A multivariate analysis system for assessing the causes of pesticide residues in organic fruit crops was developed for research on organic agriculture conducted in 2016–2017. During the project, a total of 66 organic fruit crops were examined and subjected to detailed multivariate analysis, including 25 apple, 29 raspberry, and 12 strawberry crops from various regions of Poland. Pesticide residue analysis included soil, leaves, and fruit. Pesticide residues were detected in 81.8% of the crops tested, representing 44.4% of all samples. Residues of one compound were detected in 34.8% of the crops, two compounds in 16.7%, three compounds in 18.2%, and four or more compounds in 12.1%. No residues were detected in 18.2% of the crops. The highest residue detection rate was found in soil (75.8% of crops), and the most frequently detected compounds were dichlorodiphenyltrichloroethane (DDT) isomers. Pesticide residues in leaves were found in 36.4% of crops, and in fruit in 3% of crops (2.4% of the total number of fruit samples). A total of 40 pesticide residues were detected across all samples from organic crops, including 16 fungicides and one fungicide degradation product, nine herbicides, 13 insecticides, and one synthetic repellent. Among the detected compounds, eight were withdrawn from use during the study period, and eight were not approved for use in conventional apple, raspberry, and strawberry crops. Multivariate analysis showed that in 74.2% of the crops studied, the detections concerned so-called historical compounds, in 12.1% of cases, contamination from neighbouring conventional crops, and in 39.4% of cases, the intentional use of pesticides not permitted in the organic system. Multivariate analysis indicated that the cause was deliberate use not only in the years when the samples were collected, but also in the period preceding those years. The proposed system should be regarded as a rule-based, multifactorial inference framework that integrates standard analytical methods with structured contextual information, rather than as a purely statistical multivariate model. Its conceptual structure is designed to support causal interpretation of residue findings at the crop level and could be adapted to different regulatory and geographic contexts.
Danelski et al. (Sun,) studied this question.