The latitudinal gradient hypothesis describes the observation that diversity increases as you travel from the poles towards the equator. Even at the equator diversity is not equally distributed. Understanding what spatial and temporal features best explain these local differences can help elucidate what processes drive and maintain biodiversity. The refuge hypothesis predicts that fluctuations in climate drove cyclical isolation of forest refugia, leading to isolation of populations and allopatric speciation. Historical refugia provide an enticing scenario to explain the unequal distribution of species in the tropics. To estimate centers of forest stability (refugia), we hindcasted 20 forest-obligate frog species back 2.5 million years. Stability was estimated based on how many species were present in a cell for >90% of the time slices. To explore how forest stability has shaped contemporary diversity we focused on Afrobatrachian frogs in the Lower Guinean Forests (LGF). Afrobatrachia is a highly diverse group and represents ~50% of frog species in the LGF. In total, 124 species out of the 128 Afrobatrachia known in the region were included. We used the phylogeny from Portik et al. (2019) to estimate phylogenetic diversity and phylogenetic endemism. Using linear regression models, we show that historical stability positively explains heightened Afrobatrachia PD and PE in the LGF. Directly testing the influence of historical habitat stability can help elucidate evolutionary processes, not only at the species-level, as demonstrated here but also in population genetics and functional diversity research.