Estimation of individual ancestry proportions was performed with the programs STRUCTURE[25],[45] and ADMIXMAP[24]. Since very similar estimates were obtained with both programs, only those obtained with STRUCTURE are reported here. Replicate runs of STRUCTURE used a burn-in period of 20,000 iterations followed by an additional 10,000 iterations from which parameter estimates were obtained. Ten replicate runs were carried out and the average parameter estimate retained. Population admixture proportions and mean time since admixture were calculated from the individual estimates. Spearman rank correlations (two-sided) and Wilcoxon signed rank tests between ancestry estimates and other population parameters were applied using the R statistical package (http://www.r-project.org). STRUCTURE runs used an admixture model with correlated allele frequencies and with individuals from ancestral populations assigned to K predetermined clusters (so-called “supervised analysis”). K was varied in order to examine different groupings of Native American populations while considering Europeans and Africans as single independent clusters: K = 3 when grouping all Native American data into a single cluster, K = 7 when Native American populations were grouped into five linguistic stocks and K = 28 when each Native American population was considered independently. Population assignment to linguistic stocks followed the linguistic classification proposed by Ruhlen (1991)[40].
