INTRODUCTION
Competitors, predators and parasites all may serve to limit rates of increase and act as important checks on the distribution, abundance and fitness of organisms. Although competition theory has been an important component of community ecology for half a century (Hutchinson Reference HUTCHINSON1959, MacArthur Reference MACARTHUR1972), and predation is now universally regarded as a major structuring force (Berger et al. Reference BERGER, STACEY, BELLIS and JOHNSON2001, Terborgh Reference TERBORGH1988), we have much to learn about the importance and roles of parasitism (Morand et al. Reference MORAND, KRASNOV and POULIN2006, Packer et al. Reference PACKER, HOLT, HUDSON, LAFFERTY and DOBSON2003, Thompson Reference THOMPSON1999). Richer empirical documentation offers tests of theoretical alternatives and points towards new mechanisms of interaction (Poulin Reference POULIN2007).
Bats (Order Chiroptera) are highly abundant and diverse in Neotropical forests. In places such as Guyana's Iwokrama Forest, about two-thirds of all mammals (86 of 130 species) are bats (Lim & Engstrom Reference LIM and ENGSTROM2004). Neotropical bats are also impressively abundant, and their biomass can equal that of all other mammal species combined (Tuttle Reference TUTTLE1983). Their ecological success has been traced to their utilization of diverse trophic strategies, roosting structures, locomotory patterns and sensory modalities (Kalko Reference KALKO and Ulrich1997, Patterson et al. Reference PATTERSON, WILLIG, STEVENS, Kunz and Fenton2003). Their ecological importance as pollinators, seed dispersers and insectivores extends beyond primary trophic linkages to exert numerous indirect effects on forest health and vitality (Heithaus Reference HEITHAUS and Kunz1982, Myers Reference MYERS1992). Bats also have remarkably diverse social systems, living in groupings that range from small family groups to aggregations of millions (Wilkinson Reference WILKINSON, Waal and Tyack2003). Such ecological variation may cause bat species to differ dramatically in their susceptibility to different biological controls, as well as in the prevalence and intensity of parasitism. For example, the permanence and exposure of their roosting sites is known to correlate significantly with both the prevalence and intensity of parasitism on Neotropical bats (Patterson et al. Reference PATTERSON, DICK and DITTMAR2007).
Bats are host to various parasites, both internal (Coggins Reference COGGINS and Kunz1988) and external (Whitaker Reference WHITAKER and Kunz1988). The Chiroptera are second only to rodents in the number of families and species of ectoparasitic insect they host (Marshall Reference MARSHALL1981), their ectoparasites exhibit the highest degree of niche specialization, and more orders and families of ectoparasitic arthropod are exclusive to bats than to any other mammalian order (Wenzel et al. Reference WENZEL, TIPTON, KIEWLICZ, Wenzel and Tipton1966). The most conspicuous ectoparasites are blood-sucking flies of the superfamily Hippoboscoidea (Marshall Reference MARSHALL and Kunz1982). Traditionally classified into two families, Streblidae and Nycteribiidae (but see Dittmar et al. Reference DITTMAR, PORTER, MURRAY and WHITING2006), these obligate ectoparasites live only on bats, inhabiting the fur and flight membranes. Like all Diptera, bat flies must pupate or metamorphose to complete development. But unlike other flies, bat flies give birth to a single young that is already a third-instar larva and that immediately pupates. Deposition of larvae and pupation typically happen in or near the bat's roost, and weeks later, the newly emergent flies must relocate a host bat (Dick & Patterson Reference DICK, PATTERSON, Morand, Krasnov and Poulin2006). Despite the decoupling of host and parasite at every generation, bat fly species are strikingly host specific (Dick Reference DICK2007, Dick & Patterson Reference DICK and PATTERSON2007). A recent survey of bat flies on Paraguayan bats determined that 87% of 31 streblid species were restricted to a single bat species; the remaining 13% occurred on closely related host species (Dick & Gettinger Reference DICK and GETTINGER2005).
