The harmonic mean of several Bayesian runs was not consistently higher for any partitioning scheme. Here we present the
results from a Bayesian analysis where all data were treated as a
single partition using the GTR+gamma+I model of sequence
evolution. The Posterior Probability (PP) values obtained from
this analysis differ onlymarginally fromany of the other Bayesian
analyses (data not shown). Inspection of all estimated parameter
values, including tree likelihood, plotted against generation,
showed convergence well before our set burn-in. The tree resulting from the Bayesian analysis had 18 ingroup nodes with PP
equal to or greater than 95%. This tree is fully congruent with the
single tree retrieved by the parsimony analysis.