Abstract
BACKGROUND AND AIMS: Repeated hybridization and/or polyploidization confound classification and phylogenetic inference, and multiple colonizations at different time scales complicate biogeographical reconstructions. This study investigates whether such processes can explain long-term controversies in Anthoxanthum, and in particular its debated relationship to the genus Hierochloë, the evolution of its conspicuously diverse floral morphology, and the origins of its strikingly disjunct occurrences. A hypothesis for recurrent polyploid formation is proposed. METHODS: Three plastid (trnH-psbA, trnT-L and trnL-F) and two nuclear (ITS, ETS) DNA regions were sequenced in 57 accessions of 17 taxa (including 161 ETS clones) and Bayesian phylogenetic analyses were conducted. Divergence times were inferred in *BEAST using a strict molecular clock. KEY RESULTS: Anthoxanthum was inferred as monophyletic and sister to one species of Hierochloë based on the plastid data, whereas the nuclear data suggested that one section (Anthoxanthum section Anthoxanthum) is sister to a clade including the other section (Anthoxanthum section Ataxia) as sister to the genus Hierochloë. This could explain the variation in floral morphology; the aberrant characters in Ataxia seem to result from a Miocene hybridization event between one lineage with one fertile and two sterile florets (the Anthoxanthum lineage) and one which probably had three fertile florets as in extant Hierochloë. The distinct diploid A. gracile lineage originated in the Miocene; all other speciation events, many of them involving polyploidy, were dated to the Late Pliocene to Late Pleistocene. Africa was apparently colonized twice in the Late Pliocene (from the north to afro-alpine eastern Africa, and from south-east Asia to southern Africa), whereas Macaronesia was colonized much later (Late Pleistocene) by a diploid Mediterranean lineage. The widespread European tetraploid A. odoratum originated at least twice. CONCLUSIONS: Many of the controversies in Anthoxanthum can be explained by recurring hybridization and/or polyploidization on time scales ranging from the Miocene to the Late Pleistocene. All but one of the extant species shared most recent common ancestors in the Late Pliocene to the Late Pleistocene. The disjunct occurrences in Africa originated in the Late Pliocene via independent immigrations, whereas Macaronesia was colonized in the Late Pleistocene.