Abstract
Molecular clock approaches applied previously to European alpine plants suggest that Primula sect. Auricula, Gentiana sect. Ciminalis and Soldanella diversified at the beginning of the Quaternary or well within this period, whereas Globularia had already started diversifying in the (Late-)Tertiary. In the first part of this paper we present evidence that, in contrast to Globularia and Soldanella, the branching patterns of the molecular internal transcribed spacer phylogenies of both Primula and Gentiana are incompatible with a constant-rates birth-death model. In both of these last two taxa, speciation probably decreased through Quaternary times, perhaps because of some niche-filling process and/or a decrease in specific range size. In the second part, we apply nonlinear regression analyses to the lineage-through-time plots of P. sect. Auricula to test a range of capacity-dependent models of diversification, and the effect of Quaternary climatic oscillations on diversification and extinction. At least for one major clade of sect. Auricula there is firm evidence that both diversification and extinction are a function of temperature. Intriguingly, temperature appears to be correlated positively with extinction, but negatively with diversification. This suggests that diversification did not take place, as previously assumed, in geographical isolation in high-altitude interglacial refugia, but rather at low altitudes in geographically isolated glacial refugia.