Behavioral Ecology Advance Access originally published online on August 22, 2006
Behavioral Ecology 2006 17(6):971-978; doi:10.1093/beheco/arl034
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Multiple paternity in a philopatric rodent: the interaction of competition and choice
a Departments of Biological Sciences, Purdue University, West Lafayette, IN 47907, USA b Forestry and Natural Resources, Purdue University, West Lafayette, IN 47907, USA
Address correspondence to P.M. Waser. E-mail: pwaser{at}bilbo.bio.purdue.edu.
Received 9 December 2005; revised 29 March 2006; accepted 12 July 2006.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSResultsDiscussionREFERENCES |
|---|
Paternity confusion is often suggested as the benefit that female mammals accrue by mating with multiple males, but genetic advantages are also possible. Microsatellite-based parentage analyses demonstrate that female banner-tailed kangaroo rats (Dipodomys spectabilis) commonly mate with more than one male; we asked how male and female behaviors interact to influence the characteristics of males that sire offspring. Specifically, we compared attributes (age, weight, mobility, relatedness, proximity) of the fathers of 229 known-maternity offspring with those of the other males accessible to the mothers. Adult males living adjacent to each female attempt to monopolize access to her, and the nearest male sires a plurality of offspring, but most mothers' young are fathered by more than one male and littermates are usually half-sibs. Male proximity and mobility, but not size, influence the probability of paternity, suggesting a role for competitive mate searching. Females significantly reduce the inbreeding coefficient of their offspring by mating with males other than (or in addition to) the nearest male. Fathers are less closely related to the mother in years of high density when unrelated males are more accessible to the female. Our results favor the genetic "bet-hedging" hypothesis, whereby females actively but unselectively seek matings with additional males when the male most likely to win in mate competition is costly to her (in this case, genetically less compatible). We anticipate that genetic bet hedging will be common in species whose females are defendable, especially if they are also philopatric.
Key words: Dipodomys, genetic bet hedging, kangaroo rat, mate choice, multiple mating, parentage.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSResultsDiscussionREFERENCES |
|---|
In early models of the evolution of mating systems (Emlen and Oring 1977
), mate competition plays the predominant role: ecological factors determine the spatiotemporal distribution of receptive females, and this distribution in turn determines whether males can monopolize access to them. If one male can largely exclude others from the vicinity of a female, she will have little opportunity to exercise choice among mates. The idea that female spatiotemporal distribution might influence the interplay between mate choice and competition had a major impact on discussions of mammalian mating systems in general (e.g., Clutton-Brock 1989) and of rodent mating systems in particular (e.g., Koprowski 1998; Solomon and Keane forthcoming).
An option available to females that complicates the ability of mate competition to constrain choice is that of mating with more than one male (e.g., Gowaty 2005). Among mammals, it has never been a surprise to find evidence of multiple mating by males (e.g., Orians 1969), but increasing evidence suggests that multiple mating by females is nearly as common (Wolff and Macdonald 2004). It may be that multiple mating is a side effect of mate competition, disadvantageous to females (Lee and Hays 2004). On the other hand, there are many ways in which multiple mating might benefit females (e.g., Birkhead 2000; Tregenza and Wedell 2000; Fedorka and Mousseau 2002; Stockley 2003). Stockley et al. (1993) provide a useful categorization. According to their "best-male" hypothesis, females mate with more than one male only if after their initial mating they encounter a second, superior male. Under the "genetic diversity" hypothesis, females mate multiply to maximize genetic variation among their offspring. A third alternative is the "genetic bet-hedging" hypothesis: if females cannot avoid the possibility of mating with genetically inferior males (because they have either imperfect information or imperfect control of mating), they should dilute the risk by mating with more than one.
Resource distribution, female "defendability," and mate choice have been extensively investigated among sciurids and microtines, but heteromyid rodents (kangaroo rats and pocket mice) can also serve as exemplars. In some species, like the Merriam's kangaroo rat (Dipodomys merriami), females exploit widely distributed resources, scatterhoarding seeds, and moving extensively among multiple burrows within large home ranges whose locations drift through time (Behrends et al. 1986a, 1986b; Jones 1989; Daly et al. 1992). Under such circumstances, a female might readily exert mate choice; her mobility and home-range size would make it difficult for a male to limit her contact with other males. Indeed, female D. merriami home ranges overlap those of multiple males, and females become more mobile during estrous periods. In contrast, banner-tailed kangaroo rats (Dipodomys spectabilis) are larder hoarders, each individual usually residing in a single conspicuous den whose location is stable not only from week to week but also from generation to generation (Jones 1984; Randall 1984). These circumstances should make females defendable from the perspective of a male; male and female dens are interspersed, and males repeatedly visit and attempt to exclude other males from access to close female neighbors during their brief estrous periods (Randall 1987, 1991).
Preliminary genetic data from banner-tailed kangaroo rats (Winters and Waser 2003) confirm Randall's behavioral observations in that many of each female's young are sired by the male living nearest to her. However, they also suggest that females have means of thwarting male attempts to monopolize access. Many offspring are sired by males that are not their mother's nearest male neighbor, and in such cases, the sire is generally less related to her than the nearest neighbor. In addition, some litters are sired by more than one male. Mating with multiple males has now also been reported in the ecologically similar Dipodomys ingens (Randall et al. 2002).
