Department of Geology, 114 Hofstra University, Hempstead,
Estimates of species abundance used to quantitatively describe paleocommunities are more precise and more reliable when sampling effort is distributed among many small replicate samples, rather than concentrated in the collecting of one or a few large samples. This is because sampling error is introduced by the patchy distribution of individuals within a fossil deposit. This study applies a dispersed sampling protocol to compare the fossil assemblages preserved within a marine shell bed at two different localities of the Upper Cretaceous Navesink Formation in east-central New Jersey. A spatial hierarchy of small bulk samples (replicate samples collected along an outcrop, samples from different outcrops within a locality, and samples collected from two different localities) reveals the magnitude and scale of patchiness in the distribution of macrofauna in the Navesink shell bed. Species abundance is highly variable between replicate samples and moderately variable between outcrops due to small scale patchiness. Nevertheless, estimates of species abundance generated by collecting across the patches at each locality reveal that the overall species abundance distribution for the Navesink shell bed is nearly identical between the two localities. When collecting effort is dispersed among many widely distributed samples, different patches of fossil remains are sampled and contribute to the overall estimate of average composition obtained for a locality. Comparisons of fossil assemblages between localities or horizons are rendered more reliable by decreasing the probability that compositionally different patches have been sampled within otherwise identical paleocommunities. The significance of differences detected between local paleocommunities can be assessed more confidently when replicate samples provide a measure of local variability arising from patchiness. Paleontologists sampling to describe the species abundance composition of a discrete region of a stratigraphic horizon (e.g., outcrop, locality) should define the spatial scale of the region they are describing and disperse their sampling effort within that region.