Statistical testing of community patterns: uppermost Hamilton Group, Middle Devonian (New York State, USA)

We present an extensive and rigorously controlled quantitative paleoecological study within an interval of inferred coordinated stasis. This Middle Devonian Hamilton Group study completes a 20-yr project by providing data within the unresolved upper Hamilton Group section. Together with other rigorously controlled studies, these data sets have the potential to address the larger question of coordinated stasis in the fossil record. We collected data from the Windom Member, Moscow Formation (uppermost Hamilton Group), to test different statistical approaches to define paleocommunities. We evaluate various techniques, including non-parametric multidimensional scaling and agglomerative hierarchical clustering to decipher community patterns. Additionally, we advocate regular use of cluster significance testing along with ANOSIM (i.e. analysis of similarities) when examining ecological data. Together these techniques test the significance of sample groups more rigorously than conventional testing (e.g. discriminant analysis or analysis of variance (ANOVA)). Our results indicate that communities within this upper Hamilton Group interval exhibit variable taxonomic membership within a relatively stable ecological structure. (C) 2002 Elsevier Science B.V. All rights reserved.