CALL FOR PROPOSALS:

ORGANIZERS

  • Harvey Thorleifson, Chair
    Minnesota Geological Survey
  • Carrie Jennings, Vice Chair
    Minnesota Geological Survey
  • David Bush, Technical Program Chair
    University of West Georgia
  • Jim Miller, Field Trip Chair
    University of Minnesota Duluth
  • Curtis M. Hudak, Sponsorship Chair
    Foth Infrastructure & Environment, LLC

 

Paper No. 2
Presentation Time: 1:45 PM

MULTIVARIATE ASSESSMENT OF THE MINIMUM REQUIRED SAMPLE SIZE FOR COMMUNITY PALEOECOLOGICAL RESEARCH


FORCINO, Frank L., Geological Sciences, Michigan State University, Lansing, MI 48917, Forcino@msu.edu

When conducting paleocommunity research, a sample size that is too small may lead to erroneous conclusions. An increasingly larger sample size will eventually result in diminishing returns of data. Paleocommunity research is labor- and time-intensive as it requires expertise in multiple taxonomic groups and sub-disciplines of geology to develop the stratigraphic and biological framework. Therefore, gaining the most accurate estimate of the required sample size for paleocommunity research can save money, time, and effort. Previous paleoecological studies estimated sample size requirements to be 200 or more individuals per sample. Here, using 30 datasets from the Paleobiology Database and the literature from a wide range of time periods, taxonomic groups, scales, lithologies, numbers of samples (13–124) and taxa (21–167), and initial median sample sizes (9–2441), I present evidence that a minimum sample size of 50 specimens per stratigraphic bed is sufficient to determine the relative patterns in variation among paleocommunities.

Each sample within each dataset was randomly subsampled 1000 times to five proportions (50%, 25%, 10%, 5%, 2.5%) of the original median sample size. Each of the 1000 subsampled datasets created for each proportional sample size was correlated with its corresponding 100% dataset using two multivariate methods: the Mantel test and Pearson’s Product Moment Correlations of non-metric multidimensional scaling (NMDS) ordination axis scores. All Mantel R-statistics are > 0.92 for median sample sizes > 25 individuals, and Pearson’s r-values for NMDS axis-one scores are > 0.86 for median sample sizes > 50 individuals. Thus, the multivariate paleocommunity signal, for each dataset respectively, is statistically indistinguishable for all median sample sizes > 50. This is evidence that the minimum required sample size for paleocommunity studies employing multivariate techniques to examine patterns among samples is 50 individuals. In addition, lithology, time period, dominant taxonomic group, number of taxa, or number of samples did not influence the result. This suggests that none of these variables will affect the required sample size. With a smaller required sample size, paleocommunity researchers can focus their efforts on gathering more, rather than larger, samples in less time.

Meeting Home page GSA Home Page