Roosting ecology and variation in adaptive and innate

ORIGINAL PAPER

Roosting ecology and variation in adaptive and innateimmune system function in the Brazilian free-tailed bat(Tadarida brasiliensis)

Louise C. Allen Æ Amy S. Turmelle ÆMary T. Mendonca Æ Kristen J. Navara ÆThomas H. Kunz Æ Gary F. McCracken

Received: 11 June 2008 / Revised: 12 October 2008 / Accepted: 14 October 2008 / Published online: 11 November 2008! Springer-Verlag 2008

Abstract Bats have recently been implicated as reser-voirs of important emerging diseases. However, few

studies have examined immune responses in bats, and even

fewer have evaluated these responses in an ecologicalcontext. We examined aspects of both innate and adaptive

immune response in adult female Brazilian free-tailed bats

(Tadarida brasiliensis) at four maternity roosts (two nat-ural caves and two human-made bridges) in south-central

Texas. Immune measurements included in vitro bacterici-

dal ability of whole blood and in vivo T cell mediatedresponse to mitogenic challenge. Bactericidal activity in

T. brasiliensis varied with roosting ecology, but appears to

be sensitive to colony-level effects. Blood from femalesliving at one cave had significantly lower bactericidal

ability than blood from females at three other sites. T cell

mediated response in this species was associated withvariation in roost ecology, with females from two caves

having greater responses than females from two bridges.

T cell mediated response and bactericidal activity were

negatively correlated with one another within individualsthat were tested for both. Variation in immunological

response of T. brasiliensis is important for understanding

the influence of the environment on the frequency anddistribution of immunologically competent individuals and

for understanding disease-host dynamics in this and other

colonial species.

Keywords Artificial roosts ! Bactericidal ability !Bats ! Immune response ! Phytohaemagglutinin

AbbreviationsBKA Bacterial killing ability

FBS Fetal bovine serumPBS Phosphate buffered saline

PHA Phytohaemagglutinin

Introduction

The emergence in wildlife of infectious diseases, such asAIDS, Ebola, West Nile virus, SARS, and Hantaviruses,

highlight the need to understand the ecology of reservoir

hosts to disease, and the interactions of immunologicalresponse and ecological variation. Recently, bats have

become a focus of attention as possible reservoirs of

several emerging pathogens, including SARS-like viruses(Li et al. 2005), and other Corona viruses (Dominguez

et al. 2007; Gloza-Rausch et al. 2008), Nipah and

Hendra viruses (Joharra et al. 2001; Halpin et al. 2000),Ebola virus (Leroy et al. 2005), Herpes virus (Wibbelt

et al. 2007), and others (reviewed in Calisher et al.

2006). Bats also have long been recognized as reservoirhosts, and vectors to humans and domestic animals, of

Communicated by G. Heldmaier.

L. C. Allen (&) ! T. H. KunzCenter for Ecology and Conservation Biology,Boston University, Boston, MA 02215, USAe-mail: [email protected]

A. S. Turmelle ! G. F. McCrackenDepartment of Ecology and Evolutionary Biology,University of Tennessee, Knoxville, TN 37996, USA

M. T. Mendonca ! K. J. NavaraDepartment of Biological Sciences, Auburn University,Auburn, AL 36849, USA

K. J. NavaraDepartment of Poultry Science, University of Georgia,30602 Athens, GA, USA

123

J Comp Physiol B (2009) 179:315–323

DOI 10.1007/s00360-008-0315-3

rabies virus (Messenger et al. 2003; Constantine 1967)

and related lyssaviruses (Fraser et al. 1996; Botvinkinet al. 2003). However, little is known about variation in

immune function and susceptibility to disease in free-

ranging bats (Calisher et al. 2006; Dobson 2005; but seeChriste et al. 2000). Improving our understanding of

variation in immune function among individuals and

populations of bats is important for understanding therole of bats as potential vectors of emerging diseases

(Messenger et al. 2003; Dobson 2005).Bats possess a variety of life-history traits that are

likely to affect their immune function and the role of bats

as reservoirs and vectors of disease (Calisher et al. 2006).Many species are colonial, although considerable varia-

tion in colony size exists (Kunz 1982; O’Shea and Bogan

2003). The Brazilian free-tailed bat (Tadarida brasilien-sis), for example, roosts in some of the largest

aggregations of mammals on earth, with several thousand

to several million individual bats estimated to formmaternity colonies in caves and under highway bridges

(Davis et al. 1962; McCracken 2003; Keeley and Keeley

2004; Betke et al. 2008). Colonial living brings the pos-sibility of increased exposure to infectious pathogens,

with direct links to immune defense and protection.

Relationships have been shown between coloniality andimmune responsiveness in several avian species, although

evidence is conflicting. For example, Tella et al. (2001)

found that T cell mediated immune response in fledglingsof the Magellanic penguin (Spheniscus magellanicus) was

negatively correlated with colony size, whereas Møller

et al. (2001), examining immunity at the interspecificlevel, found that highly colonial swallows and martins,

have higher levels of T and B cell response compared to

less social songbirds.Bats sometimes roost in proximity to urban and agri-

cultural landscapes (Horn and Kunz 2008), and

occasionally within human dwellings. In south-centralTexas, T. brasiliensis uses both natural and man-made

structures as roosts, including caves, bridges, buildings,

and bat houses (Kunz and Reynolds 2003; McCracken2003; Sgro and Wilkins 2003). Roosts can differ in struc-

ture, capacity and possible quality (Lausen and Barclay

2006; Neubaum et al. 2006). Given the recognized role ofbats in disease transmission, it is important to understand

variation in their immune condition in relation to ecolog-

ical variables, particularly for those species roosting inclose proximity to humans.

Innate immunity establishes an early line of defense

against invading pathogens and inhibits the progression ofinfection. In contrast, the adaptive immune responses are

induced upon antigen processing and have greater

importance for clearing infections (Goldsby et al. 2003).The innate and adaptive arms of the immune response

interact to protect individuals, although their competence

may vary among individuals (Blount et al. 2003). Mostprevious research on the effects of ecological conditions

on immune function has focused primarily on the adaptive

immune response (but see Tieleman et al. 2005 andBlount et al. 2003), by estimating T cell proliferation in

response to mitogenic challenge (Tella et al. 2001; Tella

et al. 2002). Evaluation of multiple components ofimmune response is needed to develop a more complete

understanding of immune competence in free-ranginganimals.

In the present study, we evaluate aspects of both the

innate and adaptive arms of the immune system inT. brasiliensis over a period of 5 months at four maternity

colonies (two natural caves, two man-made bridges) in

south-central Texas. Bactericidal activity of whole bloodin culture, one aspect of the innate immune response,

primarily measures complement-mediated cytotoxicity

(Merchant et al. 2003), which is a major factor in defenseagainst viruses (Blue et al. 2004). T cell mediated

response, which contains aspects of both innate and

adaptive immune components, is also important in clear-ance of viruses, including rabies virus (Hooper et al.

1998). This study was designed to characterize responses

of T. brasiliensis to two immune challenges. We postu-lated that variation in both the innate and adaptive arms

of the immune system would be affected by differences in

roosting ecology as it relates to roost environment andcolony size.

Materials and methods

Animal sampling

This study was conducted in south-central Texas from May

to September of 2005. Bactericidal activity and T cellinfiltration were assessed in free-ranging adult female

Brazilian free-tailed bats (T. brasiliensis), captured at two

natural caves (Frio Cave and Davis Cave) and two largepre-caste concrete highway bridges (Seco Creek Bridge

and East Elm Creek Bridge). Periods of data collection

corresponded to female reproductive stages: pregnancy(May–June); lactation (June–July); post-lactation/non-

reproductive (August–September). Earlier studies indicated

that mating in T. brasiliensis occurs prior to springmigration (Davis et al. 1962; Cockrum 1969), though

recent evidence suggests that mating also occurs in Texas

in March and April (Keeley and Keeley 2004). Outside ofthe mating season, however, many of these roosts are

occupied by non-territorial maternity colonies, comprised

mostly of females. Thus, we have focused on adult femalesin the present study.

316 J Comp Physiol B (2009) 179:315–323

123

Approximately 15 bats were captured on a given night at

each site, most of which were sampled within 2–4 days ineach sampling period, for a total of 327 bats. Both

immunological assays were not always performed on each

bat, due to temporal constraints of the specific protocols.Body mass, reproductive condition (non-reproductive,

pregnant and lactating) and age class (adult or juvenile;

Anthony 1988) were determined for each individual.Juveniles were not included in the analyses, due to low

sample sizes (n = 20) over the entire study period. Batswere held for up to 14 h for the immunological assays

described below and then released at the site of capture. All

individuals received a site-specific tattoo on their wings toprevent re-sampling. Tattoos on wings provide effective

markings on these bats (Lollar and Schmidt-French 1998)

that can last up to 3 years. All capture, handling andexperimental procedures were approved by the University

of Tennessee Animal Care and Use Committee (#890) to

G.F.M. and Boston University Animal Care and UseCommittee (05–012) to T.H.K., and under Texas Parks and

Wildlife Department permit SPR-0305-058 to T.H.K.

Innate immune function: bactericidal ability

Bacterial killing ability of the bat’s whole blood againstEscherichia coli was measured to represent one aspect of

the innate immune response. This technique has been used

successfully as an in vitro assay of innate immune functionin free-ranging animals (Tieleman et al. 2005). A small

amount of whole blood (6 ll) was collected in sterile

heparinized capillary tubes via venopuncture (Kunz andNagy 1988) within 2 h of capture from each bat. Although

Matson et al. (2006) found a decrease in bactericidal ability

of the blood of several species of birds at this length oftime post-capture, preliminary data on T. brasiliensisshows no reduction in bactericidal ability up to 2 h post-

capture. The blood was diluted to 1:50 in RPMI-1640media (Roswell Park Memorial Institute), supplemented

with 5% Fetal Bovine Serum (FBS). The E. coli (ATCC

8739; Microbiologics, USA) solution was diluted to1:1,000 using sterile phosphate buffered saline (PBS). A

total of 140 ll of diluted blood was mixed with 10 ll of

diluted bacteria. Once mixed, 50 ll of the combined bloodand bacterial dilution was spread onto labeled trypticase

soy agar plates (BD Diagnostic systems, USA) at both 0

and 60 min post-mixing. Two control plates consisted of5% FBS in RPMI with equivalently diluted bacteria and no

blood. All plates were incubated at 37"C for 12 h, after

which colonies of E. coli were visually counted andrecorded. The unit-less index used to calculate the bacte-

ricidal ability of the blood was standardized for both

controls of a given assay:

BKA index

¼ #1$ blood plate 60# blood plate 0

blood plate 0$ 100

! "!

# control 60# control 0

control 0$ 100

! ""

This index assigned large positive values when bats had

blood with high cytotoxic activity and negative valueswhen blood had little or no cytotoxic activity against the

bacteria. The index method controls for bacterial die-off

occurring within an assay that is unrelated to the blood’sbactericidal ability. Only data from assays where there was

less than a 15% difference between the 0 and 60 min

controls were used. Three individuals whose bactericidalindices were over three standard deviations lower than the

mean of the remaining individuals were excluded from the

data set, because these values deviated from the normaldistribution of bactericidal ability found in our samples. In

the final analyses, bactericidal indices were derived from

89 individual adult female bats.

Adaptive immune function: T cell mediated immune

response

A subcutaneous injection of phytohaemagglutinin (PHA-P

#L8754; Sigma, USA) was administered to assess T cellinfiltration in a total of 163 individual bats. Injections were

administered in the interfemoral membrane (uropatagium)

below the knee at the point of contact with the leg. Prior toinjection, the area was measured with a digital micrometer

(Mitutuyo #293-230, Japan). The experimental area on

each bat was injected with 0.05 ml of 3 mg/ml PHA inPBS. As a control, the contralateral side was then injected

with 0.05 ml of PBS. Swelling at the sites of injection was

subsequently measured at 12 hours post-challenge. Mea-surements of PHA swellings in the uropatagium of bats are

very similar to the more common measurements made in

the wing web of birds. The swelling produces thickenedcellular infiltrate between the two layers of skin adjacent to

the leg, thus our measurements are comparable in execu-tion to those conducted in birds. Although bactericidal

ability of blood was not tested in every bat, all bats were

bled within 3 min of capture, for a companion study onstress hormones, therefore all bats in our study were bled

once prior to PHA challenge. To ensure consistency in

assessing the PHA test, one of us (L.C.A.) performed allinjections, and another (A.S.T.) made all measurements.

Care was taken to conduct repeatable, blind measurements

with each swelling measured twice and averaged. Fol-lowing Navara et al. (2005), the unit-less index below was

used to determine the swelling response to the mitogenic

challenge standardized against the control response:

J Comp Physiol B (2009) 179:315–323 317

123

PHA index ¼ 12hrPHA# 12hrPBSð ÞprePHAþprePBS

2

# $

Indices greater than one designate bats with marked T cellinfiltration to the injection area, whereas smaller index

values indicate that little or no response was observed. One

individual with an index value that was over three standarddeviations beyond the normal range of indices was exclu-

ded from the data set, resulting in analyses of T cell

infiltration for 162 individual adult female bats.While this technique has been widely used in avian

immunology to assess T cell mediated response (Bon-

forte et al. 1972; Tella et al. 2001), recent studies havecautioned that the proper identification of cellular com-

ponents present in the swelling is necessary to accurately

interpret the response (Kennedy and Nager 2006; Martinet al. 2006a). In a separate group of T. brasiliensis(N = 18), we analyzed PHA treated and control tissues

with biopsies of the swellings taken 6–21 h post-chal-lenge. We documented significantly larger numbers of

lymphocytes (mean cell count = 6), as well as heter-

ophils (mean cell count = 35), in PHA-challengedtissues that were not observed in any numbers in control

tissues, demonstrating that specific cellular infiltration

had occurred in response to the PHA challenge. Wefound that measurements of the swellings at 12-h post-

injection capture the maximal swelling response of

individual bats, and represented when increased lym-phocyte and heterophil infiltration occurred (Turmelle

and Mendonca, unpublished data). The immune response

to PHA challenge in bats suggests that both innate andadaptive components are responsible for local swelling at

the injection site, as was found in birds by Martin et al.

(2006a). Because PHA elicits T-lymphocyte response atthe site of injection, this method is a good overall

measure of cell-mediated immune function in this

species.

Statistics

The bactericidal indices among all individuals included in

the final data set were normally distributed. However, T

cell responses among individuals were not normally dis-tributed, and thus the data were log-transformed to fit a

normal distribution prior to statistical analysis. Non-trans-

formed values, means ± standard error, of bactericidal andT cell indices are presented. Indices for either assay did not

differ statistically between reproductive classes (BKA,

P = 0.374; PHA, P = 0.961). Thus, all reproductiveclasses were combined prior to analyses of the data sets.

