Diversity biases in terrestrial mammalian assemblages and quantifying the differences between museum collections and published accounts: A case study from the Miocene of Nevada

Understanding diversity through time in the fossil record has primarily relied on the raw count of species within a given time interval, or species richness. These estimates are often derived from published fossil data, and standardized for sample size or geographic area. However, most methods that standardize richness by sample size are sensitive to changes in evenness, which introduces a potential problem with relying on published records: published accounts could be more even than the museum collections from which they are drawn. We address this bias in the context of mammalian paleodiversity, comparing published and museum collections of the Hemphillian Thousand Creek fauna to those of the Barstovian Virgin Valley fauna. We rarified specimen data, both number of identified specimens (NISP) and minimum number of individuals (MNI), and presence/absence data to compare published and museum data within and between faunas. Within faunas, published numbers of specimens are more even than museum samples, but the difference for localities in Virgin Valley is not significant. Neither published nor museum numbers of specimens indicate a significant difference between faunas, but the diversity pattern is reversed between the two data sets. Presence/absence rarefactions show no differences between sources; here, published data adequately sample the underlying museum records. Specimen-based evenness is not accurate in the published sample, and therefore we suggest that future studies of diversity in terrestrial mammalian assemblages must assess unpublished collections. Additionally, NISP data for Thousand Creek are more even than MNI data, suggesting that relying solely on NISP for assessing species diversity can also be misleading. Because publication bias alters richness and evenness, diversity estimates using published data must be circumspect about data sources.