Disease‐related population declines in bats demonstrate non‐exchangeability in generalist predators

Abstract The extent to which persisting species may fill the functional role of extirpated or declining species has profound implications for the structure of biological communities and ecosystem functioning. In North America, arthropodivorous bats are threatened on a continent‐wide scale by the spread of white‐nose syndrome (WNS), a disease caused by the fungus Pseudogymnoascus destructans. We tested whether bat species that display lower mortality from this disease can partially fill the functional role of other bat species experiencing population declines. Specifically, we performed high‐throughput amplicon sequencing of guano from two generalist predators: the little brown bat (Myotis lucifugus) and big brown bat (Eptesicus fuscus). We then compared changes in prey consumption before versus after population declines related to WNS. Dietary niches contracted for both species after large and abrupt declines in little brown bats and smaller declines in big brown bats, but interspecific dietary overlap did not change. Furthermore, the incidence and taxonomic richness of agricultural pest taxa detected in diet samples decreased following bat population declines. Our results suggest that persisting generalist predators do not necessarily expand their dietary niches following population declines in other predators, providing further evidence that the functional roles of different generalist predators are ecologically distinct.


| INTRODUC TI ON
Species tend to diverge and specialize over long periods of time, but generalist habits can be an asset in eras dominated by change.
On an evolutionary scale, specialization can lead to speciation, yet the persistence of generalism may offer a buffer against extinction (Dennis et al., 2011;Loxdale et al., 2011). Flexible resource requirements are often considered advantageous in the rapidly changing environments of the Anthropocene (Boyles & Storm, 2007;Colles et al., 2009;Purvis et al., 2000). As certain species decline, the question of whether persisting sympatric species can serve as ecological replacements to maintain interaction networks becomes increasingly important (Parker et al., 2010;Rubenstein et al., 2006;Tylianakis et al., 2010). Whether extant persisting taxa can compensate for the functional roles of other declining populations is generally unknown in aerial arthropodivores, largely due to methodological challenges associated with characterizing the consumption of diverse prey communities prior to population declines. Anatomical and physiological constraints often limit the flexibility of generalist predator foraging.
For example, among arthropodivorous bats, body size, flight agility and maneuverability, bite force, and echolocation call frequencies influence the type of prey that can be captured (Aguirre et al., 2003;Barclay & Brigham, 1991). Some bat species are capable of consuming prey larger than their own body size and will occasionally land and consume the preferred parts of the prey while discarding the rest (O'Shea & Vaughan, 1977;Santana & Cheung, 2016). Since most arthropodivorous bats rely on echolocation while hunting, certain types of echolocation are also adapted for specific prey and may play a role as a factor-limiting dietary niche breadth and flexibility (Arbour et al., 2019). For example, while echolocation frequency is related to the habitats in which different bats forage, the diets of bats that use lower-frequency echolocation are also more likely to include larger prey, while the diets of bats that use higher-frequency echolocation are more likely to include smaller prey (Denzinger & Schnitzler, 2013;Jones & Holderied, 2007). Such physiological constraints are generally considered to be more important determinants of bat trophic niches than direct competition with other bats in similar guilds (Schoeman & Jacobs, 2011). As such, the morphological constraints that shape bat foraging strategies call into question the potential for ecological equivalency among taxa.
In North America, many hibernating bat species have experienced rapid and precipitous declines due to white-nose syndrome (WNS), a disease caused by the fungus Pseudogymnoascus destructans (Frick et al., 2016;Lorch et al., 2011). Bat population declines from WNS present a unique circumstance under which basic ecological questions about the role of arthropodivorous bats as predators may be answered. In the eastern region of North America where WNS has been present for over a decade, little brown bats (Myotis lucifugus, Leconte 1831) have declined by more than 90% (Frick et al., 2016). Comparatively, other common bat species such as big brown bats (Eptesicus fuscus, Palisot de Beauvois 1796) experience infection from WNS yet have declined with much less severity (Frank et al., 2014;Frick et al., 2010). Additional studies have shown that some overlap in prey resource usage occurs between these two species in this study region, with molecular methods indicating a 21.3% overlap in OTUs detected in diets (Wray et al., 2021) and stable isotope methods indicating a 45% overlap between dietary profiles . These observed patterns, combined with the apparent lesser severity of big brown bat mortality due to WNS, raise the question of the extent to which persisting taxa may further expand their dietary profiles to include prey resources formerly consumed by other arthropodivorous bats that experience more severe population declines.
In this study, we quantified changes in dietary composition and niche overlap in two generalist arthropodivorous bat species following the rapid, WNS-induced decline in one species (little brown bats) and the persistence of another (big brown bats). Specifically, we tested whether declines in little brown bats would lead to dietary niche expansion among big brown bats by comparing changes in prey consumption, including agricultural pests, as measured using high-throughput amplicon sequencing methods.

