Pattern Recognition for Spatial Distribution of Heavy Metals and Hydrocarbons in a Polluted Marine Basin

Pattern recognition techniques are largely exploited for the assessment of the contamination degree within complex ecosystem. On the other side, geostatistical techniques are often used to generate spatial pollution maps at global and local scales. In this paper concentrations of ten metals and metalloids (Hg, Cd, Pb, As, Cr, Cu, Ni, Zn, Fe, Mn), 16 EPA polycyclic aromatic hydrocarbons (PAHs) and total petroleum hydrocarbons (THC), determined within bottom sediments of a heavily polluted marine basin, were analyzed coupling pattern recognition techniques (PCA - Principal Component Analysis, OPLS-DA Orthogonal Partial Least Square Discriminate Analysis) with Kriging geostatistical technique, in order to provide a comprehensive spatial and vertical distribution of the contaminants in a highly polluted marine basin. A first-screening contamination assessment was performed by using PCA exploiting the data from 66 sediment samples, collected from 22 sites, at three different depth level, 0–20 cm, 80–100 cm, and 180-200 cm. The content of total organic carbon and granulometric features of sediments were also included in the analysis aiming to investigate the effect of depositional environment. In addition, five PAHs diagnostic molecular ratios (LWM/HWM, BAP/Bghip, BaA/(BaA+Chry), IcdP/(IcdP+BghiP), Fluo/(Fluo+Pyr), Ant/(Ant+Phen) were considered aiming to getting insight on the origin of PAHs content. The analysis allowed to identify a hot-spot area in the marine track receiving the Sarno River flow, that resulted to be characterized by the intense contamination from Pb, Zn, Hg and Cd and from High Weight Molecular (HWM) PAHs. The contamination in this track seems to reach surficial (20 cm) and middle depth level (120 cm). In turn, the shoreline coastal track resulted characterized by geogenic As and Ni and high content of sand regardless depth. The track approaching offshore area resulted not interested by organic or inorganic contamination, except for relatively high content of Mn with respect to the rest of the investigated area. The combination of multivariate techniques and geostatistical tools was demonstrated to be an easy/to/use method and a reliable approach for evaluating the spatial and vertical distribution of trace elements and organic contaminants, identifying the major contamination pathway.