ISHS Acta Horticulturae 464: International Postharvest Science Conference Postharvest 96
NON-DESTRUCTIVE DISCRIMINATION OF POST-HARVEST FRUIT PROPERTIES USING VISIBLE-NEAR INFRARED SPECTROPHOTOMETRY |
| Authors: | A. D. Mowat, Philip R. Poole |
| Abstract:
New techniques are needed to measure and manage fruit variability. Visible-near infrared (VNIR) spectrophotometry is one of a number of non-destructive or rapid sampling technologies that could be used on horticultural products. This study investigates the ability of discriminant analysis techniques using VNIR spectra to differentiate populations of blueberry fruit with differences in harvest quality, and to describe the effects of shoot treatments that alter preharvest fruit development. Shoots of hybrid blueberry (Vaccinium darrowii x corymbosum) received one of four treatments: (i), shoot girdling; (ii), leaf plucking; (iii), leaf plucking and shoot girdling (P&G); and (iv), untreated control. Fully pigmented fruit were harvested 7 days after the application of shoot treatments. Diffuse body reflectance VNIR spectra (500–1000 nm) were collected from intact fruit. Discriminate and principal component analysis (PCA) techniques were used to determine treatment differences. PCA used diffuse reflectance spectra from all fruit to ascertain the major components of spectral variation. Spectra from control fruit were used to develop a discriminating algorithm. Spectra of the other treatment groups were then compared with the control group, and the Mahalanobis distance (measure of the closeness of individual fruit to the control population mean) was used to determine treatments that deviated significantly from the control group specifications. Berry coloration was similar between all treatments. The P&G treatment had significantly lower berry weight and soluble solids, and higher titratable acidity in comparison to berries from other treatments. Spectral characteristics also differed between the P&G treatment in comparison to other treatments. PCA and discriminant analysis confirmed that the VNIR spectra were significantly different between the P&G treatment and the other 3 treatments. The PCA score for the first principal component was the only principal component that significantly differed between treatments. The first principal component contained spectral features in both the visible and near infrared regions. VNIR spectra processed by PCA and discriminant analysis can distinguish some artificially-induced changes in the maturation of blueberry. The PCA analysis indicates that some of the treatment differences occur in spectral regions that are associated with both changes in flesh pigmentation and sugar/water concentrations. Although discriminant analysis of blueberry maturity is not dependent on measuring individual chemical components, the spectra reflect changes in pigmentation, soluble solids, water and refractive index. This approach has applications to the understanding of physiological processes and in principle could be used for pre-harvest crop prediction and postharvest assessment. | |
Download Adobe Acrobat Reader (free software to read PDF files) | |
