Strong effects of variation in taxonomic opinion on diversification analyses

Abstract

Summary The effects of variation in taxonomic opinion between ‘splitters’ and ‘lumpers’ have been debated intensely in conservation biology and have also been discussed in macroecology. However, the impact on diversification analyses has received little attention and has largely been ignored by many end‐users of macroevolutionary analyses. Using simulated data sets, we analysed the effects of variation in taxonomic opinion on lineage diversification analysis, focusing on two measures of phylogenetic asymmetry (β and IC), a measure of the relative age composition of nodes (γ), and the MEDUSA algorithm, which searches for clades with unique diversification histories. All measures were biased by variation in taxonomic opinion, but to different degrees. The γ‐estimate and the MEDUSA algorithm were found to be especially sensitive, with error rates high enough to make it likely that false increases in diversification rates over time or clades with apparently, but not actually, unique diversification histories would be present in many published analyses. In contrast, for the two measures of phylogenetic asymmetry the biases were likely too small to substantially change conclusions in studies ignoring variation in taxonomic opinion. Our results highlight that variation in taxonomic opinion can be an important source of error in diversification analyses. We therefore suggest that all end‐users of diversification analyses should consider whether the approaches used are sensitive to this issue and, if so, whether any of their results could be a consequence of different taxonomic treatment of different lineages rather than real biological differences.