Exposure assessment of pesticides in a shallow groundwater of the Tagus vulnerable zone (Portugal): a multivariate statistical approach (JCA)

PURPOSE: To assess groundwater exposure to pesticides, in agricultural areas of ‘Ribatejo’ region (Portugal), and the influence of some key factors in that exposure, field, laboratory and modelling studies were carried out. METHODS: The study was performed in maize, potato, sugar beet, tomato and vegetables agricultural areas, located in a shallow aquifer, with pesticides use and, in most cases, with irrigation practices. Pesticides used in the studied agricultural areas and having leaching potential were selected, being considered also other pesticides included in priority lists, defined in Europe. Evaluation of groundwater exposure to pesticides was carried out by successively: (1) groundwater sampling in seven campaigns over the period 2004–2006; (2) pesticide analysis [including isolation and concentration from the groundwater samples and further determination by gas chromatography–mass spectrometry (GC–MS) of 14 herbicides, four insecticides and two metabolites]; and (3) analysis and discussion of the results by applying joint correspondence analysis (JCA). RESULTS: From the 20 pesticides and metabolites selected for the study, 11 were found in groundwater. Pesticides and metabolites most frequently detected were atrazine, alachlor, metolachlor, desethylatrazine, ethofumesate, α-endosulfan, metribuzine, lindane and β-endosulfan. The results showed that groundwater exposure to pesticides is influenced by local factors—either environmental or agricultural, as precipitation, soil, geology, crops and irrigation practices. Spring and autumn were more associated with the detection of pesticides being more likely to observe mixtures of these compounds in a groundwater sample in these transition seasons. CONCLUSIONS: This work evidences the importance of models, which evaluate pesticides environmental behaviour, namely their water contamination potential (as Mackay multicompartimental fugacity model) and, specially, groundwater contamination potential (as GUS and Bacci and Gaggi leaching indices), in pesticide selection. Moreover, it reveals the importance to adapt proper statistical methods according to level of left-censored data. Using JCA was still possible to establish relations between pesticides and their temporal trend in a case study where there were more than 80% of data censored. This study will contribute to the Tagus river basin management plan with information on the patterns of pesticide occurrence in the alluvial aquifer system.