A comprehensive pollution survey of polychlorinated dibenzo-p-dioxins and dibenzofurans by means of principal component analysis and partial least squares regression

The combined use of principal component analysis and partial least squares regression to the interpretation of polychlorinated dibenzofuran (PCDF) and dibenzo-p-dioxin (PCDD) patterns in crabs are investigated. By exploratory principal component analysis, the PCDFs and the PCDDs are ranked according to their specificity to the emission source. Among the most specific PCDFs to the emission are the 12368/13479-PeCDF and 12478-PeCDF. The 123789-HxCDD isomer is the only specific PCDD to the emission source. Partial least squares regression is used both to explore the determination of sex specific isomers, and in confirmatory analysis by quantitatively relating the pollution waste to the distance from the emission source. 123678-HxCDF is statistically significant for the separation between male and female crabs collected at the same distance from the emission source, while no separation between male and female crabs is evident from 1234789-HpCDF, Octa-CDF, 2,3,7,8-TCDD and Octa-CDD. This observation is somewhat more shaded than what has previously been reported in the literature on differences in male and female crabs. A linear relationship exists between the logtransformed distance and the concentration of PCDFs and PCDDs in male crabs collected between 18 and 33 km from the source. This model underestimates the predicted distance of crabs collected between 33 and 56 km, possible due to a higher sedimentation rate at distances closer to the fjord and the factory.

Although applied to PCDF/PCDD data in this case, the outlined multivariate methods should be applicable to all kind of similar environmental surveys.