Winter behavior of bats and the progression of white‐nose syndrome in the southeastern United States

Abstract Understanding the winter behavior of bats in temperate North America can provide insight into how bats react to perturbations caused by natural disturbances such as weather, human‐induced disturbances, or the introduction of disease. This study measured the activity patterns of bats outside of their hibernaculum and asked how this winter activity varied by time, temperature, bat species, body condition, and WNS status. Over the course of three winters (2011–2013), we collected acoustic data and captured bats outside of five hibernacula in Tennessee, United States. During this time, Pseudogymnoascus destructans, the causative agent of white‐nose syndrome, became established in hibernacula throughout the region, allowing us to track disease‐related changes in the winter behavior of ten bat species. We determined that bats in the southeastern United States were active during winter regardless of disease. We recorded activity outside of hibernacula at temperatures as low as −13°C. Although bat activity was best determined by a combination of variables, the strongest factor was mean daily temperature (R 2 = .2879, F 1,1450 = 586.2, p < .0001). Bats that left the hibernacula earlier in evening had lower body condition than those that left 2–4 hr after sunset (F 7,932 = 7.225, p < .0001, Tukey HSD, p < .05). The number of daytime emergences from hibernacula, as determined via acoustic detection, increased the longer a site was P. destructans positive (F 3,17 808 = 124.48, p < .0001, Tukey HSD, p < .05). Through the use of passive acoustic monitoring and monthly captures, we determined that winter activity was driven by both ambient temperature and the presence of P. destructans.

Prior to the white-nose syndrome (WNS) disease epizootic, it was suggested that bats leave hibernacula in winter infrequently, and do so to switch roosts or in search of water or food (Boyles et al. 2006).
However, as WNS became established in the northeast, behaviors such as daytime and cold-weather flight during winter became indicative of infection. To date, no studies have looked at the winter behavior of bats in the lower latitudes of the southeastern USA, where warmer temperatures and available insects may allow for sustained activity through winter.
The importance of understanding the winter behavior of bats in the USA has become better appreciated due to WNS. First identified in February 2006 in a cave in New York, this disease has spread across eastern North America as far west as Missouri, south into Mississippi, and north into Canada, and was recently found further west in Washington state. To date, WNS has killed over 5 million bats of seven hibernating species (USFWS, 2014). Mortality as high as 90%-100% has been documented at some northeastern hibernacula, leading to possible regional extinction of several once-common species (Frick et al. 2010;Langwig et al. 2012). The highest rates of morbidity and mortality at sites in the northeast have been documented following the second and third winters after initial visual detection of the fungus (Titchenell 2012;Knudsen, Dixon & Amelon 2013).
Pseudogymnoascus destructans invades the muzzle, wings, and tail membrane of bats during torpor when their immune systems are likely suppressed (Moore et al. 2011;Field et al. 2015). Invasion of the wing tissue by P. destructans can cause dehydration, electrolyte imbalance, and increased arousal frequency leading to critical loss of fat stores needed to survive hibernation (Cryan et al. 2010;Reeder et al. 2012;Warnecke et al. 2012, Verant et al. 2014. Changes in winter behavior associated with WNS have been documented, with bats roosting in exposed regions of cave entrances or leaving hibernacula during the day, and flying during cold winter nights (Turner, Reeder & Coleman 2011;Foley et al. 2011;Carr, Bernard & Stiver 2014). To date, seven bat species have been confirmed with the disease via histopathology, with five additional species confirmed with P. destructans DNA on their epidermis (Meteyer et al. 2009;USFWS, 2014;Bernard et al. 2015).
Tennessee is among several southeastern states that contain hibernacula used by threatened and endangered species, including  (Samoray 2011;Carr et al. 2014). By winter 2015-2016, 54% of the counties in Tennessee (n = 52/95) were deemed WNS positive, with several high-priority M. sodalis hibernacula experiencing declines in bat populations (Campbell 2016). Myotis septentrionalis appears to be one of the hardest hit species in Tennessee, having experienced declines varying from 69% to 98.5% at known hibernacula (TWRA, unpublished data). Although declines due to WNS have not been confirmed in M. grisescens, they were thought to be vulnerable to the disease due to their propensity for forming high-density clusters during hibernation (U.S. Fish and Wildlife Service, 2009). Myotis grisescens are known to hibernate in eight caves in Alabama, Arkansas, Kentucky, Missouri, andTennessee (USFWS, 1982, 1997). In vitro growth curves suggest that P. destructans may reproduce more quickly in cave environments that maintain more moderate temperatures of 10-15°C in winter (Verant et al. 2012), which could result in increased virulence in southern hibernacula.
Our study measured the activity patterns of bats outside of caves during winter and investigated how this activity varied by time, temperature, bat species, body condition, and WNS status. To address these questions, we looked at the following hypotheses: (1) Bats will remain active throughout winter in the southeast due to warmer ambient temperatures than in the northeast and (2) Length of WNS infection at hibernacula will affect the body condition and winter activity of cave-roosting bats. Passive acoustic monitoring and bat captures during winter outside of caves allowed us to address these questions without adding additional stress to bats within hibernacula.   (Samoray, 2011;Flock, 2013).

