Inferring geographic range evolution of a pantropical tribe in the coffee family (Lasiantheae, Rubiaceae) in the face of topological uncertainty.

In this study we explore what historical biogeographic events are responsible for the wide and disjunct distribution of extant species in Lasiantheae, a pantropical group of trees and shrubs in the coffee family. Three of the genera in the group, Lasianthus, Saldinia, and Trichostachys, are found to be monophyletic, while there are indications that the fourth, Ronabea, is paraphyletic. We also address how the uncertainty in topology and divergence times affects the level of confidence in the biogeographic reconstruction. A data set consisting of chloroplast and nuclear ribosomal DNA data was analyzed using a Bayesian relaxed molecular clock approach to estimate phylogenetic relationships and divergence times, and the dispersal-extinction-cladogenesis (DEC) method to reconstruct geographic range evolution. Our results show that the Lasiantheae stem lineage originated in the neotropics, and the group expanded its range to the palaeotropics during the Eocene, either by continental migration through the boreotropics or by transatlantic long-distance dispersal. Two cases of Oligocene/Miocene over water-dispersal were also inferred, once from the paleotropics to the neotropics within Lasianthus, and once to Madagascar, concurrent with the origin of Saldinia. A lot of the diversification within Lasianthus took place during the Miocene and may have been influenced by climatic factors such as a period of markedly warm and moist climate in Asia and the aridification of the interior of the African continent. When biogeographic reconstructions were averaged over a random sample of 1000 dated phylogenies, the confidence in the biogeographic reconstruction decreased for most nodes, compared to when a single topology was used. A good understanding of phylogenetic relationships is necessary to understand the biogeographic history of a group, bit since the phylogeny is rarely completely known it is important to include phylogenetic uncertainty in biogeographic analysis. For nodes where the resolution is uncertain, the use of a single "best" topology as a basis for biogeographic analysis will result in inflated confidence in a biogeographic reconstruction which may be just one of several possible reconstructions.