UC Berkeley

Key evolutionary transitions have shaped the tree of life by driving the processes of speciation and extinction. This dissertation aims to advance statistical and computational approaches that model the timing and nature of these transitions over evolutionary trees. These methodological developments in phylogenetic comparative biology enable formal, model-based, statistical examinations of the macroevolutionary consequences of trait evolution. Chapter 1 presents computational tools for data mining the large-scale molecular sequence datasets needed for comparative phylogenetic analyses. I describe a novel metric, the missing sequence decisiveness score (MSDS), which assesses the phylogenetic decisiveness of a matrix given the pattern of missing sequence data. In Chapter 2, I introduce a class of phylogenetic models of chromosome number evolution that accommodate both anagenetic and cladogenetic change. The models reveal the mode of chromosome number evolution; is chromosome evolution occurring primarily within lineages, primarily at lineage splitting, or in clade-specific combinations of both? Furthermore, these models permit estimation of the location and timing of possible chromosome speciation events over the phylogeny. Finally, Chapter 3 demonstrates a new method of stochastic character mapping for state-dependent speciation and extinction models and applies it to test the impact of plant mating systems on the extinction of lineages. This approach estimates the timing of both character state transitions and shifts in diversification rates over the phylogeny. Confirming long standing theory, I found that self-compatible lineages have higher extinction rates and lower net diversification rates compared to self-incompatible lineages. Additionally, the method shows that the loss of self-incompatibility is followed by a short-term spike in speciation rates, which declines after a time lag of several million years resulting in evolutionary decline.