Testing similarity coefficients for analysis of the fossil record using clustering methods: the Palaeozoic flora as a study case.

Abstract

This paper reports a global methodological approach based on the similarity and clustering methods of the Palaeozoic plant fossil record using a comparative approach between two similarity measures: the Jaccard and Raup-Crick Coefficients. The results show that although the Raup-Crick Coefficients clearly have the potential for providing more robust results, the consequences of the extinction processes are better reflected in the similarity analysis based on the Jaccard Coefficients. On the other hand, the cluster analysis based on UPGMA algorithm shows four robust clusters and reveals new evidence for the singularity of Mississippian flora. Finally, the results obtained reveal that similarity and cluster analysis are powerful tools to interpret the consequence of the processes modifying the taxonomic composition of the several analyzed Palaeozoic time units.