The fossilized birth-death model for the analysis of stratigraphic range data under different speciation concepts

TL;DR

This paper introduces a comprehensive mathematical framework for the fossilized birth-death model that explicitly incorporates stratigraphic ranges and different speciation modes, enabling more accurate phylogenetic and epidemiological analyses.

Contribution

It extends the FBD model to account for stratigraphic ranges and multiple speciation modes, facilitating joint Bayesian analysis of fossil and extant data, and applicable to epidemiology.

Findings

Derived expressions for three speciation modes.

Framework supports Bayesian analysis of fossil and living data.

Applicable to epidemiological modeling of infection durations.

Abstract

A birth-death-sampling model gives rise to phylogenetic trees with samples from the past and the present. Interpreting "birth" as branching speciation, "death" as extinction, and "sampling" as fossil preservation and recovery, this model -- also referred to as the fossilized birth-death (FBD) model -- gives rise to phylogenetic trees on extant and fossil samples. The model has been mathematically analyzed and successfully applied to a range of datasets on different taxonomic levels, such as penguins, plants, and insects. However, the current mathematical treatment of this model does not allow for a group of temporally distinct fossil specimens to be assigned to the same species. In this paper, we provide a general mathematical FBD modeling framework that explicitly takes "stratigraphic ranges" into account, with a stratigraphic range being defined as the lineage interval associated with a single species, ranging through time from the first to the last fossil appearance of the species. To assign a sequence of fossil samples in the phylogenetic tree to the same species, i.e., to specify a stratigraphic range, we need to define the mode of speciation. We provide expressions to account for three common speciation modes: budding (or asymmetric) speciation, bifurcating (or symmetric) speciation, and anagenetic speciation. Our equations allow for flexible joint Bayesian analysis of paleontological and neontological data. Furthermore, our framework is directly applicable to epidemiology, where a stratigraphic range is the observed duration of infection of a single patient, "birth" via budding is transmission, "death" is recovery, and "sampling" is sequencing the pathogen of a patient. Thus, we present a model that allows for incorporation of multiple observations through time from a single patient.