Fig. 5.

Cassiopeia builds highly accurate trees from large empirical datasets. The consistency between tree reconstructions is evaluated with respect to the first split. The mean majority vote (a) and the meta purity test (b) were used for Cassiopeia-Hybrid and Cassiopeia-Greedy (both with or without priors) and neighbor joining. The statistics are plotted as a function of the number of clades at the depth of the test (i.e., the number of clades created by a horizontal cut at a given depth). All Cassiopeia approaches consistently outperform neighbor joining by both metrics