| Study species
Little brown and big brown bats are among the most common bat species in North America (Fenton, 1980;Kurta & Baker, 1990). Little brown bats are higher-frequency echolocators that tend to be generalist in their foraging habits, mostly consuming aquatic insectsparticularly those with swarming behaviors (such as chironomid midges)-although they also consume terrestrial prey including moths, true bugs, beetles, and spiders (Clare, Symondson, Broders, et al., 2014;Whitaker & Lawhead, 1992;Wray et al., 2021). In contrast, big brown bats are lower-frequency echolocators and are often speciously referred to as "beetle specialists" but are well known to consume a variety of other arthropods such as flies, caddisflies, true bugs, and moths (Agosta, 2002;Wray et al., 2021).
We selected little brown bats as a focal study species because, prior to WNS-related declines, they were the most abundant bat species in the study region (Huebschman, 2019). We selected big brown bats as a second focal study species because they are also abundant in the region but were expected to decline less from WNS (based on previous population trends observed in the eastern region of North America, e.g., Cheng et al., 2021), thus allowing for a comparative analysis of the effects of disease-related bat declines in differentially afflicted species. During the breeding season, both little brown and big brown bats also frequently form large maternity roosts within human-built structures (Voigt et al., 2016), which allows for ease of detection and monitoring between years.

| Bat guano collection and detection of arthropod DNA
We collected roost-level bat guano samples (e.g., multiple pellets from multiple individuals at the same location) weekly at five little [IQR] = 27, 32) and 29 for the second 2 years (IQR = 26, 32). Following collection, samples were stored on wet ice during transport and subsequently kept at -80°C for long-term storage. Additional samples were collected in the same manner by community scientists at three or four time periods during the summer from 2015 to 2018, yielding an additional two little brown and two big brown bat roost sites ( Figure 1a). These samples were initially stored at -20°C, then shipped overnight with a frozen ice pack and kept at -80°C for longterm storage. This sample collection effort builds upon a previous study (Wray et al., 2021), with two additional years of data and five additional roost sites (n = 296 additional samples collected). The identity of bat species was confirmed by directly observing bat appearance (roosting bats were visible at all sites) and by comparing

| Sequencing arthropod COI isolated from bat guano
DNA extraction, PCR, and high-throughput amplicon sequencing followed Jusino et al. (2019) with the same modifications presented in Wray et al. (2021). Briefly, DNA was extracted from an 80mg subsample (~8 pellets) using a Qiagen DNA Stool mini kit (Qiagen Inc.), with a 180-bp region of the COI subunit c amplified using PCR with ANML primers (Jusino et al., 2019). Thermocycler parameters followed Hebert et al. (2003), with the exception of the final extension at 72°C increased from 5 to 7 min. A single-copy mock community of 34 known arthropod constituents was also amplified under the same conditions as a positive control (Jusino et al., 2019). Negative controls were included during each extraction batch and for each set of PCR reactions, which were visualized on a 2% agarose gel and did not yield any visible bands. Similarly, positive controls were also included for each set of PCR reactions, which did yield visible bands. PCR products were purified using a Zymo Select-a-Size Clean & Concentrator kit (Zymo Research), and then quantified using a Qubit 2.0 fluorometer (Invitrogen) with a high-sensitivity dsDNA kit. Following quantification, the purified PCR products were then equilibrated. Five total equimolar libraries were then constructed, each with approximately 72 samples per library. Samples were processed in a randomized order to reduce potential batch processing biases. Sequencing was performed on an Ion Torrent Personal Genome Machine platform (PGM; ThermoFisher Scientific Inc.) according to the manufacturer's recommendations with an Ion PGM 318v2 chip. Raw sequence data were then processed using AMPtk v1.4.2 (Palmer et al., 2018). This data processing procedure includes de-multiplexing using unique barcode index sequences, stripping of forward and reverse primers, and quality filtering and denoising with the DADA2 algorithm (Callahan et al., 2016). The resulting amplicon sequence variants (ASVs) were then clustered at 97% similarity using the UCLUST algorithm employed in VSEARCH to generate operational taxonomic units (OTUs; Jusino et al., 2019). Demultiplexed sequences were mapped back to these OTUs to generate an OTU table. Taxonomy was then assigned using the built-in COI database in AMPtk. Finally, we removed all OTUs that were not identified as insects or arachnids, as well ectoparasites including mites (Mesostigmata and Trombidiformes), acari (Sarcoptiformes), and fleas (Siphonatera), which do not represent typical prey items.

