No evidence of inbreeding avoidance or inbreeding depression in a social carnivore

Behavioral Ecology Vol. 7 No. 4: 480-489

No evidence of inbreeding avoidance or inbreeding depression in a social carnivore Brian Keane," Scott R. Creel,b and Peter M. Waser0

a

Department of Biological Sciences, University of Cincinnati, Rieveschl Hall, Cincinnati, OH 45221, USA, bRockefeller University Field Research Center, Millbrook, NY 12545, USA, department of Biological Sciences, Purdue University, West Lafayette, IN 47907, USA

M

uch uncertainty exists regarding the extent and effect of inbreeding in wild populations of mammalian carnivores. Long-term behavioral studies of lions, spotted hyenas, and African wild dogs indicate that close inbreeding (between parents and offspring or siblings) is rare. Among these carnivores, as in other mammals, dispersal by young animals is widely viewed as the principal means by which matings by close relatives are avoided (Frame et al., 1979; Packer and Pusey, 1993; Smale et al., 1993). On the other hand, dispersal does not appear to prevent father-daughter mating in black bears (Rogers, 1987), while copulations outside the group apparendy allow outbreeding without dispersal in aardwolves, brown hyenas, and Ethiopian wolves (Gotelli et al., 1994; Mills, 1989; Richardson, 1987). The frequency of inbreeding in wild gray wolf populations is a matter of debate (Laikre and Ryman, 1991), and for most wild carnivore species, inbreeding rates are unknown (Rails et al., 1986). Close inbreeding has been shown to reduce offspring fitness in captive gray wolves (Laikere and Ryman, 1991), and correlational evidence links lack of genetic variation to a variety of reproductive problems in wild cheetahs (O'Brien et al., 1985) and lions (Wildt et al., 1987). But evidence of inbreeding depression in captive tigers, black bears, leopards, maned wolves, and other captive carnivore species is mixed (Lacy, 1993; Rails et al., 1988; Rogers, 1987). Lack of inbreeding depression in a population suggests a history of inbreeding, if the primary cause of depression is the presence of deleterious recessive alleles, rather than heterosis (Brewer et al., 1990; Lacy, 1993). Measures of inbreeding depression from wild carnivore populations are not generally available. Indirect evidence suggests decreased offspring production in inbred wild lions (Packer et al., 1991), but wild cheetahs show no evidence of elevated mortality caused by genetic problems (Caro and Laurenson, 1994). Shields (1982) has argued that natal philopatry, which is common in carnivores (Waser and Jones, 1983), fosters population subdivision and inbreeding, a view that is consistent with the relatively small estimates of effective population size Received 2 June 1995; accepted 21 January 1996. 1045-2249/96/55.00 O 1996 International Society for Behavioral Ecology

for several mammalian carnivores (Chepko-Sade et al., 1987) and with the lack of allozyme variation reported in a number of species (Kennedy et al., 1990; Patkau and Strobeck, 1994; Simonson, 1982). Pedigrees available for wild carnivores are generally not extensive enough to detect anything less than "close" inbreeding, mating between parents and offspring or between full siblings, but prolonged inbreeding between less closely related individuals can result in a highly inbred population (Templeton, 1987). Even when extensive pedigrees exists, diey must be viewed with caution because of the difficulty in accurately determining paternity from behavioral observations alone. Coupling long-term behavioral studies with analysis of genetic markers has the potential to gready strengthen inferences on the prevalence of inbreeding in natural carnivore populations (Packer and Pusey, 1993). Dwarf mongooses, HelogaU parvula, are small carnivores diat live in cohesive packs of 2-21 animals (Rood, 1980,1990). In general, only the oldest, most dominant pair in each pack produce young; reproduction by younger subordinates is suppressed bodi behaviorally and endocrinologically (Creel et al., 1991a,b, 1993; Rasa, 1973, 1987; Rood, 1980). Nevertheless, genetic analyses reveal diat subordinates of both sexes occasionally reproduce (Keane et al., 1994). Characteristic of relatively open savanna and woodland habitats in sub-Saharan Africa, dwarf mongooses are the most common carnivore in the Serengeti National Park, Tanzania (Waser et al., 1995), where a large population has been marked, observed, and censused since 1974 (Rood, 1980, 1987, 1990). These characteristics allow genetic, demographic, and behavioral investigations of die extent and consequences of mongoose inbreeding. Specifically, they allow us to use both pedigrees and multilocus DNA fingerprinting to estimate how closely mates are related, to determine whether patterns of mating widiin packs or of dispersal between packs reduce inbreeding, and to estimate die magnitude of several components of inbreeding depression. METHODS Study population

The population of dwarf mongooses we studied occupies at 25 + km! area of Acada-Commiphora woodland near the San-

Downloaded from http://beheco.oxfordjournals.org/ by guest on September 9, 2015

Dispersal by young mammals away from their natal site is generally thought to reduce inbreeding, with its attendant negative fitness consequences. Genetic data from the dwarf mongoose, a pack-living carnivore common in African savannas, indicate that there are exceptions to this generalization. In dwarf mongoose populations in the Serengeti National Park, Tanzania, breeding pairs are commonly related, and close inbreeding has no measurable effect on offspring production or adult survival. Inbreeding occurs because average relatedness among potential mates within a pack is high, because mating patterns within the pack are random with respect to the relatedness of mates, and because dispersal does little to decrease the relatedness among mates. Young females are more likely to leave a pack when the dominant male is a close relative but are relatively infrequent dispersers. Young males emigrate at random with respect to the relatedness of die dominant female and tend to disperse to packs that contain genetically similar individuals. Key words: carnivores, dispersal, dwarf mongoose, inbreeding avoidance, inbreeding depression, mate choice. [Behav Ecol 7:480-489 (1996)]

Keane et al. • Mongoose inbreeding, mate choice, and dispersal

Figure 1

gere River in the Serengeti National Park, Tanzania (Figure 1). Jon Rood first marked animals in this area, using individually distinctive freeze-brands (Rood and Nellis, 1980), in 1974. By 1977, most individuals were recognizable. Rood marked yearlings and any unmarked immigrants, and censused the entire population, at the beginning of each breeding season (November). During most years, he (and more recently, Scott Creel, Nancy Creel, and Lee Elliott) followed 10+ central packs throughout the entire breeding season, recording matings, pregnancies, litter sizes, and juvenile survival, as well as dispersal events. Our demographic data come from the resulting 1545 mongoose-years of data, involving 776 marked yearlings and adults and 177 litters. During 1988-1990, we additionally collected muscle biopsies (10-20 mg from the quadriceps, under metofane anesthesia) from 108 individuals, comprising nearly all members of 10 different packs during two successive breeding seasons, for DNA analyses.

were rare, and we knew both putative parents for nearly all juveniles born during the study. Of 513 yearlings whose parents were known, we knew at least one grandparent for 242 (47%) and at least on great-grandparent for 152 (30%). A few pedigrees were five generations deep. In studies of this sort, it has become common to base estimates of Wright's coefficient of relationship ron all available pedigree information, even though different individuals have pedigrees of different depths and one's estimate of r increases as the pedigree gets deeper (Hoogland, 1992). We have followed this practice. In our pedigrees, we have 3740 pairs of animals for which all four parents are known. For this sample, our estimates of r are linearly related to estimates based on one-generation pedigrees (slope = 1.01 ± 0.01, not significandy different from one) but, as expected, are slighdy higher (intercept=0.05 ± 0.02, t = 27.13, p < .001). We report inbreeding coefficients of offspring as one-half the r of their parents (Wright, 1922).

