Use of near infrared reflectance spectroscopy (NIRS) for the prediction of the chemical composition and nutritional attributes of green crop cereals

Abstract

The NIR spectra of 135 samples of the following green crop oats ( Avena sativa), barley ( Hordeum vulgare), triticale ( x Triticosecale), wheat ( Triticum durum), ryegrass ( Lolium multiflorum) and sorghum ( x Sorghum sudanense) grown in Portugal, were obtained by reflectance from oven-dried, milled whole plant material. Samples were cut at various stages of development (viz. stem elongation, boot, early heading, milk and dough stage) and were analysed by standard laboratory methods. The nutritive value was evaluated in vivo using rams in metabolic cages. The green crops showed values of ASH varying from 5.5 to 14.9%, crude protein (CP) from 5.0 to 22.3%, crude fibre (CF) from 25.3 to 40.4%, neutral detergent fibre (NDF) from 45.9 to 77.2%, acid detergent fibre (ADF) from 23.6 to 49.9%, acid detergent lignin (ADL) from 1.7 to 8.0%, dry matter digestibility (DMD) from 53.2 to 82.9%, organic matter digestibility (OMD) from 56.2 to 85.0% and dry matter intake (IVDMI) from 35.8 to 88.6 g DM kg −1 W 0.75. Spectra were used to develop NIR calibration models for ASH (SEC=0.72; r 2=0.90), CP (SEC=0.63; r 2=0.98), CF (SEC=1.02; r 2=0.94), NDF (SEC=1.41; r 2=0.97), ADF (SEC=1.43; r 2=0.96), ADL (SEC=0.49; r 2=0.87) and in vivo attributes DMD (SEC=2.61; r 2=0.86), OMD (SEC=2.36; r 2=0.88) and IVDMI (SEC=6.05; r 2=0.41). Calibration models were successful for all parameters except for IVDMI. However, animal attributes were less successfully predicted than chemical composition. Calibration models were generated omitting one of the six plant species to assess their ability to accommodate unknown green crop species not represented in the calibration model. In general, NIRS was shown to provide acceptable prediction of composition across different species of green crop cereals whether or not these were included in the calibration process.