Genetic Mark–Recapture Improves Estimates of Maternity Colony Size for Indiana Bats

Advances in the application of molecular genetic techniques to wildlife biology have made it possible to individually identify animals using DNA as a unique mark. These methods rely on the fact that every individual has a unique molecular fingerprint, which can be used in the same way that a band or tag might be used in a traditional mark–recapture study. Further, this DNA fingerprint can be used to investigate family relationships by linking an individual to its offspring or parents. It is possible to obtain DNA from numerous sources of biological material (e.g., hair, feathers, shed skin, feces, urine, etc.); and thus, genetic information can be collected noninvasively without disturbing or harming animals, which is particularly useful for rare or cryptic species or those that are difficult to trap (Waits and Paetkau 2005).

Genetic mark–recapture methods are increasingly being used to estimate population sizes for many large terrestrial mammals (e.g., Chancellor et al. 2012; Ebert et al. 2012a; Davidson et al. 2014; Gray et al. 2014; Roy et al. 2014; Stansbury et al. 2014; Dumond et al. 2015; Stetz et al. 2015). In addition to population size, survival rates have been estimated using genetic mark–recapture for Arctic fox Vulpes lagopus (Meijer et al. 2008), brown bears Ursus arctos (Boulanger et al. 2004), wolves Canis lupus (Caniglia et al. 2012), and caribou Rangifer tarandus (Hettinga et al. 2012). The promise of these approaches, however, extends far beyond large mammals to a wide variety of species for which demographic information is needed and traditional mark–recapture methods are problematic.

The Indiana bat Myotis sodalis was listed as an endangered species in 1967 with full legal protection provided by the US Endangered Species Act (ESA 1973, as amended) on account of range-wide declines. Historical declines were likely the result of multiple factors including modification and disturbance of hibernacula (caves and mines where they hibernate [Humphrey 1978; Richter et al. 1993; U.S. Fish and Wildlife Service 2007]), loss of summer habitat (Brack 1983; Sparks et al. 2004, 2005; Kurta 2005), and impacts of pesticides (O'Shea and Clark 2002; U.S. Fish and Wildlife Service 2007; Eidels et al. 2016). At the time of this study, populations at northern hibernacula were stable or increasing while southern populations were declining (Clawson 2002; U.S. Fish and Wildlife Service 2007; Thogmartin et al. 2012; Eidels et al. 2016). Populations are now declining dramatically range-wide because of the unprecedented fungal disease white-nose syndrome (Turner et al. 2011; Thogmartin et al. 2012). To date, most demographic information is based on counts at winter hibernacula (Clawson 2002; Thogmartin et al. 2012, 2013), which are important for describing overall region-wide population size. However, important information on population size, reproductive success, and survival occurring in the summer months is lacking. The continuing threats to the species make the broader goal of improving the understanding of population dynamics and life history throughout the year an essential part of this species' protection and recovery. Nonetheless, field studies of Indiana bats during the summer have focused primarily on roosting and foraging habitat requirements (e.g., Carter and Feldhamer 2005; Kurta 2005; Menzel et al. 2005; Sparks et al. 2005; Bergeson et al. 2015) rather than population dynamics.

The shortage of available demographic information from the more extensive summer range is due to biological and corresponding methodological impediments. Like many other species of social mammals, including some bats (e.g., Kerth and König 1999; Willis and Brigham 2004), reproductive colonies of Indiana bats have a fission–fusion social organization (Kurta 2005). When females emerge from hibernation, they spread throughout most of the eastern half of the United States while males stay near the hibernacula (Brack 1983; Gardner and Cook 2002). Females raise their young in maternity colonies that occupy a series of roosts in separate large, usually dead or dying trees (e.g., Humphrey et al. 1977; Callahan et al. 1997; Kurta 2005; Bergeson et al. 2015). Colony members are often not all contained within a single roost tree on any given day; instead, they are split into subgroups that coalesce and subdivide among multiple roost trees spread across the landscape (e.g., Kurta 2005; Whitaker and Sparks 2008; Timpone et al. 2010). This biological complexity makes it difficult to estimate the numbers of bats that comprise a single maternity colony through more traditional techniques such as counts of exiting bats at emergence (Silvis et al. 2014).

