A phylogenetic study of some South African representatives of the tribe Arundineae

English: This study dealt with the phylogenetic history of some representatives of the South African Arundineae. Arundineae Durnort. is one of the tribes in the grass (Poaceae) subfamily Arundinoideae Tateoka. This is known to be a heterogeneous assemblage and a taxonomically difficult group. Three techniques were used to investigate the relationships within the tribe Arundineae, namely cytogenetics, DNA amplification fingerprinting and DNA sequencing of the ITS region of the nuclear DNA. The genera Pentameris P.Beauv., Pentaschistis (Nees) Spach. and Prionanthium Desv. are the only three genera in the tribe with seven as a basic chromosome number (excluding the genus Dregeochloa Conert, which was not investigated in this study). A close affinity between these genera is, thus, recognised. This relationship is corroborated by the DNA fmgerprinting and DNA sequencing data. One Prionanthium dentatum (L.f.) Hem. specimen was, however problematic. This specimen grouped with Tribolium Desv. Clayton and Renvoize (1986) proposed this relationship, but in the light of various studies, as in this study, Prionanthium is removed from this closely related genus. Therefore, the recognition of subtribes in Arundineae, is proposed. The genera Arundo L. and Phragmites Adans. were grouped together. This is unexpected, because according to Clayton and Renvoize (1986), these genera are not as closely related, as their similarity in habit would suggest. In this study these two genera, known as arundinoid genera, are a monophyletic grouping nested within the rest of the genera, known as danthonoid genera. The genera Karroochloa Conert & Turpe, Merxrnuellera Conert and Prionanthium were shown to be polyphyletic with the sequencing data. DNA amplification fingerprinting was helpful in resolving species relationships, but failed to elucidate most of the generic relationships in the study. DNA sequencing was the more informative of the two techniques. The combined analysis was helpful in that some resolution was gained, while other was lost. Due to large amounts of missing data and the inability of DNA fingerprinting to resolve generic relationships, the combined analysis could not be used to its fullest capacity. PAUP and HENNIG86 were both used and both software programs gave the same results. In general, the trees created with PAUP were usually shorter. Both software programs can be used, but the ease of PAUP makes it the preferred analysis package. Successive weighting usually resulted in a tree with better Cl and RI values, compared to a search conducted on the equally weighted data set. The topologies of the trees, however, remained the same. Adams consensus trees were the only consensus trees, which were usually the best (or most) resolved when compared to Strict and Semistrict consensus trees. The present tribal classification for the Arundineae seems justified. However, a subdivision of the tribe Arundineae, to accommodate the clustering of the genera Pentameris, Pentaschistis and Prionanthium into a subtribe, may be justified. The position of Arundo and Phragmites is monophyletic if only the South African representatives are investigated. Some confusion still exists within the tribe and DNA sequencing studies of more specimens per species and more species per genus is advised.