These findings are of particular interest in a highly philopatric species like the banner-tailed kangaroo rat. Where dispersal is limited, neighbors are often close relatives and inbreeding may be difficult to avoid (Waser and Elliott 1991; Winters and Waser 2003). Thus, in this species, there is reason to believe that ecological factors provoke a particularly strong conflict between male and female interests because the males that most easily defend access to a female are often those that are less desirable to her.
In this study, we search for evidence of control and choice by extending the analyses of Winters and Waser, using much larger samples of animals (10 years' data rather than 4) and microsatellite loci (11 rather than 5). First, we describe patterns of parentage: do most females produce offspring sired by multiple fathers, or are most litters fathered by a single male? Second, we ask to what extent young are sired by those males likely to dominate in mate competition: do sires tend to be the mother's nearest male neighbor, or larger than average, as would be expected if proximity or size biased the outcome of male–male aggression (Randall 1991)? Or are offspring more likely to be sired by males that are mobile, as would be expected from the possibility of competitive mate searching (Schwagmeyer 1995)? In addressing this question, we also investigate female age and mobility, as well as population density—factors that might influence the relative success of mate competition and mate choice. For example, males might find it easier to monopolize access to females that are younger or more sedentary, whereas older or more mobile females may have more complete information regarding nearby males or be more capable of exerting their own preferences. Population density influences the number of males in each female's vicinity, making more choices available to her in high-density years and perhaps making it more difficult for her immediate neighbor to monopolize access.
Next, we ask to what extent patterns of paternity are those expected if females engage in mate choice. Are young sired primarily by older males (as might be expected if age is an indicator of fitness), unrelated males (as might be expected if there is inbreeding depression), or distant males (expected if nearby males are low quality in any sense)? Finally, we ask whether patterns of paternity give us any insight into what females gain by mating with multiple males. Assuming that nearby, closely related males are likely to "win" in mate competition and mate first (Randall 1991), then any of the hypotheses of Stockley et al. (1993) predict that females should seek additional matings with more distant males. Easiest to distinguish would be the bet-hedging and best-male hypotheses. The former predicts that such matings should be random choices among the males available; we can reject it in favor of the best-male hypothesis if we find (after controlling for proximity) that sires are older or less related than the alternatives.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSResultsDiscussionREFERENCES |
|---|
Field collection of genotypes and demographic data
We describe patterns of parentage in D. spectabilis based on 10 successive breeding seasons' data (1990–1999) from a 1-km2 study area in SE Arizona (31°37'N, 109°15'W). Adults in this population exhibit annual survival rates on the order of 0.4–0.5; females produce 0–4 litters of 1–3 offspring each year in a breeding season that begins in late November or December and continues for 3–6 months depending on rainfall (Randall 1991
; Waser and Jones 1991). The natural history of these animals—in particular, the fact that they build conspicuous dirt mounds over labyrinthine dens that they inhabit singly—allowed us to trap, genotype, and locate the residences of virtually 100% of adults during twice-annual mark-release–recapture censuses (Skvarla et al. 2004). Mound locations were mapped to a precision of ±1 m with a theodolite (neighboring mounds are generally separated by 10–20 m). At the beginning of each approximately week-long census, we first determined which mounds showed signs of activity (recent excavation, cleared sandbathing sites around the periphery). We then set 3 Sherman live traps at every active mound on at least 3 nights. When we first captured an animal, we cut a small snip from one ear, which we froze in liquid nitrogen within a few hours. At the same time, we recorded the individual's weight, sex, and location; gave it a uniquely numbered ear tag; and categorized it as juvenile or adult (defined by scrotal testes or conspicuous nipples). Because we first caught nearly all animals well before they reached adult weight, we could also establish year of birth and thus age. At every subsequent capture, we recorded the individual's weight, location, and reproductive status.
Using criteria developed by Jones (1984), we assigned ownership to each mound during each census and determined which mound was the primary residence of each adult. An adult's primary residence was the mound at which it was trapped more often than at any other mound and more often than any other adult. Some adults "owned" (were trapped more often than any other adult at) 1–2 additional mounds adjacent to the primary residence; we termed these secondary mounds. Radio tracking in this and other populations (Schroder 1979; Jones 1984) has demonstrated that even those adults that maintain secondary mounds spend the great majority of their time in their primary residence. Primary residences are highly stable across censuses (Jones 1987; Waser and Elliott 1991).
Our first census each year was in March, late in the breeding season. Because we knew the residence of each adult during the census, we also knew the distance of each male from each adult female. From trapping records collected during the March census, we calculated the mean weight for each adult (mean for all males = 138 g, females = 128 g). We knew the age of all animals except those that were already adult when first captured; in all years but 1990, we could characterize every adult as either young (1 year old) or old (2–6 years). Based on its trapping records, we also determined each adult's mean distance from its primary residence as a measure of its mobility.
Most juveniles are born in late winter or early spring and are first independent enough to venture into traps between March and June. In most years, we therefore did supplemental trapping in May and June so as to tag juveniles at or near their mother's residence. By our second standard census in late July/early August, juveniles were approaching adult weight (
125 g). Some first-year females showed signs of sexual activity (swollen vulvas, copulatory plugs, elongated nipples), but although females have occasionally become pregnant in other populations during their first summer (Waser and Jones 1991), we detected no birth-year pregnancies in this study. Thus, we believe that with few exceptions, young of both sexes reproduced for the first time during the winter following their birth.
