Can Grassland Chemical Quality Be Quantified Using Transform Near-Infrared Spectroscopy?

SIMPLE SUMMARY: Near-infrared spectroscopy (NIRS) has been applied to analyse the quality of forage and animal feed. However, grasslands more than other raw materials are linked to many variability factors (e.g., site, year, occurring species, etc.) that can represent strong points as well as weak points in NIRS estimation. This research is aimed at testing NIRS application for the determination of chemical characteristics of fresh, undried and unground samples of meadows and grasslands located in north-central Apennine. The interest lies in the possibility of monitoring grassland resources, supporting the decision in terms of the need of supplementation and identifying the critical periods for cutting grassland intended for animal feeding. The results indicated that FT-NIRS models could be used in the real-time quantification of crude protein, fibrous fraction and dry matter, while for lignin only a screening test could be considered. Minor components of grassland such as ash and lipids need improvement. As a practical point, a key factor of FT-NIRS in grassland chemical quality estimation is the absence of samples preparation and the importance of the parameters that have obtained the best results in animal diet formulation. ABSTRACT: Near-infrared spectroscopy (NIRS) and closed spectroscopy methods have been applied to analyse the quality of forage and animal feed. However, grasslands are linked to variability factors (e.g., site, year, occurring species, etc.) which restrict the prediction capacity of the NIRS. The aim of this study is to test the Fourier transform NIRS application in order to determine the chemical characteristics of fresh, undried and unground samples of grassland located in north-central Apennine. The results indicated the success of FT-NIRS models for dry matter (DM), crude protein (CP), acid detergent fibre (ADF), neutral detergent fibre (NDF) and acid detergent lignin (ADL) on fresh grassland samples (R2 > 0.90, in validation). The model can be used to quantitatively determine CP and ADF (residual prediction deviation-RPD > 3 and range error ratio- RER > 10), followed by DM and NDF that maintain a RER > 10, and are sufficient for screening for the lignin fraction (RPD = 2.4 and RER = 8.8). On the contrary, models for both lipid and ash seem not to be usable at a practical level. The success of FT-NIRS quantification for the main chemical parameters is promising from the practical point of view considering both the absence of samples preparation and the importance of these parameters for diet formulation.