NMR-based metabolomic identification of freshness markers reveals the working mechanism of visible and near-infrared spectroscopy to predict post-harvest broccoli freshness

Broccoli is highly perishable after harvesting. The purpose of this study was to investigate the possibility of estimating the freshness of recently harvested broccoli using visible (Vis) and near-infrared (NIR) spectroscopy. The Vis-NIR spectra of different parts of broccoli were collected as explanatory variables and the cumulative temperature of broccoli under different storage conditions was determined as an indicator of freshness. By combining PLSR analysis and the stepwise selectivity ratio method, an accurate predictive model for the freshness of broccoli florets was constructed using informative wavelengths. Based on strong correlations between the NIR and NMR signals of broccoli florets, seven key amino acids that increased during storage were identified. Among them, isoleucine, valine, asparagine, and phenylalanine were identified as common freshness-marker metabolites shared by different vegetables, such as broccoli and komatsuna. Protein degradation, detected as a change in NIR absorption, can be used to interpret the working mechanism of the calibration model constructed for freshness prediction. Thereby, the interpretable predictive model built based on the informative wavelengths is promising for reliable and accurate estimation of broccoli freshness in the supply chain.