Anthropogenic Roost Switching and Rabies Virus Dynamics in House-Roosting Big Brown Bats

Abstract

Big brown bats (Eptesicus fuscus) are the most commonly encountered rabid bat in North America and represent an important source of wildlife rabies epizootics. Urban and suburban colonies of E. fuscus are often evicted from their roosts in houses, with poorly understood consequences for bat dispersal, population dynamics, and rabies virus transmission. We combined radiotelemetry and mark-recapture of E. fuscus with enhanced surveillance to understand the frequency of rabies virus exposure in house-roosting bats and to assess the potential for behavioral responses of eviction to exacerbate viral transmission. Serology demonstrated the circulation of rabies virus in nearly all sites, with an overall seroprevalence of 12%, but no bats were excreting rabies virus at the time of capture. Bats that were excluded from roosts relocated to houses <1 km from the original roost. However, behavioral responses to eviction differed, with bats switching repeatedly among new roosts in 1 site, but fusing with a neighboring colony in another. These findings confirm the circulation of rabies virus in E. fuscus that live in close contact with humans and companion animals, suggest mechanisms through which anthropogenic disturbance of bats might influence pathogen transmission, and highlight simple strategies to balance conservation and public health priorities.

Introduction

R abies virus (RV) is a negative-sense, single-stranded RNA virus within the genus Lyssavirus that causes acute, progressive encephalitis in mammals. Most of the tens of thousands of human rabies cases that occur globally each year are attributed to dogs; however, widespread vaccination of domestic carnivores has made rabies predominately a wildlife zoonosis in the Americas (World Health Organization 2005, Blanton et al. 2010). Viral maintenance occurs in host species-associated transmission cycles in carnivores and bats (Smith et al. 1995). Among North American bat RV reservoirs, big brown bats (Eptesicus fuscus) cause most human exposures due to their large geographic range and tendency to roost in man-made structures (Kurta and Baker 1990, Agosta 2002). Despite frequent exposure to E. fuscus, human rabies transmitted by this bat species is quite rare, with only a single case reported in 1997 (Messenger et al. 2002). In contrast, cross-species transmission of E. fuscus RVs to other mammals is more common and has caused epizootics in skunks and foxes in Arizona (Shankar et al. 2005, Leslie et al. 2006, Kuzmin et al. 2012). The rarity of E. fuscus-transmitted human rabies cases may reflect the widespread availability of postexposure prophylaxis in North America, better recognition of bites from E. fuscus relative to smaller bat species, or perhaps differential virulence among bat RV variants (Messenger et al. 2003).

Managing human and companion animal rabies risk from E. fuscus requires understanding RV transmission within bat populations and limiting cross-species exposures. Controlled infection studies have shown that E. fuscus may survive abortive infections and gain a degree of immunity, or may experience acute, lethal infection (Jackson et al. 2008, Turmelle et al. 2010b). Surveillance of RV in wild E. fuscus in the northeastern and western United States demonstrated moderately high levels of RV exposure (typically 10–30%), but low frequencies of active infection (<3%), suggesting enzootic dynamics and frequent nonlethal exposures (Trimarchi and Debbie 1977, Pearce et al. 2007, Davis et al. 2012).

The dynamics of RV in E. fuscus may be complicated by anthropogenic disturbance of bat roosts. In much of its geographic range, E. fuscus is excluded from houses because of damage resulting from accumulation of guano or when bats enter human living areas, a risk for exposures to RV and potentially to vector-borne pathogens such as Borrelia and Rickettsia if they can be transmitted to humans by bat ectoparasites (Agosta 2002, Reeves et al. 2006). Exclusion typically involves installation of devices that allow bats to exit roosts, but prevent their re-entry. Little is known of the ecological or physiological effects of displacement of E. fuscus from their roosts, nor of the consequences of exclusion for RV transmission. In one radiotracking study, individual E. fuscus showed strong site fidelity prior to eviction and experienced lower reproductive fitness after relocating to nearby roosts (Brigham and Fenton 1986). However, the degree of roost fidelity and the fitness costs of eviction also likely depend on the availability of alternative roosts (Brigham 1991). The stress of eviction might enhance susceptibility to RV infection or could promote mixing of colonies, increased dispersal, or increased density due to the fusion of local roosts, each of which could influence RV transmission.

