Published online by Cambridge University Press: 08 April 2016
Two rival models of evolution in multispecies systems are tested against empirical species-level data. The two models are the Red Queen model of Van Valen as reformulated by Stenseth and Maynard Smith, which assumes that evolution is driven principally by biotic interactions, and the Stationary model of Stenseth and Maynard Smith, which assumes that evolution is propelled primarily by abiotic factors and will cease in the absence of changes in abiotic parameters. Testing refers to the models' predictions regarding the behavior of extinction and origination rates, and assumptions regarding equilibrium diversity and a constant effective environment. The data set includes the dates of origination and extinction for all coccolith, planktic foraminifer, and radiolarian species recorded in the Oligocene through Holocene, and all planktic diatom and silicoflagellate and ebridian species recorded in the Middle Miocene through Holocene in 111 DSDP sites of the low- to mid-latitude Pacific Ocean.
The condition of stable specific age distribution over geologic time is met, which allows one to perform survivorship analysis on extinction rates. The best fit survivorship curve is a decreasing function of age for both coccolith and foraminifer species, and an increasing function of age for radiolarian species. Neither model predicts age dependence of the probability of extinction. The small disparity between these curves and age-independent curves for each group indicates, however, that an age-independent interpretation of extinction probability is a reasonable first approximation. Rates of origination are analyzed in terms of species accretion, introduced to represent the cumulative origination of species within a higher taxon as a function of the age or duration of the community. Accretion analysis indicates that the probability of accretion is both diversity-dependent and absolute time-dependent.
The assumption of a constant effective environment is tested by polycohort analysis and nonparametric logistic regression analysis of true species cohorts. Both techniques indicate considerable variation in extinction probability over geologic time. When the predictions of the two evolutionary models are adjusted to take this variation into account, the results of both survivorship and accretion analysis seem to conform more closely to the predictions of the Red Queen than to the Stationary model. However, as the speed with which the effective environment changes is increased relative to speciation-extinction rates, it becomes increasingly difficult to differentiate patterns predicted by the two models. The assumption of equilibrium diversity can be neither corroborated nor rejected, since the species-level data are compatible with both an equilibrium and a nonequilibrium view of diversity behavior. Reservations concerning the basic assumptions of both models indicate an ultimate test requires that both models be reformulated to make precise and distinctive predictions under a varying effective environment.