Ryan Brewer

A fundamental characteristic of biodiversity is its uneven distribution across the tree of life. This difference in diversity between clades has been attributed to either i) time, with older clades harbouring more species as these clades have had longer to accumulate species; ii) differential diversification rates between clades due to intrinsic properties of the clades, due to, for example, morphological, environmental, and/or anatomical traits; and iii) ecological limits to diversity, where limits to resource availability or carrying capacity within a system restricts diversity. However, analysing these evolutionary patterns and processes globally is extremely complex, so smaller geographic units are often used to study these processes: islands. Islands simplify com plex processes through their distinct boundaries, isolation, and compara tively simpler, but unique, biotas relative to continental systems. The Canary Islands represent one of the best-studied archipelago systems globally, with their biota studied in impressive detail, and one of the most species rich flora of any oceanic archipelago. This knowledge makes the Canary Islands an ideal arena to study the processes underlying biodiversity. In this research, I use a combination of phylogenomics, morphological and environmental trait data, and biogeographical modelling to study how the flowering plant (angiosperm) community of the Canary Islands is distributed in time, space, and across their phylogeny. To do so, I built a dataset identifying the number of flowering plant col onisations of the Canary Islands using published literature and the number of species per lineage. Using this dataset, I generated the most complete time-calibrated phylogenomic dataset of any archipelago flora and their continental ancestors globally, using 353 nuclear genes and time-calibrated using 166 fossils, representing ~60% of the Canary Islands non-monocot angi osperm community and ~65% of the colonisation events (Chapter 2). Using a dynamic, stochastic island biogeography model, we find strong evidence of differential diversification rates between lineages comprised primarily of dif ferent traits (trait-dependent diversification). Specifically, perennial lineages and lineages that evolved woodiness after island colonisation from a herba ceous ancestor (insular woodiness) have faster rates of speciation, and lower extinction and colonisation rates than annual lineages and lineages without insular woodiness. Interestingly, woody lineages have the same speciation rates as herbaceous lineages, but lower extinction rates. We also found some evidence of ecological limits as insular woody lineages had a detectable car rying capacity (i.e., as the number of species increases, rates of speciation and colonisation decrease), and the flora was found to be at macroevolution ary equilibrium. Interestingly, there was no relationship between clade age and diversity. As Chapter 2 found growth form to be an important aspect of the Canarian flora, we looked into which environmental variables affect the distribution of species with these different growth forms. We built a novel pipeline to extract, filter, and analyse publicly available spatial data to identify the environmen tal correlates of five growth forms (annual herbaceous, perennial herbaceous, ancestral woodiness, derived woodiness, and insular woodiness). We recon structed environmental niche space for these growth forms using a phyloge netic principal component analysis with a dataset of ~600 non-monocot angi osperm species. We found considerable overlap in niche space among these growth forms, and only minor differences between a combined woody and a combined herbaceous niche. Niche space is delineated by temperature and precipitation seasonality, and topographic complexity (rugosity). Using phy logenetic ANOVAs, we find significant pairwise differences between growth forms across five environmental variables, with rugosity having the most sig nificant differences between growth forms. From this data, insular woodiness appears to be an adaptation to a variable climate, allowing insular woody spe cies to persist across a wide range of climatic conditions, likely in response to the topographically complex areas they primarily inhabit. Drought-related variables and a low, non-freezing temperature in the coldest quarter are also contributing factors. I also built a trait dataset containing 24 morphological traits and environ mental variables to assess how traits have affected diversification (speciation minus extinction) rates. We selected two traits with high data coverage (life span and growth form) and combined these traits with the phylogenomic data from Chapter 2 to test a novel trait-dependent diversification model spe cifically tailored to islands, TRAISIE (TRait-dependent Assembly of Island biota through Speciation, Immigration, and Extinction). TRAISIE is a likeli hood-based framework that allows for the estimation of colonisation, speci ation, and extinction rates per trait state, as well as transition rates between states, using phylogenetic and trait data for an island community. We apply TRAISIE first to simulated data to show that for datasets containing >100 line ages and >200 species, TRAISIE can accurately estimate parameters and per form model selection, with accuracy increasing with dataset size. Next, we applied TRAISIE to the Canary Islands angiosperm dataset. We found that woody species and perennial species have faster speciation rates and lower extinction rates than herbaceous species and annual species, respectively. We also tested for confounding effects of these traits using a three-state model (annual herbaceous, perennial herbaceous, perennial woody), and found that woody species have the fastest speciation rates, perennial herbaceous the low est extinction rates, and annual herbaceous the fastest colonisation rates. In short, this thesis finds strong evidence of trait-dependent diversification operating in the Canary Islands, providing the first empirical data supporting the hypothesis that insular woodiness promotes diversification in the Canar ies; Canary plant distributions are shaped by climatic seasonality and rugos ity, both of which likely contribute to the evolution of insular woodiness; and provide a comprehensive trait dataset required to test a novel trait-dependent diversification model specifically tailored to islands, TRAISIE.