Canopy height regulates berry quality in wine grapes: A cultivar-specific insight from metabolic profiling and machine learning

A two-year field experiment was conducted in the eastern foothills of Helan Mountain, Ningxia, China, to investigate the effects of three canopy heights (T1: 100 cm, T2: 120 cm, T3: 140 cm) on berry quality parameters of Vitis vinifera L. cvs. ‘Chardonnay’ (CH) and ‘Merlot’ (ML). Metabolic profiling, chemometrics, and machine learning methods were integrated to reveal cultivar-specific responses. Increasing canopy height significantly promoted total soluble solids (TSS) accumulation and reduced titratable acidity (TA) in both cultivars, with ML showing higher sensitivity. Phenolic and volatile organic compounds (VOCs) exhibited distinct cultivar-dependent patterns. CH accumulated higher contents of total polyphenols and non-anthocyanin phenolics under T1 and T2, while ML was richer in polyphenols, anthocyanins, and non-anthocyanin phenolics under T2 and T3. T3 significantly enhanced methylated anthocyanins, whereas T2 favored acylated anthocyanins in ML. Total VOCs were the highest in CH under T1, but in ML under T2 and T3, with alcohols and esters being the most responsive classes. Based on SHAP analysis of the optimal support vector machine (SVM) for CH and random forest (RF) models for ML, trans -ferulic acid, 1-butanol, and ethyl caprylate served as key biomarkers for CH, while gentisuric acid, malvidin-3- O -(6- O -coumaroyl)-glucoside, and TSS represented those for ML. This study provides a theoretical basis for precise canopy management and targeted improvement of wine grape quality in this region.