SUBFAMILIAL AND TRIBAL Birgitta Bremer2,RELATIONSHIPS IN THE Katarina Andreasen2, and Daniel Olsson2RUBIACEAE BASED ON rbcLSEQUENCE DATA'ABSTRACT A parsimony analysis of rbcL sequences from 49 Rubiaceae genera (representing 23 tribes) and 7 outgroup taxawas performed. Species representing 48 genera of Rubiaceae were sequenced: Anthospermum, Antirhea, Bertiera,Bouvardia, Calycophyllum, Catesbaea, Cephalanthus, Cinchona, Coffea, Cubanola, Deppea, Enterospermum,Erithalis, Exostema, Gardenia, Guettarda, Haldina, Hallea, Hamelia, Hillia, Hintonia, Hoffmannia, Hydnophy-tum, Ixora, Keetia, Luculia, Meyna, Mitriostigma, Morinda, Mussaenda, Nauclea, Nertera, Oldenlandia, Ophior-rhiza, Parapentas, Pentagonia, Pentas, Pentodon, Pinckneya, Pogonopus, Psychotria, Rachicallis, Rogiera,Rubia, Sarcocephalus, Theligonum, Uncaria, and Vangueria. The cladistic analysis resulted in 24 equally parsi-monious trees with a consistency index (C.I.) of 0.38. The results were analyzed to test higher-level classification andreconstruction of Rubiaceae phylogeny, and to place taxa with disputed taxonomic positions. There are three groupsof taxa more or less corresponding to three of Robbrecht's four subfamilies: Cinchonoideae s. str., Ixoroideae s.l.,and Rubioideae. There is no support for the subfamily Antirheoideae, the taxa of which are nested within Cinchonoideaes. str. and Ixoroideae s.l. The positions of Luculia and Hintonia are uncertain. The tribal positions (sensu Robbrecht)are supported for a majority of the genera, but other indicated relationships contradict earlier classification. The data,although based on a limited number of taxa, support the monophyly of the tribes Anthospermeae, Chiococceae s.l.,Guettardeae, Hamelieae, Hedyotideae, Naucleeae s.l., Psychotrieae, and Vanguerieae, but there is no support for thepositions and/or narrow circumscriptions of Cephalantheae and subtribe Mitragyninae (Cinchoneae), or for a widecircumscription of Cinchoneae. The rbcL data also give useful suggestions for taxonomic positions of the followinggenera with uncertain affiliations: Bertiera, Catesbaea, Deppea, Hillia, Ophiorrhiza, Rachicallis, and Theligonum. The Rubiaceae are one of the most speciosefamilies, especially in the tropics, with about 10,000species (Mabberley, 1990). Biologically and mor-phologically they are diverse, with many differentlife forms and reproductive traits. The life formsvary from tiny herbs, epiphytes, lianas, and shrubsto tall trees, and the various kinds of flowers havedifferent pollination systems, where pollen is spreaddirectly from the stamens by insects, birds, or windor secondarily from the upper part of the styles(stylar pollen presentation) by insects. There is agreat variety of fruits and seeds dispersed by dif-ferent agents, e.g., dry capsules with wind-dis-persed seeds, dry dehiscent or indehiscent meri-carps, or fleshy and animal-dispersed berries ordrupes. The different fruit traits have been muchused for higher-level classification. The Rubiaceaeare known to be a family with difficult intrafamilialclassification. A persistent problem with the differ-ent classification schemes proposed has been thatone or a few characters have been used as absolutemarkers for the different taxonomic groups, andthis has led to unnatural groupings in many cases. Schumann (1891) divided the Rubiaceae intotwo subfamilies, Cinchonoideae and Coffeoideae,based on a single character, the number of ovulesper locule. This character and Schumann's clas-sification were almost totally rejected by later au-thors (Table 1 presents a comparison between tribaland subfamilial classification schemes historicallyused). Bremekamp (1954, 1966) instead empha-sized testa structure, occurrence of albumin in theseeds, raphides, and the "ixoroid pollen presen-tation mechanism" (Ixoroideae), and he recognizedeight subfamilies. Three of these subfamilies, theCinchonoideae, the Rubioideae, and the Guettar-doideae (= Antirheoideae), were accepted by Verd-court (1958). He utilized the same characters asBremekamp, but rejected the pollen presentationmechanism as a subfamilial character; instead, he ' We are grateful for many kinds of help and support; for providing plant material: Y. B. Linhart, L. McDade, T.McDowell, E. Robbrecht, R. Sanders, O. Seberg, and the directors of the botanical gardens listed in Table 2; forcomments on the manuscript: K. Bremer, C. Taylor, and one anonymous reviewer; and for technical assistance: P.Jalonen and J. R5nnholm. This research was supported by Swedish Natural Science Research Council grant B-BU1487 to BB. 2 Department of Systematic Botany, Uppsala University, Villav. 6, S-752 36 Uppsala, Sweden. ANN. MISSOURI BOT. GARD. 82: 383-397. 1995.