For both immune indices, a nested mixed ANOVAmodel, with one covariate (general linear model; using

JMP 5.0.1), was used for comparison of roost type. By

roost type we refer to the structure of the roost (bridge or

cave), whereas by colony we refer to the location at whichthe individual bats were sampled (Frio Cave, Davis Cave,

Seco Creek Bridge and East Elm Creek Bridge). Because

we were only able to sample four colonies, a nested designwas used (comparing roost types, with colony nested

within roost type) to examine if there is significant varia-

tion in roost types beyond variation due to differencesamong colonies alone. Roost type, and colony nested

within roost type were independent variables and bodycondition, a ratio of body mass to right forearm length, was

a covariate in the overall model. We also treated colony as

a random effect (i.e., we are interested in generalizing toother colonies of Brazilian free-tailed bats); roost type

(bridge or cave), a fixed effect, was tested over colony

nested within roost type. Because colony size varies withroost type (the majority of cave colonies, including those

studied here, have more bats living in them than bridge

colonies), we were unable to directly test for the effects ofpopulation size on immune function, but we will explore

this hypothesis in our discussion below. Lastly, we tested

for a linear association between bactericidal ability and thelog-transformed T cell response data, within individuals

subjected to both assays (n = 74), using Pearson’s product-

moment correlation (a = 0.05).

Results

Innate immune function: bactericidal ability

The overall model of roost type, colony within roost type

and body condition significantly explained variation in

bacterial killing ability (P = 0.03, F = 2.88(3,88),R2 = 0.12). Within the model, colony (within roost type)

was the only significant predictor of the variation in bac-

tericidal ability (P \ 0.05; Fig. 1a). Blood from females atDavis Cave had significantly lower killing ability

(20.19 ± 4.32) than the three other colonies; Frio Cave

(35.46 ± 4.60), Seco Creek Bridge (44.31 ± 7.59), andEast Elm Creek Bridge (38.68 ± 6.73). Other variables

including body condition (P = 0.77) and roost type [bridge

(41.15 ± 5.15) and cave (27.35 ± 3.22); P = 0.06] werenot significant predictors of bactericidal ability in female

bats, although roost type borders on significance. There

were no interactions between roost type or colony nestedwithin roost type and body condition.

Adaptive immune function: T cell mediated immuneresponse

The overall model of roost type, colony within roost typeand body condition significantly explained variation in

318 J Comp Physiol B (2009) 179:315–323

123

PHA response (P \ 0.01, F = 8.34(3,161), R2 = 0.18).

Within the model roost type and colony (within roost type)were both significant predictors of PHA response

(P \ 0.001 and P = 0.02, respectively). Females from

caves (1.68 ± 0.07) exhibited a greater T cell responsethan those living in bridges (1.05 ± 0.12). Females from

Davis Cave (1.54 ± 0.08) and Frio Cave (1.90 ± 0.10)

had larger responses than those at the bridge colonies, EastElm Creek (1.14 ± 0.21) and Seco Creek (1.01 ± 0.15;

Fig. 1b). Variation in T cell response was not explained by

body condition (P = 0.33). Again, there were no interac-tions between roost type or colony nested within roost type

and body condition.

Combined immune function

Within individual females, there was a weak but significantnegative correlation between bactericidal ability and T cell

response to PHA (Pearson’s q = -0.190, P = 0.05).

Females that produced larger swellings in response to PHAhad blood with lower bactericidal ability (Fig. 2).

Discussion

This study demonstrates that individual T. brasiliensisdiffer in their ability to mount functional immunological

responses, and that roosting ecology significantly impacts

immune function. We also provide evidence of a negativeassociation between innate and adaptive immune responses

within individuals.

Blood from female bats at Davis Cave had lower bac-terial killing ability compared to females at Frio Cave and

the two bridge sites (Fig. 1a). This trend is also present

when roost types are compared; suggesting that there areinherent differences between bridges and caves that likely

affect immunity. Although this result is not linked to body

condition, there may be additional factors that affect innateimmunity which are not included in our model. One pos-

sible factor could be differences in ectoparasitism among

the colonies surveyed. High parasite loads affect immunefunction adversely in a variety of vertebrates (Sheldon and

Verhulst 1996; Møller et al. 1999). Examination of infes-

tations of ectoparasitic mites on T. brasiliensis at thesesame sites show that individuals at Davis Cave, the site

with the lowest mean BKA index, had the highest degree of

mite cover, followed by individuals at Frio Cave and thenat bridge sites (Turmelle 2005). Although these types of

immune responses are not directly related, we suggest that

0

10

20

30

40

50

60B

KA

Inde

x

*

a

a,b

cb,c

0

0.5

1

1.5

2

2.5

Frio Davis Seco CreekBridge

East Elm Bridge

PH

A In

dex

CaveCave

a

b

Fig. 1 a BKA Index (bactericidal ability of blood; mean ± SE) atthe four colonies sampled, blood from bats at Davis Cave hadsignificantly lower bactericidal ability than blood from bats at theother colonies (P \ 0.05). Caves versus bridges (P = 0.06). Samplesizes for Frio and Davis Caves and Seco Creek and East Elm CreekBridges are n = 30, 34, 11, 14, respectively. b PHA Index (responseto PHA injection at 12 h; mean ± SE) at the four colonies sampled.Letters denote statistical differences between colonies, estimatedusing post-hoc analysis. Bats sampled at two cave colonies hadsignificantly greater response to PHA challenge than those at the twobridges (colony, P = 0.02; roost type, P = 0.001). Sample sizes forFrio and Davis Caves and Seco Creek and East Elm Creek Bridges aren = 50, 76, 24, 12, respectively

-50

0

50

100

-1 -0.5 0 0.5 1 1.5

PHA Index

BK

A In

dex

p = 0.05

Fig. 2 BKA index (bactericidal ability of blood) as a function ofPHA index (response to PHA injection at 12 h) in individual adultfemale bats tested for both. There was a negative correlation betweenBKA and PHA indices (Pearson’s q = -0.190, P = 0.05)

J Comp Physiol B (2009) 179:315–323 319

123

higher parasitism of bats in the colony at Davis Cave may

reflect a weakened immune system overall; however, theexact causative mechanisms are currently unknown.

Research on the relationships between ectoparasitism and

immune responses in T. brasiliensis is in progress.The lower T cell responses in females occupying bridges

compared to their cave-dwelling counterparts indicates that

adaptive immune response is also influenced by roost type(Fig. 1b). This supports our hypothesis that environmental

differences may exist between roost types. Bats living inbridges may experience physiological stress associated

with living in man-made roosts that negatively impact

adaptive immune responses. As reviewed by Nelson et al.(2000) immune suppression associated with stress hor-

mones has been demonstrated in several species of

vertebrates. Notwithstanding, evidence linking stress hor-mone levels and immune system function in the Brazilian

free-tailed bat is currently lacking. Alternatively, this pat-

tern may also reflect differences in colony size between theroost types. While our model does not directly test for the

effects of colony size on swelling response to PHA, a post-

hoc test shows a positive correlation (P \ 0.0001,F = 19.79(3,161), R2 = 0.11; Fig. 3) between mean PHA

response and the best current estimates of colony size

(Betke et al. 2008). Because colony size varied with roosttype, and the effects of both variables may be working in

combination, additional research will be needed to tease

apart these effects. Similarly, in an interspecific studyMøller et al. (2001) found that highly colonial swallows

and martins had more robust T cell responses compared to

less social songbirds, and they suggested that higher levelsof parasitism observed in larger colonies may cause

individuals to allocate more resources to combat infection.

Tella et al. (2001), found the opposite relationship, exam-ining colonies of the Magellanic penguin (Spheniscusmagellanicus), with larger colonies having reduced T cell

responses compared to smaller ones. The authors point outthat density dependent food limitations and crowding

negatively affected body condition and immunocompe-

tence in fledglings. Since we found PHA response toincrease with colony size, it is likely that these bats may

not be particularly sensitive to crowding and foodlimitations.

Because the adaptive arm of the immune system

responds to chronic infection (Klasing 2004), greaterinvestment in immunity is expected in organisms threa-

tened by a larger number of pathogens (Read and Allen

2000). Evidence suggests that bats in caves have higherectoparasite loads than those roosting beneath bridges

(Turmelle 2005). Differences in exposures to other patho-

gens may also vary by roost type. For example, once thespores of Histoplasma capsulatum and other fungi that

grow in the accumulated guano below bats become air-

borne they can infect lungs and mucosal membranes(McMurray and Russel 1982). Resistance to H. capsulatuminfection in mammals is dependent on a cellular immune

response primarily mediated by T cells (Cain and Deepe1998; Deepe 1994). One critical determinant of the course

of infection is the inflammatory response evoked in

response to host-pathogen interactions. The inability toevoke the appropriate inflammatory response can lead to

disease progression. Compared to bats roosting in caves,

bats roosting under bridges may be less exposed to fungidue to smaller accumulations of guano and greater air

circulation between the guano and the roosting bats. Thus,

we expect that higher pressures from parasites and fungi incaves may result in cave-roosting bats investing more

energy in mounting memory-related, adaptive, resistance to

the pathogens they commonly face. These findings are incontrast with those of Christe et al. (2000) who found that

reproductive females of Myotis myotis had lower T cell

responsiveness and higher mite loads than non-reproduc-tive females, and that during lactation immunocompetence

was positively correlated with body mass. In the current

study, we found no evidence for differences in T cellresponse due to body condition or reproductive stage.

Moreover, we found evidence for a positive relationship

between ectoparasitic mite loads and T cell response,although this pattern may be indirect owing to differences

in roosting condition in bridge and cave-roosting bats.

Results from our study have shown a negative correla-tion between PHA response and bactericidal ability within

individual bats (Fig. 2). Forsman et al. (2008) found sim-

ilar results, and showed that humoral bactericidal activitywas negatively related to cutaneous immune activity (PHA

0

0.5

1

1.5

2

0 500K 1000K 1500K

Colony Size

PH

A In

dex

p < 0.0001

Fig. 3 Mean PHA index (response to PHA injection at 12 h) as afunction of estimated colony size. Colonies estimates in order fromlargest to smallest are Frio Cave, Davis Cave, Seco Creek Bridge andEast Elm Creek Bridge. The positive correlation between colony sizeand mean PHA Index was significant (P \ 0.0001, F = 19.79(3,161), R2 = 0.11). Sample sizes for Frio and Davis Caves and SecoCreek and East Elm Creek Bridges are n = 50, 76, 24, 12,respectively

320 J Comp Physiol B (2009) 179:315–323

123

assay) in house wrens. Moreover, Martin et al. (2006b)

found that some immune responses can negatively affectother recent immunological activity in female white-footed

mice, when examining simultaneous wound healing and

cutaneous immune response. Our data suggests that indi-viduals may not be able to maximally activate all aspects of

immunity owing to competing costs and the variety of

strategies associated with a multiple-component immunesystem (Klasing 2004). T cell mediated inflammation aids

in the clearance of viruses (Hooper et al. 1998); thereforeindividuals at greater risk of coming in contact with viru-

ses, such as rabies in bats, may be selected to invest more

into adaptive immune defenses, potentially at the expenseof reduced innate immune responsiveness.

In addition to Rabies and other lyssaviruses, Brazilian

free-tailed bats may be incidental hosts to various arbovi-ruses, including the flaviviruses (Family Flaviviridae,

Genus Flavivirus) Rio Bravo Virus (RBV; Constantine and

Woodall 1964), St Louis encephalitis (SLE; Allen et al.1970; Herbold et al. 1983), and West Nile Virus (WNV;

Davis et al. 2005; Pilipski et al. 2004), and the alphaviruses

(Family Togaviridae, Genus Alphavirus) Eastern EquineEncephalitis (EEE), and Western Equine Encephalitis

(WEE) (Constantine 1970). The infection cycle for these

arboviruses occurs primarily between birds and arthropods,with incidental infection of mammals generally not pro-

ducing significant viremia to act as amplifying hosts or

facilitate transmission. Enzootic foci for arboviruses aregenerally swampy areas, drainages, or irrigated agricultural

land, with infection occurring primarily during the summer

months. Both risk factors are relevant for roosting aggre-gations of Brazilian free-tailed bats, particularly colonies

that roost over standing water (bridges) or near irrigated

agricultural lands (both caves and bridges). Infection withflaviviruses or alphaviruses should induce a viral neutral-

izing antibody response that confers lifelong immunity

against the specific virus responsible for infection. In thecontext of our data, we would predict that bats which

are unable to mount strong T cell mediated antibody

responses would be more likely to succumb to viralinfection. In subclinical cases, bats with lower T cell

mediated response may develop limited protection against

future or related exposures.Determining the ecological factors that predict variation

in the ability of individual Brazilian free-tailed bats to

exhibit functional immunological responses is important tounderstand disease dynamics and population health in this

and other species of colonial animals. A number of vari-

ables can affect the ability of organisms to combatpathogens and viral infections, and this study is a first step

toward understanding susceptibility related to immuno-

competence. A companion paper investigating linksbetween roosting ecology and disease exposure (Rabies

virus) is currently in review. Evidence from this current

study suggests that immune responsiveness and presumablydisease susceptibility are linked to the roosting ecology of

the host. We predict that the immunotypic composition of a

colony will influence pathogen transmission and persistencein free-ranging bats (Dimitrov et al. 2006, 2007).

Energetically costly aspects of immunological defense

may also impose tradeoffs and energetic limitations forother seasonal behaviors and life-history functions that are

important for survival. From this perspective the emergingfield of ecoimmunology has contributed to our deeper

understanding of the role of immunological competence in

population regulation and the evolution of coloniality(Lochmiller 1996). The implications of such studies are

further enhanced by examining species that are known to

play a role in infectious disease transmission.

Acknowledgments We thank Nickolay Hristov, Jonathan Reichard,Sarah Duncan, Cynthia Schmaeman, and Laura Borrelli for countlesshours of assistance with fieldwork. We thank Irving Marbach,Dwayne Davis, and Bill Cofer for allowing access to their property.Special thanks to Pat Morton and Meg Goodman (Texas Parks andWildlife), Mark Bloschock (Texas Department of Transportation),Bain Walker (Hill Country Adventure Tours), and Barbara French(Bat Conservation International) for support in the field, and toMargaret and J. David Bamberger (Selah-Bamberger Ranch) forhospitality. Additionally, we thank Christopher Richardson and JamesSullivan for help with statistical analysis. Funding was provided bygrant EID 0426082 from NSF/NIH Program for the Ecology ofInfectious Disease to G.F.M. and T.H.K. All experimental procedurescomply with the current laws of the United States of America.