Our long-term trapping data allowed us to examine patterns of paternity both overall and also with respect to maternal age and mobility. Thus, we examined paternity separately for juveniles whose mothers were 1 year versus 2–6 years old (data from 1990 were excluded). Similarly, we examined patterns separately for "mobile" females (those trapped away from their primary residence at least once during a census) and sedentary ones.
Finally, we examined patterns of parentage over a range of population densities. The mounds that we censused were divided into 3 clusters of 12–46 mounds, referred to elsewhere as R1E, R1W, and SSW (Skvarla et al. 2004). Clusters were colonized from neighboring areas in 1987 (R1E, R1W) and 1993 (SSW) and grew in a logistic fashion during the 1990s (Waser and Ayers 2003). Between 1990 and 1999, the total adult population size varied from 17 to 65, and the adult sex ratio varied from 0.8 to 1.2 females/male. Both population size and sex ratio might be expected to influence the intensity of mate competition and the ability of a female to exert choice. An annual index reflecting both factors is the median distance between the primary residence of an adult female and that of her nearest adult male neighbor.
Microsatellite markers and genotyping methods
We used a set of 11 microsatellite loci (Ds1, Ds3, Ds19, Ds28, Ds46, Davis et al. 2000; Ds98, Ds107, Ds109, Ds163, Ds222, Ds281, Waser et al. 2006) developed for this species. Polymerase chain reaction (PCR) primers were designed for each locus and screened for polymorphism using individuals from our study site. Amplifications for all loci were conducted in 8-µl PCRs consisting of 0.5 µM each primer, 250 µM dNTPs, 1.88 mM MgCl2, 1.25x reaction buffer (Promega), and 0.75 units Taq DNA polymerase (Promega). Thermal profiles are described in Waser et al. (2006). Individual genotypes were determined by PCR amplification using a single end-labeled primer for each locus, followed by electrophoresis in a 4.75% polyacrylamide gel on an ABI377. Genescan 3.1 and Genotyper 2.1 software from ABI were used to assign locus-specific genotypes.
Parentage and relatedness analyses
For juveniles genotyped at 4 or more loci, we assessed parentage using a combination of exclusion and likelihood-based inference (Winters and Waser 2003; Waser et al. 2006). As candidate parents for each juvenile, we initially included all adults whose primary residence was within 100 m of that juvenile's first capture location, either during the March census of the juvenile's birth year or the preceding August census. Our first step was strictly exclusionary based on 2 X-linked loci, Ds19 and Ds222 (see Winters and Waser 2003). We then used CERVUS (Marshall et al. 1998) to infer parentage based on 4–9 (mean 7.75) autosomal loci. As recommended by Morrissey and Wilson (2005), we set the error rate for likelihood calculations at 0.001, below the observed genotype error rate of 0.01. We accepted parental candidates only if CERVUS assigned them a confidence >95%. In 8 cases, these last 2 rounds of analysis produced 2 different "best pairs" with confidence >95%. When this happened, we assigned the pair with the shortest mother–father distance as parents.
If CERVUS did not find parents within 100 m, we repeated the analysis including as candidates all males with residences <250 m from the female. Finally, we searched using candidate males within 500 m. In these latter cases, however, we accepted parents only if CERVUS assigned them at >95% confidence, genotype comparisons were based on 8 or more loci, and there were no mismatches. As described in Winters and Waser (2003), we independently assigned juveniles to "field-determined" mothers if they were captured repeatedly in the primary residence of a living adult female, an approach that genetic parentage assignment showed to be highly reliable. For 12 juveniles, we had no genotype for the field-determined mother, CERVUS found no candidate mothers within 100 m, so we accepted the field-determined mother and the CERVUS-determined father as the true parents.
We considered siblings whose estimated birth dates were within 20 days of each other to be littermates (the shortest interbirth interval we observed was 35 days). Our birth date estimates were based on the age–weight relationship exhibited by animals collected on a nearby site during 1982–1985. During those years, females were trapped at least weekly between February and July, allowing us to determine birth dates of particular litters. An exponential function fit the age–weight relationship well for animals weighing <120 g (R2 = 0.86) and intersected the y axis at 8 g, approximately the weight of a neonate (PM Waser, unpublished data).
Using program KINSHIP (Goodnight and Queller 1999), we constructed a matrix of relatedness estimates in each year, including all adults genotyped at 4 or more autosomal loci.
Statistical methodology
To examine whether multiple paternity was influenced by a female's age, weight, or mobility or by population density, we used Generalized Linear Mixed Models (GLMM), (the SAS GLMMIX macro, Littell et al. 1996). Our dependent variable was the number of different males that sired a female's young during a breeding season. We performed a parallel analysis for males, asking how a male's characteristics influenced the number of females that gave birth to his young. We treated age and mobility as categorical variables. Because we had data from some individuals in more than 1 year, we used the REPEATED statement with individuals as subjects, assuming compound symmetric covariance structure. We used a backward stepwise procedure, entering all interactions into the initial model, then removing interaction terms (more complex first) that did not reach significance at alpha = 0.25 and rerunning the model.
