Much of what is known about the long-term history of biodiversity and rates of taxonomic evolution in the fossil record derives from literature-based compilations of fossil stratigraphic ranges. It has been suggested that taxonomic and stratigraphic errors in these compilations are randomly distributed and, therefore, introduce no significant bias to macroevolutionary patterns. Here we compare a new, comprehensive global database of Ordovician and Early Silurian crinoids to Sepkoski's global genus compendium.
Approximately 44% of the crinoid genera resolved to substage in Sepkoski's compendium are taxonomically inaccurate (i.e., invalid, nomina dubia, or column genera) or have incorrect first and/or last occurrences. Errors in Sepkoski's compendium result from incomplete coverage of existing taxonomic work and incorrect stratigraphic correlations that, in some cases, are propagated throughout the taxonomic literature. Stratigraphic range errors are nonrandomly distributed among substages in Sepkoski's compendium. The result is underestimated richness in the Early Silurian and significantly overestimated rates of extinction in the Late Ordovician. There is no similar bias in Sepkoski's substage origination rates for crinoids.
At the stage-level of temporal resolution, Sepkoski's crinoid data are more accurate. In this case, only 32% of the compendium's crinoid genera contain some stratigraphic or taxonomic inaccuracy. However, errors still result in incorrect macroevolutionary patterns, particularly with respect to rate of origination in the Ashgill, which is significantly underestimated in Sepkoski's compendium. Genera described since the completion of Sepkoski's compendium have had relatively little effect on estimated rates of evolution at both stage and substage resolution.
These results suggest that macroevolutionary patterns among some taxa in Sepkoski's compilation may be significantly influenced by nonrandomly distributed taxonomic inaccuracies and stratigraphic range errors. In the case of the apparent end-Ordovician mass extinction among crinoids, the revised history reveals a dramatically reduced role for extinction at the substage-level of temporal resolution. At the stage level, Sepkoski's original compilation strongly exaggerates the excess of extinction over origination in the Ashgill. Although biases inherent in the stratigraphic record remain unaccounted for, removing taxonomic and stratigraphic errors in Sepkoski's compendium substantially changes our understanding of the nature of large-scale biotic change for an important Paleozoic taxon during the end-Ordovician.