Dry matter determination in 'Hass' avocado by NIR spectroscopy

'Hass' avocado harvested at four different times during a growing season, was analysed by both reflectance and interactance NIR spectroscopy to establish its reliability for the non-destructive determination of fruit dry matter (DM). Mean DM increased from 27.7 to 36.8% during the course of the study. Relationships between spectral wavelengths and DM were evaluated by application of chemometric techniques to partial least squares (PLS) and multiple linear regression (MLR) models, using calibration and validation statistics of predictions to compare their efficacy. The interactance mode was a superior predictor of fruit DM compared with reflectance. A PLS model applied 2 to interactance data from fruit from all harvests combined (n = 239) gave a goodness-of-fit (RP) between predicted and actual DM measurements of 0.88, and an error of prediction (RMSEP) of 1.8% DM. MLR models incorporating four wavelengths returned equivalent validation statistics. Three of these wavelengths, in the vicinity of 900-920 nm, are consistent with the C-H absorbance band, leaving only a minor role for water-related absorbances. These observations suggest on-line commercial NIR systems capable of operating in an interactance mode have the potential to usefully grade avocado on the basis of their DM content, to improve taste and oil content. (C) 2003 Elsevier B.V. All rights reserved.