The Mating System of Tadarida brasiliensis (Chiroptera: Molossidae) in a Large Highway Bridge Colony

Abstract

We observed mating by Brazilian free-tailed bats (Tadarida brasiliensis) in central Texas between 21 March and 5 April 1998. We documented copulations in large and small day roosts and in temporary night roosts. Focal animal sampling at a highway bridge revealed an aggressive and a passive male copulation strategy that may function as adaptations to different roost conditions. During aggressive copulation, the male separates a female from a roost cluster and restricts her movements during mating while he emits characteristic calls. During passive copulation, the male moves very slowly onto a female that roosts in a dense cluster. Passive copulations occur without resistance from the female and without male vocalizations. Both males and females mate with multiple partners, suggesting that mating is promiscuous. The mating system in a large highway bridge colony is characterized as mating aggregations or swarming because mating occurs in large, temporally unstable multimale and multifemale mating groups, with no apparent male territories or defense of females.

Resumen

Entre el 21 de marzo y el 5 de abril de 1998 observamos la conducta reproductiva de Tadarida brasiliensis en Texas central. Documentamos cópulas en refugios diurnos grandes y pequenos, y en refugios nocturnos. La observation de individuos reveló la existencia de dos estrategias de copula, unaagresiva y otra pasiva, que pueden ser adaptaciones a diferentes condiciones de refugio. En el apareamiento agresivo el macho separa uma hembra del grupo y limita sus movimientos durante la cópula, mientras emite llamados característicos. Las cópulas passivas ocurren sin resistencia por parte de la hembra y sin vocalizaciones del macho. Ambos machos y hembras copulan com parejas múltiples lo cual sugiere que el apareamiento es promiscuo. Caracterizamos el sistema reproductivo en una colonia numerosa de un puente de autopista como de agregación o enjambre ya que la cópula ocurre en grandes grupos de apareamiento multimacho y multihembra, los cuales son temporalmente inestables, y en los que al parecer los machos no defienden hembras ni territorios.

copulation strategies, copulatory plugs, mating system, Tadarida brasiliensis

Bats may have a greater array of mating systems than mammals in any other order (Altringham 1996; McCracken and Wilkinson 2000). Classification of these behaviors is often difficult because they occur along a continuum (Clutton-Brock 1989). Emlen and Oring (1977) categorized mating systems on the basis of the potential for polygamy, degree of parental care required, distribution of resources, temporal distribution of mates, and operational sex ratio. Clutton-Brock (1989) presented a model based on 3 variables that influence variation in mammalian mating systems: whether the male assists in rearing young, female ranging behavior, and size and stability of female groups. Three major mating systems are recognized: monogamy, resource-defense polygyny, and temporary mating aggregations (McCracken and Wilkinson 2000).

The Brazilian free-tailed bat (Tadarida brasiliensis) is widely distributed throughout North, Central, and South America (Wilkins 1989). This species has been characterized as a temperate cycle bat (Bradbury 1977) based on the division of its annual cycle into 3 phases: gender segregation during parturition and lactation, mixed associations of both sexes in winter, and formation of mating groups (Constantine 1967; Davis et al. 1962; Wilkins 1989). Even though the species is common and forms colonies that can number l the millions, little is known about the mating system. Sherman (1937) studied T. b. cynocephala in Florida and noted that adult bats left the roost for a week in March during which copulations occurred. Davis et al. (1962) found copulatory plugs in a few females in a cave in Texas in March and suggested that the mating period was brief and highly synchronized. Several authors have speculated that mating of North American summer populations occurs in the Mexican winter range (Constantine 1967; Davis et al. 1962; Eads et al. 1957).

The lack of breeding information may be the result of the difficulty of studying individual bat behavior in a highly gregarious species within roosts that are dark and relatively inhospitable and inaccessible to humans. However, a highway bridge close to Austin, Texas, proved ideal for studying the mating behavior of T. brasiliensis. This study describes the mating system of the Brazilian free-tailed bat, including male mating behaviors, the duration of the mating period, and the types of roosts selected for mating.

