Generic concepts in Styphelieae: resolving the limits of Leucopogon

Abstract

Sequence data for the chloroplast-encoded atpβrbcL intergenic spacer were assembled for 43 representatives of Leucopogon R.Br., the largest genus of Epacridaceae. Cladistic analysis rooted on Prionotes, and including representatives of the sister Tribe Epacrideae and a range of other Styphelieae, revealed Leucopogon in all senses to be polyphyletic. There is strong support for the placement of L. pluriloculatus (Leucopogon E) with Lissanthe, but the monophyly of the segregates Leucopogon A, C and D finds no support. Leucopogon hookeri and L. maccraei cluster strongly with Cyathodesand Leptecophylla, whereas the remainder of Leucopogon A form a separate well-supported clade. Two clades defined by base chromosome numbers of x ₂= 6 and x ² = 4 are resolved, the former including all representatives assigned to ‘Gynoconus’ (Leucopogon C) and some of ‘Axonanthus’ (Leucopogon D), the latter comprising the remainder of ‘Axonanthus’ as well as Styphelia viridis, Astroloma ciliatum and A. humifusum. The relationship of the monotypic Croninia (Leucopogon B) to these last two clades is unresolved. A data base of 41 non-molecular characters supported the association of L. pluriloculatus, L. hookeri and L. maccraei with Cyathodes, Leptecophylla and Lissanthe, and also resolved a clade containing the remaining Leucopogon A. When superimposed on the molecular tree, many non-molecular characters appeared homoplastic. However, data for chromosome number and pollen morphology were highly congruent with the molecular tree. Present generic concepts are clearly inadequate. The analysis indicates that L. hookeri and L. maccraei should be removed from Leucopogon s.str. to a new genus; the remainder of Leucopogon A constitute Leucopogon s.str. A sterile anther appendage characterises most species of Leucopogon s.str., but is neither a necessary nor sufficient condition for membership. The genus can only be circumscribed using a combination of morphological features. Further data and more intensive sampling are needed before the limits of a more satisfactory set of genera for the remaining segregates can be defined.