Are Sepkoski’s evolutionary faunas dynamically coherent?

For more than two decades, Jack Sepkoski’s hypothesis of three great ‘evolutionary faunas’ has dominated thinking about the Phanerozoic evolution of marine animals. This theory combines pattern description with process modelling: diversity trajectories of major taxonomic groups are sorted into three categories, and the trajectories are predicted by coupled logistic equations. Here I use a re-creation of Sepkoski’s classic three-phase coupled logistic model and an empirical analysis of his genus-level compendium to re-examine his claims about diversity dynamics. I employ a ‘focal-group’ variant of the proportional volatility G-statistic to determine whether variation in turnover rates of focal taxonomic groups can be explained by the average rates for each group through time combined with average rates across all groups within each temporal bin. If growth is exponential and ecological interactions between pairs of groups are always similarly strong, then groups will wax and wane very predictably, and these statistics will always be insignificant. If instead growth is density-dependent and there are no interactions, significant volatility should be confined to periods of rapid radiations, such as those following major mass extinctions. Finally, if unusually strong pairwise interactions directly cause certain groups to succeed or fail, then significant volatility in each competing group should be present during replacement episodes that are not tied to overall radiations or extinctions. Additionally, if clustering groups into faunas is informative, then summed faunal diversity histories will replicate the observed volatility of all groups treated separately. To illustrate the focal-group method, I apply it to diversity data for the major groups of Cenozoic North American mammals. Surprisingly, the tests show that although some orders experience significant radiations and extinctions, orders with visually similar trajectories such as archaic, mostly Paleocene mammals fail to share dynamic properties. Volatility is far greater in Sepkoski’s marine data, with almost every class showing significant and strong deviations from background turnover rates. However, Sepkoski’s three-phase model predicts these patterns inconsistently. As expected, the Cambrian and Paleozoic faunas show high volatility during the hand-off between them. However, the hypothesized twin late Paleozoic and Jurassic/early Cretaceous hand-offs between the Paleozoic and Modern faunas are not marked by excessively rapid declines and increases. Instead, the Modern evolutionary fauna shows highly unusual dynamic behaviour starting in the mid-Cretaceous, coincident with the Mesozoic marine revolution and well after the Paleozoic fauna’s decline. Sepkoski’s categorization also generally fails to summarize overall volatility during long stretches of the Paleozoic and Mesozoic. * e-mail: [email protected] Consult the copyright statement on the inside front cover for non-commercial copying policies. Evolutionary Ecology Research, 2004, 6: 1–32