Analysis of sample-level properties along a paleoenvironmental gradient: The behavior of evenness as a function of sample size

The assessment of fossil biodiversity is influenced by a range of sampling effects such as sample size, evenness and lithological character. Many paleontologists have worked to understand and mitigate for such biasing effects, generating standardized assessments of local, regional and global biodiversity. However, when assessing biodiversity among individual communities with small sample sizes, such as those from lithified Paleozoic samples, it is important to understand how these properties relate to one another qualitatively and quantitatively. Here, the properties of richness and evenness were investigated at the sample level along an ecological gradient within a stratigraphically and geographically constrained interval, the Fulton submember of the Kope Formation (Upper Ordovician) of Kentucky and Ohio. Using a replicate subsampling procedure analogous to rarefaction, evenness values were generated using four metrics (J, PIE, Ess and Essmin). Simulations using real and simulated data revealed that the PIE metric generates the most stable assessments of evenness at very small sample sizes, and relative evenness can be assessed at any sample size, provided that replicate subsampling is performed. Comparisons of richness, evenness and lithology revealed no directional trend along the ecological gradient, but two distinct groups of samples were identified by differences in evenness as well as the number of sampled individuals. Such differences in abundance structure influence perceptions of biodiversity when samples from low-evenness and high-evenness assemblages are aggregated. Increasing the sampling intensity in low-evenness communities may help to mitigate for these differences and more thoroughly capture the richness contained within small samples.