Formal tests for population mixture.

Notes: We analyzed data from all West Eurasian populations with ≥5 samples. Regions are abbreviated: I – Northwest Europe, ECE – East-Central Europe, SE – Southern Europe and L – Levant. We used a Block Jackknife (block size of 5cM) to correct for LD among SNPs and to estimate a Z-score that reports the number of approximately normally distributed standard deviations that the correlation coefficient differs from 0. For the 4 Population Test, we interpret |Z|>3 as significant evidence for mixture (we test the tree ((Px-PCEU)(PPapuan-PYRI), and do not show the tests of the two alternative trees, although all |Z|-scores are >16). For the 3 Population Test, we interpret Z<−3 as significant evidence for mixture; a positive score for the 3 Population Test is possible even in the presence of population mixture, since genetic drift after mixture can mask the signal (for example, Bedouin-g2). Scores that are significant are highlighted in bold. For further study of sub-Saharan African mixture, we chose populations with a significantly negative score by the 4 Population Test (bold).