Identification of the geographical origin of Ecolly (Vitis vinifera L.) grapes and wines from different Chinese regions by ICP-MS coupled with chemometrics

[Display omitted] •Element profile of Ecolly grape and wine from six regions was analyzed by ICP-MS.•Ba, Cs, Li, Mg, Na, Rb, and Sr in wine were positively related to those in grape.•Element profile of Ecolly grape and wine was mainly driven by geographical origin.•ICP-MS coupled with ML models was...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:Journal of food composition and analysis 2022-01, Vol.105, p.104248, Article 104248
Hauptverfasser: Gao, Feifei, Hao, Xiaoyun, Zeng, Guihua, Guan, Lingxiao, Wu, Hao, Zhang, Liang, Wei, Ruteng, Wang, Hua, Li, Hua
Format: Artikel
Sprache:eng
Schlagworte:
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:[Display omitted] •Element profile of Ecolly grape and wine from six regions was analyzed by ICP-MS.•Ba, Cs, Li, Mg, Na, Rb, and Sr in wine were positively related to those in grape.•Element profile of Ecolly grape and wine was mainly driven by geographical origin.•ICP-MS coupled with ML models was a useful tool to trace grape and wine origins.•Feedforward neural network was the most superior classifier with 100 % accuracy. The authenticity of the geographical origin of grapes and wines is vital for ensuring food safety in the wine market. To authenticate the geographical origin of Ecolly grapes and wines from different Chinese regions, forty-six elements were determined by inductively coupled plasma mass spectrometry (ICP-MS). Moreover, the relationships of element content between grapes and wines and regional differences in element profile were analyzed by multivariate statistics. K, Na, Cs, V, Li, Sc, In, Mn, Mg, Al, and Li, Na, Mg, K, Ca, Mn, Fe, Ni, Cu, Zn were important characteristic variables to distinguish grape and wine origins based on chi-squared tests, respectively. A feed-forward neural network (FNN), random forest (RF), and support vector machine (SVM) exhibited 98.33 %, 96.67 %, and 100 % accuracy for distinguishing the geographical origin of Ecolly grapes using these variables, respectively. All models obtained 100 % accuracy for distinguishing wines. FNN exhibited superior performance compared with RF and SVM according to a comprehensive evaluation based on the overall accuracy, the receiver operating characteristic curve (ROC), an area under the curve (AUC), and tenfold cross-validation. Consequently, this study provides powerful evidence of the potential application of ICP-MS coupled with machine learning models to discriminate the geographical origin of Chinese grapes and wines.
ISSN:0889-1575
1096-0481
DOI:10.1016/j.jfca.2021.104248