Assessing phylogenetic information content and redundancy in hominin craniodental traits

Abstract

Paleoanthropological phylogenetic inference is based on characters assumed to be phylogenetically informative and mutually independent, but our understanding of which traits fulfill these requirements is limited. We assess the cladistic information content (CIC) of each trait in a widely-used hominin craniodental character matrix (107 traits) and quantify the degree of unique phylogenetic information contributed by each trait, groups of traits from a given anatomical unit (AU), and operational taxonomic unit (OTU) using the branch and bound (BAB) algorithm. We devise metrics based on permutation of character values along the taxa to quantify the phylogenetic information from a given trait, and removal of individual characters or of an AU to test for redundancy of each trait or an AU, respectively. We quantify the amount of phylogenetic information contributed by a given OTU by measuring the degree of topological change in the inferred tree when that OTU is removed. We use Robinson-Foulds distances to quantify the topological distance between these modified trees to a baseline tree inferred using all traits and taxa. Out of 107 traits, we only identify 23 traits (21.5%) that contribute unique phylogenetic information. Additionally, eight traits are informative but redundant, and thus can be removed from the character matrix without affecting the inferred topology. No single unit contains sufficient information for accurate phylogenetic inference. Because the challenge in identifying phylogenetically informative characters is present throughout paleontological clades, our approach and findings should be more broadly applicable to other paleontological taxa.