This study conducted enhanced surveillance of RV infection and exposure in urban and suburban E. fuscus colonies in Georgia to assess the behavioral responses of bats to exclusion from houses and the potential for this disturbance to affect RV transmission. We used mark-recapture, radiotelemetry, and sampling of bats to simultaneously monitor the short-term movements and interannual population dynamics of bat colonies evicted from houses and to follow RV transmission dynamics. We hypothesized that bat colonies would be commonly exposed to RV and would remain loyal to the neighborhoods of exclusions, but that roost switching and long-term use of new roosts might depend on the quality of alternative roosts nearby.

Materials and Methods

Capture and sampling of wild big brown bats

Bats were captured in 6 neighborhoods (hereafter, sites) across 4 counties of Georgia in the United States (Fig. 1). In some sites, multiple nearby roosts were identified, leading to a total of 15 houses sampled between June, 2005, and May, 2006. Capture was limited to the spring and summer months when bats were present and active in the region, but bats were not captured from mid-July through mid-August to minimize disturbance during the nursing period. Diurnal capture was undertaken within the attics of houses using hand nets or from the outside of attic vents using funnel traps or large padded forceps (Kunz and Kurta 1988). In sites E and F, where bats were scheduled to be excluded from roosts by wildlife control workers, 1 and 2 bats, respectively, were fitted with 0.5-g radio-transmitters (BD-2N, Holohil Ltd.) immediately before exclusion and followed for up to 24 days. In site E, a second bat was captured and radiotracked in March to May, 2006. Newly discovered roosts were monitored by capture-recapture for 1 to 10 months, with an average of 5 visits per roost (range, 1–8).

FIG. 1.

Study areas for capture of bats in Georgia, United States. Each point represents 1 of the 6 sites visited between June, 2005, and May, 2006. Capture occurred at multiple houses in sites E and F (white triangles) after roosts were identified by radiotracking. Shading on the map shows counties sampled.

For each individual, we recorded the sex and weight. For a subset of bats (n=72 individuals), age was classified as adult or juvenile based on the degree of ossification in the phalangeal epiphyses (Anthony 1988). Individuals captured in sites where ongoing monitoring took place were marked with passive integrated transponder (PIT) tags (Wimsatt et al. 2005). For serological studies of RV exposure, a maximum of 200 μL of blood was collected with a heparinized capillary tube following puncture of the propatagial vein with a sterile 23-gauge needle. Blood samples were stored on ice packs until arrival at the Centers for Disease Control and Prevention (CDC) Rabies laboratory, where they were centrifuged for 10 min at 5903×g to isolate serum and stored at −20°C. A maximum of 1 blood sample per week was collected from recaptured bats. Saliva was collected by allowing bats to chew on sterile polyester swabs for 10–15 s. Paired samples from each individual were stored in TRIzol and minimum essential medium (MEM-10) for RNA extraction and RV isolation, respectively. Swabs were stored on dry ice and transferred to −80°C freezers upon arrival in the laboratory. Procedures for animal capture and sampling were approved by the CDC Animal Care and Use Committee and the Georgia Department of Natural Resources (permit 29-WSF-05-14).

Serological and RT-PCR assays

Detection of RV neutralizing antibodies (VNA) in bat sera used the modified rapid florescent focus inhibition test (RFFIT) using the challenge virus standard (CVS-11) strain of RV as described by Jackson et al. (2008). Neutralization results were calculated as end point titers using the Reed–Muench method (Reed and Muench 1938). A positive VNA titer was considered as complete neutralization of the challenge dose by a serum dilution of at least 1:4, or an approximate end-point titer of ≥0.05 IU/mL.