Pedigrees From census and reproductive data, we constructed pedigrees based on the assumption that all young in a pack are the offspring of the dominant pair at the time of conception (Raps, 1987; Rood, 1980; the validity of this assumption is examined below using a genetic analysis of parentage). When observers were absent at conception, and a dominant individual had turned over since the most recent prior observations, that parent was scored as unknown. Such gaps in the data

Genetic techniques DNA extraction, digestion, Southern transfer and hybridization with radiolabelled human minisatellite probe 33.6 have been described previously (Keane et al., 1994). The probe detected a mean (±SE) of 16.5 ± 0.38 scorable fragments per individual. We ran DNA from each animal adjacent to a lane containing BstE II digested lambda DNA that served as a size standard. Gels were scanned and analyzed using the NCSA

Downloaded from http://beheco.oxfordjournals.org/ by guest on September 9, 2015

Pack home ranges and successful dispersal within the main study area, 1974-1990. The major drainage is the Sangere River, just NE of the Serengeti Wildlife Research Centre. Home ranges of all packs detected in this area during the study are shown, but home range boundaries are approximate. Shaded home ranges are those belonging to packs whose members were biopsied. Arrows are proportional in width to the number of animals that dispersed between the packs they connect When more than one animal dispersed, n's are given at the tips of the arrows, but the animals they represent did not necessarily disperse together. Besides the dispersal events shown, 55 animals immigrated into these packs from unknown locations, and 10 emigrants were found in other packs off the mapped area.

Behavioral Ecology Vol. 7 No. 4

482

GelReader 2.0 program (NCSA, University of Illinois), which uses relative mobilities to estimate the molecular weight of DNA fragments of unknown size from known size fragments according to a method described by Elder and Southern (1983). We considered bands in two individuals' fingerprints to be shared if dieir molecular weight estimates, based on their mobility relative to that of die standards, differed by less than three standard deviations (Baird et al., 1986). We used proportions of bands shared (Wetton et al., 1987) to measure genetic similarity between individuals.

r

O

0.0

0.2

0.4

0.6

0.8

Coefficient of relationship Figure 2

Bandsharing as a function of Wright's coefficient of relatedness, r, as determined from pedigree analysis. See text for regression equauon. Each point represents the mean proportion of bands shared among individuals with a particular value of r. Sample sizes are r = 0, n = 4070 pairs; r = .063, n = 38; r = .125, n = 162; r = .188, n = 30; r = .250, n = 216; r = .375, n = 299; r = .5, n = 306; T = .625, n = 15; r = .75, n = 28.

bination widi odier independent variables, we used stepwise logistic regression (SAS Proc LOGIST, SAS Institute, 1990). The logic of diis procedure is analogous to diat of ordinary stepwise multiple regression when die dependent variable is binary. In all analyses, tests were two-tailed and we considered alpha=0.05. Because our sample sizes for some comparisons are small, it is important to note that Type II errors are possible. Where we find no evidence for inbreeding depression, for example, it is possible diat weak inbreeding depression exists. Similarly, where we find no evidence diat considerations of inbreeding influence male dispersal decisions, it remains possible diat larger sample sizes would detect a weak effect. Pedigrees and bandsharing Theory predicts die relationship between bandsharing and relatedness to be linear (Lynch, 1988), but some investigators have found a non-linear relationship between die two estimates (Gilbert et al., 1991; Jones et al., 1991). We calibrated bandsharing estimates against pedigree-based estimates of kinship (Buder et al., 1994; Haig et al., 1994; Piper and Parker, 1992) to determine die reliability widi which bandsharing reflects relatedness between individuals. Our calibration curve showed that bandsharing was closely and linearly related to our estimates of r as calculated from pedigrees based on die assumption diat all young in a pack are die offspring of die dominant pair at die time of conception (Figure 2; proportion of bands shared=0.32+0.22Xrelatedness, p = .001, R2 = 0.84). This result suggests diat bandsharing can be useful for discriminating closely related from unrelated individuals. However, as noted previously, multilocus fingerprinting of 45 offspring showed that 12% to 40% of young have subordinate fathers (95% confidence limits) and 6% to 30% have subordinate mothers (Keane et al., 1994). Consequendy, subordinate reproduction may affect the accuracy of die pedigree t's. Because we have genetic data only from 1987 to 1990, we could not correct entire genealogies for subordinate reproduction on a case-by-case basis. However, when we regressed die pedigree-based r values between offspring and dominants against die rvalues between offspring and dieir actual parents (as determined by DNA fingerprinting), we found diat die slope was not significandy different from one (Creel and Waser, 1994). As will be noted below, in almost all cases of subordinate parentage the r between offspring and dominants

