Generic concepts in Australian Gardenieae (Rubiaceae): a cladistic approach (original) (raw)

1999, Australian Systematic Botany

Cladistic analyses of a morphological database for 37 Australian species of Gardenia Ellis and Randia Houst. ex L. demonstrate the polyphyly of both genera. Five clades are identified, with Randia audasii C. T. White placed in an isolated position. Four of these clades are assigned to extant genera: Aidia Lour., Atractocarpus Schltr. et K. Krause, Gardenia s.s. and Kailarsenia Tirveng. The following genera are considered as falling within Atractocarpus: Neofranciella Guillaumin, Sukunia A. C. Sm., and Trukia Kaneh. Pelagodendron Seem. belongs with Aidia. Whereas Randia audasii falls within the current concept of Rothmannia Thunb., available data suggest that genus is paraphyletic, and should be confined to the African taxa. Introduction The Gardenieae and Coffeeae together include one-fifth of all rubiaceous genera, following revised delimitation of the tribes by Bremekamp (1934, 1966) and Verdcourt (1958). These tribes were initially associated by their contorted corolla lobes (Bremekamp 1934) and later studies (e.g. Robbrecht 1979) have reinforced this link over the earlier system based on number of ovules. While many genera with large flowers and fruits and numerous seeds immersed in pulp were referable to the Gardenieae, and others with small flowers and fruits, and 1-seeded locules with no pulp were referable to the Coffeeae, many intermediate taxa remained difficult to assign (Verdcourt 1981). Although many small, often monotypic, genera have been described (e.g. Griffithia Thwaites, Xeromphis Raf., Basanacantha Hook. f.), they have often been separated by rather artificial characters (Standley 1929) and few have been widely accepted (e.g. Rothmannia Thunb.). For almost two centuries Randia Houst. ex L. and Gardenia Ellis have served as reservoirs for difficult elements with contorted aestivation within the tropical woody Rubiaceae, species with two locules in the ovary being placed in the former, and those with one in the latter (Bentham 1861, 1867; Hooker 1873). Whereas Keay (1958), Bridson et al. (1980) and later workers have radically divided Randia and Gardenia in tropical Africa, these narrower generic concepts have generally been rejected in the neotropics (e.g. Standley and Williams 1975; Dwyer 1980; Rogers 1987). Most recent workers on the palaeotropical taxa, however, have accepted the conclusion of Fagerlind (1943), Taylor (1944), Bremekamp (1957) and Keay (1958), that Randia is restricted to the Americas; a recent cladistic analysis of Ixoroideae revealed no close relationship between Randia s.s. and the two representatives of Australian Randia s.l. included in the study (Andreasen and Bremer 1996). There has been a proliferation of new genera in the eastern palaeotropics (e.g. Tirvengadum and Sastre 1979; Tirvengadum 1983) but no testing of the concepts. Indeed, few taxa have been described or revised in such detail that one can be confident of their generic identity (e.g. Tirvengadum 1983). The aim of this study is to resolve the generic placement of the Australian species of Gardenia s.l., Randia s.l. and several new taxa in light of the narrower generic delimitations of both Gardenia and Randia.