To assess the extent to which nearest males monopolize females (and whether sires are larger, more mobile, older, or less related), we 1) compared nearest males that sired offspring with those that did not, using 2-sample t-tests after checking for normality and 2) compared nearest males that did not sire offspring with sires, using paired t-tests. Unless otherwise noted, we report means ± standard errors.
It was less straightforward to compare the characteristics of sires with those of the average male accessible to a female because each mother was surrounded by a different selection of males. Therefore, we developed a randomization test—a simulation of hypothetical females whose neighbors were identical to those of our juveniles' true mothers, but that mated at random. For each offspring, our simulation located the mother, chose randomly among the males available to her, recorded the chosen male's characteristics (proximity, weight, age, mobility, and relatedness), and then repeated the process for the next offspring. When all simulated mothers had mated, the program determined the mean characteristics of those randomly chosen males. This procedure was then repeated 1000 times. Finally, we compared the mean characteristics of true sires with the 1000 random samples. P values were directly determined from this comparison. For example, if the mean age of the actual sires was greater than the mean ages of males in 957 out of our 1000 random replicates, we concluded that the probability of our result arising by chance was 0.04.
Trapping, radio-tracking, and spool-and-line–tracking data indicate that banner-tailed kangaroo rats can move >100 m from their residences during the course of a night, though they rarely do so (Winters and Waser 2003; Steinwald 2004). To be conservative, the first version of our simulation assumed that all adult males with known genotypes whose residences were within 250 m of the mother's were potential mates. In effect, this criterion assumes that a female has access to any male within the mound cluster she lives in. Because we found that most offspring were sired by nearby males, in a second version of the simulation, our hypothetical females' preferences were weighted by distance to match the pattern observed in real females but were random with respect to other male characteristics. This approach allows us to ask whether male size, age, mobility, or relatedness predict paternity while controlling for the effects of proximity.
|
Results |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSResultsDiscussionREFERENCES |
|---|
Over the 10 years of the study, we trapped 645 different individuals and genotyped 524 (81%) of them at 4 or more loci. Of the 433 genotyped individuals that we first trapped as juveniles, we were able to assign mothers for 306 (71%), fathers for 280 (65%), and both parents for 214 (49%). In addition, we found both parents for 15/72 animals that we first trapped as adults after the initial year of the study and that we therefore believed to be 1-year-olds.
Patterns of parentage
The data confirm that male banner-tailed kangaroo rats are highly polygynous. Of the 55 fathers in our sample that produced at least 2 known-maternity offspring during their lifetimes, 53 (96%) sired them with 2 or more different females (median 3 females, range 1–8). Within a breeding season, polygyny was nearly as pervasive. Males producing at least 2 known-maternity offspring fathered young with multiple females during 45/52 male-years (87%, median 2 females, range 1–5, 1 male-year is 1 male observed over the course of 1 year).
At least as striking, however, is the degree of polyandry exhibited by females. Of the 57 mothers that produced 2 or more known-paternity offspring, 46 (81%) had mated with more than one male (median 2 males, range 1–6) during their lifetimes. Within a breeding season, females conceived offspring by multiple fathers in 44/55 female-years (80%; median 2 males, range 1–4). Given their propensity for polyandry, it is not surprising that females often produce litters sired by multiple males. We were able to assess the possibility of multiple paternity within a litter for 52 litters of 2 and 12 litters of 3; 42/64 litters (66%) were sired by more than one male, and in 4 litters of 3, each juvenile had a different father.
Based on the 55 female-years for which we detected >1 offspring, GLMM indicated that a female's tendency to mate with multiple males was influenced by how mobile she was (the young of females trapped away from their residences were fathered by 2.4 ± 0.1 different males, sedentary females' young were fathered by 1.9 ± 0.1 males, P = 0.01) and by her age (females >1 year old had young sired by 2.4 ± 0.1 males, 1-year-old females' young were sired by 1.8 ± 0.1 males, P = 0.03) but not by her weight or by population density (both P 
0.65). In an equivalent analysis for males, we detected no tendency for a male's age, mobility, or weight or for population density to influence whether he mated with multiple females (all P 0.22). No significant interactions were detected in either analysis.
Do males control female mating opportunities?
If mating patterns are determined primarily by mate competition, we would expect most offspring to be fathered by the nearest male (Randall 1991) or by males that were larger or more mobile. Instead, the median distance between mother and father was 58 m, more than 3 times the median distance of 18 m between a mother and her nearest male neighbor. Somewhat surprisingly, we found that this distance was approximately constant across years and independent of or even positively related to population density (the slope of the relationship between mother–father distance and population density was marginally positive, t = 2.20, P = 0.06). The average mother had nearly 4 times as many males within 60 m during the highest density year, yet the male that sired her offspring was just as far away.
Of the 229 known-parentage offspring, only 53 (23%) were sired by the mother's nearest male neighbor. The ranked distance between mothers and sires is fit by a negative exponential function (Figure 1). This pattern closely approximates the geometric distribution that would be expected if offspring were sired by the nearest male with a probability of 0.25, by the second-nearest male with the same probability if the nearest is rejected, and so on (Miller and Carroll 1989).
|
Nearest male neighbors that sired offspring were neither larger nor more mobile than nearest male neighbors that did not (Table 1). Comparisons between sires and nearest neighbors for the 176 offspring that were not sired by the nearest neighbor produced the same result. In this analysis, nearest male neighbors were just as large and mobile as the more distant males that sired the offspring (Table 2).
|
|
Is there evidence of mate choice?