Use of more intensive methodology for determining maternity colony size and dynamics may not be feasible or justifiable across large summer ranges because of cost, field effort, and the potential for harming or harassing bats (Kunz et al. 2009; Clement et al. 2015). Such methods include monitoring multiple bats at multiple roost trees through short-term radiotelemetry, banding and recapturing bats, or tagging bats with passive integrated transponders. We had four objectives in this study. For the first two objectives we investigated the potential use of DNA from fecal samples (fecal DNA) collected at a roost to obtain genetically based mark–recapture estimates of 1) colony size and 2) survival rates, for an Indiana bat maternity colony in Indianapolis, Indiana. The third objective was to compare our genetically based colony-size estimates with emergence counts conducted at the same roost tree to evaluate the genetic mark–recapture method. Our fourth objective was to use fecal DNA to estimate levels of relatedness among individuals sampled at the roost for comparisons with other studies that focus on relatedness within colonies of bats with fission–fusion social systems (Kerth et al. 2002; Metheny et al. 2008a, 2008b; Patriquin et al. 2013).

This study was conducted on properties in Marion and Hendricks counties, Indiana, managed for conservation of the Indiana bat by the Indianapolis Airport Authority as part of a Habitat Conservation Plan. The study area is now a nature park consisting of many small forest remnants enclosed in a matrix that grades from subdivisions in the north to agricultural fields in the south (Sparks et al. 2009). Also present at the southern end of the area is a complex of wetlands and forests protected by the Indianapolis Airport Authority to provide habitat for Indiana bats (Whitaker et al. 2004; Sparks et al. 2009). Conservation measures applied in these areas include preservation of existing mature forest, use of artificial roosts to supplement roosting habitat within these forests, and creation of new forest and wetland habitats. Effectiveness of these restoration efforts is monitored each year using a combination of emergence counts at known roost trees to estimate the number of bats present, along with capture and radiotelemetry of Indiana bats to identify new roosting and foraging areas. The monitoring program is described in Whitaker et al. (2004) and Sparks et al. (2009). Thus, this study area had the unusual combination of urban and suburban development, active restoration of native habitats, and an ongoing effort by private businesses and government to preserve remaining wild habitats within an urban matrix.

The current study leveraged the annual monitoring program described above in several ways. First, data from these efforts provided us with a well-studied population of Indiana bats where key roosting and foraging areas were already well-known (Whitaker et al. 2004; Sparks et al. 2009). Second, telemetry efforts and associated emergence counts at known roosts allowed us to collect samples at roosts that were known to be occupied during key times during the summer. The regular monitoring effort included emergence counts before, after, and between dates when we collected fecal samples and thus represent a season-long data set obtained using a traditional technique (emergence counts) with which we could compare the results of the genetic mark–recapture analysis.

As part of the monitoring program, biologists completed emergence counts (Kunz 2003; Whitaker and Sparks 2008; Kunz et al. 2009) at roosts known or suspected of being used by Indiana bats. These counts focused on primary roost trees (i.e., those roost trees that are used by large portions of the colony for multiple days). Biologists counted all active primary roost trees at least twice per week with at least one of these counts scheduled such that all active primary roost trees were counted on the same day. Biologists would arrive at a tree ≥30 min before expected emergence and continued counts until activity at the roost ceased or it had become too dark to observe the bats. If bats were seen, the biologist would record the time the first bat emerged, the time the last bat emerged, and the total number of bats that emerged. These counts were also supplemented by counts at trees occupied by radiotagged bats, and spot-checks of historically important but apparently abandoned roosts, including artificial roosts. Prior to completing an emergence count on an artificial roost, biologists would use a spotlight to check for occupancy, which also allowed them to count small numbers of bats when they were present. We used the emergence count data from the specific roost trees that we targeted for our fecal pellet collection.

