(4) Dispersal patterns
To examine if dispersal distances differ between sexes, we compared the diversity of mtDNA haplotypes, relatedness, and heterozygosity among adult females and males. MtDNA haplotype and nucleotide diversity was calculated and compared using a permutation test implemented in R function genetic_diversity_diff 1.0.6 (https://github.com/laninsky/genetic_diversity_diffs). We included 12 sampled adult females and 12 adult males in this analysis, plus 2 females that could not be sampled but whose haplotypes were inferred from the haplotypes of their offspring (the adult female of Group 4, who supposedly had been replaced before the study period, and the adult female of Group 9, who was present during the study period but could not be sampled). Relatedness was calculated using Wang’s estimator r and compared between sexes using 1000 bootstrapping samples in Coancestry 1.0.1.9 (Wang, 2011). In this analysis, as well as in the tests described below, we included 12 sampled adult females and 12 males. Female genotypes included 16 autosomal and one X-linked loci, while male genotypes included only the 16 autosomal loci. Individual heterozygosity was calculated using HL estimator (homozygosity by locus, see above) and compared between sexes using a paired t-test.
To evaluate spatial genetic structure, we conducted a spatial autocorrelation analysis following Smouse and Peakall (Smouse & Peakall, 1999) in PopGenReport 2.2.2 (Adamack & Gruber, 2014), separately for adult females and males. The analysis allows to estimate the correlation between pairwise genetic distances, calculated using microsatellite genotypes with the method of Smouse and Peakall (Smouse & Peakall, 1999), and pairwise spatial distances. If a spatial genetic structure is present, neighboring individuals will be more closely related, and negative correlation between r and spatial distances is expected; the shorter distances individuals travel the stronger the correlation should be. As a measure of spatial distances, we used distances between centroids of home ranges estimated using the 95% fixed kernel density method with ArcGIS Desktop 10.6 (ESRI).
To further evaluate spatial genetic structure in females, we conducted a test similar to spatial autocorrelation analysis using mtDNA haplotype distances, correlating the number of nucleotide differences between haplotypes with spatial distances. For this test, if a spatial genetic structure is present, a positive correlation between haplotype and spatial distances is expected. We used Mantel tests with 10,000 permutations in R package ecodist (Goslee & Urban, 2007).
(6) Kinship within and between groups
Kinship was assigned conservatively based on r values and pedigree reconstruction in Colony. First-degree kinship (full-sibling and parent-offspring, without distinguishing between these two categories) was assigned to dyads with r > 0.487 (mean r for simulated parent-offspring dyads). Second-degree kinship (dyads sharing 25% of the genome, such as half-sisters or uncle-nephew dyads) was assigned for dyads with r > 0.247 (mean r for simulated half-offspring dyads). Dyads with lower r values were categorized as unrelated.