Phylogenetic relationships of Combretaceae inferred from nuclear and plastid DNA sequence data: implications for generic classification

The putative complexity of Combretaceae and lack of information on phylogenetic relationships within the family led us to explore relationships between genera of Combretaceae by means of combined analyses of plastid and nuclear sequences. We collected DNA sequence data from the nuclear ribosomal internal transcribed spacer region and plastid rbcL, psaA‐ycf3 spacer and psbA‐trnH spacer for 14 of the 17 genera of Combretaceae. The current classification of the family into two subfamilies, Strephonematoideae and Combretoideae, is corroborated. Within Combretoideae, division into two tribes, Laguncularieae and Combreteae, is strongly supported. Within Combreteae subtribe Terminaliinae, relationships between genera are largely unresolved. Terminalia is not supported as monophyletic and two groups were identified, one containing mainly African species and another of mostly Asian species. Pteleopsis, Buchenavia and Anogeissus are embedded within Terminalia, and we suggest that all genera of Terminaliinae, with the exception of Conocarpus, should be included in an expanded circumscrition of Terminalia. Within subtribe Combretinae, a clade formed by the two monotypic genera Guiera and Calycopteris is sister to the rest of the subtribe. Groupings in Combretinae are consistent with recent results based on morphological data. Combretum is currently divided into three subgenera: Apethalanthum, Cacoucia and Combretum. The last two were included in this study and supported as monophyletic if Quisqualis is included within subgenus Cacoucia. Meiostemon is sister to subgenus Combretum. We recommend that subgenus Combretum should be expanded to include Meiostemon and subgenus Cacoucia to include Quisqualis. The sectional classification within Combretum proposed in earlier morphological studies is confirmed except for the exclusion of C. imberbe from section Hypocrateropsis in a separate and monotypic section and the inclusion of C. zeyheri (section Spathulipetala) in section Macrostigmatea. In order to accommodate C. imberbe, a new section is suggested. The reinstatement of previously recognized sections Grandiflora and Trichopetala, both of which had been sunk into subgenus Cacoucia section Poivrea, is proposed. © 2010 The Linnean Society of London, Botanical Journal of the Linnean Society, 2010, 162 , 453–476.