Apart from the three mismatches in the case of the unassigned paternity, we found only two cases of allelic mismatches (Supplementary Table S1). Father-daughter dyad from Group 1 had a mismatch at locus chr10a. Since the daughter was homozygous at this locus, this was most likely a result of allelic dropout or genotyping error. Father-daughter dyad from Group 9 had a mismatch at locus chrXa; the daughter was not homozygous at this locus but considering the high likelihood of parentage given by the other loci, we assumed that this mismatch was due to a genotyping error.
Of eight sampled social mothers, seven were identified as genetic mothers of all offspring in their family groups (17 offspring in 7 groups, 1 to 5 offspring per group). One inferred case of female replacement was detected, as the adult female of Group 4 was not identified as the genetic mother of the group’s juvenile offspring; they did not share the mtDNA haplotype and had 11 allelic mismatches. No other female in our sample was identified as the most likely mother for this offspring or shared a mtDNA haplotype with it. The social father of this juvenile was indicated as the genetic father.
All assignments were made with a 95% confidence level in Cervus software and confirmed with Colony software (Supplementary Table S1). The assignments did not change when the set of known mother-offspring pairs was excluded from the priors. Colony also yielded strong support for full-sib relationships between all offspring from the same groups, confirming correct parentage assignments.
Is mate choice based on relatedness or heterozygosity?
We found no evidence for relatedness-based mate choice. There was no difference between relatedness of real mating pairs and randomly generated mating pairs (-0.048 vs. -0.021, p = 0.565; n = 10 pairs, breeding pool of 12 females and 12 males). Likewise, we found no evidence for heterozygosity-based mate choice, as homozygosity by loci (HL) was not significantly correlated between pair mates (r = -0.527, n = 10 pairs, p = 0.118).
Despite the lack of evidence for active inbreeding avoidance via mate choice, relatedness (Wang’s r) between mating partners was generally low, averaging -0.033, and none of the pair mates shared the same mtDNA haplotype (Supplementary Table 1, Fig. 1). Only in one pair were the partners found to be second-degree kin (Group 6, r = 0.285). The mtDNA haplotype network (Fig. 2) showed no clear pattern of haplotype similarity between pair mates: some had closely related haplotypes (e.g., Groups 4, 5, 9), while others had only distantly related haplotypes (e.g., Groups 1, 11).
Do both sexes disperse and does one sex disperse further than the other?
Our results indicate that both sexes dispersed similar distances. There were no significant differences between adult females and males in mean mtDNA haplotype diversity (0.945 in females, 0.924 in males, permutation test p = 0.766), mtDNA nucleotide diversity (0.027 in females, 0.029 in males, permutation test p = 0.699), mean relatedness r (-0.013 in females, -0.056 in males, mean difference -0.040, lying within the 95% confidence interval (-0.048 – 0.054) obtained by bootstrapping) or mean heterozygosity HL (0.184 in females, 0.216 in males, paired t-test p = 0.438).
We did not find evidence for spatial genetic structure in our study population, suggesting that dispersal is most likely opportunistic. The correlation between genetic and spatial distances was not significant for either sex, as the 95 % CI of autocorrelation r values overlapped zero for all distance classes (Supplementary Materials Table 2, Fig. S1). The correlation between mtDNA haplotype distances and spatial distances in females was not significant either (Mantel correlation = 0.048, n = 91 dyads, right-tailed p = 0.342).