Hibiscus is a pantropical genus encompassing a wide range of morphological diversity. As features used to delimit the genus are shared with other genera in the tribe, the circumscription of Hibiscus is problematic. Furthermore, several genera have been segregated from within Hibiscus. At tribal level, the segregation of Gossypieae and Decaschistieae from Hibisceae has highlighted the problems with tribal circumscription. Only a limited number of morphological features have been used to delimit the tribes, which therefore stand in need of testing with other data. We present the results of cpDNA (ndhF, rpl-16 intron) and low-copy nDNA (rpb2) sequence analyses which are congruent with each other, but conflict with the current classification of the family. Hibiscus is shown to be paraphyletic; it contains several genera from Hibisceae and from two other tribes nested within it. Other Hibisceae genera are not most closely related to Hibiscus s.l., but diverge early from the majority of the family. Within the broad clade which contains all Hibiscus sampled thus far, some sections are monophyletic (eg. Furcaria, Azanzae), but others are not (eg. Ketmia, Bombicella). Further, genera of Malvavisceae are present in two clades within Hibiscus, indicating the paraphyly of this tribe and three separate origins of schizocarpic fruit in the family. Further character evolution will be discussed. We also found evidence for the duplication of the nuclear rpb2 gene in Hibiscus s.l. but not in Gossypieae or Malveae. Chromosome numbers suggest that polyploidy may be the cause of the gene duplication, however alternative explanations are considered. Our findings suggest that major changes in the classification of Malvaceae are needed in order to accommodate our increased understanding of the evolutionary history of these plants. Tentative suggestions for such changes are made.

Key words: classification, cpDNA, Hibiscus, Malvaceae, nDNA