The fitness costs of bat fly parasitism are not well understood. By definition, parasites are symbionts that derive benefit from the host at some expense to it (Roberts &Janovy Reference ROBERTS and JANOVY2000). But it can be difficult to distinguish the effects of parasitism on host condition from the effects of host health on the virulence of the parasites (Bize et al. Reference BIZE, JEANNERET, KLOPFENSTEIN and ROULIN2008). Actual energetic costs of bat fly parasitism remain poorly defined, but the flies take blood meals up to eight times per hour and typically die after being separated from their hosts for only several hours (Fritz Reference FRITZ1983). Heightened grooming rates, reduced resting time, and elevated metabolism are all consequences of higher mite infestations of European bats (Giorgi et al. Reference GIORGI, ARLETTAZ, CHRISTE and VOGEL2001). Both species co-occurrence analyses and host-choice experiments suggest that the number of flies on a given host bat are limited (Dick & Dick Reference DICK and DICK2006, Tello et al. Reference TELLO, STEVENS and DICK2008). Hosts spend more time grooming as fly numbers increase (Overal Reference OVERAL1980), and grooming is thought to be a major source of fly mortality (Marshall Reference MARSHALL1981). As many as six bat fly species may coexist on a single host bat (Wenzel Reference WENZEL1976), although one to three species is more typical. In cases of multiple infestation, fly species partition the bat's fur and flight membranes and exhibit highly structured ecomorphology (Dick Reference DICK2005). Bat flies are not currently known to transmit disease pathogens or other parasites.
Despite great variation in bat social systems and their daily and seasonal aggregations, few studies have examined sex differences in bat fly parasitism, and these have produced equivocal results. Neither abundance nor prevalence of bat fly parasitism varied with the sex of bulldog bats (Noctilio leporinus) in southern Brazil (Moura et al. Reference MOURA, BORDIGNON and GRACIOLLI2003) or Paraguay (Presley Reference PRESLEY2007). A study in São Paulo, Brazil yielded ambiguous results, with some fly species more abundant on female bats and others more abundant on males (Bertola et al. Reference BERTOLA, AIRES, FAVORITO, GRACIOLLI, AMAKU, PINTO and ROCHA2005). In Minas Gerais, Brazil, the most abundant bat fly (Trichobius joblingi) was found at higher intensity and prevalence on male hosts (Komeno & Linhares Reference KOMENO and LINHARES1999) but in São Paulo this species was differentially found on females (Bertola et al. Reference BERTOLA, AIRES, FAVORITO, GRACIOLLI, AMAKU, PINTO and ROCHA2005). The only study to date to offer a fauna-wide assessment of infestations by sex was of Paraguayan bats, where females were generally more heavily infested than males (Presley & Willig Reference PRESLEY and WILLIG2008). Because Paraguay straddles the Tropic of Capricorn and its modest-sized bat fauna contains both tropical and temperate elements (López-González Reference LÓPEZ-GONZÁLEZ2004), we decided to assess the generality of this pattern using surveys of hyperdiverse tropical bat faunas and their associated bat flies.
METHODS
The Smithsonian Venezuela Project (SVP) was conducted from 1965–1968 and was one of the largest coordinated surveys of vertebrates and their ectoparasites ever conducted. That effort sampled 38 213 mammals representing 270 species, including 24 979 sexed bats of 133 species. Most mammals were captured and held in individual paper bags, fumigated with ether, and their parasites collected in 70% ethanol (Handley Reference HANDLEY1976). All mammals were identified and curated by Charles O. Handley at the National Museum of Natural History (USNM). Host names were recently reviewed using computerized records of the National Museum of Natural History to revise names in accordance with subsequent taxonomic revisions and re-identifications. Thus, host identifications are accurate, consistent, and can be re-evaluated, all critical attributes for host-parasite studies (Poulin Reference POULIN1998, Wenzel et al. Reference WENZEL, TIPTON, KIEWLICZ, Wenzel and Tipton1966).
Bulk parasite samples were collected and organized by individual host. At the Field Museum of Natural History (FMNH), Rupert L. Wenzel sorted, identified and enumerated the streblid bat flies (Diptera: Streblidae). Collections included 36 663 streblids, representing 22 genera and 116 species; two genera and 45 species were new to science (Wenzel Reference WENZEL1976). Records analysed here consisted of the number of a given bat fly taxon recovered from a host individual; such aggregations have been termed an ‘infrapopulation’ in earlier parasite surveys (Bush et al. Reference BUSH, LAFFERTY, LOTZ and SHOSTAK1997). Multiple records were used for bat individuals that hosted two or more species of flies. We excluded SVP records not catalogued at the USNM (and hence lacking re-confirmed identifications), as well as those that lacked bat or bat fly identifications, gender specifications, or counts of parasite abundance. A total of 24 979 bats, representing 133 species in nine families, was sampled for parasitic bat flies.
