The effect of freezing and thawing on grass silage quality predictions based on near infrared reflectance spectroscopy

This study was undertaken to determine if freezing and thawing of fresh grass silage influenced near infrared reflectance spectroscopy (NIRS) predictions. Approximately I kg each of 19 grass silages was used. Each sample was split into three sub-samples. One sub-sample was scanned fresh, wrapped in cling film (triplicate packages) and these were regarded as the reference samples (PF). These packages were retained frozen for I week at -20degreesC, thawed, scanned, re-frozen and re-scanned after 3 months frozen. At the same time one of the bulk samples was frozen for I week and the other bulk sample was frozen for 3 months and thawed and scanned as described for the packaged samples. NIRS predictions of the five treatments were compared for metabolizable energy intake, alcohol toluene corrected dry matter (ATDM), pH, nitrogen (N), digestible organic matter in the dry matter (D-value), neutral detergent fibre (NDF), acid detergent fibre (ADF), lactic acid and total acids. Although statistical significant differences were recorded for most treatments the prediction errors were acceptable within the NIRS cross-validation errors for potential intake, pH, D-value, lactic acid, total acids, NDF and ADF Spectral changes, due to treatments, were studied by plotting difference spectra after standard normal variate (SNV) and de-trend (DT) had been applied. The main areas of variation were found to be around the two water peaks and from 2000 to 2200 urn which is an area usually associated with protein. (C) 2002 Elsevier Science B.V. All rights reserved.