearson's productmoment correlations to test for a linear relationship between skin temperature and substrate temperature, and test for a linear relationship between skin temperature and roost temperature.
We tested our hypothesis that skin temperatures and roost temperatures of tri-colored bats hibernating in cold climates in northern Texas (Southwestern Tablelands) would be colder than those in the warm climates of southern Texas (Chihuahuan Deserts, Edwards Plateau, South Central Plains) using Kruskal-Wallis H-test. We performed post-hoc comparisons of groups using Dunn's tests, adjusting P-values using the Benjamini-Hochberg method to decrease the false discovery rate. Because we only found cave myotis in two ecoregions, we used a Wilcoxon rank-sum test to test our hypothesis that the skin temperatures of cave myotis in cold climates would be colder than the skin temperatures of cave myotis in warm climates. Finally, to compare roost temperatures and skin temperatures of tri-colored bats hibernating in culverts to those in caves between ecoregions, we used a Wilcoxon rank-sum test. All means are reported with standard deviations (± SD). We considered a P-value � 0.05 significant for all tests. All analyses were performed in Program R v. 3.4.1 [58].
Site temperature was highly correlated with external temperature for both occupied caves and culverts (Fig 2; P � 0.05). Between October and March, the average roost temperatures for the two caves ranged from 2.29-20.90˚C (Cave 1) and 12.38-21.31˚C (Cave 2) (Fig 2). During the same period, the average roost temperatures for the two culverts ranged from 0.19-23.10C (Culvert 1) and 6.21-23.00˚C (Culvert 2) (Fig 2). Roost temperatures differed among species (K = 30.27, d.f. = 6, P = < 0.001; Fig 3A). Rafinesque's big-eared bats were found at hibernating in areas with significantly higher roost temperatures ( � X = 22.20 ± 4.21) than all other species except Mexican free-tailed bats and southeastern myotis ( � X = 19.67 ± 3.93; � X = 17.36 ± 2.49, respectively). The roost temperatures at which Mexican free-tailed bats were found did not differ significantly from the roost temperatures at which tri-colored bats, big brown bats, and cave myotis were found ( � X = 15.79 ± 3.69; � X = 11.73 ± 5.57; � X = 14.39 ± 4.87, respectively). Townsend's big-eared bats were found at significantly lower roost temperatures ( � X = 11.18 ± 4.36) than Rafinesque's big-eared bats, southeastern myotis, tri-colored bats, and Mexican free-tailed bats but were not significantly different to the roost temperatures at which big brown bats and cave myotis were found.

Discussion
We found that bats hibernate under diverse conditions across Texas, and observed broad differences in selected microclimates and torpid skin temperatures among species. Our data showed that hibernacula temperatures in Texas were suitable, and often within the optimal growth range of P. destructans, placing several bat species at risk of developing WNS. Although we did not investigate microclimate variability across the winter season at all sites, which constrains our ability to fully understand the available microclimates and species' susceptibility to WNS, our data add significantly to our knowledge about microclimates used by bats during winter and their suitability to the growth of P. destructans.
As expected, we found a strong relationship between internal and external temperatures for both caves and culverts. However, hibernacula temperatures were not simple reflections of surface temperatures and the four hibernacula that we monitored with dataloggers for the entire winter illustrate that a diversity of thermal habitats can be found in Texas. This variation can result from several factors, including aspects of local topography and the structure of the site and its entrances [59][60]. Temperatures within each hibernacula varied throughout the winter, but minimally so during the middle of the hibernation period (December-February). Thus, although it was not feasible for us to deploy microclimate dataloggers in all hibernacula, roost temperature data taken at these locations were similar to internal site temperatures indicating that our single-visit surveys captured conditions representative of the winter season. Although these temperatures are warmer than those reported for caves and mines in temperate regions, they are similar to those reported from subtropical zones [8,11,55,61].
In a review of hibernacula temperatures in North America, Perry [60] found that temperatures less than 10˚C are most suitable for hibernation, and that hibernacula in areas with mean annual surface temperatures (MAST) greater than 10˚C are too warm. Our data show that knowledge about hibernation ecology from the temperature region of North America may yield misleading conclusions about hibernation in the subtropical region, as we frequently found bats hibernating in temperatures greater than 15˚C. Overall, we found that microclimates within hibernacula used by all species overlapped the temperature range of growth for P. destructans [20]. In fact, 22 of the 97 (22.7%) sites surveyed had microclimates within the optimal growth range (12.5-15.8˚C) [20]. Four of the species found within these sites, the big brown bat, cave myotis, southeastern myotis, and tri-colored bat, are known to be susceptible to WNS [62]. While these data suggest high risk of WNS mortality, and P. destructans has been present in Texas since 2017 [31], diagnostic symptoms of WNS have yet to be documented within the state. Although WNS may yet develop in Texas, we urge researchers to investigate the possibility that hibernating bats in Texas mitigate risk to WNS through differences in winter physiology and behavior in comparison to populations farther north.
