Résumé
Two theoretical models have been proposed to describe long-term dynamics of diversification: the equilibrium model considers the Earth as a closed system with a fixed maximum biological carrying capacity, whereas the expansion model hypothesizes a continuously increasing diversification of life. Based on the analysis of the fossil record of all organisms, Benton [1] suggested contrasting models of diversity dynamics between marine and continental realms. Diversity in marine environments is characterized by phases of rapid diversification followed by plateaux, i.e., an equilibrium model [2-4] directly derived from insular biogeography theories [5, 6], whereas diversity in continental environments is characterized by exponential growth. Previous studies that aimed at testing these models with empirical data were based on datasets extracted directly from the reading of the vagaries of the raw fossil record, without correcting for common fossil record biases (preservation and sampling). Although correction of datasets for the incompleteness of the fossil record is now commonly performed for addressing long-term biodiversity variations [7, 8], only a few attempts [9] have been made to produce diversity curves corrected by phylogenetic data from extant and extinct taxa. Here we show that phylogenetically corrected diversity curves for "fish'' (actinopterygians and elasmobranchs) during the last 200 million years fit an equilibrium model in the marine realm and an expansion model in the freshwater realm. These findings demonstrate that the rate of diversification has decreased for marine fish over the Cenozoic but is in sharp expansion for freshwater fish.