RELATEDNESS AND GENETIC STRUCTURE OF BIG BROWN BAT (EPTESICUS FUSCUS) MATERNITY COLONIES IN AN URBAN-WILDLAND INTERFACE WITH PERIODIC RABIES VIRUS OUTBREAKS

Rabies virus (RABV) has the highest fatality rate of infectious diseases, annually resulting in tens of thousands of human deaths worldwide (Hampson et al. 2015). Dogs are the predominant RABV reservoir globally, but in the Americas, where vaccination programs target domestic and wild carnivores, bat-mediated rabies is responsible for most human rabies cases (Vigilato et al. 2013; Franka and Wallace 2018). The main species imparting infections is the common vampire bat (Desmodus rotundus) in Latin America (Banyard et al. 2020). In the northwestern and southeastern US, silver-haired (Lasionycteris noctivagans) and tricolored (Perimyotis subflavus) bats have RABV variants that are associated with a disproportionate number of terrestrial mammal and human deaths despite being infrequently encountered (Messenger et al. 2003). Big brown bats (Eptesicus fuscus) are the species associated with the most human exposures because they often roost in houses and other structures and because of their large geographic distribution (Agosta 2002; Pieracci et al. 2020). Although transmission of RABV from big brown bats to humans only occurs rarely, due in part to postexposure prophylaxis, transmission to wildlife species, such as carnivores, can occur more frequently (Kuzmin et al. 2012). In a national survey of bats submitted for RABV testing, Patyk et al. (2012) found that bats from the southwest US had a higher chance of being rabid. Arizona consistently ranks among the prominent states for rabid wildlife, with bats and skunks being the most common reservoirs (Arizona Department of Health Services 2020; Ma et al. 2020).

The urban-wildland interface, where houses and other structures are built within or close to wildland vegetation (Manzello 2020), in the arid southwest US may increase risk for sylvatic RABV transmission (Theimer et al. 2015) because of aggregations of wildlife species due to increased water availability (Bazelman 2016). Flagstaff, Arizona, is a high-elevation city (2,170 m) of 75,000 people and is surrounded by 751,000 ha of national forest. Wildlife populations of bats, striped skunks (Mephitis mephitis), and gray foxes (Urocyon cinereoargenteus), in and around the city, have experienced periodic RABV outbreaks for at least 15 yr. Most of these outbreaks have been driven by the big brown bat RABV variant introduced into the two mesocarnivore species' populations, independently, during each outbreak (Kuzmin et al. 2012). The exception is a gray fox rabies variant among species that appeared in 2019 (Arizona Department of Health Services 2020). The mechanism of RABV transmission among bats, skunks, and foxes is unknown. Suburban wildlife-human interface zones, such as golf courses, provide habitat for multiple species because of abundant water (e.g., ponds), food (e.g., for bats, insects; for mesocarnivores, bird feeders and pet food), and shelter (homes). This is amplified in semiarid climates, such as in northern Arizona, where water outside the suburban matrix is limited. In hot and arid urban areas in Phoenix, Arizona, golf courses had the highest bat species richness because of water availability (Bazelman 2016). Hence, such areas may provide opportunities for RABV spillover from bat bites or ingestion of infected bat carcasses (Theimer et al. 2015, 2017a, b). During summer, female big brown bats may form maternity roosts in houses neighboring golf courses. Bats roost and raise young in soffits or cracks in chimneys (Vonhof et al. 2008; O'Shea et al. 2010). Neighborhoods with access to rich resources become potential enzootic rabies hotspots. As the weather cools during October, bats hibernate or depart, possibly to lower elevations, until the following spring. In a Colorado system, a model showed that the virus was maintained between years because of cold temperatures during hibernation, which reduced viral activity, in combination with long incubation periods; both enabled the virus to persist into the summer when transmission to naïve individuals, such as pups, occurred (George et al. 2011).

