Abstract
Animal fats are among the most highly-prized animal by-products in feed manufacture, for both nutritional and commercial reasons. To meet market requirements and legislative demands regarding the use and marketing of animal fats in an efficient, rapid, innovative and low-cost manner requires a major advance in the analytical control of fats and oils. Near infrared (NIR) spectroscopy has proved to be an ideal candidate for this purpose. This paper reports on the development and evaluation of quantitative NIR chemometric models for predicting the percentage of different animal-origin fats in fat blends. Fat samples (88 for the calibration set and 20 for the validation set) were analysed in a Foss NIRSystems 6500 I spectrophotometer, using folded-transmission gold reflector cups with a pathlength of 0.1 mm. Calibrations were developed using modified partial least squares (MPLS) regression to predict percentages of vegetable oil and animal fat in a fat sample; for both parameters, values for the coefficient of determination (R2) were over 0.99. Excellent results were obtained for the external validation of prediction equations, with R2 v values of 0.99 and standard error of prediction (SEP) values of 1.3%, in both cases. The high degree of precision and accuracy afforded by the NIR equations, which enabled detection of amounts as low as 0.5% animal fat in soy oil, confirm that NIR technology could be used to ensure compliance with current legislation prohibiting the use of animal origin fats.
Get full access to this article
View all access options for this article.
