Polygynandry, Extra-Group Paternity and Multiple-Paternity Litters In

Home , Polygynandry, Reproductive system

Molecular Ecology (2007) 16, 5294–5306 doi: 10.1111/j.1365-294X.2007.03571.x

Polygynandry,Blackwell Publishing Ltd extra-group paternity and multiple-paternity litters in European badger (Meles meles) social groups

HANNAH L. DUGDALE,*† DAVID W. MACDONALD,* LISA C. POPE† and TERRY BURKE† *Wildlife Conservation Research Unit, Department of Zoology, University of Oxford, Tubney House, Abingdon Road, Tubney OX13 5QL, UK, †Department of Animal and Plant Sciences, University of Sheffield, Western Bank, Sheffield S10 2TN, UK

Abstract The costs and benefits of natal philopatry are central to the formation and maintenance of social groups. Badger groups, thought to form passively according to the resource dispersion hypothesis (RDH), are maintained through natal philopatry and delayed dispersal; however, there is minimal evidence for the functional benefits of such grouping. We assigned parentage to 630 badger cubs from a high-density population in Wytham Woods, Oxford, born between 1988 and 2005. Our methodological approach was different to previous studies; we used 22 microsatellite loci to assign parent pairs, which in combination with sibship inference provided a high parentage assignment rate. We assigned both parents to 331 cubs at ≥ 95% confidence, revealing a polygynandrous mating system with up to five mothers and five fathers within a social group. We estimated that only 27% of adult males and 31% of adult females bred each year, suggesting a cost to group living for both sexes. Any strong motivation or selection to disperse, however, may be reduced because just under half of the paternities were gained by extra-group males, mainly from neighbouring groups, with males displaying a mixture of paternity strategies. We provide the strongest evidence to date for multiple- paternity litters, and for the first time show that within-group and extra-group males can sire cubs in the same litter. We investigate the factors that may play a role in determining the degree of delayed dispersal and conclude that the ecological constraints hypothesis, benefits of philopatry hypothesis, and life history hypothesis may all play a part, as proposed by the broad constraints hypothesis. Keywords: cervus, Extra Pair Copulation (EPC), mating system, microsatellites, Mustelidae, reproductive skew Received 4 April 2007; revision received 24 June 2007; accepted 5 September 2007

the environment, the minimum defendable territory that Introduction provides sufficient resources for a minimum social unit may Mammals exhibit a diverse array of mating systems, chara- also accommodate more individuals. Once social groups form, cterized by the ecological and behavioural opportunities the factors under which selection will favour natal philopatry for individuals to monopolize mates and the way in which (Waser & Jones 1983) are central to the maintenance of social mates are acquired (Emlen & Oring 1977). Breeding system groups (Macdonald & Carr 1989). Delayed dispersal may properties determine group composition (Ross 2001). occur due to ecological constraints on dispersing and breeding As individuals are expected to act in a way that maximizes elsewhere (Emlen 1982), benefits of philopatry (Stacey & their lifetime inclusive fitness (Hamilton 1964), understanding Ligon 1991), certain life history traits, such as a low rate of why individuals live in groups requires knowledge of the breeder mortality that increases habitat saturation (Arnold costs and benefits of group living. The resource dispersion & Owens 1998), or a combination of these factors (the broad hypothesis (RDH) (Macdonald 1983; Carr & Macdonald constraints hypothesis, Hatchwell & Komdeur 2000). 1986) proposes that when resources are spaced patchily in Social species that encounter a variety of environmental conditions may exhibit variation in their mating system Correspondence: Hannah L. Dugdale, Fax: +44 (0) 1865 393101. (Taylor et al. 2000). The European badger is a promising E-mail: [email protected] species for understanding such variation because it has a

© 2007 The Authors Journal compilation © 2007 Blackwell Publishing Ltd BADGER MATING SYSTEM 5295 widespread distribution from the British Isles to Japan and Our study combines parentage assignment and sibship from the southern Mediterranean to the Russian Arctic inference, a novel approach recommended by Garant & Circle. It exhibits large variation in social organization over Kruuk (2005), to test whether (i) badger social groups exhibit this range, living singly, or in pairs and defending territories plural breeding; (ii) females breed with extra-group males; in parts of mainland Europe, whereas in lowland England (iii) males have different paternity strategies; and (iv) multiple- badgers are typically group living (Johnson et al. 2000). Despite paternity litters occur. We discuss our results in relation to their sociality, there is minimal evidence of the functional the four proposed hypotheses for the occurrence of delayed significance of grouping in badgers (Johnson et al. 2004), dispersal. In particular, we ask: Is group living costly to whereas there is evidence of an ecological basis of group liv- badgers? Does extra-group paternity reduce inbreeding? ing that does not invoke cooperation (Johnson et al. 2001a, b). Do multiple-paternity litters provide fitness benefits? Do Badger social groups have been reported to vary in size delayed implantation and the potential for superfetation from 2 to 29 individuals (da Silva et al. 1994). One proposal, (conception during pregnancy, Yamaguchi et al. 2006) facilitate first advanced for badgers by Kruuk (1978) is that groups multiple-paternity litters? form according to the RDH (Macdonald 1983) and are maintained by natal philopatry (Cheeseman et al. 1988; da Silva Materials and methods et al. 1994). Dispersal is restricted (Pope et al. 2006), with only 20% of the badgers trapped in Wytham Woods at any given Study site and population demography time having ‘dispersed’ (i.e. resident in more than one group), with residence defined as trapped in the same group in two Our data come from Wytham Woods, Oxfordshire (51°46′N, consecutive trapping events, and at least one of two trapping 01°19′W), that cover 4 km2 and consist of deciduous events prior to that (D. W. Macdonald, C. Newman, C. D. woodland, surrounded by permanent pasture and mixed Buesching & P. J. Johnson, unpublished). Dispersal may be arable land (Kruuk 1978). Enclosing features potentially costly; females that dispersed failed to produce cubs (da limit badger movement in and out of the study area Silva et al. 1993) and reduced fecundity was associated with (Macdonald & Newman 2002). From June 1987 to November increased dispersal rates in a culled compared with a con- 2005, trapping events have usually been undertaken at trol population (Tuyttens et al. 2000). Costs of group living also least four times a year, in January, June, August and include low levels of cub productivity (Cheeseman et al. 1987; November (Macdonald & Newman 2002). The badger Cresswell et al. 1992; Woodroffe & Macdonald 1995; Rogers population ranged from 60 to 228 adults and 23–61 cubs et al. 1997; Macdonald & Newman 2002) and a decreased (1987–96) with the highest density being 44 badgers/km2 proportion of lactating females in a control vs. culled popu- (Macdonald & Newman 2002). lation (Tuyttens et al. 2000). Few benefits to badgers of group living have been identified (Johnson et al. 2004). Although Sample collection cooperative breeding has been suggested in badgers (Woodroffe 1993), this has not been confirmed. The number Badgers were sedated by an intramuscular injection of of nonbreeding females had a negative effect on mean approximately 0.2 mL/kg ketamine hydrochloride. Cubs litter size, after controlling for territory quality (Woodroffe were first trapped around 16 weeks of age; those judged & Macdonald 2000); however, longer-term benefits have to weigh ≤ 2 kg were considered too small for sedation not been investigated. and were released after a hair sample was plucked. On Badgers give birth once a year, around February, with initial capture, badgers were marked with a unique tattoo cubs remaining underground for the first 8 weeks. This, number, through which recaptures were identified. Badgers along with the presence of many potential parents, means were classified as cub or adult based on their size and that it is not possible to identify individuals that have bred tooth wear; they were then sexed. Tooth wear was graded successfully using conventional ecological methods. Our on a subjective scale of 1 (no tooth wear: white teeth, study uses microsatellite data to assign parentage in a long- pointed canines and unworn ridges on molars) to 5 term study in Wytham Woods, Oxford, UK. Microsatellites (extreme tooth wear: canines broken or missing and molars were also used to determine parentage in Woodchester Park, worn down to the dentine). Approximately 3 mL of blood Gloucestershire, UK, which is a similarly high-density was collected from the jugular vein of each badger using population (Carpenter et al. 2005). Prior to this, multiple a vacutainer containing EDTA, and mixed immediately. maternity and extra-group paternity had been shown Blood was transferred into two 1.5-mL microcentrifuge (Evans et al. 1989; da Silva et al. 1994; Domingo-Roura et al. tubes and frozen immediately at –4 °C. Additionally, 2003), but parentage could not be assigned to individuals. from June 2002 onwards, approximately 100 guard hairs Behavioural observations have previously provided limited were plucked from each badger, and stored at 4 °C in 80% evidence for a polygynandrous mating system (Johnson ethanol. Ear tissue samples were collected from five road 2001) and extra-group paternity (Christian 1995). kills.

