A model with many small shifts for estimating species-specific diversification rates
Abstract
Understanding how and why diversification rates vary through time and space and across species groups is key to understanding the emergence of today’s biodiversity. Phylogenetic approaches aimed at identifying variations in diversification rates during the evolutionary history of clades have focused on exceptional shifts subtending evolutionary radiations. While such shifts have undoubtedly affected the history of life, identifying smaller but more frequent changes is important as well. We developed ClaDS—a new Bayesian approach for estimating branch-specific diversification rates on a phylogeny that relies on a model with changes in diversification rates at each speciation event. We show, using Monte Carlo simulations, that the approach performs well at inferring both small and large changes in diversification. Applying our approach to bird phylogenies covering the entire avian radiation, we find that diversification rates are remarkably heterogeneous within evolutionarily restricted species groups. Some groups such as Accipitridae (hawks and allies) cover almost the full range of speciation rates found across the entire bird radiation. As much as 76% of the variation in branch-specific rates across this radiation is due to intraclade variation, suggesting that a large part of the variation in diversification rates is due to many small, rather than few large, shifts.
What the paper shows and why it matters
(AI-generated)
Most methods that look for variation in diversification rates along phylogenies (family trees of species) have focused on a few exceptional shifts, in which one lineage starts a radiation. Such shifts have surely shaped the history of life, but the paper argues that smaller and more frequent changes in rate matter as well. It therefore asks how much of the variation in diversification rates comes from many small changes rather than a few large ones.
The paper argues that much of the variation in diversification rates among birds arises from many small shifts within closely related groups, and it introduces ClaDS, a Bayesian method, to show this. In the underlying model the speciation rate of each lineage changes at every speciation event, and the method estimates a rate for every branch. Simulations indicate that it recovers both small and large changes reliably on average, although extreme rates are somewhat smoothed. Applied to 42 bird clades (mostly families), rates are very heterogeneous within groups: on a logarithmic scale, Accipitridae (hawks and allies) cover a large part of the range of speciation rates across all birds, and 76% of the variation in branch-specific rates lies within groups. A method built around discrete shifts would place only 46% of the variance within groups, which the paper takes to mean that many small shifts escape such methods.
The paper reads these results as suggesting that a large part of the variation in diversification rates is due to many small rather than a few large shifts. Small, frequent changes are then important alongside exceptional shifts.