Relatedness within urban big brown bat maternity roosts and the amount of roost switching during summer months could provide clues about intraspecies RABV transmission dynamics, which may assist with management to mitigate and model rabies outbreaks. Whether maternity roosts within buildings are composed of groups of related females or whether bats in northern Arizona have high fidelity to roosts when raising offspring or commonly switch roosts is unknown. Fidelity to roosts or high relatedness within a colony would suggest that RABV outbreaks could be limited to the roost area, rather than spread throughout a larger area involving multiple roosts. O'Shea et al. (2010) found high roost fidelity by big brown bats between years in an urban area of Colorado. Brigham and Fenton (1986) determined that female big brown bats inhabiting maternity roosts in buildings in Ottawa, Canada, exhibited high site fidelity and suggested that the roosts may be composed of cohesive social groups, rather than random assortment of individuals. In forest-dwelling big brown bats in Saskatchewan, Canada, Willis and Brigham (2004) found that females were loyal to a small area within years, and that in some cases this extended between years. They also found nonrandom roosting associations, with 40% of bats roosting with the same associates the following year and pairs often switching roosts together. However, in this same Saskatchewan site, Metheny et al. (2008) determined that associating pairs were not more related than expected by chance.

In our study, we assessed roost fidelity and kinship in a maternity colony system in a Flagstaff golf course, hypothesizing, based on the above studies, that relatedness is higher within, rather than among, colonies and that roost switching is infrequent. We evaluated short-term movements by tracking individuals with microsatellite DNA and radiotelemetry and determined whether longer-term kin relationships were present by evaluating parentage, genetic relatedness, and genetic structure within and among colonies.

We focused our study in a 1.7-km² rabies-outbreak area (Kuzmin et al. 2012) in Flagstaff, Arizona (35°11′57″N, 111°37′52″W). We captured female big brown bats from four maternity roosts in the fascia of houses and nearby ponds between May and August 2012 and 2013. Telemetry data that we collected in 2012 identified potential roost selection for genetic data collection in 2013. Roosts were selected based on the presence of bats, dispersion in the study area, accessibility, and permission from homeowners. For houses, we deployed nets with funnels (Fig. 1) at dusk and left them open for up to 4 h, and captured all, or nearly all (>90%), bats at roosts. We netted each roost once during the study. At ponds, we captured bats with mist nets, which we opened at dusk for 4 h. We placed bats in holding bags until processing, whereupon we recorded species, sex, reproductive status, and mass (g) for each bat (Hinman and Snow 2003; Racey 2009). For big brown bats, we collected a buccal swab (Whatman Omniswabs, Whatman International Ltd., Maidstone, UK) for genetic analysis. We gently rotated swabs in the mouth for 1 min before placing them in 1.5-mL tubes containing 500 µL of RNAlater (Ambion, Austin, Texas, USA). Total time of bat capture was 30 min or less. We stored samples at –80 C until DNA extraction. All bat handlers used personal protective equipment and had pre-exposure rabies vaccinations.

We attached radiotransmitters (BD-2 model, Holohil Systems Ltd., Carp, Ontario, Canada) using nontoxic latex glue between the scapulae of female big brown bats captured at a pond. We limited radiotagging to two individuals per night to avoid locating animals from a single roost. All transmitters weighed ≤5% of the mass of the bat (Neubaum et al. 2005). We located each bat daily when possible until transmitters fell off (7–14 d). We used an omnidirectional car-top whip antenna (RA-5A, Telonics, Mesa, Arizona, USA), followed by a handheld directional H antenna (RA-23K, Telonics) to identify roosts for radiotagged animals. We captured and handled bats under guidelines of the American Society of Mammalogists (Sikes and Animal Care and Use Committee 2016) and with approval of Northern Arizona University's Institutional Animal Care and Use Committee (protocol nos. 07-006-R1 and 07-006-R2). No bat suffered injury or mortality as part of this study.