We tested for RV excretion in saliva using a heminested RT-PCR targeting the RV nucleoprotein (N) gene using primers 1066F and 304R in the primary RT-PCR and primer 1087F in the heminested reaction (Crepin et al. 1998, Jackson et al. 2008). This resulted in a 377-bp fragment at the 5′ end of the N gene in positive controls. Amplicons consistent in size with positive controls were sequenced using primers 1087F and 304R. We also performed an independent RT-PCR reaction targeting E. fuscus 18S ribosomal RNA to confirm the presence of amplifiable RNA in our samples using primers 18S F (5′-TCAAGAACGAAAGTCGGAGG-3′, positions 1026–1045 of the Mus musculus 18S gene, GenBank accession no. NR003278) and 18S R (5′- GGACATCTAAGGGCATCACA-3′, positions 1494–1514) using the same RT-PCR conditions.

Statistical analysis

We used generalized linear mixed modeling (GLMM) with binomial errors to identify factors associated with RV exposure. The model contained fixed effects describing the sex and weight (log transformed) of individual bats, as well as the month of capture and colony size, estimated as the minimum number of bats alive. Age was not included because data were unavailable for many individuals. Site was included as a random effect to account for the nonindependence of individuals in the same area. For bats that were resampled over time, we analyzed a single randomly selected date of capture for which a blood sample was analyzed. We conducted a separate binomial GLMM to test monthly variation in seroprevalence, where site was again treated as a random effect, but samples from the same individuals captured in different months were included. The GLMMs used the lme4 package of R v. 2.14.1 (Bates et al. 2011, R Development Core Team 2011). Model selection was performed using the Akaike information criterion (AIC) with individual term significance determined by stepwise removal of terms, starting with interactions, followed by nested likelihood ratio tests (Zuur et al. 2009).

Results

Capture and movement histories of big brown bats

Across all sites and sampling periods, we captured 245 E. fuscus, representing 191 unique individuals and 54 recaptured individuals. Of the 4 individuals that were radiotagged, we were able to track the movements of 3 for the full 3-week monitoring period. Each of these bats was located in all subsequent searches (5–15 visits, depending on the site), confirming their presence in the study areas. The fourth bat was tracked for 9 days after release, but was not found in later searches. In the weeks following exclusion, bats demonstrated strong fidelity to neighborhoods, always relocating to houses that were <1 km from the site of exclusion (mean=271 meters, standard error [SE]=63, range=67–551; Table 1). Recaptures of marked bats in the roosts identified through telemetry gave broader insights into the frequency of roost switching at each site. In site E, individuals recaptured after the first capture following exclusion (when a roost switch was guaranteed), switched roosts almost constantly, with 15/21 recaptures indicating a roost switch since the previous capture (Fig. 2A). In that site, all roosts found after exclusion lacked substantial accumulation of guano, suggesting that these houses were not common roosts prior to displacement. In contrast, in site F, bats joined a larger colony (as indicated by the abundance of unmarked individuals and large accumulation of guano) and switched roosts less frequently in the weeks after exclusion (Table 1, Fig. 2B).

FIG. 2.

Patterns of roost switching after exclusion of bats from 2 houses in Georgia. Each line shows the movement of an individual bat, inferred from recapture histories. Triangles represent the houses from which bats were excluded (as in Fig. 1), and circles indicate roosts identified by radiotelemetry. Doubled lines indicate recaptures in the same roost. Movements in both sites E (A) and F (B) occurred over a time span of 23 days, with the exception of the dashed arrow in A, which occurred between September, 2005 and March, 2006.

Table 1.

Summary of Movement Histories of Radiotracked E. fuscus in Suburban Neighborhoods in Georgia

Bat	Sex	Site	No. roost switches	No. searches	No. days ^a	Min. distance (km)	Max. distance (km)
876–887^b	Male	E	2	5	14	0.152	0.274
859–105	Female	E	4	14	24	0.082	0.419
305–1407	Female	F	2	10	23	0.068	0.424
814–1409	Male	F	1	10^c	23	0.551	0.551

The number of days between the release of the radiotagged bat and the last search.

Radiotracking performed in March to May, 2006, the year following exclusion.

Bat was not detected after search number 5, 10 days after initial release.

In site E, we were able to capture bats in the year after exclusion and found a substantially smaller bat population. All former roosts were surveyed, but only 6 individuals were found in 6 surveys between March and May, 2006. Radiotracking of 1 of these individuals failed to reveal a larger roost nearby, suggesting that bats had abandoned the area (Table 1).

