_e0134207_Fig_8a.jpg)
Monte Carlo simulations of genomic phylogeography in Harp Seals (Pagophilus groenlandicus)
The four breeding populations of Harp Seals (Pagophilus groenlandicus) can be arranged in a linear stepping-stone model, with the dispersal cost among populations. In Model C, the dispersal cost is the number of populations through which any seal must move to get from one population to another. The 4x4 matrix shows these symmetrical costs. In Model d, the Southern Gulf and Newfoundland Ice Front populations are combine as a single population (dispersal cost 0), and the matrix is recalculated accordingly. Given the topology of the Maximum Likelihood tree, seals from different breeding grounds are distributed over 10,000 random trees (a Monte Carlo simulation).
The numbers of dispersal events required in these random trees are compared with that required in the observed tree. In Model C, the observed tree requires 26 dispersal events, as compared with a mode of 32. This places the observed three in the 1.3% tail of the random trees. In Model D, the observed tree require 18 events as compared with the modal 22 events, and falls in the 3.2% tail. In both cases, the probability of obtaining the observed tree by chance is p < 0.05. The results suggest that gene flow among harp seal breeding populations is non-random, and follows a stepping-stone model.