Fig. 1 Flow chart of the cervus parentage assignment and colony sibship reconstruction rules. Cubs that were assigned to sibship groups were included in the population parameter estimates at ≥ 80% confidence. M, number of cubs assigned a mother; P,number of cubs assigned a father. N, number of cubs that followed that path: N1, with one parent inferred, and N2,with both parents inferred. *inferred parent only accepted if the inference was logical given the cervus assignments.

chromatogram data twice and checked for errors using the DNA extraction excel microsatellite toolkit 3.11 (Parks 2001). Individuals DNA was extracted from blood using a GFX genomic blood genotyped at fewer than 16 loci were excluded, except for DNA purification kit (Amersham Biosciences) or from a one cub in 2005 that was genotyped at 11 loci. minimum of 20 hairs with visible roots using a Chelex We tested for deviations from Hardy–Weinberg equilibrium protocol (Walsh et al. 1991). An ammonium acetate protocol (HWE), per locus within cohorts, and for linkage equilib- (Bruford et al. 1998) was used to extract DNA from 20 to rium, between pairs of loci within cohorts, with exact tests 50 mg of each road-kill tissue sample. (1000 dememorizations, 1000 batches and 1000 iterations) using genepop 3.4 (Raymond & Rousset 1995). Within each cohort candidate parents and cubs were tested separately Genotyping to reduce multigenerational effects. Additionally, we analysed We genotyped individuals for 22 polymorphic microsatellite a sample of one male and one female candidate parent loci described by Pope et al. (2006). Each 10-μL polymerase from each social group to reduce social structure effects. chain reaction (PCR) contained approximately 20 ng of False discovery rate (FDR) control for type I error corrections genomic DNA, 0.25 μm of each primer, 0.1 mm of each was used to generate adjusted P values, to account for dNTP, 1.5 or 2.5 (Mel101, Mel104, Mel108, Mel110, Mel112 multiple tests (Benjamini & Hochberg 1995). Candidate and Mel115) mm MgCl2 and 0.175 unit of Taq DNA poly- parent genotypes in each cohort were used to estimate the merase (Thermoprime Plus, ABgene), in 1× PCR buffer amount of power available to distinguish between indi- containing: 20 mm (NH4)SO4, 75 mm Tris-HCl pH 8.8, 0.01% viduals by generating PI(unbiased) (Paetkau et al. 1998) and PI(sib) (w/v) Tween. If Mel101 did not amplify, we performed a (Evett & Weir 1998; Waits et al. 2001) using gimlet 1.3.3 12.5-μL PCR that was identical to the 10-μL PCR, except (Valière 2002). that the concentrations of genomic DNA and Taq DNA polymerase were increased to c. 60 ng and 0.2 unit, respectively. Parentage analysis Mel105 and Mel106 were multiplexed together in a single PCR, as were Mel103 and Mel107, Mel111 and Mel113, We determined parentage within each cohort through a Mel112 and Mel115, Mel114 and Mel117, Mel1 and Mel12, likelihood based approach using cervus 3.0.1.8 (Kalinowski and Mel10 and Mel14. Amplification then followed the et al. 2007) and colony 1.2 (Wang 2004). cervus assigns the methods described by Pope et al. (2006). most likely parent, whereas exclusion methods may result PCR products were separated on either an ABI Prism 377 in more than one individual matching or exclusion of the DNA Sequencer or an ABI Prism 48-capillary 3730 DNA true parent, if errors are present or relatives are candidate analyser. We analysed data from the 377 sequencer using parents (Marshall et al. 1998). Analyses were run system- genescan 3.1 and genotyper 2.5 software, whereas atically according to strict rules (Fig. 1). Genotypes were genemapper 3.5 software (Applied Biosystems) was applied first analysed in cervus to assign a parent pair, or if a pair to the 3730 data. We ran one set of 95 samples on both could not be assigned then we assigned either maternity sequencers to ensure the compatibility of results. We scored or paternity alone. cervus enables the presence of relatives,

