Maximum-likelihood analyses were conducted under the GTR molecular substitution model.
For choosing the molecular substitution model we analysed the data using MrModelTest v2 [ 98], and based on the Akaike Information Criterion, the most parameter-rich model GTR+G+I was suggested.
To calculate the fitness of each tentative solution, we used the matrix of genetic distances among samples, corrected with the best fit molecular substitution model (GTR+ Γ +I) [ 5, 18].
To do this we used an already published tree as starting point [ 12], and a matrix of genetic distances among samples, corrected with the best fit molecular substitution model, in the fitness computation.
Second, molecular dating requires accurate substitution models for the genes under consideration over billion-year time-scales as well as models that account for substitution rate variation across the tree branches [ 1, 2, 19, 20].
Table 1 lists the parameter estimates and log-likelihood values for the eight codon substitution models of molecular evolution.
We unlinked the substitution models and molecular clock for the all-loci and nuclear gene tree models, and linked the two mitochondrial tree models.
For a better understanding of evolution, it is of interest whether these mutation regularities are human specific or present in other vertebrates, as their presence might affect the validity of currently used substitution models and molecular clocks.
A previous study of Lagenorhynchus cyt b [ 12] suggests the evolution of this locus is clock-like; therefore dates of divergence were calculated under the assumption of a molecular clock with substitution model parameters predetermined by Modeltest.
Dispersion (an indicator of spread) is assessed on the one-dimensional Euclidean distance between the raw pair-wise sequence distances and those corrected according to a Jukes-Cantor model of molecular substitution.