Downloaded from http://beheco.oxfordjournals.org/ by guest on September 9, 2015

Statistical analyses Estimates of mean bandsharing for within/between pack comparisons and for comparisons between classes of pack members of different social status (e.g., dominant versus subordinate) were based on die adult members of each of die 10 biopsied packs during die 1990 breeding season, die last year during which DNA samples were collected. Bandsharing estimates between dispersing individuals and members of die packs diey left, joined, and did not join were based on only diose adult pack members present at emigration (packs left) or immigration (packs joined or did not join). To avoid artificially inflating sample sizes for statistical analysis by using die same individual in many comparisons, we used group means while comparing bandsharing widiin/between packs, between classes of pack members of different social status, and between dispersing individuals and members of die packs diey left and joined. Since estimates of bandsharing similarity may be biased if they include pairwise comparisons diat are not independent, we used die mediod developed by Lynch (1990) to calculate an unbiased estimate of die variance in bandsharing values before statistical analysis. We compared mean bandsharing values by ANOVA, using die statistical program STATVIEW (Abacus Concepts, 1992) after testing for normality and homoscedasticity. All comparisons met the assumption of homoscedasticity, and nearly all bandsharing samples (25 of 30) met die assumption of normality. Where diey did not, we tried to transform die data to meet parametric assumptions; failing diis, we used a KruskalWallis one-way analysis of variance. In all cases, the results of die non-parametric analysis, or die ANOVA results, on transformed data were identical to the results on untransformed data, as expected since ANOVA is relatively robust to mild deviations from normality (Sokal and Rohlf, 1981). For simplicity, we dierefore report only die ANOVA results based on untransformed data. Where die overall F was significant, we calculated post hoc ("protected") t values for pairwise comparisons (Zar, 1984). We used SYSTAT (Wilkinson, 1990) to perform linear regressions. To examine litter size as a function of die inbreeding coefficient of parents, we first used SYSTAT to regress litter size against group size, calculating residual litder sizes. We dien regressed diese against inbreeding coefficients. We followed die same procedure to determine die relationship between die number of litters produced by a pair per year and die pair's inbreeding coefficient. We also used SYSTAT to compare means, for example when comparing litter sizes between inbred and oudsred parents. We used a t test for variables diat were normally distributed, a {/test odierwise. We tested whedier emigration tendencies were independent of die relatedness of potential mates using die log-likelihood test, or where samples were small, die Fisher exact test (Sokal and Rohlf, 1981). We also used diese tests for other contingency analyses, for example, to determine whedier die chances of subordinate parentage was independent of die relatedness of die dominant pair. To explore whedier emigration tendencies might be influenced by inbreeding in com-

0.5

Keane et al. • Mongoose inbreeding, mate choice, and dispersal

and the r between offspring and their actual parents was identical.

483

Table 1 Dispersal by subordinates as a potential inbreeding avoidance mechanism

RESULTS Is there evidence that mates are related?

Is emigration more likely when potential mates are close relatives?

Dominant, breeding dwarf mongooses, males and females, can retain their dominant status for up to 7 years (Rood, 1990). As a result, adult subordinates are often offspring of one or both of the dominants. Most males, and many females, emigrate from their natal pack during their first few years of life (Rood, 1987; Waser et al., 1994); 92% of males and 57% of females that become dominant breeders do so after dispersal (Rood, 1990). To determine whether subordinates emigrate to avoid inbreeding, we first determined whether a subordinate's dispersal decision was independent of its relatedness to its most likely mate (the opposite-sexed dominant) if it stayed in its pack. We considered only cases for which the relationship between the subordinate and the opposite-sexed dominant was known from pedigree data, and for which the subordinate's dispersal decision was unambiguous (that is, we excluded data from subordinates that simply disappeared). Forty-three percent of subordinate males dispersed during years when dominant females were their first-degree relatives (mothers or full sisters), compared with 39% when dominant females were less related (Table 1). There is no evidence, therefore, that subordinate males emigrated to reduce inbreeding (G = 0.12, p = .72). The results are the same if the analysis is confined to second-ranking subordinates. Beta males were no more likely to emigrate if die dominant female was a first-degree relative (47% versus 50%, p = 1.00, Fisher test). Females, however, were marginally more likely to disperse if the dominant male was a first-degree relative (18%) dian when he was not (10%, G = 4.02, p = .05, Table 1). Thirteen

Disappear

All subordinate males Dominant female is mother or sister Dominant female not mother or sister Beta males only Dominant female is mother or sister Dominant female not mother or sister All subordinate females Dominant male is father or brother Dominant male not father or brother Beta females only Dominant male is father or brother Dominant male not father or brother

70 14

53 9

88 14

9 4

8 4

10 0

149 123

33 14

47 59

14 44

2 0

5 16

Entries are die numbers of individual-years during which subordinates survived in their natal groups for die entire year ("stay") or during which subordinates were seen leaving their natal groups or were located later in another group ("emigrate"). The "disappear" column tallies the numbers of natal subordinates that disappeared under unknown circumstances: elsewhere (Waser et al., 1994) we show that roughly one-third of these animals probably died in their natal groups. The balance presumably emigrated when no observers were present and either died or settled off die study site. Since it seemed unlikely to us that die pattern of emigration versus relatedness would be influenced by whether we witnessed the emigration process or not, we report only analyses based on known emigrants in die text. When disappearing animals are scored as emigrants, all emigration "decisions" are independent of relatedness (G or Fisher tests, all p > .05).

percent of beta females left when the dominant male was a close relative, while none left when he was not (p = .07, Fisher test). Subordinate dispersal decisions may be influenced by many factors other than relatedness to the opposite-sexed dominant; age, pack size, "queue length" (the number of samesexed individuals that are older and, therefore, more likely to obtain dominant, breeding status should the same-sexed dominant disappear), and r to the same-sexed dominant (Creel and Waser, 1994). To determine how inbreeding-avoidance tendencies might have been associated with these other variables, we used stepwise logistic regression to analyze dispersal decisions. Our model included all the above variables, as well as squared terms to incorporate possible non-linear effects. We also included the possibility of two-way interactions among age, queue length, pack size, and both rvalues. As in our first analysis, logistic regression indicated that inbreeding avoidance influenced dispersal decisions in females, but not in males. For subordinate males, none of the independent variables or their interactions entered the (forwardstepping) model at p < .15. Subordinate males, on the other hand, were more likely to disperse if the dominant male was closely related (p = .002). A significant quadratic effect of r to the dominant male (p = .01) indicated that a female's tendency to emigrate was disproportionately strong if the dominant male was very closely related. The only other variable to enter the model for females was the interaction between age and queue length; females were more likely to disperse if they were older and in longer queues (p =. 03).

Downloaded from http://beheco.oxfordjournals.org/ by guest on September 9, 2015

An analysis of pedigrees, constructed assuming that the dominant pair were the parents of all offspring born in the pack, estimated r between dominants in the same pack to be 0.12 ± 0.02 (n = 86 litters). While most parents were only moderately related, close inbreeding did sometimes occur. Of 241 offspring whose parents' pedigrees were extensive enough to detect mating among first-degree relatives, 35 (14%) had inbreeding coefficients of at least 0.25. Eleven of these were apparently produced by matings between full siblings, 11 by father-daughter matings, and 13 by mother-son matings. Among 25 fingerprinted offspring for whose parents we had pedigrees, six were products of full sibling matings and two of father-daughter matings. Genetic evidence reinforces the conclusion from pedigrees that parents are often related. DNA fingerprinting allowed us to determine the parents of 45 offspring (Keane et al., 1994). We compared bandsharing among these parents to bandsharing among individuals we considered to be unrelated because we knew that they shared no common ancestors in the previous two generations. Parents determined by DNA fingerprinting shared almost half again as many bands (x± SE; 0.45 ± 0.03) as did non-relatives (0.31 ± 0.01; t = 3.65; df = 251; p < .0003). From the rates of bandsharing among fingerprinted parents, we estimated Wright's coefficient of relatedness, r, to be 0.19 ± 0.04 using the method of Lynch (1991). Both estimates of r between mates, that based on bandsharing and that based on pedigree analysis, indicate that moderate inbreeding is widespread in this population.

