Figure 9.

Schematic world map of the “flow” of microsatellite alleles. Colored boxes represent regions of the world, positioned geographically. Links entering into a geographic region indicate the percentages of distinct alleles from the geographic region found in other regions (and an edge with the number x is drawn proportionately in width to x/4–8). For example, averaging across loci, 87% of alleles observed in Europe are also observed in Africa, whereas 74% of alleles observed in Africa are also observed in Europe. More precisely, following Conrad et al. (2006), for a given locus, a sample size g, and a pair of regions A and B, the expected number α of distinct alleles that will be found in a sample of size g from region A is computed as in Figure 2A. The expected number π of distinct alleles that will be found in a sample of size g from region A but not in a sample of size g from region B is computed as in Figure 2C, as the number of alleles private to region A, averaging over all possible subsamples that contain g alleles from region A and g alleles from region B. The fraction of the alleles in region A that are found in region B is then obtained as 1-π/α. The value g=40 was used for all computations, and for a given pair of regions, only those loci with sample sizes of at least g in both regions were considered.