Statistical evaluation of the effect of sample preparation procedure on the results of determinations of selected elements in environmental samples. Honey bees as a case study.

Honey bees became a frequently studied environmental research object due to the fact that they are considered as indicators of the environmental contamination with metals. Such studies require reliable methods of sample preparation that allow comparison of the results obtained in different laboratories. In this paper, different variants of sample preparation were examined, including washing, milling and mineralization. Determination of total contents of Cd, Pb, Co, Cr, Cu, and Zn was performed using ICP MS. Each procedure of the sample preparation was characterized in terms of repeatability and recovery and the resulting combined standard uncertainty was evaluated. The relative UNC ranges for the treatment procedures examined are: 11-16% (washing, open digestion), 13-14% (open digestion), 13-51% (washing, closed digestion), 12-37% (washing, milling, closed digestion), 13-68% (closed digestion), and 12-51% (milling, closed digestion). In almost all cases the lower limits of these ranges correspond to Zn determinations while the upper limits to Pb. Analytical results of metal determination obtained for samples prepared according to six different procedures do not differ significantly when compared within the expanded uncertainty (±2×u(c¯)). The procedures were compared statistically using one-factor analysis of variance (ANOVA) and chemometrically on the basis of Principal Component Analysis (PCA), which allowed to demonstrate some general recommendations regarding environmental monitoring for matrices other than honey bees.