A QUANTITATIVE ASSESSMENT OF CONGRUENCE BETWEEN MOLECULAR AND NONMOLECULAR ESTIMATES OF PHYLOGENY

An important means of assessing the validity of phylogenetic hypotheses is to measure congruence between different studies of the same group. We applied statistical methods to assess patterns of congruence among phylogenies taken from the literature, using strict-consensus and quartet statistics to measure congruence for 48 pairs of phylogenies of various groups of birds and mammals. The strict-consensus measures were higher on average for distance-distance comparisons than for distance-parsimony or parsimony-parsimony comparisons, and for molecular-molecular comparisons than for molecular-morphological or morphological-morphological comparisons. However, the only factor that was consistently important statistically was the number of taxa involved in the comparison, with congruence decreasing as the number of taxa increased. Quartet indices, which measure network similarity regardless of root position, also showed an effect based on number of taxa but exhibited no trend toward higher congruence in molecular-molecular and distance-distance comparisons. The number of studies with jointly unresolved quartets was small, indicating that data sets varied in the degree to which they could resolve relationships at different phylogenetic levels. The levels of congruence between phylogenies, including those employing single-copy nuclear DNA hybridization data, appear to be higher than expected in random sets of trees, cannot be explained by nonindependence of data sets, and thus provide empirical support for the validity of both distance and parsimony methods of phylogenetic inference. In specific instances, low congruence pointed to mistakes in the application of certain methods, and to the existence of problem taxa in need of additional study. © 1990 Springer-Verlag New York Inc.