The assembly of the Cape flora is consistent with an edaphic rather than climatic filter

The Cape flora is compositionally biased, being dominated by a few fynbos clades (such as Iridaceae, Ericaceae, Proteaceae and Restionaceae) that make up major part of the distinct heathland vegetation in the Cape Floristic Region. Uncertainty exists concerning what excluded the subtropical to tropical palm-dominated woodland/forest vegetation that was the dominant component in the CFR in the Paleocene and allowed the fynbos clades, which are largely derived from outside Africa, to establish and radiate. Two filters have been proposed. The first postulates that the establishment of the Mediterranean climate driven by the late Miocene initiation of the cold-water Benguela Upwelling System (BUS) eliminated the African lineages and allowed the establishment and radiation of sclerophyllous plant clades ("the Mediterranean climate model", MCM). Alternatively, the "oligotrophic soils model" (OSM) postulates that the oligotrophic soils, gradually exhumed by post-Gondwanan Late Cretaceous - early Cenozoic erosion, acted as a filter excluding the African lineages. In this study, we re-calibrate the fynbos clade Phylica (Rhamnaceae), the genus initially used to test the MCM, using new fossil data to test if the crown age precedes the Late Miocene. Our results indicate that we cannot significantly reject a crown age of Phylica consistent with the MCM. We compare the MCM and OSM model for the Cape fynbos flora by compiling the crown ages of 22 fynbos clades. We show that crown ages are not clustered in time around the initiation of the BUS but, are dispersed throughout the Cenozoic. This suggests that oligotrophic soils, rather than summer drought, acted as a filter. Consequently, we argue that the fynbos clades radiated separately in expanding edaphically controlled heathland patches in the Cape mountains as sandstone exhumation after the Gondwanan break-up progressed.