Identifying the vintage of French wine using stable isotopes, elemental fingerprints, and a data-driven but explainable approach

Food Chem. 2025 Feb 1;464(Pt 3):141907. doi: 10.1016/j.foodchem.2024.141907. Epub 2024 Nov 4.

Abstract

Stable isotopes and elemental fingerprints were employed as indicators to evaluate the vintage of French wine using climate factors and data-driven models. δ13C of wine ethanol and glycine, and δ18O of wine water and 16 elements were determined in wine from Bordeaux, Burgundy, and Languedoc-Roussillon. Results revealed that isotopic and elemental signatures from various vintages were influenced by precipitation and temperature. If there was less precipitation and higher temperatures during the grape ripening phase, isotopic and elemental signatures had a positive impact on the grapes, resulting in superior quality French wine. Data-driven models achieved excellent accuracy to identify vintages with the identification accuracy up to 72.0 %, and even higher accuracy (up to 95.0 %) from the Bordeaux region. Explainable methods were employed to select the top-5 and top-10 variables for each region under different data-driven models, yielding results comparable to those from a full set of variables.

Keywords: Authentication; Data-driven models; Elemental analysis; Explainable analysis; Stable isotopes, wine vintage identification.

MeSH terms

  • Carbon Isotopes* / analysis
  • Ethanol / analysis
  • Ethanol / chemistry
  • France
  • Fruit / chemistry
  • Fruit / growth & development
  • Glycine / analysis
  • Glycine / chemistry
  • Oxygen Isotopes / analysis
  • Vitis* / chemistry
  • Vitis* / growth & development
  • Wine* / analysis

Substances

  • Carbon Isotopes
  • Oxygen Isotopes
  • Ethanol
  • Glycine