Long‐term changes in occurrence, relative abundance, and reproductive fitness of bat species in relation to arrival of White‐nose Syndrome in West Virginia, USA

Abstract White‐nose syndrome (WNS) is a disease caused by the fungus Pseudogymnoascus destructans which has resulted in the deaths of millions of bats across eastern North America. To date, hibernacula counts have been the predominant means of tracking the spread and impact of this disease on bat populations. However, an understanding of the impacts of WNS on demographic parameters outside the winter season is critical to conservation and recovery of bat populations impacted by this disease. We used long‐term monitoring data to examine WNS‐related impacts to summer populations in West Virginia, where WNS has been documented since 2009. Using capture data from 290 mist‐net sites surveyed from 2003 to 2019 on the Monongahela National Forest, we estimated temporal patterns in presence and relative abundance for each bat species. For species that exhibited a population‐level response to WNS, we investigated post‐WNS changes in adult female reproductive state and body mass. Myotis lucifugus (little brown bat), M. septentrionalis (northern long‐eared bat), and Perimyotis subflavus (tri‐colored bat) all showed significant decreases in presence and relative abundance during and following the introduction of WNS, while Eptesicus fuscus (big brown bat) and Lasiurus borealis (eastern red bat) responded positively during the WNS invasion. Probability of being reproductively active was not significantly different for any species, though a shift to earlier reproduction was estimated for E. fuscus and M. septentrionalis. For some species, body mass appeared to be influenced by the WNS invasion, but the response differed by species and reproductive state. Results suggest that continued long‐term monitoring studies, additional research into impacts of this disease on the fitness of WNS survivors, and a focus on providing optimal nonwintering habitat may be valuable strategies for assessing and promoting recovery of WNS‐affected bat populations.