Materials and Methods

The main study site was an overpass bridge in Williamson County, Texas (30°31′N, 97°42′W), where bats occupy parallel crevices (1.4-3.1 cm wide × 66 cm deep) between concrete beams. The bridge houses tens of thousands of Brazilian free-tailed bats in winter, serves as a temporary day roost during spring and fall migration, and, as estimated from our observations, serves as a nursery roost for approximately 750,000 bats. During spring migration from March to May, numbers of bats present can fluctuate greatly. We used an incandescent flashlight headlamp (Petzl, Springfield, Utah) for behavioral observations. Disturbance caused by the headlamp was minimal and did not appear to affect bat behavior drastically because the bats were accustomed to ambient light throughout the day. During the mating period, we distinguished males in the crevices from females based on the obvious, often wet, gular gland, a secretory gland on the chest, and characteristic behaviors (described below). We marked individual males by painting ears and wrists with distinctive combinations of different colors of nontoxic tempera paint (Derivan, Toronto, Ontario, Canada). We applied the color with a paint brush attached to a 60-cm-long wooden dowel (1.9-cm diameter), which allowed marking without capturing the bats. The markings lasted 1–3 days, but we could repaint bats easily.

We conducted behavioral observation sessions for an average of 2 h in early morning (0600-0900 h), late morning (0900-1200 h), early afternoon (1200-1500 h), and late afternoon (1500-1800 h) every 2 days from 23 to 26 March and 30 March to 4 April 1998. All mating activities that we analyzed statistically were observed between 0700 and 0830 h. On 27 March, we observed the bats from 0100 to 0600 h and from 2100 to 2400 h. Total observation time was 53 h. To document individual male movements and spatial distribution patterns, we recorded locations of males within a section of a crevice marked by reference points. For statistical analysis of the spatial distribution patterns we recorded the distance of 1 male to the next male on his right side, and pooled data from 1 date.

During observation periods, we recorded behaviors associated with mating simultaneously and continuously for 6–18 individually marked focal males. Specific behaviors observed included rubbing of the gular gland and penis on the wall, searching for females, aggressive female acquisition, aggressive copulation, passive female acquisition, passive copulation, and male-male antagonistic behaviors.

To document changes in sex ratio and the development of body mass and sizes of testes and gular glands, we captured 1,138 T. brasiliensis at the highway bridge in Williamson County. Out of these, we weighed 617 males and measured their testes and gular glands before release. We made these collections at least twice weekly between 1100 and 1500 h between 28 February and 17 May 1998 and daily during the observed mating period. We captured bats by gently pushing them out of crevices into a cloth bag by using 2 padded wooden dowels. In addition to the highway bridge, we checked 3 caves (1 each in Mason, Comal, and Gillespie counties, Texas), a bat house in Gillespie County, and 2 bridges (1 each in Gillespie and Hays counties, Texas) for mating activity.

We determined body mass to the nearest 0.25 g with a Pesola spring balance (model 20100, Baar, Switzerland), and measured the width and length of the gular gland and testes to the nearest 0.1 mm with dial calipers. We measured body mass in the afternoon, except for once when bats were captured soon after returning from an evening emergence.

We tested divergence of the sex ratio from a 1:1 ratio by a chi-square test (Zar 1998). We assessed changes in body mass, testis size, and gular gland size over time by regression analyses. We compared body mass of 2 groups of males (males before and after the evening emergence) by a 2-sample Student's Mest. We examined relationships between body mass, testis size, and gular gland size by Pearson's correlation. We used a Kendall rank correlation analysis (Zar 1998) to assess correlations between behaviors of individual males (measured in events per hour). The Kendall rank correlation coefficient x is a measure of intensity of association between, in this case, 2 behaviors. To obtain data for this analysis, we observed marked males and noted behavioral events during an observation period. We summed up the number of specified behaviors (successful copulation, aggression towards other males, aggressive female acquisition attempts, and rubbing of gular gland and penis on the wall) that occurred in an observation period on a per-male basis and then standardized them in events per hour.

