Partial least square models for non-destructive prediction of 'Marsh' grapefruit (Citrus × paradisi MacFad) susceptibility to postharvest rind pitting using Vis/NIR spectroscopy

Postharvest rind pitting (RP) is a progressive physiological disorder developing in 3 to 5 weeks during storage, reducing the external quality and causing major postharvest losses on fresh citrus fruit. Its progressive nature makes it difficult to detect during sorting and packing. In this study, the ability of using visible to near infrared radiation spectroscopy (Vis/NIRS) for non-destructive prediction of fresh 'Marsh' grapefruit susceptibility to RP was investigated. Vis/NIRS spectral data was acquired in reflectance mode using a laboratory bench-top monochromator NIR Systems Model XDS equipped with a quartz halogen lamp and lead sulfide detector. Vis/NIRS spectral data was correlated to rind biochemical parameters related to rind quality and RP. The data was also correlated to visual scores of RP that were scored after six weeks of cold storage simulating commercial chain and one week at room temperature. Partial least square regression models were developed using Unscrambler^® chemometric software. Good prediction of RP was obtained with the regression coefficient R²=0.78, root mean square error of prediction RMSEP=1.41 and residual predictive deviation RPD=2.01. Rind biochemical parameters were also predicted accurately with all R² values ranging from 0.88-0.99. Additionally, the susceptibility of fruit from outside was higher than fruit from inside the canopy. Principal component analysis successfully assigned fruit spectra based on the fruit origin within the tree canopy and susceptibility to RP. The application of Vis/NIRS and associated chemometric data analysis was recommended as a tool for discriminating fruit with high chances of developing RP during sorting and packing.