Abstract

Reconstructing the tree of life involves more than identifying relationships among lineages; it also entails accurately estimating when lineages diverged. Paleontologists typically scale cladograms to time a posteriori by direct reference to first appearances of taxa in the stratigraphic record. Some approaches use probabilistic models of branching, extinction, and sampling processes to date samples of trees, such as the recently developed cal3 method, which stochastically draws divergence dates given a set of rates for those processes. However, these models require estimates of the rates of those processes, which may be hard to obtain, particularly for sampling. Here, we contrast the use of cal3 and other a posteriori time-scaling approaches by examining a previous study that documented a decelerating rate of morphological evolution in pterocephaliid trilobites. Although aspects of the data set make estimation of branching, extinction, and sampling rates difficult, we use a multifaceted approach to calculate and evaluate the rate estimates needed for applying cal3. In agreement with previous simulation studies, we find that the choice of phylogenetic dating method impacts downstream macroevolutionary conclusions. We also find contradictory evolutionary inferences between analyses on ancestor–descendant contrasts (based on ancestor trait reconstruction methods) and maximum-likelihood parameter estimates. Ancestral taxon inference in cal3 corroborates previously hypothesized ancestor–descendant sequences, but cal3 suggests greater support for budding cladogenesis than anagenesis. This case study demonstrates the potential and wide applicability of the cal3 method and the benefits afforded by choosing cal3 over simpler a posteriori time-scaling approaches.

You do not currently have access to this article.