We used G-tests to evaluate differences in proportionality between several measures of parasitism on male and female bats. Each fly has the opportunity to colonize either a male or female host, so that departures from equal proportions can be tested directly with this statistic if those decisions are assumed to be statistically independent. To avoid spurious effects of small sample size, we restricted tests to 47 bat species with ≥ 20 individuals infested with streblid flies. We tested differences between male and female infestations (non-zero counts of the number of bat flies on a given bat), the total number of flies (summed infestations), and number of fly infrapopulations (the number of bat fly species comprising an infestation). Two different sets of expectations were used to assess observed values of these variables: capture frequencies (fractions of male and female bats in captures sampled for ectoparasites) and infestation frequencies (proportions of males and females among those bats bearing streblid flies). Thus, if twice as many female bats of a given species were captured as males, ‘proportionate to capture’ tests expected twice as many females to be infested. ‘Proportionate to infestation’ tests used expectations based on sex ratios of infested individuals. To control experiment-wide errors at P < 0.05 in a survey involving so many independent tests, we used both Bonferroni (α < 0.00106) and Sidak adjustments (α < 0.00109) to assess the significance of individual tests (http://www.quantitativeskills.com/sisa/index.htm), although some would argue this conservative treatment raises the risk of ignoring some biologically significant differences (Moran Reference MORAN2003). We tabulated actual G statistics to allow readers to apply their own standard of significance.
Sexual size dimorphism of these species was estimated by Linares (Reference LINARES1998), who judged 15 of 46 species (Dermanura glauca was not evaluated) to show notable dimorphism. We evaluated whether significantly biased loads were associated with sexual dimorphism using a binomial test, where ‘success’ was defined as larger infestations on the larger sex, ‘trials’ were defined as the significantly dimorphic bat species (n = 15), and the hypothetical probability of success was equal to the proportion of bat species with significantly biased parasitic loads. Our alpha values report the probability of finding as few or fewer agreements between significant sex bias and appreciable sexual dimorphism as were actually observed.
RESULTS
Bat flies were recovered from 6935 bats, representing eight of nine families and 89 of the 133 bat species captured (Appendix 1). Streblid flies were recovered from all sampled species of Mormoopidae, Noctilionidae, Natalidae and Furipteridae, most species of Phyllostomidae, and select species of Emballonuridae, Molossidae and Vespertilionidae. No streblids were recovered from 11 individuals of one species of Thyropteridae, Thyroptera tricolor. Over all bat species, mean prevalence (proportion of infested individuals) averaged 27.8% (28.5% for females, 26.9% for males). Forty-seven species were captured frequently enough (≥ 20 infested individuals) to permit reliable estimates of streblid parasitism on males and females; prevalence among well-sampled species averaged 33.9% (36.4% for females, 31.4% for males).
Both captures and infestations were biased towards females (Appendix 1). Females comprised 53.2% of captures and 54.6% of infestations over all species, and values for well sampled species were comparable (50.9% and 54.6%, respectively). The excess of infested females in both groupings is significant in tests of proportionality between captures and infestations: G = 67.8 for all species and 99.8 for well-sampled species (both P < 0.0001).
Relative to capture frequencies, bat fly infestations of males and females in most species did not differ. However, six species showed disproportionately high infestations of females (Table 1; all P < 0.05 after Bonferroni correction). In no species were males infested more heavily than females. The intensity of parasitism also differed between sexes in 21 of the 47 species (Table 1). In five cases, males carried disproportionately heavy loads, whereas in 16 cases, females hosted significantly more flies than males. Finally, eight species showed significant sex differences in the number of fly infrapopulations supported by males and females, including all six of the species with prevalence differences (Table 1). In seven of the eight species, females supported significantly more species of bat flies than males.
Tests for sex differences in bat fly parasitism. G statistics and the predominating sex are highlighted where deemed significant (P < 0.05) using Sidak's Bonferroni adjustment for experiment-wide error rates (P < 0.00109 in individual tests). Tests assess infestations, the total number of flies, and the number of fly infrapopulations against expectations based on capture frequencies (first three columns) and based on infestation frequencies (final two columns).
Most of these sex differences in parasitism appear to be consequences of differential infestation. When fly loads of males and females are compared relative to levels of infestation for each sex, 12 species show significant differences between sexes (Table 1); in seven, females carried heavier loads whereas in five, males did. No species carried unbalanced infrapopulation loads after adjusting for differential infestation frequencies (Table 1).
