Measuring phylogenetic signal between categorical traits and phylogenies

Motivation: Determining whether a trait and phylogeny share some degree of phylogenetic signal is a flagship goal in evolutionary biology. Signatures of phylogenetic signal can assist the resolution of a broad range of evolutionary questions regarding the tempo and mode of phenotypic evolution. However, despite the considerable number of strategies to measure it, few and limited approaches exist for categorical traits. Here, we used the concept of Shannon entropy and propose the δstatistic for evaluating the degree of phylogenetic signal between a phylogeny and categorical traits. Results: We validated δas a measure of phylogenetic signal: the higher the δ-value the higher the degree of phylogenetic signal between a given tree and a trait. Based on simulated data we proposed a threshold-based classification test to pinpoint cases of phylogenetic signal. The assessment of the test's specificity and sensitivity suggested that the δapproach should only be applied to 20 or more species. We have further tested the performance of δin scenarios of branch length and topology uncertainty, unbiased and biased trait evolution and trait saturation. Our results showed that δmay be applied in a wide range of phylogenetic contexts. Finally, we investigated our method in 14 360 mammalian gene trees and found that olfactory receptor genes are significantly associated with the mammalian activity patterns, a result that is congruent with expectations and experiments from the literature. Our application shows that δcan successfully detect molecular signatures of phenotypic evolution. We conclude that δrepresents a useful measure of phylogenetic signal since many phenotypes can only be measured in categories. Availability and implementation: https://github.com/mrborges23/delta-statistic. Supplementary information: Supplementary data are available at Bioinformatics online.