Animal feed formulation: Rapid and non-destructive measurement of components from waste by-products

Abstract

The major cost (60–80%) of animal production is attributed to feed but most growers are yet to accept and adopt alternative materials like by-products due to their vast variations in nutrient components. Feed and animal production methods are currently considered as unsustainable -with environmental issues related to by-products disposal. Rapid and non-destructive models for quantifying sugars, organic acids, amino acids and other nutrients in alternative materials and a model for precision animal feed production were developed. Consequently, we investigated the nutrient components of by-products using line-scan hyperspectral imaging (HSI) technique. Hyperspectral images of by-products were acquired in the spectral range of 1000–2500 nm. The spectral data were extracted and preprocessed to develop a prediction model using partial least square regression (PLSR) analysis. The PLSR models developed resulted in the following acceptable prediction accuracies (R2p); sugars (0.76–0.94), organic acids (0.72–0.75), amino acids (0.55–0.84), and other nutrients content (0.69–0.96). The root means square error of predictions (RMSEP) obtained were sugars (0.076–0.524 mg/mL), organic acids (0.360–0.626 mg/mL), amino acids (0.007–0.052 mg/mL), and other nutrients content (0.403–1.035 %). The results obtained from the PLSR models showed reliable performance for quantifying chemical components of different by-products. Further, the generated PLSR-based chemical-mapped images facilitated the visual assessment of the chemical concentration and distribution in by-products. Thus, based on the results, the application of HSI in combination with multivariate analysis method of PLSR in a commercial setting may be feasible. This can ultimately enable cost-saving in breeding by curtailing overfeeding and post-production losses and significantly mitigate environmental issues related to by-products disposal.