We conducted a goodness-of-fit test of the Poisson distribution to determine whether males were distributed randomly, uniformly, or clumped in the crevices (Zar 1998). We used the computer program SPSS (SPSS Inc., Chicago, Illinois) to perform the statistical tests.

Results

Mating period, sex ratio, and male morphology.—We observed copulations at the highway bridge from 21 March to 5 April 1998. By listening for the distinctive mating calls (French and Lollar 1998; B. French, in litt.) and by visual observations we also documented copulations on 28 and 29 March in 6 other central Texas Brazilian free-tailed bat colonies (3 caves, a bat house, a night roost under a 2nd bridge, and an expansion joint of a 3rd bridge).

In the 2 weeks before the mating period, the sex ratio at the bridge was significantly skewed towards males (Fig. 1; males : females: 7.76, n = 149; χ² = 88.76, d.f. = 1, P < 0.001). However, during the mating period, more females than males were present at the bridge (males : females: 0.32, n = 388; χ² = 105.16, d.f. = 1, P < 0.001). In the 2 weeks after the mating period, the sex ratio at the bridge was approximately 1:1 (males : females: 0.92, n = 232; χ² = 76.52, d.f. = 1, P < 0.001).

Fig. 1

Changes in sex ratio (males : females) during the mating period in a colony of Brazilian free-tailed bats in Williamson County, Texas. Horizontal line is 1:1 sex ratio; numbers above bars indicate sample siz

Testes in adult males were scrotal and visible from the beginning of the study until 4 weeks after the end of the mating period. Testes decreased significantly in size throughout the mating period (R² = 0.489, F = 704.85, d.f = 1,P< 0.0001; Fig. 2); however, body mass of males remained almost constant (R² = 0.000, F = 0.222, d.f. = 1, P < 0.638; Fig. 2). On 26 March, 13 males captured immediately before the nightly emergence at 1815 h had a significantly lower body mass than 8 males captured about 2.5 h after the emergence at 2300 h on 27 March (X̄₁ = 10.94 g; X̄₂ = 12.72 g; t = 5.238, d.f. = 19, P < 0.001).

Fig. 2

Mean body mass and testis size for male Brazilian free-tailed bats collected at a highway bridge in Williamson County, Texas. Symbols represent mean ± SD.

Gular gland dimensions did not change over the mating period (R² = 0.002, F = 1.41, d.f. = I, P < 0.235); however, in most males the gland changed in morphology and secretory function. Just before and during the mating season, the gular glands were donut-shaped and hairless and secreted clear oily, musky smelling fluid. After the mating season, the glands flattened, were dry, and hair had started to regrow.

Two days before mating was 1st observed, a positive correlation was found between mean testis size and body mass (r = 0.443, d.f. = 33, P = 0.01) and a negative correlation was found between gular gland size and body mass (r = −0.543, d.f. = 33, P = 0.001); sizes of gular gland and testis were not correlated (r = −0.154, d.f. = 33, P = 0.393). After the mating period, no correlation was observed among mean testis size, body mass, and gular gland size (r = −0.060, d.f. = 25, P = 0.778; r = 0.221, d.f = 21, P = 0.336; r = −0.108, d.f. = 21, P — 0.640, respectively).

