Metabolomic analysis combined with machine learning algorithms enables the evaluation of postharvest pecan color stability.

Nut kernel color is a crucial quality indicator affecting the consumers first impression of the product. While growing evidence suggests that plant phenolics and their derivatives are linked to nut kernel color, the compounds (biomarkers) responsible for kernel color stability during storage remain elusive. Here, pathway-based metabolomics with machine learning algorithms were employed to identify key metabolites of postharvest pecan color stability. Metabolites in phenylpropanoid, flavonoid, and anthocyanin biosynthetic pathways were analyzed in the testa of nine pecan cultivars using liquid chromatography-mass spectrometry. With color measurements, different machine learning models were compared to find relevant biomarkers of pecan color phenotypes. Results revealed potential marker compounds that included flavonoid precursors and anthocyanidins as well as anthocyanins (e.g., peonidin, delphinidin-3-O-glucoside). Our findings provide a foundation for future research in the area, and will help select genes/proteins for the breeding of pecans with stable and desirable kernel color.