Résumé
The varietal aroma of most wines from non-aromatic grapes is partly dependent on the qualitative and quantitative composition of glycosidic precursors in grapes. The only rapid method to assess these glycoconjugates (Red-Free Glycosyl-Glucose (G-G) method) allows only their total quantitation. We developed a new method using Fourier-transform infrared spectrometry (FT-IR) and chemometric techniques allowing these glycosidic precursors to be determined more finely. Vitis vinifera cv. Melon B. grapes grown in different areas of Muscadet vineyard (Northwest France) and harvested at different maturity stages were used to demonstrate the potentiality of this analysis method. Predictive partial least squares (PLS) regressions were established using 39 samples, representative of the glycoside variability. These models allowed the total levels of C13-norisoprenoidic and monoterpenic glycoconjugates, the most relevant aroma glycoconjugates for Muscadet wines, to be determined with a predictive error of 14 and 15%, respectively, and the total levels of the other classes of glycoconjugates with predictive errors ranging from 22% for volatile phenols to 36% for aliphatic and aromatic alcohols