| INTRODUC TI ON
White-nose syndrome (WNS) is a deadly disease caused by the fungus Pseudogymnoascus destructans (Pd) which affects hibernating bats. WNS was first discovered in a cave in New York in 2007; by 2012, it had been confirmed as present in caves across most of the eastern United States and was estimated to have killed 5.7-6.7 million bats (US Fish & Wildlife Service, 2012). Since then, the disease and bat death toll has spread across much of North America, with new species being affected as the disease crosses into different geographic ranges. Population impacts have varied by species, with some of the most heavily impacted bat species presently including the endangered Myotis sodalis (Indiana bat), Perimyotis subflavus (tri-colored bat), and Myotis lucifugus (little brown bat; Figure 1). In West Virginia, hibernacula counts for these species have declined by ~97%, 90%, and 97%, respectively, since WNS was first detected in the state in 2009 (West Virginia Division of Natural Resources [WVDNR], unpublished data). Hibernacula counts have been the predominant means of tracking the spread and impact of this disease (e.g., Frick, Pollock, et al., 2010;Langwig et al., 2012;Powers et al., 2015;Turner et al., 2011). However, some species, such as Myotis septentrionalis (northern long-eared bat), are also highly vulnerable to WNS but are less conspicuous in hibernacula, making their populations difficult to track with winter counts. Monitoring of populations outside the hibernacula season may provide the best means of tracking changes in these populations.
Monitoring of summer bat populations using either acoustic or capture surveys provides a means to assess relative changes in population numbers over time. Several studies have detected declines in Myotis summer populations following WNS arrival in an area (e.g., Brooks, 2011;Dzal et al., 2011;Moosman et al., 2013;Reynolds et al., 2016) that often track those seen in hibernacula. As such, the impacts of WNS on population trends of several species are well documented during both winter and summer across geographic areas where Pd has been present for many years. However, there is a paucity of information regarding the impacts to reproduction and body condition, particularly in the natural environment and outside of hibernation. Higher body fat stores in hibernating bats are likely advantageous for both over-winter survival and successful reproduction the following year (Cheng et al., 2019;Kunz et al., 1998); this is especially true for bats infected with Pd, as they are subject to more frequent and longer arousal periods and resultant increases in body fat depletion during hibernation (Lilley et al., 2017;Reeder et al., 2012). Higher body fat in early winter could reduce WNS mortality by 58%-70%, providing higher energy stores to be depleted throughout hibernation to deal with the demands of fighting Pd infection during that period (Cheng et al., 2019). Furthermore, M. lucifugus appears to have evolved mechanisms to conserve fat reserves during hibernation by maintaining normal torpor patterns in the presence of Pd (Frank et al., 2019). In addition to the potential loss of fat reserves during hibernation, energy demands for Pd-infected individuals upon emergence from hibernacula are particularly high as they attempt to recover from infection-related tissue (wing) damage with a potential associated reduction in flight and, thus, foraging efficiency (Fuller et al. 2020).
Over time, steep declines in adult survival as a result of WNS may be ameliorated in different ways, such as developing resistance to pathogen infection or physiological adaptations in survivors (Frank et al., 2019;Langwig et al., 2017). However, population-level impacts of WNS likely extend well beyond those reflected by adult mortality rates, as WNS-infected individuals emerging from hibernacula are faced with increased energetic demands to cope with healing of skin and wing damage and heightened stress levels (Davy et al., 2017;Meierhofer et al., 2018), in addition to the normally high energy requirements of bats during spring emergence. Given the high energetic cost to females of gestation and raising pups, it is likely that there is some energetic trade-off between adult annual survival and successful reproduction for WNS-affected bats emerging from hibernacula. Life history theory predicts that in organisms with high adult survival, such as bats, females should forego reproduction and allocate resources to their own maintenance and survival when resources are limited and the probability of survival of their young is particularly low (Barclay et al., 2004). Thus, it is possible that WNS-affected females facing the high energetic cost of recovery upon emergence might forego reproduction and instead allocate resources to recovery and increasing body mass prior to entering hibernacula the following winter.
A better understanding of the fitness of WNS-affected individuals and populations requires an assessment of demographic parameters such as age, sex, reproductive state, and body condition, collected in the field as part of capture surveys. Such detailed demographic data are critical to our understanding of how WNS affects individuals and populations that do survive infection, enabling F I G U R E 1 An adult Myotis lucifugus (little brown bat), released on a tree after mist-net capture in Grant County, West Virginia, USA. Photograph used with permission from Keith Christenson more accurate population viability analyses and informing potential recovery strategies for species already impacted. Unfortunately, long-term capture data are rarely available for bat species and even fewer are available for assessing bat communities, though preand post-WNS capture data have been used to detect declines in Myotis species in New Hampshire (Moosman et al., 2013), Indiana (Pettit & O'Keefe, 2017), Kentucky (Thalken et al., 2018), and the Great Smoky Mountains of North Carolina and Tennessee (O'Keefe et al., 2019). When such data do exist, the numbers are generally too small to assess more than population trends. Analyses of demographic parameters such as changes in reproductive state or body condition pre-and post-WNS are especially difficult since numbers of those species most-affected decline so precipitously that sample sizes post-WNS are very small (Lacki et al., 2015).
Understanding the condition and reproductive status of WNSaffected bats that survive hibernation can also help illuminate the trade-offs that may affect persistence of populations post-WNS, inform vulnerability assessments for species most-affected, and help to focus conservation efforts in habitats used outside the hibernation season. Here, we present data from a long-term summer bat monitoring program at the Monongahela National Forest (MNF) in West Virginia used to assess changes in the bat community and population parameters of multiple species in the MNF pre-and post-WNS, including an analysis of changes in body condition as well as reproductive status and timing in response to WNS.
We predicted that summer populations of WNS-vulnerable species in our study area would reflect the steep declines observed in hibernacula counts for species such as M. lucifugus and P. subflavus in the region (Powers et al., 2015;Turner et al., 2011;WVDNR, un-published data) and the general lack of M. septentrionalis captures and recordings across the summer landscape in the northeast post-WNS (Moosman et al., 2013;Reynolds et al., 2016). Species breeding in the study area that are not considered to be vulnerable to WNS were expected to remain relatively stable, with the exception of Corynorhinus townsendii virginianus (Virginia big-eared bat), which had been steadily increasing prior to WNS invasion as a result of conservation actions (Stihler, 2011) and was predicted to continue on that track. There are few published data available to inform predictions on reproductive status and body condition post-WNS. However, given the energetic costs faced by Pd-infected individuals emerging from hibernacula (Fuller et al., 2020;Meierhofer et al., 2018), we expected to see adverse impacts to WNS-vulnerable species reflected in reproduction, body condition, or both.

| Study area
This study was conducted across the Monongahela National Forest

| Data collection
The MNF has conducted annual summer bat mist-netting since 1997. Mist-net poles and nets were set up prior to dusk and nets deployed at dusk. A minimum of two net locations were used at each site (≥30 m apart), and nets were operated for a minimum of 5 hr each night. Net-set heights and lengths ranged from 2.6 m (singlehigh) to 10.4 m (quadruple-high) and from 3 to 18 m, respectively, depending on the physical characteristics of the site. Nets were checked every 8-10 min while deployed.
Nets were operated for 2 nights/year at each site. When inclement weather prevented completion of a full survey at a given site, additional surveys were conducted until 2 full survey nights were completed. Individual bat data from partial survey nights were included in reproductive state and body condition analyses, but partial survey nights were excluded from presence and relative abundance analyses so that sampling effort was consistent across sites. At Data recorded for each bat captured included: species, sex, age, mass (g), reproductive condition, and forearm length (mm). Age (juvenile or adult) was determined by degree of ossification of the finger joints (Kunz & Anthony, 1982), and reproductive condition was recorded as pregnant, lactating, postreproductive, or nonreproductive for females and scrotal or nonreproductive for males (Racey, 1988).
Surveys followed U.S. Fish and Wildlife Service WNS protocols (e.g., U.S. Fish & Wildlife Service, 2018) and handling guidelines for bats, Monongahela National Forest protocol, and WV Division of Natural Resources permit requirements. Bat morphological data were collected following the Mammal Collectors' Manual (Nagorsen & Peterson, 1980

