Abstract: To elucidate the dynamic evolution of volatile flavor metabolites and their interconversion mechanisms with non-volatile matrix components during goji wine brewing, headspace solid-phase microextraction coupled with gas chromatography-mass spectrometry (HS-SPME-GC-MS) was integrated with multivariate statistical analysis. This study systematically characterized the metabolite profiles of samples from the pretreatment, fermentation, and aging stages and deciphered the mechanism of volatile-nonvolatile interconversion. A total of 82 volatile metabolites were identified during the brewing process, including 13 aldehydes, 6 ketones, 23 alcohols, 16 esters, 4 acids, 2 ethers, 8 hydrocarbons, 3 phenolics, and 7 miscellaneous compounds. Multivariate statistical analysis identified 19, 54, and 20 differential metabolites in the pretreatment, fermentation, and aging stages, respectively. Pearson correlation analysis revealed competitive inhibition between lipid oxidation and ester synthesis. Phenolics served a dual function: flavonoid glycosides were positively correlated with esters, whereas phenolic acids (e.g., ethyl protocatechuate) were negatively correlated with β-ionone. Non-volatile metabolites critically functioned as pivotal precursors and regulators in flavor formation. This study establishes key theoretical targets for the targeted flavor regulation of goji wine.

Key words: goji wine; headspace solid-phase microextraction-gas chromatography-mass spectrometry; volatile metabolites; differential metabolites; correlation analysis