Updated site concordance factors minimize effects of homoplasy and taxon sampling.

Abstract

Site concordance factors (sCFs) have become a widely used way to summarize discordance in phylogenomic datasets. However, the original version of sCFs was calculated by sampling a quartet of tip taxa and then applying parsimony-based criteria for discordance. This approach has the potential to be strongly affected by multiple hits at a site (homoplasy), especially when substitution rates are high or taxa are not closely related. Here, we introduce a new method for calculating sCFs. The updated version uses likelihood to generate probability distributions of ancestral states at internal nodes of the phylogeny. By sampling from the states at internal nodes adjacent to a given branch, this approach substantially reduces-but does not abolish-the effects of homoplasy and taxon sampling. Updated sCFs are implemented in IQ-TREE 2.2.2. The software is freely available at https://github.com/iqtree/iqtree2/releases. Supplementary information is available at Bioinformatics online.