Emigrate

Stay

484

Bandsharing (mean ± SE) among all adult mongooses within the same pack (black), among adult mongooses from adjacent packs (white) and among adult mongooses from non-adjacent packs (hatched) in 1990. Trends are similar whether we considered bandsharing among all adults irrespective of sex (ANOVA, df = 2,52; F = 31.72; p < .0001); bandsharing among opposite-sexed individuals only (ANOVA, df = 2,51; F = 20.67; p < .001); or bandsharing only between opposite-sexed dominants (ANOVA, df = 2,77; F = 4.53; p < .05). An asterisk indicates a significant difference between means. Bars indicated pairwise values that differ significantly based on post hoc (protected) t tests at p < .05 (*), p < .01 (••), or p < .001 (*•*). Do emigrants join packs whose members are unrelated?

Even for male subordinates whose emigration tendency is uninfluenced by the presence of close, opposite-sexed relatives, dispersal might decrease inbreeding if emigrants join packs whose members are distantly related. Between 1974 and 1990, 160 mongooses immigrated into marked study packs, 95 (59%) from other packs on the study site (Figure 1). Between 1987 and 1990, 16 animals (all males) dispersed between fingerprinted packs, and eight animals immigrated into fingerprinted packs from unknown natal packs. Bandsharing data suggest that dispersal, even to immediately adjacent packs, has the potential to reduce inbreeding substantially (Figure 3). In 1990, adult members of adjacent packs shared a significandy lower proportion of bands (0.35 ± 0.01) than was shared between adults members of the same pack (0.47 ± 0.02). Adult members of non-adjacent packs were also significandy less similar genetically (0.32 ± 0.01) than members of the same pack. Bandsharing between adult members of adjacent packs was as low as that between adult members of non-adjacent packs. The same patterns were found if bandsharing values are measured for opposite-sexed individuals only (Figure 3). Opposite-sexed members of adjacent (0.35 ± 0.02) and non-adjacent (0.31 ± 0.01) packs shared a much lower proportion of bands than opposite-sexed members of the same pack (0.46 ± 0.03). Opposite-sexed members of adjacent and non-adjacent packs were equally similar to each other. Since most of the reproduction is by the dominant male and female of a pack, it may also be relevant to compare a

dominant animal's similarity to the opposite-sexed dominant in its own versus other packs (Figure 3). In 1990, dominants in the same pack shared significantly more bands (0.42 ± 0.04) than did opposite-sexed dominants in adjacent packs (0.32 ± 0.03) or non-adjacent packs (0.29 ± 0.02). These data indicate that, if dominant animals were to disperse even to adjacent packs, they would substantially decrease their genetic similarity to their mates. To determine whether dispersing males take advantage of the opportunity to join packs whose members are distandy related, we examined our sample of fingerprinted (male) dispersers and immigrants. Although Figure 3 demonstrates that the potential to reduce inbreeding by dispersing clearly exists, our bandsharing data provide no evidence that dispersing animals take advantage of it (Figure 4). Instead, the fingerprinted dispersers joined packs whose adult members shared significantly more bands with them (0.40 ± 0.03) than did members of packs they did not join (0.31 ± 0.02). In fact, dispersers chose packs whose adult members were as'similar to them as their natal packmates (0.40 ± 0.03). Furthermore, for this sample of males the process of dispersal did not reduce genetic similarity with females (Figure 4). Dispersers entered packs whose adult females shared as many or more bands with them (0.38 ± 0.03) as the females in their natal packs (0.35 ± 0.03). The same pattern occurs if only dominant females are considered: males were at least as similar to dominant females in the packs they joined (0.36 ± 0.03) as the packs they left (0.32 ± 0.04). Surprisingly, males appeared to pass up opportunities for outbreeding. Females in packs that males did not enter shared fewer bands with them (0.31 ± 0.02 for all females, 0.28 ± 0.03 for dominant females). Dispersing males also joined packs whose males resembled them (0.4O ± 0.02) almost as closely as males in their natal groups (0.45 ± 0.02) and more than males in packs they did not join (0.32 ± 0.02). The pattern was similar considering bandsharing between dispersers and dominant males alone (dominant males in natal packs, 0.39 ± 0.05; in new packs, 0.36 ± 0.04; in packs not entered, 0.29 ± 0.02; Figure 4). Does the mating pattern reduce relatedness between mates?

To determine whether patterns of mating within packs were random with regard to relatedness, we first examined the possibility that dominants are more likely to mate with subordinates when die opposite-sexed dominant is closely related. We then asked whether subordinates were less likely to become or stay dominant if they were close relatives of their probable mates. We could not detect any evidence that subordinate parentage is more likely when dominants are close relatives. Pedigree-based rvalues are available for both the true (fingerprinted) parents and die putative (dominant) parents of 23 young. Subordinate parentage was not significandy more likely in packs whose dominant male and female were first-degree relatives (4 of 7 fingerprinted young in such packs had at least one subordinate parent) dian in packs whose dominants were less closely related (6 of 16 young, Fisher p = .34). Moreover, when subordinate madng occurs, it does not markedly reduce inbreeding. Of 10 cases of subordinate parentage detected by DNA fingerprinting, eight had no effect on inbreeding. Subordinate parentage decreased the inbreeding coefficients of 2 of 7 offspring in packs whose dominants were first-degree relatives and of 0 of 16 offspring in packs with less related dominants. This difference is not significant (Fisher p= .08). Genedc evidence reinforces the conclusion that dominants do not decrease inbreeding by mating with subordinates. In

Downloaded from http://beheco.oxfordjournals.org/ by guest on September 9, 2015

Figure 3

Behavioral Ecology Vol. 7 No. 4

485

Keane et al. • Mongoose inbreeding, mate choice, and dispersal

05

•O 0.5

C 3 0.3

0.2

8.'

