Molecular polytomies

Abstract

This paper focuses on polytomies, especially molecular polytomies. The distinction between molecular and species polytomies is important, but is often not made. Likelihood ratio tests are an easier method for detecting molecular polytomies than other methods cited herein. Simulation shows that parsimony will generally falsely resolve molecular polytomies, which is worrisome because a simple mathematical model described herein predicts that molecular polytomies will occur often when the mean branch length is small. A test of the model using several real molecular data sets indicates that molecular polytomies may actually occur more often than predicted by the model. This suggests that at least some published molecular parsimony trees contain clades that are false resolutions of polytomies. Finally, a possible method for detecting species polytomies from molecular data is described.