© 2007 The Authors Journal compilation © 2007 Blackwell Publishing Ltd BADGER MATING SYSTEM 5297 genotyping error, and the proportion of unsampled indi- for parentage analysis in year Z. If they were only trapped in viduals to be incorporated; however, only sampled individuals year Z-1 at setts belonging to one of the newly formed groups can be assigned as parents. Cubs that were not assigned a in year Z, then they were assigned to that group in year Z-1 mother and/or father using cervus were therefore included for paternity analysis in year Z. If they were trapped at setts in a sibship inference using colony. in year Z-1 that were assigned to different groups at the split in year Z, then they were assigned to both groups in year Z-1. When calculating the number of candidate parents Selection of candidate parents per social group as a parameter in the parentage analysis, Parentage was not known a priori; therefore, the power to individuals assigned to more than one group (mean = 22 ± 2 assign parentage from the whole population was low and males, and 19 ± 2 females per year) were counted in all we selected candidate parents using mark–recapture of the groups that they were assigned to, to provide a data and biological rules, following Carpenter et al. (2005). conservative estimate. Candidate mothers were reproductive females (aged two or more), present in the cub’s social group in the year when Parentage simulations the cub was born. Candidate fathers were all males older than one year and present in Wytham Woods in the year We ran cervus simulations for 100 000 cycles, using the before the cub was born, as due to delayed implantation allele frequencies of all genotyped badgers (Table S1, females conceive in the year prior to birth. To minimize Supplementary material) and yearly parameters (Table 1). error from badgers that were present but were not caught, We estimated the proportion of loci that were typed we included adults and yearlings for an extra 2 years, incorrectly by re-genotyping 5% of the population. Of the rather than 3 (Carpenter et al. 2005), after their date of last 823 single-locus genotypes compared, we observed three capture, if their death date was unknown. This is because allelic dropouts each at different loci and one false allele, 95% of the intertrap intervals were within 525 days (n = 6193). giving an estimate of 0.005 loci typed incorrectly. pedant Badgers last caught as a cub were only included for one 1.3 (Johnson & Haydon 2007) estimated a mean allelic extra year, as the greatest mortality is generally seen in cubs dropout rate per allele of 0.005 and per heterozygote of (Macdonald & Newman 2002). This reduced the possible 0.009, and a mean false allele rate per genotype of < 0.001. assignment of a full-sibling instead of the true parent as, on We entered the proportion of loci mistyped as 0.005, in both average, full-siblings produce higher log-likelihood values the simulations and likelihood calculations. than the true parent (Thompson 1976). Candidate fathers assigned to more than one social Sibship reconstruction group within a year were assigned to the group closest to the cub(s) they sired, in the paternity distance analysis. Cubs that were not assigned a mother and/or father using Badgers first trapped as adults with tooth wear of 4–5, were cervus, were included in a full- and half-sibling inference judged to be at least 2 years old, otherwise they were with all other cubs in their social-group-year, using colony. judged to be at least 1 year old (da Silva & Macdonald We based estimations on the allele frequencies of all geno- 1989). The social group in which a cub was trapped was typed badgers. The allelic dropout rate per allele, estimated considered its natal group. Cubs trapped in two social using pedant, was 0.024 for Mel101, 0.073 for Mel10 and groups (n = 25) had the candidate mothers from both 0.010 for Mel110, with the mean being 0.005, which we used social groups included in their maternity analysis. The for the other loci and for the other typing error rate. colony natal groups of nine cubs contained no candidate mothers, assigns sibships, taking into consideration any assigned according to the trapping data, and the social group of one parents (from the cervus analyses). colony assumes that cub was not recorded. All candidate mothers in the popu- unknown parents are not those that are already known lation that year were included in the maternity analyses (from the cervus analyses), and that one sex is polygamous for these cubs. Assigned mothers were accepted only if they and the other monogamous. We first tested for siblings of had not bred in another group that year. cubs that had one unknown parent. If, say, the mother was Candidate parents, trapped successively in different unknown and the father was known then we analysed all social groups, were recorded as resident in both groups for cubs in that social-group-year with the same father and a the intervening period (see Carpenter et al. 2005). If a social known mother specifying the identity of the known father, group split into two groups over this period, or during the while entering all mothers as unknown and monogamous. extra years that an individual was assumed present, then a If this did not resolve the mother, or there were no other candidate parent was assigned to both from the time of the cubs with the same known parent, then we analysed all the split during those intervening years. If a social group split cubs in that social group, specifying all known fathers, and in year Z, candidate fathers that were present but not all mothers as unknown and polygamous. If parentage was trapped in year Z-1 were assigned to both groups in year Z-1 still unresolved or for cubs with both parents unknown, we

R ¶ R the Proportion Proportion relatives male in population§ = it was 48.2%), whereas er of candidate parents in er of candidate parents 2 R number is given in Table 1), = 0.001, the offspring’s social group as social group the offspring’s P = 14.91, 5.0.8 (Queller & Goodnight 1989). The average Proportion Proportion female relatives/ social group§ 1,16 F and a mean 82% trappability over a year. ‡Number and a mean 82% trappability over year. present each year (the actual present group ( group opulation each year. §Mean numb opulation each year. relatedness ing of the candidate fathers in Proportion Proportion candidate fathers sampled‡ parents (Macdonald & Newman 2002) parents the number of candidate mothers in er of candidate parents present in the p present er of candidate parents tion or social group. ¶Estimated using tion or social group. No. of candidate fathers in population† s were estimated to represent 94% of the actual number estimated to represent s were Proportion Proportion candidate mothers sampled‡ given year that were candidate given year that were = to the offspr we used the average relatedness 1.2%). Therefore, 2 R No. of candidate mothers in population† s that year. †Trapping record †Trapping s that year. 16 loci, as a percentage of the actual numb 16 loci, as a percentage cub was significantly negatively affected by cub was significantly negatively affected = 0.669, divided by the mean number in popula P parentage simulations parentage = 0.19, 1,16 No. of candidate mothers per social group† F cervus Proportion Proportion loci typed* Parameters for the 198819891990 0.951991 0.951992 0.951993 0.971994 0.97 31995 0.98 51996 0.99 61997 0.99 71998 0.99 71999 0.99 72000 0.99 72001 47 0.99 82002 62 0.99 92003 73 0.98 92004 78 0.98 92005 77 0.98 8Mean 0.88 95 0.98 8 100 0.84 0.98 all of the genotyped badger 7*Calculated from 119 0.98 0.87 8 136 0.88 8 154 0.90 7 156 33 0.90 7 0.88 154 57 7 0.90 143 79 0.91 141 76 0.91 155 82 0.86 161 90 106 0.85 146 137 0.84 139 161 119 0.81 0.94 178 0.84 0.92 180 0.86 0.84 170 0.90 0.81 157 0.94 0.81 137 0.88 0.91 124 0.92 135 0.91 0.70 130 0.91 0.78 124 0.91 0.83 120 0.93 0.86 0.88 0.85 0.80 0.85 0.86 0.79 0.87 0.11 0.80 0.88 0.07 0.76 0.89 0.09 0.83 0.89 0.07 0.88 0.88 0.86 0.08 0.88 0.06 0.06 0.26 0.87 0.05 0.19 0.88 0.05 0.17 0.87 0.05 0.19 0.86 0.05 0.21 0.85 0.05 0.20 0.85 0.19 0.05 0.20 0.05 0.18 0.05 0.20 0.04 0.20 0.04 0.20 0.04 0.13 0.06 0.22 0.19 0.11 0.15 0.20 0.19 each social group minus one (the parent), minus one (the parent), each social group which we calculated from the percentage of the population in a the percentage which we calculated from genotyped for at least of candidate parents of the candidate mothers to relatedness for both male and female relatives of the offspring. for both male and female relatives Year not for the candidate fathers ( Table 1

Table 2 Number of mismatches observed between parents assigned in cervus and their offspring

No. of mismatches Trio ≥ 80% Trio ≥ 95% Mother–cub ≥ 80% Mother–cub ≥ 95% Father–cub ≥ 80% Father–cub ≥ 95%

0 374* 276 29 18 34 18 181382092 255160010 31910000 4300000 Total 532† 331 31 18 44 20

