Evolutionary Shifts Associated with Substrate Endemism in the Western American Flora

This study investigated how habitat specialization affects the evolution and ecology of flowering plants. Specifically, a phylogenetic framework was used to investigate how trait evolution, lineage diversification, and biogeography of the western American flora are affected by two forms of substrate endemism: (1) edaphic specialization onto serpentine soils, and (2) host specialization of non-photosynthetic, holoparasitic Orobanchaceae. Previous studies have noted a correlation between presence on serpentine soils and a suite of morphological and physiological traits, one of which is the tendency of several serpentine-tolerant ecotypes to flower earlier than nearby closely related populations not growing on serpentine. A phylogenetically uncorrected ANOVA supports this hypothesis, developed predominantly through previously published comparisons of conspecific or closely related ecotypes. However, comparisons among three models of trait evolution, as well as phylogenetic independent contrasts across 24 independent clades of plants that include serpentine tolerant species in California and with reasonably resolved phylogenies, revealed no significant affect of flowering time in each of these genera. Taken together, these results suggest eco-evolutionary scale dependence of flowering time, and that flowering time may be an exaptation that facilitates colonization of serpentine habitats.To better understand the role that different substrates may play in diversification of the western American flora, additional studies were performed on a clade of parasitic plants in the Orobanchaceae endemic to the New World, long recognized as Orobanche sections Gymnocaulis and Nothaphyllon. First, a densely sampled phylogeny from specimens across the native geographic and host ranges of this clade was inferred using ribosomal nuclear DNA, a portion of the low-copy nuclear gene waxy, and three plastid regions. Several taxonomic species were found to be either non-monophyletic, or monophyletic but including several strongly supported sub-clades. In each case, these least-inclusive clades showed unique host associations, suggesting that host-switching may be an important driver of diversification. At a deeper phylogenetic scale, the monophyly of New World broomrapes, coupled with phylogenetic evidence that the Eurasian genus Diphelypaea is nested within Orobanche sensu lato supports the resurrection of the genus Aphyllon to include all New World taxa traditionally recognized as Orobanche and diagnosable by a five-toothed calyx. Finally, comparisons between the ranges of Aphyllon species and those of their hosts suggest that host specificity can also constrain biogeographic patterns in parasitic plants. Most strikingly, the closely related species A. ludovicianum and A. chilense both parasitize species of Grindelia in North and South America, respectively. Chronograms for both the tribe Orobancheae (Aphyllon and its holoparasitic relatives) and Grindelia were constructed using fossil and secondary calibration points, one of which was an inferred horizontal gene transfer event from an ancient leguminous host into the common ancestor of Orobanche and Phelipanche. These chronograms were used to reconstruct the historic biogeography using a dispersal-extinction-cladogenesis model. The dispersal of host and parasite lineages were both found to be from North America to South America, recapitulating a biogeographical pattern seen in many other lineages of land plants, with the dispersal of the Aphyllon lineage parasitic on Grindelia taking place somewhat after Grindelia began to diversify in South America. From a methodological standpoint, this study also demonstrates the general utility of using horizontal gene transfer events from well-dated clades to calibrate phylogenies in the absence of a fossil record