| Acoustic activity
Ultrasonic bat detectors were deployed near the entrance of each hibernaculum. Acoustic data were recorded in zero-crossing mode onto

| Bat captures
During years 2 and 3, bats were captured near the entrance of each cave once a month using 6-, 9-, and 12-meter mist nets (Avinet, Dryden, New York, USA; 75/2 mesh size, 2.6 m high, 4 shelves, black polyester for bats). Nets were deployed for a total of 6,899.1 net hr/ m 2 over both winters. Each site had designated equipment to prevent F I G U R E 1 Cave locations within county boundaries in Tennessee, United States. Color coding of counties corresponds to the year Pseudogymnoascus destructans was confirmed using either real-time PCR (Muller et al. 2013) or histopathology (Meteyer et al. 2009) the spread of P. destructans among sites. Mist nets were deployed 30 min before civil sunset and remained open for 5 hr, until 30 bats were captured, or temperatures dropped below freezing (0°C). Bats were held individually in paper bags and placed in a large insulated box with four hand warmers (HotHands ® , Dalton, Georgia, USA) for 30-60 min prior to processing. Each captured bat was identified to species, and reproductive condition (scrotal, pregnant, lactating, postlactating, nonreproductive), right forearm length (mm), weight (g), and wing damage index (WDI) score were recorded. Guano was collected when possible. Wing damage was classified from 0 to 3, where WDI score = 0 indicates no obvious scaring or discoloration on the membranes, WDI score = 1 denotes light damage covering less than 50% of the membranes, WDI score = 2 indicates moderate damage greater than 50% of the wing membrane covered in scar tissue, and WDI score = 3 signifies heavy damage with deteriorated wing membranes and necrotic tissue (Reichard & Kunz 2009

| Data analysis
Acoustic activity was quantified as the number of files or "bat passes" recorded per 24-hr period. A "bat pass" was defined as a file containing a search-phase echolocation sequence of ≥2 echolocation pulses (Gannon, Sherwin & Haymond 2003). This metric is an index of overall activity and does not correlate to the numbers of individual bats actively flying outside of the cave (Kunz et al. 2007;Schwab & Mabee 2014). All call files were analyzed in AnalookW (version 3.9c, Titley Scientific) using antinoise filters. Filters were used to eliminate files that contained calls with less than two pulses, calls with short duration (<4 ms) or low frequency noise (i.e., wind, insect noise, or rain).
Once all noise files were removed, each file was manually vetted to determine whether the remaining call files fit the "bat pass" criteria.
Because we were attempting to quantify bat activity at each site regardless of species, and due to the difficulties in distinguishing the closely similar calls of several Myotine species (Barclay 1999;Britzke et al. 2011), species identification of each call was not attempted for this study.
Generalized linear mixed models (GLMM) with site as a random effect were calculated using the package lme4 (version 1.1-11) in R (version 3.2.4; R Core Development Team, 2016). Negative binomial GLMMs were employed to account for overdispersion and repeated sampling. Differences in bat acoustic activity among years were tested using daily and hourly "bat pass" totals. Fixed effects included year, sampling season (winters 2011-2012, 2012-2013, and 2013-2014)

| Acoustic activity
Useable acoustic data were collected on 88.4% or 2,091 of 2,366 detector nights. Loss of data resulted from malfunctioning Anabat/ ZCAIM units, SM2Bat+ external AC adaptor failure and a microphone cord failure. The recordings contained a total of 566,000 bat passes with bat activity recorded on 1,695 of the 2,091 recording days. Mean activity at sites varied considerably (Table 1), presumably due to differences in species compositions and population sizes at each location. Daily bat activity was best predicted by the model that included the interaction between month and temperature at emergence with the additive effects of year since P. destructans was confirmed and moon illumination (Table 2). Hourly bat activity was best described by the model that included the interaction of time of day (day or night classification) with mean hourly temperature, and year since P. destructans was confirmed (Table 2).
Bat calls were recorded on nights where temperatures at emergence were below 0°C on 23 nights at Blount cave, 27 nights at Campbell cave, eight nights at Hawkins cave, three nights at Warren cave, and three nights at White cave. The lowest temperature at which bats were detected was −13°C (n = 2 calls), and bats were not detected on nights when the temperature at emergence was below −8°C (n = 49 nights). Bat activity was significantly positively correlated with mean daily temperature (R 2 = .2879, F 1,1450 = 586.2, p < .0001).
The highest mean numbers of nightly calls were recorded during year 1, with a reduction in total calls recorded in subsequent years (Table 1). There was an increase in daytime activity at Blount cave in year 2, 2 years post-WNS confirmation, followed by a sharp decline in total acoustic activity by the end of year 3. However, at all sites, nighttime acoustic activity always exceeded daytime activity. The highest number of bat calls recorded per day (mean ± SE; 1,252.53 ± 39.14 bat calls) was recorded at caves that were P. destructans negative during the first year of monitoring (Campbell and

| Bat captures
A total of 947 individuals of 10 species were captured (

| DISCUSSION
This is the first study to demonstrate that bats are active throughout winter in the southeastern USA. Activity throughout winter was not restricted to bats in WNS-infected caves; however, increases in activity associated with low temperatures and diurnal flight were attributed to the disease. Variation in body condition across all species was found throughout a capture session, across seasons, and in relation to how long a hibernaculum was P. destructans positive.