*72 cubs could be placed in more than one trio (i.e. mother–father–cub) with 0 mismatches (max = 8 trios) and 60 of these had one trio assigned at ≥ 80% confidence. †For 75 cubs, trios were not assigned with confidence, but one parent was assigned using cervus (see Fig. 1). entered all cubs in the social group, specifying all parents for all 22 microsatellite loci. Of the 735 badgers first trapped as unknown, to test for full-sib assignment to a cub with as cubs in 1988–2005, 630 (86%) were genotyped. Mel104 in both parents known. If the unknown parent could not be a 1993 and Mel110 in 1995 showed significant departure from parent of any other cubs in the social group then we assigned HWE in both the cubs and candidate parents (m = 22, it as an unknown reconstructed parent (using the colony α = 0.05, adjusted P = 0.050–0.002), after adjusting for estimated genotype), when calculating multiple-paternity multiple testing by FDR control (Benjamini & Hochberg litters and the number of breeders per social group. 1995). These samples did not depart from HWE when one male and one female adult from each social group, in that cohort, were analysed to reduce the effect of social structure. Statistical analyses Deviations from linkage equilibrium for a pair of loci were We used minitab 14 for the majority of the statistical seen in 5 out of the 18 cohorts, after FDR control, but these analyses; however, we used sas 9.1 to conduct Generalized were not the same pair of loci in each cohort. When one Linear Mixed Models (GLMM) with Kenward–Roger male and one female per social group were analysed (n =39 denominator degrees of freedom method (Littell et al. 2006). badgers), only Mel105 and Mel12 deviated from linkage GLMMs allow the fitting of fixed effects, random effects equilibrium, in 1995. All loci were therefore included in (that model variance between experimental units), and analyses that required HWE and independence of data repeated measures (that model covariance between measures points, as no pair of loci consistently deviated from linkage –13 –14 on the same experimental unit). Repeated measures recorded equilibrium. PI(unbiased) (2.43 × 10 ± 1.4 × 10 ) and PI(sibs) closer in time are likely to be more correlated than those (2.94 × 10–6 ± 8.9 × 10–8) indicated a high probability of further apart in time (Littell et al. 1998). We therefore distinguishing individuals. examined graphically the covariance between pairs of Both parents were assigned to 595 (94%) cubs at ≥ 80%, observations on the same experimental unit at different and 331 (53%) at ≥ 95% confidence. Paternity was assigned times to determine the most appropriate covariance structure. to a further 16 and 7 cubs, resulting in 611 (97% at ≥ 80%) The GLIMMIX procedure was run with a Poisson error and 338 (54% at ≥ 95%) cubs that were assigned a father. distribution, log link, and social group as a repeated factor Maternity was assigned to an additional 7 and 5 cubs, with compound symmetry covariance structure, or as both resulting in a total of 602 (96% at ≥ 80%) and 336 (53% at a random factor, and a repeated factor with autoregressive ≥ 95%) cubs that were assigned a mother. A maximum of covariance structure (Littell et al. 2006). Where parametric three mismatches occurred between an assigned parent and tests were used, we tested differences for normality using cub (n = 2); however, 88% had no mismatches. Considering the Anderson–Darling test. We tested for homogeneity of trios, the maximum was four, with 70% having no mismatches group variances using Levene’s test. Means are given at ≥ 80%, and 83% at ≥ 95% confidence (Table 2). Parent-pair ±95% confidence interval, unless otherwise stated. assignment rates were generally lower than expected from simulations, when assigning the mother alone, father alone ≥ and parent pairs at 80% (t17 = –5.35, P < 0.001; t17 = –2.99, Results ≥ P = 0.008; t17 = –3.04, P = 0.007) and 95% (t17 = –2.32, We genotyped 915 (85%) badgers trapped and marked P = 0.033; t17 = –1.34, P = 0.199; t17 = –4.93, P < 0.001) confid- in Wytham Woods, Oxford, between 1987 and 2005. One ence, respectively. This may indicate the presence of more badger’s sample did not amplify and 165 were not sampled, unsampled individuals, more relatives, different allele the majority of which were only trapped once. Genotyping frequencies or a higher error rate than simulated (Marshall was 98% complete, with 77% (706) of the badgers genotyped et al. 1998; Kalinowski et al. 2007).

Fig. 2 Frequency distribution of the number of candidate mothers (white bars) and candidate fathers (black bars) per social-group-year (1988–2005). Data include adults for two years after their last capture date. If an individual was trapped in more than one social group within a year, it was split between these, unless it bred, in which case it was assigned to that social-group-year. Excludes social-group-years with no candidate.

Distribution of parentage The mean numbers of candidate mothers and candidate fathers per social-group-year were 5.6 ± 0.4 (median = 5) and 5.8 ± 0.4 (median = 5), respectively (Fig. 2). The number of candidate mothers or fathers per social-group-year increased as the density of candidate mothers or fathers increased over the study period (GLMM with Poisson- distributed error and autoregressive covariance structure:

F1,124.6 = 19.4, P < 0.0001; F1,303.6 = 29.7, P < 0.0001, respectively). The maximum number of mothers and fathers assigned within a social-group-year was 7 at ≥ 80% and 5 at ≥ 95% confidence (Fig. 3). The mean number of mothers per social-group-year was 1.9 ± 0.1 (n = 222, mode = 1) at ≥ 80% and 1.6 ± 0.1 (n =167, mode=1) at ≥ 95% confidence. The mean number of fathers per social-group-year was 1.9 ± 0.1 (n = 222, mode = 1) at ≥ 80% and 1.5 ± 0.1 (n = 167, mode = 1) at ≥ 95% confidence. A GLMM with Poisson-distributed error and compound symmetry covariance structure showed no significant difference in the numbers of mothers and fathers per social-group-year ≥ using the assignments at 80% (F1,418 = 0.00, P = 0.96) or ≥ 95% confidence (F1,311 = 1.03, P = 0.31). Analysing only those social-group-years in which all cubs were assigned a mother or a father, the mean number of mothers and fathers per social-group-year was 1.9 ± 0.2 and 1.9 ± 0.2 (n = 156 & 157, respectively, mode = 1 at ≥ 80%), and Fig. 3 The number of (a) females and (b) males assigned as parents within each social-group-year with an assigned parent(s). Grey bars 1.4 ± 0.2 and 1.3 ± 0.2 (n =51, mode=1 at ≥ 95% confi- represent parentage assignment at ≥ 80% confidence (n = 222), dence). Again, there was no significant difference in the and black bars at ≥ 95% confidence (n = 167). Fewer parents were numbers of mothers and fathers within a social-group-year assigned at ≥ 95% confidence, resulting in more social groups with ≥ ≥ ≥ (F1,280 = 0.20, P = 0.65, 80%; F1,76 = 0.36, P = 0.55, 95%). just one mother or father than with the 80% confidence assignments. The mean percentage of candidate mothers that bred was estimated as 28 ± 5% (range = 8–47%, median = 25%) at ≥ 80%, and 31 ± 6% (range = 8–51%, median = 28%) at mode = 1, n = 427 litters) at ≥ 80% and 1.3 ± 0.06 (range = 1–3, ≥ 95% confidence (Fig. 4). In contrast 25 ± 6% (range = 8–69%, mode = 1, n = 262) at 95% confidence. Considering just median = 23% at ≥ 80%), and 27 ± 6% (range = 8–60%, those social groups where all cubs were assigned a mother, median = 24% at ≥ 95%) of the candidate fathers bred. The the mean litter size was very similar: 1.4 ± 0.07 (range = 1–4, proportions of candidate fathers and candidate mothers mode = 1, n = 293 at ≥ 80%) and 1.3 ± 0.12 (range = 1–3, mode estimated to have bred each year were significantly different =1, n = 71 at ≥ 95%). Males assigned paternity to at least at both ≥ 80% (Z = –2.7, n = 18, P = 0.006) and ≥ 95% confi- one cub in a given year sired a mean of 1.6 ± 0.11 cubs at dence (Z = –2.9, n =18, P = 0.003). ≥ 80% (range = 1–9, mode = 1, n = 377 paternal litters) The mean postemergence litter size of females that were and 1.5 ± 0.12 cubs at ≥ 95% (range = 1–6, mode = 1, n = 221) assigned at least one cub was 1.4 ± 0.06 (range = 1–4, confidence.

Fig. 5 Number of paternities assigned according to the number of social-group boundaries that must have been crossed in order to gain paternity at ≥ 80% (grey bars, n = 585) and ≥ 95% (black bars, n = 338) confidence.

by extra-group males and six had both an extra-group and same-group father.