We extracted genomic DNA following the buccal-swab protocol described in Walker et al. (2016). We PCR-amplified seven microsatellite loci: MMG9 (Castella and Ruedi 2000), COTOF09 (Piaggio et al. 2009), and EUMA18, EUMA29, EUMA39, EUMA43, EUMA55 (Walker et al. 2014) on MJ Research PTC-200 thermal cyclers in 15-µL reactions. Reactions contained 1× Mg-free PCR buffer (Invitrogen, Thermo Fisher Scientific, Waltham, Massachusetts, USA), 2 mM MgCl₂ (Invitrogen, Thermo Fisher Scientific), 0.2 mM deoxynucleoside triphosphates, 0.1 U/µL Platinum Taq DNA polymerase (Invitrogen, Thermo Fisher Scientific), 0.04 µg/ µL Ultrapure nonacetylated bovine serum albumin (Ambion, Austin, Texas, USA), 0.2 µM fluorescently labeled forward primer, 0.2 µM reverse primer (Integrated DNA Technologies, Inc., Coralville, Iowa, USA), H₂O, and 5 µL of DNA template. We used a standard annealing temperature of 54 C for all loci, except EUMA18, EUMA39, and EUMA43, which had an annealing temperature of 61 C (Walker et al. 2014). Cycling conditions began with a denaturation step of 94 C for 2 min, followed by 35 cycles of 30 s at 94 C, 30 s at 54 or 61 C, and 1 min at 72 C, then concluded with a final extension step of 72 C for 10 min. We diluted PCR products 1:50 before fragment analysis on an Applied Biosystems 3130 Genetic Analyzer (Applied Biosystems, Foster City, California, USA). We used GeneMapper 4.0 software (Applied Biosystems) to score alleles, and GenAlEx 6.5 (Peakall and Smouse 2012) to identify recaptured bats.

We examined deviations from the Hardy-Weinberg equilibrium and linkage disequilibrium between pairs of loci with program GENEPOP 4.7.3 (Rousset 2008). In addition to the nondirectional Hardy-Weinberg test, we assessed heterozygote or homozygote deficits, and applied a sequential Bonferroni procedure (Rice 1989) to adjust for multiple tests, where appropriate.

We employed genotypic data to calculate genetic relatedness (Queller and Goodnight 1989) between pairs of individuals in each roost using GenAlEx 6.5 (Peakall and Smouse 2012). We performed randomization tests to determine whether there was a difference in pairwise relatedness among the three largest roosts (10,000 permutations via Resampling Stats 4.0 [Simon 1990]). We investigated population structure with analysis of molecular variance via GenAlEx 6.5 (Peakall and Smouse 2012), which uses permutation methods (999 iterations) to test for significance of the variance components of within- and among-roost levels of genetic structure. We also assessed genetic differentiation with FSTAT 2.9.3 (Goudet 2001) to calculate Nei's genetic distance between roost pairs and to examine significance via 1,000 permutations of the data.

We used the log-likelihood–based program Cervus 3.0 (Kalinowski et al. 2007) to identify mother–adult daughter pairs. The average probability of excluding a randomly selected unrelated bat from parentage was >0.999. Because we had no prior knowledge of maternity, each female was both a candidate offspring and a candidate mother, in 100,000 simulations. Parameters included 45% of candidate mothers sampled, 100% of loci genotyped, and an error rate of 2%. We assigned parentage when all of the following criteria were satisfied: 1) logarithm of the odds greater than 3.0 (Marshall et al. 1998); 2) delta>95%; and 3) no allelic mismatches.

We captured 339 bats representing 11 species at five ponds from May and November 2012 and April to July 2013 (Supplementary Material Table 1). The two most common species, Arizona myotis (Myotis occultus) (61% of captures) and big brown bats (17% of captures), often roost in human structures (Fenton 1997). We radiotagged one to four big brown bats per capture session, totaling 32 individuals (25 female and seven male), and located 25 roosts within our study area (Table 1). We identified 17 unique roosts in 2012 and eight unique roosts in 2013. Of those, six were used both years. In 2012 and 2013, four and two roosts, respectively, were used at least twice by a different individual on a different day during the radiotracking period. Bats roosted in 24 houses (fascia, chimney caps) and one bat house. All roosts were <500 m from a pond on a golf course. Exit counts at maternity roosts used by bats with transmitters from 30 min before dusk until about 1.5 h after sunset, gave a mean (SD) of 26 (4) bats (range, 1–56). In contrast, males roosted singly or in small groups, with a mean (SD) of 4 (2) bats (range, 1–13).