References

Allen R, Taylor SK, Sulkin SE (1970) Studies of arthropod-bornevirus infections in Chiroptera: VIII. Evidence of natural St.Louis encephalitis virus infection in bats. Am J Trop Med Hyg19:851–859

Anthony ELP (1988) Age determination in bats. In: Kunz TH (ed)Ecological and behavioral methods for the study of bats.Smithsonian Institution Press, Washington, DC, pp 47–58

Betke M, Hirsch DE, Makris NC, McCracken GF, Procopio M,Hristov NI, Teng S, Bacchi A, Reichard JD, Horn JW, CramptonS, Cleveland CJ, Kunz TH (2008) Thermal imaging revealssignificantly smaller Brazilian free-tailed bat colonies thanpreviously estimated. J Mammal 89:18–24

Blount JD, Houston DC, Møller AP, Wright J (2003) Do individualbranches of immune defense correlate? A comparative case studyof scavenging and non-scavenging birds. Oikos 102:340–350

Blue CE, Spiller OB, Blackbourn DJ (2004) The relevance ofcomplement to virus biology. Virology 319:176–184

Bonforte R, Topilsky M, Siltzbach L, Glade P (1972) Phytohemag-glutinin skin test: a possible measure of cell-mediated immunity.J Pediatr 81:775–780

Botvinkin A, Poleschuk E, Kuzmin I, Borisova T, Gazaryan S, YagerP, Rupprecht C (2003) Novel lyssaviruses isolated from bats inRussia. Emerg Infect Dis 9:1623–1625

Cain JA, Deepe GS Jr (1998) Evolution of the primary immuneresponse to Histoplasma capsulatum in murine lung. InfectImmun 66:1473–1481

J Comp Physiol B (2009) 179:315–323 321

123

Calisher CH, Childs JE, Field HE, Holmes KV, Schountz T (2006)Bats: important reservoir hosts of emerging viruses. ClinMicrobiol Rev 19:531–545

Christe P, Arlettaz R, Vogel P (2000) Variation in intensity of aparasitic mite (Spinturnix myoti) in relation to the reproductivecycle and immunocompetence of its bat host (Myotis myotis).Ecol Lett 3:207–212

Cockrum EL (1969) Migration in the guano bat Tadarida braziliensis.Miscellaneous Publications, Museum of Natural History, Uni-versity of Kansas, vol 51, pp 303–336

Constantine DG (1967) Bat rabies in the south-western United States.Public Health Rep 82:867–888

Constantine DG (1970) Bats in relation to the health, welfare, andeconomy of man. In: Wimsatt WA (ed) Biology of bats.Academic Press, New York, pp 319–449

Constantine DG, Woodall DF (1964) Latent infection of Rio BravoVirus in salivary glands of bats. Public Health Rep 79:1033–1039

Davis RB, Herreid CFII, Short HL (1962) Mexican free-tailed bats inTexas. Ecol Monogr 32:311–346

Davis A, Bunning M, Gordy P, Panella N, Blitvich B, Bowen R(2005) Experimental and natural infections of North Americanbats with West Nile Virus. Am J Trop Med Hyg 73:467–469

Deepe GS (1994) The immune response to Histoplasma capsulatum:unearthing its secrets. J Lab Clin Med 123:201–205

Dimitrov DT, Hallam TG, McCracken GF (2006) Modeling theeffects of stresses on dynamics of bat rabies. In: Tian HQ (ed)Environmental modeling and simulation. ACTA Press, Calgary,pp 107–112

Dimitrov DT, Hallam TG, Rupprecht CE, Turmelle AS, McCracken GF(2007) Integrative models of bat rabies immunology, epizootiol-ogy, and disease demography. J Theor Biol 245:498–509

Dobson AP (2005) What links bats to emerging infectious diseases?Science 310:628–629

Dominguez SR, O’Shea TJ, Oko LM, Holmes KV (2007) Detectionof group 1 coronaviruses in bat in North America. Emerg InfectDis 13:1295–1300

Forsman AM, Vogel LA, Sakaluk SK, Grindstaff JL, Thompson CF(2008) Immune-challenged house wren broods differ in therelative strengths of their responses among different axes of theimmune system. J Evol Biol 21:873–878

Fraser GC, Hooper PT, Lunt RA, Gould AR, Gleeson LJ, Hyatt AD,Russell GM, Kattenbelt JA (1996) Encephalitis caused by alyssavirus in fruit bats in Australia. Emerg Infect Dis 2:327–331

Gloza-Rausch F, Ipsen A, Seebens A, Gottsche M, Panning M,Drexler JF, Petersen N, Annan A, Grywna K, Muller M,Pfefferle S, Drosten C (2008) Detection and prevalence patternsof group 1 coronaviruses in bats, Northern Germany. EmergInfect Dis 14:626–631

Goldsby RA, Kindt TJ, Osborne BA (2003) Kuby Immunology. W!H.Freeman and Company, New York

Halpin K, Young P, Field H, Mackenzie J (2000) Isolation of Hendravirus from pteropid bats: a natural reservoir of Hendra virus.J Gen Virol 81:1927–1932

Herbold JR, Huschele WP, Berry RL, Parsons MA (1983) Reservoirof St. Louis encephalitis virus in Ohio bats. Am J Vet Res44:1889–1893

Hooper DC, Morimoto K, Bette M, Weihe E, Koprowski H,Dietzschold B (1998) Collaboration of antibody and inflamma-tion in clearance of rabies virus from the central nervous system.J Virol 72:3711–3719

Horn JW, Kunz TH (2008) Analyzing NEXRAD Doppler radarimages to assess nightly dispersal patterns and population trendsin Brazilian free-tailed bats (Tadarida brasiliensis). Integr CompBiol 48:24–39

Joharra M, Field H, Rashdi A, Morrissy C, van der Heide B, Rota P,Adzhar A, White J, Daniels P, Jamaluddin A, Ksiazek T (2001)

Nipah virus infection in bats (Order Chiroptera) in peninsularMalaysia. Emerg Infect Dis 7:439–441

Keeley ATH, Keeley BW (2004) The mating system of Tadaridabrasiliensis (Chiroptera: Molossidae) in a large highway bridgecolony. J Mammal 84:113–119

Kennedy MW, Nager RG (2006) The perils and prospects of usingphytomaemagglutinin in evolutionary ecology. Trends Ecol Evol21:653–655

Klasing KC (2004) The costs of immunity. Acta Zool Sin 50:961–969Kunz TH (1982) Roosting ecology of bats. In: Kunz TH (ed) Ecology

of bats. Plenum Publishing, New York, pp 1–55Kunz TH, Nagy KA (1988) Energy budget analysis. In: Kunz TH (ed)

Ecological and behavioral methods for the study of bats.Smithsonian Institution Press, Washington, DC, pp 283–285

Kunz TH, Reynolds DS (2003) Bat colonies in buildings. In: O’Shea TJ,Bogan MA (eds) Monitoring trends in bat populations of theUnited States and Territories: problems and prospects USGeological Survey, Biological Resources Discipline, Informationand Technology Report, USGS/BRD/ITR-2003-003, pp 91–102

Lausen CL, Barclay RMR (2006) Benefits of living in a building: bigbrown bats (Eptesicus fuscus) in rocks versus buildings.J Mammal 87:362–370

Leroy EM, Kumulungui B, Pourrut X, Rouquet P, Hassanin A, YabaP, Delicat A, Paweska JT, Gonzalez JP, Swanepoel R (2005)Fruit bats as reservoirs of Ebola Virus. Nature 438:575–576

Li W, Shi Z, Yu M, Ren W, Smith C, Epstein JH, Wang H, CrameriG, Hu Z, Zhang H, Zhang J, McEachern J, Field H, Daszak P,Eaton BT, Zhang S, Wang L (2005) Bats are natural reservoirs ofSARS-like Coronaviruses. Science 310:676–679

Lochmiller RL (1996) Immunocompetence and animal populationregulation. Oikos 76:594–602

Lollar A, Schmidt-French B (1998) Captive care and medicalreference for the rehabilitation of insectivorous bats. Bat WorldPublication, Mineral Wells

Martin LBII, Han P, Lewittes J, Kuhlman JR, Klasing KC, WikelskiM (2006a) Phytohemagglutinin-induced skin swelling in birds:histological support for a classic immunoecological technique.Funct Ecol 20:290–299

Martin LBII, Weil ZM, Kuhlman JR, Nelson RJ (2006b) Trade-offswithin the immune systems of female white-footed mice,Peromyscus leucopus. Funct Ecol 20:630–636

Matson KD, Tieleman BI, Klasing KC (2006) Capture stress and thebactericidal competence of blood and plasma in five species oftropical birds. Physiol Biochem Zool 79:556–564

McCracken GF (2003) Estimates of population sizes in summercolonies of Brazilian free-tailed bats (Tadarida brasiliensis). In:TJ O’Shea, MA Bogan (eds) Monitoring trends in bat popula-tions of the United States and Territories: problems andprospects US Geological Survey, Biological Resources Disci-pline, Information and Technology Report, USGS/BRD/ITR-2003-003, pp 21–30

McMurray DN, Russel LH (1982) Contribution of bats to themaintenance of Histoplasma capsulatum in a cave microfocus.Am J Trop Med Hyg 31:527–531

Merchant ME, Roche C, Elsey RM, Prudhomme J (2003) Antibac-terial properties of serum from the American alligator (Alligatormississippiensis). Comp Biochem Physiol B 136:505–513

Messenger SL, Rupprecht CE, Smith JS (2003) Bats, emerging virusinfections, and the rabies paradigm. In: Kunz TH, Fenton MB(eds) Bat ecology. The University of Chicago Press, Chicago, pp622–679

Møller AP, Christe P, Lux E (1999) Parasitism, host immunefunction, and sexual selection. Q Rev Biol 74:3–20

Møller AP, Merino S, Brown CR, Robertson RJ (2001) Immunedefense and host sociality: a comparative study of Swallows andMartins. Am Nat 158:136–145

322 J Comp Physiol B (2009) 179:315–323

123

Navara KT, Hill GE, Mendonca MT (2005) Variable effects of yolkandrogens on growth, survival, and immunity in eastern bluebirdnestlings. Physiol Biochem Zool 78:570–578

Nelson RJ, Demas GE, Klein SL, Kriegsfield LJ (2000) SeasonalPatterns of Stress, Immune Function and Disease. CambridgeUniversity Press, Cambridge

Neubaum D, O’Shea T, Wilson K (2006) Autumn migration andselection of rock crevices as hibernacula by big brown bats inColorado. J Mammal 87:470–479

O’Shea TJ, Bogan MA (2003) Monitoring trends in bat populations ofthe United States and territories: problems and prospects. USGeological Survey, Biological Resources Discipline, Informa-tion and Technology Report, USGS/BRD/ITR-2003-003

Pilipski JD, Pilipski LM, Risley LS (2004) West Nile Virus antibodiesin bats from New Jersey and New York. J Wildlife Dis 40:335–337

Read AF, Allen JE (2000) The economics of immunity. Science290:1104–1105

Sgro MP, Wilkins KT (2003) Roosting behavior of the Mexican free-tailed bat (Tadarida brasiliensis) in a highway overpass. West NAm Nat 63:366–373

Sheldon BC, Verhulst KT (1996) Ecological immunology: costlyparasite defenses and trade-offs in evolutionary ecology. TrendsEcol Evol 11:317–321

Tella JL, Forero MG, Bertellotti M, Donazar JA, Blanco G, CeballosO (2001) Offspring body condition and immunocompetence arenegatively affected by high breeding densities in a colonialseabird: A multiscale approach. Proc R Soc London B268:1455–1461

Tella JL, Scheuerlein A, Ricklefs RE (2002) Is cell-mediatedimmunity related to the evolution of life-history strategies inbirds? Proc R Soc London B 269:1059–1066

Tieleman BI, Williams JB, Ricklefs RE, Klasing KC (2005)Constitutive innate immunity is a component of the pace-of-life syndrome in tropical birds. Proc R Soc London B 272:1715–1720

Turmelle AS (2005) Ecology of parasitism on Brazilian free-tailedbats (Tadarida brasiliensis). Bat Res News 46:83–84

Wibbelt G, Kurth A, Yasmum N, Bannert M, Nagel S, Nitsche A,Ehlers B (2007) Discovery of herpesviruses in bats. J GenerVirol 88:2651–2655

J Comp Physiol B (2009) 179:315–323 323

123

Birth size and postnatal growth in cave- and bridge-roostingBrazilian free-tailed batsL. C. Allen1, C. S. Richardson1, G. F. McCracken2 & T. H. Kunz1

1 Department of Biology, Center for Ecology and Conservation Biology, Boston University, Boston, MA, USA

2 Department of Ecology and Evolutionary Biology, University of Tennessee, Knoxville, TN, USA

Keywords

bats; human-made roosts; intraspecific

variation; postnatal growth; roost quality.

Correspondence

Louise Allen, Boston University,

5 Cummington St., Boston, MA 02215,

USA.

Email: [email protected]

Editor: Virginia Hayssen

Received 24 February 2009; revised 16 July

2009; accepted 22 July 2009

doi:10.1111/j.1469-7998.2009.00636.x

Abstract

As the human population continues to expand, increased encroachment on naturallandscapes and wildlife habitats is expected. Organisms able to acclimate tohuman-altered environments should have a selective advantage over those unableto do so. Over the past two decades, bats have increasingly begun to roost and raiseoffspring in spaces beneath pre-cast concrete bridges. Few studies have examinedthe health or fitness of individuals living in these anthropogenic sites. In the presentstudy, we examined birth size and postnatal growth, as surrogates of reproductivesuccess, in Brazilian free-tailed bat pups born at a natural and a human-maderoost. Based on putative stress-related conditions (noise from vehicular traffic,chemical pollutants and a modified social environment) present at bridges, wepredicted that bats at these sites would have reduced reproductive success.Contrary to our prediction, pups born at a bridge site were on average heavierand larger at birth and grew faster than those born at a cave site. Also, both birthsize and growth rates of pups differ between years. We attribute observeddifferences to a combination of roost-related conditions (i.e. roost temperatureand proximity to foraging areas), climate and maternal effects with larger mothersraising larger pups. Thus, some bridge roosts, at least in the short term, aresuitable, and in some cases may provide better conditions, for raising young batpups than cave roosts.

Introduction

Human population growth and associated exploitation ofnatural resources may be the greatest threat to wildlifeacross the globe. Population expansion leads to increasedencroachment on natural landscapes and wildlife habitats,both directly (e.g. human/wildlife conflicts) and indirectly(e.g. habitat degradation). Individuals within a species thatacclimate to human-altered landscapes should have a selec-tive advantage over those unable to do so. Human actionsmay have a positive influence on wildlife by generatingfeeding and roosting opportunities (Fenton, 1997; Cleve-land et al., 2006; Lausen & Barclay, 2006). Human activities,such as road construction, tourism and habitat fragmenta-tion can have significant effects on behavior and stress levelsin diverse wildlife (Creel et al., 2002; Walker, Boersma &Wingfield, 2005a; Bejder et al., 2006; Barja et al., 2007).However, the extent to which these activities directly orindirectly affect long-term fitness is poorly understood(Walker, Boersma & Wingfield, 2005b; Arlettaz et al.,2007). Thus, studies that directly measure the consequencesof anthropogenic factors on fitness are important.