Patterns of paternity Fig. 4 The annual (a) number of candidate parents and cubs, and (b) estimated percentage of candidate parents that bred Fathers from known social groups were assigned to 569 (extrapolating from the number of cubs assigned parents and the (≥ 80%) and 331 (≥ 95% confidence) cubs that had a mother number of parents to which they were assigned, in comparison to from a known social group in the year of conception, and the total number of cubs and the total number of candidate parents in each year). Although the number of cubs assigned a parent was around half of these fathers were extra-group males (50% lower at the ≥ 95% than the ≥ 80% confidence level, if the mean and 42%, respectively). Neighbouring males gained 74% number of cubs assigned to each parent was lower at the ≥ 95% (37% of all paternities at ≥ 80%), and 86% (36% of all than the ≥ 80% confidence this led to a higher estimated paternities at ≥ 95% confidence) of the assigned extra-group proportion of candidate parents that bred at the stricter confidence paternities (Fig. 5). The mean distance between the main level. The figure includes individuals that were not genotyped. sett of a cub’s social group and its father’s social group in The dip in cub number in 2002 is potentially a result of the dry the year of conception was 214 ± 26 m (range = 0–1976 m conditions during that cub-rearing season. at ≥ 80%) and 155 ± 26 m (range = 0–1657 m at ≥ 95% confidence). Examining only extra-group paternities, the mean distance was 440 ± 40 m at ≥ 80% and 376 ± 44 m at ≥ 95% confidence. The greatest distance between a cub’s social group and its father’s social group was 2.0 km (maximum Multiple-paternity litters possible = 2.2 km); the father was 2 years old, and the natal There were 143 litters of more than one cub that had both groups of father and cub were located at opposite ends of parents assigned, and 64 (45%) of these were multiple- Wytham Woods. Seventeen further inferred mating sorties paternity litters, comprising 49 twins, 11 triplets and 4 were between 1.0 and 2.0 km. The mean distance between quadruplets. Three litters had more than two fathers: one the main setts of neighbouring social groups, based on triplet and one quadruplet had three different fathers, and bait-marking (Kruuk 1978) data from years with the least one quadruplet had four different fathers. Examining the (1987) and most (2005) social groups, were 486 ± 111 m (n =27 LOD scores of the fathers assigned paternity within these neighbouring pairs) and 373 ± 60 m (n =63), respectively. 64 multiple-paternity litters, to the other cubs in the litter, The natal groups of both parents of 186 cubs (≥ 80% revealed negative LOD scores in 44 litters, indicating that confidence) were known, and the parents of 31 (17%) of these males were unlikely have sired the whole litter. This these cubs were born in the same social group as each other. therefore suggested multiple paternity in 44/143 (31%) A similarly low number of parents of cubs assigned at ≥ 95% litters. Multiple paternity was observed in 20/65 (31%) confidence had the same natal group (15/108; 14%). litters assigned at ≥ 95% confidence, all of which were twins. Of the 134 males assigned paternity to more than one cub The LOD scores revealed strong evidence of multiple in a given year at ≥ 80% confidence, 53 sired cubs only in paternity in 16/65 (25%) litters at ≥ 95% confidence. Of their resident social group, 52 sired cubs only in other groups, these multiple-paternity litters, six were sired solely by and 29 had a mixed outcome. Similar values were seen at males from the same social group as the mother, four solely ≥ 95% confidence: 35, 20, and 17, respectively.

Table 3 Litter size and percentage of females breeding in lower-density populations. Dash (—) indicates data not stated

Density Females Mean litter Study population (badgers/km2) breeding (%) size Data Reference

Suburban Bristol, England 4.4–7.5 48 2.7 Capture–mark–recapture, Harris & Cresswell 1980–84 (1987), table 4 Central Sweden 2.4–3.2 26–45 — 26 females Data in Anderson & Trewhella (1985) Doñana area, 0.23–0.67 65 — Capture–mark–recapture, Revilla et al. (1999) south-west Spain 1983–98, 17 females County Offaly, Ireland — 35–40 — Culling, 1989–90 Whelan & Hayden (1993) East Germany 2–4 — 3.3 14 litters Data in Anderson & Trewhella (1985) Holland 1 — 2.4 15 litters Data in Anderson & Trewhella (1985) Serra de Grândola, 0.36–0.48 — 4 3 litters Rosalino et al. (2004) south-west Portugal Bialowieza Primeval 0.21 — 2.4 16 litters Kowalczyk et al. (2003) Forest, Poland

evidence that cubs in the same litter can have extra-group Discussion and within-group fathers. We discuss our results in rela- Our study documents the mating system of the European tion to the four proposed hypotheses for the occurrence of badger in Wytham Woods, Oxford, from 1988 to 2005, delayed dispersal. using 22 microsatellite loci. Carpenter et al. (2005) also examined the mating system of a similarly high-density Plural breeding badger population in Woodchester Park, Gloucestershire, over a similar time period, 1989–2002. They used 17 On average, there were six males and six females of breeding microsatellite loci to assign parentage to 425 cubs from 10 age within a social-group-year, of which a mean of 1.5 males social groups, with maternity assigned to 185 cubs, and 1.6 females bred, similar to the estimates in Woodchester paternity to 64 cubs and both parents to 58 (14%) cubs at Park of 1.2 males and 1.8 females (Carpenter et al. 2005). ≥ 95% confidence. In comparison, we assigned parentage Plural breeding in Wytham Woods had not been quantified to 630 cubs with maternity assigned to 336 cubs, paternity before; previous genetic analyses revealed at least three to 338 and both parents to 331 (53%) at ≥ 95% confidence. mothers within one studied social group (da Silva et al. Our study utilized cervus 3.0.1.8, which uses revised 1994; Domingo-Roura et al. 2003) and field data suggested likelihood equations that correct the way in which the a mean of two or three (Woodroffe & Macdonald 1995; genotyping error rate is interpreted, which had resulted in Rogers et al. 1997). Breeding among males has been harder conservative assignment rates previously (Kalinowski to quantify, and prior to the Woodchester Park study, only et al. 2007). cervus 3.0.1.8 also jointly assigns parent-pairs, Evans et al. (1989) demonstrated that more than one male which is more powerful and robust than the previous may sire cubs within a group. method, in which offspring were first assigned one parent, The 27% of candidate fathers and 31% of candidate and then the second parent was assigned dependent upon mothers estimated to have bred each year were similar to the first. We also reconstructed sibships of cubs that were calculations from Woodchester Park (18–31% and 29%, not assigned one or both parents using cervus. These respectively, at ≥ 80% confidence, Carpenter et al. 2005) and approaches provided a high rate of parentage assignment, Wytham field data (29% of candidate mothers, Macdonald which is notable given that no parents were known a priori, & Newman 2002). This is reduced compared to the estimates there was a large number of candidate parents that in lower-density populations (Table 3), suggesting a cost to included close relatives, and some candidates were not living in large groups in terms of the proportion of females genotyped. that breed each year. Post-mortem studies have also reported Our discussion focuses upon the results calculated from a decline in per adult productivity with group size (Cresswell the more robust parentage assignments at ≥ 95% con- et al. 1992). This may be due to female–female competition fidence. We provide strong evidence for the occurrence of within social groups, as the number of mothers in a group plural breeding and extra-group paternity, the strongest declines at higher latitudes, where population density is evidence, to date, of multiple-paternity litters, and the first low (Woodroffe & Macdonald 1995).

