Crude protein prediction of heterogeneous mountain grassland with visible-near-infrared spectroscopy

Abstract

Species-rich Nardus grasslands are some of the most widespread habitats in the European mountainous areas and its Crude Protein (CP) content is one of the nutritional properties studied for sustainable grazing management. Near-Infrared Reflectance Spectroscopy (NIRS) and Visible-Near-Infrared Reflectance Spectroscopy (VNIRS) were reported to be alternative technologies to measure CP content in grass and forage. Despite being a fast, cheap and environmental-friendly method, issues related with sample presentation need to be considered, foremost in field conditions. The objective of this study was to evaluate the potential of VNIRS to estimate the CP content in heterogeneous species-rich Nardus grasslands for non-processed samples in field conditions (on-site) and processed samples scanned in laboratory conditions. The hypothesis was that accurate calibration models could be developed to predict CP content in heterogeneous grasslands for both sample presentation conditions. In this study, 103 samples of heterogeneous grass were scanned and collected for chemical analysis. Samples points were selected following a completely randomized experimental design considering spatial and temporal variability. Successful results (Rc2³0.90) were obtained for CP calibrations for both on-site and laboratory conditions. However, better predictive ability and better ratio of prediction deviation (RPD) were obtained for processed samples in laboratory conditions (Rp2=0.93 and RPD=4.01) compared to non-processed samples in field conditions (Rp2=0.81 and RPD=2.43). According to the results, it was concluded that accurate calibration model for CP estimation in heterogeneous species-rich Nardus grasslands was possible for processed samples in laboratory conditions. Conversely, calibration equation for non-processed samples in field conditions could be useful when quick response is needed in decision-making process. Thus, VNIRS can be used as an alternative technique to estimate CP content in heterogeneous grasslands for both field and laboratory conditions and it could be a useful tool in decision-making for sustainable grazing management