Abstract

 Environmental risk of contaminated sediments can be evaluated using a number of approaches, including Sediment Background, Equilibrium Partitioning, Apparent Effects Threshold and Sediment Quality Triad Approach etc. In this study I have focused on establishing links between concentration of contaminants in natural sediment and effects on resident fauna, i.e. combining elements from the SLC and AET approaches.
  Structure of benthic communities are determined by several factors and processes, and moreover individual components (e.g. number of individuals within species) often are interrelated due to predation and competition. Therefore, to reveal relationships between sediment contaminants and structure of benthic communities the analysis must be carried out including all available information characterising the system. Only multivariate techniques fulfil these requirements.
  Using EMAP data from 5 coastal regions along the East coast of USA multivariate statistic (PLS Partial Least Squares) was used to describe the relationship between predictor variables (i.e. contaminant concentrations, sediment characteristics, water quality, depth, salinity) and response variables describing the community structure (e.g. species richness and diversity) and sediment toxicity.
  The number of significant predictor variables ranged between 25 and 40 with heavy metals generally being the most influential on benthic structure followed by sediment characteristics, WQ and PAH´s. In comparison, ERL/ERM predicted PCB's and chlorinated pesticides to have the highest impact. Depending on the actual response variables PLS models were able to describe between 40 and 60 % of the variation in benthic structure while predictions of whole sediment toxicity consistently were low. The PLS regression models were applied in an artificial clean-up exercise of sediments from the Chesapeake Bay region: by reducing concentrations of heavy metals in Cu contaminated sediments to the average level indicated by the Me-TOC relation the B-IBI (index of benthic quality) increased from -1.4 to -0.15. Hence, verified PLS models can be used to predict benthic quality in connection to mitigation/restoration efforts.