We detected no difference in age between nearest males that sired offspring and those that did not, a result that was mirrored by the similarity of ages between nearest male neighbors and sires (Tables 1 and 2). On the other hand, both comparisons indicated that mating patterns reduce inbreeding. Nearest neighbors that fathered young were significantly less related to the mothers than those that did not (Table 1). Where young were not fathered by the neighbor, mothers and neighbors were as related as first cousins, but mothers and sires were essentially unrelated (Table 2).
Simulation 1: do sires differ from randomly chosen males?
Not surprisingly, with this simulation we found that the proximity of a male to the mother was much lower than expected by chance (P < 0.001, randomization tests). The median distance of 58 m to fathers was much shorter than the median distance of 135 m to a randomly chosen male in the same mound cluster.
This effect of proximity overshadowed any other differences between fathers and random males. Fathers did not differ from randomly chosen males in traits that might be related either to competitive ability (weight, mobility) or to female preference (age, relatedness). The only male attributes that came close to statistical significance were mobility (fathers were trapped farther on average from their residences than random males, P = 0.06) and relatedness (fathers were less related to the female than random males, P = 0.09).
Simulation 2: do sires differ from males that are random in all respects except distance from the mother?
Our second simulation asked how sires differed from males that females would have chosen had they been influenced by proximity in the same negative exponential fashion that we observed but not by male age, weight, mobility, or relatedness. In this comparison, fathers were somewhat more mobile than random males (P = 0.03). As before, fathers were random with regard to age and weight. The tendency for fathers to be less related to the mother than random approached significance (P = 0.06, Table 3).
|
We repeated the second simulation after breaking offspring into subsets (because these analyses represent multiple comparisons based on the same data, P values above 0.01 should be treated with caution). Separating 1-year-olds from older mothers, we found that the tendency for fathers to be more mobile held more strongly for young females (P = 0.01, Table 3). Separating mothers that were always trapped at their primary residence from those that were sometimes trapped away from it, we found that the offspring of more mobile mothers were slightly less inbred than those of sedentary mothers. However, mobile mothers also had less related male neighbors overall. Only for sedentary females did the data suggest that offspring were less inbred than expected by chance (P = 0.02, Table 3).
Finally, running the simulation separately by years, we found that a female's randomly chosen male neighbor was slightly more related to her when population density was high than when it was low. Sires, however, were slightly less related when population density was high. Discrimination against related males was therefore evident only in years of high density, when a female's male neighbors were only a short distance away (P = 0.01, Figure 2).
|
|
Discussion |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSResultsDiscussionREFERENCES |
|---|
Our results make clear that male banner-tailed kangaroo rats do not monopolize female mating opportunities, even though behavioral observations (Randall 1987
, 1991) indicate clearly that they try, and despite the fact that females appear highly defendable, concentrating nearly all their time at a single, stable location (Schroder 1979; Jones 1984; Steinwald 2004). On the one hand, a female's nearest male neighbor was more likely to father her offspring than a male of any other distance rank (Figure 1). On the other hand, more distant males (taken together) fathered three-quarters of all juveniles, and these males had no size advantage over the nearest neighbors they "replaced." Male kangaroo rats can be highly aggressive to each other (Eisenberg 1963; Randall 1993), but we found no evidence that larger males fathered young with more females or that they were more likely to father a female's young than a randomly chosen neighbor.
The spacing pattern of banner-tailed kangaroo rats, with its interspersion of solitary males and females, coupled with the short estrous period of females (as short as a single night, Wilson et al. 1985; Randall 1991), would seem to favor males that monitor and remember the estrous condition of multiple females. The results of our second simulation suggest a tendency for more mobile males to father a disproportionate number of offspring, especially of young females (Table 3). This tendency is consistent with the competitive mate search hypothesis (Schwagmeyer 1995), if mobile males are generally better at finding estrous females, but older females are either harder to find or better able to avoid monopolization than younger ones.
We found no evidence that females selectively mated with older males, but some data are consistent with the possibility that females can assess kinship. Offspring are less likely to be sired by the nearest male when that male is closely related to the mother (Table 1), and when females mate with more distant males, those males are significantly less related than the nearest neighbor (Table 2). On the other hand, randomization tests provide only weak evidence that, after controlling for the effects of proximity, sires are less related than random males (Table 3). In our simulations, the strongest evidence for kin discrimination and inbreeding avoidance comes from sedentary females, an apparently paradoxical result in that these females would seem less able to avoid monopolization than mobile ones. A possible explanation derives from the way that young D. spectabilis disperse. When mothers control secondary residences, their offspring begin using them early in life. As the young mature, mothers gradually restrict their own activity, ceding those secondary residences to their offspring (Jones 1984, 1987). Sedentary females may be those that have recently abdicated control of their secondary residences, so that they are confined to a single mound, and their neighbors are their own offspring from the previous year. If mothers recognize their own adult sons and avoid mating with them, one would predict that 1) male neighbors of sedentary females would be their particularly close relatives and 2) discrimination by such females against such males as fathers would be particularly strong. Indeed, the data support both these contentions (Table 3).