0.2

a- o.o O 0.1 CL 0.0

All adults

Adult females

Dominant females

Adult males

nun

Dominant males

packs with subordinate parentage, bandsharing between the true parents (0.43 ± 0.06) was slightly, though not significantly, greater than bandsharing between dominants in the same pack (0.39 ± 0.04; paired t test, df = 12, / = 0.55, p = .06). Figure 5 shows that, across all packs, dominants and parents shared virtually identical proportions of bands (parents, 0.45 ± 0.04; dominants, 0.42 ± 0.04). Another way that mating patterns could reduce inbreeding would be through the differential acquisition by unrelated subordinates of dominant, breeding positions. The best predictor of dominance among dwarf mongooses is age (Creel et al., 1991a). Is it possible, nevertheless, that the processes that lead a mongoose to become dominant are influenced by r, so that the dominant male and female are less related to each other than random members of the "pool" of possible mates in their pack? One process that could lead to non-random mating has been described: the increased tendency of subordinate female mongooses to emigrate when their most likely mate (the dominant male) is closely related. Figure 5 demonstrates, however, that despite the apparently selective emigration of some subordinate females, dominant male-dominant female bandsharing within a pack if as high as bandsharing between the dominant male and subordinate females. Dominant males share 0.44 ± 0.04 of their bands with beta female packmates and 0.38 ± 0.03 with lower-ranking females. A third way that one might imagine mating patterns influencing the chances of inbreeding is through female mate choice. Female preferences could influence which of several male subordinates acquires dominant status. Observations of female matings suggest that all females, including the alpha female, prefer to mate with the alpha male (Creel et al., 1991a, esp. Table IV). Again, however, Figure 5 shows that mate choice does not decrease inbreeding. Dominant females are as related to dominant males as they are to other subordinate males that might in principle have acquired dominant

^

Figure 5 Comparison of the proportion of bands shared (mean ± SE) between parents established by DNA fingerprinting (n = 9), the opposite-sexed dominants in the same pack (n = 7), the dominant female and beta male in the same pack (n = 8), the dominant male and beta female in the same pack (n = 7), the dominant female and all other adult male subordinates in the same pack (n = 15), and the dominant male and all other adult female subordinates in the same pack (n = 12). None of these bandsharing values differ significantly from each other (ANOVA, F = 0.39; df = 5,51; p = .85).

status. Dominant females share 0.44 ± 0.04 of their bands with beta male packmates and 0.41 ± 0.03 with lower-ranking males. Overall, bandsharing among dominants is comparable to that among random opposite-sexed packmates (0.41 ± 0.03). A final aspect of mating pattern that could potentially reduce inbreeding could be a tendency for dominants to abandon their mates more quickly if they were close relatives. The distribution of pair tenure lengths was strongly right-skewed for both related and unrelated dominants, with one or both members of most pairs turning over with less than a year and only a few pairs persisting for more than 3 years. Nevertheless, dominant animals did not turn over more rapidly if they were related and "inbred" dominant pairs persisted for just as long as unrelated pairs. Four dominant pairs that were first-degree relatives had a median tenure length of 6 months, while 12 pairs that were unrelated had a median duration of 5.5 months (Mann-Whitney U = 22, p = .81). Is there evidence for inbreeding depression? Litter counts revealed little cost to close inbreeding. Mongoose litters can first be counted at approximately 4 weeks, when they emerge from the natal den. At this age, we had counts of 63 litters whose parents were outbred (pedigreebased inbreeding coefncient=0) and 18 litters whose parents were first-degree relatives (full siblings or parent and offspring; inbreeding coefficient = 0.25). Litter sizes were slightly but not significantly lower when parents were first-degree relatives (2.1 ± 0.3 versus 2.7 ± 0.2, t = 1.29, df = 79, p = .20). Even this hint of a difference vanishes by the time litters are weaned. At age 3 months, when weaning usually occurs, 43 litters with unrelated parents had a mean size of 1.7 ± 0.2, while 17 litters whose parents were first-degree relatives had a mean size of 1.6 ± 0.3 (t = 0.18, df = 58, p = .86). Litter size at weaning is influenced significantly by pack size (Creel et al., 1991b; Rood, 1990), but statistically removing this effect did not reveal any evidence of inbreeding depression (Figure 6). Inbreeding depression might also be expressed through differential failure of inbred pregnancies or the loss of entire

Downloaded from http://beheco.oxfordjournals.org/ by guest on September 9, 2015

Figure 4 Bandsharing between male natal dispersers (including 16 animals that dispersed between fingerprinted packs and eight that immigrated into fingerprinted packs from elsewhere) and members of the packs they left, joined, and did not join. In all cases, but particularly with respect to females, males left groups whose members were genetically similar to them, but then joined other groups whose members were also genetically similar in preference to groups with less-related members. ANOVA showed significant differences for three of the five comparisons and approached significance for the other two. For bandsharing between dispersing males and all adult group members, ANOVA df = 2,60; F = 8.879, p = .0004. For bandsharing with adult females, df = 2,60; F = 3.812; p = .03. For bandsharing with dominant females only, df = 2,50; F= 3.125; p = .05. For bandshare with adult males, df = 2,59; F = 10.834; p = .0001. For bandsharing with dominant males only, df = 2,48; F = 3.053; p = .06. As in Figure 3, bars indicate the results of post hoc (protected) ttests.

Behavioral Ecology Vol. 7 No. 4

486

bO

0 0 —

0

o

I 06

0.00

0.10

0.20

_______^ o

1

Figure 6

Relationship between litter size at weaning and the inbreeding coefficients of juveniles (the inbreeding coefficient of a juvenile is half the relatedness of its parents, which was estimated from pedigrees). Litter sizes were recorded at 3 months (n =71). Since litter size is known to increase with pack size, we plot residuals of litter size at weaning after removing the effects of pack size. The slope of the regression of residual litter size on inbreeding coefficient is not significantly different from 0 (b = —1.06 ± 1.66, t = —0.64, p = .52). Curves are 95% confidence limits.

inbred littlers before our first counts. We checked for this possibility by counting the numbers of litters produced by packs that were under observation, and whose dominants did not turn over, throughout entire breeding seasons. Again, we found no evidence of inbreeding depression. Unrelated dominants produced 0-4 litters per year (median = 2, n = 19), while first-degree relatives produced 2-3 (median = 2, n = 4, U = 35, p = .80). We counted the numbers of yearlings produced in packs whose dominants did not turn over during the breeding season to determine the summed effects of inbreeding from conception to age 1. This measure, which should detect any tendency for related dominants to produce fewer litters, wean smaller litters, or produce offspring with lower survival between weaning and age 1, again detects no evidence of inbreeding depression. First-degree relatives produced 1-4 yearlings per year (median=2, n=4) while unrelated dominants produced 0-13 (median=2, n=41, [7=78.5, p=.89). Corrected for pack size, the number of yearlings produced per year remained unrelated to inbreeding coefficient (Figure 7). At ages 1-2, many mongooses disappear, and disappearance may reflect either death or emigration off the study site. Yearling and adult survival rates on the study site are, however, almost identical between inbred and outbred mongooses. Of yearlings with an inbreeding coefficient of 0.25, 3 of 18 (17%) survived to age 5; of outbred yearlings, 14 of 76 (18%) survived that long (Fisher p = .58). Because genetic data are available only for animals born late in the study, survival of animals with parentage confirmed through fingerprinting can be estimated only between the time of capture and age 16 months. In this sample, 4 of 5 offspring (80%) with inbreeding coefficients of 0.25 survived, as did 6 of 7 offspring (86%) with inbreeding coefficients of

1

0.00

0.30

Inbreeding coefficient

o

1

0.10

0.20

•

1

0.30

Inbreeding coefficient

Figure 7

Relationship between the number of yearlings produced per year and their inbreeding coefficients. We counted yearlings in packs whose dominants did not turn over during the previous breeding season (n = 44). As in Figure 6, we plot residuals of the number of yearlings in each pack and year after removing die effects of pack size. The slope of the regression of residual number of yearlings on inbreeding coefficient is not significantly different from 0 (b = -2.57 ± 5.88, / = -0.44, p = .66). Curves are 95% confidence limits.

