White-nose syndrome-related changes to Mid-Atlantic bat communities across an urban-to-rural gradient

Background White-nose Syndrome (WNS) has reduced the abundance of many bat species within the United States’ Mid-Atlantic region. To determine changes within the National Park Service National Capital Region (NCR) bat communities, we surveyed the area with mist netting and active acoustic sampling (2016–2018) and compared findings to pre-WNS (2003–2004) data. Results The results indicated the continued presence of the threatened Myotis septentrionalis (Northern Long-eared bat) and species of conservation concern, including Perimyotis subflavus (Tri-colored bat), Myotis leibii (Eastern Small-footed bat) and Myotis lucifugus (Little Brown bat). However, we documented a significant reduction in the abundance and distribution of M. lucifugus and P. subflavus, a decrease in the distribution of M. septentrionalis, and an increase in the abundance of Eptesicus fuscus (Big Brown bat). Conclusions Documented post-WNS M. septentrionalis recruitment suggests that portions of the NCR may be important bat conservation areas. Decreases in distribution and abundance of P. subflavus and M. lucifugus indicate probable extirpation from many previously occupied portions of the region.

White-nose Syndrome does not impact all bat species equally. Roosting behaviors, body sizes, hibernacula colony sizes, environmental conditions of particular hibernacula, WNS exposure levels and disease virulence can positively or negatively affect survival rates following exposure to WNS [24,25]. Although WNS can negatively impact Eptesicus fuscus (Big Brown bat) [19], many individuals appear to be unaffected or are able to survive at higher rates than other species, perhaps because their larger body size allows for greater fat reserves than smaller bat species [26], or they are hibernating within warmer anthropogenic roosts unfavorable to growth of P. destructans [27]. Despite high mortality rates, remnant populations of M. lucifugus, M. septentrionalis, M. sodalis, and P. subflavus can persist in WNS-affected areas [28][29][30].
When populations experience declines or extirpation, competitive release may result in other species altering their habitat use, increasing their range, or increasing their population size; for example, some relaxation in niche partitioning has been observed for bats in WNSimpacted landscapes [31]. Along the East Coast, increases in E. fuscus, Lasiurus borealis (Eastern Red bat), and Lasiurus cinereus (Hoary bat) have corresponded with decreases in M. lucifugus, M. septentrionalis, M. sodalis and P. subflavus [12,19,31]. However, total intra-night activity indicated that other bat species were not expanding into temporal niches formerly dominated by M. lucifugus post-WNS [23,31].
Urban environments can increase contact between individuals because of highly-concentrated resources, such as limited foraging patches and limited day-roost availability, or impact individuals' physical condition through use of marginal day roosts, increased competition for food resources, and pollution [32]. Urbanization can impact bat activity patterns and foraging ecology [33]. Therefore, urbanization may exacerbate winter WNS population declines, as better body condition entering hibernation can increase WNS survival [34,35] or compound WNScaused population declines with urbanization-caused decreases in reproductive success [12,15,29,36]. Additionally, highly-urbanized areas often have lower bat diversity and activity levels than less-developed, rural, semi-natural, or natural areas [36][37][38][39][40][41][42][43].
Hibernacula near the NCR have shown large declines in M. lucifugus, M. sodalis, and P. subflavus populations [17,44]. Accordingly, we sought to document how summer bat communities and habitat associations in the NCR have changed since WNS onset, in terms of overall bat activity and changes to community membership. In 2003-2004, Johnson et al. [45] surveyed bat communities within 11 NCR National Park Service (NPS) units to assess community patterns along an urban-to-rural gradient using a combined active acoustic and mist-netting sample approach. Both methods allow for detection of bat species and evaluation of relative activity or abundance. Acoustic sampling records bat echolocation calls that can be identified to species, and mist netting captures and identifies individual bats [46].
Our first objective was to compare the sampling effort required to document presence of species pre-and post-WNS. We expected that the effort required to document the same species would be higher post-WNS than pre-WNS [46]. Our second objective was to determine the impacts of WNS on bat communities within the NCR by considering relative abundance measures and distribution of detections in the NCR. We predicted shifts in NCR community structure relative to the pre-WNS findings of Johnson et al. [45], including possible population decreases, population extirpations, and reduced distribution of species highly impacted by WNS (M. lucifugus, M. septentrionalis, M. sodalis, and P. subflavus). Competitive release of larger-bodied, less-affected species, such as E. fuscus, or non-hibernating migratory species, such as L. borealis, could result in increased abundance or changes in habitat associations. Similar to Johnson et al. [45], we predicted that the highest species richness would be found in less-fragmented, rural habitats with more forest cover, as fewer bat species occurred within higher urbanization levels.

