Locally adaptive Bayesian birth-death model successfully detects slow and rapid rate shifts

Abstract

Birth-death processes have given biologists a model-based framework to answer questions about changes in the birth and death rates of lineages in a phylogenetic tree. Therefore birth-death models are central to macroevolutionary as well as phylodynamic analyses. Early approaches to studying temporal variation in birth and death rates using birth-death models faced difficulties due to the restrictive choices of birth and death rate curves through time. Sufficiently flexible time-varying birth-death models are still lacking. We use a piecewise-constant birth-death model, combined with both Gaussian Markov random field (GMRF) and horseshoe Markov random field (HSMRF) prior distributions, to approximate arbitrary changes in birth rate through time. We implement these models in the widely used statistical phylogenetic software platform RevBayes, allowing us to jointly estimate birth-death process parameters, phylogeny, and nuisance parameters in a Bayesian framework. We test both GMRF-based and HSMRF-based models on a variety of simulated diversification scenarios, and then apply them to both a macroevolutionary and an epidemiological dataset. We find that both models are capable of inferring variable birth rates and correctly rejecting variable models in favor of effectively constant models. In general the HSMRF-based model has higher precision than its GMRF counterpart, with little to no loss of accuracy. Applied to a macroevolutionary dataset of the Australian gecko family Pygopodidae (where birth rates are interpretable as speciation rates), the GMRF-based model detects a slow decrease whereas the HSMRF-based model detects a rapid speciation-rate decrease in the last 12 million years. Applied to an infectious disease phylodynamic dataset of sequences from HIV subtype A in Russia and Ukraine (where birth rates are interpretable as the rate of accumulation of new infections), our models detect a strongly elevated rate of infection in the 1990s.

The rise of grasslands is linked to atmospheric CO2decline in the late Palaeogene

“…The support of the uncorrelated model over the two autocorrelated models (GMRF and HSRMF), although the autocorrelated models were favored when using time-varying diversification rates without environmental variables (Fig. S7), could stem from the use of vague prior distribution which allows for more rate variation in autocorrelated models 21 . However, regardless of the specific environmentally-dependent diversification model, we inferred a negative correlation between diversification rates and environmental CO 2 (Fig.…”

“…We develop a novel Bayesian approach for detecting diversification-rate shifts that incorporates a more realistic (non-uniform) model of species sampling and implemented it in the open-source software RevBayes 17 . Our model builds on the episodic birth-death process, where speciation and extinction rates are constant within an interval but may shift instantly to new rates at a rate-shift episode [18][19][20][21] . Furthermore, we tested for a correlation between diversification rate and two environmental variables -atmospheric CO 2 concentration and average global paleo-temperature-using one existing [22][23][24][25][26][27] and three new environmentally-dependent diversification models.…”

Impact of K-Pg Mass Extinction Event on Crocodylomorpha Inferred from Phylogeny of Extinct and Extant Taxa

“…tips (Wilberg et al 2019;Oaks 2011). To account for the possible effect of background rate variation, and for the possibility that a mass extinction is followed by a rapid burst of speciation (Crisp and Cook 2009), we employed time-varying priors on the speciation, extinction, and sampling rates through time (Magee et al 2020). We computed Bayes factors to test for the signal of a mass extinction event anytime along the phylogeny.…”

“…To account for background variation in the rates of speciation, extinction, and sampling, we apply horseshoe Markov random field (HSMRF) prior distributions (Magee et al 2020), which have been shown to be able to both detect rapid shifts in speciation rates and to reject time-varying models in favor of effectively constantrate models. In a HSMRF model, we must specify prior distributions on the initial rates λ 1 , µ 1 , and φ 1 .…”

“…We use a reversible-jump Markov chain Monte Carlo algorithm (Green 1995) to infer whether a mass extinction event has impact our study group or not. We use Horseshoe Markov random field (HSMRF) smoothing priors (Magee et al 2020) on speciation, extinction and fossilization rates to allow for abrupt changes while favoring constant rates over rate variation (Occam's razor principle). We infer an overwhelming support for a mass extinction event about 68 Mya.…”