FIELD METHODS FOR QUANTITATIVE SAMPLING IN PALEOECOLOGY: STRATEGIES FOR THE COLLECTION AND ANALYSIS OF FOSSIL ASSEMBLAGE DATA

The quantitative sampling and statistical analysis of fossil assemblages is an increasingly important tool in paleontology. To illustrate, twenty five percent of all research studies published in the journal PALAIOS from 2001 to 2003 employed the statistical analysis of new quantitative data sampled from fossil assemblages. In these studies, paleontologists used assemblage data to assess community stability through time, characterize and track depositional environments, reconstruct paleocommunities, measure predation intensity, and analyze taphonomy. A review of the sampling methods employed in these studies reveals that, although most provided data that are adequate for addressing the research questions posed, the majority of studies would have benefited greatly from a better understanding of how sampling methods impact the quality of the data for statistical analysis. All studies based on the sampling of fossil assemblages should explicitly identify the scale (both spatial and temporal) of the target population(s) being sampled. Field workers must also be aware of the limitations of bulk sampling, particularly when fossil assemblages are patchy. Bulk sampling is most effective when many small samples are taken, rather than a few large samples. Bulk samples should also be consistent in area or volume within a study. This allows fossil assemblage data to be analyzed by density in addition to proportional abundance, which can sometimes be misleading if specimen counts vary greatly from sample to sample. Statistical analyses are often predicated on random sampling, which is usually impractical, if not impossible, to carry out in field surveys of fossil assemblages. Most paleontologists sample at regular intervals along a transect through the target population (such as along the face of an outcrop). When sampling is not random, replication must be employed to yield reliable data for statistical analysis. There is no simple formula for determining the correct size, number and distribution of samples; the optimal sampling scheme will be different for each research project. If paleontologists understand the sampling requirements of statistical analysis they can better plan their sampling in the field to yield data that provide maximum analytical power with minimum noise and effort.