Quantifying and visualising divergence between pairs of phylogenetic trees: implications for phylogenetic reconstruction

Abstract

ABSTRACTTree topologies are a primary output of phylogenetic analysis. It is important to understand how such outputs are influenced by the choice of phylogenetic method, and the nature and quality of input data. This often entails the measurement of how closely an output corresponds to an ideal tree topology. The symmetric difference (partition) metric is the only widely-used measure that is defined for trees that contain polytomies, but expresses undesirable behaviour in certain situations.Here I propose a divergence metric for unweighted tree topologies based on quartet statements, which measures the topological information common to two trees. This metric reflects the contributions of both accuracy and precision to tree quality; these components can be decomposed through the use of ternary diagrams. This approach is implemented in a new package for the R statistical environment, and applied to a recent controversy, where it provides a new perspective on the relative merits of Bayesian and parsimony approaches to morphological phylogenetics.