We found that offspring are less inbred in years of high male density (Figure 2). This result too is consistent with the possibility that females have some ability to recognize and discriminate against related males (or their sperm)—assuming that their ability to do so is greater when nonrelatives are more easily accessible. An alternative hypothesis is that inbred offspring are more vulnerable to early mortality (either in utero or while they are too young to trap). If this were the case, then at low densities, all available males would be relatively closely related to the female; as a result, all offspring would be relatively inbred and vulnerable to early mortality and fathers would be a random sample of males. In contrast, at higher densities, some nearby males would be unrelated to the female; their offspring would therefore be more viable than those of related males, so that even if females did not mate preferentially with them, unrelated males would be overrepresented as fathers of the offspring we capture.
We remain uncertain whether female kangaroo rats can directly assess male relatedness, but this is not the only way that a female could exert "choice." The strongest indicators that female mating patterns actively counteract attempts at monopolization are the high frequency of multiple paternity and the fact that most litters are not fathered by the nearest male neighbor. One of the more surprising results of our analyses is that the average father lives approximately 60 m from the mother, whatever the population density—despite the fact that the number of males living within that radius varies across years by a factor of 4. One process that could generate this pattern is that males, females, or both make excursions to mate away from their residences but that mobility is independent of density. Randall (1991), observing interactions at estrous females' mounds, documented some visits and one copulation by more distant males after the female mated with the nearest male. Radio-tracking and livetrapping data both indicate that males tend to move somewhat more than females (e.g., Jones 1984). On the other hand, there is evidence from spool-and-line tracking that estrous females actively visit and enter male mounds as much as 100 m away (Steinwald 2004).
We find it intriguing that in our GLMM analyses, the extent of multiple mating is influenced by how mobile the mother is, but not the father. Suppose females indeed mate with the nearest male with some probability, then find (or are approached by) the second-nearest male and mate with him with the same probability, and so on, but travel on average the same absolute distance no matter what the population density is. This scenario (implied by the negative exponential distribution of mother–father distances, Figure 1) is consistent with the stable relationship of proximity to paternity across years. Even if the scenario is not taken literally, the density-independent 60-m separation between mothers and fathers is consistent with the possibility that females make mating forays or at least advertise their estrous status over a constant distance.
Discussions of the potential advantages of polyandry in vertebrate taxa commonly focus on attracting paternal investment and, especially for mammals, on deterring infanticide (van Noordwijk and van Schaik 2000; Wolff and Macdonald 2004). But we are unaware of any reports suggesting infanticide in heteromyid rodents, and there is no evidence of paternal care in banner-tailed kangaroo rats. Thus, the possibility that females gain a genetic advantage, and in particular an advantage related to genetic compatibility (Tregenza and Wedell 2000; Stockley 2003), becomes an attractive one. A scenario that is consistent with all our results is that males attempt to monopolize matings and that nearby males have an advantage in mate competition (especially if they are mobile) but females actively though unselectively seek matings from additional males. This is the genetic bet-hedging hypothesis of Stockley et al. (1993): females mate with multiple males when they cannot exert complete control over the quality of the first male they mate with. Given the highly philopatric nature of banner-tailed kangaroo rats, the first male a female contacts is likely to be a close relative, and it is easy to imagine that seeking additional, more distant mates would increase the mean quality—meaning genetic compatibility—of mates.
Our data cannot reject the genetic diversity hypothesis (Stockley et al. 1993) in that the number of loci required to predict genome-wide heterozygosity is far greater than available for these types of analyses (Mitton and Pierce 1980; DeWoody YD and DeWoody JA 2005). However, they confirm our earlier report that 40% of females have nearest male neighbors with r > 0.25 (Winters and Waser 2003). Thus, mating with the nearest male will often mean mating with a close relative, a problem that mating with additional, more distant males demonstrably reduces. We prefer the bet-hedging hypothesis to the best-male hypothesis because our randomization tests provide at best weak evidence that females assess kinship. Relatedness of a female to her neighbors drops rapidly with distance (Winters and Waser 2003), so that choosing randomly among more distant males generates offspring that are on average less inbred.
For inbreeding depression or other forms of genetic incompatibility to create advantages to multiple mating, the relationship between a female's fitness and the proportion of "costly" (in this case, closely related) males among her mates must be concave down (Lorch and Chao 2003). This is often interpreted to mean that there must be some mechanism by which sperm from more distantly related males are more likely to fertilize eggs or that females can discriminate against and quickly terminate investment in offspring fathered by incompatible mates (Tregenza and Wedell 2000; Lorch and Chao 2003; Stockley 2003). Whether such mechanisms exist in banner-tailed kangaroo rats is a question for further research, but we wonder also whether this condition must necessarily be met in a system like this. If the scenario we imagine is correct, females do not choose mates based on kinship per se, instead seeking only more distant matings. In a viscous population, however, this behavior should generate a tendency for later matings to be of higher quality.
