An evaluation of the importance of the sampling step to the total analytical variance -: A four-system field-based sampling intercomparison study for hydrophobic organic contaminants in the surface waters of the open Baltic Sea

Correctly assessing the concentrations of hydrophobic organic contaminants in aquatic samples is faced with significant analytical challenges in terms of contamination, fractionation and sorptive losses to the sampling system. While little studied from an analytical perspective, the sampling step is likely to be the largest source of uncertainty in the total analytical chain. In this article, four different water-sampling systems for collection of particle-bound and dissolved polychlorinated biphenyls (PCBs) were evaluated in the surface water of the Baltic Sea. Individual concentrations of sixteen PCB congeners sampled with stainless steel tubing (system A) spanned between 0.04 and 3 fM (0.02 and 1 pg/L) in the particulate fraction and between 0.04 and 8 fM (0.01 and 2 pg/L) in the dissolved fraction. The sampling variance introduced to the data, expressed as relative standard deviation (RSD %) for single PCB congeners in system A varied between 14 and 77% (average 35%) in the dissolved fraction and between 8 and 20% (average 13%) in the particulate fraction. The implication of the constrained sampling and total variances is that variations in environmental data that are larger than about 50% RSD may be interpreted as reflecting natural processes as opposed to merely methodological variance.