Significantly biased parasite loads were not obviously related to sexual size dimorphism. Fifteen species show significant sexual size dimorphism (Table 1); the 23 species showing at least one significant bias in parasitism included seven where females were larger, 13 where the sexes were subequal, and three where males were larger. Although male biases were observed in Noctilio leporinus, where males are larger than females, they were also observed in Pteronotus davyi and Leptonycteris curasoae where the sexes are equal in size, and in Chrotopterus auritus and Anoura latidens where females are larger. Binomial tests determining whether the larger sex was significantly more heavily parasitized are tallied in Table 2; all four possible tests indicate that sex bias in parasitism and host size dimorphism are unrelated.
Binomial tests of association between sex-biased parasitism and sexual dimorphism in body size. The first three columns designate test parameters while the last tallies the probability that associations between the two are at random.
DISCUSSION
This extensive survey of a major neotropical bat fauna uncovered significant differences between sexes in levels of infestation and parasite loads involving streblid bat flies. In most cases and by most measures, tests showed females were more frequently and heavily parasitized. Despite some methodological differences and our analysis of a much richer fauna, this study confirms the main conclusions of Presley & Willig (Reference PRESLEY and WILLIG2008) that female bats are generally more heavily infested by ectoparasites. Because it encompassed many more infested bat species (89 versus 19–22 species in Paraguay), surveyed many more bats (24 979 versus 2909 in Paraguay), and restricted analysis to a single family of parasites (versus 11 in Paraguay, where 55% of ectoparasites were macronyssid mites), this survey refines the sex-bias pattern and strengthens this conclusion.
Higher infestation of bat flies on female hosts is a result that contrasts with studies of other mammalian ectoparasites. Across many host–parasite systems, when the sexes are differentially infested, it is commonly males that exhibit higher parasite loads. A meta-analysis of sex differences in parasitism showed that biases tended towards male hosts, particularly where arthropod parasites were involved (Schalk & Forbes Reference SCHALK and FORBES1997). Male-biased infections by nematodes are also widespread and well documented in field studies (Poulin Reference POULIN1996). But at least one other group of bat ectoparasites, mites, also typically shows female-biased infestations (Christe et al. Reference CHRISTE, GLAIZOT, EVANNO, BRUYNDONCKX, DEVEVEY, YANNIC, PATTHEY, MAEDER, VOGEL and ARLETTAZ2007, Whitaker et al. Reference WHITAKER, WALTERS, CASTOR, RITZI and WILSON2007).
Many bat species exhibit slight sexual dimorphism in body size, females in many species being 1–7% larger than males in linear dimensions (Ralls Reference RALLS1976). Other factors being equal, larger size means more ‘habitat’ for the parasite and perhaps larger refuges from host grooming. Although a third of the sampled species were sexually dimorphic in size, dimorphism was not associated with significant sex biases (Table 2). Elsewhere, size dimorphism was not a significant component of variation in Paraguayan bat fly infestations (Presley & Willig Reference PRESLEY and WILLIG2008). Sex differences in body size of these bat species are evidently minor compared to other ecological and behavioural differences affecting parasite loads.
Many factors – physiological, morphological, behavioural and social – may differentially expose either sex to parasitism, and the ecological diversity of bats makes them excellent subjects for teasing these variables apart. However, many relevant variables were not tracked in the SVP surveys, handicapping our ability to assess them. In fact, most bats sampled by SVP were captured with mist nets while they foraged, a technique that obscures their social groupings, associations, and roosts. As discussed below, physical, social and physiological contexts are vital for evaluating various explanations for parasite loads, so a number of interesting hypotheses could not be tested with these data. Future analyses of fly parasite loads in bats should focus on parasitism levels among individuals of known social groupings, preferably collected at the roost where inter-relationships of individuals could be assessed.
One class of explanations for sex bias in parasitism centres on immunological mechanisms. Immunocompetence has been proposed to explain the higher intensity of spinturnicid (wing mite) parasitism on reproductive and pregnant female bats versus non-reproductive females (Christe et al. Reference CHRISTE, ARLETTAZ and VOGEL2000). High testosterone levels in males may be responsible for immunosuppression and lead to male-biased infestation patterns (Folstad & Karter Reference FOLSTAD and KARTER1992). Hormones may also influence the levels and seasonality of parasitic infestations independently of immunology, as rabbit fleas depend on the hormones of reproducing female hosts to complete their own ovarian maturation (Mead-Briggs & Rudge Reference MEAD-BRIGGS and RUDGE1960, Rothschild & Ford Reference ROTHSCHILD and FORD1964).
