Non-destructive determination of soluble solids in apple fruit by near infrared spectroscopy (NIRS)

The viability of a near infrared (NIR) method based on a dual-beam, fibre-optic portable spectrophotometer was tested to determine soluble solids non-destructively in apples (Malus domestica Borkh,). A total of 340 apples (190 of cv. 'Golden Delicious' and 150 of cv. 'Jonagold') sampled from two positions in the canopy to increase variation in soluble solids concentration, were measured after 5-months storage and the measurements randomly assigned to a calibration data set and a prediction data set. Thus the calibration set and the prediction set represented exactly the same distribution. The calibration data set was used to select the wavelengths best correlated with degrees Brix and to fit a multiple linear regression (MLR) equation that was applied to calculate the degrees Brix value in the prediction data set. The most significant R-2 (0.56) was found with the first derivative of log(1/R) (where R = reflectance), yielding standard error of calibration (S.E.C.) = 1.01 degrees Brix, standard error of prediction (S.E.P.)= 1.14 degrees Brix and bias = - 0.13 degrees Brix. When MLR was carried out separately on each cultivar, the R-2 value was higher for 'Golden Delicious' (0.65), but not for 'Jonagold' fruit. Analysis of variance performed on the actual and the predicted degrees Brix showed no difference in statistical significance, although the bias was higher (+/- 5% of actual degrees Brix) at the extremities of the degrees Brix range. The scoring of the fruits according to soluble solids content measured by NIR enabled correct classification of 72 and 76% of the fruits, with a threshold of 12 and 13 degrees Brix, respectively. This NIR method seems reliable for determining soluble solids contents non-destructively, and could prove useful in the orchard. (C) 1998 Elsevier Science B.V. All rights reserved.