Table 4.

The number of subsets selected using the iterative k -means algorithm for 20 simulated alignments in which 2 independent subsets were simulated

Simulation replicate	k -means subsets	Number of subsets consisting of > =95% sites from same model
1	14	12
2	12	12
3	16	15
4	14	13
5	13	12
6	16	14
7	15	15
8	13	12
9	14	13
10	13	12
11	14	12
12	15	13
13	14	12
14	12	11
15	14	13
16	15	13
17	14	14
18	24	20
19	13	12
20	14	13