Sampling statistic limitations on benthic foraminifera chemical and isotopic data

Abstract

Analyses of benthic foraminiferal isotopic and chemical properties are typically done on a small number of individuals (often 5–10). Because benthic sediment mixing produces a mix of individuals whose properties were determined at times encompassing several thousand years, a sampling of several individuals does not adequately represent the average mixture. A quantitative statistical analysis shows that this effect places a fundamental limit on measurement precision in cores, especially where rapid environmental changes occur over short vertical distances. The fundamental parameters defining this limit are the rate of change of the property (per unit distance) and the effective mixing depth. In cores found in the Vema Channel/Rio Grande Rise region, where sedimentation rates are about 1 cm ka −1 and mixing depths are about 3 cm, a single sampling of six individuals can produce a result that has an expected one sigma standard deviation of 6–25% of the observed range. Precision can be improved by using more individuals or running more replicate analyses, but the result only improves as the square root of the number of individuals or replicates. This effect is illustrated by Cd Ca and isotopic data from low (0.7 cm ka −1) and moderate (1.5 cm ka −1) sedimentation rate cores. Although precision is compromised the most in low sedimentation rate cores, this effect contributes to imprecision in high sedimentation rate cores as well.