Geographical identification of strawberries based on stable isotope ratio and multi-elemental analysis coupled with multivariate statistical analysis: A Slovenian case study. (1st July 2022)

• Record Type: Journal Article
• Authors: Strojnik, Lidija; Potočnik, Doris; Jagodic Hudobivnik, Marta; Mazej, Darja; Japelj, Boštjan; Škrk, Nadja; Marolt, Suzana; Heath, David; Ogrinc, Nives
• Is Part Of: Food chemistry. Volume 381(2022)
• Publication Date: 2022-07-01
• Subjects: Strawberries -- Geographical origin -- Authenticity -- Stable isotope -- Element composition -- DD-SIMCA; Food -- Analysis -- Periodicals; Food -- Composition -- Periodicals; 664
• DOI: 10.1016/j.foodchem.2022.132204
• Languages: English
• ISSNs: 0308-8146

Abstract

Graphical abstract: Highlights: Distinguishing between production years and origin of strawberries is possible. Important variables for origin separation were P, Na, K, As, Ba, Sr, Mo, Cr and Zn. DD-SIMCA year-to-year model proved highly sensitive and specific. A reduction in variables did not significantly improve the model specificity. Based on DD-SIMCA model 39% of commercial samples were classified as non-Slovenian. Abstract: The geographical classification and authentication of strawberries were attempted using discriminant and class-modelling methods applied to stable isotopes of light elements and elemental composition. The work involved creating a database of 92 authentic Slovenian strawberry samples and 32 imported samples. All samples were harvested between 2018 and 2020. A good geographical classification of Slovenian and non-Slovenian strawberries was obtained despite different production years using discriminant approaches. However, for verifying compliance with a given specification (geographical indications), a class-modelling approach was used to build an unbiased verification model. Class models generated by data-driven soft independent modelling of class analogy (DD-SIMCA) had high sensitivity (96% to 97%) and good specificity (81% to 91%) on a yearly basis, while a more generalised model combining total yearly data gave a lower specificity (63%). Of the 33 commercially available samples (test samples) with declared Slovenian origin, 39% were from outside of Slovenia.