Linking speciation and the niche: taxonomy, phylogeny and niche evolution in neotropical Begonia

Begonia L. is one of the ten largest plant genera with >1,830 currently accepted species distributed throughout the tropical Americas, Africa and Asia. This exceptional diversity offers the opportunity to study the processes underlying recent and rapid radiations, including the theory of adaptive radiations, but also poses major challenges. Methods to test adaptive radiation theory require sufficient occurrence data to produce species distribution models for the majority of species in a group. In many groups, including Andean, Central American and Mexican Begonia, this criterion is not met. The availability of specimen data on GBIF for species distribution modelling in vascular plants as a whole was assessed. The potential contribution of databased and freely available but indetermined or non-georeferenced specimens to the field of species distribution modelling was estimated. More than half of vascular plant species are currently unavailable for species distribution modelling. Our results indicate 22% of currently unavailable species could be modelled with already available herbarium data that is not yet georeferenced or determined to species. We argue that a greater focus on georeferencing and identification skills are vital if herbaria are to contribute fully to the growing field of species distribution modelling. We highlight those families and geographic regions that would most benefit from this approach and discuss the historical factors have influenced differences among regions. Another prerequisite for species distribution modelling is a stable species level taxonomy. In preparation for investigation involving SDM’s in Begonia, five taxonomic papers are presented, including revised species concepts, distribution data suitable for use in species distribution models, seven new species and a new section of Begonia. Three plastid markers for 574 species and 809 accessions of Begonia were used to produce the largest, most representative phylogeny of Begonia published to date. An updated sectional classification of the genus is provided. The sections of Begonia are used frequently as analogues to genera in other families but, despite their taxonomic utility, few of the current sections have been examined in the light of molecular phylogenetic analyses. The relationships among some species and sections are poorly resolved, but many sections and deeper nodes receive strong support. We recognise 77 sections of Begonia including four new sections: Astrotricha, Jackia, Kollmannia, and Stellanthera; five sections are reinstated from synonymy: Australes, Exalabegonia, Latistigma, Pereira and Poecilia; and four sections are newly synonymised. The new sectional classification is discussed with reference to identifying characters and previous classifications. Central to the theory of an adaptive radiation is that the majority of speciation events are adaptive, implying speciation driven by adaptation to different environments and resulting in niche shifts. Species in an adaptive radiation should display distinct ecological niches and environmental disparity across the group’s phylogeny should best fit a kappa ‘niche-shift’ or ‘speciational’ model of character evolution. These characteristics were tested in two clades of Neotropical Begonia. Species were compared through the comparison of their distributions in environmental space calculated from species distribution models. The fit of four models of environmental evolution (a kappa ‘niche-shift’ model, a Brownian motion model, an early-burst model, and a white noise model) to the observed niche disparity in the group were assessed using the Akaike Information Criterion Patterns of ecological diversification in the two clades of Begonia examined strongly diverge from those predicted by an adaptive model, and we conclude Begonia is not an adaptive radiation with respects to climatic niche. Ecological disparity within Begonia clades best fits the predictions of a Brownian model of niche evolution. The characteristics of a Brownian model include constant niche evolution over the phylogeny, no increases in niche evolution during speciation, and independent niche evolution in sister lineages. This is incompatible with the predictions of an adaptive radiation. We suggest the remarkable diversity of the genus has developed through geographic speciation, and subsequent adaptation to local environments. This phenomenon may be widespread among plant radiations in topographically- and environmentally-heterogeneous areas, and a re-evaluation of putative adaptive radiations throughout the world’s montane regions may be necessary