We captured and collected genetic samples from 92 female big brown bats from four roosts and nearby ponds, representing 88 unique individuals and four recaptures. Three recaptures were at different roosts, and one was at a pond. The number of bats at each roost was 28, 22, 18, and 9, with an additional 11 at ponds. Females were in various reproductive conditions of gestating (n=38), lactating (n=30), and nonreproductive (n=14); six were unknown. All were volant, with no or few parasites. We saw no evidence of disease.

All loci adhered to Hardy-Weinberg expectations for the probability test as well as for heterozygote or homozygote deficits, except for loci EUMA39 and EUMA43, which we dropped from further analyses (Bonferroni-adjusted α=0.016). No locus pairs were in linkage disequilibrium (Bonferroni-adjusted α=0.003), and there was no evidence for null alleles for any locus. Mean (SD) expected heterozygosity across roosts was 0.835 (0.19), and the mean (SD) number of alleles per locus was 11.4 (1.09); the data are shown in Supplementary Material Table 2. Multilocus genotypes are available in Supplementary Material Table 3.

We examined genetic relatedness and structure among the three largest roosts; the fourth roost was dropped from these analyses because of small sample size. There was no difference in mean pairwise genetic relatedness among maternity colonies (roost 2, –0.013; roost 3, 0.009; roost 4, –0.013; P=0.50). We found no significant genetic structure among the three largest maternity colonies (analysis of molecular variance, Table 2; Nei's genetic distance, Table 3).

The delta criterion was 5.72 at the 95% confidence level and 3.62 at the 80% confidence level. Two mother-daughter pairs met the mother-daughter criteria at the 95% confidence level: one pair was captured at the same roost, and members of the other pair were captured at different roosts. Of the 6 pairs at the 80% confidence level, 67% were from different roosts. All members of pairs were adults.

We detected an adult mother-daughter pair at different roosts genetically and found via genetic recaptures that three females in varying reproductive condition moved between multiple roosts that were 0.47 and 0.12 km apart (Figure 2a). Using telemetry, we identified six cases of bats shifting roosts (five females moved to a second roost, one male moved among three roosts) in 2012 and two cases of females switching roosts in 2013 (Figure 2b). The mean (SD) distance between female roosts used by the same individual in 2012 was 0.295 (0.093) km (range, 0.107–0.524 km), and distance between male roosts averaged 0.154 (0.015) km (range, 0.139–0.169 km). Roosts of the two roost-switching females in 2013 were 0.153 km and 0.523 km apart. Roosts were in houses (gaps in soffits and fascia, within chimney caps); one roost was an artificial roost attached to the side of a house (Table 1).

The lack of elevated genetic relatedness within big brown bat maternity colonies, lack of genetic structure among colonies, detection of mothers and their adult daughters at both the same and different maternity roosts, and observations of roost switching all suggest a high level of movement at this geographic scale. There was no suggestion that disturbance from capture contributed to this movement because bats did not immediately abandon roosts after capture at the roost and capturing at ponds did not affect roost movements. The genetic and telemetry evidence indicates that females at this location are not selecting their summer maternity roosts based on kinship. The absence of genetic relatedness having a role in group and pairwise associations was also found by Metheny et al. (2008) for tree-roosting big brown bats. They suggested that transfer of information and social thermoregulation may provide more benefit than associating with close kin and may be the norm for bats, with some exceptions (Rossiter et al. 2002). Other factors can drive roosting decisions in this species, for example, ambient temperature (Ellison et al. 2007), and for other Chiroptera (Lewis 1995; Fagan et al. 2018). That said, our finding of return of a mother and adult daughter to the same summer maternity colony may indicate that active preferential associations can be present (Willis and Brigham 2004) or may be a passive association generated by attraction to the same roost.