One component of fitness is reproductive success. Becausea direct measure of lifetime reproductive success is often

difficult, a variety of substitutes, including number of off-spring produced (Verhulst, van Balen & Tinbergen, 1995),offspring survival (Sockman & Schwabl, 2000), birth orweaning size (Harding et al., 2005) and postnatal growth(Setchell et al., 2001), have been used. Birth size and postnatalgrowth of young may be the most appropriate indicators ofreproductive success for species in which the clutch or littersize is one or where following individuals past weaning isdifficult, due to natal dispersal or large colony sizes.

Various factors influence postnatal growth rates andsurvivorship in young mammals, including climate, foodavailability, roost temperature, age, nutritional and hormo-nal condition of the mother, parasite loads and colony size(Kunz, Adams & Hood, 2009). To our knowledge, nostudies have quantified possible effects of anthropogenicfactors, such as man-made roosts or human-altered foraginghabitats, on postnatal growth rates. The production ofoffspring is an energetically costly activity for mammals(Gittleman & Thompson, 1988; Liu, Wang & Wang, 2003),particularly in bats (Barclay, 1994; Kunz et al., 2009), inwhich newborn pups, on average, can weigh a quarter oftheir mother’s body mass (BM), and mothers nurse theirpup(s) until they are almost adult size (Kunz & Robson,1995; Kunz et al., 2009).

Journal of Zoology

Journal of Zoology 280 (2010) 8–16 c! 2009 The Authors. Journal compilation c! 2009 The Zoological Society of London8

Journal of Zoology. Print ISSN 0952-8369

mailto:[email protected]

The Brazilian free-tailed bat Tadarida brasiliensis is anexcellent model species to investigate the effects of anthro-pogenic factors on reproductive success because it occupiesa variety of roost types – some of which have been con-structed by humans. Over the past two decades this specieshas begun to rear pups in the expansion joints, or crevices,formed by the construction of pre-cast concrete bridges.These roosts may provide important refuges (Keeley &Tuttle, 1999). However, limited research has evaluated therelative health or fitness of individuals roosting in thesestructures compared with those that roost in natural caves(Allen et al., 2009; Turmelle et al., in press). Examining birthsize and postnatal growth at human-made roosts can lendinsight into what influence anthropogenic pressures have onboth parents and offspring.

Bats that roost beneath highway bridges are exposed toloud noises and vibrations associated with vehicular traffic.These novel stimuli could represent environmental stressorsthat may negatively alter physiology and reproduction inthese bats. Increased stress levels are associated with re-duced growth in diverse taxa (fish, Van Weerd & Komen,1998; Gregory & Wood, 1999; rats, Lordi et al., 2000;primates, Schneider et al., 2003). If conservation is apriority, and action is to be taken to mitigate possiblepopulation declines in this and other species, knowledgeabout reproductive success and the variables influencingpostnatal growth and survival is important for understand-ing how species may adapt to human-dominated landscapes.

The goals of the present study were to investigate intras-pecific variation in size at birth and postnatal growth of free-ranging Brazilian free-tailed bats born in two different roosttypes and to examine how environmental factors influencethese life-history traits. We postulate that environmentalfactors, both large scale (i.e. climatic conditions) and smallscale (i.e. roost conditions), will influence postnatal growth.We predicted that cave roosts are warmer and more ther-mally stable than bridges, owing to differences in rooststructure. We also predicted that because of expected roosttype differences, bats born and reared in bridges would beborn smaller and grow slower than those born and raised incaves.

Materials and methods

Study species and sites

In our study area, the Brazilian free-tailed bat T. brasiliensismigrates fromMexico to Texas each spring and congregatesin large maternity colonies where pups are born and nursed(Wilkins, 1989). Females give birth to a single pup each year.Like most North American bat species, females do not carrytheir young with them but leave them in clusters, or creches,with other similar aged pups (Kunz & Robson, 1995), whilethe adults emerge and forage at night (Davis, Herreid &Short, 1962; Kunz, Whitaker & Wadanoli, 1995; Reichardet al., 2009). Thus, once females give birth, we assume thatmother and offspring remain at the same site until the pupreaches independence or until one dies.

Our study was conducted during the summers of 2005and 2006 in south-central Texas. Study sites included onenatural cave (Davis Blowout Cave, Blanco County) and onehuman-made bridge (McNeil Bridge, Williamson County).Davis Blowout Cave is home to over 400 000 T. brasiliensis(Betke et al., 2008). The cave is surrounded by hills, cattlepastures and landscapes dominated by live oak and mes-quite. McNeil Bridge houses an estimated 600 000 T. brasi-liensis (L. C. Allen and T. H. Kunz, unpubl. data). Thebridge is situated along Interstate 35, with eight lanes (fournorth- and four south-bound) over the roosting bats. Thisbridge is surrounded by a suburban landscape and byremnant pasture and croplands.

Birth size and growth rate measurements

To measure growth rates, neonates were hand-capturedimmediately after evening emergence of adults. Pups wereplaced into cloth bags and their original location noted. Thebags were then placed on commercially available foot war-mers (Grabbers, Grand Rapids, MI, USA) to ensure thatthe enclosed pups did not become cold while outside theroost. Individuals were assumed to be a day old (day=1) ifthey still possessed an attached umbilical cord (Kunz, 1973).Each pup was marked with an individually numbered2.9mm flanged, metal alloy wing band (Porzana Ltd., EastSussex, UK). Any placenti still attached were removedbefore to weighing the neonates. Bats were weighed to thenearest 0.01 g with an electronic balance (SP202, Ohaus, PineBrook, NJ, USA). Length of the right forearm was measuredto the nearest 0.01mm with digital calipers. Length of thetotal epiphyseal gap (gap; the cartilaginous metacarpal–pha-langeal epiphyseal growth plates of the fourth digit) wasmeasured to the nearest 0.1mm using an ocular micrometerfitted to a zoom dissecting microscope (Kunz & Robson,1995; Kunz et al., 2009). An LED headlamp (Black Dia-mond, Salt Lake City, UT, USA) and the substage mirror ofthe microscope were used to transilluminate the right wing toallow accurate measurement of the gap. Pups were returnedwithin 2 h to the site of original capture before the return ofadults from foraging. In 2005, we captured 94 and 37 one-day-old bat pups from Davis Blowout Cave and McNeilBridge, respectively. Additionally, in 2006 we captured 171and 107 one-day-old pups from the same sites, respectively.To maximize our 1-day-old sample size to include indivi-duals whose umbilical cord may have fallen off either beforecapture or before processing while being held in the collec-tion bags, we included individuals without an umbilicus, butwhose length of forearm (FL) was equal to themean plus onestandard deviation (mean+SD) in size of those captured withan umbilical cord (Hoying & Kunz, 1998). This cutoff isconservative (compared with Kunz &Robson, 1995), and weretained a large sample of 1-day-old pups that were laterrecaptured (2005, n=149; 2006, n=232).

We returned to each site at 2- to 4-day intervals andrecaptured as many marked pups as possible, repeating themorphological measurements. As young became capableof flight, we recaptured marked bats and recorded

Journal of Zoology 280 (2010) 8–16 c! 2009 The Authors. Journal compilation c! 2009 The Zoological Society of London 9

Roosting ecology and postnatal growth in batsL. C. Allen et al.

measurements beginning c. 3 h before evening emergence.This made possible the capture of older, volant pups thatmay have been missed by only sampling following nightlyemergence. The capture protocol caused minimal distur-bance to adults in the colony because, during lactation,adult female T. brasiliensis roost separately from theiryoung when not nursing (Kunz & Robson, 1995). Onlymeasurements recorded on pups that were recaptured atleast once were included in the analyses of growth rates. Allcapture, handling and experimental procedures were per-formed in a humane manner, following national legislationand published animal use guidelines (Gannon & Sikes, 2007)and were under protocols approved by the Boston Univer-sity Animal Care and Use Committee (05-012) to T. H. K.and under Texas Parks and Wildlife Department permitSPR-0305-058 to T. H. K.

Roost temperature and weather

Roost temperatures near creches of pups were recordedduring two 3-week periods in June and August, using dataloggers (iButtons, Maxim, Sunnyvale, CA, USA). Ambienttemperatures were obtained from loggers placed in shadedlocations outside the roost. Additionally, local monthlytemperature and precipitation data (September–August)were obtained from databases maintained by a local weatherstation (Austin-Bergstrom International Airport; KAUS)located within 35 and 80 km of the two sites (MesoWest, TheUniversity of Utah; http://www.met.utah.edu).

Statistical analysis

FL is the most important character for measuring growthand estimating age during the early postnatal period (age1–21days) and the change in gap length is important forestimating age during the later portion of postnatal growth(age 21–42days) (Kunz & Robson, 1995; Kunz et al., 2009).For statistical analyses of all growth variables, we used thelinear region of the curves because the complete empirical

growth curves for both sites are curvilinear (i.e. they do notfit a simple linear model). A linear model is the traditionalmethod for determining whether growth rates differ betweengroups. We determined the linear region of growth curves byinspection of current data and from published growth curvesfor this species (Kunz & Robson, 1995). The coefficient ofdetermination (R2) from simple regressions of the linearregions of the growth curves for each year and each site werehigh (R240.85, Po0.001; Table 1). Thus, a linear analyticalmodel was appropriate. Because we found no significantdifferences between sexes at birth (ANOVA on variables atbirth: FL: F1,407=1.31, P=0.25; BM: F1,407=0.34,P=0.56; length of gap: F1,407=0.73, P=0.22), and becausea previous study on this species found no sex differences ingrowth rates (Kunz & Robson, 1995), we combined data forboth males and females in all analyses.

Before testing for site differences, we tested for differencesbetween years in variables at birth and in growth rates foreach site. If years differed within a site, we compared bothsites separately for each year. For each growth variable, wecompared the sites using a generalized estimating equation(GEE) model. A GEE model is based on an iterativeprocedure to estimate regression coefficients and P-values(Zeger & Liang, 1986) and accounts for correlations amongobservations on the same individuals.

We compared sites using GEE regression models thatinclude year or site as a grouping factor, age as a covariate,and an interaction term (age" year or site). For each model,we used the partial F-statistic for the interaction term toperform a test for parallelism to compare the slopes describ-ing growth rates (variable vs. age) for each site and thepartial F-statistic for the site term to perform a test for equalintercepts to compare the estimated size at day 1 (y-inter-cept) of each growth variable for each site. Additionally, wetested for differences at birth in mass and FL between sitesusing ANOVA. All data were normally distributed. Statis-tical analyses were performed using SAS version 6.12 (SASInstitute Inc., Cary, NC). Independent variables werejudged to be significant at Po0.05.

Table 1 Linear regression of postnatal growth parameters on agea in Tadarida brasiliensis for each site in each year (Po0.001 for all comparisons)

Growth parameter Year, site d.f.b Interceptc Growth rate R2

Length of forearm 2005, McNeil Bridge 139 17.90 mm +0.89 mm day#1 0.96

2005, Davis Cave 225 17.60 mm +0.76 mm day#1 0.95

2006, McNeil Bridge 320 17.12 mm +0.88 mm day#1 0.96

2006, Davis Cave 360 16.69 mm +0.84 mm day#1 0.95

Body mass 2005, McNeil Bridge 139 3.43 g +0.35 g day#1 0.91

2005, Davis Cave 225 3.00 g +0.29 g day#1 0.88

2006, McNeil Bridge 317 3.12 g +0.33 g day#1 0.92

2006, Davis Cave 360 2.91 g +0.31 g day#1 0.87

Length of epiphyseal gap 2005, McNeil Bridge 56 5.77 mm #0.10 mm day#1 0.85

2005, Davis Cave 74 5.82 mm #0.10 mm day#1 0.86

2006, McNeil Bridge 133 6.20 mm #0.11 mm day#1 0.86

2006, Davis Cave 107 6.29 mm #0.11 mm day#1 0.86

aLength of forearm and body mass age range: 1–21 days; gap age range: 22–42 days.bFor residual sources of variation.cFor gap length, intercept refers to estimated value at age=0, however this regression is only used on bats Z22 days.


Roosting ecology and postnatal growth in bats L. C. Allen et al.

Results

Contrary to our initial predictions pups born at the bridgesite were larger at birth and grew at a faster rate than pupsborn at the cave site.

Birth size

Pups were born at 39.5–42% (FL) and 26–29% (BM) ofadult size, dependent on site and year of birth. Birth sizevaried between the 2 years (FL, Po0.001; BM, Po0.001),with both FL (17.20$ 0.04mm) and BM (2.98$ 0.02 g)significantly smaller in 2006 than in 2005 (FL=17.90$0.06mm; BM=3.08$ 0.03 g). Thus, because of differencesin size at birth between years at each site, we compared sitesseparately for each year. Roost was a significant predictor ofbirth size in both years (Figs 1a and b). In 2005, bats born atthe bridge (18.10$ 0.12mm) were slightly larger than batsborn at the cave (17.83$ 0.07mm; P=0.054; Fig. 1a). Inthe same year, bats born at the bridge (3.29$ 0.04 g) were c.10% heavier at birth than those born at the cave(2.99$ 0.03 g; Po0.001; Fig. 1b). In 2006, bats at the bridge

(17.38$ 0.07mm) were slightly larger at birth than batsborn at the cave (17.09$ 0.06mm; P=0.001; Fig. 1a). Alsoin 2006, bats born at the bridge (3.07$ 0.03 g) were c. 5%heavier than those born at the cave (2.92$ 0.02 g; Po0.001;Fig. 1b).

Postnatal growth

Because differences exist between years for growth para-meters at one or both sites, we compared sites for each yearseparately. When the growth rates were different betweenyears, bats born in 2006 grew significantly slower than thoseborn in 2005. During initial stages of postnatal growth(1–21 days) for both sites, BM and FL increased linearly(Table 1, Fig. 2). In the later portion of the postnatal growthperiod (22–42 days), the length of the total gap decreasedlinearly (Table 1, Fig. 3). However, the initial sizes(y-intercepts) and rates of growth (slope of variable on age)of these parameters differed between the two sites (Table 2).For FL and BM, in each year, tests of parallelism (interac-tion term) and tests for equal intercepts (site term) weresignificant (Table 2), except for BM in 2006 (P=0.073).Thus, for each site in each year, the growth rate lines startwith different intercepts and the lines diverge (Fig. 2). Fortotal gap length, in both years, tests for equal intercepts (siteterm) were significantly different (P % 0.05), while tests ofparallelism (interaction term or growth rate) were not(Table 2). Thus, for each site in each year, the growth ratelines start with different intercepts, but are parallel (Fig. 3).