Copulatory behavior.—We observed 2 different strategies (aggressive and passive) used by male T. brasiliensis to gain copulatory access to females. In the aggressive copulation strategy, the male slowly approached a nearby female, aggressively grabbed her by an ear, the jaw, or the neck, and pulled her out of a cluster. If he was successful, he quickly moved onto her back and held her by biting the scruff of her neck. The female usually resisted strongly, and the male emitted faint audible chirps (mating calls) that were easily distinguishable from normal colony vocalizations (French and Lollar 1998; B. French, in litt.). The female often squeaked as well. We considered copulations that were observed to be successful if they were uninterrupted and we could observe characteristic rhythmic lower back movements in the male. After copulation, females as well as males sometimes bled from wounds on the face. On 11 occasions, 2 different males copulated with the same female within minutes of each other. In 3 instances, the 2nd male removed a copulatory plug before mating, apparently by using his inserted penis. During the mating period, we found thousands of these plugs, which look like opaque grains of rice, 2–3 mm long, beneath roosts in the guano.

We observed the passive copulation strategy only when females roosted in dense clusters and only during the last 4 days of the mating period. In this behavior, the male moved slowly onto the back of a female with his ears held low and eyes closed. Copulations occurred without neck-biting or the emission of audible calls. The female did not resist; females sometimes groomed themselves or appeared to sleep. During 16 h of continuous observation, we observed 3 marked males that used both strategies in mating, 5 males that used only the passive copulation strategy, and 7 males that used only the aggressive copulation strategy. Eight marked males did not copulate with any females.

Other behaviors observed in males only during the mating period involved rubbing the gular gland and penis on the wall of the roost surface. We observed male-male antagonistic interactions, which included 2 males only a few centimeters apart facing each other with their mouths wide open, and, even less frequently, 1 male briefly biting another male's neck, or 2 males locking their jaws and pulling each other in opposite directions. Males that we observed mating during the day were later quietly resting with eyes closed.

The frequency of mating activity declined rapidly from 24 March to 3 April. On average, 62% of marked males engaged in mating activity on 6 days from 24 March through 3 April. The percentage of males observed in copulation during our observations decreased from 89% (24 March) to 0% (1 April). In contrast, the percentage of males involved in unsuccessful mating activity increased from 11% (24 March) to 56% (1 April). Analyzed on a per-male basis, the number of successful copulations was weakly correlated with aggression towards other males, aggressive female acquisition attempts, and rubbing the gular gland and penis on the wall (τ = 0.262, n = 300, P < 0.001; τ = 0.255, n = 300, P < 0.001; τ = 0.238, n = 300, P < 0.001, respectively).

As number of males per meter of crevice length increased, mean distance between adjacent males decreased over the mating period (Table 1). Early in the mating period (23–26 March), distribution of males along the length of crevices was uniform; however, later distribution of males was clumped (30 March-1 April; Table 1). Of 56 males that were marked from 23 March to 1 April, 18 were seen on at least 2 of the 5 days when observations were made (X̄ = 2.72 days ± 1.02 SE). Average distance between subsequent roost site locations of the same male was 0.38 m (n = 34, SD = 0.42 m, maximum = 1.87 m, minimum = 0.00 m). Throughout the mating period, marked males did not remain in the same location or an extended time but moved within and between crevices, and possibly between roosts. Movements occurred not only from day to day, but males were seen at new locations on the same day. On 26 March, we noted locations for 16 males at 1200, 1400, and 1600 h (Fig. 3). Ten marked males were present at all 3 observations, 4 males were seen at 2 observations, and 2 males were seen once. Individual males roosted on average 0.21 m apart (n = 24, SD = 0.17 m, range 0–0.48 m).

Fig. 3

Movements of male Brazilian free-tailed bats within 1 crevice at 3 times in 1 day. Closed circles represent locations of males; dark gray bars represent locations of clusters of females. Locations of individual males are connected by lines between time periods. Observations were at a highway bridge in Williamson County, Texas, on 26 March 1998. Distances (m) along crevice are indicated.

Table 1

—Distribution of male Tadarida brasiliensis in bridge crevices during the mating period. The distribution was tested by using a goodness-of-fit chi-square test of the Poisson distribution (if sigma²/ mean is smaller than 1, the distribution is uniform; if sigma²/mean is larger than 1, the distribution is clumped; sigma² is the variance). A Student's t-test verified whether sigma²/mean departed significantly from 1.

