Journal: Internations Journal of Plant Sciences
Volume: 180
Pages: 386-402


Premise of research. With wind-pollinated flowers and partly temperate distribution, the tribe Anthospermeae stands out in the otherwise mostly animal-pollinated and tropical coffee family (Rubiaceae). Nevertheless, few attempts to resolve the phylogeny of the group have been made, and inter- and infrageneric relationships have been only partly addressed. Here we investigate evolutionary relationships and generic and subtribal delimitations of Anthospermeae. We assess the influence of alternative evolutionary rate models on topology and node support.

Methodology. Using sequence data from the nuclear (nrITS and nrETS) and plastid (atpB-rbcL, ndhF, rbcL, rps16, and trnT-trnF) genomes collected for a broad sample of taxa, we conducted Bayesian analyses using nonclock, strict clock, and relaxed clock models. The resulting topologies and support values were compared, and the relative fit of evolutionary models to our data was evaluated. Marginal likelihood estimates were used to discriminate between the competing rate models.

Pivotal results. The monophyly of Anthospermeae was confirmed with Carpacoce resolved as sister to the remaining species. We found several cases of supported topological conflict between results based on nuclear and plastid data, but the deepest splits of the tribe were congruent among all analyses and incompatible with traditional subtribal delimitations of Anthospermeae. Monophyly of the genera Anthospermum, Nenax, and Coprosma was not supported. While the relaxed clock model was consistently favored over the nonclock and strict clock models for all data sets, the use of the different models had little impact on phylogenetic results.

Conclusions. We propose a revised subtribal classification of Anthospermeae, including a new subtribe, the monogeneric Carpacocinae. Introgression/hybridization and incomplete lineage sorting are the most likely causes for the plastid-nuclear incongruences detected for Anthospermeae, but their relative contribution could not be concluded.