Roost temperature

We examined roost temperatures during two 3-week peri-ods, one at the start of lactation and one at the end oflactation (during June and early-August). The mean roosttemperatures at the bridge site were 2 1C warmer than thoseat the cave site (bridge=33.4$ 0.25 1C, cave=31.4$ 0.25 1C; Po0.001). The temperature range (differencebetween minimum and maximum temperatures during aday) did not differ between the two sites (P=0.97). Ambienttemperature recorded in shady locations outside each roostdid not differ between the two sites (P=0.20).

Discussion

We predicted, due to roosting conditions, that bats born atthe bridge would be smaller at birth and grow more slowlythan those raised in the cave. However, our results contra-dict these hypotheses. Bats born in McNeil Bridge werelarger at birth in both BM and FL compared to bats born inDavis Blowout Cave (Fig. 1). Also during the first 3weeks ofpostnatal growth, pups born at the bridge site grow at afaster rate than pups born at the cave site, as measured byFL (Fig. 2b). The same factors may influence differences inboth birth size and postnatal growth between the sites;however, these ecological factors are not mutually exclusive.

Roost temperature can influence birth size and postnatalgrowth. Thermal stability influences roost choice in bats(Lourenco & Palmeirim, 2004) and low temperatures can

16

16.5

17

17.5

18

18.5

3

2005 2006

(a)

(b)

Leng

th o

f for

earm

(mm

)B

ody

mas

s (g

)

2.6

2.8

3.2

3.4

Figure 1 Comparison of mean birth size of Tadarida brasiliensis born

at a cave and a bridge. Dark bars represent bats born at the cave site

and light bars represent bats born at the bridge site. (a) Size at birth, as

measured by length of forearm in millimeters (2005, P=0.054; 2006,

Po0.001). (b) Size at birth as measured by body mass in grams (2005,

Po0.001; 2006, Po0.001).



delay gestation, reduce birth size (Hood, Bloss & Kunz,2002; Willis, Brigham & Geiser, 2006) or slow postnatalgrowth (Tuttle, 1975; Hoying & Kunz, 1998; Reiter, 2004).The Brazilian free-tailed bat has a high thermal neutralzone; estimated between 35 and 38 1C (Herreid & Schmidt-Nielsen, 1966). We found warmer roost temperatures in thebridge crevices compared to Davis Blowout Cave. Althoughthe temperatures recorded during our two sampling periodswere at the lower end of the thermal neutral zone, tempera-tures are probably elevated within groups of bats at bothsites. Thus, while pups born at the cave site experiencedlower roost temperatures, were smaller at birth, and grewmore slowly, the energetic costs of thermoregulation may besimilar at both roosts for both parents and offspring, due toclustering. Other important ecological factors may alsoinfluence the observed colony level differences in postnatalgrowth.

Female bats living in bridges appear to be in better bodycondition (ratio of BM to FL) than females at caves (L. C.Allen, unpubl. data). Although we did not capture andmeasure mother–pup pairs in this study, analysis of mea-surements from a separate group of mothers and pups,sampled from Davis Blowout Cave in 2006, found that aftercontrolling for pup age, females in better body conditionraise pups in better body condition (R2=0.32, Po0.0001,F=10.27). Hood et al. (2002) showed that bats born toolder mothers (middle-age and older) had greater growthrates compared to those born to young mothers, however

they did not relate age differences to differences in maternalbody condition. Possibly heavier bats may simply give birthto and raise larger offspring (Ralls, 1976; Festa-Bianchet,Gaillard & Jorgenson, 1998).

Differential energy expenditure for foraging may contri-bute to larger birth size at bridges. NEXRAD (Next Gen-eration RADAR) data of nightly dispersal from severalcolonies over a large spatial scale in Texas show that batsliving under bridges have shorter nightly foraging commutesthan bats living in caves (Horn & Kunz, 2008). Many of thebridges occupied by Brazilian free-tailed bats in Texas arelocated close to prime foraging sites in agricultural areas,whereas most cave sites are not. At bridges, shortenedcommuting and foraging times would allow pregnant fe-males to allocate more resources into reproductive growthand spend less energy on foraging. Proximity to foragingwould also influence the differential growth rates observedat the two sites. Mothers that spend less time foraging couldallocate more energy and nutrients to milk production. Inour system, the unique environment and location of bridgesites may give pups born at these sites the advantage ofwarmer roosts, without the disadvantage of increased in-traspecific competition for food that may occur with in-creased colony size.

Bats born in 2005 were larger at birth than those born in2006. This may reflect prolonged drought conditions presentin south-central Texas during the 2 years of our study.Although average monthly temperatures did not differ from

0

2

4

6

8

10

12

2006

2005

0

2

4

6

8

10

12

0 10 15 20 25

15

20

25

30

35

402005

2006

15

20

25

30

35

40

0 10 15 20 25

(a) (b)Le

ngth

of f

orea

rm (

mm

)Le

ngth

of f

orea

rm (

mm

)

Bod

y m

ass

(g)

Bod

y m

ass

(g)

Age (days)Age (days)

5 5

Figure 2 A comparison of age versus size in Tadarida brasiliensis at two sites, based on the early linear portion of postnatal growth (days 1–21).

The cave site is represented by open circles and a solid line and the bridge site is represented by closed circles and a dashed line. (a) Age versus

length of forearm [2005 generalized estimating equation (GEE) slope, Po0.001; 2006 GEE slope, Po0.05]. (b) Age versus body mass (2005 GEE

slope, Po0.01; 2006 GEE slope, P=0.07).



area norms, rainfall did. While other studies (Hoying &Kunz, 1998; Hood et al., 2002) report decreased birth size orgrowth rates associated with cold and/or wet weather, wefound the opposite relationship. During 2005 and 2006,

precipitation levels were well below the average yearly rain-fall for the region. The yearly precipitation measured at anearby weather station was 520mm in 2005 and 660mm in2006, or 39 and 22% below 30-year average values, respec-tively. Although precipitation was slightly higher in 2006than in 2005, 2 continuous years of drought conditions mayhave greatly reduced insect abundance and foraging oppor-tunities for aerial-feeding insectivorous bats. By contrast,precipitation in 2004 was above or near normal. Theseconditions may have delayed the decrease in insect abun-dance in 2005 and thus led to smaller birth size and slowergrowth rates in 2006. In the south-western United States,arid and severe drought conditions may decrease overallinsect abundance, if insect food resources and breedingopportunities are limited where water is scarce.

Although our data did not show significant differencesin ossification rates between sites (i.e. rate of change of gaplength), during the last 3 weeks of postnatal growth, thegap is smaller at any given age in both 2005 and 2006 inbats born at the bridge site (Fig. 3). In addition, pups bornat the bridge site attained a greater proportion of adult size(measured by FL) than pups born at the cave site. Thisdifference is more pronounced in 2005, but the differencein growth rates between the sites was also significant in2006 (Table 2; Fig. 2). Tadarida brasiliensis attains firstflight with a FL on average 82.6% of adult size (Pagels &Jones, 1974; reviewed by Barclay, 1994). If pups attainflight at this size, then in 2005 pups born at the bridge sitewould have attained flight nearly 5 days earlier, during a42-day growth period, than those born at the cave site.These bats may be at a selective advantage. Because pups,to some extent, still nurse from their mothers as they learnto forage, attaining flight at an earlier age would allowbats to put on more BM before being weaned. Pressures onpostnatal growth may be relaxed for migrating species anda high mass at weaning may not be vital for over-wintersurvival in due to adequate food abundance in wintering

0

2.5

5

2006

2005

0

2.5

5

20 25 30 35 40 45

Leng

th o

f gap

(m

m)

Leng

th o

f gap

(m

m)

Age (days)

Figure 3 A comparison of age versus epiphyseal gap length at two

sites, based on the latter linear phase of postnatal growth (days

22–42). The rate of change did not differ across sites in either year

[slope, generalized estimating equation (GEE); P40.05], but the start

and end points are differed in both years (2005 GEE intercept,

P=0.02; 2006 GEE intercept, P=0.052). The cave site is represented

by open circles and a solid line and the bridge site is represented by

closed circles and a dashed line.

Table 2 Outcome of generalized-estimating equation (GEE) analysis of growth parameters on age, site and age by site in Tadarida brasiliensis in

2005 and 2006

Growth parameter Source of variation

2005 2006

d.f. P d.f. P

Length of forearm Site 1 0.020 1 o0.001

Age 1 o 0.001 1 o 0.001

Interaction (age" site) 1 o 0.001 1 0.041

Residual 149a 232a

Body mass Site 1 o 0.001 1 o 0.001

Age 1 o 0.001 1 o 0.001

Interaction (age" site) 1 0.002 1 0.073

Residual 149a 232a

Length of epiphyseal gap Site 1 0.020 1 0.052

Age 1 o 0.001 1 o 0.001

Interaction (age" site) – –b – –b

Residual 76a 140a

aNumber of clusters used in GEE analysis, not degrees of freedom (a cluster is a set of observations for one individual).bInteraction term was not significant and thus removed from model.



ranges (Koehler & Barclay, 2000). This argument couldhold for Brazilian free-tailed bats that migrate to Mexicoin the fall. However, even in migrating species, young thatcomplete growth earlier and achieve foraging indepen-dence sooner should have a survival advantages in theirfirst year over slower developing pups, especially duringfall migration.

The comparison of closely related species and distantpopulations of single species may document fine-tuning ofgrowth rates to environmental conditions (Case, 1978;Arendt, 1997). We provide evidence of environmental con-ditions influencing the plastic and adaptive trait of postnatalgrowth in a single species within a single population,between different colonies. Postnatal growth in bats may behighly influenced by environmental factors, even at themicro-geographic or within-population (i.e. between col-ony) level. Further, both reproductive success and environ-mental variables should be considered when addressingconservation needs. We initially predicted that bats born atbridges would be born smaller and grow more slowly thanthose born and raised in caves. However, our results contra-dict these hypotheses. Results from our study indicate thatsome bridges may make suitable roosts, at least in the shortterm and may be better for raising young bat pups.Although, we were only able to assess reproductive successat one roost of each type, we have found consistent roost-type patterns in other measures (immune function and stresshormone levels) and we suggest that ecological variation inhealth measures may be influenced by predictable differ-ences in roost environment (Allen, 2009; Allen et al., 2009).However, the long-term effects of roosting in man-madestructures on overall population health and sustainability iscurrently unknown.

Acknowledgments

We thank Nickolay Hristov, Kaitlin Farrell and LauraBorrelli for many hours of assistance in the field. We thankDwayne Davis for allowing access to Davis Blowout Cave.Special thanks to Mark Bloschock (Texas Department ofTransportation), and Barbara French (Bat ConservationInternational) for support in the field, and to Margaret andJ. David Bamberger (Selah-Bamberger Ranch) for theirhospitality. We thank Bram Lutton, Nickolay Hristov andanonymous reviewers for comments on earlier versions ofthis paper. Funding was provided by grant EID 0426082from NSF/NIH Program for the Ecology of InfectiousDisease to G.F. McCracken and T.H. Kunz.

References

Allen, L.C. (2009). Immunity, reproductive success, and stress

in cave- and bridge-roosting Brazilian free-tailed bats (Ta-darida brasiliensis): implications for conservation. PhD

dissertation. Boston University, 151pp.

Allen, L.C., Turmelle, A.S., Mendonca, M.T., Navara, K.J.,Kunz, T.H. & McCracken, G.M. (2009). Roosting ecology

and variation in adaptive and innate immune systemfunction in the Brazilian free-tailed bat (Tadarida brasi-

liensis). J. Comp. Physiol. B Biochem. Syst. Environ. Phy-siol. 179, 315–323.

Arendt, J.D. (1997). Adaptive intrinsic growth rates: an

integration across taxa. Q. Rev. Biol. 72, 149–177.Arlettaz, R., Patthey, P., Baltic, M., Leu, T., Schaub, M.,

Palme, R. & Jenni-Eiermann, S. (2007). Spreading free-riding snow sports represent a novel serious threat for

wildlife. Proc. Roy. Soc. Lond. Ser. B: Biol. Sci. 274,1219–1224.

Barclay, M.R. (1994). Constraints on reproduction byflying vertebrates: energy and calcium. Am. Nat. 144,

1021–1031.Barja, I., Silvan, G., Rosellini, S., Pineiro, A., Gonzalez-Gil,

A., Camacho, L. & Illera, J.C. (2007). Stress physiological

responses to tourist pressure in a wild population ofEuropean pine marten. J. Ster. Biochem. Mol. Biol. 104,

136–142.Bejder, L., Samuels, A., Whitehead, H., Gales, N., Mann, J.,

Connor, R., Heithaus, M., Watson-Capps, J., Flaherty, C.& Krutzen, M. (2006). Decline in relative abundance of

bottlenose dolphins exposed to long-term disturbance.Conserv. Biol. 20, 1791–1798.

Betke, M., Hirsh, D.E., Makris, N.C., McCracken, G.F.,Procopio, M., Hristov, N.I., Tang, S., Bagchi, A., Reich-ard, J.D., Horn, J.W., Crampton, S., Cleveland, C.J. &

Kunz, T.H. (2008). Thermal imaging reveals significantlysmaller Brazilian free-tailed bat colonies than previously

estimated. J. Mammal. 89, 18–24.Case, T.J. (1978). On the evolution and adaptive significance

of post-natal growth rates in terrestrial vertebrates.Q. Rev.Biol. 53, 243–282.

Cleveland, C.J., Frank, J.D., Federico, P., Gomez, I., Hallam,T.G., Horn, J., Lopez, J., McCracken, G.F., Medellin,

R.A., Moreno-V, A., Sansone, C., Westbrook, J.K. &Kunz, T.H. (2006). Economic value of the pest controlservice provided by Brazilian free-tailed bat in south-

central Texas. Front. Ecol. Environ. 4, 238–243.Creel, S., Fox, J.E., Hardy, A., Sands, J., Garrott, B. &

Petterson, R.O. (2002). Snowmobile activity and glucocor-ticoid stress responses in wolves and elk. Conserv. Biol. 16,

809–814.Davis, R.B., Herreid, C.F. & Short, H.L. (1962). Mexican

free-tailed bats in Texas. Ecol. Monogr. 32, 311–346.Fenton, M.B. (1997). Science and the conservation of bats.

J. Mammal. 78, 1–14.Festa-Bianchet, M., Gaillard, J.M. & Jorgenson, J.T. (1998).

Mass- and density-dependent reproductive success andreproductive costs in a capital breeder. Am. Nat. 152,367–379.