	Distance to male on right (cm)
Date	n	Mean	SE	χ2	sigma2/mean	t-test	Distribution
23 March	24	17.7	3.91	7.81***	0.23	−2.61***	Uniform
24 March	15	14.0	1.90	9.49*	0.18	−2.18**	Uniform
26 March	26	15.5	2.35	11.07**	0.48	−1.83*	Uniform
30 March	49	8.3	0.96	1.00***	6.61	11 9***	Clumped
1 April	31	6.7	1.60	5 99***	17.48	28.55***	Clumped

*P < 0.05, **P < 0.01, ***P < 0.001.

Discussion

The bridge colony had several advantages for behavioral observations over colonies located in caves. Observations of undisturbed behaviors are only possible in caves with a low ceiling and, because caves are dark day and night, require a night-vision scope or an infrared video camera to avoid disturbing bats (McCracken and Gustin 1991). Further, we have found that in a cave, capturing, marking, and observing a specific bat within densely packed clusters (as many as 3,387 bats/m²—McCracken and Gustin 1991) when using an infrared-sensitive device is difficult. In contrast, observations of bats with incandescent headlamps at the highway bridge, where the bats were accustomed to higher levels of ambient light throughout the day, caused minimal disturbance. In the bridge crevices, marking and observing several males at the same time was facilitated by the linear distribution of bats and by reduced density of roosting bats.

Our observations of mating activity at the highway bridge and 6 other roosts suggest that mating of the Brazilian free-tailed bat in Texas occurs during a brief but concentrated period from mid-March to early April. The timing of this period and changes in observed sex ratios before, during, and after the mating period support earlier suggestions that mating occurs during the spring migration, which also happens in March and April (Constantine 1967; Eads et al. 1957; Glass 1982; Villa-R. and Cockrum 1962). Earlier attempts to observe mating behavior in 6 Brazilian free-tailed bat caves in Nuevo Leon, Tamaulipas, and Coahuila in northeastern Mexico between 10 and 23 February 1998 revealed no mating activity or evidence of copulatory plugs, which confirms the observations of McCracken et al. (1994). In contrast to other species of mammals in which males lose weight during a brief, concentrated period of mating (e.g., Pipistrellus pipistrellus—Gerell and Lundberg 1985; Nycticeius humer alls—Bain and Humphrey 1986; Didelphis virginiana—Ryser 1992; Vulpes vulpes—Cavallini 1998), weight loss was not observed in male T. brasiliensis. Several explanations for this are possible. First, it appears that resting males sleep or possibly even enter torpor between bouts of daily activity, allowing them to conserve energy. Second, mating activity in the highway bridge colony does not involve energetically expensive behaviors such as searching for females over great distances, or advertising and courting behaviors. Third, in contrast to other species mentioned above, no evidence was found that male T. brasiliensis aggressively defended territories against other males. Evidence for the lack of territorial behavior included spatial overlap of movements by sexually active males, frequent observations of males roosting in body contact with other males, and the rare occurrence of intense or protracted male-male antagonistic behaviors. As females emerge from the roost in the evening to feed and return after a few hours, the absence of male territoriality may allow males to leave the roost to forage as well.

In contrast to the bridge colony, observations by French and Lollar (1998) from 2 small captive colonies in Texas containing 20–30 Brazilian free-tailed bats revealed that sexually active males occupied and aggressively defended roost space against other males. Rubbing of the gular gland and the penis appear to be associated with marking and defending the territory. Females roosted in a male's territory for several days before moving into another male's territory (French and Lollar 1998). The mating system in these captive colonies appears to conform to resource-defense polygyny with mating territories (Bradbury and Vehrencamp 1977; Emlen and Oring 1977), which could be an artifact of captivity. Another explanation is that differences between the mating systems of the highway bridge colony and captive colonies are a result of differences in densities at the 2 sites (Emlen and Oring 1977). Perhaps it is energetically feasible for males to defend roost sites against other males in small, low-density captive colonies with high availability of food resources, whereas it may not be possible in large, high-density wild colonies where defense of roosting sites or females by males would be impractical and costly.

