Hermannia, commonly known as dolls roses, is a genus of plants predominantly found in South Africa. While four species occur within North and Central America and one species in Australia, most species are distributed across Africa from Cape Verde eastwards to Saudi Arabia, with the majority restricted to Southern Africa. The taxonomy of the genus is not well understood, species concepts are poorly defined, and there have not been any comprehensive studies on its phylogeny, distribution, diversity or ecology. With the limited knowledge of the genus prior to this thesis, it was not possible to assess how Hermannia has evolved and dispersed, nor to gauge its value as an indicator of diversity or climate change. In this thesis, a study of the systematics of Hermannia is undertaken, utilizing molecular analysis of DNA samples obtained from plant material. Sequencing and phylogenetic analysis is undertaken for 141 species based primarily on the nuclear gene region ITS4-5m. The result is used to create a phylogenetic tree of a representative portion of the known species of Hermannia. The phylogenetic hypothesis is then used as a framework for analysing the locality and morphological information for each available accession. The principle outputs of these analyses include 1) a taxonomic framework for the genus, 2) distributions of clades within Hermannia, 3) molecular dating of the phylogeny, 4) reconstructions of ancestral states for a species included in the phylogeny for 13 characters, 5) maps of richness and diversity of Hermannia, 6) phytochoria for Hermannia and climatic modelling. In addition, a taxonomic treatment is provided for the subgenus Mahernia and a revised nomenclatural account for the subgenus Hermannia. Morphological characters for a selection of over 100 species of subgenus Mahernia are encoded and used to produce an electronic key. Data from the molecular dating and reconstructions are analysed in order to address the origins of the genus, modes and directions of dispersal and morphological support for the molecular-based phylogeny. The locations of high richness and diversity are related to persistence and speciation of the genus and patterns of distribution. The phytochoria are compared to known vegetation types and subjected to bioclimatic and prediction modelling in order to identify climatic variables associated with the distributions of various clades and to predict alteration of their current ranges under predicted future climatic conditions. The phylogenetic analysis confirms the monophyly of two subgenera that were initially proposed at generic level by Linnaeus, namely the widely distributed subgenus Hermannia containing approximately 127 species and subgenus Mahernia containing approximately 112 species. The newly proposed taxonomy consolidates more than 600 species names into some 250 species, and includes descriptions and depictions of close to 70 new species. The earliest split between clades within Hermannia is postulated to have occurred around 27 million years ago, with subsequent splits between 17 and 4.5 million years ago. Reconstruction of characters provides strong indirect support for the proposed phylogeny, unambiguous ancestral states for most characters, and reveals four clear synapomorphies. High levels of richness and diversity are found for a number of small regions across Southern Africa, with a concentration in the Cape Floral Region, despite the genus being almost absent from the mountains where most fynbos diversity resides. At a broad level, the cluster analysis of Hermannia displays similar boundaries to the biomes recently derived from a comprehensive assessment of species and environmental factors for the Vegetation Map of South Africa (Vegmap), suggesting that Hermannia may be a good proxy for overall phylogeographic processes within Southern Africa. Analysis of the climatic envelope for each species shows that Hermannia is a genus primarily of species that occupy arid habitats, with subsequent adaptations to more moderate environments. Prediction of changes in distribution resulting from climate change models shows some generalized reaction patterns, with a shrinking of overall distribution for nearly all species. The section Hermannia is found to fit Linder's (2005) criteria for a Cape Clade, and provides independent support for the concept of a Greater Cape Floristic Region (Born, Linder et al. 2007). The results of this thesis provide a good general framework for species relationships within Hermannia on a molecular and morphological basis, patterns of distribution for its newly defined clades, and an initial assessment of broader issues of diversity and climate change via analysis of the genus.