The phylogenetic analyses using parsimony and maximum likelihood supported the results from Bayesian inference on the combined five-locus dataset.
The first model is the null model without adaptive evolution and the second model is the alternative model with adaptive evolution, so a significant improvement in maximum likelihood supports positive selection.
Maximum-likelihood support was calculated for the trees inferred from the three data supermatrices.
Maximum-likelihood support for the Bayesian maximum-clade-credibility trees was estimated using 1,000 bootstrap replicates.
Maximum-likelihood support for the inferred topology was generated from 100 bootstrap replicates of the data with PhyML3.0 (63) (LG matrix; GTR+Γ+I, as described above).
For a small number of nodes, notably within Pitheciidae, there was no maximum-likelihood support for the nodes estimated using Bayesian analysis.
As larger amounts of data were incorporated into the analyses, however, this picture was modified – using complete 12S rRNA and cytochrome b sequences, Tougard et al. [ 8] found high support (maximum likelihood bootstrap support of 97%) for the phylogeny as outlined by geography (although they could not, using a Kishino-Hasegawa test, reject outright the horn topology).
Indeed, phylogenetic analysis of the EF1 family with bacterial SelB and other long branches subgroups excluded shows a relationship of EF-Tu with a/eEF1A to the exclusion of aIF2g and aSelB, albeit with low statistical support (maximum likelihood bootstrap support of 51%, Additional file 4).
Support levels are shown for each node; from left-to-right: Maximum Parsimony bootstrap support values (>70), Maximum Likelihood bootstrap support values (>70), Bayesian posterior probabilities (>90), respectively.
