Prediction of chemical composition and peroxide value in unground pet foods by near‐infrared spectroscopy |
| |
| Authors: | M De Marchi F Righi M Meneghesso D Manfrin R Ricci |
| |
| Affiliation: | 1. Department of Agronomy, Food, Natural Resources, Animals and Environment, University of Padova, Legnaro (Padova), Italy;2. Department of Veterinary Science, University of Parma, Parma, Italy;3. FOSS Italia, Padova, Italy;4. Department of Animal Medicine, Productions and Health, University of Padova, Legnaro (Padova), Italy |
| |
| Abstract: | The massive development of the pet food industry in recent years has lead to the formulation of hundreds of canine and feline complete extruded foods with the objective of meeting both the needs of the animals and numerous demands from pet owners. In the meantime, highly variable raw material compositions and the industry's new production techniques oblige manufacturers to monitor all phases of the extrusion process closely in order to ensure the targeted composition and quality of the products. This study aimed at evaluating the potential of infrared technology (visible and near‐infrared spectrophotometer; 570–1842 nm) in predicting the chemical composition and peroxide value (PV) of unground commercial extruded dog foods. Six hundred and forty‐nine commercial extruded dog foods were collected. For each product, an unground aliquot was analysed by infrared instrument while a second aliquot was sent to a laboratory for proximate analysis and PV quantification. The wide range of extruded dog food typologies included in the study was responsible for the wide variability observed within each nutritional trait, especially crude fibre and ash. The mean value of the 208 pet foods sampled for PV quantification was 17.49 mEq O2/kg fat (min 2.2 and max 94.10 mEq O2/kg fat). The coefficients of determination in cross‐validation of NIRS prediction models were 0.77, 0.97, 0.83, 0.86, 0.78 and 0.94 for moisture, crude protein, crude fat, crude fibre, ash and nitrogen‐free extract (NFE) respectively. PV prediction was less precise, as demonstrated by the coefficient of determination in cross‐validation (0.66). The results demonstrated the potential of NIRS in predicting chemical composition in unground samples, with lower accuracy for moisture and ash, while PV prediction models suggest use for screening purposes only. |
| |
| Keywords: | dog food chemical composition near‐infrared reflectance spectroscopy peroxide value |
|
|