Spatial, temporal, and individual patterns of RV exposure

VNAs were detected in house-roosting E. fuscus in all sites in all months, with the exception of site D. However, that site was sampled only once with relatively few individuals tested (n=10 individuals; Table 2). Seroprevalence in other sites ranged from 6% to 50%, but values between 15% and 25% were more typical (Table 2). Among the 22 individuals that were sampled on multiple occasions, 16 were seronegative in all capture occasions and 5 were seropositive across multiple captures (separated by 14–25 days). Loss of apparent VNA occurred in only 1 individual, which had a weak positive titer (0.05 IU/mL) in September, 2005 but was seronegative in a subsequent sample in March, 2006. No seroconversion was observed in bats that were captured and sampled multiple times during the study period.

Table 2.

Prevalence of Rabies Virus Neutralizing Antibodies in E. fuscus from 6 Sites, 2005–2006

			2005			2006
Site	County	Overall seroprevalence ^a	Jul ^b	Aug	Sept	Mar	Apr	May
B	Forsyth	0.1 (20)	0.10 (20)
C	Spalding	0.29 (17)	0.29 (17)
D	Fulton	0 (10)	0 (10)
E	Henry	0.14 (36)		0.15 (20)	0.2 (20)	0.25 (4)	0.5 (2)
F	Fulton	0.06 (32)			0.06 (32)
L	Fulton	0.13 (16)						0.13 (16)
All	NA	0.12 (125)	0.15 (47)	0.15 (20)	0.11 (52)	0.25 (4)	0.5 (2)	0.13 (16)

Proportions in each cell are the fraction seropositive and numbers in parentheses are the total number of unique individuals tested (i.e., removing multiple samples from the same individual in the specified time frame).

Seroprevalence calculated using first capture data from recaptured individuals.

Monthly seroprevalence calculated using a single sample per month for recaptured individuals.

Abbreviation: NA, not applicable.

The GLMM revealed no statistically significant predictors of RV exposure. The best model according to AIC contained an effect of sex (AIC=100.081), but this model was not significantly better than a model containing only a y-intercept term (AIC=101.244; likelihood ratio test: χ²=3.23, degrees of freedom [df ]=1, p=0.073). The model that included observations of individuals in multiple months similarly showed no strong variation in RV exposure by month (χ²=1.97, df=5, p=0.85). Particularly in 2006, this analysis was limited by small samples sizes, so our power to detect differences was low (Fig. 3). In both excluded sites, seroprevalence increased in the months following eviction, but these changes were not statistically significant (Site E: pre=0.15, post=0.20, χ²=0.17, p=1; site F: pre=0.0, post=0.11, χ²=1.46, p=0.5).

FIG. 3.

Monthly patterns of rabies virus seroprevalence across all sites. Solid bars on points are standard errors; the dashed line indicates the overall seroprevalence, averaging over all sites in the study period (12%).

RT-PCR detection of virus shedding in saliva

Saliva swabs were tested for both the presence of RV RNA and bat 18S rRNA. Of the 210 swabs tested, 22 (10.5%) contained no amplifiable 18S rRNA, suggesting that RNA had been lost during extraction or the presence of RNases or PCR inhibitors. Heminested RT-PCR of the remaining 188 saliva samples from 157 unique individuals found no instances of viral excretion.

Discussion

In North America, exclusion of bats from their roosts is a common tool to nonlethally remove bats from houses to reduce damage to homes and avoid human exposures to zoonotic pathogens, including RV. By conducting active surveillance of 6 urban and suburban E. fuscus colonies in Georgia, our study demonstrated widespread circulation of RV in house-roosting bats and suggested several mechanisms by which anthropogenic disturbance might influence bat populations and RV transmission.

