Seeing through the glass: through-bottle VIS-NIR screening for the real-time detection of light exposure in bottled rosé wines

Abstract

Ensuring the quality of rosé wine throughout the supply chain is a major challenge, as light exposure can rapidly induce the “light-strike” defect. While chemical markers can quantify the damage, their analysis is destructive and time-consuming. There is an urgent need for non-invasive diagnostic tools capable of verifying the storage history of sealed bottles, enabling quality control at every stage, from the winery to the retail shelf. This study evaluates in-bottle VIS-NIR spectroscopy (450-1050 nm) as a rapid, non-destructive tool for detecting light exposure. Thirty-six commercial rosé wines were monitored over 30 days of LED exposure (400-500 nm) or dark storage. To isolate storage-driven changes from glass-related interference, a T0-subtraction preprocessing was applied to baseline-corrected spectra. A Partial Least Squares Discriminant Analysis (PLS-DA) model was developed to classify wines based on their exposure status. Light exposure induced a distinct spectral response, characterized by decreased absorbance in the 450–500 nm region, consistent with riboflavin photodegradation. Despite the spectral interference of bottle glass and the compositional variability among 36 commercial wines, the T0-subtraction preprocessing effectively isolated light-induced changes. The PLS-DA model achieved 97.4% accuracy (training) and 98.0% (validation), with reliable classification already within 10 days of exposure. Our results demonstrate that through-bottle VIS-NIR spectroscopy is a reliable methodology for the real-time assessment of light exposure. This non-invasive screening tool could allow producers and distributors to implement proactive quality control along the entire supply chain, identifying “at-risk” lots before they reach the consumer. This approach paves the way for a new standard in shelf-life verification and brand protection for light-sensitive beverages.

References

Fracassetti D., Di Canito A., Bodon R., Messina N., Vigentini I., Foschino R., Tirelli A., Trends Food Sci. Technol., 2021, 112, 547–558.