Evidence that intraspecific variation can occur in mating systems also comes from banana bats (Pipistrellus nanus). Males in populations roosting in palm-thatched roofs show territoriality and aggressive competition for spacious roosts (O'Shea 1980), whereas males in populations roosting in banana leaves appear not to defend territories aggressively (Happold and Happold 1996). This difference might be based on the availability of preferred roosts. The number of spacious, semipermanent roosts in thatched roofs is limited, and males that successfully defend a territory may demonstrate greater fitness. In contrast, the number of suitable roosts in banana plants may not be a limiting factor, rendering choice of roost irrelevant to male mating success (O'Shea 1980).

In the highway bridge colony, male movements seemingly were related to movements of female clusters. Occasional male-male antagonistic behaviors and rubbing of the penis and gular gland on the wall may serve the function of dispersing males among available females because these behaviors are clearly associated with mating activity (Davis et al. 1962; French and Lollar 1998; Gutierrez and Aoki 1973). Other possible functions of the gular gland include advertising male fitness and providing females with a basis for choice among males, as suggested for Molossus ater (Rasweiler 1987). Similar to M. ater and M. sinaloae, gular glands in males were functional only during the mating season, likely reflecting changes in hormone levels (Skully et al. 2000). Further studies are needed to clarify the exact functions of the gular gland, penis rubbing, and associated behaviors.

Two alternative explanations exist for the function of copulatory plugs: they may serve to prevent sperm leakage from the female's reproductive tract, or they may function as barriers to further insemination (Fenton 1984). Vaginal plugs can be male- or female-generated (Fenton 1984). In Rhinolo-phus ferrumequinum, the plugs are formed from male secretions, possibly serving to prevent further mating and thus precluding sperm competition during the time between the autumn mating period and fertilization in spring. However, females appear to be able to eject their plugs, thus allowing further mating (Rossiter et al. 2000). In T. brasiliensis, fertilization occurs immediately after copulation (Hill and Smith 1992; Sherman 1937), and if copulatory plugs (whether of male or female origin) function as barriers to further insemination, they may not need to persist long to be effective in preventing other males from fertilizing a female. However, we observed multiple copulations by females within a short period of time on several occasions, with males evidently being able to remove the copulatory plug of the previous copulation. This would open the door for sperm competition in T. brasiliensis, a topic that requires further study.

Alternative copulation tactics within the behavioral repertoire of individuals are quite frequent in animals (Hogg and Forbes 1997). However, individual males do not typically apply both strategies simultaneously. For example, in Atlantic salmon (Salmo salar) and sockeye salmon (Oncorhynchus nerka), mating territories are held by large anadromous males, whereas tiny precocious parr or jack salmon gain fertilizations by sneaking in to ejaculate while anadromous males and females spawn (Foote et al. 1997; Gage et al. 1995). In the multimale, multifemale social groups of Japanese macaques (Macaca fuscatafuscata), subordinate males copulate with females while hidden from view, whereas dominant males copulate openly. With change in social status over time, the mating tactic of a male macaque can thus change (Soltis et al. 1997). Rocky Mountain bighorn (Ovis canadensis canadensis) rams, in contrast, have 3 copulation strategies, all of which may be used by the same male during a mating season (Hogg and Forbes 1997).

