NIR Spectroscopy for Predicting Physicochemical and Functional Quality Attributes of Berry (Aronia, Haskap, and Goji) Fruits.

This study evaluated the potential of a miniaturized portable near-infrared spectroscopy (NIRS) device for the non-destructive prediction of the physicochemical and functional quality attributes of red berries. A total of 145 samples from three berry species (aronia, haskap and goji), representing different harvest years and ripening stages, were analyzed. Spectra were acquired over the 908-1676 nm range using a MicroNIR™ 1700 OnSite-W spectrophotometer, and partial least squares regression models were developed to predict total soluble solids, moisture content, pH, total phenolic content and antioxidant capacity. The calibration models achieved coefficients of determination in cross-validation (R2CV) ranging from 0.83 to 0.92, with Root Mean Square Error of Cross-Validation (RMSECV) between 0.281 for pH and 2.085 g Trolox kg-1 FW for antioxidant capacity. External validation confirmed the robustness of the models, yielding R2EV values between 0.76 and 0.88 and Root Mean Square Error of Validation (RMSEV) ranging from 0.381 for pH to 2.095% for moisture content. The highest predictive performance was obtained for total soluble solids (R2EV = 0.88; RMSEV = 1.391), followed by moisture content, pH and antioxidant capacity, whereas total phenolic content showed the lowest predictive accuracy (R2EV = 0.76; RMSEV = 1.914 mg GAE g-1 FW). The Residual Prediction Deviation (RPD) and Range Error Ratio (RER) values further supported the practical applicability of the models for approximate quantitative prediction. Overall, these results demonstrate that portable NIRS is a rapid, non-destructive and reliable tool for the integrated assessment of the physicochemical and functional quality attributes of emerging red berry species.