Multivariate assessment of the required sample size for community paleoecological research

When conducting paleocommunity research, a sample size that is too small may lead to erroneous ecological conclusions, while an increasingly larger sample size will eventually result in diminishing returns. Using 30 datasets from the Paleobiology Database and the literature, I present evidence that the median required sample size for paleocommunity research is 25 to 50 individuals. Each sample within each dataset was randomly subsampled 1000 times to five proportional sizes (50%, 25%, 10%, 5%, 2.5%). Each of the 1000 subsampled datasets created for each proportional sample size was correlated with its corresponding complete dataset using three multivariate methods: the Mantel Test, Pearsonʹs Product Moment Correlations of non-metric multidimensional scaling ordination axis scores, and the Procrustean Randomization Test (PROTEST). For median sample sizes greater than 50 individuals, all Mantel R-statistics are greater than 0.93, Pearsonʹs r-values for NMDS axis-one scores are greater than 0.86, and PROTEST m2-values are greater than 0.82. Between a median sample size of 25 and 50, the R-statistics, r-values, and m2-values are relatively plateaued, and then subsequently plummet at median sample sizes less than 25. Thus, the multivariate paleocommunity signal, within each dataset is effectively preserved when all median sample sizes are greater than 25 to 50 individuals.