Gannon, W.L. & Sikes, R.S. (2007). Guidelines of the amer-ican society of mammalogists for the use of wild mammals

in research. J. Mammal. 88, 809–823.



Gittleman, J.L. & Thompson, S.D. (1988). Energy allocationin mammalian reproduction. Am. Zool. 28, 863–875.

Gregory, T.R. & Wood, C.M. (1999). The effects of chronicplasma cortisol elevation on the feeding behaviour,

growth, competitive ability, and swimming performanceof juvenile rainbow trout. Physiol. Biochem. Zool. 72,286–295.

Harding, KC., Fujiwara, M., Axberg, Y. & Harkonen, T.(2005). Mass-dependent energetics and survival in harbour

seal pups. Funct. Ecol. 19, 129–135.Herreid, C.F. & Schmidt-Nielsen, K. (1966). Oxygen con-

sumption, temperature, and water loss in bats from differ-ent environments. Am. J. Physiol. 211, 1108–1112.

Hood, W.R., Bloss, J. & Kunz, T.H. (2002). Intrinsic andextrinsic sources of variation in size at birth and rates of

postnatal growth in the big brown bat Eptesicus fuscus(Chiroptera: Vespertilionidae). J. Zool. (Lond.) 258,355–363.

Horn, J.W. & Kunz, T.H. (2008). Analyzing NEXRADDoppler radar images to assess nightly dispersal patterns

and population trends in Brazilian free-tailed bats (Tadar-ida brasiliensis). Integr. Comp. Biol. 48, 24–39.

Hoying, K.M. & Kunz, T.H. (1998). Variation in size at birthand post-natal growth in the insectivorous bat Pipistrellus

subflavus (Chiroptera: Vespertilionidae). J. Zool. (Lond.)245, 15–27.

Keeley, B. & Tuttle, M. (1999). Bats in American bridges.Resource Pub 4. Austin, TX: Bat Conservation Interna-tional Inc.

Koehler, C.E. & Barclay, R.M.R. (2000). Post-natal growthand breeding biology of the hoary bat (Lasirius cinereus).

J. Mammal. 81, 234–244.Kunz, T.H. (1973). Population studies of the cave bat (Myotis

velifer): reproduction, growth, and development. Occas.Pap. Mus. Nat. Hist. Univ. Kansas 15, 1–43.

Kunz, T.H., Adams, R.A. &Hood,W.R. (2009). Methods forassessing size at birth and postnatal growth and develop-

ment. In Ecological and behavioral methods for the study ofbats: 273–314. Kunz, T.H. & Parsons, S. (Eds). Baltimore:Johns Hopkins University Press.

Kunz, T.H. & Robson, S.K. (1995). Postnatal growth anddevelopment in the Mexican free-tailed bat (Tadarida

brasiliensis mexicana): birth size, growth rates and ageestimation. J. Mammal. 76, 769–783.

Kunz, T.H., Whitaker, J.O. Jr & Wadanoli, M.D. (1995).Dietary energetics of the Mexican free-tailed bat (Tadarida

brasiliensis) during pregnancy and lactation.Oecologia 101,107–115.

Lausen, C.L. & Barclay, R.M.R. (2006). Benefits of living in abuilding: big brown bats (Eptesicus fuscus) in rocks versusbuildings. J. Mammal. 87, 362–370.

Liu, H., Wang, D.H. & Wang, Z.W. (2003). Energyrequirements during reproduction in female Brandt’s

voles (Microtus brandtii). J. Mammal. 84, 1410–1416.

Lordi, B., Patin, V., Protais, P., Mellier, D. & Caston, J.(2000). Chronic stress in pregnant rats: effects on growth

rate, anxiety and memory capabilities of the offspring. Int.J. Psychophysiol. 37, 195–205.

Lourenco, S.I. & Palmeirim, J.M. (2004). Influence of tem-perature in roost selection by Pipistrellus pygmaeus (Chir-optera): relevance for the design of bat boxes. Biol.

Conserv. 119, 237–243.Pagels, J.F. & Jones, C. (1974). Growth and development of

the free-tailed bat, Tadarida brasiliensis cynocephala.Southwest. Nat. 19, 267–276.

Ralls, K. (1976). Mammals in which females are larger thanmales. Q. Rev. Biol. 51, 245–276.

Reichard, J.D., Gonzalez, L.E., Casey, C.M., Allen, L.C.,Hristov, N.I. & Kunz, T.H. (2009). Evening emergence

behavior and seasonal dynamics in large colonies of Brazi-lian free-tailed bats. J. Mammal, in press.

Reiter, G. (2004). Postnatal growth and reproductive biology

of Rhinolophus hipposideros (Chiroptera: Rhinolophidae).J. Zool. (Lond.) 262, 231–241.

Schneider, M.L., Roughton, E.C., Koehler, A.J. & Lubach,G.R. (2003). Growth and development following prenatal

stress exposure in primates: an examination of ontogeneticvulnerability. Child Dev. 70, 263–274.

Setchell, J.M., Lee, P.C., Wickings, E.J. & Dixson, A.F.(2001). Growth and ontogeny of sexual size dimorphism in

the mandrill (Mandrillus sphinx). Am. J. Physiol. Anthro-pol. 115, 349–360.

Sockman, K.W. & Schwabl, H. (2000). Yolk androgens and

offspring survival. Proc. Roy. Soc. Lond. Ser. B: Biol. Sci.267, 1451–1456.

Turmelle, A.S., Allen, L.C., Jackson, F.R., Kunz, T.H.,Rupprecht, C.E. McCracken, E.F. (2009). Ecology of

rabies virus exposure in colonies of Brazilian free-tailedbats (Tadarida brasiliensis) at natural and man-made roots

in Texas. Vector Borne Zoonotic Dis, doi: 10.1089/vbz.2008.0163.

Tuttle, M.D. (1975). Population ecology of the gray bat(Myotis grisescens): factors influencing early growthand development. Occas. Pap. Mus. Natl Hist. 36,

587–595.Van Weerd, J.H. & Komen, J. (1998). The effects of chronic

stress on growth in fish: a critical appraisal. Comp. Bio-chem. Physiol. A: Mol. Integr. Physiol. 120, 107–112.

Verhulst, S., van Balen, J.H. & Tinbergen, J.M. (1995).Seasonal decline in reproductive success of the greattit: variation in time or quality? Ecology 76,2392–2403.

Walker, B.G., Boersma, P.D. & Wingfield, J.C. (2005a).Physiological and behavioral differences in Magellanicpenguin chicks in undisturbed and tourist-visited Loca-

tions of a colony. Conserv. Biol. 19, 1571–1577.Walker, B.G., Boersma, P.D. &Wingfield, J.C. (2005b). Field

endocrinology and conservation biology. Integr. Comp.Biol. 45, 12–18.



Wilkins, K.T. (1989). Tadarida brasiliensis. Mamm. Species331, 1–10.

Willis, C.K.R., Brigham, R.M. & Geiser, F. (2006). Deep,prolonged torpor by pregnant, free-ranging bats. Natur-

wissenschaften 93, 80–83.

Zeger, S.L. & Liang, K.Y. (1986). Longitudinal data analysisfor discrete and continuous outcomes. Biometrics 42,

121–130.



Contributed Paper

Variation in Physiological Stress between Bridge- andCave-Roosting Brazilian Free-Tailed BatsLOUISE C. ALLEN,∗†∗∗ AMY S. TURMELLE,‡ ERIC P. WIDMAIER,∗ NICKOLAY I. HRISTOV,∗§GARY F. MCCRACKEN,‡ AND THOMAS H. KUNZ∗

∗Center for Ecology and Conservation Biology, Boston University, Boston, MA 02215, U.S.A.†Department of Biology, Salem College, Winston-Salem, NC 27101, U.S.A.‡Department of Ecology and Evolutionary Biology, University of Tennessee, Knoxville, TN 37996, U.S.A.§Winston Salem State University, Winston-Salem, NC 27106, U.S.A.

Abstract: Since the late 1980s, Brazilian free-tailed bats (Tadarida brasiliensis) have increasingly used bridgesas roosts in the southern United States. We examined differences in blood cortisol levels, body condition, andparasite load, as measures of physiological stress in bats roosting in bridges and bats roosting in caves. Wecollected data during three periods, coinciding with female phases of reproduction. For all measures, batswere captured during the nightly emergence from the roost and immediately sampled. Cortisol levels weresignificantly higher during pregnancy and lactation and in individuals with lower body-condition scores(length of forearm to mass ratio) and significantly higher in bats roosting in caves than in those roosting inbridges. Thus, we concluded that individuals of this species that roost in bridges are not chronically stressedand seem to be unaffected by human activities present at bridges. This is a rare documented instance wherea human-dominated environment does not appear to be adversely affecting the physiological health of afree-ranging animal.

Keywords: artificial roosts, bats, conservation physiology, cortisol, disturbance, stress

Variacion en el Estres Fisiologico de Murcielagos Tadarida brasiliensis que Perchan en Puentes y Cuevas

Resumen: Desde fines de la decada de 1980, murcielagos (Tadarida brasiliensis) han incrementado el uso depuentes como perchas en el sur de los Estados Unidos. Examinamos las diferencias en los niveles de cortisol enla sangre, condicion del cuerpo y carga de parasitos, como medidas del estres fisiologico en murcielagos queperchan en puentes y murcielagos que perchan en cuevas. Recolectamos datos durante 3 perıodos, coincidentescon las fases reproductivas de hembras. Para todas las medidas, los murcielagos fueron capturados al salir desus perchas y procesados inmediatamente. Los niveles de cortisol fueron significativamente mayores duranteel embarazo y la lactancia y en individuos con valores bajos en la condicion del cuerpo (relacion longituddel antebrazo – masa) y significativamente mayores en murcielagos que perchan en cuevas que en los queperchan en puentes. Por lo tanto, concluimos que los individuos de esta especie que perchan en puentes noestan estresados cronicamente y parece que las actividades humanas en el puente no les afectan. Esta es unarara instancia en la que un ambiente dominado por humanos parece no afectar negativamente la saludfisiologica de una especie de libre movimiento.

Palabras Clave: cortisol, estres, fisiologıa de la conservacion, perchas artificiales, perturbacion

Introduction

Measures of endocrine function, in particular levels of thestress hormones cortisol and corticosterone (Wikelski &

∗∗email [email protected] submitted November 13, 2009; revised manuscript accepted August 1, 2010.

Cooke 2006; Breuner et al. 2008; Bonier et al. 2009), arenow commonly used as indicators of the physiologicalstatus of animals exposed to environmental stress (Wing-field et al. 1997; Munns 2006; Busch & Hayward 2009).

374Conservation Biology, Volume 25, No. 2, 374–381C⃝2010 Society for Conservation BiologyDOI: 10.1111/j.1523-1739.2010.01624.x

Allen et al. 375

Human activities, such as road construction, tourism,and the conversion of natural habitat have substantialeffects on behavior, stress levels, and health in a vari-ety of free-ranging animals (Creel et al. 2002; Acevedo-Whitehouse & Duffus 2009; Barber et al. 2009). Never-theless, the extent to which these activities directly orindirectly affect long-term fitness is poorly understood(Walker et al. 2005), perhaps because paired disturbedand undisturbed populations often do not exist for eval-uation (but see Wasser et al. 2004; Partecke et al. 2006;French et al. 2008). Brazilian free-tailed bats (Tadaridabrasiliensis) use a variety of roost types, including naturaland human-made structures, which allows direct compar-isons of individuals occupying different types of roosts.Natural caves were the only roosts historically availableto this species; however, over the past 30 years thisspecies and other species of bats (e.g., Myotis velifer)have increasingly occupied crevices in concrete bridges.Bat roosts are thought to differ in quality (Lausen & Bar-clay 2006) on the basis of variation in temperature orthermal stability, proximity to food sources, access towater, presence of contaminants, and noise and vibra-tion levels. Keeley and Tuttle (1999) suggest bridges pro-vide important roosts for Brazilian free-tailed bats, whoseabundances may be declining (Betke et al. 2008). To thebest of our knowledge, however, prior to our work noone has evaluated the relative health and fitness of batsroosting in bridges relative to those roosting in naturalcaves (Allen et al. 2009; Allen et al. 2010; Turmelle et al.2010).

We studied the Brazilian free-tailed bat because it is anecologically and economically important species (Cleve-land et al. 2006; Lee & McCracken 2005). Historic esti-mates of the size of maternity colonies of this speciesin south central Texas are in the millions to tens of mil-lions of individuals (Davis et al. 1962, McCracken 2003).More recent estimates of colony size at these same lo-cations are an order of magnitude smaller (Betke et al.2008). Because females give birth to one offspring eachyear (Kunz & Robson 1995), unexpected declines inabundance could affect probability of persistence of thespecies more than declines in abundance of other small-bodied mammals with higher reproductive output (Bar-clay & Harder 2003).

We investigated the association between roost type(bridges and caves) and health of individual Brazilian free-tailed bats. We used differences in baseline and acutestress-induced plasma cortisol as an indicator of habitatquality (Mostl & Palme 2002; Homan et al. 2003). Wealso compared body condition and ectoparasite loads be-tween bats roosting in bridges and caves. Bridges areassociated with light, noise, and vibrations from mo-tor vehicle and railway activities that are not typicallypresent in or near caves. These factors may negativelyaffect cortisol levels, body condition, and ectoparasiteload.

Methods

Study Species and Sites

Brazilian free-tailed bats are small (approximately 12 g)and range from Argentina northward into the UnitedStates (Wilkins 1989). The species forms large mater-nity colonies in caves and bridges (Wilkins 1989). Wesampled bats at three caves and three bridges in south-central Texas. Frio Cave and Seco Creek and East ElmCreek bridges are located in Uvalde and Medina coun-ties, in landscapes dominated by intensive crop agricul-ture. Davis Blowout and Eckert James River caves arein Blanco and Mason counties, respectively. They arelocated in areas dominated by live oak (Quercus vir-giniana) and mesquite (Prosopis sp.) savannahs that areused primarily for grazing livestock. McNeil Bridge is inWilliamson county in a suburban residential area. Colonysizes of Brazilian free-tailed bats vary seasonally (Hristovet al. 2010), but estimates of colony size at our study sitesare 1.2 million in Frio Cave, 600,000 in McNeil Bridge,500,000 in Eckert James River Cave, 400,000 in DavisCave, 250,000 in Seco Creek Bridge, and <5,000 in EastElm Creek Bridge (Betke et al. 2008; L.C.A. & T.H.K.,unpublished data).