0.125 and 2 of 7 outbred offspring (29%): The differences were not significant (comparing inbred and outbred, Fisher p = .12). Results are comparable if survival is examined as a function of parental bandsharing; 6 of 11 offspring (55%) whose parents shared more than the median number of bands survived to age 16 months, versus 7 of 11 offspring (64%) whose parents shared fewer bands (Fisher p = .5). DISCUSSION

Our analysis of long-term pedigree data, constructed assuming that the dominant pair were the parents of all offspring born in a pack, indicated moderate levels of inbreeding to be pervasive in this population of dwarf mongooses. Though it has recendy been shown that subordinates produce some offspring (Keane et al., 1994), the evidence of inbreeding suggested by the pedigrees was reinforced by our analysis of DNA fingerprints. Minisatellite bandsharing revealed genetic similarity between actual parents (determined by fingerprinting) to be very similar to that of putative parents from pedigrees and significandy higher dian bandsharing among non-relatives. Estimates of parental coefficients of relatedness, r, derived from pedigrees and from DNA fingerprints were comparable and indicate diat the average relatedness of parents is between diat of first cousins and half-siblings. However, our pedigree-based estimates of r are incomplete. In a population such as this with a history of inbreeding, they are likely to miss paths of common ancestry between mates. In addition, since Lynch's method of calculating relatedness from minisatellite bandsharing data is based on die proportion of bands shared between non-relatives, our "genetic" estimate of parental relatedness is also biased downward. Individuals assumed to be non-relatives according to two-generation pedigrees probably shared more distant common ancestors. For

Downloaded from http://beheco.oxfordjournals.org/ by guest on September 9, 2015

C/3

•

Keane et al. • Mongoose inbreeding, mate choice, and dispersal

Both of these mechanisms may operate in this population of dwarf mongooses, but the second seems particularly likely. Of the 16 dispersing males in our genetic sample, only two joined packs containing relatives that they could have encountered in their natal pack (unpublished data). On the other hand, Figure 1 indicates that most dispersal was between contiguous packs. Most clusters of dispersal events fell in areas linked by rock outcrops or strips of vegetation along watercourses, and it is easy to imagine that predation might restrict potential dispersal routes. Risk of predation is high for dwarf mongooses, especially for animals in small packs (Rasa, 1987; Rood, 1990), and more than half of mongooses that attempt dispersal probably die (Waser et al., 1994). Since subordinates do reproduce on occasion (Keane etal., 1994), matings involving subordinates could in principle decrease inbreeding within a pack. There was no evidence, however, that subordinate matings did or could reduce inbreeding in our study packs. In packs with subordinate parentage, bandsharing between the subordinate and its mate was not different from bandsharing between dominants in the same pack. Across all packs, dominant females shared similar proportions of bands with lower-ranking adult males (either "beta" or more subordinate) as they did with dominants. Similarly, dominant males were as similar genetically to subordinate females, either beta or more subordinate, as to dominant females. Thus, inbreeding in this population of dwarf mongooses seems to be an unavoidable result of the high average genetic similarity among adult pack members. The net effect of dispersal patterns between packs and mating patterns within packs is a syndrome of pervasive, mild inbreeding. Not surprisingly for a population with a history of inbreeding (Templeton, 1987), our analyses revealed litde cost to close inbreeding. Related dominants did not produce fewer litters, smaller litters, or fewer yearlings than unrelated mates. Adults judged to be inbred from pedigrees survived at rates identical to outbred adults. Juveniles with parentage known from DNA analyses survived at rates independent of their inbreeding coefficient. The lack of any evidence of inbreeding depression suggests that many generations of inbreeding have purged most of the deleterious recessive alleles from this population and that heterosis has not had a significant effect on fecundity and offspring survival. Theory suggests that animals will not necessarily evolve to avoid inbreeding, since avoiding inbreeding may have costs that outweigh the benefits (Bengtsson, 1978; Waser et al., 1986). In social carnivores, the costs of avoiding inbreeding include the risk of starvation or predation during dispersal, the risk of aggression during immigration into packs containing non-relatives, and the loss of indirect fitness. For Serengeti dwarf mongooses, these costs are now high. If they were equally high in the ancestors of present-day dwarf mongooses, restricted dispersal and repeated dispersal within local clusters of packs could have been favored despite the genetic cost of mating with relatives. If the cost to inbreeding in ancestral mongooses was caused by the unmasking of deleterious alleles, then continued inbreeding would have reduced the cost of inbreeding in succeeding generations. Lack of disassortatrve mating within packs would then reflect both a lack of opportunities for outbreeding and a lack of strong selection against inbreeding. In 1983, Cheney and Seyfarth suggested that non-random dispersal among primate groups could potentially elevate, rather than reduce, relatedness within a group. Our results confirm that their suggestion also applies to a pack-living carnivore. Mechanisms that could produce this result include selective transfer to groups containing siblings, to adjacent groups, or to particular groups linked by safe corridors. These patterns characterize many mammals. Thus the expectation