Our finding of interactions between colonies by females at the time of year that they occupy the Flagstaff urban-wildland interface, which is also attractive to mesocarnivores, implies a potential risk of sylvatic spread of RABV in a larger geographic area. Aggregations of bats and mesocarnivores at high density because of water availability in a semiarid or arid environment (Bazelman 2016), suggests a high potential for RABV transmission. In summer roosts in urban Colorado, female big brown bats exhibited higher seroprevalence of rabies virus–neutralizing antibodies than males or juveniles had (O'Shea et al. 2014). Within the Flagstaff urban landscape, female bats may likewise be the most relevant sex for RABV because of greater exposure within large maternity colonies. In our study, adult males were present but did not form large colonies, and fewer were captured than females. This pattern was also found in an artificial roost study in ponderosa pine forest 16 km west of Flagstaff where more female than male big brown bats were captured and recaptured over 7 yr (Diamond et al. 2015). This species exhibits male sex-biased dispersal (Turmelle et al. 2011), and in common vampire bats, dispersing males bring RABV to isolated female populations (Streicker et al. 2016). In northern Arizona, big brown bats abandon maternity roosts and disperse when pups become volant, so male-biased dispersal may not be as strong of a driver for transporting RABV to the area. Further studies incorporating mitochondrial, nuclear, and viral genetic markers over a larger geographical area and across multiple urban areas would address this question and increase the localized scale of the present study.

Webber et al. (2016) used social network analyses to predict hypothetical pathogen transmission in big brown bat maternity colonies. Their models indicated that a pathogen could move faster in a building roost than in a tree roost. However, the common practice of excluding bats from building roosts may not be advantageous from a public health perspective because studies have found that big brown bats increased movements after roost eviction (Brigham and Fenton 1986; Streicker et al. 2013), with bats settling in new roosts within 1 km (Brigham and Fenton 1986). In a European badger (Meles meles)–tuberculosis system, increased intermixing and movements caused pathogen spread into neighboring areas (Donnelly et al. 2006; Woodroffe et al. 2006). Pseudogymnoascus destructans, the fungus responsible for white-nose syndrome in bats has been found on flying male big brown bats in the summer, illustrating the potential of disease spread to new locations (Carpenter et al. 2016). A larger telemetry study to radiotrack bats in an area adjoining our <2 km² rabies outbreak site would determine the geographic extent of movements across the area and the scale of fidelity to particular areas (Willis and Brigham 2004). Additional sites surrounding other golf courses in the Flagstaff area would broaden our understanding of the implications of this work, as would inclusion of serologic studies.

It is illegal in Arizona to exclude bats from roosts during the maternity season. Exclusion of big brown bats before pups are born is associated with a decrease in reproduction (Brigham and Fenton 1986). Given the current stresses on North American bats, such as white-nose syndrome (Frick et al. 2010), it is important to align public health and conservation priorities. Streicker et al. (2013) suggested that installation of artificial roosts in urban areas would decrease risks to humans and companion animals. Further, they found that bat boxes may reduce roost switching after exclusion from buildings, thereby minimizing RABV transmission to neighboring colonies and to mesocarnivores in these areas. Artificially heated bat boxes have also been suggested as a management strategy for bat populations declining because of white-nose syndrome or other factors; little brown bats (Myotis lucifugus) with the disease preferentially used heated boxes and had a predicted energy savings of about 80% (Wilcox and Willis 2016). Eliminating den sites of skunks via sealing access under houses, decreasing pet food availability, and baiting with oral rabies vaccines under bird feeders are potential measures for reducing bat-mesocarnivore transmission (Theimer et al. 2015, 2017a, b). These measures, in addition to maintaining current rabies vaccinations of pets, will assist with mitigating the potential implications of this study: Big brown bats are vagile, with the potential to circulate RABV to neighboring areas.

Funding was provided by the State of Arizona Technology and Research Initiative Fund, Arizona Biomedical Research Commission, US Department of Agriculture Animal and Plant Health Inspection Service Wildlife Services, and a Northern Arizona University Hooper Undergraduate Research Award (to C.E.P.). Thanks to E. Saunders-Considine and the Keeley family for assistance with mist netting and radiotracking, and to J. Jenness for map generation.

Supplementary material for this article is online at http://dx.doi.org/10.7589/JWD-D-20-00112.