Female kangaroo rats in general and banner-tailed kangaroo rats in particular have very short periods of behavioral receptivity, preceded several days earlier by vaginal opening and swelling of the perivaginal area (Pfeiffer 1960; Wilson et al. 1985; Randall 1991). The duration of the period during which females advertise impending receptivity is, however, highly variable, to the extent that it has proven difficult for human observers to predict behavioral estrus during captive breeding attempts (Daly et al. 1984; Roest 1991). It would be interesting to know whether the exact timing of estrus is as cryptic to male kangaroo rats as it is to humans; one might predict this to be true if it is to females' advantage to avoid reproductive monopolization. Other striking characteristics of the reproductive biology of this species include the large size of the testes and accessory structures and the presence of conspicuous copulatory plugs, traits that are clearly compatible with an evolutionary history of multiple mating and sperm competition (Kenagy and Trombulak 1986; Randall 1991).
Intense mate competition in banner-tailed kangaroo rats appears to have created a situation in which males gain a proximity advantage, so that male–male aggression persists even though females gain nothing by mating with competitively dominant males. At the same time, each female is tightly tied to one or a few mounds, making her easy for males to locate and defend. A male's proximity to a female seems unlikely to be related to his viability, and the situation is made worse for the female because of the philopatric nature of the species—mating with nearby males is likely to result in inbred offspring. Under the circumstances, mating with multiple males appears to be a simple and relatively effective female counterstrategy. We suspect that Stockley's genetic bet-hedging hypothesis is widely applicable to species like D. spectabilis in which females' foraging ecology leads them to occupy static locations that can be easily monopolized by males, especially if the species is also philopatric.
|
ACKNOWLEDGEMENTS |
|---|
We thank J. Busch, B. Keane, C. McCormick, M. Steinwald, B. Swanson, and J. Winters for persistence in genotyping; numerous members of the Waser Lab for livetrapping; J. Lucas for help with SAS; H. Sahm for help with data analysis; and J. Cooper, B. Keane, M. McColgin, J. Randall, M. Steinwald, D. Triant, S. Turner, R. Williams, and especially J. Busch, M. Elgar, and 2 anonymous reviewers for constructive comments on earlier versions of the manuscript. The work was funded in part by the National Science Foundation.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSResultsDiscussionREFERENCES |
|---|
Behrends P, Daly M, Wilson MI. (1986a) Aboveground activity of Merriam's kangaroo rats (Dipodomys merriami) in relation to sex and reproduction. Behaviour 96:210–26.
Behrends P, Daly M, Wilson MI. (1986b) Range use patterns and spatial relationships of Merriam's kangaroo rats (Dipodomys merriami). Behaviour 96:187–209.
Birkhead T. (2000) Promiscuity(Harvard University Press, Cambridge, MA).
Clutton-Brock TC. (1989) Mammalian mating systems. Proc R Soc Lond B Biol Sci 236:339–72.[Medline]
Daly M, Wilson MI, Behrends P. (1984) Breeding of captive kangaroo rats, Dipodomys merriami and Dipodomys microps. J Mammal 65:338–41.[CrossRef]
Daly M, Wilson MI, Behrends P, Jacobs LF. (1992) Sexually differentiated effects of radio transmitters on predation risk and behavior in kangaroo rats Dipodomys merriami. Can J Zool 70:1851–5.
Davis C, Keane B, Swanson BJ, Loew S, Waser PM, Strobeck C, Fleischer RC. (2000) Characterization of microsatellite loci in banner-tailed and giant kangaroo rats, Dipodomys spectabilis and Dipodomys ingens. Mol Ecol 9:642–4.[CrossRef][Medline]
DeWoody YD and DeWoody JA. (2005) On the estimation of genome-wide heterozygosity using molecular markers. J Hered 96:85–8.
Eisenberg JT. (1963) The behavior of heteromyid rodents. Univ Calif Publ Zool 69:1–114.
Emlen ST and Oring LW. (1977) Ecology, sexual selection, and the evolution of mating systems. Science 197:215–23.
Fedorka KM and Mousseau TA. (2002) Material and genetic benefits of female multiple mating and polyandry. Anim Behav 64:361–7.[CrossRef]
Goodnight KF and Queller DC. (1999) Computer software for performing likelihood tests of pedigree relationship using genetic markers. Mol Ecol 8:1231–4.[CrossRef][Medline]
Gowaty PA. (2005) Beyond extra-pair paternity: constraints, fitness components, and social mating systems. In Lucas JR and Simmons LW (Eds.). Essays in animal behaviour(Elsevier/Academic Press, San Diego, CA) pp. 221–54.
Jones WT. (1987) Dispersal patterns in kangaroo rats Dipodomys spectabilis. In Chepko-Sade BD and Halpin ZT (Eds.). Mammalian dispersal patterns: the effects of social structure on population genetics(University of Chicago Press, Chicago) pp. 119–27.
Jones WT. (1984) Natal philopatry in banner-tailed kangaroo rats. Behav Ecol Sociobiol 15:151–5.
Jones WT. (1989) Dispersal distance and the range of nightly movements in Merriam's kangaroo rats. J Mammal 70:27–34.[CrossRef]
Kenagy GJ and Trombulak SC. (1986) Size and function of mammalian testes in relation to body size. J Mammal 67:1–22.[CrossRef]
Koprowski JL. (1998) Conflict between the sexes: a review of social and mating systems in the tree squirrels. In Steele MA, Merritt JF, Zegers DA (Eds.). Ecology and evolutionary biology of tree squirrels(Virginia Museum of Natural History, Martinsville, VA) pp. 33–42.