We targeted three primary maternity roost trees (based on historical patterns of use) for fecal pellet collection during three time periods (11–18 June, 26 June–2 July, and 16–22 July) in the summer of 2008. Young become volant at the study area in mid-to late July (Whitaker and Sparks 2008): during sampling periods 1 and 2 fecal pellets collected below the roost should nearly all be from adult females, whereas during sampling period 3 some volant young were likely to be included in the sample. To collect pellets, we placed a guano trap (similar to those described in Murray and Kurta 2002) under each roost. Each guano trap was a 1-m² elevated wooden quadrat to which tulle (a fabric used to make bridal veils) could be easily attached and replaced with a staple gun. We collected pellets each morning (<12 h old) and froze them immediately upon collection. To avoid cross-contamination between individual pellets, we collected each pellet using a clean wooden toothpick and placed it in either a 1.5-mL tube or a clean plastic bag. We cleaned guano screens between sampling bouts. We collected 4,085 pellets across three primary roost trees throughout the summer of 2008. Only one of the roosts trees was used consistently throughout the three time periods, and thus we focused our analyses and inferences on bats using that roost tree. We collected 3,372 pellets at that primary roost tree. Individual bats likely defecate multiple pellets each day; therefore, we subsampled the total number of pellets so that we could minimize the number of duplicate samples (samples collected from the same individual) in the same day. To implement our subsampling process, we randomly chose 1,008 pellets from the primary roost tree such that we sampled 30% of pellets collected each day and included only those samples in our study. In accordance with a mark–recapture robust design (Kendall et al. 1997), which is a combination of the Cormack–Jolly–Seber live recapture model and the closed-capture models, we collected pellets from these traps every day during each of the three sampling periods.

We isolated DNA from 1,008 fecal pellets using a QIAamp DNA Stool Mini Kit (QIAGEN) following the manufacturer's protocol with the exception of a modified elution step (eluted in 60 μL Buffer AE after 5 min of incubation at room temperature). We performed all DNA extractions in a sterile biological safety cabinet to prevent contamination. To monitor for possible contamination, we looked for more than two alleles in each sample (evidence of cross-contamination), and we included negative controls without fecal material added each time an extraction was performed. We extracted samples in sets of either 11 (11 samples plus a negative extraction control) or 23 (23 samples plus a negative extraction control).

We amplified five microsatellite loci (IBat CA11, IBat CA47, IBat M23, IBat CA19, and IBat CA38) specifically developed for the Indiana bat (Oyler-McCance and Fike 2011). We chose these five loci because in a screen of 30 individuals they were highly polymorphic, ranging from 10 to 39 alleles/loci (Oyler-McCance and Fike 2011). Further, a calculation of the probability of identity for siblings with these loci revealed that the probability that even related individuals mistakenly would be considered the same individual was extremely low (0.0027). We amplified these five loci using the multiplex preamplification method described by Piggott et al. (2004), which is a two-step procedure that requires an initial Polymerase Chain Reaction (PCR) using multiple sets of primers with a final concentration of 0.01 μM. This initial step involved a set of unlabeled primers for all five loci and we performed it following the conditions outlined in Piggott et al. (2004) with the exception of using 10 μL fecal DNA as the template for the 50-μL reaction. The second step used 3 μL of the PCR product produced in the first step as template for 12.5μL reactions containing 100 μM each of dNTP, 1X GoTaq Flexi Buffer (Promega), 1.25mM MgCl₂, 1X BSA, 0.5 μM of each primer (dye-labeled forward), and 1 U of Taq DNA polymerase (Promega). The amplification conditions for the second step were as follows: 94°C for 2 min, then 94°C for 30 sec, annealing temp (60°C for IBat CA38 and IBat CA19, 58°C for the remaining three loci) for 30 sec, 72°C for 30 sec for 40 cycles, then 60°C for 45 min, and a final extension at 72°C for 10 min. For each sample, we combined PCR products from primers IBat CA11 and IBat CA47 in one well of a 96-well plate. Similarly, we combined PCR products from primers IBat CA19 and IBat 38. For IBat M23, we loaded PCR products into wells separately and did not combine them with any other PCR products. We added the S400 size standard (Beckman Coulter) to all wells on every plate. We ran plates on a CEQ8000 XL DNA Analysis System (Beckman Coulter) and analyzed them using the Frag 3 default method. We ran positive and negative controls on each plate to maximize quality and consistency of genotyping.

