MULTIPLE FACTORS IN EXTINCTION RISK: TESTING MODELS OF EXTINCTION SELECTIVITY IN EOCENE BIVALVES USING PATH ANALYSIS

Ecological characteristics such as abundance and geographic range size are widely believed to influence extinction risk and to have influenced diversity dynamics of clades. While the importance of such factors has been established for many extant and extinct taxa, most studies have assumed independence among factors when measuring their effects on extinction rates. Here I use the Eocene marine fossil record of the U.S. Gulf Coastal Plain to assess the covariance between species ecological characteristics - abundance, body size, and geographic range - and their combined effects on extinction rates. Contrary to the general assumption of independence, preliminary results for a broad sampling of bivalve species show a positive covariance between body size and geographic range and between body size and abundance. Given the non-independence of abundance, body size, and geographic range at this broad taxonomic scale, I use path analysis to examine the direct and indirect effects of these multiple factors on extinction rates for a subset of bivalve superfamilies (Carditoidea, Pectinoidea, and Veneroidea) that differ in basic aspects of life history and ecology. Species-level data for these clade-based analyses are gathered through quantitative sampling in the central and western Gulf Coastal Plain and use of existing museum collections and literature records. Using a model selection approach, I test the generality of extinction rate predictors among these three groups and find that geographic range explains much of the variation in extinction probability with body size having relatively little effect. Given the interactions between ecological characteristics, multivariate approaches are essential for establishing the generality of factors influencing extinction risk and for assessing their relative importance over geologic time.