Lee PLM and Hays GC. (2004) Polyandry in a marine turtle: females make the best of a bad job. Proc Natl Acad Sci USA 101:6530–5.
Littell RC, Milliken GA, Stroup WW, Wolfinger RD. (1996) SAS system for mixed models(SAS Institute, Cary, NC).
Lorch PD and Chao L. (2003) Selection for multiple mating in females due to mates that reduce female fitness. Behav Ecol 14:679–86.
Marshall TC, Slate J, Kruuk LEB, Pemberton JM. (1998) Statistical confidence for likelihood-based paternity inference in natural populations. Mol Ecol 7:639–55.[CrossRef][Medline]
Miller GL and Carroll BW. (1989) Modeling vertebrate dispersal distances: alternatives to the geometric distribution. Ecology 70:977–86.[CrossRef]
Mitton JB and Pierce BA. (1980) The distribution of individual heterozygosity in natural populations. Genetics 95:1043–54.
Morrissey MB and Wilson AJ. (2005) The potential costs of accounting for genotypic errors in molecular parentage analyses. Mol Ecol 14:4111–21.[CrossRef][Medline]
Orians GH. (1969) On the evolution of mating systems in birds and mammals. Am Nat 103:589–603.[CrossRef]
Pfeiffer EW. (1960) Cyclic changes in the morphology of the vulva and clitoris of Dipodomys. J Mammal 41:43–8.[CrossRef]
Randall JA. (1984) Territorial defense and advertisement by footdrumming in bannertail kangaroo rats (Dipodomys spectabilis) at high and low population densities. Behav Ecol Sociobiol 16:11–20.
Randall JA. (1987) Observations of male competition and mating in Merriam's and bannertail kangaroo rats. Am Midl Nat 117:211–3.[CrossRef]
Randall JA. (1991) Mating strategies of a nocturnal, desert rodent (Dipodomys spectabilis). Behav Ecol Sociobiol 28:215–20.
Randall JA. (1993) Behavioural adaptations of desert rodents (Heteromyidae). Anim Behav 45:263–87.[CrossRef]
Randall JA, Hekkala ER, Cooper L, Barfield J. (2002) Familiarity and flexible mating strategies of a solitary rodent, Dipodomys ingens. Anim Behav 64:11–21.[CrossRef]
Roest AI. (1991) Captive reproduction in Heermann's kangaroo rat, Dipodomys heermanni. Zoo Biol 10:127–38.[CrossRef]
Schroder GD. (1979) Foraging behavior and home range utilization of the bannertail kangaroo rat (Dipodomys spectabilis). Ecology 60:657–65.[CrossRef]
Schwagmeyer PL. (1995) Searching today for tomorrow's mates. Anim Behav 50:759–67.[CrossRef]
Skvarla JL, Nichols JD, Hines JM, Waser PM. (2004) Modeling interpopulation dispersal by banner-tailed kangaroo rats. Ecology 85:2737–46.
Solomon NG and Keane BN. Reproductive strategies in female rodents. In Wolff JO and Sherman PW (Eds.). Rodent societies(University of Chicago Press, Chicago) Forthcoming.
Steinwald MC. (2004) Following the bannertail: a study of home range utilization and the mating system of Dipodomys spectabilis via spool-and-line tracking (MS Thesis)(Purdue University, West Lafayette, IN).
Stockley P. (2003) Female multiple mating behaviour, early reproductive failure and litter size variation in mammals. Proc R Soc Lond B Biol Sci 271–8.
Stockley P, Searle PJ, Macdonald DW, Jones CS. (1993) Female mating behaviour in the common shrew as a strategy to reduce inbreeding. Proc R Soc Lond B Biol Sci 254:173–9.[Medline]
Tregenza T and Wedell N. (2000) Genetic compatibility, mate choice and patterns of parentage: invited review. Mol Ecol 9:1013–27.[CrossRef][Medline]
van Noordwijk MA and van Schaik CP. (2000) Reproductive patterns in mammals: adaptations against infanticide? In van Schaik CP and Janson CH (Eds.). Infanticide by males and its implications(Cambridge University Press, Cambridge, UK) pp. 322–60.
Waser PM and Ayers JM. (2003) Microhabitat use and population decline in banner-tailed kangaroo rats. J Mammal 84:1031–45.[CrossRef]
Waser PM, Busch JD, McCormick CR, DeWoody JA. (2006) Parentage analysis detects cryptic pre-capture dispersal in a philopatric rodent. Mol Ecol 15:1929–37.[CrossRef][Medline]
Waser PM and Elliott LF. (1991) Dispersal and genetic structure in kangaroo rats. Evolution 45:935–43.[CrossRef]
Waser PM and Jones WT. (1991) Survival and reproductive effort in banner-tailed kangaroo rats. Ecology 72:771–7.[CrossRef]
Wilson MI, Daly M, Behrends P. (1985) The estrous cycle of two species of kangaroo rats (Dipodomys microps and D. merriami). J Mammal 66:726–32.[CrossRef]
Winters JB and Waser PM. (2003) Gene dispersal and outbreeding in a philopatric mammal. Mol Ecol 12:2251–9.[CrossRef][Medline]
Wolff JO and Macdonald DW. (2004) Promiscuous females protect their own offspring. Trends Ecol Evol 19:127–34.[CrossRef][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||