First, we found seropositive bats in all sites that were adequately sampled (Table 1), but no evidence for active excretion of RV. These results are consistent with previous surveys of RV in E. fuscus and other bat species, which found comparable rates of seroprevalence with very low frequencies of active infection (Steece and Altenbach 1989, Turmelle et al. 2010a, Davis et al. 2012). These apparently nonlethal exposures that cause seroconversion may have important impacts on RV epizootiology by granting some degree of immunity; however, further study is needed to understand the degree of protection afforded by VNAs (Turmelle et al. 2010b). A previous study in Colorado found RV exposure in E. fuscus to vary by year, colony size, and to a lesser extent age (Pearce et al. 2007). In our study, RV exposure was not strongly predicted by any of the factors that we measured (sex, mass, colony size, or month of capture). Year was not tested because we sampled different parts of the season in 2005 versus 2006, and missing data for the age of individuals precluded the inclusion of an age term in models. The lack of variation in RV exposure by month is perhaps surprising given that human exposures to rabid E. fuscus peak in late summer to early fall, so seroprevalence would be expected to peak in August or September, then decrease as bats return from hibernation (Constantine 1967). However, field studies in Mexican free-tailed bats, Tadarida brasiliensis suggest that seasonal differences in infection may be most pronounced in juvenile bats (Turmelle et al. 2010a, Steece and Altenbach 1989). Because our survey included blood samples from relatively few juveniles (n=12), we had little power to detect seasonal differences if they were driven by transmission in juveniles. Because mathematical models of RV in E. fuscus suggest that birth pulses of susceptible juveniles could be important for long-term viral maintenance, future field surveillance should emphasize sampling of young bats, particularly in the months before dispersal to hibernacula (George et al. 2011).

A major contribution of our study was the ability to follow colonies after exclusion from maternity roosts. We observed two distinct short-term responses to exclusion that could potentially influence the transmission dynamics of RV. First, in site E, bats switched frequently between 4 alternative roosts, none of which emerged as predominant (Fig. 2A). Although reports of roost switching in forest-dwelling E. fuscus suggest the possibility that the roost switching that we observed could be unrelated to exclusion, E. fuscus in buildings tended to have higher roost fidelity and changing roosts diminished reproductive fitness (Brigham and Fenton 1986, Brigham 1991, Lewis 1995, Willis and Brigham 2004). The absence of accumulated guano at newly discovered roosts in site E further supported a shift in behavior after eviction. Importantly, increased dispersal after disturbance can have unanticipated consequences for disease transmission by homogenizing host populations and enabling pathogen spread to adjoining areas (Donnelly et al. 2006, Woodroffe et al. 2006). Similar effects could result as bats disperse and relocate after exclusion. Second, in site F, bats joined a neighboring colony and roost switching was less frequent (Fig. 2B). In this scenario, the short-term stress of eviction on bats may have been alleviated, but contact rates and RV transmission might be increased by larger colony sizes (Pearce et al. 2007). Thus, both of the posteviction roosting strategies that we observed (frequent dispersal and roost fusion) might have consequences for RV transmission that deserve further attention through more comprehensive surveillance of larger numbers of bat colonies before and after exclusion.

Finally, although we were only able to follow a single bat colony in the year after exclusion, that site was almost completely abandoned in the following spring, suggesting that most of the colony relocated to a different area following hibernation. Interestingly, this was the same colony where bats frequently switched roosts after eviction (Fig. 2A). Abandonment of this area is therefore consistent with the absence of suitable alternative roosts implied by frequent roost switching in the previous year. In other areas where alternative roosts are abundant, bats might be more likely to return to the area of exclusion. Therefore, our results highlight the potential importance of the presence of high quality alternative roosts for the long-term consequences of exclusion.

Implications for bat conservation and rabies management

Given the mounting threats to bats in eastern North America through habitat loss associated with urbanization and the spread of white-nose syndrome, eliminating additional challenges to bats is a key priority (Fenton 2003, Blehert et al. 2009). Humane exclusion of bats from roosts is justified when they enter human living spaces or generate excessive property damage. Management must therefore embrace strategies that limit the impacts of exclusion on bat population dynamics and RV transmission. One potential solution may be to encourage use of alternative roost structures, such as artificial “bat boxes,” to bats that are excluded from houses. When properly designed and installed, bat boxes provide suitable roosts for the most commonly excluded bat species in the US, E. fuscus and Myotis lucifugus (Brittingham and Williams 2000). In addition to minimizing zoonotic disease risk by reducing contact between bats and/or their ectoparasites and humans, our results suggest that installation of bat boxes might minimize roost switching after exclusion, encourage bats to return to areas following hibernation, and limit the potential exacerbation of RV transmission associated with enhanced dispersal and/or fusion with neighboring colonies. Pairing well-prescribed exclusion from houses with effective installation of bat boxes may therefore provide an effective balance of management priorities for bat conservation and zoonotic disease risk.

