Fig. 2

Log-likelihood (a) and ΔK (b) plots for reanalysis of microsatellite data from populations belonging to the major cluster for K = 5 in Fig. 1 (omitting Malaysia), with 20 replicate runs of 500 000 burn-in iterations and 1 000 000 further iterations, summarised by Structure Harvester [31]. Below are ancestry coefficients (c) for K = 2 and K = 3, averaged across runs and plotted by CLUMPAK [24, 32]