PERFORMANCE OF LRI METHOD ON DETECTING TEMPO AND MODE OF TRAIT EVOLUTION ON PHYLOGENETIC TREES

Current approaches in macroevolution enable the calculation of rates of phenotypic change from both the fossil record and phylogenetic trees. However, the data used in these methods (fossil occurrences or the phylogenetic tree and corresponding traits) are not readily interchangable. One method developed to explore the tempo and mode of phenotypic evolution from the fossil record is the log-rate, log-interval (LRI) method. The LRI method allows for the simultaneous determination of mode (directional, random, or stasis) as well as the tempo of phenotypic evolution by computing all the pairwise differences from an evolutionary sequence. A recent study applies the LRI method to phylogenetic trait data to explore the effects of correcting for the non-independence of traits on measuring rates of phenotypic evolution. The author found phylogenetic independent contrasts return abnormally high rates of phenotypic evolution compared to the rate calculated from pairwise differences of tips. However, the conclusions of the study relied on a few empirical phylogenies, and it is unclear why rate calculated from the LRI method using independent contrasts produces a much higher extrapolated step rate compared to the one from intervals. Here we systematically explore the behavior of the LRI method on phylogenetic trait data on detecting tempo and mode. We simulated traits under various modes of evolution to test whether LRI plots could distinguish between modes of phylogenetic trait evolution: Brownian motion (BM), Ornstein-Uhlenbeck (OU), and early burst (EB) as well as directional, random and stasis in a phylogenetic context. We also investigated if the rates of phenotypic evolution differed when comparing tip differences with independent contrasts across these various evolutionary modes. For extreme parameter values, the LRI method was able to distinguish early burst, but is unable to differentiate other evolutionary modes. Additionally, rates of phenotypic evolution calculated from independent contrasts were not significantly elevated compared to those calculated from tip differences. These results suggest the LRI method should not be applied to phylogenetic trait data.