Figure 4. Phoronid phylogeny. Bayesian relaxed clock analysis of rDNA sequences. At each node is shown the relative node height, the 95% highest posterior density (HPD) limits, and the node label. Labels at brachiopod nodes are as used previously (Cohen & Weydmann, 2005). The 24tx alignment file was loaded into the utility BEAUTI to prepare an xml input file for BEAST and taxon groups corresponding to nodes B, D, F, M, N, and R were defined. The general time reversible + gamma + invariant sites model was selected, with four gamma categories and the mean substitution rate was fixed = 1.0 (the default). The uncorrelated lognormal clock rate variation model was selected, with all priors set to 'Uniform'. Operators that would change tree topology were switched off, and auto-optimize selected. The xml file was edited to include the maximum likelihood starting tree, which had been exported from PAUP*4 in Newick format with branch lengths. Multiple analyses were run with closely comparable results; the particular analysis shown was stopped after 1.3 ¥ 107 generations with default data-logging, when inspection of the log file in TRACER revealed that all parameter distributions had smooth, quasi-normal curves and the lowest effective sample size was> 450. The best resulting tree was selected in TreeAnnotator, with 10% discarded as burnin, displayed in FigTree as relative heights (root = 1.0) with 95% HPD node bars, and the graphics file exported for editing. Diagonal lines mark a truncated node bar.