Climate and edaphic (soil) properties are widely recognised as key environmental drivers of the evolution, adaptive specialisation, and development of edaphically distinct floras on atypical geological substrates. In South Africa, the mountain ranges of Griqualand West (GW), a recognised centre of plant endemism within the Kalahari Bushveld Bioregion of the Savanna Biome, harbour floras associated with such substrates. However, these mountains remain understudied, with limited understanding of the interactions among vegetation, soil, and climate. The aim of this study was to determine the primary environmental drivers of plant diversity across landscapes underlain by dolomite, banded ironstone and quartzite. We assessed 19 bioclimatic variables and 14 edaphic properties. We calculated four plant diversity indices across four mountain landscapes (Ghaap Plateau, Kuruman Hills, Asbestos Hills, and Langberg). Based on the productivity-diversity hypothesis, using rainfall and soil fertility as proxies for productivity, we hypothesised that quartzite sites (characterised by low rainfall and nutrient-poor soil) would exhibit the lowest diversity and richness, while dolomitic sites (with higher rainfall and fertile soil) would exhibit the highest. Mountain plant communities differed significantly in their bioclimatic conditions, edaphic properties and levels of diversity. Generalised linear models (GLMs) identified isothermality, minimum temperature of the coldest month, mean annual precipitation, precipitation of the warmest quarter, Al, Ca, Ca:Mg ratio, cation exchange capacity (CEC), Mg, soil pH, S and Ti as significant predictors of species diversity. Communities on ironstone were the most diverse, whereas those on dolomite were the most species-rich, but less diverse. Herbaceous diversity and species richness were lowest on nutrient-poor quartzite soil. These findings partially support our initial hypothesis and highlight the importance of substrate-specific edaphic filtering effects linked to parent geology. Given the increasing threats posed by land use and climate change to edaphically unique floras globally, we advocate for the incorporation of both climatic and edaphic variables into species distribution models (SDMs). Such integration will enhance predictive accuracy regarding the distribution of edaphic endemics under scenarios of environmental change. Furthermore, detailed soil sampling and analysis remain critical to improving our understanding of the edaphic determinants of plant diversity in edaphically complex landscapes.