NIR Determination of Major Constituents in Tropical Root and Tuber Crop Flours.
Tropical root and tuber crops (cassava, sweet potato, taro, and yam) are staples in developing countries where rapid urbanization is strengthening the demand for flour based foods. The objectives of this study were to calibrate Near-infrared spectroscopy (NIRS) for routine analysis of flours and to test its reliability to determine their major constituents.
Authors
Vincent Lebot*†, Antoine Champagne‡, Roger Malapa§ and Dan Shiley
† CIRAD, P.O. Box 946, Port-Vila, Vanuatu
‡ Universit de Saint Etienne, 42023 Saint-Etienne, France
§ VARTC, P.O. Box 231, Santo, Vanuatu
ASD Inc., 2555 55th Street, Boulder, Colorado 80301
J. Agric. Food Chem., 2009, 57 (22), pp 10539–10547
DOI: 10.1021/jf902675n
Publication Date (Web): October 30, 2009
Copyright © 2009 American Chemical Society
Abstract
Tropical root and tuber crops (cassava, sweet potato, taro, and yam) are staples in developing countries where rapid urbanization is strengthening the demand for flour based foods. Quality control techniques are still under development, and when available, laboratory analyses are too expensive. The objectives of this study were to calibrate Near-infrared spectroscopy (NIRS) for routine analysis of flours and to test its reliability to determine their major constituents. Flours prepared from 472 accessions (traditional varieties and breeding lines) were analyzed for their starch, total sugars, cellulose, total nitrogen, and ash (total minerals) contents. The near-infrared (350−2500 nm) spectra of all samples were measured. Calibration equations with cross and independent validation for all analytical characteristics were computed using the partial least squares method. Models were developed separately for each of the four crop species and by combining data from all spp. to predict values within each of them. The quality of prediction was evaluated on a test set of 94 accessions (20%) by standard error of prediction (SEP) and r2 parameters between the measured and the predicted values from cross-validation. Starch, sugar, and total nitrogen content could be predicted, respectively, with 87%, 86%, and 93% confidence, whereas ash (minerals) could be predicted with 71%, and cellulose was not predictable (r2 = 0.31). The statistical parameters obtained for starch, sugars, and total nitrogen are of special interest for flour quality control. These constituents are quantitatively the most important in the chemical composition of flours, and starch content is negatively correlated with sugars and total nitrogen. NIRS is a low cost technique well adapted to the conditions in developing countries and can be used for the high-throughput screening of a great number of samples. Possible applications are discussed.