Amino acid composition as a taxonomic tool for molluscan fossils: An example from Pliocene-Pleistocene Arctic marine deposits

Amino acid compositions of more than 550 individuals of three common subarctic/arctic molluscan genera, Hiatella, Mya, and Astarte, were measured to determine the feasibility of using such data to classify unidentifiable shell fragments belonging to one of these genera. To account for diagenetically induced variation in amino acid composition of the fossils, equations were developed to reconstruct the original composition of each shell based on its extent of amino acid epimerization. The classification procedure involved calculating the multivariate similarity (coefficient of similarity (CS)) between the adjusted amino acid composition of an unknown shell and that of a set of references. CSs were calculated for an additional 141 identifiable shells of the three common genera to evaluate the performance of the classification method; the procedure correctly classified 93% of the shells. In addition, CSs were calculated for 162 identifiable shells of five other common subarctic/arctic genera (Portlandia, Spisula, Macoma, Cardita, and Arctica) to test the ability of the method to discriminate shells belonging to genera not included in the reference set and are therefore necessarily misclassified. Overall, CSs for correctly classified shells ( x ̄ = 0.891) are higher than for misclassified shells ( x ̄ = 0.809), and the ratio of correctly to incorrectly classified individuals increases with increasing CS. Based on these empirical tests, only those classifications supported by CSs greater than 0.88 were considered reliable; and of these, approximately 6% may be misclassified. The classification procedure was applied to sixty-four unidentifiable molluscan shell fragments recovered from boreholes drilled onshore of Nome, Alaska, of which twenty-six were classified as Hiatella, Mya, or Astarte with a CS greater than 0.88.