Fecal samples can have lower quantities of DNA that is often of lower quality, increasing the probability of genotyping error (Taberlet et al. 1999). To cope with such errors, DNA from fecal samples are amplified multiple times and genotypes from those amplifications are compared to identify and quantify errors. Potential errors can include false alleles (where the genotype obtained is incorrect as a result of PCR errors, problems with electrophoresis, or human error such as cross-contamination between samples) or allelic dropout (where one allele of a heterozygote does not amplify), and comparisons among amplifications can help identify these errors and determine the correct genotype (Broquet and Petit 2004).

We amplified each sample multiple times (at least twice, but often three or four times) at every microsatellite locus and compared genotypes across these multiple amplifications. If after two amplifications at a particular locus, a sample had a matching genotype, we retained that genotype. If not, we performed a third amplification and compared the three genotypes to check for a matching genotype. Mismatches consisted of either failure to amplify (no additional genotype with which to compare), amplifying genotypes with three or four different alleles when comparing two amplifications (evidence of false alleles), or having two genotypes where one was a heterozygote and the other was a homozygote with one of the two alleles identified in the heterozygote (evidence of allelic dropout). Often after three amplifications, we obtained a matching genotype; but if not, we performed a fourth amplification. Samples that did not have matching genotypes after four amplifications or that amplified only once were treated as missing data at that locus. To be conservative, we used only those samples that had genotypes at all five loci (i.e., no missing data) even though mark–recapture methods can deal with missing data.

We identified individuals using GenAlEx ver. 6.41 (Peakall and Smouse 2006) and used Program DropOut (McKelvey and Schwartz 2005) to assess allelic dropout. In a small number of comparisons among pairs of samples (22 of 2,445 or 0.9%), we found evidence of allelic dropout. In these cases two samples had genotypes that were identical except at one locus. At the locus where they were different, we scored one sample as heterozygous and we scored the other sample as homozygous for one of the two matching alleles. It is highly unlikely to have two individuals whose genotypes matched at all but one locus (McKelvey and Schwartz 2005), so we identified these as instances of allelic dropout and considered them to be the same individual (i.e., recaptures).

We used a robust design mark–recapture approach (Kendall et al. 1997) to analyze the genetic mark–recapture data to estimate detection and apparent survival probabilities and population size. To examine the possibility that detection probability may be different for adults and juveniles,individual bats were split into two groups. The first group contained all individuals detected in sampling periods 1 and 2 (including those that were detected in those two periods as well as subsequently detected again in the third). Individual bats detected during periods prior to volancy of young, were all considered to be adult bats. The second group contained genotypes only detected in the sampling period 3. These individuals may include adults not previously detected, but are most likely volant juveniles entering the population. We fit a two-point finite mixture model for detection probability to allow for varying detection probability between two unobservable groups (Pledger 2000). We conducted analyses in Program MARK (White and Burnham 1999).

We considered three models for the detection process. The first model allowed detection probability to vary by group (genotypes from periods 1 and 2 or genotypes from period 3) and period, the second model to vary by group, and the third model held detection probability constant. We based model selection on AIC_c (Burnham and Anderson 2002). All models included a constant survival probability and did not allow temporary emigration. We qualitatively compared the mark–recapture estimates with the minimum number of individuals detected based on unique DNA profiles, and with simple summary statistics for the emergence counts.

We also estimated relatedness among all samples to examine relationships among individuals. To this end, we used our genetic data to document relatedness among all pairs of individuals using the maximum-likelihood method implemented in ML-Relate (Kalinowski et al. 2006) adjusting for the possibility of null alleles. We inferred the most likely relationship (unrelated or related) between all pairs of individuals with 95% confidence using 10,000 repetitions. For the purpose of this study, we considered all pairs to be related if the most likely relationship was half sib, full sib, or parent–offspring.