| Presence and relative abundance
We used two-part zero-inflated Poisson generalized additive models (ZIPGAM) to estimate temporal patterns in presence and relative abundance (i.e., site detection and count, respectively) for each bat species from 2003 to 2019 (Wood et al., 2016). We chose zeroinflated models because the survey data contained a large proportion of zero counts (i.e., 56% of sites across all species and survey years), and the count data distributions did not satisfy assumptions of standard Poisson or negative binomial distributions. We chose generalized additive models rather than generalized linear models to allow for potentially complex trend dynamics, which is common for long-term monitoring data (e.g., Fedy & Aldridge, 2011;Fewster et al., 2000). The two-part models included binary (presenceabsence) and continuous (count) states, where the binary state was modeled using a binomial distribution and the continuous state was modeled using a truncated Poisson distribution (Cunningham & Lindenmayer, 2005). To estimate the influence of year on presence and relative abundance, we included year as a smoothed continuous predictor for both the binary and continuous states. The smooth parameter was modeled using restricted maximum likelihood with Laplace approximation, which has been shown to reduce undersmoothing compared with generalized cross-validation (Reiss & Ogden, 2009;Wood, 2011).
To identify time periods of significant population increases and decreases, we computed 200 first-order derivatives (i.e., slope of the tangent line) across the 17-year period using the finite difference method (Simpson, 2018), and considered periods of significant change to be those in which the 95% point-wise confidence intervals did not overlap 0. We assessed significance of model terms using Wald tests with α = 0.05 (Zuur et al., 2009). We examined model assumptions using residual diagnostic plots (Jones & Wrigley, 1995;Zuur et al., 2014), as well as ensured model convergence was reached and that the smoothing parameter was not overfit based on the k-index test (Wood, 2017). We examined model performance by comparing modeled trend estimates to annual summaries of survey For the reproductive state models, we included day of year as the null model, and assessed whether WNS invasion status was a supported additive predictor using likelihood-ratio tests (α = 0.05; Zuur et al., 2009). We modeled probability of being nonreproductive or reproductive to investigate changes in reproductive potential. We also modeled probability of being in four states: nonreproductive, pregnant, lactating, and postreproductive, to investigate potential shifts in reproductive timing. We used logistic regression for the two-state models and ordinal regression for the four-state models (Agresti, 2007). We examined model assumptions and performance using residual diagnostic plots and receiver operating characteristic (ROC) curves (Agresti, 2007;Pardoe & Cook, 2002).
We used body mass rather than a body condition index (BCI) to assess body condition for bats in this study. While BCI (body mass divided by forearm length) has often been used to assess body condition in bats, McGuire et al. (2018) found that forearm length does not accurately correct for intraspecific variation in body size and that body mass alone is a better predictor of fat mass in bats. For the body mass models, we included reproductive state as the null model, and assessed whether WNS invasion status was a supported interaction predictor of body mass using likelihood-ratio tests (α = 0.05).
This model estimates the influence of WNS on body mass of bats in each reproductive state. We used linear regressions with a Gaussian distribution for the body mass analyses (Zuur et al., 2009). We performed model validation using residual diagnostic plots and removed 8 outlier observations among the species to satisfy assumptions of normality and homoscedasticity.

| Bat community
Changes to bat populations following the onset of WNS were characterized by dramatic decreases in species that had previously com- where (Cope et al., 1991). Since the post-WNS decline of M. lucifugus populations in Pennsylvania, state biologists have also documented shifts to E. fuscus colonies in 6 maternity roost sites that previously supported large M. lucifugus colonies (G. Turner, Personal communication). Future research efforts examining conditions and relative use of maternity roosts by these and other species would be useful to better inform our understanding of optimal maternity colony conditions and potential competition across species.

| Reproductive state and body mass
Myotis lucifugus females in our study area were less likely to be reproductively active post-WNS (Figure 5c), though not significantly so (likely due to small post-WNS sample sizes), suggesting that WNS-related stressors may have an adverse effect on the reproductive capacity of this species. M. lucifugus and other WNS-affected species that survive Pd exposure are faced with extreme energetic imbalance associated with healing of wing damage and reduced foraging efficiency (Fuller et al., 2020;Meierhofer et al., 2018). Captive

M. lucifugus that survived Pd infection in hibernation had increased
wing tissue damage and elevated mass-specific resting metabolic rate compared with Pd-uninfected bats, suggesting greater energetic costs during spring in WNS survivors (Meierhofer et al., 2018).
In addition, chronic stress levels in free-ranging bats indicate that physiological consequences of Pd infection persist even after infected bats have emerged from hibernation and recovered from wing damage and other outward signs of disease (Davy et al., 2017).