Study area
Our study area consisted of the 11 NPS units that Johnson et al. [45] [16].
The region includes forests, pasture areas, suburban development, and high-density urban development. The study area largely has an east-west (and, to a lesser extent, south-north) urban-to-rural gradient. The most rural portions of the study area include CATO (a contiguous forest unit surrounded by both private and public forestland) and northwestern portions of CHOH in central and western Maryland. As the longest unit with the most habitat variation, CHOH follows the D.C. and Maryland border with Virginia and West Virginia from downtown D.C., through suburban development, agricultural fields, and private and state forests and wildlife management areas to Cumberland, Maryland. The CHOH is adjacent to HAFE (a mix of rural mountainous forest and open battlefields), near ANTI and MONO, and adjacent to small portions of GWMP. The battlefield units (ANTI, MANA, MONO) consist mainly of agricultural fields and small forest woodlots. As a performing arts center, WOTR is a combination of developed land and forest fragments surrounded by a developed suburban landscape. Near D.C., GWMP primarily follows the Virginia border and Potomac River. It contains larger contiguous forest stands (e.g., Great Falls Park, approximately 13 km north of D.C.) and linear forest strips along motorways. Units partially or entirely within D.C. (ROCR, CHOH, NAMA, NACE) are highly fragmented, with the exception of ROCR. This unit primarily consists of 2 large contiguous forested patches which follow 2 water courses to the Potomac River, Rock Creek and Foundry Branch [47]. Dominated by open areas and National Mall monuments, NAMA contains minimal forest cover. The parks of NACE include units on the southeastern bank of the Anacostia River, as well as a thin linear forest strip running along the Baltimore-Washington Parkway up to Greenbelt Park, a small contiguous forest stand northeast of the city.
Currently, the region is dominated by a mix of oakhickory and mesophytic forest types, e.g., Quercus spp. (oak species), Acer spp. (maple species), Liriodendron tulipifera (yellow poplar) and to a lesser extent Pinus spp. (pine species) [48]. Within the NCR units, forest stands primarily consisted of mature and late successional forests, more so than the surrounding landscape [47]. Deciduous tree species that dominate the upland areas on NCR units include Acer rubrum (Red Maple), Fagus grandifolia (American Beech), L. tulipifera, and Quercus spp. Riparian forests along larger watercourses contain Platanus occidentalis (Sycamore), Acer negundo (Boxelder), and Liquidambar styraciflua (Sweetgum).

Acoustic and mist-netting surveys
We replicated the active acoustic methods used by Johnson et al. [45] to allow a direct comparison of echolocation call data between 2003 and 2004 (pre-WNS) and 2017 (post-WNS). Using the same acoustic recording methods was important for directly comparing pre-WNS call data recorded to post-WNS call data, as the methods can impact detection and habitat associations [49]. We used Anabat II (Titley Electronics, Ballina, Australia) zero-crossing, frequency-division equipment to actively survey at each sample site for one 20-min period between sunset and 0200. If a bat was "observed" as evidenced by audible echolocation sounds from the recorder or by sight, we attempted to follow its direction with the microphone to record a high-quality call sequence [50]. Active survey sites were in forested, wetland, and open habitats with a mean distance between survey locations of 886 m ± 982 m [45]. We avoided active acoustic recording and netting during heavy precipitation or low temperatures, as these can reduce bat activity [51].
In Department of Natural Resources #56152, Virginia Department of Game and Inland Fisheries #061519, and West Virginia Division of Natural Resources #2018.180 scientific collecting permits. We identified bats to species using a dichotomous key to the bats of this region [54], and recorded sex, age (adult or juvenile), and reproductive condition [55,56]. We disregarded samples when heavy precipitation or fog event (as defined by onthe-ground observations) disrupted sampling.