But many ecological factors are likely to influence sex-biased parasitism, and some of these may take precedence over immunological factors in explaining variation in levels of parasitism (Krasnov et al. Reference KRASNOV, MOUILLOT, SHENBROT, KHOKHLOVA and POULIN2005). Heavier loads on females could be attributable to differential roost-site selection by male and female bats (Zahn & Rupp Reference ZAHN and RUPP2004). Bats roosting in more permanent and enclosed shelters exhibit a higher prevalence, intensity and diversity of parasitism than bats that roost in exposed, transient structures (Patterson et al. Reference PATTERSON, DICK and DITTMAR2007). Permanent roosts, such as caves, mines, and other large cavities, typically house larger colonies of bats than do ephemeral roosts such as tree leaves and smaller cavities (Kunz Reference KUNZ and Kunz1982). The permanence of bat roosts is also closely tied to the fidelity of bats roosting there (Lewis Reference LEWIS1995), which in turn determines the dynamics of roost-switching. Roost-switching may serve as a behavioural strategy for avoiding parasites altogether or reducing parasite loads (Reckardt & Kerth Reference RECKARDT and KERTH2006, Timm Reference TIMM, Patterson and Timm1987).
Females might select higher-quality roosts that attract larger numbers of bats, which in turn facilitate higher levels of infestation by making them more vulnerable to infestation (Krasnov et al. Reference KRASNOV, SHENBROT and KHOKHLOVA2002, Zahn & Rupp Reference ZAHN and RUPP2004). However, the density of roosting bats is critical to transmission, as large roosts offer more space for bats to spread out, thus decreasing physical contact and the likelihood of parasite transfers (Dick et al. Reference DICK, GANNON, LITTLE and PATRICK2003). The colony-size effect has been observed to affect louse infestations of certain Galapagos birds, where amblyceran lice responded positively to an increase in the degree of host sociality (Whiteman & Parker Reference WHITEMAN and PARKER2004). Females in many bat species roost in dense maternity colonies (McCracken Reference MCCRACKEN1984); the dense crowding of these colonies helps to maintain high body temperatures needed for the rapid growth of juvenile bats but may also facilitate transmission of parasites to colony members. Bat flies are known to move between host individuals when they roost so closely together that their fur is touching (Overal Reference OVERAL1980). Females in many species of mammals enjoy higher instantaneous survivorship rates and so might be a better prospect for a colonizing parasite or alternatively be more likely to accumulate a non-selective one.
Curiously, documented effects of bat aggregations seem to enhance their parasitism by streblids, not reduce it. Reductions might be expected if allo-grooming were widespread and important in controlling parasite loads (Wohland Reference WOHLAND2000). Although grooming may be the principal source of bat fly mortality (Marshall Reference MARSHALL and Kunz1982) and appears to influence host-site selection by the flies (ter Hofstede et al. Reference TER HOFSTEDE, FENTON and WHITAKER2004), grooming behaviour itself does not predict parasitism levels in bats. In fact, harem male Artibeus lituratus in Venezuela spend half as much time during the day grooming as do harem females (Muñoz-Romo Reference MUÑOZ-ROMO2006), yet the male A. lituratus in our sample enjoyed significantly lower infestation levels (Table 1). Again, knowing the social affiliations of the bats sampled would help to illuminate the role of these variables and the behavioural, physiological and social conditions promoting susceptibility. Future field surveys should incorporate social biology into their experimental design.
ACKNOWLEDGEMENTS
We owe a special debt of gratitude to Charles O. Handley, Jr. and Rupert L. Wenzel for their remarkable taxonomic expertise and great efforts to sort, identify and describe the mammals and streblids of the SVP. We are grateful to Harvey Golden for insightful discussions on literature, Matt Dean and Jamie Bender for their help in developing a database of the SVP records, and the National Science Foundation (DBI-0545051 and DEB-0640330/1) and Stephanie Ware for support to complete their development. Four reviewers, including L. Rozsa and J. O. Whitaker, Jr., offered insightful commentary on an earlier draft that helped to strengthen our arguments.
Records of streblid bat flies on Venezuelan bats, by sex. Taxonomy follows Simmons (Reference SIMMONS, Wilson and Reeder2005) except as discussed by Solari et al. (Reference SOLARI, PACHECO, LUNA, VELAZCO, PATTERSON, Patterson, Stotz and Solari2006). Captures record all sexed, identified, and catalogued bats that were processed for ectoparasites, while infested bats reflect the number of those harbouring one or more streblid flies. Prevalence and mean intensity of all streblids calculated separately from male and female totals. Species marked with an asterisk were deemed to have been adequately sampled for statistical tests.
