INTRODUCING EXPLORATORY DATA ANALYSIS TECHNIQUES IN INTRODUCTORY PALEONTOLOGY: DIVERSITY INDICES, SIMILARITY COEFFICIENTS, AND CLUSTER ANALYSIS

The distribution of fossil floras and faunas provides insight into the temporal and spatial relationships in which these organisms lived, thrived, and were preserved. Additionally, geographic distributional patterns provide evidence about the relationships between organisms, communities, and biomes. Linking current analytical methods routinely applied in ecology to the deep time record often is overlooked at the undergraduate level, relegating such approaches to graduate courses. But, the introduction of such techniques at the undergraduate level provides students with basic data analytical approaches that can be applied to other geoscience disciplines. A laboratory exercise is presented in which students use basic diversity measures, correlation coefficients, and cluster analysis to reconstruct the complexities of a Carboniferous peat forest preserved as a T₀ assemblage.

The data set originates from the Lower Pennsylvanian (Langsettian) Mary Lee Coal zone in the Black Warrior Basin, Alabama, and provides a temporally and spatially constrained perspective on a Pennsylvanian peat-mire ecosystem at a community scale. Rapid coseismic subsidence resulted in estuarine (tidalite) deposition that buried the Blue Creek forest in growth position, preserving an autochthonous leaf litter associated and erect, siltstone-cast trunks. Strip mining exposed this horizon that was traced and assessed over 0.5 km² area, from which 47 form taxa were recognized from seventeen 10 m² quadrats.

Principles behind the Shannon-Weiner, Simpsons, and Marglef diversity measures are introduced, and students are asked to calculate each indice. Subsequently, they compare the metrics against the original data to determine which factors control the calculated results. The Sorenson’s, Jaccard, and Simple Matching binary (presence/absence) similarity coefficients are introduced, and calculations made to understand which assemblages are most similar. Subsequently, these are used in agglomerative hierarchical methods in which R-mode clustering results in an hierarchical arrangement such that the relationships between the different groups are apparent and dendrograms constructed. These relationships then are spatially plotted providing an understanding of the mosaic pattern in the assemblage.