Phylogenetic analysis of combined LSU and SSU rRNA. Scale bars indicate number of substitutions per site. The dotted branches leading to eubacteria are arbitrary lengths. (*A*) Consensus tree of 16,000 trees obtained from the posterior distribution of an MCMC analysis with homogeneous composition across the tree ([GTR+Γ]x2) log_{e}(*L*_{m}) = −23,960.00. Nodes highlighted with dots were supported by ≥95% PP. The 3 values indicate support for a monophyletic archaebacteria from homogeneous composition MCMC (73% PP, Fig. S52 in *SI Appendix*), equally weighted maximum parsimony (95% BS, Fig. S53 in *SI Appendix*) and ML (55% BS, *SI Text* and Fig. S54 in *SI Appendix*). (*B*) Consensus tree of 10,000 trees obtained from the posterior distribution of an MCMC analyses with heterogeneous composition across the tree (NDCH model: [GTR+Γ+2CV]x2, log_{e}(*L*_{m}) = −23,507.36). The posterior predictive simulations of *X*^{2} for the NCDH model were: SSU: original statistic = 468.06, *P* = 0.3810 (range of simulated stat under the model = 186.57–1,014.93, mean = 449.61), LSU: original statistic = 759.69, *P* = 0.7515 (range of simulated statistic under the model = 475.46–1,589.75, mean = 845.98). By contrast, posterior predictive simulations of *X*^{2} for the homogeneous model were; SSU: range = 29.55–210.79, mean = 70.49, and LSU: range = 27.03–151.45, mean = 63.13. Nodes highlighted with dots were supported by >95% PP. The 2 values indicating support for the eocyte tree are posterior probabilities for the NDCH analyses as described (75% PP, Fig. S55 in *SI Appendix*), and for an MCMC analysis with the CAT model (95% PP, +Γ, Fig. S56 in *SI Appendix*).

## PubMed Commons