Browsing by Author "Cronje, Paul J."

Now showing 1 - 1 of 1
  • Thumbnail Image
    ItemOpen Access
    Prediction of ‘Nules Clementine’ mandarin susceptibility to rind breakdown disorder using Vis/NIR spectroscopy
    (Elsevier, 2012-07-15) Magwaza, Lembe Samukelo; Opara, Umezuruike Linus; Terry, Leon A.; Landahl, Sandra; Cronje, Paul J.; Nieuwoudt, Helene; Mouazen, Abdul Mounem; Saeys, Wouter; Nicolai, Bart M.
    The use of diffuse reflectance visible and near infrared (Vis/NIR) spectroscopy was explored as a non-destructive technique to predict ‘Nules Clementine’ mandarin fruit susceptibility to rind breakdown (RBD) disorder by detecting rind physico-chemical properties of 80 intact fruit harvested from different canopy positions. Vis/NIR spectra were obtained using a LabSpec® spectrophotometer. Reference physico-chemical data of the fruit were obtained after 8 weeks of storage at 8 °C using conventional methods and included RBD, hue angle, colour index, mass loss, rind dry matter, as well as carbohydrates (sucrose, glucose, fructose, total carbohydrates), and total phenolic acid concentrations. Principal component analysis (PCA) was applied to analyse spectral data to identify clusters in the PCA score plots and outliers. Partial least squares (PLS) regression was applied to spectral data after PCA to develop prediction models for each quality attribute. The spectra were subjected to a test set validation by dividing the data into calibration (n = 48) and test validation (n = 32) sets. An extra set of 40 fruit harvested from a different part of the orchard was used for external validation. PLS-discriminant analysis (PLS-DA) models were developed to sort fruit based on canopy position and RBD susceptibility. Fruit position within the canopy had a significant influence on rind biochemical properties. Outside fruit had higher rind carbohydrates, phenolic acids and dry matter content and lower RBD index than inside fruit. The data distribution in the PCA and PLS-DA models displayed four clusters that could easily be identified. These clusters allowed distinction between fruit from different preharvest treatments. NIR calibration and validation results demonstrated that colour index, dry matter, total carbohydrates and mass loss were predicted with significant accuracy, with residual predictive deviation (RPD) for prediction of 3.83, 3.58, 3.15 and 2.61, respectively. The good correlation between spectral information and carbohydrate content demonstrated the potential of Vis/NIR as a non-destructive tool to predict fruit susceptibility to RBD.