Potential of Laboratory VIS–NIR–SWIR Spectroscopy to Estimate Dry Matter, Crude Protein, and Neutral Detergent Fiber in Urochloa brizantha Tropical Pastures

Pastures are the main feed source for beef cattle production, a sector in which Brazil plays a prominent global role. This study aimed to develop predictive models for dry matter yield (DM yield, kg ha−1), crude protein (CP), and neutral detergent fiber (NDF) in Urochloa brizantha tropical pastures using laboratory VIS-NIR-SWIR spectroscopy, and to identify spectral patterns associated with these variables. Samples were collected from a commercial pasture area of approximately 200 ha, subdivided into 19 paddocks cultivated with Urochloa brizantha cv. Marandu and managed under rotational grazing during 2023. Forage samples were oven-dried, ground, and spectrally measured using a FieldSpec spectroradiometer (350–2500 nm). Partial least squares regression (PLSR) models were calibrated and evaluated using cross-validation, and informative wavelengths were identified using Variable Importance in Projection (VIP) scores. DM variability was mainly associated with near-infrared regions, CP with visible and near-infrared regions, and NDF with the visible region. Models calibrated with VIP-selected wavelengths achieved acceptable performance for CP (R2CV = 0.74) and NDF (R2CV = 0.72), whereas the general full-spectrum models showed moderate performance for CP (R2CV = 0.57) and acceptable performance for NDF (R2CV = 0.75). Temporal transferability varied among sampling periods, with greater robustness for CP and NDF than for DM. Overall, DM prediction remained limited and showed poor temporal transferability.