Figure 4

The appearance of phylogenetic artifacts in the RPB1 phylogeny due to "long-branch attraction" (LBA), "short-branch exclusion" (SBE), and CTD-based functional constraint on evolutionary change. A. Adapted from Felsenstein's original four taxon demonstration of "long-branch attraction" [66], this tree shows graphically the phenomena of LBA and SBE. A combination of large differences in substitution probabilities among branches, combined with short internodes, leads to artificial grouping of the more rapidly evolving sequences. This, by definition, results in an additional artificial grouping of shorter branches, at some level, which we call "short-branch exclusion." B. Unrooted phenogram of RPB1 tree from figure 2 more graphically demonstrating the large variation in inferred substitution probabilities in terminal branches, along with the generally short internodes throughout the tree. The four most basal lineages (as viewed in the rooted phenogram in figure 2) are consistent with a LBA artfifact, while the presumably artificial clustering of Glaucocystis, Acanthamoeba and Dictyostelium is most consistent with SBE. Both clades are highlighted in gray. C. Topological features of the global RPB1 tree that are consistent with the three kinds of artifacts discussed. With the complete data set, only the extreme long-branch features of the four most rapidly evolving basal sequences are obvious (Figure 3). With subsequent rounds of analysis, in each case removing the most long-branch sequences identified in the prior round, the branching pattern of subsections of the tree are shown to be consistent with one of the artifacts.