| Winter activity
Hibernating mammals arouse periodically throughout winter to presumably maintain physiological balance and activate suppressed immune systems (Prendergast et al. 2002;Field et al. 2015); however, prior to the arrival of WNS, bats in northern latitudes were rarely found leaving the hibernacula (Whitaker and Rissler, 1992  during winter in more mild southern climates, and fecal samples from bats captured outside of caves verify that bats were leaving the hibernacula to forage (Bernard et al., unpublished). We captured bats representing every species of cave-dwelling bat known to hibernate in the region, as well as tree and foliage-roosting species, demonstrating that it is not only the latter that remain active during winter (Boyles et al. 2006). Both cave and tree or foliage-roosting species captured were active throughout winter at a range of temperatures, indicating that bats overwintering in the southeast exhibit different behaviors than northern populations (Boyles et al. 2006). This is an important factor for understanding latitudinal differences in WNS susceptibility and mortality (Bernard et al., unpublished).
We found evidence for a strong correlation of temperature with the third year of sampling, with an increase in the number of P. subflavus found roosting outside of the cave. We did not see the same mortality effects of WNS as seen in the northeast in years 2-3 until 4-5 years after the disease was detected in the southeast. Therefore, hibernating bat populations in the southeast may continue to experience delayed declines caused by WNS due to regional differences in climate-related life-history traits such as shorter hibernation periods (Sherwin, Montgomery & Lundy 2013), abbreviated torpor bouts Twente, Twente, and Brack, 1985;Jonasson & Willis 2012), and foraging opportunities throughout winter (Bernard et al., unpublished).

| Bat captures and body condition
Over the course of two winter seasons, we captured twice as many males as females flying outside of each hibernaculum. One explanation for sex bias in captures of bats active in winter is the "thrifty female hypothesis" (Jonasson and Willis 2011). This theory suggests that adult females minimize energy loss by relying more on deep torpor during hibernation, whereas adult males have more energy to expend and rely less on torpor. The capture of more active males than females is consistent with this idea. However, we did not detect any difference in body condition between males and females, nor did we find any seasonal patterns in the rate of decline in BCI for males versus females, as would be expected if females were remaining in caves to preserve energy. The "thrifty female hypothesis" may best explain sex ratio bias in bats hibernating in more northern latitudes and may function best where winters are more severe and intermittent foraging is not feasible like it is in the southeast.
An alternative explanation may be due to disproportionate sex ratios of bats within hibernacula (Whitaker & Gummer 2000;Parsons et al. 2003). Previous studies suggest sex ratios at winter sites are biased toward males at more northern roosts and become more evenly distributed the farther south a hibernacula is located, with sites in Florida reported to have an equal distribution of males and females (Davis 1959;Tinkle & Milstead 1960;Elder & Gunier 1978). Samanie Loucks and Caire (2007), however, described hibernating colonies of M. velifer in Oklahoma (n = 42 caves) to be female-biased throughout winter, suggesting seasonal movements of males and females throughout the hibernation period can change sex ratios frequently. Although we netted at each site once per month throughout winter, a single sex ratio determination for a hibernacula may not accurately reflect the true value for the entire hibernation season. While sample sizes were small, we saw no obvious gender bias in C. rafinesquii which uses abbreviated torpor during hibernation and continue to forage throughout winter (Johnson et al. 2012), or in L. borealis and L. noctivagans, which hibernate in more thermally unstable environments (Table 3). These observations suggest that both sexes of these species may be similarly active in winter in foraging or searching for alternative roosts.

| CONCLUSIONS
In the southeastern USA, bats are active outside of hibernacula and feed on insects throughout winter regardless of WNS disease status. Activity was found to be strongly correlated with temperature; however, activity during daytime and subfreezing temperatures increased following the confirmation of P. destructans at a site, suggesting that vulnerable bat species in the southeast exhibit behaviors similar to those seen at WNSinfected hibernacula in the northeast. The onset of aberrant behavior and mortality occurred 4-5 years after WNS confirmation, which was substantially later than what has been documented in the North. We found regional differences in the bat faunas affected by WNS. Caves dominated by M. septentrionalis, M. sodalis, and P. subflavus displayed increased aberrant behaviors and reduced activity toward the end of year 3, whereas M. grisescens populations showed no such effects. We also identified differences in how bats prepare for winter, with individuals in the southeast entering hibernation with lower mean BCI than bats studied in the northeast. Mild winter climates and available insects in the southeast can allow for shorter hibernation and continued foraging, likely reducing the demands of fat storage prior to winter. The regional differences in bat behavior we identified in this study provide a better understanding of the adaptive variations of bats to climatic perturbations and diseases. These findings are relevant to how bat populations across North America will react to infection by P. destructans and can allow for more targeted intervention and disease management.