At the primary roost tree where fecal pellets were collected and analyzed, the mean emergence count during period 1 was 100 (Table 1; 95% CI = 41–159, seven nights, range = 2–155 bats each night). During period 2, the mean emergence count was higher at 134 (Table 1; 95% CI = 68–200, six nights, range = 12–184 bats). Only one count was possible for period 3 (215 bats; Table 1).

We excluded 328 (33%) of the 1,008 pellets from which DNA was extracted because they did not amplify in at least three loci, making individual identification problematic. Of the 680 remaining samples, 489 had no missing data and had genotypes that were confirmed by multiple amplifications. Although mark–recapture analyses can deal with missing data, we included only those samples with complete genotypes (489 samples) for this analysis to be conservative. Samples with some missing genotypes could be used as long as individuals can be identified with confidence, yet we did not take that approach in this study. Data are presented in Data S1 (Supplemental Material) and in Oyler-McCance et al. (2017).

From those 489 samples, we documented 180 individuals. Of those 180 individuals, 83 were detected only once during one of the three time periods. The period with the highest number of single unique detections (45) was period 3, which occurred late in the summer corresponding to the time when young of the year are flying (Table 2). The unadjusted number (minimum known alive) of individuals identified that were likely adults (present in sampling periods 1 and 2 or in period 3 in combination with 1 or 2) was 135.

For the genetic mark–recapture analysis, the best model (selected using AIC_c; Table 3) included varying detection probability across both groups (p̂_adult = 0.126, = 0.009 and p̂_mixed = 0.077, = 0.023; ΔAIC_c = 1.05). Given this detection probability, adult bats had only a 0.15 probability of not being detected in at least one of the first two sampling periods. Therefore, most individuals in the adult and juvenile group in sampling period 3 likely are juveniles unless previously unidentified adults moved into the roost tree.

The maximum estimated number of bats present in any single period using genetic mark–recapture was 266 in sampling period 3, higher than the 215 documented through emergence counts in period 3 (Table 1). These bats were composed of adults estimated to be present in prior sampling periods (151, SE = 13.34) and juveniles emerging from roost trees as well as potential new adults arriving at the roost tree (115, SE = 31.86). The number of estimated adults in period 3 (151, SE = 13.34) is similar to the estimated number in period 1 (163, SE = 11.52), suggesting that there is not a large influx of new adult bats; moreover, the apparent survival probability of 0.994 (SE = 0.04) between primary sampling periods suggests few bats died, shifted to unsampled roost trees, or permanently emigrated during the study.

Microsatellite markers used in this study had high diversity with observed heterozygosity values ranging from 0.917 to 0.967 (Table 4). Relatedness estimates, r, between all pairs of individuals averaged 0.055 ranging from 0 to 0.779, suggesting that most pairs of individuals were not closely related to one another. Only 4.5% of the pairs of individuals (723 comparisons) were assigned to status of related, and comparison of genotypes for 0.9% of pairs (174 comparisons) were consistent with a parent–offspring relationship (i.e., shared at least one allele at each of the five loci). A parent–offspring relationship among some pairs of individuals is to be expected in a maternity colony where females are raising young.

We used fecal DNA in a genetic mark–recapture framework and generally revealed greater and more precise numbers of bats at a summer roost tree compared with emergence counts at the same tree. The estimated number of bats present in the roost tree based on mark–recapture analysis in period 1 was 163 bats, whereas the mean of emergence counts made on seven nights at this roost tree (100, 95% CI = 41–159) was substantially lower, with a lower maximum and wider confidence interval (Table 1). In period 2 the mark–recapture estimate was 122 bats, which was slightly lower than the mean emergence counts on 6 nights (134, 95% CI = 68–200). The 95% confidence intervals for both estimates overlapped, yet confidence intervals for the mean emergence counts were much wider (Table 1). In period 3 the combined mark–recapture estimate (266) was higher than the one emergence count (225). Unfortunately, simple emergence counts do not allow partitioning of the sources of variation contributing to these counts. In this case, concurrent emergence counts at other known roost trees revealed that many bats belonging to this colony were simply in other roost trees.

