Many features of global diversity compilations have proven robust to continued sampling and taxonomic revision. Inherent biases in the stratigraphic record may nevertheless substantially affect estimates of global taxonomic diversity. Here we focus on short-term (epoch-level) changes in apparent diversity. We use a simple estimate of the amount of marine sedimentary rock available for sampling: the number of formations in the stratigraphic Lexicon of the United States Geological Survey. We find this to be positively correlated with two independent estimates of rock availability: global outcrop area derived from the Paleogeographic Atlas Project (University of Chicago) database, and percent continental flooding. Epoch-to-epoch changes in the number of formations are positively correlated with changes in sampled Phanerozoic marine diversity at the genus level. We agree with previous workers in finding evidence of a diversity-area effect that is substantially weaker than the effect of the amount of preserved sedimentary rock. Once the mutual correlation among change in formation numbers, in diversity, and in area flooded is taken into consideration, there is relatively little residual correlation between change in diversity and in the extent of continental flooding. These results suggest that much of the observed short-term variation in marine diversity may be an artifact of variation in the amount of rock available for study. Preliminary results suggest the same possibility for terrestrial data.
Like the comparison between change in number of formations and change in sampled diversity, which addresses short-term variation in apparent diversity, the comparison between absolute values of these quantities, which relates to longer-term patterns, also shows a positive correlation. Moreover, there is no clear temporal trend in the residuals of the regression of sampled diversity on number of formations. This raises the possibility that taxonomic diversity may not have increased substantially since the early Paleozoic. Because of limitations in our data, however, this question must remain open.