Hibernacula temperatures were not always within the optimal growth range for P. destructans, and in many cases temperatures less than 10˚C were available and occupied. As we hypothesized, tri-colored bats and cave myotis in northern Texas were in deeper torpor-as indicated by lower skin temperatures-than bats of the same species in southern Texas. Tricolored bats and cave myotis were found roosting at temperatures ranging 5.50-25.56º C and 4.67-23.67º C, respectively, and were always found torpid, as indicated by the strong correlation between skin temperature and roost temperature, and skin temperature and substrate temperature. This large range in hibernacula microclimates used by tri-colored bats is similar to values reported in previous studies (e.g., [8,11,48,[63][64]). The average temperature used by this species, 15.8˚C, is at the upper end of the optimal growth range for P. destructans, suggesting this species is at risk in Texas. The range of microclimates used by cave myotis in our study was similar to values contained in a review by Webb and colleagues [8]. Similar to tricolored bats, the average roost temperature for cave myotis, 14.35˚C, suggesting these bats select microclimate conducive to growth of P. destructans.
Compared to tri-colored bats and cave myotis, Rafinesque's big-eared bats, Mexican freetailed bats and southeastern myotis were consistently found in warmer areas of the hibernacula and had warmer skin temperatures, although these values did not always differ significantly from other species. This result is consistent with previous studies of Mexican free-tailed bats, which have shown this species employs shallow torpor during winter months [41, [65][66]. Unlike the Mexican free-tailed bat, Rafinesque's big-eared bat and the southeastern myotis have limited distributions within Texas, and are only found in regions of warmer climates. In particular, Rafinesque's big-eared bat is only found in the South Central Plains [41]. Compared to other bats hibernating in this region, Rafinesque's big-eared bats had the highest skin temperatures. This result is similar to that of Johnson and colleagues [42], who found this species' winter skin temperature of 13.9˚C ± 0.6 among hibernating big-eared bats in Kentucky. Johnson and colleagues [42] also found this species frequently switched among nearby hibernacula in winter. Although we did not track movements of this or any other species, the relatively high torpid skin temperatures we observed suggests this species may also be active during winter in Texas.
Contrary to predictions, there was no difference in roost VPD between species. This indicates that-when variation in temperature is taken into consideration-the bat species studied in this manuscript do not differ in their required or preferred level of roost moisture. Across species, VPD was low, suggesting that bats were found in roosts with high levels of internal atmospheric moisture. As hibernating bats are as risk of evaporative water loss (EWL), it makes sense to find these bat species in moist environments. Although not significant, the data did suggest that Mexican free-tailed bats were found in hibernacula with slightly higher VPDs than other species. As a species that tends to engage in shallow torpor during winter [41, [65][66], Mexican free-tailed bats are not as much at risk from EWL and thus can tolerate somewhat drier roosts.
Our work adds to previous studies documenting the use of culverts as hibernacula for tricolored bats in Texas [48][49]67]. Culverts may act similarly to artificial caves in regions where cave-forming karst is scarce, increasing roosting options. Contrary to our prediction, tri-colored bats hibernating in culverts did not have significantly different skin temperatures than those hibernating in caves. However, the range of skin temperatures of tri-colored bats roosting within caves was larger than those in culverts. Although not reported here, our observations suggest that tri-colored bats may be selecting to roost within a given temperature range within culverts as a result of their body condition, EWL, and rates of heat loss through convection and conduction [5,9,68]. Raesly and Gates [69] suggested that there is an interplay between external hibernacula characteristics and internal characteristics that influence roost use. This relationship, along with the lack of caves in the region, may drive occupancy of certain culverts.
The roost temperatures, skin temperatures and actual water vapor pressures we reported herein depict microclimates selected by bats within hibernacula. Further, these data speak to the vulnerability to developing P. destructans infections but cannot predict WNS mortalities, as additional factors such as length of torpor bouts, winter food availability and foraging patterns must also be considered. Nevertheless, there are some important takeaways from our study. Based on our data, and provided that P. destructans continues to spread in Texas, we hypothesize that Mexican free-tailed bats, southeastern myotis, and Rafinesque's big-eared bats will not develop WNS in Texas. Although our data suggest that cave myotis and tri-colored bats in Texas will be affected by WNS, the lack of any WNS-affected bats in Texas to-date demands further research. Future studies should focus on gathering information on the torpor patterns (e.g., torpor bout lengths) of bats in southern US. This work will help to improve the current understanding about differences in hibernation ecology and the potential impacts of WNS on these southern occupants.