Evaluation of near infrared spectra for the prediction of post harvest quality in canning peaches

Abstract

Huge production losses occur because of deterioration of quality of clingstone peaches during cold storage, rendering the fruit unsuitable for canning. Near infrared (NIR) spectra recorded from intact clingstone peaches were evaluated to predict post harvest quality of these peaches. The NIR spectra were recorded, before storage, on the fresh peaches while the subjective post-storage quality, fruit firmness and diameter (reference data) were determined after either three- (2002 season) or two-week (2003 season) cold storage periods at -0.5ºC. Poor post-storage quality was indicated by softening of the flesh, loosening of the skin and adhesion of the flesh to the stone after destoning. Soft independent modeling by class analogy (SIMCA) gave correct total classifications of between 53 and 60% in comparison to the 57 to 65% obtained by multivariate adaptive regression splines (MARS). However, when predicting only poor post-storage quality, correct classfication rates of between 60 and 80% were obtained using MARS. Using classification trees, the fruit were classified into good and poor post-storage quality classes according to fruit firmness and diameter. Only reasonable results were obtained due to the poor relationship between the NIR spectra and firmness measurements.