Patterns of Trait and Range Size Variation in an Evolutionary Radiation: The Role of Environmental Gradients

Current threats to biodiversity are driving a research focus on understanding how environmental changes influence patterns of biodiversity at global and continental scales, deemphasizing research done at regional scales. I examined the role of environmental gradients in driving diversity patterns in the Greater Cape Floristic Region (GCFR) of South Africa, a biodiversity hotspot, renowned for its floristic diversity and high endemism levels. The plant genus Pelargonium is centered in the winter-rainfall region of the GCFR, and varies considerably in morphology and growth form: remarkably, leaf size varies over 15000-fold across species. Pelargonium comprises 6 well-supported subclades that have diversified within the last 30 Ma. This diversity is arrayed along steep environmental gradients of temperature and amount and seasonality of rainfall. Chapter 1 examined trait-trait and trait-environment associations across ~120 Pelargoniumspecies. Bayesian phylogenetic mixed-effects models revealed strong associations between climate and functional traits: height, leaf area, leaf mass per area (LMA), and leaf nitrogen content. Relationships between environments and these traits were subclade-dependent, i.e., subclades differed significantly in the signs of their trait-environment associations. Chapter 2 explored how environmental gradients shape trait variation within the widespread species, Pelargonium scabrum. A Bayesian multiple-response multiple-regression model showed that P. scabrum leaves tended to decrease in size, effective leaf width and LMA with increasing winter minimum temperatures, suggestive of local adaptation. Trait-environment associations were influenced by rainfall seasonality, and by changing covariation between environmental variables across the range. Chapter 3 evaluated niche breadth-range size relationships in Pelargonium. I generated Bayesian models that simulated random environments and preserved the spatial structure of the original temperature and precipitation climate layers. Observed niche breadth-range size relationships were no stronger than expected based on 999 random temperature and precipitation variables. Spatial autocorrelation reduces estimates of niche breadth, positively biasing estimates of niche breadth-range size relationships. Overall, I demonstrate that evolutionary history matters for trait-environment associations at regional scales, and that local adaptation is mediated by the covariation among climate variables, and I show that spatial autocorrelation leads to a bias in estimates of niche breadth-range size relationships, and needs to be considered before inferring any causal relationship