Downloaded from http://beheco.oxfordjournals.org/ by guest on September 9, 2015

these reasons, the absolute value of r between dwarf mongoose mates may be even higher than our estimates suggest. Dispersal in this population of dwarf mongooses seems ineffective in reducing inbreeding. We found no indication that subordinate male emigration was contingent on inbreeding risk. There was no evidence that second-ranking subordinates were more likely to acquire dominant status if they were unrelated to the dominant female or that dominants remained together longer if they were unrelated. Patterns of dispersal did not significantly reduce bandsharing between males and the dominant female. These findings are in marked contrast to the situation in lions (Packer and Pusey, 1993), hyenas (Smale et al., 1993), and wild dogs (Frame et al., 1979), in which members of the dispersing sex typically emigrate when potential mates are close kin. Dwarf mongooses are unusual among social carnivores in that both sexes commonly disperse (Rood, 1987). Since the sex providing more parental investment has more to lose from inbreeding (Waser et al., 1986), it is of interest that females do show a tendency to emigrate more often when potential mates are closely related. Nevertheless, this tendency is weak. Only the closest categories of relatives appear to provoke emigration, with females showing no tendency to avoid halfbrothers, uncles, or nephews. Moreover, the majority of females, even second-ranking females, remain in their natal packs even when the dominant male is a first-degree relative. Wolves, which like dwarf mongooses are cooperative breeders in which both sexes disperse, are another social carnivore in which individuals often fail to emigrate when the group contains opposite-sex breeders that are close kin (Mech, 1987). The majority of males and females that acquire dominant, breeding positions do so after dispersing from their natal pack. Bandsharing data indicate that mongooses could significantly reduce their genetic similarity to potential mates by dispersing even to neighboring packs. Nevertheless, dispersing mongooses do not appear to take advantage of this opportunity because they do not enter packs at random with respect to genetic similarity. Males that dispersed between biopsied packs emigrated disproportionately into packs whose members were genetically similar to them. Surprisingly, this trend was caused by emigrant males joining packs that contained genetically similar females, as well as males. Males apparently immigrate into packs where mates will be as closely related to them as females in their natal packs, even though packs with less-related females exisL There are at least two possible mechanisms by which male dispersal into genetically similar packs might arise. First, dispersing males could disproportionately join packs in which they have living relatives. Aggression toward immigrating mongooses might be less if the immigrants were recognized as relatives, and mortality during the process of immigration appears to be a substantial component of the cost of dispersal (Waser et al., 1994). In addition, indirect fitness is an important component of a subordinate mongoose's inclusive fitness. Emigrating to packs containing relatives would increase this fitness component, and dispersed subordinates indeed have non-trivial indirect fitness (Creel and Waser, 1994). That dispersing males particularly joined packs with closely related alpha males further increases indirect fitness. Second, there might be microgeographic barriers that favor dispersal between particular neighboring packs. For instance, some pairs of packs are separated by open areas within which dispersers might be highly vulnerable to predators, while others are linked by corridors containing rocky outcrops or dense vegetation, providing more cover, holes, or temporary den sites. Repeated use of such corridors, even if dispersal were infrequent, might elevate genetic similarity among particular pairs of packs.