| Data cleaning and statistical analyses
Post-processing, OTU tables were converted into weighted percent occurrence (wPO), a presence-based metric, and relative read abundance (RRA), a read-based metric, following Deagle et al.
(2019). Since reads do not necessarily reflect abundance or biomass, occurrence-based metrics are often considered more conservative, yet can be sensitive to overinflating the influence of rare taxa (Deagle et al., 2013(Deagle et al., , 2019. For this reason, we present both read-based and occurrence-based metrics for visual and qualitative comparison but use wPO for statistical analyses. Within major arthropod orders (Araneae, Coleoptera, Diptera, Ephemeroptera, Hemiptera, Hymenoptera, Lepidoptera, and Trichoptera), we com-   (Hurlbert, 1978;Levins, 1968;MacArthur & Levins, 1967). To visualize diet communities in multivariate space, we performed non-metric multidimensional scaling (NMDS) using the metaMDS function in the R package "vegan" (Oksanen et al., 2013) with a modification of the raupcrick function as described by Chase et al. (2011), which was performed separately for presence/ absence matrices at the OTU and family levels. For both taxonomic levels, we excluded any taxon groups that were not detected at least five times in order to reduce the influences of infrequently detected diet items. To focus on general trends in community composition, outliers were also sequentially removed following a visual inspection of NMDS plots (n = 12 samples removed). To assess whether the bat diet communities differed by species, time period, collection site, and ordinal week, we used non-parametric PERMANOVA tests (Anderson, 2001) that were performed by the "adonis" function with 999 replicates and assessed the influences of multivariate dispersion (Anderson, 2006) using the "betadisper" function to separately test each predictor variable. We also searched taxonomy tables for certain arthropod taxa that are known agricultural pests in the area. To compare the incidence and taxonomic richness of pest taxa detected in samples collected between 2015 and 2016 with samples collected between 2017 and 2018, we used chi-squared tests and Welch's ttests, respectively, which were conducted separately for each bat species. All post-OTU table data cleaning and analyses were conducted in R version 4.1.0 (R Core Team, 2021). Additional packages used for data processing and visualization include "dplyr", "ggplot2", "tidyverse", "wesanderson", and "reshape2" (Ram & Wickham, 2018;Wickham, 2007Wickham, , 2016Wickham et al., 2019).  Table 1). The most common prey, as measured by wPO, and relative reads were generally consistent, except for big brown bats where Trichoptera: Hydropsychidae had a much higher RRA in comparison to wPO (Table 1).

| RE SULTS
Overall, we found that little brown and big brown bat dietary composition was distinct, and intraspecific differences in dietary composition did not change substantially between pre-and post-WNS time periods for either bat species. Little brown bats diets contained a higher richness of Diptera, Lepidoptera, and Hemiptera, and big brown bat diets contained a higher richness of Coleoptera.
The mean wPO within arthropod orders differed by bat species but remained similar between time periods within each bat species ( Figure 1b; Table 2). Hemiptera were less common in both little brown and big brown bat guano samples in the later time periods, and for little brown bats, other prey groups did not change (Figure 1b; Table 2). In the later time period, Hymenoptera were marginally less common in big brown bat guano samples ( Figure 1b; Table 2). The arthropod families most commonly detected for each bat species also remained similar between time periods (Table 3).
In general, little brown and big brown bats showed interspecific differences in family-level dietary niche breadth, which did not change substantially between the pre-and post-WNS time periods.
Little brown bats displayed higher niche breadth in comparison to big brown bats, and total interspecific niche overlap was 0.281 (Table 4).  (Table 4).
Interspecific dietary niche overlap was similar between time periods, increasing by only 2.3% from 0.281 to 0.288 (Table 4). NMDS plots indicated visually that diet composition differed more between species than between time periods (Figure 2a). PERMANOVA demonstrated that species was the best predictor of variation at the family and OTU levels ( Figure 2a; Table 5). At both the family and OTU levels, ordinal week was also a significant predictor of variation, while collection site and time period were also significant predictors at the OTU level (Table 5). At both the family and OTU levels, there were significant differences in multivariate dispersion between bat species and between collection sites, and at the OTU level, there were also significant differences in multivariate dispersion between time periods and ordinal week (Figure 2a; Table 5). These results suggest that while bat species was the most important factor in determining diet community structure and dispersion, other factors were also somewhat influential, particularly at the OTU level. prior to WNS-related population declines, which is also consistent with a long-term study using stable isotope analyses . The observed decrease in niche breath following declines in roost sizes suggests that both bat species may display some degree of individual specialization (Bolnick et al., 2002(Bolnick et al., , 2003, with the individuals within a population consuming different prey resources that ultimately contribute to a broader population-level dietary niche.

