Fagraea

Abstract: The intrafamilial relationships of the Gentianaceae are investigated by means of a cladistic analysis based on morphological and to a lesser extent on chemical data. The 21 genera that are selected for the analysis represent all tribes and subtribes except Leiphaimeae, Rusbyantheae and Voyrieae. The large genus Gentiana is represented by three of its sections. The former loganiaceous genera Anthocleista and Fagraea are used as outgroups. Standard parsimony analyses and analyses using weights that are based on the cladistic reliability of the characters give congruent results as far as the global relationships are concerned. The best supported clade contains Eustoma (Tachiinae) and all included Gentianinae, Erythraeinae and Chironiinae. The basal division in this clade is between Ixanthus and the other genera. In this way Ixanthus, an endemic of the Canary Islands, connects the mostly woody tropical and the mostly herbaceous temperate taxa. Subtribe Gentianinae (excluding Ixanthus) is monophyletic, unlike Erythraeinae and Chironiinae. In most analyses, however, both subtribes together (and including Eustoma) are the sister-group of Gentianinae. Possibly Erythraeinae, Chironiinae and Eustoma should be merged. The basal parts of the cladograms, involving the woody tropical representatives and Exacum, are poorly resolved. More extensive sampling, especially among the tropical representatives, is necessary to elucidate these basal relationships. The tropical ancestry of the family, the switch from a woody to a herbaceous life form, and the position of critical taxa, such as Swertia and Halenia or Tripterospermum, are discussed. The Gentianaceae is a cosmopolitan family of medium size, with 76 genera (Brummitt 1992) and about 1200 species (Mabberley 1990; see Table 1). Its oldest known fossils are from the Eocene of North and Central America (Crepet and Daghlian 1981; Graham 1984). Recent cladistic analyses based on rbcL sequence data (Olmstead et al. 1993; Bremer et al. 1994), restriction site variation of the chloroplast genome (Downie and Palmer 1992) and morphological, anatomical, embryological and chemical data (Struwe et al. 1994) indicate that Gentianaceae are one of the principal families of the monophyletic order Gentianales. Results in Bremer et al. (1994) and Struwe et al. (1994) are consistent with the hypothesis (e.g. Downie and Palmer 1992; Bremer and Struwe 1992) that Loganiaceae sensu Leeuwenberg and Leenhouts (1980) are a paraphyletic assemblage with members showing closest relationships to other families both within and outside of the Gentianales. As far as Gentianaceae is concerned, Struwe et al.'s (1994) main conclusion is to formally include Potalia Aubl., Fagraea Thunb. and Anthocleista Afzel. ex R. Br. (tribe Potalieae of Loganiaceae sensu Leeuwenberg and Leenhouts 1980) in the Gentianaceae. This transfer had already been proposed by Bureau (1856) in the previous century and more recently by Fosberg and Sachet (1980) on the basis of gross morphology (although monographers of the Loganiaceae disagreed, e.g. Leeuwenberg and Leenhouts 1980) and by Jensen (1992) on the basis of the presence of advanced iridoid glucosides. It should be noted that the inclusion of Anthocleista and Fagraea increases the woody paleotropic representation of the family, that is otherwise restricted to Gentianothamnus Humbert (Humbert 1937). While a consensus seems to be emerging about the monophyly of the Gentianales and the inclusion of Potalieae in Gentianaceae, much work remains to be done concerning the interfamilial relationships within the order, including the relationships of the smaller families often included in Gentianales (e.g. Saccifoliaceae, Dialypetalanthaceae) and concerning the infrafamilial relationships of the bigger Gentianales families. We focus on the Gentianaceae. Because the broad-based cladistic analyses (e.g. Downie and Palmer 1992; Olmstead et al. 1993; Bremer et al. 1994; Struwe et al. 1994) to date