Prediction of the chemical composition of locally available brans for ruminants using near-infrared (NIR) spectroscopy in West Africa

Ruminant production systems in West Africa are constrained by seasonal feed shortages and variable nutritional quality of locally available feed resources. Cereal and soybean brans are important dietary supplements, but their evaluation is limited by the cost and time required for conventional laboratory analyses. This study assessed near-infrared spectroscopy (NIR) for rapid prediction of the chemical composition of cereal and soybean brans used in animal feeding. A total of 320 samples were collected from processing units and feed manufacturing facilities in seven municipalities of Benin. Samples were analyzed by standard laboratory methods and scanned using a Tellspec Enterprise handheld NIR spectrometer operating at 900–1700 nm. Partial least squares regression models were developed in WinISI, using 70% of samples for calibration and 30% for internal validation. The samples showed wide compositional variability, with dry matter ranging from 88.45 to 92.00%, crude protein from 60 to 120 g kg⁻¹, and metabolizable energy from 1350 to 2200 kcal kg⁻¹. Internal validation showed strong performance for DM, ME, CF, and CP. External validation with 50 field samples confirmed model robustness. Lower accuracy was observed for Ca and P. NIR spectroscopy offers a rapid tool for feed quality monitoring and ration formulation in West Africa.