The implications of interrelated assumptions on estimates of divergence times and rates of diversification

A timescale for placental mammal diversification based on Bayesian modeling of the fossil record

The timing of the placental mammal radiation has been the focus of debate over the efficacy of competing methods for establishing evolutionary timescales. Molecular clock analyses estimate that placental mammals originated before the Cretaceous-Paleogene (K-Pg) mass extinction, anywhere from the Late Cretaceous to the Jurassic. However, the absence of definitive fossils of placentals before the K-Pg boundary is compatible with a post-Cretaceous origin. Nevertheless, lineage divergence must occur before it can be manifest phenotypically in descendent lineages. This, combined with the non-uniformity of the rock and fossil records, requires the fossil record to be interpreted rather than read literally. To achieve this, we introduce an extended Bayesian Brownian bridge model that estimates the age of origination and, where applicable, extinction through a probabilistic interpretation of the fossil record. The model estimates the origination of placentals in the Late Cretaceous, with ordinal crown groups originating at or after the K-Pg boundary. The results reduce the plausible interval for placental mammal origination to the younger range of molecular clock estimates. Our findings support both the Long Fuse and Soft Explosive models of placental mammal diversification, indicating that the placentals originated shortly prior to the K-Pg mass extinction. The origination of many modern mammal lineages overlapped with and followed the K-Pg mass extinction.

The Implications of Incongruence between Gene Tree and Species Tree Topologies for Divergence Time Estimation

Phylogenetic analyses are increasingly being performed with data sets that incorporate hundreds of loci. Due to incomplete lineage sorting, hybridization, and horizontal gene transfer, the gene trees for these loci may often have topologies that differ from each other and from the species tree. The effect of these topological incongruences on divergence time estimation has not been fully investigated. Using a series of simulation experiments and empirical analyses, we demonstrate that when topological incongruence between gene trees and the species tree is not accounted for, the temporal duration of branches in regions of the species tree that are affected by incongruence is underestimated, whilst the duration of other branches is considerably overestimated. This effect becomes more pronounced with higher levels of topological incongruence. We show that this pattern results from the erroneous estimation of the number of substitutions along branches in the species tree, although the effect is modulated by the assumptions inherent to divergence time estimation, such as those relating to the fossil record or among-branch-substitution-rate variation. By only analyzing loci with gene trees that are topologically congruent with the species tree, or only taking into account the branches from each gene tree that are topologically congruent with the species tree, we demonstrate that the effects of topological incongruence can be ameliorated. Nonetheless, even when topologically congruent gene trees or topologically congruent branches are selected, error in divergence time estimates remains. This stems from temporal incongruences between divergence times in species trees and divergence times in gene trees, and more importantly, the difficulty of incorporating necessary assumptions for divergence time estimation. [Divergence time estimation; gene trees; species tree; topological incongruence.].

exTREEmaTIME: a method for incorporating uncertainty into divergence time estimates

We present a method of divergence time estimation (exTREEmaTIME) that aims to effectively account for uncertainty in divergence time estimates. The method requires a minimal set of assumptions, and, based on these assumptions, estimates the oldest possible divergence times and youngest possible divergence times that are consistent with the assumptions. We use a series of simulations and empirical analyses to illustrate that exTREEmaTIME is effective at representing uncertainty. We then describe how exTREEmaTIME can act as a basis to determine the implications of the more stringent assumptions that are incorporated into other methods of divergence time estimation that produce more precise estimates. This is critically important given that many of the assumptions that are incorporated into these methods are highly complex, difficult to justify biologically, and as such can lead to estimates that are highly inaccurate. This article has an associated First Person interview with the first author of the paper.