Alternative copulation strategies may allow males of differing capabilities to copulate successfully, thus increasing the genetic variability in a population (Cavallini 1998; Soltis et al. 1997; Wilkinson 1985). Alternative copulation strategies also may be adaptive adjustments to differences in the environment. In the nonmigrating precocious parr salmon, the alternative mating strategy eliminates time costs and risks of mortality incurred by anadromous males during migration before reproduction (Gage et al. 1995). In birds, extrapair copulations of generally monogamous species are elicited by females, possibly to obtain genes from larger or older males (e.g., Wagner et al. 1996).

The 2 copulation strategies observed in Brazilian free-tailed bats were used by the same male. One strategy might be more appropriate under different roost conditions than the other. In the bridge colony, where females usually roost in rather loose clusters, a male can more easily pull a female from the group for copulation. However, in a cave it may be energetically too costly if not energetically impossible to gain access to a female roosting in the middle of dense clusters by using the aggressive copulation strategy. The passive copulation strategy might allow males to copulate under conditions typical for caves. However, in the bridge colony, the passive strategy might be advantageous only when females are tightly packed, and therefore this strategy was seldom used.

Although 2 strategies for male copulation were observed, female choice also may play a role in the mating behavior of the Brazilian free-tailed bat. In aggressive copulation, females appear to have little choice in mate selection, except possibly to test a male's strength by resisting, allowing only the strongest males to successfully copulate. During passive copulation, it sometimes appeared that the female did not notice the copulation at all, which if true would eliminate female choice altogether. It was not possible to determine the cause of unsuccessful passive copulatory attempts. As behavioral studies on the little brown bat (Myotis lucifugus) have shown, mating appears to be random with respect to mating success. However, genetic studies have revealed differential male mating success (Thomas et al. 1979; Wai-Ping and Fenton 1988; Watt and Fenton 1995). Clearly, further behavioral and genetic studies are needed to unravel the role of female choice and differential male mating success in T. brasiliensis.

The audible mating call (B. French, in litt.) was emitted by males of T. brasiliensis only during active copulation. As in M. lucifugus, the mating call may function in courtship (Eberhard 1996), conveying the male's sexual rather than aggressive intention to the female, thus synchronizing their sexual behaviors (Barclay and Thomas 1979). When male M. lucifugus copulate with hibernating nonresisting females, males do not emit the mating call (Barclay and Thomas 1979). Similarly, during passive copulation when female T. brasiliensis are not struggling, males did not emit audible mating calls.

In summary, it appears that T. brasiliensis is behaviorally flexible and that male copulation strategy and social mating system change with colony size and roosting conditions. Although the mating system in small captive colonies can be described as resource-defense polygyny with mating territories, the pattern described for wild colonies of T. brasiliensis resembles that of the gray sac-winged bat (Balantiopteryx plicata), a species that also forms large colonies (Bradbury and Vehrencamp 1977). In the latter species, low group stability and frequent changes of roosting and foraging sites would render defense of roosting sites or females by males energetically costly. We suggest that the behavioral aspect of the mating system of T. brasiliensis in a large highway bridge colony consisting of tens of thousands of bats may best be described as promiscuous. The mating strategy during the short mating period can be characterized as mating aggregations or swarming.

Acknowledgments

We thank Bat Conservation International, Inc., for supplying equipment and allowing access to Bracken Cave. The Nature Conservancy and D. Davis permitted access to Eckert-James-River Cave and Davis Blowout Cave, respectively. B. French and G. F. McCracken offered many useful suggestions and advice. G. F. McCracken, T. H. Fleming, M. B. Fenton, J. T. Baccus, R. W. Manning, T. R. Simpson, and 2 anonymous reviewers read earlier drafts of the manuscript and made many helpful comments. Adrian Tejedor kindly translated the abstract into Spanish. Throughout the study, we followed the American Society of Mammalogists Guidelines on Animal Care and Use (available online at http://www.mammalogy.org/committees/index.asp).

Author notes

Present address of ATHK: 2727 N Jamison Boulevard, Flagstaff, AZ 86004, USA

Associate Editor was Mark D. Engstrom.