Untargeted Analysis of Volatile Compounds in Wine as Predictors of Fermentation Dynamics

<ns3:p>Wine fermentation is a complex biochemical process that has garnered significant interest, particularly in wine production and quality control. Monitoring changes in the volatile profile during fermentation provides critical insights into the biochemical transformations that shape the final product. Volatile compounds, characterised by their high vapour pressure at room temperature, are key contributors to the aroma and flavour of wine. However, their analysis presents challenges due to the intricate composition of wine, which contains a vast array of compounds with diverse chemical properties. Factors such as varying polarity, solubility, volatility, pH sensitivity, low concentrations, and the instability of certain aromatic compounds further complicate the assessment of volatile constituents.This research investigated four wine varieties, two red and two white, by sampling them daily throughout the fermentation process. Volatile compounds were extracted using liquid-liquid extraction with three successive dichloromethane extractions. The extracts were then analysed using gas chromatography-mass spectrometry (GC-MS) in splitless mode, with helium as the carrier gas at a flow rate of 1 mL/min. A DB-wax column (60 m × 0.25 mm × 0.25 m) was used as the stationary phase, and compound identification was performed using NIST library. The total ion chromatograms were compiled into a matrix format for chemometric analysis.Principal component analysis (PCA) was applied to evaluate variance within the dataset, while partial least squares (PLS) regression identified key volatile compounds that exhibited significant changes during fermentation. Based on PLS regression loadings, peaks were annotated to determine the compounds present in the dichloromethane extracts. Across all four wine varieties, three key compounds—phenylethyl alcohol, undecanoic acid, and 2,5-dimethylbenzaldehyde—were consistently detected.To enhance the reliability of the analysis, data preprocessing was performed using five spectral transformation methods. Among these, baseline standard normal variate (SNV) preprocessing demonstrated the best performance, yielding the lowest root mean square error of prediction (RMSE) at 2.6 days. This result indicates that GC-MS-based analysis of volatile constituents can serve as a predictive tool for assessing the progression of wine fermentation.</ns3:p>