Overall, the mark–recapture results suggest that the use of fecal DNA for individual identification can help to account for colony members that may not be present at the roost tree during emergence counts on a particular night by adjusting for detection probabilities. Even when detection probabilities are unaccounted for, the DNA-based simple minimum number known alive at the roost tree is in concordance with the ranges of high emergence counts of Indiana bats that are accessible in the literature. For example, the estimated minimum number of adults known alive using the roost tree when young were not volant (135; sums from Table 2) was within the range of the largest emergence counts previously reported for the period of nonvolancy in Indiana (85–184; Whitaker et al. 2006; this study), and elsewhere in the central United States (e.g., 107 in Illinois, 138 in Missouri [Carter 2003; Hendricks et al. 2005]). The estimated apparent survival rates were comparable to or slightly higher than the only other short-term daily survival rate estimates for temperate zone vespertilionid bats of which we are aware, big brown bats Eptesicus fuscus monitored by passive integrated transponder tags automatically detected at roosts (Wimsatt et al. 2005; Ellison et al. 2006). Detection probabilities in our study were lower than in those studies, but might be improved in future studies of Indiana bats by increasing the sampling period lengths, including additional primary or secondary roost trees, or increasing the proportion of fecal pellets sampled for DNA analysis. Increasing detection probabilities should provide narrower confidence intervals for apparent survival estimates. The cost of conducting emergence counts is less (∼US$21,000 for two people hired commercially, counting two trees every night for seven nights, three times throughout the summer) compared with genetic analysis (∼US$26/fecal pellet for supplies and equipment, ∼US$20/pellet for analysis and interpretation, could be less for a student—∼US$46,000 for analysis of 1,000 fecal pellets not including pellet collection in the field); yet the precision of the estimates from the genetic mark–recapture analysis is greater, and there is added information about survival and detection probability.

These genetic mark–recapture estimates show that emergence counts at a single primary roost tree likely underestimate colony size in endangered Indiana bats (Silvis et al. 2014), and this probably holds true for other species of bats with fission–fusion social systems. In such cases genetic techniques might be considered as a research tool that can extend to capture–recapture analysis for estimation of apparent survival and colony size. However, our use of the robust model in this exploratory analysis did not allow temporary emigration, which clearly can be of importance in deriving estimates of colony size and apparent survival in bats with fission–fusion group dynamics. Additional statistical modeling or direct estimation should be incorporated in future studies to estimate temporary emigration and its effects on estimates (e.g., Kendall et al. 1997; Williams et al. 2002; Bird et al. 2014). Further, our study focused on only one maternity roost tree in one population and, as such, may not be effective or efficient in all types of maternity roosts. For example, genetic mark–recapture techniques may not be practical for some cavernicolous species. Sample size requirements might limit applicability to species such as Brazilian free-tailed bats Tadarida brasiliensis that roost in the hundreds of thousands or millions (McCracken 2003), or species that roost in caves and are highly sensitive to local disturbances from sampling (e.g., perhaps some populations of Corynorhinus townsendii; Pearson et al. 1952; Humphrey and Kunz 1976). Similarly, our approach would not work for noncolonial bats, yet might work better for bat species that show high fidelity to roosts as well as those with fission–fusion dynamics.

We found that most pairs of Indiana bats at the roost tree we studied were not closely related. We did find a group of individuals whose relationships were consistent with parent–offspring, likely reflecting juveniles and their mothers in this maternity roost tree. The low levels of relatedness found here are consistent with other studies of fission–fusion tree-roosting bats, although examining mitochondrial DNA also might provide additional information on the presence of matrilines (Kerth et al. 2002; Metheny et al. 2008b; Patriquin et al. 2013). The Indiana bats at this roost tree comprise a social group of largely unrelated individuals, suggesting that maternity colonies of Indiana bats may be less differentiated genetically than would be expected if maternity colonies were made up of highly related individuals that move together during fission events (Metheny et al. 2008a).

