Influence of growth type and season on the prediction of the metabolisable energy content of herbage by near-infrared reflectance spectroscopy

Abstract

Two experiments are reported where near-infrared reflectance spectroscopy (NIRS) has been used to predict the metabolisable energy (ME) content of the organic matter (OM) of four herbage populations. The herbages comprised 158 spring primary growths harvested between 1980 and 1984 (Population 1), 32 spring primary growths harvested in 1986–87 (Population 2), and 13 summer regrowths (Population 3) and 24 autumn regrowths (Population 4) both harvested in 1986–87. In Experiment 1, calibration was carried out using a sub-set of Population 1 (104 samples). The best four calibration relationships were then validated separately on Populations 2, 3 and 4. All four calibrations validated best on the spring herbages (Population 2), but all contained spectral terms with F ratios less than 10. In Experiment 2, calibration was carried out on sub-sets of samples (total 149) from all four populations. A calibration relationship ( R 2 = 76.9%, standard error of calibration 0.583) containing five spectral terms was selected as the best for validation. This relationship validated well on all the remaining samples ( R 2 = 70.5, standard error of prediction 0.695, slope 0.956, bias 0), although when tested on the autumn herbages (Population 4) alone, the validation was poor. It is believed that this was due in large part to the small variation in the ME content of the autumn herbages. The wavelengths at the spectral segment centres of the five term equation were 2280, 1658, 1718, 1378 and 2308 nm. The prediction errors for spring primary growths recorded in these experiments were shown to be similar to those obtained earlier using the pepsin-cellulase in vitro technique. It is concluded that NIRS can predict the ME content of herbage with similar accuracy to the best available in vitro procedure, but with much increased speed and reduced unit cost. There appear to be some advantages in developing spectral relationships from herbage containing a mixture of different growth types rather than spring primary growth alone.