Table 1.

Correlation coefficients between the estimates of STRUCTURE and the singular value decomposition (SVD)-based method from the full data of Rosenberg et al. [5]

	Cluster/population
K a	1	2	3	4	5	6
2	0.9602	0.9602
3	0.9756	0.9695	0.9826
4	0.9824	0.9853	0.9667	0.9708
5	0.9602	0.9836	0.9470	0.9719	0.5715
6	0.5688	0.9473	0.9473	0.9719	0.9642	0.9596

^aK is the number of clusters/populations.

Note: For the SVD-based method, the first five principal components were used. The mixture model was used for clustering.