The use of DNA from samples collected noninvasively has become increasingly routine (Ebert et al. 2012a, 2012b; Row et al. 2015), particularly for species that are rare or elusive (Piggott and Taylor 2003; Waits and Paetkau 2005; Stanton et al. 2015; Vergara et al. 2014). Bats represent the second largest order of mammals, yet often basic information regarding population status of species is lacking. Arguably, sampling of bats through fecal DNA could provide much needed information for many bat species. Fecal DNA from bats has been shown to be particularly useful for bat species identification (Zinck et al. 2004) and dietary analysis of bats (Clare et al. 2009, 2011; Razgour et al. 2011; Zeale et al. 2011; Alberdi et al. 2012; McCracken et al. 2012). The use of bat fecal DNA has been tested and advocated for studies of population genetics and population size estimation (Puechmaille et al. 2007; Puechmaille and Petit 2007; Boston et al. 2012); yet, to the best of our knowledge, examples of such applications remain limited. We demonstrate here the promise of using fecal DNA for an endangered bat to estimate demographic information that is highly relevant to its management and conservation. Further, we offer that such techniques may be widely applicable to other species of bats, particularly those with fission–fusion colony dynamics.

Please note: The Journal of Fish and Wildlife Management is not responsible for the content or functionality of any supplemental material. Queries should be directed to the corresponding author for the article.

Data S1. Genotypes of all samples amplified from DNA extracted from Indiana Bat Myotis sodalis fecal pellets at a roost tree in Indianapolis, Indiana, in the summer of 2008. Genotypes are defined by the size of the DNA fragment measured in number of base pairs. We amplified samples at five highly polymorphic microsatellite loci and used these to uniquely identify individuals for the genetic mark–recapture framework. Period 1 was 11–18 June (adults only flying), period 2 was 26 June–2 July (adults only flying), and period 3 was 16–22 July (adults and volant young flying).

Found at DOI: http://dx.doi.org/10.3996/122016-JFWM-093.S1 (38 KB XLSX); also available at https://doi.org/10.5066/F728063T (October 2017).

Reference S1. Kunz TH. 2003. Censusing bats: challenges, solutions, and sampling biases. Pages 9–20 in O'Shea TJ, Bogan MA, editors. Monitoring trends in bat populations of the United States and Territories: problems and prospects. Washington, D.C.: U.S. Geological Survey, Information Technology Report, UGSG/BRD/ITR-2003-003.

Found at DOI: http://dx.doi.org/10.3996/122016-JFWM-093.S2 (7,392 KB PDF); also available at https://pubs.usgs.gov/itr/2003/0003/report.pdf (October 2017).

Reference S2. McCracken GF. 2003. Estimates of population sizes in summer colonies of Brazilian free-tailed bats (Tadarida brasiliensis), Pages 21–30 in O'Shea TJ, Bogan MA, editors. Monitoring trends in bat populations of the United States and Territories: problems and prospects. Washington, D.C.: U.S. Geological Survey, Information and Technology Report USGS/BRD/ITR-2003-0003.

Found at DOI: http://dx.doi.org/10.3996/122016-JFWM-093.S3 (9,550 KB PDF); also available at https://pubs.usgs.gov/itr/2003/0003/report.pdf (October 2017).

Reference S3. U.S. Fish and Wildlife Service. 2007. Indiana Bat (Myotis sodalis) Draft recovery plan: first revision. Fort Snelling, Minnesota: U.S. Fish and Wildlife Service.

Found at DOI: http://dx.doi.org/10.3996/122016-JFWM-093.S4 (1,395 KB DOCX); also available at http://endangered.fws.gov (October 2017).

We thank our field crew. The Indianapolis Airport Authority provided access and partial support. We thank two anonymous reviewers and the Associate Editor for their helpful comments on earlier versions of this paper.

Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the U.S. Government.