PARSIMONY ANALYSIS ANDCLADISTICRECLASSIFICATION OF THERELHANIA GENERIC GROUP(ASTERACEAE-GNAPHALIEAE)Arne A. Anderberg' and K�re Bremer2ABSTRACT Parsimony analysis of the Relhania group of genera (Asteraceae-Gnaphalieae-Relhaniinae) yields a revisedhypothesis of their phylogeny. The genera Leysera L. and Oedera L. are demonstrated to have their closest relativeswithin Relhania L'H�r., which is shown to be paraphyletic as hitherto circumscribed. A revised generic classificationis proposed. Relhania and Leysera are redefined more narrowly. Rhynchopsidium DC. and the monotypic NestleraSprengel are reest:ablished. Oedera is amended to include some former Relhania species. The new genus ComborhizaAnderb. & Bremer, with 2 species, is described, and 14 new combinations in Oedera and Comborhiza are made.Because parsimony analysis produces better phylogenetic hypotheses than limited selections of alleged synapomorphies,and because the latter approach hitherto has been more common in generic classification, it is argued that, withimproved phylogenetic hypotheses derived from parsimony analysis, generic reclassifications will become necessaryalso in other groups. In this paper we present a parsimony analysisand propose a cladistic reclassification of a groupof South African Asteraceae genera belonging tothe tribe Gnaphalieae. The genera were revisedduring the 1970s and were then the subjects ofsome of the first cladistic approaches in botany(Bremer, 1976a, b, 1978a, b). At that time,uniquely derived, qualitative characters that couldbe interpreted as synapomorphies, unequivocallyindicating monophyletic groups were sought. Thus,selected features, mainly of the pappus structurecrowning the fruits, were used as putative synapo-morphies defining the genera. Today, cladistics ismethodologically much more sophisticated and isalso considered a standard technique for analyzingphylogenies (Hull, 1989). Application of the par-simony criterion (Farris, 1983) in computerizedprograms has made it possible to use even largeand homoplastic data sets including ail kinds ofinformation, minimizing a priori assumptions. Par-simony analysis of the Relhania group of generayields a revised hypothesis of their phylogeny, andhence a basis for a cladistic reclassification. It is not only the theoretical and methodologicalbackground to the study of the phylogeny that haschanged during the past 15 years, but also theknowledge of the phylogeny of the tribe Gnaphalie-ae as a whole (Anderberg, 1991). Furthermore,another genus, not considered in Bremer's 1976-1978 revisions, has recently been added to thegroup (Anderberg & K�llersji, 1988). Bremer (1976a, b, 1978a, b) revised the tax-onomy and nomenclature of the genera Relhania,Rosenia, Leysera, Oreoleysera, and Antithrixia.In trying to circumscribe correctly these generaas monophyletic groups, Bremer relied on partic-ular characters as synapomorphies defining them.Thus, Relhania, with 29 species, was distinguishedby its pappus of more or less connate scales andno bristles. Bremer also pointed out the hypothet-ical nature of the generic delimitation: Admittedly the loss of pappus bristles might have occurred several times. However, since there is no evidence that Relhania should be polyphyletic in its present circumscription, I believe we must for the time being rely on this character as uniquely derived and keeping the genus together (Bremer 1976a: 9). Relhania was amended by Bremer to includemost of the species of the genus Nestlera, whichprior to Bremer's work was used to house all thespecies with epaleate receptacles. Bremer showed ' Department of Phanerogamic Botany, Swedish Museum of Natural History, P.O. Box 50007, S-104 05 Stockholm,Sweden. 2 Department of Systematic Botany, University of Uppsala, P.O. Box 541, S-751 21 Uppsala, Sweden. ANN. MISSOURI BOT. GARD. 78: 1061-1072. 1991.