Flavonoid metabolic variation and rapid prediction of dihydrochalcone contents in Lithocarpus litseifolius leaf color phenotypes

Leaf color phenotypes in medicinal plants are often accompanied by secondary metabolite profiles. To characterize changes in flavonoid metabolism associated with leaf color phenotypic variation in Lithocarpus litseifolius and develop a rapid, color-based model for predicting key secondary metabolites, spring leaves from yellow (YL) and red-brown (RBL) phenotypes were compared with those of green phenotype (GL). Differences in leaf color parameters, pigment composition, and metabolomic profiles were analyzed, and the resulting metabolite-color associations were used to construct predictive models for phlorizin and trilobatin contents.1) Significant differences in leaf color parameters and pigment contents were observed among the phenotypes ( P < 0.05). Differences in chlorophyll depletion and anthocyanin accumulation collectively explained the variation in leaf color phenotypes. 2) Metabolomic analysis showed that flavonols were significantly enriched in YL, accounting for a 23.66% higher proportion than in GL, whereas anthocyanins in RBL were 7.79-fold higher than in GL. 3) Partial correlation analysis showed significant positive correlations between phlorizin and anthocyanin contents, and negative correlations between phlorizin with flavonols, as well as negative correlations between trilobatin and specific flavanols, dihydroflavonols, and flavonols. 4) Based on these pigment–metabolite relationships, we constructed predictive models: the phlorizin model obtained a training R² of 0.834 and a 5-fold CV R² of 0.827 (Δ = 0.007), with negligible overfitting; the trilobatin model yielded a training R² of 0.637 and a CV R² of 0.559 ( Δ = 0.078), indicating acceptable predictive capacity. This study demonstrates that leaf color phenotypes can serve as reliable predictors for dihydrochalcone contents in L. litseifolius . The proposed approach provides a practical tool for high-throughput, efficient, and minimally invasive quality evaluation and germplasm selection of L. litseifolius , and can serve as a reference for phenotype-based prediction studies in woody plants.