WNS survivors faced with increased energetic costs associated with
wing tissue damage repair and other aspects of disease recovery also may be unable to forage as efficiently as they normally would due to impaired flight performance (Fuller et al., 2020). Such Pd survivors, with increased energetic demands and decreased foraging efficiency, may be unable to expend the additional high energetic costs associated with successful rearing of young.
Although M. lucifugus populations in the northeast may be evolving a resistance to Pd mortality during hibernation (Frank et al., 2019;Langwig et al., 2017), populations that fail to meet their pre-WNS reproductive potential likely will continue to decline. Studies of M. lucifugus maternity colonies in the vicinity of the WNS epicenter show successful reproduction and colony growth, with increasing survival probabilities post-WNS (Dobony & Johnson, 2018;Ineson, 2020).
However, our results suggest that the proportion of reproductively active females may be lower for WNS-affected M. lucifugus in West Virginia post-WNS. The timing of our long-term surveys precludes the ability to detect changes in the proportion of juveniles across years, but caution must also be taken in assessing reproductive rates from maternity colonies and extrapolating to populations (Barclay et al., 2004). Thus, a combination of summer mist-net monitoring combined with maternity colony studies may be required to gain a fuller understanding of the reproductive fitness of recovering M. lucifugus populations in a given area.
As compared to the laboratory and field studies that have been conducted on M. lucifugus, including those examining the impacts of WNS, there is a paucity of data regarding the impacts of WNS on reproduction or body condition of M. septentrionalis. Our results indicated the proportion of reproductively active M. septentrionalis females was nearly identical pre-and post-WNS, suggesting that TA B L E 1 Coefficient estimates (β) and 95% confidence intervals (CI) for effects of white-nose syndrome (WNS;pre [2003-2008 and post [2011][2012][2013][2014][2015][2016][2017][2018][2019]) on body mass of female Eptesicus fuscus (big brown bat) and Myotis septentrionalis (northern long-eared bat) in the Monongahela National Forest (MNF), West Virginia, USA. Models include an interaction effect between WNS invasion status and reproductive state (nonreproductive, pregnant, lactating, and postreproductive). The intercept represents predicted body mass (g) for the pre-WNS nonreproductive state, and coefficients represent predicted change in body mass from the intercept for each parameter and interaction. Based on likelihood-ratio tests, the WNS predictor was supported for both species, but WNSassociated coefficient CI's overlapped 0. Coefficient estimates with a CI that did not overlap 0 are bolded  (Stepanian & Wainwright, 2018). Climate change is likely a cause for at least some of these observed changes in phenology.
However, WNS also could be playing a role in these shifts for WNSaffected species. Ineson (2020) also found some evidence for an advance in reproductive timing for M. lucifugus post-WNS independent of climate, though both spring temperature and precipitation also contributed to changes in reproductive phenology.
Earlier reproduction could be beneficial for WNS survivors that do attempt to raise pups as young born earlier in the summer may have a higher probability of surviving their first year . However, environmental changes associated with global climate change in some regions (e.g., higher precipitation rates during foraging hours or lower precipitation rates leading to fewer water sources, decreases or changes in phenology of insect resources, decreases in optimal or suitable roosts, unsuitably hot or cold temperatures) also could exacerbate WNS-related bat population declines (Hayes & Adams, 2017

| Management implications
Given the higher physiological demands of Pd-infected species in spring and early summer, the ability to quickly heal wing dam- Habitat management focused on helping WNS-affected bats achieve energy balance upon emergence from hibernacula could be critical (Fuller et al., 2020). Active-season habitat management includes foraging habitat as well as roosting habitat and can range from sound forest management practices to providing roosts that minimize thermoregulatory costs of recovering bats (Wilcox & Willis, 2016). While habitat preferences are not uniform across North American bat species, some forest habitat features are beneficial for most WNS-affected species, and guidance is available to help land managers meet those habitat needs at a variety of spatial scales (e.g., Johnson et al., 2018;Silvis et al., 2016). Given the need to better understand the impacts of WNS during the nonhibernation season and the eagerness of land managers to proactively help in the fight to save WNS-affected bats, research and management focused on posthibernation recovery and enhancing reproductive success in WNS-affected bats could yield population-level benefits, especially in areas where WNS has long been established and the focus is on population recovery.

ACK N OWLED G M ENTS
We thank the many MNF biologists and technicians, volunteers, and

CO N FLI C T O F I NTE R E S T
None declared.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data used in this study are archived in the Dryad data repository: https://doi.org/10.5061/dryad.r4xgx d2cv.