Such a cost would only occur if litter size does not increase compensating for poor quality mates, and obtaining good with density. Males and females that were assigned parent- genes (Jennions 1997; Jennions & Petrie 2000; Johnson 2001; age had a mean of 1.5 and 1.3 assigned cubs, respectively, Chapman et al. 2003; Wolff & Macdonald 2004). Males, similar to Woodchester Park (1.3 for fathers at ≥ 95% and conversely, may not benefit from mate-guarding females 1.5 for mothers at ≥ 80% confidence, Carpenter et al. 2005); if females conceive over a long time-period, and indeed, however, these may be underestimates, as not all cubs were there is no behavioural evidence of mate guarding. As social assigned parentage and some cubs may have died before groups consist of related individuals (H. L. Dugdale et al., they could be trapped. Teat-size data produced a similar unpublished), males may not benefit from mate-guarding litter size of 1.6 (Macdonald & Newman 2002). Mean foetal females from within-group males. Instead, males may seek litter size measured using ultrasound in Wytham was to prevent extra-group males from mating, through indirect 1.9 ± 0.1 (1993–2005, unpublished data). If the trapped territorial defence and direct aggression around the sett. females are representative of the population, this may suggest Given the level of promiscuity and that badgers forage embryo reabsorption (38%, Woodroffe & Macdonald 1995; solitarily, males cannot always mate-guard females. Yamaguchi et al. 2006), pre-emergence mortality (up to 25%, Additionally, as females can produce multiple-paternity Anderson & Trewhella 1985), and/or that not all of the litters, with a mixture of same-group and extra-group fathers, cubs were caught. Mean litter sizes in lower-density popu- it may benefit males to make sorties into neighbouring lations were higher (Table 3), reinforcing the cost to group territories to gain extra-group paternities, although exactly living for females at high densities. where extra-group mating events occur is unknown. The percentage of breeding individuals in the population is high compared to other social mammals, such as meerkats Spatial patterns of paternity and dispersal (Suricatta suricatta, Griffin et al. 2003) and Ethiopian wolves (Canis simensis, Randall et al. 2007), where reproductive Extra-group paternity has been suggested, but not quantified, suppression is thought to occur and the majority of offspring using allozymes (Evans et al. 1989; da Silva et al. 1994). are assigned to the dominant pair. More egalitarian systems, Extra-group males obtained 42% of the assigned paternities with limited evidence of reproductive suppression, are seen and 86% of these extra-group paternities were by neigh- in banded mongooses (Mungos mungo; 71% of females give bouring males, which is similar to Woodchester Park (45% birth, Cant 2000) and spotted hyenas (Crocuta crocuta; all and 67%, respectively, Carpenter et al. 2005). Neighbouring females and two-thirds of males were assigned parentage, setts were on average 430 m apart, whereas extra-group Engh et al. 2002). Reproductive skew in high-density badger paternities occurred over a mean of 376 m. Badgers scent- groups may be affected by limited reproductive suppression, mark boundaries to advertise their sex and reproductive although other factors affect this (H. L. Dugdale, D. W. Mac- status (Buesching et al. 2002), which increases the likelihood Donald, L. C. Pope, P. J. Johnson & T. Burke, unpublished). of extra-group parentage occurring between neighbouring Further studies of badger mating systems in lower-density mates, rather than mates further afield. A maximum of 2 km, populations are required to enhance our understanding. or four social groups, was crossed and both assigned parents were trapped in the year of conception. Including badgers not trapped in the year of conception may exaggerate the Multiple-paternity litters frequency of extra-group paternities; however, examining There is strong potential for the occurrence of multiple- only assigned parents trapped in the year of conception paternity litters in badgers, due to the presence of more produced similar results in Woodchester Park. Furthermore, than one adult or yearling male per social group and specific considering only the paternity success of genotyped males features of the badger’s reproductive system (delayed trapped in the population will underestimate gene flow. implantation and superfetation, Yamaguchi et al. 2006). As Short-term, intergroup movements of badgers of both badgers can conceive throughout the year, cubs within a sexes are common (D. W. Macdonald et al., unpublished), litter may be sired by more than one male. We found the and may represent a search for mates or a group to disperse strongest evidence, to date, of multiple paternity in 16/65 to. Extra-group sorties may occur throughout territories; (25%) litters at ≥ 95% confidence, compared to 5/31 (16% at therefore, the actual frequency is likely to be higher than ≥ 80%) and 0/7 at ≥ 95% confidence in Woodchester Park that revealed by trapping data, which only records pre- (Carpenter et al. 2005). sence at a sett. Extra-group mountings have been observed Multiple-paternity litters may provide females with fitness (Christian 1995) and both sexes have been radio-tracked in advantages, such as direct benefits like fertilization neighbouring groups (Christian 1994). Extra-pair paternity assurance, reduced harassment from males or reduced may increase the genetic diversity of litters and increase infanticide risk from males, and genetic benefits such as offspring heterozygosity in alpine marmots that live in family promoting sperm competition, possibly increasing litter groups (Cohas et al. 2007). In Ethiopian wolves, however, genetic diversity, reducing genetic incompatibilities, where female-biased dispersal may reduce inbreeding, there