487

488

Behavioral Ecology Vol. 7 No. 4

1994. Subordinate reproduction in dwarf mongooses. Anim Behav 47:65-75. Kennedy PK, Kennedy ML, Clarkson PL, Uepens IS, 1990. Genetic variability in natural populations of the gray wolf, Canis lupus. Can J Zool 69:1183-1188. Lacy RC, 1993. Impacts of inbreeding in natural and captive populations of vertebrates: implications for conservation. Persp Biol 36: We thank the National Geographic Society and the National Science 480-496. Foundation (BSR 88180040) for their support; the Serengeti Wildlife Laikre L, Ryman N, 1991. Inbreeding depression in a captive wolf Research Centre and Tanzania National Parks for permission to study (Canis lupus) population. Conserv Biol 5:33-40. mongooses; and Jon and Hazel Rood, Nancy Creel, and Lee Elliott Lynch M, 1988. Estimation of relatedness by DNA fingerprinting. Mol for help and encouragement. Dennis Minchella, Rick Howard, Jeff Biol Evol 5:584-599. Lucas, Brad Swanson, Cathy Mossmann, William Olupot, Steve Travis, and an anonymous reviewer provided helpful comments on earlier Lynch M, 1990. The similarity index and DNA fingerprinting. Mol versions of the manuscript All animal handling methods were apBiol Evol 7:478-484. proved by the Purdue University Animal Care and Use Committee. Lynch M, 1991. Analysis of population genetic structure by DNA fingerprinting. In: DNA fingerprinting: approaches and applications (Burke T, Dolf G, Jeffreys A, Wolff R, eds). Basel: Birkhauser, 113REFERENCES 126. Mech LD, 1987. Age, season, distance, direction, and social aspects of Baird M, Balazs I, Giusti A, Miyazaki L, Nicholas L, Wexler K, Kanter wolf dispersal from a Minnesota pack. In: Mammalian dispersal patE, GlassbergJ, Allen F, Rubinstein P, Sussman I, 1986. Allele freterns (Chepko-Sade BD, Halpin ZT, eds). Chicago: University of quency distribution of two highly polymorphic DNA sequences in Chicago Press; 55-74. three ethnic groups and its application to the determination of Mills MGL, 1989. The comparative behavioral ecology of the hyaenas. paternity. Am J Hum Genet 39:489-501. The importance of diet and food dispersion. In: Carnivore behavBengtsson BO, 1978. Avoiding inbreeding: at what cost? J Theor Biol ior, ecology and evolution (Gitdeman J, ed). Ithaca, New York: Cor73:439-444. nell University Press; 125-152. Brewer BA, Lacy RC, Foster ML, Alaks G, 1990. Inbreeding depression O'Brien SJ, Roelke ME, Marker L, Newman A, Winkler CA, Meltzer in insular and central populations of Peromyscus mice. J Hered 81: D, Colly L, Evermann JF, Bush M, Wildt DE, 1985. Genetic basis 257-266. for species vulnerability in the cheetah. Science 227:1428-1434. Buder MA, Templeton AR, Read B, 1994. DNA fingerprinting in SpePacker C, Pusey AE, 1993. Dispersal, kinship, and inbreeding in Afke's gazelle: a test for genetic distinctness, and the correlation berican lions. In: The natural history of inbreeding and outbreeding tween relatedness and similarity. Mol Ecol 3:355-361. (Thornhill NW, ed). Chicago: University of Chicago Press; 375-391. Caro TM, Laurenson ML, 1994. Ecological and genetic factors in conPacker C, Pusey AE, Rowley H, Gilbert DA, Martenson J, O'Brien SJ, servation: a cautionary tale. Science 263:485-486. 1991. Case study of a population botdeneck; Lions of the NgoronCheney DL, Seyfarth RM, 1983. Nonrandom dispersal in free-ranging goro Crater. Conserv Biol 5:219-230. vervet monkeys: social and genetic consequences. Am Nat 122:392Patkau D, Strobeck C, 1994. Microsatellite analysis of genetic variation 412. in black bear populations. Mol Ecol 3:489-495. Chepko-Sade BD, Shields WD, Berger J, Halpin ZT, Jones WT, Rogers Piper WH, Parker PC, 1992. Use of fragment-sharing estimates from LL, Rood JP, Smith AT, 1987. The effects of dispersal and social DNA fingerprinting to determine relatedness in a tropical wren. structure on effective population size. In: Mammalian dispersal patMol Ecol 1:69-78. terns (Chepko-Sade BD, Halpin ZT, eds). Chicago: University of Rails K, Ballou JD, Templeton A, 1988. Estimates of lethal equivalents Chicago Press; 287-321. and the cost of inbreeding in mammals. Conserv Biol 2:185-193. Creel SR, Creel NM, Wildt DE, Monfort SL, 1991a. Behavioural and Rails K, Harvey PH, Lyles AM, 1986. Inbreeding in natural populaendocrine mechanisms of reproduction suppression in Serengeti tions of bird and mammals. In: Conservation biology: the science dwarf mongooses. Anim Behav 43:231-245. of scarcity and diversity (Soule ME, ed). Sunderland, Massachusetts: Creel SR, Monfort SL, Wildt DE, Waser PM, 1991b. Spontaneous lacSinauer Associates; 35-56. tation is an adaptive consequence of pseudopregnancy. Nature 351: Rasa OAE, 1973. Intra-familial sexual repression in die dwarf mon660-662. goose, Hclogale parvula. Naturwissen 60:303-304. Creel SR, Monfort SL, Wildt DE, 1993. Aggression, reproduction and Rasa OAE, 1987. The dwarf mongoose: a study of behavior and social androgens in wild dwarf mongooses: a test of the challenge hystructure in relation to ecology in a small, social carnivore. AdvStud pothesis. Am Nat 141:816-825. Behav 17:121-163. Creel SR, Waser PM, 1994. Inclusive fitness and reproductive strateRichardson PRK, 1987. Overt cuckoldry in an apparendy monogagies in dwarf mongooses. Behav Ecol 5:339—348. mous mammal. S Afr J Sci 83:405—410. Elder JK, Southern EM, 1983. Measurement of DNA lengdi by gel Rogers LL, 1987. Factors influencing dispersal in die black bear. In: electrophoresis II: Comparison of methods for relating mobility to Mammalian dispersal patterns (Chepko-Sade BD, Halpin ZT, eds). fragment length. Anal Biochem 128:227-231. Chicago: University of Chicago Press; 75-84. Frame LH, Malcolm JR, Frame GW, van Lawick H, 1979. Social orRood JP, 1980. Mating relationships and breeding suppression in die ganization of African wild dogs (Lycaon pictus) on the Serengeti dwarf mongooses. Anim Behav 28:143—150. Plains, Tanzania, 1967-1978. Z Tierpsychol 50:225-249. Rood JP, 1987. Dispersal and intergroup transfer in die dwarf monGilbert D, Packer C, Pusey AE, Stephens JC, O'Brien SJ, 1991. Anagoose. In: Mammalian dispersal patterns (Chepko-Sade BD, Halpin lytical DNA fingerprinting in lions: parentage, geneuc diversity and ZT, eds). Chicago: University of Chicago Press; 85-103. kinship. J Hered 82:378-386. Rood JP, 1990. Group size, reproduction, and routes to breeding in Gotelli D, Sillero-Zubiri C, Applebaum GD, Girman D, Roy M, Garciadwarf mongooses. Anim Behav 39:566-572. MorenoJ, Ostrander E, Wayne RK, 1994. Molecular genetics of the Rood JP, Nellis D, 1980. Freeze marking mongooses. J Wild Manage most endangered candid, the Ethiopian wolf Canis simensis. Mol 44:500-502. Ecol 3:301-312. SAS Institute, 1990. SAS/STAT user's guide, version 6. Cary, North Haig SM, Ballou JD, Casna NJ, 1994. Identification of kin structure Carolina: SAS Institute. among Guam rail founders: a comparison of pedigrees and DNA Shields WM, 1982. Philopatry, inbreeding and the evolution of sex. profiles. Mol Ecol 3:109-119. Albany, New York: State University of New York Press. Hoogland JL, 1992. Levels of inbreeding among prairie dogs. Am Nat 139:591-602. Simonson V, 1982. Electrophoretic variation in large mammals. II. The red fox, Vulpes vulpa, the stoat, Mustela crminca, die weasel, Jones CS, Lessells CM, Krebs JR, 1991. Helpers-at-uie nest in EuroMustda nivalis, the polecat, Mustda putonus, the pine marten, pean bee-eaters (Merops apiaster): a genetic analysis. In: DNA finMaries martes, die beech marten, Maries foina, and the badger, gerprinting: approaches and applications (Burke T, Dolf G, Jeffreys Meles meles. Hereditas 96:299-305. A, Wolff R, eds). Basel: Birkhauser; 169-192. Keane B, Waser PM, Creel SR, Creel NM, Elliott LF, Minchella DJ, Smale L, Frank L, Holekamp KE, 1993. Ontogeny of dominance in

that carnivore populations, or populations of other mammals, are necessarily outbred should be viewed with caution, particularly where the risks associated with dispersal are likely to be high.

Downloaded from http://beheco.oxfordjournals.org/ by guest on September 9, 2015

Keane et al. • Mongoose inbreeding, mate choice, and dispersal

free-living spotted hyaenas: juvenile rank relationships with adult females and immigrant males. Anim Behav 46:467-477. Sokal RR, Rohlf FJ, 1981. Biometry. New York: WH Freeman Co. Templeton AR, 1987. Inferences on natural population structure from genetic studies on captive mammalian populations. In: Mammalian dispersal patterns (Chepko-Sade BD, Halpin ZT, eds). Chicago: University of Chicago Press; 257-272. Waser PM, Jones WT, 1983. Natal philopatry among solitary mammals. Q Rev Biol 58:355-390. Waser PM, Austad SN, Keane B, 1986. When should animals tolerate inbreeding? Am Nat 128:529-537. Waser PM, Creel SR, Lucas JR, 1994. Death and disappearance, estimating mortality risk associated with philopatry and dispersal. Behav Ecol 5:135-141. Waser PM, Elliott LF, Creel NM, Creel SR, 1995. Habitat variation and

489

mongoose demography. In: Serengeti II, dynamics, management, and conservation of an ecosystem (Sinclair ARE, Arcese P, eds). Chicago: University of Chicago Press; 426—447. Wetton JH, Carter RE, Parkin DT, Walters D, 1987. Demographic study of a house sparrow population by DNA fingerprinting. Nature 327:147-149. Wildt D, Bush M, Goodrowe GL, Packer C, Pusey AE, Brown JL,Joslin P, O'Brien SJ, 1987. Reproduction and genetic consequences of founding isolated lion populations. Nature 329:328-331. Wilkinson L, 1990. SYSTAT: the system for statistics. Evanston, Illinois: SYSTAT, Inc. Wright S, 1922. Coefficients of inbreeding and relationship. Am Nat 56:330-336. Zar JH, 1984. Biostatistical analysis, 2nd ed. Englewood Cliffs, New Jersey: Prentice-Hall International.

Downloaded from http://beheco.oxfordjournals.org/ by guest on September 9, 2015