Mating patterns: population genetic summary statistics
We measured genetic diversity within localities by calculating per-site
nucleotide diversity (θπ, i.e. average number of
pairwise differences between sequences) using ANGSD. We generated a
theta file for each locality based on the locality-specific site
frequency spectra. We intersected the theta files of all localities of
each species in order to only compare SNPs that were shared by all
localities following Peñalba et al. (2015). From these intersected sites
files, we calculated nucleotide diversity using thetaStat as implemented
in ANGSD (Maas et al. 2018). We divided θπ estimates by
the number of sites and tested for significant differences between
localities using Kruskal-Wallis ANOVA and Dunn tests with Bonferroni
correction for multiple comparisons (Dinno 2017). For calculating
heterozygosity, we calculated unfolded site frequency spectra per
individual and divided per-individual heterozygosity by the total number
of sites. Again, we used Kruskal-Wallis ANOVA and Dunn tests to evaluate
differences among localities.
We calculated per-individual inbreeding coefficients (F ) on
genotype likelihoods as the degree of deviation from Hardy-Weinberg
equilibrium (ngsF v.1.2.0-STD,
Vieira et al. 2013). We performed two runs in ngsF, the first run to
calculate reliable starting values of F per individual and a
second performing a deeper search, allowing for a maximum of 1,500
iterations.
We assessed localized (within locality) mating patterns by testing for
isolation-by-distance (IBD) between individuals using Mantel’s tests on
pairwise genetic distances (calculated using ngsDist v.1.0.9, Vieira et
al. 2016b) and log-transformed geographic distances between individuals
(mantel.randtest , 10,000 permutations, ADE4 v.1.7-13;
Dray & Dufour 2007).