408
M.e. Gennaro . S. Angelino
required to solve a mixing model. With respect to PCA, PVA offers the advantage to
report the final results as percentages rather than as abstract scores or loadings: this
makes the end-member chemistry easier to interpret in a scientific context and allows
the evaluation of concentration gradients by end-member mapping. Using the data of
the cruise of April 1995, five separate PCB congener fingerprints were identified and
their presence was correlated to the different locations in the bay. As concerns the different cruises, both within and between them, there is a large range of values, not correlated with seasonality.
On 26-28 November 1986, through a fracture in the Claymore pipeline close to its
junction with the main Piper/Flotta oil line (approximately 10 km north of the Claymore platform), 2000-3000 t of crude oil leaked into the North Sea (Grahl-Nielsen
and Lygre 1990). In the same period a typical industrial spill also happened: an overflow of a slop tank ended up in Byfjorden, Bergen. A study was carried out in 1990 for
the identification of oil related to the two spills. The identification of samples from a
spill becomes increasingly difficult the longer the oil has been exposed to light, air,
water and microorganisms. A method based on the separation of the aliphatic fraction of the oil followed by GC-MS of polycyclic alkanes of low volatility was used for
the analysis. In addition to samples from these two spills, samples of other North Sea
crudes were analysed by PCA. The pairwise distances between the samples in the
multivariate space was calculated for the most doubtful cases by the k*-nearestneighbour (k*-NN) method. The plot of PC1 vs. PC2 and the k*-NN results show significant difference between the samples of different origin.
By PCA and FA treatments (Machado 1993), the dependence on pH and copper(II)
concentration of fluorescence signals of fulvic acids present in coastal marine waters
collected in the Atlantic Ocean at Mindelo, Vila do Condo (20 km to the north of Porto)
was evidenced.
With particular attention to sites devoted to hatcheries of edible mussels, the degree of pollution in the Gulf of Trieste (Northern Adriatic Sea), in particular in areas
receiving industrial, agricultural and urban wastes, was studied with the help of
chemometric methods (Reisenhofer et al.1996). Four trace metals (Zn, Cd, Pb, Cu) and
5 nutrients (phosphate, silicate, NH 3 , NOi, NO;-) were determined in the filtered fraction of coastal surface water. Sampling was effected with a monthly frequency in the
year 1993 in 6 representative stations to obtain information about the effect of possible seasonal trends and of the influence of meteorological factors. The use of PC
method allows us to identify the sources and the typology of pollution and to dis crimi -
nate between the quality of the water sampled at the different sites. In this area, largely
inhabited and with low agricultural activity, the first and the most important factor
resulted related to NO;- and silicate concentrations. This fact was attributed to the
leachate of farmlands, whereas urban wastes (NH3 from sewage and phosphate from
domestic detergents) were related to the third factor. The heavy metal content was
related to the second factor. It is worthwhile to mention the result obtained for Zn,
which is not related to the other metal ions and seems therefore to have a different
origin. A similar study was performed in the Marano lagoon: three sites were chosen
in correspondence to hatcheries of edible clams to evaluate the environmental quality
of the waters. The distribution of heavy metals (Cu, Pb, Cd, Zn, Cr, Mn, Ni) in sediments,
in sea water at different depths, particulate matter and soft tissues of mollusks was evaluated. The results ofHCA analysis and LDA (linear discriminant analysis) indicate a simi-
M.e. Gennaro . S. Angelino
required to solve a mixing model. With respect to PCA, PVA offers the advantage to
report the final results as percentages rather than as abstract scores or loadings: this
makes the end-member chemistry easier to interpret in a scientific context and allows
the evaluation of concentration gradients by end-member mapping. Using the data of
the cruise of April 1995, five separate PCB congener fingerprints were identified and
their presence was correlated to the different locations in the bay. As concerns the different cruises, both within and between them, there is a large range of values, not correlated with seasonality.
On 26-28 November 1986, through a fracture in the Claymore pipeline close to its
junction with the main Piper/Flotta oil line (approximately 10 km north of the Claymore platform), 2000-3000 t of crude oil leaked into the North Sea (Grahl-Nielsen
and Lygre 1990). In the same period a typical industrial spill also happened: an overflow of a slop tank ended up in Byfjorden, Bergen. A study was carried out in 1990 for
the identification of oil related to the two spills. The identification of samples from a
spill becomes increasingly difficult the longer the oil has been exposed to light, air,
water and microorganisms. A method based on the separation of the aliphatic fraction of the oil followed by GC-MS of polycyclic alkanes of low volatility was used for
the analysis. In addition to samples from these two spills, samples of other North Sea
crudes were analysed by PCA. The pairwise distances between the samples in the
multivariate space was calculated for the most doubtful cases by the k*-nearestneighbour (k*-NN) method. The plot of PC1 vs. PC2 and the k*-NN results show significant difference between the samples of different origin.
By PCA and FA treatments (Machado 1993), the dependence on pH and copper(II)
concentration of fluorescence signals of fulvic acids present in coastal marine waters
collected in the Atlantic Ocean at Mindelo, Vila do Condo (20 km to the north of Porto)
was evidenced.
With particular attention to sites devoted to hatcheries of edible mussels, the degree of pollution in the Gulf of Trieste (Northern Adriatic Sea), in particular in areas
receiving industrial, agricultural and urban wastes, was studied with the help of
chemometric methods (Reisenhofer et al.1996). Four trace metals (Zn, Cd, Pb, Cu) and
5 nutrients (phosphate, silicate, NH 3 , NOi, NO;-) were determined in the filtered fraction of coastal surface water. Sampling was effected with a monthly frequency in the
year 1993 in 6 representative stations to obtain information about the effect of possible seasonal trends and of the influence of meteorological factors. The use of PC
method allows us to identify the sources and the typology of pollution and to dis crimi -
nate between the quality of the water sampled at the different sites. In this area, largely
inhabited and with low agricultural activity, the first and the most important factor
resulted related to NO;- and silicate concentrations. This fact was attributed to the
leachate of farmlands, whereas urban wastes (NH3 from sewage and phosphate from
domestic detergents) were related to the third factor. The heavy metal content was
related to the second factor. It is worthwhile to mention the result obtained for Zn,
which is not related to the other metal ions and seems therefore to have a different
origin. A similar study was performed in the Marano lagoon: three sites were chosen
in correspondence to hatcheries of edible clams to evaluate the environmental quality
of the waters. The distribution of heavy metals (Cu, Pb, Cd, Zn, Cr, Mn, Ni) in sediments,
in sea water at different depths, particulate matter and soft tissues of mollusks was evaluated. The results ofHCA analysis and LDA (linear discriminant analysis) indicate a simi-