Not essential elements as tracers of geographical provenience of Sorrento PGI lemon juices

The mineral composition of agri-food products is useful to define their provenance for fraud protection. The potential of mineral composition to define the geographical provenance of high-value PGI agri-food products was explored in order to protect them from fraud. The Sorrento lemon (Citrus limon (L.) Burm. f. cv. Ovale di Sorrento), is known for its characteristic cultivation on terraces in the Sorrento peninsula and Capri island of Campania region (South Italy). In this environment, the peculiar soil and climatic features and the traditional cultivation on terraces have contributed not only to high-quality lemon productions but also to protect the landscape. The geographical conformation of the territory leads to different microclimates and habitats even at a very small scale. In this work, the multielement fingerprinting (essential and not essential elements) is proposed for discrimination lemon juices of six different cultivars (Femminello Ovale di Sorrento, Femminello Zagara Bianca, Femminello Siracusano 2KR, Femminello Sfusato Amalfitano, Femminello Adamo, and Femminello Cerza), grown in the PGI area of Sorrento lemon and in other two Campania region areas (no-PGI), according to the cultivars and their geographical origin on regional territory scale in two years (2018 and 2019). The explorative analysis by PCA on the mineral profile of the lemon juices showed natural grouping according to provenance at the expense of different cultivars. This suggests that the juice mineral composition depends slightly on cultivars, but strongly on the features of the cultivation environments. The applied discriminant model S-LDA, according to territorial provenance of lemon juices, showed 97.73% correct classification, 98.48% accuracy, and 93.83% external validation, and Mo, Ba, Rb, Mg, Co, Ca, Fe and Sr as discriminant elements. However, the annual variation of discriminant elements regarding many nutrients, the correlation of lemon juices/soil of some not essential elements (Ba, Rb, and Sr) which also discriminate juices and soils according to areas in both years, suggested the use of not essential elements as stable indicators of lemon juice provenance. In support of this suggestion, we applied S-QDA, more stringent than S-LDA, on only the determined, not essential elements (Ti, Co, Rb, Ba, and Sr). The results were discrimination of lemon juices according to provenance by all not essential elements, with 87.50% correct classification and 83.95% validation, despite the low number of variables. An increasing number of not essential elements is expected to improve the discrimination models.