65 6. Conclusions and Future Research 6.1 Conclusions Very Few

6. Conclusions and future research

6.1 Conclusions Very few morphological characters separate genera within Thymelaeaceae and the literature has a long history of discussing the relative merits of characters and where the generic limits in this family should be drawn. The family has recently been the focus of several molecular systematic studies, e.g. Fay et al. (1998), Van der Bank et al. (2002) and Galicia-Herbada (2006), in the hope of resolving the relationships within. In this molecular phylogenetic study (based on rbcL, trnL-F and ITS sequence data), the family Thymelaeaceae has been found to be monophyletic comprising four well supported subfamilies, viz. Gonystyloideae, Synandrodaphnoideae, Aquilarioideae and Thymelaeoideae. Three major clades were observed in Thymelaeoideae (I) tropical Africa and south-eastern Asia; (II) non-African taxa; and (III) Gondwanan taxa. The Gondwana clade comprise one of the largest genera within the Thymelaeaceae, namely Gnidia. The most important conclusion drawn from this molecular phylogeny is that Gnidia is not monophyletic but comprises at least four distinct lineages, each related to the other genera within the Gondwana clade of the Thymelaeoideae. The first lineage consists of Gnidia species sister to Drapetes muscoides, as described in 1843 by Hooker. In the second lineage Gnidia pinifolia and G. racemosa were grouped with Struthiola. Although this molecular evidence clearly supports transferring Gnidia pinifolia and G. racemosa to Struthiola, a lack of morphological synapomorphies in this expanded concept of the genus exist. However, the precise affinities of Gnidia pinifolia and G. racemosa to other Gnidia species based on morphological similarities are also not clear. Furthermore, Gnidia pinifolia is the type species of Gnidia and the consequences of transferring this species to Struthiola and appointing a new type species for Gnidia could have far reaching taxonomic repercussions. The implications to typification are currently being considered. Any modification to nomenclature will take into account the significant broader impact of the changes to this important genus. The third lineage, including Gnidia pilosa (previously known as Englerodaphne pilosa), together with Gnidia species, are sister to the genus Pimelea. A monophyletic clade of Gnidia species previously classified in the old genus Epichroxantha, are also included. More taxon sampling is needed as well as morphological and anatomical studies to support the reinstatement of Epichroxantha. In the fourth lineage, taxa previously known as Lasiosiphon, grouped together with a few Gnidia species. The question is now raised whether the old genus Lasiosiphon should be reinstated. Just because the molecular data indicates that Lasiosiphon and other Gnidia species are grouped together, can we use old names like Lasiosiphon when the original

6. Conclusions and future research circumscription of this generic name has changed? It is not Lasiosiphon sensu Fresenius because he did not include four-merous taxa in his circumscription and the molecular analyses place four- and five-merous taxa together. So, on the other hand are these results proof that Lasiosiphon sensu Fresenius is not a natural group? As mentioned earlier, Rogers intends to place Dais cotinifolia into synonymy under a broadly circumscribed Gnidia, including the genus Lasiosiphon. According to these results, the genus Dais is not part of the four lineages of the polyphyletic Gnidia, instead the species D. cotinifolia and Phaleria capitata are grouped together, due to documented cases of heterostyly. A complete molecular study with the addition of sequence data for the only other species (Dais madagascariensis) of the genus, in collaboration with a morphological revision is suggested before any decisions can be made. A good example of Gondwanic affinity is presented here, when observing the geographical distribution of Gnidia and closely related genera (Drapetes, Passerina, Lachnaea, Struthiola and Pimelea) within the subfamily Thymelaeoideae. All of these distributions occur in the Southern Hemisphere, thus the Gondwana region.

6.2 Future research Additional taxon sampling of representatives within the subfamily Thymelaeoideae is needed, with the addition of a non-coding plastid gene, the rps16 intron, which provided successful phylogenetic analyses on specific-levels in previous studies (Robinson, 2004). Taxonomic decisions to adjust the generic limits to fit the criterion of monophyly needs to be strengthened by including more samples of type species of the segregate genera of Gnidia s.l., for example Atemnosiphon coriaceae for the genus Atemnosiphon. Although more resolution is required, future estimations of the root node age, as well as the rates of speciation of the genus Gnidia, would provide insight and lead to a better understanding of the radiation of this polyphyletic genus and the relationship to other genera within the Thymelaeoideae.