Assessment and management of pesticide pollution at a river basin level part II: Optimization of pesticide monitoring networks on surface aquatic ecosystems by data analysis methods.

Abstract The high cost of extensive pesticide monitoring studies, required for the protection of water resources, and the necessity of early identification of environmental threats, highlighted the need for prioritization of pesticides and sampling sites to be monitored. The aim of this study was to develop an optimum surface water monitoring network at a catchment scale including only the sites of a catchment vulnerable to pesticide pollution. The identification of sampling sites vulnerable to pesticide pollution (VPS) was based on the data of an intensive monitoring survey of 302 pesticides in 102 stationary sampling sites located on the surface water network of a river basin. In the proposed methodology the left-censored data of the analytical results derived from the above mentioned monitoring campaign were included in the statistical analyses by transforming all the raw data into categorical variables and arranging them in ordinal scales based on ecotoxicological thresholds derived from pesticide toxicity tests on aquatic non-target organisms. The categorized data were subjected to Categorical Principal Component Analysis with Optimal Scaling. For the identification of the VPS, the Squared Mahalanobis Distance criterion was applied on the extracted values (scores) of the significant principal components. With this methodology a 46% reduction in the number of the monitoring stations was achieved. This approach will be valuable in establishing more cost effective monitoring schemes in the future in other basins and in developing targeted measures to eliminate or limit the effect of critical pollution sources in surface aquatic systems. Moreover, by applying the proposed methodology, historical monitoring data can be used to initiate more efficient pesticide monitoring campaigns in the future. Graphical abstract Unlabelled Image Highlights • Optimization of an existing pesticide surface water monitoring network. • The methods used pesticide data, ecotoxicity thresholds, and in situ inspection. • The methodology based on CATPCA and Squared Mahalanobis Distance methods. • Identification of monitoring stations vulnerable to pesticide pollution. • A reduction of 46% of the stationary sampling stations was accomplished. [ABSTRACT FROM AUTHOR]