We sampled bats bimonthly from May to October 2005.During the spring, bats migrate north from Mexico andfemales select a site in which to give birth and nurse theiryoung. We collected data during three periods: gestation(May–mid-June), lactation (mid-June–July), and after lac-tation (August–early October). Males are nonreproduc-tive during these female phases of reproduction. It islikely that individual bats move between colonies duringspring and autumn migration (Russell & McCracken 2006;Horn & Kunz 2008); thus, we did not know whether thegravid bats we sampled consistently occupied a givensite. Scale and Wilkins (2007), however, report high lev-els of roost fidelity during the spring and summer, and thisspecies does not transport its offspring in flight (Wilkins1989). Therefore, females are tied to a given roost loca-tion until their offspring become independent in early-to mid-August (Kunz & Robson 1995). Thus, we are con-fident that bats sampled during lactation and immedi-ately after lactation ceased were residents of the colonyin which they were captured. We captured bats withhand nets when they emerged from their roosts (approx-imately 18:00–19:00).

Blood Sampling

From each captured bat, we collected, via venipuncture,a small quantity of whole blood (25–40 µL) in a ster-ile, heparinized capillary tube (Kunz & Nagy 1988). Weseparated plasma from cellular blood components in aportable centrifuge and immediately froze the plasmain dry ice and then later placed it in a −20 ◦C freezer

Conservation BiologyVolume 25, No. 2, 2011

376 Stress in Cave- and Bridge-Roosting Bats

until processing. We shipped frozen plasma samples indry ice to Boston University (Boston, Massachusetts) forhormonal analyses. For all bats captured, we recordedmorphological traits (body mass, length of forearm, sex,reproductive status, and age class), time of capture, andtime required to acquire the blood sample. Estimatedbody condition was the ratio of body mass to length ofthe forearm. This index allows for comparison of bodymass among individuals when their skeletal sizes differ(Allen et al. 2009). To prevent resampling we applied asite-specific tattoo to the wing of each captured individ-ual (Allen et al. 2009). All procedures were conductedunder protocols approved by the Boston University An-imal Care and Use Committee (05-012) to T.H.K. andunder Texas Parks and Wildlife Department permit SPR-0305-058 to T.H.K.

Blood samples were collected from each bat (n = 524)within 3 min of capture; thus, we considered these sam-ples would provide baseline levels of the stress hormonecortisol (Romero & Reed 2005). Plasma cortisol levelswere not significantly associated with time between cap-ture and acquisition of blood (when time was <3 min)(p = 0.23). For the acute stress treatments, after wetook baseline blood samples, we placed a subset of bats(n = 110; two sites of each type, Frio and Davis Blowoutcaves; and East Elm Creek and McNeil bridges) in individ-ual plastic restraint tubes custom made for bats of 9–14g to restrict movement but not respiration. After 60 min,we took a blood sample from each bat within 3 min ofremoval from the restraint tube.

Ectoparasite Loads

We did not evaluate mite abundance in bats from whichwe took blood samples. We determined the relative levelsof mite abundance in captured bats (n = 261) betweenJune and August 2005 at four sites: Frio and Davis Blowoutcaves and East Elm Creek and McNeil bridges. To estimatethe relative mite load per adult bat, we photographedthe tail membrane (uropatagium) with the same focaldistance (0.25 m) and image resolution (2048 × 1536) foreach bat. We transformed the digital images to grayscalein Photoshop (Adobe Systems, San Jose, California) andcalculated percent cover of white pixels in the processedimages as parasite loads.

Hormonal Assays

Cortisol is the primary glucocorticoid stress hormone inbats (Reeder et al. 2004; Reeder & Widmaier 2009). We as-sayed its level in plasma in duplicate, without extraction,in a volume of 2.5 or 5 µL plasma with a radioimmunoas-say (RIA) kit from MP Biomedicals (Irvine, California),as previously validated for this species (E.P.W. & L.C.A.,unpublished data). Sample volumes were small and weexpected high hormone levels, so we diluted samples by

1:10 or 1:5 to a volume of 25 µL. Negligible cross reac-tivity (<5.5%) of the cortisol assay with corticosteroneis reported by the kit manufacturer, and this claim isconsistent with findings in another bat species, Myotislucifugus (Reeder et al. 2004). Intra- and interassay vari-ation in results was 2% and <8%, respectively, at cortisollevels present in samples. The least detectable level ofcortisol was 12.5 ng/mL.

Statistical Analyses

We found no difference in baseline levels of cortisol be-tween postlactating and nonreproductive adult females(n = 39 and 15, respectively; p = 0.31), between sexesin nonreproductive adults (n = 54 females, 97 males;p = 0.19), between sexes within juveniles (n = 31females, 40 males; p = 0.43), or between nonrepro-ductive adults and juveniles (n = 151, 71 respectively;p = 0.15). Thus, we combined all nonreproductive in-dividuals into one group for further analyses. Addition-ally, consistent with other mammals (Reeder & Kramer2005), reproductive status (gestation, lactation, and non-reproduction) explained significant variation in baselinecortisol (n = 133, 169, and 222, respectively; p < 0.001).Thus, we separated subsequent analyses of cortisol byreproductive class. Cortisol values were not normally dis-tributed. Thus, values were log transformed for statisticalanalyses.

We used a nested analysis of variance (ANOVA) model(general linear model; implemented in JMP 5.0.1) to com-pare cortisol levels (baseline and stress treatment) be-tween roost types (bridges and caves). Colony refersto the specific bridge or cave from which the individ-ual bats were sampled. Because we only sampled fourto six colonies, we used a nested design to examinewhether there was significant variation in roost type af-ter accounting for variation due to colony. Roost typeand colony nested within roost type were independentvariables, whereas body condition and the interaction ofbody condition and roost type were covariates in theoverall model. Because we were interested in general-izing to other colonies of Brazilian free-tailed bats, wetreated roost type as a fixed effect and colony as a ran-dom effect.

To test for differences between roost types in ec-toparasite load, we used a nested ANOVA with per-cent mite cover as a dependent variable, roost type andcolony as independent variables, and body condition asa covariate. Because sexes did not differ in ectopara-site load (n = 170 females, 91 males; p = 0.14), wecombined them in this analysis. Mite load differed sig-nificantly during lactation and postlactation, so we ana-lyzed them separately (roost type by period interaction;p < 0.001).


Allen et al. 377

Results

Baseline and Stress-Induced Cortisol

For gravid females (n = 133), colony was the only vari-able we measured that explained significant variationin baseline levels of cortisol (p < 0.001, R2 = 0.42,F = 18.35). Gestating females from Davis Blowout Cave(mean [SD] = 2795.2 ng/mL [229.7]) and McNeil Bridge(mean = 3452.9 ng/mL [235.2]) had significantly higherbaseline cortisol levels than gravid females from EckertJames River Cave (mean = 922.3 ng/mL [196.7]), FrioCave (mean = 1201.2 ng/mL [187.6]), East Elm Creek(mean = 1198.4 ng/mL [235.1]), and Seco Creek bridges(mean = 1039.1 ng/mL [407.3]) (Fig. 1).

Roost type, colony, and body condition all explainedsignificant variation in baseline cortisol in lactating fe-males (whole model p < 0.001, R2 = 0.36, F = 15.2; p <

0.01, p < 0.05, and p < 0.001, respectively; n = 169). Sig-nificant differences between roost types beyond variationdue to colony existed, and during lactation bats sampledat the three cave sites (n = 129) had higher baseline cor-tisol levels than bats sampled at the three bridge sites(n = 40) (cave, mean [SD] = 680 ng/mL [46.7]; bridge,mean = 250.9 ng/mL [80.3]) (Fig. 2). Body conditionand baseline cortisol were negatively correlated; bats inbetter body condition had lower baseline cortisol levels.Nevertheless, roost type affected the strength of the latterrelation (roost type∗body condition interaction; p = 0.06;Fig. 3a).

For nonreproductive bats (n = 222), roost type,colony, body condition, and the interaction betweenbody condition and roost type all explained significantvariation in baseline cortisol (whole model p < 0.0001,R2 = 0.24, F = 11.08; p < 0.05, p = 0.014, p < 0.001, andp < 0.001, respectively). Bats sampled at caves (n = 93)had higher baseline levels of cortisol than bats sam-

0

200

400

600

800

NonreproductiveLactation

Cor

tisol

(ng/

mL)

Figure 2. Baseline plasma cortisol levels in Brazilianfree-tailed bats sampled at bridges (black bars) andcaves (white bars) during lactation (n = 40 and 129;p < 0.01) and in nonreproductive individuals(n = 129, 93; p < 0.05).

pled at bridges (n = 129) during the nonreproductiveperiod (cave, mean [SD] = 298 ng/mL [26.4]; bridge,mean = 170.5 ng/mL [20.5]) (Fig. 2). Body condition andbaseline cortisol were negatively correlated, and bats inbetter body condition had lower baseline cortisol levels.Nevertheless, the strength of this relation varied betweenbridges and caves; this inverse correlation was strongerfor bats that roosted in caves (Fig. 3b).

Body condition was the only variable that explained sig-nificant variation in stress-induced cortisol levels (stresstreatment) in all three reproductive stages. Stress-inducedcortisol levels and body condition in gravid bats (n = 21)were significantly and positively related; heavier bats(later in gestation) had higher levels of cortisol follow-ing restraint (p < 0.001, R2 = 0.35, F = 15.5). By con-trast, the correlation between body condition and cortisol

0

500

1000

1500

2000

2500

3000

3500

4000

East ElmCreek Bridge

McNeilBridge

Seco CreekBridge

DavisCave

Frio Cave

James RiverCave

Cor

tisol

(ng

/mL)

Figure 1. Baseline plasma cortisollevels in female Brazilian free-tailed bats sampled at six sitesduring gestation (mean and SE)(p < 0.001; n = 21, 21, 7, 22, 32,and 30).



1.0

2.0

3.0

4.0

(a)

(b)

Log

corti

sol (

ng/m

L)

1.0

2.0

3.0

4.0

0.15 0.2 0.25 0.3 0.35

Body condition (mass/FL)

Log

corti

sol (

ng/m

L)

Figure 3. Correlation between bodycondition (mass/length of forearm [FL])and baseline cortisol levels in (a) lactatingfemale (n = 169; p < 0.001; interactionterm, p = 0.06) and (b) nonreproductive(n = 222; p < 0.001; interaction term,p < 0.001) Brazilian free-tailed bats (filledcircle and dashed line, samples from batsat bridge sites; open circle, solid line,samples from bats at cave sites).

following restraint was negative in lactating (n = 31) andnonreproductive bats (n = 58); bats in better body con-dition produced less cortisol in response to restraint (lac-tation, p < 0.001, R2 = 0.48, F = 17.4; nonreproductive,p = 0.01, R2 = 0.10, F = 6.57) than those with poorbody condition. Roost type, colony, and the interactionbetween roost type and body condition did not explainsignificant variation in stress-induced plasma cortisollevels (p > 0.05).

Ectoparasite Loads

Body condition did not explain significant variation inmite cover (p > 0.05); thus, we removed it from thenested ANOVA. During the sampling period that coin-cided with female lactation (n = 151), both roost type(F = 116.1) and colony (F = 29.94) explained significantvariation in mite cover (p < 0.001, both variables). Batsroosting in caves (n = 76) had higher ectoparasite loadson their tail membrane (mean [SD] = 0.49% [0.03]) thanbats roosting in bridges (n = 75; mean = 0.09% [0.03]).

During the postlactation sampling period (n = 110),only colony explained variation in mite cover (p < 0.01,F = 5.65). Mean percent mite cover did not vary betweenroost types during this later period (p > 0.89). Bats fromEast Elm Creek Bridge (n = 28) had the highest percentmite cover (mean [SD] = 0.18% [0.03]), whereas bats

from McNeil Bridge (n = 31) had the lowest mite cover(mean = 0.08% [0.02]). Across the season, mean mitecover on bats sampled at caves was lower at the end oflactation (p < 0.001).

Discussion

Overall, baseline cortisol levels in bats that roosted inbridges during lactation and postlactation periods weresignificantly lower than levels in bats that roosted in caves(Fig. 2), which suggests these bats are not more stressedthan bats that roost in caves. Roosting in human-madestructures may improve the health of Brazilian free-tailedbats (Horn & Kunz 2008). It is also possible that thisspecies is not stressed by, for example, noise and light oracclimates quickly to these factors. Therefore, we suggestthat for Brazilian free-tailed bats, bridge roosts are a novelenvironment in which human activities can be toleratedand where bats are released from stressors present incaves (e.g., parasites, extended foraging ranges, greatercompetition for resources). Results of several previousstudies show negative effects of human activities on free-ranging animals (Wasser et al. 1997; Creel et al. 2002;Romero & Wikelski 2002). Nevertheless, our results wereconsistent with those of French et al. (2008), who foundthat tree lizards (Urosaurus ornatus) living in natural


Allen et al. 379

areas had higher stress hormone levels than those in ur-ban environments.

High levels of glucocorticoids are traditionally inter-preted as an indicator of physiological stress (Romero& Wikelski 2001; Creel et al. 2002; Pride 2005). Resultsof recent studies, however, show that both baseline andstress-induced corticosterone levels were lower in exper-imentally stressed birds than in unstressed control birds(Rich & Romero 2005; Cyr & Romero 2007). Examiningmultiple measures of health in a single species can aidin the interpretation of hormone levels and the effect ofenvironmental variables on animals (Tracy et al. 2006). Inour analysis, roost type or colony was not associated withstress-induced cortisol levels, which indicated that hor-monal response to stress was not altered across roosts.Furthermore, in all reproductive classes, the body condi-tion of individuals sampled at bridges was better than thebody condition of bats in caves. Additionally, we foundsignificant negative correlations between body conditionand both baseline and stress-induced cortisol in lactatingand nonreproductive bats (Fig. 3). Thus, the traditionalinterpretation that high baseline cortisol levels signifystress may be appropriate in this species. Results fromour other studies (Allen et al. 2009, 2010) are consistentwith the conclusion that bats that roost in bridges are notchronically stressed.

There are a number of plausible explanations for thisunexpected result. Covariates that likely differ amongroosts include colony size, roosting density, parasite load,foraging distances, and energy budgets. Bats from smaller,less dense colonies with low ectoparasite loads will likelyhave lower cortisol levels. Furthermore, bats that roostin warm, thermoneutral environments (an environmentthat enables an animal to maintain optimal body tem-perature with minimal energy expenditure and oxygenrequirements) and that forage efficiently are expected toallocate more time and energy to growth, reproduction,and maintenance and thus to experience lower levels ofphysiological stress.

Colony size, in addition to roost structure and expo-sure to human disturbance, is the principal differencebetween caves and bridges. Most cave colonies, includingthose we sampled, have more bats than bridge colonies(Betke et al. 2008; Allen et al. 2009), with the exceptionof McNeil Bridge. Group size and indices of health are re-lated in several taxa (body condition and immunocompe-tence in penguins [Spheniscus magellanicus], Tella et al.2001; stress hormones in lemurs [Lemur catta], Pride2005; stress hormones in bats [Pteropus sp.], Reeder et al.2006). We expect that Brazilian free-tailed bats in largercolonies are subject to greater intraspecific competitionfor food and roost space than bats in smaller colonies andthat this pressure contributes to their relatively higherlevels of cortisol. Nevertheless, we found no direct cor-relation between cortisol and colony size (p = 0.63).

