Beyond the Binary: Generalised Trees for Phylodynamics

Abstract

Phylodynamic methods leverage molecular and morphological data to reconstruct phylogenetic trees and understand the underlying evolutionary, epidemiological, or environmental processes that shaped them. However, many of such processes violate the assumption, prevalent among these methods, of strictly bifurcating, non-oriented trees. Here, I address this issue by generalising the existing Bayesian phylodynamic models to include non-binary topology elements or oriented branching events, or both. First, I expand the coalescent-based reticulate evolution model for segmented viruses to account for population structure as defined by, for example, geographic separation of host-type. Using the avian influenza example, I then present a framework to analyse the interaction between reticulation and host-type switching events. Next, inspired by previous work in multi-species coalescent, I tackle the problem of discordance between the genomic trees of a pathogen and its directed transmission pathway. The accuracy of this method is demonstrated by a simulation study and the reconstruction of a well-established HIV transmission chain. Finally, this thesis describes the implementation of a model that accounts for the known temporal range of an individual and the branching direction in a birth-death process. The wide applicability of this model is shown by analysing two macroevolutionary datsets, consisting of fossil and extant taxa, as well as inferring the transmission trees for slow and fast evolving pathogens. Overall, the aim of this thesis is to expand the phylodynamic method toolkit for researchers in macroevolution and epidemiology.