Data analysis
We developed species accumulation curves for both our acoustic and netting datasets, which provided an assessment of species richness based on the level of sample effort [57]. We developed species accumulation curves for each pre-or post-WNS dataset using the R package vegan with random selection [58]. In the two acoustic species-accumulation curves, we used all samples and 10 acoustically-detected species. For the pre-and post-WNS species accumulation curves based on mist-net sampling, we selected the nights with at least 4 h of netting and no precipitation within the netting event, based on on-the-ground observations. We created 2 sets of species accumulation curves for the netting data: 1 with the 6 species that were found both pre-and post-WNS and 1 for all 9 species captured post-WNS.
To determine whether the number of echolocation calls and number of bat captures for each species changed based on WNS, we performed paired and unpaired t-tests in program R [59][60][61]. We used paired t-tests for all acoustic sites that were sampled both pre-and post-WNS. We used unpaired t-tests for analysis of all netting sample data because we did not resample the exact same locations pre-and post-WNS. The netting t-tests used nightly captures per net to account for higher netting efforts post-WNS.

Results
We recorded 4967 echolocation calls (2003-2004 = 3637; 2017 = 1330), which were identified to 10 species (Table 1). Species richness declined in every NPS unit except NACE, with all units losing detections of at least 1 species severely impacted by WNS (Table 2) (Tables 2 and 3). Post-WNS, we captured M. lucifugus and P. subflavus only in northwestern Maryland (CHOH), and their absences from all other units were the primary drivers to post-WNS decreases in unit-level species richness for both netting and acoustics data ( Table 2). We noted no evidence of reproduction in the M. lucifugus and P. subflavus captures, and we only documented evidence of reproduction (pregnant and lactating females) and successful recruitment (juvenile bats) in M. septentrionalis populations within ROCR.
Acoustic species accumulation curves indicated that more effort was required to detect all species post-WNS than pre-WNS, despite more capture success post-WNS. Pre-WNS, 94 active acoustic samples were required to document a species richness of 10 (SD ± 0), whereas post-WNS, the species richness was 9.31 (SD ± 0.68) for the same effort (Fig. 2). Even with the focused post-WNS surveys, pre-WNS netting required 65 netting nights for a species richness of 6 (SD ± 0), but that effort for the same 6 species post-WNS yielded a species richness of 5.07 (SD ± 0.76; Fig. 2). For the post-WNS netting species accumulation curve of all 9 extant species captured, 65 nights yielded a species richness of 7.28 (SD ± 1.20; Fig. 2). All 3 post-WNS species accumulation curves required almost every sample included in the analysis in order to reach the maximum species richness with a standard deviation of 0 (acoustics = 147, both netting = 109).
Based on t-tests and overall detections, we observed significant post-WNS increases in E. fuscus and significant distribution decreases occurred for M. lucifugus and P. subflavus (P < 0.05; Tables 1 and 3). We also observed significant post-WNS increases in the mean number of captures of Lasionycteris noctivagans and decreases in the recorded acoustic calls for M. leibii and M. sodalis (Tables 1 and 3). Both pre-and post-WNS netting results indicated that species richness was highest in western portions of CHOH, where the NCR is the least urbanized and contains the highest proportions of rural and forested landscapes. However, we did not determine a uniformly-negative trend with regard to urbanization, as we also recorded high species richness in highly-urban units. We captured all M. septentrionalis in ROCR, except for a single capture in CATO; 76.9% of L. noctivagans in ROCR (n = 10); and all but 1 N. humeralis in GWMP (Great Falls Park). Eptesicus fuscus were active throughout the NCR, with the least activity coinciding with open, agricultural, or suburban portions of CHOH, MONO, and MANA. Most activity was in the D.C. area within the larger contiguous forested portions of ROCR and NACE. We captured more Lasiurus borealis west of Great Falls Park (along the geological fall line between the Piedmont and Coastal Plain); however, distribution varied greatly across the landscape.

