1H NMR spectroscopy coupled with multivariate analysis was applied to investigate Italian cherry tomatoes metabolic profile. (6th January 2020)

• Record Type: Journal Article
• Authors: Masetti, Olimpia; Nisini, Luigi; Ciampa, Alessandra; Dell'Abate, Maria Teresa
• Is Part Of: Journal of chemometrics. Volume 34:Number 1(2020)
• Page Start: n/a
• Page End: n/a
• Publication Date: 2020-01-06
• Subjects: 1H NMR profiling -- cherry tomatoes geographical origin -- kNN algorithm -- PCA and PLS‐DA analysis -- tomato lipophilic metabolites; Chemistry -- Mathematics -- Periodicals; Chemistry -- Statistical methods -- Periodicals; 542.85
• DOI: 10.1002/cem.3191
• Languages: English
• ISSNs: 0886-9383

Abstract

Abstract: Nuclear magnetic resonance (NMR) spectroscopy, in combination with different chemometric methods, was widely used for metabolomic profiling in the geographical determination of food origin. In the present study, spectra data of cherry tomatoes, collected from Pachino (Sicily) and Sabaudia (Latium), were analyzed by principal component analysis (PCA), k nearest neighbors (kNN), and partial least‐squares discriminant analysis (PLS‐DA) in order to discriminate the samples according to their geographical provenance. The PCA analysis of 1 H NMR spectra of Sabaudia cherry tomatoes showed significant differences linked to the production year: phospholipids had higher levels in 2004, but less amounts of polyunsaturated acids and lycopene were observed with respect to the year 2005. Despite the annual differences in 1 H NMR metabolic profile of Sabaudia cherry tomatoes, using unsupervised (PCA) and supervised (PLS‐DA, kNN) approaches, the geographical origin differentiation was obtained. In fact, the kNN algorithm correctly classified approximately 84% to 87% of Pachino samples and 77% of Sabaudia ones with recognition ability varied from 82% to 84.4% and prediction ability (CV) of 76.2% and 94.7%. The PC1 component, with 53% of total variance, greatly separated Pachino cherry tomatoes from Sabaudia ones and PLS‐DA model showed a good degree of separation with recognition ability of 100% and prediction ability (CV) of 93% to 100%. PCA and PLS‐DA combined analysis highlighted the most prominent spectral areas that well separated the two groups of samples. So, phytosterols were found discriminating compounds according to PCA and PLS‐DA and differences in aroma components were observed mainly in PCA analysis. Abstract : NMR and multivariate data analysis was proposed to distinguish cherry tomatoes geographical origin. Discrimination between samples, grown in two different regions of Italy, has been achieved by applying principal component analysis (PCA), k Nearest Neighbors (kNN) and partial least‐squares discriminant analysis (PLS‐DA). All these methods distinguished the provenance of analyzed cherry tomatoes collected in the two studied years with a good performance. In addition, during this study, a very interesting effect related to annual climatic changes was recorded. So, in characterizing geographical origin of cherry tomatoes, the production year could be a very significant variable and it should be considered together with cultivar and seasonality in multivariate statistical analysis