ASSESSING THE PRECISION AND ACCURACY OF OXYGEN ISOTOPE SEASONALITY RECONSTRUCTIONS DERIVED FROM MIDDEN MOLLUSKS

Interpretation of oxygen isotope profiles from mollusk shells excavated in archaeological middens is complicated by a number of unique challenges. These include the likelihood that the species are from estuaries with unknown and variable δ¹⁸O_water values. Furthermore, humans may have transported the shells over some unknown part of a salinity gradient, adding more uncertainty to δ¹⁸O_water estimates. Additionally, ancient peoples typically exploited species which are subject to irregular growth rates and periods of growth cessations. These challenges limit the utility of molluscan δ¹⁸O profiles for climate reconstructions and season of capture estimates, and preclude calculations of absolute paleotemperature or salinity in most circumstances. Alternatively, fine-scale analysis of δ¹⁸O values may provide useful insight into the relative seasonal amplitude of salinity or temperature variation. However, this approach assumes a clear understanding of the impact of shell growth patterns on δ¹⁸O profiles. While advances in clumped isotope analysis may someday permit circumvention of δ¹⁸O_water uncertainty, the problems concerning shell transport and growth rate variation will remain.

Samples of modern and ancient hard clam (Mercenaria spp) and American oyster (Crassostrea virginica) from the marshes surrounding Sapelo Island, Georgia, USA were analyzed to assess the potential impact of these uncertainties on seasonal-scale analysis of δ¹⁸O in mollusk shells. Environmental time series data from multiple locations surrounding the barrier island are compared using oxygen isotope modeling to assess similarities and differences in temperature, salinity, and δ¹⁸O_water between potential habitats. In turn, these models are compared to δ¹⁸O_shell profiles measured in ancient shells and from “blind” modern control samples in which the time and exact location of collection were unknown to those performing the analysis.

The results suggest that even under ideal circumstances, incorrect assessment of seasonal temperature amplitude and season of capture will be made on individual specimens, yet useful reconstructions can still be gleaned. The likelihood of error is related to habitat, the degree of growth rate variation, occurrence of growth cessations, and ontogenetic age.