Table 1. Accuracy of multiple estimators under Models Split and Tree.

Model	Split					Tree
K	1	2	3	4	5	3	4	5
		0.495	0.502	0.493	0.492	0.486	0.507	0.501
DIC	1.00	1.00	1.00	1.00	1.00	1.00	1.00	0.98
STRUCTURE	0.90	1.00	1.00	0.86	0.80	0.98	0.94	0.72
STRUCTURE, F model	0.90	0.98	0.94	0.82	0.54	0.90	0.82	0.62
		1.00	0.94	0.70	0.64	0.80	0.86	0.64
, F model		1.00	0.90	0.78	0.50	0.84	0.92	0.54
Eigenanalysis,	0.97	0.89	0.86	0.86	0.96	0.96	0.92	0.90
Eigenanalysis,	1.00	0.96	0.91	0.93	0.99	0.98	0.94	0.92
Eigenanalysis,	1.00	1.00	0.96	0.96	1.00	1.00	0.96	0.96
Structurama, noninformative prior	1.00	1.00	0.82	0.18	0.02	0.88	0.22	0.00
Structurama, correct prior	1.00	1.00	0.82	0.18	0.02	0.82	0.22	0.00
BAPS	1.00	1.00	1.00	0.82	1.00	1.00	1.00	0.96

Performance assessment of methods including DIC, STRUCTURE, , Eigenanalysis, Structurama and BAPS. “” is the population differentiation statistic estimated by SmartPCA [11] averaged across 50 data sets. STRUCTURE's performance is evaluated based upon both the original model and the correlated alleles or “F” model. Similarly tested is the statistic that relies on STRUCTURE. Eigenanalysis is tested at three significance levels (). Structurama is assessed using both a noninformative prior on and the true value as the starting point. BAPS is evaluated using the individual clustering mode. Blank values in the table indicate that a program did not generate a result.