Dinosaurs Still in Long-Term Net Speciation Decline Before the K-Pg Boundary

Abstract

A recent study in Nature Communications used ecological niche modelling (ENM) to project suitable habitat for dinosaurs during the latest stages of the Cretaceous Period (83-66 million years [Myr] ago) from detailed climate data, apparently refuting the hypothesis that dinosaurs were in a long-term decline before the Cretaceous-Paleogene (K-Pg) mass extinction event. However, we note here that: 1) suitable habitat does not necessarily equate to taxonomic diversity (i.e., number of species); and 2) lack of a decline in diversity across time intervals is not evidence against a gradual decline in net speciation through time – diversity and net speciation measure two separate things. Diversity measures the number of species within discrete time intervals, while phylogenetically inferred net speciation measures the dynamic relationship between speciation and extinction through time. Additionally, net speciation is estimated in a model framework that accounts for statistically undesired effects of shared ancestry, whereas diversity is not. Thus, diversity and net speciation are not directly comparable, and a lack of a decline in diversity does not necessarily refute a decline in net speciation through time.

Strong support for a heterogeneous speciation decline model in Dinosauria: a response to claims made by Bonsor et al . (2020)

“…Recently, Bonsor et al [1] criticized our selection of a model of diversification in dinosaurs that supported a long-term decline in the rate at which speciation events accumulated across the dinosaur tree of life through time (time-quadratic model) [2]. Core to their criticism, Bonsor et al [1] applied our analytical approach to nine additional trees that were published subsequently to our 2016 paper [2], claiming that the timequadratic model is not as well supported as we originally reported. They go on to list several criticisms of our choices in selecting the time-quadratic model as the preferred model of diversification.…”

“…In order to test three alternative hypotheses relating to how nodes accumulate (fig. 1 from [2] but also reproduced as fig. 1 in [1]), we compared three models, each differing in how the temporal effects were modelled as the predictor variables: Model A, the time-linear (null) model, N Nodes = Time; Model B, the time-square root model, N Nodes = √Time; and Model C, the time-quadratic model, N Nodes = Time + Time 2 .…”

“…Thus, we fitted six models in total: Model A-Dino, Dinosauria time-linear model; Model B-Dino, Dinosauria time-square root model; Model C-Dino, Dinosauria time-quadratic model; Model A-5G, 5group time-linear model; Model B-5G, 5-group time-square root model; and Model C-5G, 5-group time-quadratic model. In actuality, we fitted and compared various other iterations and grouping structures (such as the 3-group model), but these six are the models relevant to our discussion here and in our original paper [2].…”

“…We compared model fit using deviance information criterion (DIC) and selected the model with the lowest DIC score, at least 4 less than the next lowest DIC score, as the preferred model [2]. We found that Model C-5G had the lowest DIC score and we thus selected this model as the preferred model.…”

“…Additionally, we fitted and compared the three models on a sample of 100 meta-trees of dinosaurs (N = 961) [3] and demonstrate that the group-wise time-quadratic model is still strongly supported over the time-square root model and time-linear model. Sakamoto et al (2016) First, we will clarify the models we fitted and compared in our 2016 study [2]. We modelled how speciation events accumulated across phylogeny through time by taking the number of nodes from the root to each tip of the phylogeny (N Nodes ) as the response variable in a phylogenetic Poisson model, fitted through a Bayesian generalized linear mixed model (MCMCglmm [4]).…”