Rapid feed component assessment to enhance livestock productivity and reduce emissions

The rising global demand for livestock products necessitates innovations in feed management to enhance efficiency and sustainability. Variability in livestock feed nutrient composition highlights the need for real-time monitoring to ensure animals receive proper nutrition. In this study, we first used Analysis of Variance (ANOVA) to measure differences in moisture content (MC), crude protein (CP), and neutral detergent fiber (NDF) across eight feed types from seven factories in the Gyeongsang region, Republic of Korea. Then, we developed a non-destructive system to measure nutrient components using a Short-Wave Infrared (SWIR) Hyperspectral camera and Partial Least Squares Regression (PLSR). The PLSR models, each optimized with a specific preprocessing method, showed strong predictive accuracy: maximum normalization was used for MC (R²p = 0.84, RMSEP = 0.49%, 23 LVs); mean normalization for CP (R²p = 0.86, RMSEP = 1.81%, 7 LVs); and a Savitzky-Golay first derivative for NDF (R²p = 0.79, RMSEP = 2.26%). At the same time, ANOVA was used to confirm significant nutritional differences among the factories, emphasizing that relying on standard book values is inadequate. Overall, these results demonstrate the potential of Hyperspectral Imaging (HSI) as an effective tool for real-time feed quality assessment. This ability enables more precise feed management, which is crucial for maximizing livestock productivity and reducing environmental impacts, such as methane emissions resulting from inefficient feed use.