Discussion
We determined that active acoustic sampling, as employed in this study, was not suitable for detecting rare bat species on the current post-WNS landscape. Our species accumulation curves reflect decreases in WNS-impacted bat species and higher variation in detections. Similarly, other studies have reported that post-WNS sampling effort should be higher than pre-WNS efforts [46,62]. The small sampling window for active acoustic methods likely limited detection, especially post-WNS. The limitations of our active acoustic sampling are underscored by not observing post-WNS M. septentrionalis echolocation calls in ROCR, whereas our mist-netting results indicated continued presence. There was also a non-significant decline in M. septentrionalis echolocation calls across the NCR, despite our capture data indicating a significant decline in post-WNS occurrence. Although our active acoustic sampling techniques may continue to be options for comparing community activity to historical data [46,63] or for detecting common bats, the sampling period for these methods appears inadequate to fully document post-WNS bat communities. We recorded significant declines in relative abundance and distribution of WNS-impacted species M. lucifugus and P. subflavus. These results are similar to those of other summer surveys comparing pre-to post-WNS activity within the northeastern U.S. [21][22][23]. Though our    [21], 43% increases in a New York hibernaculum 2-3 years after WNS detection [26] and 12% increases in captures approximately 1-3 years after WNS detection in Indiana [29]. Eptesicus fuscus populations may have avoided WNS-related declines based on hibernation conditions, e.g., uninfected hibernacula, temperature within hibernacula or shorter hibernation periods, as well as being physiologically less challenged by or resistant to P. destructans [24,25,64,65]. Lasiurus borealis activity was similar between preand post-WNS sampling periods, indicating no expansion into WNS-emptied habitat niches [31] or potential wind-energy driven population declines [66]. Our results are similar to those of pre-WNS versus post-WNS comparisons of bat communities in South Carolina and New York [23,67].
Whereas urbanization influenced some species' distributions, it has not appeared to have exacerbated WNS declines. We found our only evidence of recruitment among highly WNS-impacted species in D.C. Some combination of disease resistance, summer habitat, overwintering patterns and a small pre-WNS population of WNS-susceptible species may have led to ROCR being a potential M. septentrionalis refugium within the Mid-Atlantic region. Pre-WNS, both M. septentrionalis and M. lucifugus captures were relatively rare in ROCR compared to most other units, yet M. septentrionalis populations appear to be successfully reproducing only within ROCR. Although M. septentrionalis' survival mechanisms are not yet known, remnant M. lucifugus populations along the East Coast have exhibited genetically based resistance and avoidance behavior that may reduce WNS impacts [28,68]. As a mostly-contiguous unit with oldergrowth, late-successional forest [47], ROCR contained a diverse mix of tree sizes and a high density of large standing snags and trees with cavities and bark conditions conducive to M. septentrionalis day-roosting. Given the mature forest, year-round water sources, and surrounding established neighborhoods with numerous tree and anthropogenic day-roost opportunities, ROCR may provide an optimal balance of forest and edge habitat, access to water resources, and both natural and anthropogenic roosting and hibernacula options. Recent overwinter detections of M. septentrionalis along coastal areas to the southeast of ROCR [69][70][71][72] provide evidence that some individuals are not migrating to P. destructans-infected hibernacula to the west in the Appalachian Mountains. The authors' capture of a M. septentrionalis in ROCR on 1 November 2018 provides additional support for this hypothesis. In the more easterly urban areas, we observed slightly higher E. fuscus activity overall. Our E. fuscus results are similar to those of a landscape-scale survey in Georgia and the Carolinas that reported higher E. fuscus abundance within NPS units with high levels of developed land within 5 km [73]. Still, within the D.C. area, we saw higher activity within larger forest patches, similar to observations in Mexico City [38]. We suggest caution regarding the interpretation of pre-and post-WNS comparisons in the NCR and other WNS-affected regions. When comparing post-WNS communities to pre-WNS communities, it is necessary to be cognizant of how improved sampling methods have potentially increased detection and capture rates, whether through compliance with USFWS M. sodalis survey guidelines [52] or technological improvements (e.g., passive vs. active acoustics, use of automated identification software, netting equipment). Documented bat echolocation calls and capture rates also may have been influenced by the habitat structure in more urban, fragmented forest stands, with travel corridors concentrating bat activity more than in largely forested landscapes. Moreover, annual weather variation may have influenced differences in acoustic and capture data as well as yearto-year demographic changes [74][75][76]; 2018 had the most netting samples and had higher levels of precipitation than any other year. The differences in pre-and post-WNS netting species accumulation curves are likely greater than the species accumulation curves demonstrate, as differences in netting effort on each sample night were unaccounted for.

Conclusions
Based on the documented declines in bat passes and capture rates of M. lucifugus and P. subflavus populations, these species may be functionally extirpated throughout much of the NCR. Acoustic echolocation calls and netting capture rates for E. fuscus increased significantly. Consequently, long-term monitoring would be required in the region to determine E. fuscus population increases, and intensive future surveys would be required to assess possible habitat expansion of bats less impacted by WNS (E. fuscus and L. borealis) into vacant niches. Furthermore, additional assessments in the Mid-Atlantic region are needed to determine whether or how M. septentrionalis persistence is linked to atypical overwintering activities, such as use of coastal forests to the east and southeast of the NCR rather than cave hibernacula, high-quality maternity habitat conditions in residual NCR forest patches, or other genetic or behavioral resistance mechanisms.