Population genetics analyses
We constructed a neighbor-joining tree using the pairwise genetic distance matrix data of all individuals calculated using TreeBest v1.9.2 (Vilella et al. 2009), and the number of bootstrap replicates was set to 1000 to assess the statistical support for nodes in the tree.
Principal components analysis (PCA) based on SNP markers was performed using GCTA tools (Yang et al. 2011). We transformed the population genotypes into a matrix that included the numbers 0, 1, and 2, where 0 was used to represent a genotype that is homozygous for the reference allele, 1 is a genotype heterozygous for the reference allele; and 2 is a genotype homozygous for the non-reference allele. We calculated the sample covariance of the matrix that contained the information for all individuals (with the numbers 0, 1, and 2). Finally, we calculated the eigenvector decomposition of the matrix and plotted the PCA using GCTA tools.
We used PLINK to generate the map files necessary for downstream analyses (Purcell et al. 2007). Population structure was determined using the program FRAPPE (Tang et al. 2005). We increased the pre-defined parameter K from 2 to 5 to cover the maximum numbers of lineages that could be identified in the tree, representing the assumed groups of a simulated population in ancient times. Population genetic structure and individual ancestry proportions were finally inferred using FRAPPE.