2.3 Population diversity, differentiation, and divergence
Each population was described by calculating haplotype frequencies, inter-haplotype distances, haplotype diversity and nucleotide diversity (π), as well as expected and observed heterozygosity for nuclear markers using DNAsp v.5.10.01 (Librado & Rozas 2009) and Genetix v4.05.2 (Belkhir et al. 2004). We evaluated signals for departures from neutrality or demographic changes by estimating Tajima’s D (Tajima 1989) and Fu’s Fs (Fu 1997) for each locus, with Arlequin v.3.1 (Excoffier et al. 2005). Differentiation among populations was estimated by performing AMOVAs, and calculating pairwise F ST and ‎ΦST and the population average pairwise differences DXY, using Arlequin. For AMOVAs, samples were stratified into five groups, corresponding to the five nominal lineages (lherminieri , boydi and baroli in the Atlantic,nicolae and bailloni in the Indian Ocean), and populations (i.e. sampling localities; Fig. 1) within these groups. The matrix of genetic distances among all pairs of haplotypes was computed using the K2P model of substitution for concatenated mitochondrial markers, and TN93 for concatenated nuclear markers, as determined using jModelTest2. We used a Mantel test to measure the level of correlation among genetic distances and geographic distances (Smouse et al. 1986). Geographic distance was calculated as the shortest distance between two populations without crossing land. Statistical significance (AMOVAs, PairwiseF ST and Mantel tests) was estimated using 1000 permutations. To visualize relationships among lineages, we inferred NeighborNet networks using SplitsTree v 4.14.2 (Huson & Bryant 2006), with different dataset combinations: all markers independently, concatenated mitochondrial markers, concatenated nuclear markers.