Order
The decreasing prevalence of agricultural pest arthropods in bat diets following WNS-related declines may be an artefact of overall decreasing dietary niche breadth but could also suggest that individuals select for other prey items, potentially as a result of reduced inter-or intraspecific competition. These results are consistent with one previous study which also suggested that big brown bat dietary niche breadth may be driven by individual specialization (Cryan et al., 2012), although the contributions of individuals to population-level dietary breadth have been seldom explored in arthropodivorous bats despite evidence of within-population variation in several other taxa (Johnston & Fenton, 2001).
While niche breadth decreased for both bat species, niche overlap did not change, but rather, some niche overlap (0.281) was observed prior to declines in bat populations. One previous study, which compared pre-and post-WNS dietary composition based on stomach contents of bat carcasses, found that overlap in dietary composition did increase following WNS-related bat declines and suggested that this may indicate increasing competition (Morningstar et al., 2019).
Considering the observed variation between sites and between weeks, as well as the known high spatial and temporal turnover in the diets of both bat species (Wray et al., 2021), the comparison of TA B L E 2 Changes in mean weighted percent occurrence (wPO) of prey orders between time periods pre-and post-WNS bat diets may not be appropriate for individual bats collected at different sites or during different seasonal time periods. Interannual variation in diets may also influence conclusions for studies with limited or uneven year-to-year sampling schemes.
Separate studies also suggested that pre-and post-WNS changes in bat acoustic activity could indicate shifts in temporal and spatial TA B L E 3 Top 20 family-level prey items detected in big brown and little brown bat guano samples, ranked by weighted percent occurrence (wPO). Changes in top family-level prey items between time periods are highlighted in bold niche partitioning (Jachowski et al., 2014) or reduced interspecific competition (Mayberry et al., 2020). However, just as co-occurrence does not necessarily imply the presence of ecological interactions (Blanchet et al., 2020), decreasing habitat or dietary niche overlap does not necessarily imply decreasing competition. The principle of competition relies on the supposition that resources shared by two   species must be limiting for competition to occur, and coexistence has been shown to persist under many cases where the assumptions of competitive exclusion are not met (Chase et al., 2002;Holt, 1977).
As such, the results of our study suggest that WNS-related declines in little brown bats likely do not lead to increases in niche overlap in this study area, but rather demonstrates that interspecific niche overlap between little brown and big brown bats has remained fairly consistent in recent time.  (Agosta, 2002). We observed complete roost abandonment at two little brown bat roost sites by 2018, and if these roosts were later adopted by big brown bats, it may be unlikely that little brown bats could reoccupy them upon population recovery. As such, further exploration into the potential competition between little brown and big brown bats for roost space or other habitat requirements may be warranted and are potentially more important than limitations due to food resources.
We characterized the functional role of bats as predators with the goal of assessing the extent to which widespread, flexible, and comparatively successful sympatric species have the potential to serve as ecological replacements for other declining species. While big brown bats likely cannot fully fill the trophic role of little brown bats, possibly due to morphological or other physiological constraints, other bat species may be more ecologically similar to little brown bats. However, most of these species also experience severe declines due to WNS, and non-affected species such as migratory bats do not cluster in large colonies in this study region (Huebschman, 2019), and as such probably do not influence prey communities in the same manner. While other studies have demonstrated the successful reintroductions of extirpated predators leading to restoration ecosystem functioning (e.g., Mittelbach et al., 1995;Ripple & Beschta, 2012), such efforts often rely on the possibility of conservation strategies such as captive breeding or translocation-none of which have successfully been implemented for little brown bats or other bat species severely affected by WNS (Davy & Whitear, 2016).
As such, it is unlikely that the functional role of the little brown bat can be restored either naturally or through management strategies, and several bat species severely impacted by WNS are expected to face extirpation in many regions (Frick et al., 2010;Thogmartin et al., 2013). The growing body of evidence regarding the function of arthropodivorous bats as ecologically important predators thus raises serious concerns regarding potential top-down consequences of WNS-related bat declines. These findings highlight the importance of continuing to support little brown bat population recovery, while also emphasizing the need for conservation of bats and other aerial arthropodivores in general due to the probability that each unique species cannot necessarily be replaced by another.  Finally, we thank two anonymous reviewers for their helpful suggestions for improving an earlier version of this manuscript. This study was funded by USDA Hatch Formula Fund WIS01841.

CO N FLI C T O F I NTE R E S T
The authors declare no conflicts of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The final OTU