Fossiliferous shallowing-upward parasequences in the early Oligocene Byram Formation of Mississippi provide an ideal opportunity to explore the effects of differences in data preparation, format, and analysis in testing for faunal persistence. Ten replicate bulk samples collected from each of three lithologically similar horizons in successive parasequences yielded >13,000 molluscan individuals. Assemblages were first compared using more traditional descriptive statistics that examined aspects of taxonomy, abundance, and guild structure, and were then compared using multivariate techniques with similarity coefficients that employed both compositional and abundance data. Different procedures for identifying and excluding rare taxa prior to analysis were also investigated. Simple numerical methods emphasized similarities among assemblages; hierarchical clustering, nonmetric multi-dimensional scaling (NMDS), and analysis of similarities (ANOSIM) revealed differences among the faunal horizons. Multivariate tests indicated stasis only when using presence-absence data with a large number of rare taxa excluded. When abundance data are utilized, however, faunal change through time is suggested, regardless of how rare taxa are treated. The choice of data format, rare taxon exclusion policy, and analytical approach all affect the interpretation of results and the outcome of hypothesis tests for stability, suggesting that differences in methodology partly contribute to disagreement among studies examining faunal persistence. Furthermore, various properties related to ecosystem structure are ascribed different degrees of importance depending on the study. Thus, comparisons of patterns should be made on equal footing to ensure that differences in methodology are not contributing to differences in interpretation.
