Communities and networks : examining ecological structures through a phylogenetic trait framework

Advances in phylogenetic scope and resolution present the opportunity to address questions about how phylogeny can inform not only our understanding of the past but our understanding of present and future ecologies. In this dissertation, I investigate the relationship between phylogenetic structure and patterns in community membership as well as interactions between species, advancing our understanding of eco-phylogenetic inference and the fundamental relationship between phylogenetic and network structure. I begin by considering how phylogeny and community membership relate to evolutionary histories underlying community assembly. Using simulations, I develop two techniques—one using a metric of clustering on a phylogeny, the other using machine learning—that predict the effective phylogeny of assembly-determining traits. These techniques advance the field by showing that the effective phylogenetic structure of traits is represented implicitly in the structure of organized data and can be estimated with straightforward methods. Using these techniques, I estimate an evolutionary trajectory of freeze tolerance in angiosperms consistent with historical glaciation. Continuing at the interface of phylogeny and machine learning, I next shift to patterns of pest-host interaction. Using a phenomenological model of host range expansion across a host phylogeny, I show that machine learning can predict the degree to which pests are phylogenetically constrained on their hosts. I apply this modeling approach to empirical data to describe the relative risk tree genera face from host range expansion events by different pests and pathogens. This work advances measures of phylogenetic constraint by explicitly incorporating them into the process of host range expansion. Finally, I explore the relationship between phylogeny and patterns of species interaction under the lens of network ecology. Using an extensive empirical dataset of networks of frugivory, I performed Bayesian analysis to examine how local evolutionary distinctiveness, a measure of how little evolutionary history a species shares with others in its community, is associated with patterns of interaction. I find that low local evolutionary distinctiveness is associated with more specialized partners. To build on this novel finding, I used mathematical approaches to show that such a pattern should exist when species interact by trait-matching on traits with phylogenetic signal.