Multivariate analyses for investigating highly polluted marine ecosystem: The case study of Mar Piccolo (Taranto, South Italy)

Abstract

This study presents the benefits of the application of multivariate techniques for the hazard assessment of a heavily polluted marine ecosystem. The study area, named Mar Piccolo, near Taranto city (Southern Italy), is a shallow marine basin located nearby an industrial compromised area, declared by the national government as Contaminated Site of Environmental Interest (SIN) due to the presence of long-lasting large industrial settlements that have severely impacted the marine environment. Besides the anthropogenic pressures, the marine basin is characterized by high productivity of several species at different trophic levels of the food chain, that confers to the bottom sediments an unusually high organic matter content. The latter is even enhanced by the presence of freshwater springs in the marine basin. The dynamism of the ecosystem demands for advanced evaluation tools for its correct characterisation. Multivariate ANOVA allied by Hierarchical Clustering are applied in this research to provide a readable picture of the quality status of the sediments, aiming at identifying different loading factors and getting insights into their simultaneous effects. The innovative approach adopted circumvents inefficient time-consuming procedures, usually required by the conventional univariate analyses of each parameter selected for sediment characterisation. A comprehensive hazard assessment was possible thanks to a clear graphical representation of the hazard distribution that supported the identification of the hazard controlling factors, confirming the efficiency of the adopted approach. The tools proposed herein can thus be recommended for the decision makers in investigating and interpretation of the quality status of complex polluted marine eco-systems.

Keywords:

• polluted sediments

Acknowledgments

ISPRA is gratefully acknowledged for chemical analyses and the availability of dataset.