CHAPTER 22 . Chemometric Applications to Seawater Analysis
415
markers to individuate the sources and to assess the transformation processes of organic material in various environments. Subsurface (0.2 m depth) water samples were
collected and several hundred data were treated by the multivariate method of correspondence factorial analysis (CFA) and non-parametric HCA. The chemometric treatments permitted to distinguish between different inputs in the dissolved and in particulate material and to classify the samples in two main groups: one contains planktonic markers and the second can further discriminate between a zooplanktonic and
a microbiological pole.
PCA was employed to categorise sources of variability in rainwater data, and PLS
(partial least-square) regression was used to compare the data for the same trace elements in crust, sea salt and smelter "signatures" (Vong 1993). 131 rainwater samples were
analysed for eighteen trace elements, principally metals. PCA extracted three significant components: PCl characterizes mainly the insoluble crustal elements, PC2 discriminates soluble sea salts and PC3 is mainly related to As, Sb, Cu and Se concentrations.
22.2.1.5
Particulate
In the northern Baltic Sea (Strandberg et al. 1998) the composition profile and the
spatial differences of many pollutants that are globally distributed by transport through
air and waters were studied, namely PCBs, hexachlorocyclohexanes (HCHs),
hexachlorobenzene (HCBz), DDT and related metabolites, chlordane-related compounds (CHLs), dieldrin. The Baltic Sea contains high levels of halogenated organic
compounds that, depending on their hydrophobicity and organic carbon-water partition coefficients, can be dissolved in the water phase or be associated with particles.
Since atmospheric inputs often predominate in remote water areas, and since a limited biotransformation level takes places in particles, the composition of settling particulate matter can give indications of the situation of the air and of the contamination load of seawaters. The sampling of settling particulate matter was carried out in
remote coastal and in offshore stations in the Bothnian Bay and in the Bothnian Sea,
characterized by a large water exchange with open sea. Sediments were collected in
coastal and in offshore stations. Surface bottom sediments were sampled in the coastal
stations at Harufjarden (HF) and Simpnas (SN) in autumn and spring, to compare the
contamination load at the coastal stations to that of surface bottom sediments. At one
coastal station sediment samples (SED) were collected in autumn and in spring, in
order to individuate seasonal variations. The levels of settling particulate matter (SPM)
at the offshore stations were 10 times higher on the average than the coastal stations,
likely because of an increased carbon content in the offshore settling particulate matter. The analysis was performed after cleanup procedures by GC/MS. The data set was
then treated by PCA. Figure 22.5 reports a combined plot of scores (samples, in bold)
and loadings (compound class, in italic) with the combined origins for the first two
PCs. The plot permits one to study the distribution of the samples and to determine
how the individual variables influence the samples. The profiles of the settling particulate matter from the four stations show great similarity, the major difference being the greater concentration of HCHs in the coastal stations (SN and HF), while PCBs
are the most abundant pollutants at the offshore stations (SRS and F9). A more densely
populated area might cause the elevated levels of DDTs in both stations. The plot also
415
markers to individuate the sources and to assess the transformation processes of organic material in various environments. Subsurface (0.2 m depth) water samples were
collected and several hundred data were treated by the multivariate method of correspondence factorial analysis (CFA) and non-parametric HCA. The chemometric treatments permitted to distinguish between different inputs in the dissolved and in particulate material and to classify the samples in two main groups: one contains planktonic markers and the second can further discriminate between a zooplanktonic and
a microbiological pole.
PCA was employed to categorise sources of variability in rainwater data, and PLS
(partial least-square) regression was used to compare the data for the same trace elements in crust, sea salt and smelter "signatures" (Vong 1993). 131 rainwater samples were
analysed for eighteen trace elements, principally metals. PCA extracted three significant components: PCl characterizes mainly the insoluble crustal elements, PC2 discriminates soluble sea salts and PC3 is mainly related to As, Sb, Cu and Se concentrations.
22.2.1.5
Particulate
In the northern Baltic Sea (Strandberg et al. 1998) the composition profile and the
spatial differences of many pollutants that are globally distributed by transport through
air and waters were studied, namely PCBs, hexachlorocyclohexanes (HCHs),
hexachlorobenzene (HCBz), DDT and related metabolites, chlordane-related compounds (CHLs), dieldrin. The Baltic Sea contains high levels of halogenated organic
compounds that, depending on their hydrophobicity and organic carbon-water partition coefficients, can be dissolved in the water phase or be associated with particles.
Since atmospheric inputs often predominate in remote water areas, and since a limited biotransformation level takes places in particles, the composition of settling particulate matter can give indications of the situation of the air and of the contamination load of seawaters. The sampling of settling particulate matter was carried out in
remote coastal and in offshore stations in the Bothnian Bay and in the Bothnian Sea,
characterized by a large water exchange with open sea. Sediments were collected in
coastal and in offshore stations. Surface bottom sediments were sampled in the coastal
stations at Harufjarden (HF) and Simpnas (SN) in autumn and spring, to compare the
contamination load at the coastal stations to that of surface bottom sediments. At one
coastal station sediment samples (SED) were collected in autumn and in spring, in
order to individuate seasonal variations. The levels of settling particulate matter (SPM)
at the offshore stations were 10 times higher on the average than the coastal stations,
likely because of an increased carbon content in the offshore settling particulate matter. The analysis was performed after cleanup procedures by GC/MS. The data set was
then treated by PCA. Figure 22.5 reports a combined plot of scores (samples, in bold)
and loadings (compound class, in italic) with the combined origins for the first two
PCs. The plot permits one to study the distribution of the samples and to determine
how the individual variables influence the samples. The profiles of the settling particulate matter from the four stations show great similarity, the major difference being the greater concentration of HCHs in the coastal stations (SN and HF), while PCBs
are the most abundant pollutants at the offshore stations (SRS and F9). A more densely
populated area might cause the elevated levels of DDTs in both stations. The plot also