© 2007 The Authors Journal compilation © 2007 Blackwell Publishing Ltd 5304 H. L. DUGDALE ET AL. was no evidence that extra-pair paternity reduced inbreeding Research Council (NERC) Molecular Genetics Facility. This work (Randall et al. 2007). Badger groups correspond to kinship was generously supported by the People’s Trust for Endangered groups (H. L. Dugdale et al., unpublished), with low rates Species and NERC. of delayed dispersal (D. W. Macdonald et al., unpublished) and natal philopatry. Motivation or selection for dispersal References may be lowered through the potential for extra-group matings, which may have initially evolved as an inbreeding Anderson RM, Trewhella W (1985) Population dynamics of the Meles meles avoidance mechanism, although the fact that delayed dis- badger ( ) and the epidemiology of bovine tuberculosis (Mycobacterium bovis). Philosophical Transactions of the Royal Society persal is most likely between neighbouring badger groups of London. Series B, Biological Sciences, 310, 327–381. (D. W. Macdonald et al., unpublished) reduces this likelihood. Arnold KE, Owens IPF (1998) Cooperative breeding in birds: a Further research is required into dispersal and extra-group comparative test of the life history hypothesis. Proceedings of the paternity in lower-density badger populations. Royal Society of London. Series B, Biological Sciences, 265, 739–745. Badger social groups may initially form on a purely Benjamini Y, Hochberg Y (1995) Controlling the false discovery ecological basis (Johnson et al. 2001a, b), as there are few rate: a practical and powerful approach to multiple testing. Journal of the Royal Statistical Society. Series B (Methodological) 57 demonstrated benefits to group living (Johnson et al. 2004). , , 289–300. Once formed, social groups are fostered through kinship ties; Bruford MW, Hanotte O, Brookfield JFY, Burke T (1998) Multilocus however, group living may be costly, especially to females and single-locus DNA fingerprinting. In: Molecular Genetic in terms of a reduced likelihood of direct reproduction and Analysis of Populations: a Practical Approach (ed. Hoelzel AR), reduced litter size compared to lower-density populations. pp. 287–336. IRL Press at Oxford University Press, Oxford, UK. This may represent a difference in dispersal propensity as, Buesching CD, Waterhouse JS, Macdonald DW (2002) Gas- at high density, badgers tend to move to groups with fewer chromatographic analyses of the subcaudal gland secretion of Meles meles competing individuals of the same sex (D. W. Macdonald the European badger ( ) Part I: chemical differences related to individual parameters. Journal of Chemical Ecology, 28, et al., unpublished); however, research is required into 41–56. dispersal and mating patterns in lower-density populations. Cant MA (2000) Social control of reproduction in banded mon- If cooperative breeding occurs, which it may (Woodroffe gooses. Animal Behaviour, 59, 147–148. 1993) or may not (Woodroffe & Macdonald 2000), then this Carpenter PJ, Pope LC, Greig C et al. (2005) Mating system of the cost to group living may be reduced. Additionally, there Eurasian badger, Meles meles, in a high density population. are undoubted costs to dispersing in high-density popu- Molecular Ecology, 14, 273–284. lations (da Silva et al. 1993; Tuyttens et al. 2000), so any direct Carr GM, Macdonald DW (1986) The sociality of solitary foragers: a model based on resource dispersion. Animal Behaviour, 34, or indirect reproduction benefits of natal philopatry may 1540–1549. be enhanced by indirect resource benefits, such as access Chapman T, Arnqvist G, Bangham J, Rowe L (2003) Sexual conflict. to a known high-quality territory. Any strong selection to Trends in Ecology & Evolution, 18, 41–47. disperse may be reduced further through the occurrence of Cheeseman CL, Cresswell WJ, Harris S, Mallinson PJ (1988) extra-group matings that increase gene flow. That there are Comparison of dispersal and other movements in 2 badger potential benefits to natal philopatry (Stacey & Ligon 1991), (Meles meles) populations. Mammal Review, 18, 51–59. as well as ecological constraints on dispersal (Emlen 1982), Cheeseman CL, Wilesmith JW, Ryan J, Mallinson PJ (1987) Badger population dynamics in a high-density area. In: Mammal Popu- plus life-history constraints on dispersal in the form of lation Studies: Proceedings of a Symposium at the Zoological Society of longevity that increases habitat saturation (Arnold & Owens London (ed. Harris S), pp. 279–294. Clarendon Press, Oxford, UK. 1998), suggests that all of these factors may play a role in Christian SF (1994) Dispersal and other inter-group movements in determining the degree of delayed dispersal, as proposed badgers, Meles meles. Zeitschrift für Säugetierkunde–International by the broad constraints hypothesis (Hatchwell & Komdeur Journal of Mammalian Biology, 59, 218–223. 2000). Christian SF (1995) Observations of extra-group mating and mate- defence behaviour in badgers, Meles meles. Journal of Zoology, 237, 668–670. Acknowledgements Cohas A, Yoccoz NG, Allainé D (2007) Extra-pair paternity in alpine marmots, Marmota marmota: genetic quality and genetic We would like to thank everyone who participated in the Wytham diversity effects. Behavioral Ecology and Sociobiology, 61, 1081– Woods badger project, without whom this study would not have 1092. been possible. Kristien Erven, Helen Hipperson, and Andy Krupa Cresswell WJ, Harris S, Cheeseman CL, Mallinson PJ (1992) To offered invaluable advice during laboratory work. We thank Ben breed or not to breed: an analysis of the social and density- Sheldon for kindly providing laboratory space for hair extractions. dependent constraints on the fecundity of female badgers (Meles We are very grateful to Paul Johnson, Tristan Marshall, Deborah meles). Philosophical Transactions of the Royal Society of London. Randall and Jinliang Wang for their helpful advice on our analyses, Series B, Biological Sciences, 338, 393–407. and Dave Coltman, James Malcolm, Chris Newman and two anony- da Silva J, Macdonald DW (1989) Limitations to the use of tooth mous referees for valuable comments on the paper. The laboratory wear as a means of aging Eurasian badgers, Meles meles. Revue work was completed at the Sheffield node of the Natural Environment D Ecologie–La Terre et la Vie, 44, 275–278.

