THE INFLUENCE OF SAMPLE SIZE ON THE PERCEPTION OF FOSSIL COMMUNITY DIFFERENCES BETWEEN IN-PLACE AND SURFACE COLLECTED SAMPLES
We collected 3 in-place 4 L samples at 1 m stratigraphic intervals and 3 corresponding surface samples. An experienced paleontologist collected surface samples by exhaustively picking all specimens to fill a 2 L bag. Forcino and Stafford (2017) found median sample sizes of 220 and 607 for in-place and surface samples, respectively. To produce surface subsamples of a similar size to the in-place samples, we randomly mixed, then twice halved the three surface samples, producing two 25% subsamples each. The fossils were sorted, identified to the genus level, and counted.
The median sample size for the 25% subsamples was 201, similar to the in-place sample size. The median richness in the subsamples was 25, closer to the in-place samples (25) than the 100% surface samples (45). Conversely, the median evenness for the subsamples was 0.80, closer to the 100% surface samples (0.74) than the in-place samples (0.58). The diversity metrics indicate the 25% subsamples may not contain all rare taxa found in the 100% surface samples, but the abundance distribution is preserved. In NMDS ordination, surface samples separate from in-place samples, regardless of sample size. The relative abundances of brachiopods and mollusks are more similar between the 25% and 100% surface samples than the in-place samples. Overall, the 25% subsamples are not similar enough to the in-place samples to suggest that sample size drove the observed differences between the 100% surface and in-place samples. The differences are substantial enough that paleontologists should base their collection method on what the study demands, rather than what is easier or more cost effective.