Individual species are unevenly distributed throughout even the most homogeneous deposit due to paleoenvironmental and geographic variation, as well as simple chance. It has been suggested by some workers (e.g. Cobabe and Allmon 1990) that very diverse fossil deposits, such as some of those from the Gulf and Atlantic Coastal Plains of North America, can not be adequately sampled for paleocommunity studies without taking enormous collections. However, because of patchiness in the occurrence of paleocommunities and variation in single paleoenvironmental settings, entire formations can never be sampled adequately by digging one hole at one horizon at one locality. However, with a more involved sampling scheme it is possible to adequately analyze the diversity patterns within and between paleocommunities .
I performed a species accumulation analysis on both the entire Yorktown data set and on several subsets of the data set. I iteratively added individual samples together in order observe the species accumulate as more and more samples are taken. For each analysis, I first added the most diverse sample to the next most diverse sample, all the way down to the least diverse sample. I then did the same iterative addition, but I started with the least diverse sample, added to the next least diverse sample, etc. These two extremes should yield species accumulation curves with the fastest and slowest accumulation rates, respectively, for any particular data set. I then graphed the running total of specimens counted versus the species accumulation for each data set analyzed.
For the entire data set, the species accumulation analysis (Fig. 1) reveals that after an initial rapid rate of species accumulation (as every sample yields new species), the species accumulation rate drops quickly. By the time that 28,8787 specimens have been counted, new species are being acquired at a rate of one new species per 3,125 specimens counted. Those species acquired after the initial rapid rate are inevitably uncommon to extremely rare (less than 20 total specimens).
Figure 1: Species accumulation curve determined by samples iteratively with the most diverse samples added first for the entire Yorktown data set. After 100 species have been tabulated, the rate of new species acquisition drops precipitously. Those new species found are always represented by fewer than 10 specimens, or less than .03% of the total specimens present.
Statistically coherent groups of samples were identified using principle components analysis. Each group contained samples from the same stratigraphic interval, and represents one paleoenvironment or sub-environment. For the Yorktown Formation, a combination of the three most diverse samples within each group was usually sufficient to characterize the diversity of lower diversity groups (15-30 species), while four samples were required for higher diversity groups (50-60 species). Additional samples yielded only a few new rare species (<5%) or none at all.
Each sub-environment at each locality yields a limited number of common species, regardless of how many samples are taken, or how large the samples are. In order to collect a large percentage of the several hundred species described from the Yorktown formation, bulk samples of the total suite of contained sub-environments need to be taken at several different geographic locations. With the addition of samples from more sub-environments from either different localities or different stratigraphic intervals at the same locality, a large percentage of the total number of species described from the Yorktown Formation can be collected.
Complete characterization of the fauna from a single paleoenvironment or sub-environment is possible with limited geographic sampling even for high diversity fossil assemblages like those found in the Yorktown Formation. Characterization of an entire stratigraphic unit requires that all contained sub-environments be sampled, but not that enormous individual samples need to be taken. The total aggregate sample size needed to characterize a particular local sub-environment will vary with the total contained diversity, but by replicate sampling it is possible to characterize the diversity.
Material originally presented at NAPC V, June 1996, Washington, D.C.
Back