© 2007 The Authors Journal compilation © 2007 Blackwell Publishing Ltd BADGER MATING SYSTEM 5305 da Silva J, Macdonald DW, Evans PGH (1994) Net costs of group Kalinowski ST, Taper ML, Marshall TC (2007) Revising how the living in a solitary forager, the Eurasian badger (Meles meles). computer program CERVUS accommodates genotyping error Behavioral Ecology, 5, 151–158. increases success in paternity assignment. Molecular Ecology, 16, da Silva J, Woodroffe R, Macdonald DW (1993) Habitat, food 1099–1106. availability and group territoriality in the European badger, Kowalczyk R, Zalewski A, Jedrzejewska B, Jedrzejewski W (2003) Meles meles. Oecologia, 95, 558–564. Spatial organization and demography of badgers (Meles meles) Domingo-Roura X, Macdonald DW, Roy MS et al. (2003) Con- in Bialowieza Primeval Forest, Poland, and the influence of firmation of low genetic diversity and multiple breeding females earthworms on badger densities in Europe. Canadian Journal of in a social group of Eurasian badgers from microsatellite and Zoology, 81, 74–87. field data. Molecular Ecology, 12, 533–539. Kruuk H (1978) Spatial organisation and territorial behaviour of Emlen ST (1982) The evolution of helping. 1. An ecological the European badger Meles meles. Journal of Zoology, 184, 1–19. constraints model. American Naturalist, 119, 29–39. Littell RC, Henry PR, Ammerman CB (1998) Statistical analysis of Emlen ST, Oring LW (1977) Ecology, sexual selection, and the repeated measures data using SAS procedures. Journal of Animal evolution of mating systems. Science, 197, 215–223. Science, 76, 1216–1231. Engh AL, Funk SM, Van Horn RC et al. (2002) Reproductive skew Littell RC, Milliken GA, Stroup WW, Wolfinger RD, Schaben- among males in a female-dominated mammalian society. berger O (2006) SAS for Mixed Models, 2nd ednn. SAS Institute Behavioral Ecology, 13, 193–200. Inc., Cary, North Carolina. Evans PGH, Macdonald DW, Cheeseman CL (1989) Social struc- Macdonald DW (1983) The ecology of carnivore social behaviour. ture of the Eurasian badger (Meles meles): genetic evidence. Nature, 301, 379–384. Journal of Zoology, 218, 587–595. Macdonald DW, Carr GM (1989) Food security and the rewards of Evett I, Weir B (1998) Interpreting DNA Evidence: Statistical Gen- tolerance. In: Comparative Socioecology: The Behavioural Ecology etics for Forensic Scientists. Sinauer Associates, Sunderland, of Humans and Animals (eds Standen V, Foley R), pp. 75–99. Massachusetts. Blackwell Scientific Publications, Oxford, UK. Garant D, Kruuk LEB (2005) How to use molecular marker data Macdonald DW, Newman C (2002) Population dynamics of badgers to measure evolutionary parameters in wild populations. (Meles meles) in Oxfordshire, UK: numbers, density and cohort Molecular Ecology, 14, 1843–1859. life histories, and a possible role of climate change in population Griffin AS, Pemberton JM, Brotherton PNM et al. (2003) A genetic growth. Journal of Zoology, 256, 121–138. analysis of breeding success in the cooperative meerkat (Suricata Marshall TC, Slate J, Kruuk LEB, Pemberton JM (1998) Statistical suricatta). Behavioral Ecology, 14, 472–480. confidence for likelihood-based paternity inference in natural Hamilton WD (1964) The genetical evolution of social behaviour. populations. Molecular Ecology, 7, 639–655. I. Journal of Theoretical Biology, 7, 1–16. Paetkau D, Waits LP, Clarkson PL et al. (1998) Variation in genetic Harris S, Cresswell WJ (1987) Dynamics of a suburban badger (Meles diversity across the range of North American brown bears. meles) population. In: Mammal Population Studies: Proceedings of Conservation Biology, 12, 418–429. a Symposium at the Zoological Society of London (ed. Harris S), Parks SDE (2001) Trypanotolerance in west African cattle and the pp. 295–311. Clarendon Press, Oxford, UK. population genetic effects of selection. PhD thesis, University of Hatchwell BJ, Komdeur J (2000) Ecological constraints, life history Dublin, Dublin. traits and the evolution of cooperative breeding. Animal Pope LC, Domingo-Roura X, Erven K, Burke T (2006) Isolation by Behaviour, 59, 1079–1086. distance and gene flow in the Eurasian badger (Meles meles) at Jennions MD (1997) Female promiscuity and genetic incompatibility. both a local and broad scale. Molecular Ecology, 15, 371–386. Trends in Ecology & Evolution, 12, 251–253. Queller DC, Goodnight KF (1989) Estimating relatedness using Jennions MD, Petrie M (2000) Why do females mate multiply? A genetic-markers. Evolution, 43, 258–275. review of the genetic benefits. Biology Reviews of the Cambridge Randall DA, Pollinger JP, Wayne RK et al. (2007) Inbreeding is Philosophical Society, 75, 21–64. reduced by female-biased dispersal and mating behavior in Johnson DDP (2001) Models of coexistence, conflict and cooperation Ethiopian wolves. Behavioral Ecology, 18, 579–589. DPhil Thesis, University of Oxford, UK. Raymond M, Rousset F (1995) genepop (Version 1.2): population Johnson DDP, Baker S, Morecroft MD, Macdonald DW (2001a) genetics software for exact tests and ecumenicism. Journal of Long-term resource variation and group size: a large-sample Heredity, 86, 248–249. field test of the resource dispersion hypothesis. BMC Ecology, 1, 2. Revilla E, Delibes M, Travaini A, Palomares F (1999) Physical and Johnson DDP, Macdonald DW, Newman C, Morecroft MD population parameters of Eurasian badgers (Meles meles L.) (2001b) Group size versus territory size in group-living badgers: from Mediterranean Spain. Zeitschrift für Säugetierkunde– a large-sample field test of the resource dispersion hypothesis. International Journal of Mammalian Biology, 64, 269–276. Oikos, 95, 265–274. Rogers LM, Cheeseman CL, Mallinson PJ, CliftonHadley R (1997) Johnson PCD, Haydon DT (2007) Maximum-likelihood estimation The demography of a high-density badger (Meles meles) popu- of allelic dropout and false allele error rates from microsatellite lation in the west of England. Journal of Zoology, 242, 705–728. genotypes in the absence of reference data. Genetics, 175, 827– Rosalino LM, Macdonald DW, Santos-Reis M (2004) Spatial 842. structure and land-cover use in a low-density Mediterranean Johnson DDP, Macdonald DW, Dickman AJ (2000) An analysis population of Eurasian badgers. Canadian Journal of Zoology, and review of models of the sociobiology of the Mustelidae. 82, 1493–1502. Mammal Review, 30, 171–196. Ross KG (2001) Molecular ecology of social behaviour: analyses of Johnson DDP, Stopka P, Macdonald DW (2004) Ideal flea con- breeding systems and genetic structure. Molecular Ecology, 10, straints on group living: unwanted public goods and the emergence 265–284. of cooperation. Behavioral Ecology, 15, 181–186. Stacey PB, Ligon JD (1991) The benefits-of-philopatry hypothesis

for the evolution of cooperative breeding: variation in territory Woodroffe R, Macdonald DW (2000) Helpers provide no detectable quality and group size effects. American Naturalist, 137, 831–846. benefits in the European badger (Meles meles). Journal of Zoology, Taylor AC, Cowan PE, Fricke BL, Cooper DW (2000) Genetic 250, 113–119. analysis of the mating system of the common brushtail possum Yamaguchi N, Dugdale HL, Macdonald DW (2006) Female recep- (Trichosurus vulpecula) in New Zealand farmland. Molecular tivity, embryonic diapause and superfoetation in the European Ecology, 9, 869–879. badger (Meles meles): implications for the reproductive tactics of Thompson EA (1976) A paradox of genealogical inference. males and females. Quarterly Review of Biology, 81, 33–48. Advances in Applied Probability, 8, 648–650. Tuyttens FAM, Macdonald DW, Rogers LM, Cheeseman CL, Roddam AW (2000) Comparative study on the consequences of Hannah Dugdale conducted this research as part of her DPhil culling badgers (Meles meles) on biometrics, population dynamics supervised by David Macdonald. Lisa Pope supervised the and movement. Journal of Animal Ecology, 69, 567–580. laboratory work and holds a postdoctoral position at the University Valière N (2002) GIMLET: a computer program for analysing of Sheffield. This study was initiated and led by David Macdonald, genetic individual identification data. Molecular Ecology Notes, 2, who heads the Wildlife Conservation Research Unit at the University 377–379. of Oxford and Terry Burke, who heads a molecular ecology group Waits LP, Luikart G, Taberlet P (2001) Estimating the probability of and the NERC Molecular Genetics Facility at the University of identity among genotypes in natural populations: cautions and Sheffield. guidelines. Molecular Ecology, 10, 249–256. Walsh PS, Metzger DA, Higuchi R (1991) Chelex 100 as a medium for simple extraction of DNA for PCR-based typing from forensic material. BioTechniques, 10, 506–513. Supplementary material Wang J (2004) Sibship reconstruction from genetic data with typing The following supplementary material is available for this article: errors. Genetics, 166, 1963–1979. Waser PM, Jones WT (1983) Natal philopatry among solitary Table S1 Number of alleles per locus, allele sizes, and frequencies mammals. Quarterly Review of Biology, 58, 355–390. for each of the 22 microsatellite loci, obtained from the genotypes Whelan R, Hayden TJ (1993) The reproductive cycle of the female of all of the 915 genotyped badgers badger (Meles meles L.) in East Offaly. In: The Badger (ed. Hayden This material is available as part of the online article from: TJ), pp. 64–77. Royal Irish Academy, Dublin, Ireland. http://www.blackwell-synergy.com/doi/abs/ Wolff JO, Macdonald DW (2004) Promiscuous females protect 10.1111/j.1365-294X.2007.03571.x their offspring. Trends in Ecology & Evolution, 19, 127–134. (This link will take you to the article abstract). Woodroffe R (1993) Alloparental behaviour in the European badger. Animal Behaviour, 46, 413–415. Please note: Blackwell Publishing are not responsible for the Woodroffe R, Macdonald DW (1995) Female/female competition content or functionality of any supplementary materials supplied in European badgers Meles meles: effects on breeding success. by the authors. Any queries (other than missing material) should Journal of Animal Ecology, 64, 12–20. be directed to the corresponding author for the article.