Footnotes

Acknowledgments

For helpful discussions on the manuscript we thank Julie Rushmore, Jamie Winternitz, and 2 anonymous reviewers. We thank Josh Self, Jesse Blanton, and other members of the CDC Rabies Program for assistance in capturing and sampling bats, and Tom O'Shea for providing PIT tags and a scanner. We thank the homeowners in each of our sites for allowing us access to their homes. D.S. was supported by a CDC-APHL Emerging Infectious Diseases Training Fellowship.

Author Disclosure Statement

No competing financial interests exist. The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the Centers for Disease Control and Prevention.

References

Agosta

. Habitat use, diet and roost selection by the big brown bat (Eptesicus fuscus) in North America: A case for conserving an abundant species. Mamm Rev, 2002; 32:179–198.

Anthony

ELP

. Age determination bats. Kunz

. Ecological and Behvioral Methods for the Study of Bats. Washington DC: Smithsonian Institution Press, 1988; 47–58.

Bates

, Maechler

, Bolker

. lme4: Linear mixed-effects models using S4 classes. 2011.

Blanton

, Palmer

, Rupprecht

. Rabies surveillance in the United States during 2009. J Am Vet Med Assoc, 2010; 237:646–657.

Blehert

, Hicks

, Behr

, Meteyer

et al.

Bat white-nose syndrome: An emerging fungal pathogen?

Science, 2009; 323:227–227.

Brigham

. Flexibility in foraging and roosting behaviour by the big brown bat (Eptesicus fuscus) Can J Zool, 1991; 69:117–121.

Brigham

, Fenton

. The influence of roost closure on the roosting and foraging behaviour of Eptesicus fuscus (Chiroptera: Vespertilionidae) Can J Zool, 1986; 64:1128–1133.

Brittingham

, Williams

. Bat boxes as alternative roosts for displaced bat maternity colonies. Wildl Soc Bull, 2000; 28:197–207.

Constantine

. Bat rabies in the southwestern United States. Pub Hlth Repts, 1967; 82:867–888.

10.

Crepin

, Audry

, Rotivel

, Gacoin

et al. Intravitam diagnosis of human rabies by PCR using saliva and cerebrospinal fluid. J Clin Microbiol, 1998; 36:1117–1121.

11.

Davis

, Gordy

, Rudd

, Jarvis

et al. Naturally acquired rabies virus infections in wild-caught bats. Vector-Borne Zoonotic Dis, 2012; 12:55–60.

12.

Donnelly

, Woodroffe

, Cox

, Bourne

et al. Positive and negative effects of widespread badger culling on tuberculosis in cattle. Nature, 2006; 439:843–846.

13.

Fenton

. Science and the conservation of bats: Where to next? Wildl Soc Bull, 2003; 31:6–15.

14.

George

, Webb

, Farnsworth

, O'Shea

et al. Host and viral ecology determine bat rabies seasonality and maintenance. Proc Natl Acad Sci USA, 2011; 108:10208.

15.

Jackson

, Turmelle

, Farino

, Franka

et al.

Experimental rabies virus infection of big brown bats (Eptesicus fuscus)

J Wildl Dis, 2008; 44:612–621.

16.

Kunz

, Kurta

. Capture methods and holding devices. Kunz

. Ecologial and Behavioral Methods for the Study of Bats. Washington, DC: Smithsonian Institution Press, 1988.

17.

Kurta

, Baker

. Eptesicus fuscus. Mamm Spec, 1990; 356:1–10.

18.

