Exploring the use of Near-infrared spectroscopy as a tool to predict quality attributes in prickly pear (Rosa roxburghii Tratt) with chemometrics variable strategy

Prickly pear (Rosa roxburghii Tratt) is a regional deciduous shrub fruit that grows in southwestern China. The total flavonoids and vitamin C rich in prickly pear are important criteria for judging the ripening status and quality. In this study, a nondestructive analytical method was developed to predict titratable acid (TA), soluble solids (SS), total flavonoids (TFs) and vitamin C (VC) in prickly pear using near-infrared spectroscopy (NIRS) coupled with chemometrics. Different spectral pretreatments and variables selections were involved to optimize the NIRS models. Through comparative investigation, the best performance was finally attained from DBiPLS-CARS with Rp of 0.9685, RPD of 4.388 for TFs, and Rp of 0.9461, RPD of 3.045 for VC, respectively. Prediction of TA and SS were relatively less satisfactory but acceptable. Overall results in this study demonstrated that fast quantitative analysis of the essential quality attributes can be achieved by NIRS modeling approach.