Nutritional evaluation of commercial dry dog foods by near infrared reflectance spectroscopy

• Published in: Journal of animal physiology and animal nutrition 2006-06, Vol.90 (5-6), p.223-229
• Main Authors: Alomar, D., Hodgkinson, S., Abarzúa, D., Fuchslocher, R., Alvarado, C., Rosales, E.
• Format: Article
• Language: English
• Subjects: Amino Acids - analysis; Animal Feed - analysis; Animal Nutritional Physiological Phenomena; Animals; Calibration; Dietary Fats - analysis; Dietary Fiber - analysis; Dietary Proteins - analysis; Dogs; Minerals - analysis; Nutritive Value; Predictive Value of Tests; Spectroscopy, Near-Infrared - methods; Spectroscopy, Near-Infrared - standards; Spectroscopy, Near-Infrared - veterinary
• ISSN: 0931-2439; 1439-0396
• DOI: 10.1111/j.1439-0396.2005.00585.x

Summary Near infrared reflectance spectroscopy (NIRS) was used to predict the nutritional value of dog foods sold in Chile. Fifty‐nine dry foods for adult and growing dogs were collected, ground and scanned across the visible/NIR range and subsequently analysed for dry matter (DM), crude protein (CP), crude fibre (CF), total fat, linoleic acid, gross energy (GE), estimated metabolizable energy (ME) and several amino acids and minerals. Calibration equations were developed by modified partial least squares regression, and tested by cross‐validation. Standard error of cross validation (SECV) and coefficient of determination of cross validation () were used to select best equations. Equations with good predicting accuracy were obtained for DM, CF, CP, GE and fat. Corresponding values for and SECV were 0.96 and 1.7 g/kg, 0.91 and 3.1 g/kg, 0.99 and 5.0 g/kg, 0.93 and 0.26 MJ/kg, 0.89 and 12.4 g/kg. Several amino acids were also well predicted, such as arginine, leucine, isoleucine, phenylalanine–tyrosine (combined), threonine and valine, with values for and SECV (g/kg) of 0.89 and 0.9, 0.94 and 1.3, 0.91 and 0.5, 0.95 and 0.9, 0.91 and 0.5, 0.93 and 0.5. Intermediate values, appropriate for ranking purposes, were obtained for ME, histidine, lysine and methionine–cysteine. Tryptophan, minerals or linoleic acid were not acceptably predicted, irrespective of the mathematical treatment applied. It is concluded that NIR can be successfully used to predict important nutritional characteristics of commercial dog foods.