Skeletons in the Forest: Using Entropy-Based Rogue Detection on Bayesian Phylogenetic Tree Distributions

Abstract

In a phylogenetic analysis, rogue taxa and clades are those that, due to their uncertain placement, contribute disproportionally to the variance in a set of phylogenetics trees. They can obscure clear evolutionary relationships and complicate the interpretation of results. While existing rogue detection algorithms focus on improving the consensus tree for a set of trees, we introduce algorithms based on phylogenetic entropy to improve the analysis in a Bayesian framework. In particular, we introduce a tool that extracts a skeleton distribution of the posterior tree distribution that contains the core relationships by removing a minimal subset of rogues. Additionally, we enable detailed analyses of rogues at both the taxon and clade levels, including the visualisation of different rogue placements.