Uncertainty in Paleoecological Studies of Mercury in Sediment Cores

Increased recognition of the ecological damage of mercury (Hg) has focused attention on quantifying spatial and temporal patterns of Hg deposition. Studies are commonly based on core chronologies and use a combination of techniques to measure parameters such as bulk density, percent solids, Hg concentration, and radionuclide activity. Little attention is generally devoted to the propagated error associated with these measurements. We identified the impact of sources of uncertainty on stratigraphic Hg determinations for Florida Everglades and Lake Erie cores. Large errors may be introduced by converting wet sample Hg content to dry-weight concentrations. Drying of sediments at 55 °C caused Hg losses of 18%. Samples, air-dried at room temperature, retained considerable moisture and required corrections for remaining water content. Frozen sediments did not lose Hg during a 72-day storage. Random error in radionuclide analysis of cores resulted in dating uncertainty of ±1.2 yr in 10 yr old deposits. This error increased to ±20 yr in 100 yr old sediments. Propagation of small errors in each step of the analysis (while adhering to strict QA/QC criteria) produced compounded uncertainties of ±11 and ±29% in Hg concentrations under different analytical rigor, and errors of up to ±73% in Hg accumulation rates in older sediments. Enrichment factors, comparing uncertain recent and historic Hg accumulation rates, differed by as much as ±48%. Uncertainty in paleoecological studies of mercury needs to be documented in order to correctly evaluate trends and remediation efforts.