Table 1.

Comparison of log likelihoods, AIC and BIC values among different partitioning schemes (from 1 to 13 partitions)

Number of partitions*	number of parameters	LML	AIC	Δi	BICML
1	9	164,396	328,810	9894.730	328,825
2	19	162,890	325,819	6903.326	325,850
4A	39	163,108	326,295	7379.667	326,360
4B	39	161,931	323,941	5026.042	324,006
5	49	162,673	325,445	6529.681	325,527
13	129	159,328	318,915	0.000	319,131

For each type of analysis the following results are shown: total number of parameters; log likelihood calculated using RAxML (L_ML); AIC values; the difference in AIC values among model i and the best model (Δi = AICi - AICmin); BIC_ML values.

*1 partition = all dataset; 2 partitions = mitochondrial (16S + CytB) and nuclear (Myh6 + Rag1 + Rag2); 4 partitions A = 16S and 1st, 2nd, and 3rd codon position of protein coding genes; 4 partitions B = 16S + CytB and 1st, 2nd, and 3rd codon position of nuclear genes; 5 partitions = by each gene (16S + CytB + Myh6 + Rag1 + Rag2); 13 partitions = 16S + each codon position of each protein coding genes (1st, 2nd, and 3rd codon position of CytB; 1st, 2nd, and 3rd codon position of Myh6; 1st, 2nd, and 3rd codon position of Rag1; 1st, 2nd, and 3rd codon position of Rag2).