THE STRATIGRAPHY OF MASS EXTINCTION

The stratigraphic pattern of last occurrences in local stratigraphic sections is widely used to infer whether the temporal pattern of mass extinction was gradual, stepwise, pulsed, or abrupt. Numerical models of fossil occurrences based on principles of sequence stratigraphy, marine ecology, and evolution indicate that last occurrences primarily reflect stratigraphic architecture, not times of extinction. Sequence-stratigraphic architecture causes last occurrences to be gradually arrayed in parts of the stratigraphic record, but clustered in other stratigraphically predictable positions, such as sequence-bounding unconformities, surfaces of forced regression, flooding surfaces, and condensed intervals. Facies change, hiatuses, and non-deposition are equally responsible for such clustering. These models demonstrate that the stratigraphic record of last occurrences is typically far more complex than the simple backward-smearing suggested by the Signor-Lipps Effect. Clustering of last occurrences at sequence-stratigraphic surfaces is a common occurrence in the fossil record, even during background extinction, and mass extinction simply amplifies the strength of these clusters at and below the interval of extinction. Double-pulses of last occurrences are a common result in these models in depositionally downdip regions, whereas single pulses of last occurrences are typical in depositionally updip areas. Comparison with the big-5 mass extinctions and other extinction events indicates that most mass extinctions in the fossil record are consistent with prolonged (several hundred kyr) elevated extinction, with one or more pulses of last occurrences that are stratigraphically controlled. The sole exception to this pattern is the end-Cretaceous, for which the cluster of last occurrences appears to have no connection to sequence stratigraphic architecture, except in depositionally updip settings, where it lies at a sequence-bounding unconformity. These models indicate that geochronological bracketing of clusters of last occurrences may produce misleading estimates for the age and duration of the extinction episode. Geochemical proxies of environmental changes are also affected by sequence architecture.