Kuzmin

, Shi

, Orciari

, Yager

, Velasco-Villa

, Kuzmina

, Streicker

, Bergman

, Rupprecht

. 2012. Molecular inferences suggest multiple host shifts of rabies viruses from bats to mesocarnivores in Arizona during 2001–2009. PLoS Pathogens, 8,6:e1002786, 10.1371/journal.ppat.1002786

19.

Leslie

, Messenger

, Rohde

, Smith

et al. Bat-associated rabies virus in skunks. Emerging Infect Dis, 2006; 12:1274–1277.

20.

Lewis

. Roost fidelity of bats: A review. J Mammal, 1995; 76:481–496.

21.

Messenger

, Smith

, Rupprecht

. Emerging epidemiology of bat-associated cryptic cases of rabies in humans in the United States. Clin Infect Dis, 2002; 35:738–747.

22.

Messenger

, Smith

, Orciari

, Yager

et al. Emerging pattern of rabies deaths and increased viral infectivity. Emerging Infect Dis, 2003; 9:151–154.

23.

Pearce

, O'Shea

, Shankar

, Rupprecht

. Lack of association between ectoparasite intensities and rabies virus neutralizing antibody seroprevalence in wild big brown bats (Eptesicus fuscus), Fort Collins, Colorado. Vector-Borne Zoonotic Dis, 2007; 7:489–495.

24.

R Development Core Team. R. A language and environment for statistical computing. R Foundation for Statistical Computing, 2011.

25.

Reed

, Muench

. A simple method of estimating fifty per cent endpoints. Am J Epidemiol, 1938; 27:493–497.

26.

Reeves

, Streicker

, Loftis

, Dasch

. Serologic survey of Eptesicus fuscus from Georgia, USA for Rickettsia and Borrelia and laboratory transmission of a Rickettsia by bat ticks. J Vector Ecol, 2006; 31:386–389.

27.

Shankar

, Orciari

, De Mattos

, Kuzmin

et al. Genetic divergence of rabies viruses from bat species of Colorado, USA. Vector-Borne Zoonotic Dis, 2005; 5:330–341.

28.

Smith

, Orciari

, Yager

. Molecular epidemiology of rabies in the United States. Semin Virol, 1995; 6:387–400.

29.

Steece

, Altenbach

JJS

. Prevalence of rabies specific antibodies in the Mexican free-tailed bat (Tadarida brasiliensis mexicana) at Lava Cave, New Mexico. J Wildl Dis, 1989; 25:490–496.

30.

Trimarchi

, Debbie

. Naturally occurring rabies virus and neutralizing antibody in two species of insectivorous bats of New York State. J Wildl Dis, 1977; 13:366–369.

31.

Turmelle

, Allen

, Jackson

, Kunz

et al. Ecology of rabies virus exposure in colonies of Brazilian free-tailed Bats (Tadarida brasiliensis) at natural and man-made roosts in Texas. Vector-Borne Zoonotic Dis, 2010a. 10:165–175.

32.

Turmelle

, Jackson

, Green

, McCracken

et al. Host immunity to repeated rabies virus infection in big brown bats. J Gen Virol, 2010b. 91:2360–2366.

33.

World Health Organization. WHO expert consultation on rabies. Tech Rep Ser, 2005; 931:1–88.

34.

Willis

CKR

, Brigham

. Roost switching, roost sharing and social cohesion: Forest-dwelling big brown bats, Eptesicus fuscus, conform to the fission-fusion model. Anim Behav, 2004; 68:495–505.

35.

Wimsatt

, O'Shea

, Ellison

, Pearce

et al. Anesthesia and blood sampling of wild big brown bats (Eptesicus fuscus) with an assessment of impacts on survival. J Wildl Dis, 2005; 41:87–95.

36.

Woodroffe

, Donnelly

, Jenkins

, Johnston

et al. Culling and cattle controls influence tuberculosis risk for badgers. Proc Natl Acad Sci USA, 2006; 103:14713–14717.

37.

Zuur

, Ieno

, Walker

, Saveliev

et al. Mixed Effects Models and Extensions in Ecology with R. Springer Verlag, 2009.