Binary coefficients and clustering in biostratigraphy

The large data arrays common in biostratigraphy make subjective groupings difficult. Because of this, biostratigraphers have commonly based conclusions on the occurrences of relatively few species. The use of binary similarity coefficients, cluster analysis techniques, and digital computers allows a polythetic approach to biostratigraphy. The approach is free of circular and a priori reasoning in that it is established that assemblage zones (major clusters in a dendrogram) are present before the biostratigraphic utility of various taxa for the recognition of the zones is determined. A measure for determining the biostratigraphic fidelity of a species for established zones is proposed. Published data on the distribution of ostracodes and foraminifers from the Eocene and Oligocene of Mississippi and Alabama are used to demonstrate a method of using coefficients and clustering in establishing a zonation.