We found evidence that parasite load was influencedby roost type and reproductive condition. During thelactation period, bats that roosted in caves had signifi-cantly higher mean percent mite coverage than bats thatroosted in bridges. There were no differences in mitecoverage between roost types during the postlactationperiod. These results suggest that during reproductiveactivity Brazilian free-tailed bats allocate more resourcesto reproduction at the expense of maintenance. This isconsistent with the findings of Christe et al. (2000), whoreport that reproductively active female Myotis myotishave higher mite loads than females that are not repro-ductively active. Because of differences in roost structure,we suggest that cave colonies are likely to support greaterdensities of parasites than bridge colonies. Increased aircirculation around the roost, less guano accumulation,and less frequent contact with roostmates may accountfor the lower parasite loads on bats in bridge roosts thanon bats in caves, although parasite loads in the guano andother substrates at these sites have not been determined.

Allen et al. (2010) found that McNeil Bridge had higherroost temperatures than Davis Blowout Cave. But it is un-known whether this pattern is consistent across otherbridges and caves. We also postulate that bats that spendless effort and time foraging have lower stress levels, dueto reduced energetic expenditure. Horn and Kunz (2008)found that mean dispersal distance of bats varies amongcolonies. Bats emerging from Frio Cave (the largestcolony in both their study and ours) travel the longestdistance (18.6 km), and bats emerging from CongressAvenue Bridge (not in our study) travel the shortest dis-tance (10 km). While conducting related work (Allenet al. 2010), we observed that cave- and bridge-roostingbats spend different amounts of time foraging. On sev-eral nights during the lactation period, we noted thatbats traveling to and from one of our cave sites (DavisBlowout) were absent from the roost up to 2 h longerthan bats traveling to and from a bridge site (McNeil)(L.C.A., unpublished data). If bats roosting in bridgesspend less time thermoregulating or foraging than theircave-dwelling counterparts, then this may explain someof the differences in body condition and stress hormonelevels between the roosts.

Although the populations of Brazilian free-tailed batswe studied appear to be thriving in anthropogenically al-tered environments, we did not evaluate specific costs as-sociated with roosting in bridges (e.g., exposure to pollu-tants, vulnerability to vandalism, persistence of the rooststructure, flooding) that may affect long-term sustainabil-ity of colonies. Wasser et al. (2004) found low baselinelevels of fecal glucocorticoid in bears (Ursus sp.) in ar-eas with high levels of human activity, yet bears in theseareas were more vulnerable to human-caused mortality,such as collisions with motor vehicles and poaching. It isplausible that Brazilian free-tailed bats that roost in bridge



crevices are more vulnerable to human-associated mor-tality than individuals that roost in caves.

Human-made structures that can be used as roosts mayallow the Brazilian free-tailed bat to expand its range asclimate changes, whereas the availability of both insectprey and suitable caves may have historically restrictedthe geographic distribution of this species. Agriculturallandscapes can support foraging of this species (Lee &McCracken 2005; Horn & Kunz 2008), whereas highlyurbanized areas are likely to adversely affect foraging suc-cess even if roosting needs are met.

Acknowledgments

We thank J. Reichard, S. Duncan, K. Farrell, C. Casey, L.Gonzolez, and L. Borrelli for assistance with fieldwork.We thank I. Marbach, D. Davis, and B. Cofer for access totheir property. We especially thank M. Bloschock and B.French for field support and D. and the late M. Bambergerfor their hospitality. We are grateful to C. Richardson, R.Medellın, and three anonymous reviewers for help withimproving this manuscript. Funding was provided by theNational Science Foundation (EID 0426082 and ITR-EIA0326483to G.F.M. and T.H.K.) and the American Societyof Mammalogists GIAR (to L.C.A.).

Literature Cited

Acevedo-Whitehouse, K., and A. L. J. Duffus. 2009. Effects of environ-mental change on wildlife health. Philosophical Translations of theRoyal Society B 364:3429–3438.

Allen, L. C., A. S. Turmelle, M. T. Mendonca, K. J. Navara, T. H. Kunz,and G. M. McCracken. 2009. Roosting ecology and variation in adap-tive and innate immune system function in the Brazilian free-tailedbat (Tadarida brasiliensis). Journal of Comparative Physiology B179:315–323.

Allen, L. C., C. R. Richardson, G. M. McCracken, and T. H. Kunz. 2010.Birth size and postnatal growth in cave- and bridge-roosting Brazilianfree-tailed bats. Journal of Zoology 280:8–16.

Barber, J. R., K. R. Crooks, and K. M. Fristrup. 2009. The costs ofchronic noise exposure for terrestrial organisms. Trends in Ecology& Evolution 25:180–189.

Barclay, R. M. R., and L. D. Harder. 2003. Life histories of bats: life inthe slow lane. Pages 209–253 in T. H. Kunz, editor. Bat ecology.University of Chicago Press, Chicago.

Betke, M., et al. 2008. Thermal imaging reveals significantly smallerBrazilian free-tailed bat colonies than previously estimated. Journalof Mammalogy 89:18–24.

Bonier, F., P. R. Martin, I. T. Moore, and J. C. Wingfield. 2009. Do base-line glucocorticoids predict fitness? Trends in Ecology & Evolution24:634–642.

Breuner, C. W., S. H. Patterson, and T. P. Hahn. 2008. In search of re-lationships between the acute adrenocortical response and fitness.General and Comparative Endocrinology 157:288–295.

Busch, D. S., and L. S. Hayward. 2009. Stress in a conservationcontext: a discussion of glucocorticoid actions and how levelschange with conservation-relevant variables. Biological Conserva-tion. 142:2844–2853.

Christe, P., R. Arlettaz, and P. Vogel. 2000. Variation in intensity of a par-asitic mite (Spinturnix myoti) in relation to the reproductive cycleand immunocompetence of its bat host (Myotis myotis). EcologyLetters 3:207–212.

Cleveland, C. J., et al. 2006. Economic value of the pest control serviceprovided by Brazilian free-tailed bat in south-central Texas. Frontiersin Ecology and the Environment 4:238–243.

Creel, S., J. E. Fox, A. Hardy, J. Sands, B. Garrott, and R. O. Petterson.2002. Snowmobile activity and glucocorticoid stress responses inwolves and elk. Conservation Biology 16:809–814.

Cyr, N. E., and L. M. Romero. 2007. Chronic stress in free-living Euro-pean starlings reduces corticosterone concentrations and reproduc-tive success. General and Comparative Endocrinology 151:82–89.

Davis, R. B., C. F. Herreid, and H. L. Short. 1962. Mexican free-tailedbats in Texas. Ecological Monographs 32:311–346.

French, S. S., H. B. Fokidis, and M. C. Moore. 2008. Variation instress and innate immunity in the tree lizard (Urosaurus ornatus)across an urban–rural gradient. Journal of Comparative PhysiologyB 178:997–1005.

Homan, R. N., J. V. Regosin, D. M. Rodrigues, J. M. Reed, B. S. Wind-miller, and L. M. Romero. 2003. Impacts of varying habitat qualityon the physiological stress of spotted salamanders (Ambystomamaculatum). Animal Conservation 6:11–18.

Horn, J. W., and T. H. Kunz. 2008. Analyzing NEXRAD doppler radarimages to assess nightly dispersal patterns and population trendsin Brazilian free-tailed bats (Tadarida brasiliensis). Integrative andComparative Biology 48:24–39.

Hristov, N. I., M. Betke, D. Hirsh, A. Bagchi, and T. H. Kunz. 2010. Sea-sonal variation in colony size of Brazilian free-tailed bats at CarlsbadCavern using thermal imaging. Journal of Mammalogy 91:183–192.

Keeley, B., and M. Tuttle. 1999. Bats in American bridges. Resourcepublication 4. Bat Conservation International, Austin, Texas.

Kunz, T. H., and K. A. Nagy. 1988. Energy budget analysis. Pages283–328 in T. H. Kunz, editor. Ecological and Behavioral Methodsfor the Study of Bats. Smithsonian Institution Press, Washington,D.C.

Kunz, T. H., and S. K. Robson. 1995. Postnatal growth and developmentin the Mexican free-tailed bat (Tadarida brasiliensis mexicana):birth size, growth rates and age estimation. Journal of Mammalogy76:769–783.

Lausen, C. L., and R. M. R. Barclay. 2006. Benefits of living in a building:big brown bats (Eptesicus fuscus) in rocks versus buildings. Journalof Mammalogy 87:362–370.

Lee, Y. F., and G. F. McCracken. 2005. Dietary variation of Brazilian free-tailed bats links to migratory populations of pest insects. Journal ofMammalogy 86:67–76.

McCracken, G. F. 2003. Estimates of population sizes in summercolonies of Brazilian free-tailed bats (Tadarida brasiliensis). Pages21–30 in T. J. O’Shea and M. A. Bogan, editors. Monitoring trends inbat populations of the United States and territories: problems andprospects. Information and technology report ITR-2003-003. Biolog-ical Resources Discipline, Geological Survey, Washington, D.C.

Mostl, E., and R. Palme. 2002. Hormones as indicators of stress. Domes-tic Animal Endocrinology 23:67–74.

Munns, W. R., Jr. 2006. Assessing risks to wildlife populations frommultiple stressors: overview of the problem and research needs.Ecology and Society 11:23–35.

Partecke, J., I. Schwabl, and E. Gwinner. 2006. Stress and the city:urbanization and its effects on the stress physiology in EuropeanBlackbirds. Ecology 87:145–152.

Pride, E. R. 2005. Optimal group size and seasonal stress in ring-tailedlemurs (Lemur catta). Behavioral Ecology 16:550–560.

Reeder, D. M., N. S. Kosteczko, T. H. Kunz, and E. P. Widmaier. 2004.Changes in baseline and stress-induced glucocorticoid levels duringthe active period in free-ranging male and female little brown my-otis, Myotis lucifugus (Chiroptera: Vespertilionidae). General andComparative Endocrinology 136:260–269.

Reeder, D. M., N. S. Kosteczko, T. H. Kunz, and E. P. Widmaier. 2006.The hormonal and behavioral response to group formation, seasonalchanges, and restraint stress in the highly social Malayan flying fox(Pteropus vampyrus) and the less social little golden-mantled flying


Allen et al. 381

fox (Pteropus pumilus) (Chiroptera: Pteropodidae). Hormones andBehavior 49:484–500.

Reeder, D. M., and K. M. Kramer. 2005. Stress in free-ranging mam-mals: integrating physiology, ecology, and natural history. Journalof Mammalogy 86:225–235.

Reeder, D. M., and E. P. Widmaier. 2009. Hormone analysis in bats.Pages 554–563 in T. H. Kunz and S. Parsons, editors. Ecological andbehavioral methods for the study of bats. Johns Hopkins UniversityPress, Baltimore, Maryland.

Rich, E. L., and L. M. Romero. 2005. Exposure to chronic stress down-regulates corticosterone responses to acute stressors. The AmericanJournal of Physiology – Regulatory, Integrative and ComparativePhysiology 288:1628–1636.

Romero, L. M., and J. M. Reed. 2005. Collecting baseline corticosteronesamples in the field: is under 3 min good enough? ComparativeBiochemistry and Physiology A 140:73–79.

Romero L. M., and M. Wikelski. 2001. Corticosterone levels pre-dict survival probabilities of Galapagos marine iguanas during ElNino events. Proceedings of the National Academy of Science98:7366–7370.

Romero, L. M., and M. Wikelski. 2002. Exposure to tourism reducesstress-induced corticosterone levels in Galapagos marine iguanas.Biological Conservation 108:371–374.

Russell, A. L., and G. F. McCracken. 2006. Population genetic structureof very large populations: the Brazilian free-tailed bats Tadaridabrasiliensis. Pages 227- 247 in Z. Akbar, G. F. McCracken, and T. H.Kunz, editors. Functional and evolutionary ecology of bats. OxfordUniversity Press, New York.

Scale, J. A., and K. T. Wilkins. 2007. Seasonality and fidelity in roostuse of the Mexican free-tailed bat, Tadarida brasiliensis, in an urbansetting. Western North American Naturalist 67:402–408.

Tella, J. L., M. G. Forero, M. Bertellotti, J. A. Donazar, G. Blanco, andO. Ceballos. 2001. Offspring body condition and immunocompe-

tence are negatively affected by high breeding densities in a colonialseabird: a multiscale approach. Proceeding of the Royal Society ofLondon B 268:1455–1461.

Tracy, C. R., K. E. Nussear, T. C. Esque, K. Dean-Bradley, C. R. Tracy,L. A. DeFalco, K. T. Castle, L. C. Zimmerman, R. E. Espinoza, and A.M. Barber. 2006. The importance of physiological ecology in con-servation biology. Integrative and Comparative Biology. 46:1191–1205.

Turmelle A. S., L. C. Allen, F. R. Jackson, T. H. Kunz, C. E. Rupprecht,and G. F. McCracken. 2010. Ecology of rabies virus exposure incolonies of Brazilian free-tailed bats (Tadarida brasiliensis) at nat-ural and man-made roosts in Texas. Vector-Borne and Zoonotic Dis-eases 10:165–175.

Walker, B. G., P. D. Boersma, and J. C. Wingfield. 2005. Field endocrinol-ogy and conservation biology. Integrative and Comparative Biology45:12–18.

Wasser, S. K., K. Bevis, G. King, and E. Hanson. 1997. Noninvasivephysiological measures of disturbance in the northern spotted owl.Conservation Biology 11:1019–1022.

Wasser, S. K., B. Davenport, E. R. Ramage, K. E. Hunt, M. Parker, C.Clarke, and G. Stenhouse. 2004. Scat detection dogs in wildliferesearch and management: application to grizzly and black bearsin the Yellowhead Ecosystem, Alberta, Canada. Canadian Journal ofZoology 82:475–492.

Wikelski, M., and S. J. Cooke. 2006. Conservation physiology. Trendsin Ecology & Evolution 21:38–46.

Wilkins, K. T. 1989. Tadarida brasiliensis. Mammal Species 331:1–10.

Wingfield, J. C., K. Hunt, C. Breuner, K. Dunlap, G. S. Fowler, L. Freed,and J. Lepson. 1997. Environmental stress, field endocrinology, andconservation biology. Pages 95–131 in J. R. Clemmons and R. Buch-holz, editors. Behavioral approaches to conservation in the wild.Cambridge University Press, Cambridge, United Kingdom.


Documents

Roosting ecology and variation in adaptive and innate