2.6 Demographic history analysis
To reconstruct historical relationships among North Atlantic Harbour porpoises, we inferred a maximum likelihood bifurcating population tree using Treemix (v1.13) (Pickrell & Pritchard, 2012). The genotypes (vcf ) were filtered to retain only sites with no missing data. We performed the Treemix analysis 1,000 times at the population level with a window size of 1,000 SNPs, specifying the Black Sea subspecies as outgroup and with the option-noss to turn off sample size correction, as suggested by the authors. We obtained a consensus tree and bootstrap values with the R package BITE (Milanesi et al., 2017).
We also inferred changes in effective population sizes (Ne ) through time with the software SMC++(Terhorst et al., 2017). We randomly chose three individuals per population, except for the Iberian porpoises, where only one individual was selected for the analysis as the other IBE porpoise was identified as a migrant. This analysis was run only using scaffolds larger than 1Mba (160 scaffolds) and no MAF filtering was applied on the vcf . As repetitive and excessive coverage regions were removed from thebam files, these uncalled regions were marked as missing data, as suggested by the authors (Terhorst et al., 2017). These regions could be misidentified as very long runs of homozygosity, erroneously decreasing the N e estimate, and hence compromising the power to infer true population contractions. In addition, we formed composite likelihoods by varying the distinct individual (-d ) among the three individuals, also as suggested by the authors. Then, the population size histories were computed by using the optionestimate with the default settings, a generation time of 11.9 years (Taylor et al., 2007) and a mutation rate of 2.56x10-8 (Yim et al., 2014).