• biodiversity hotspot;
  • biogeography;
  • conservation;
  • divergence time estimation;
  • molecular phylogenetics;
  • taxonomy

Madagascar is renowned for its unparalleled species richness and levels of endemism, which have led, in combination with species extinction caused by an unprecedented rate of anthropogenic deforestation, to its designation as one of the most important biodiversity hotspots. It is home to 10 650 species (84% endemic) of angiosperms in 1621 genera (19% endemic). During the last two centuries, botanists have focused their efforts on the provision of a taxonomic framework for the flora of the island, but much remains to be investigated regarding the evolutionary processes that have shaped Madagascan botanical diversity. In this article, we review the current state of phylogenetic and biogeographical knowledge of the endemic angiosperm genera. We also propose a new stratified biogeographical model, based on palaeogeographical evidence, allowing the inference of the spatio-temporal history of Madagascan taxa. The implications of past climate change and extinction events on the evolutionary history of the endemic genera are also discussed in depth. Phylogenetic information was available for 184 of the 310 endemic genera (59.3%) and divergence time estimates were available for 67 (21.6%). Based on this evidence, we show the importance of phylogenetic clustering in the assemblage of the current Madagascan diversity (26% of the genera have a sister lineage from Madagascar) and confirm the strong floristic affinities with Africa, South-East Asia and India (22%, 9.1% and 6.2% of the genera, respectively). The close links with the Comoros, Mascarenes and Seychelles are also discussed. These results also support an Eocene/Oligocene onset for the origin of the Madagascan generic endemic flora, with the majority arising in the Miocene or more recently. These results therefore de-emphasize the importance of the Gondwanan break-up on the evolution of the flora. There is, however, some fossil evidence suggesting that recent extinctions (e.g. Sarcolaenaceae, a current Madagascan endemic, in southern Africa) might blur vicariance patterns and favour dispersal explanations for current biodiversity patterns. © 2013 The Linnean Society of London