Data analyses
Hymenoptera males are haploid, and females are diploid. We analysed the genetic structure of female wasps only (N=195), to avoid overestimating the impact of the haploid males, from the five localities and across the 22 years of collection using three analyses implemented in the BAPS software (Corander, Marttinen, Sirén, & Tang, 2008; Corander, Waldmann, & Sillanpää, 2003). First, we ran a Bayesian clustering of individuals analysis, using the whole Åland as a unique population (Npop=1), to identify the maximum number of genetic clusters across all our samples. Then, we ran two independent spatial Bayesian clustering analyses with the ‘clustering of groups of individuals’ (Npop=5 or Npop=12), and with respective admixture analyses settings (Figure 3) to evaluate the degree of admixture at each locality, and in each locality through time. Multiple ‘k’ parameters were provided for each run and the parameter was not set as a strict limit; and the Bayesian model in BAPS estimated the posterior probabilities for the optimal number of clusters (Appendix 1-3) (Corander et al., 2003).
We used the software GenoDive 3.04 (Meirmans & van Tienderen, 2004) to calculate population genetics measures using only the samples that were successfully genotyped for at least seven microsatellite markers using either only diploid females (N=195), or combined diploid females and haploid males dataset (N=313). We tested Hardy-Weinbergequilibrium at each locus in each locality, and for each of the 12 spatio-temporal groups, using the Least-Square Amova Method with 999 permutations (estimator Fis given in Appendix 1C and 3C). We calculated the F ST values between the five localities and between the 12 spatio-temporal groups. Finally, we calculated the observed (HO ) and expected heterozygosity (HE ) at each locus, as well as the inbreeding coefficient (Gis ) between each locality (N=5), and each spatio-temporal group (N=12) (Appendix 1B and 3B).
To test whether samples from the Seglinge-Kumlinge islands were more often Wolbachia -infected, or carry the T-mitotype over the C-mitotype more than the rest of the Åland population, we used Fisher exact tests for count data (two-sided) in R 3.6.3 (R Core Team, 2020).