Effects of oven- or freeze-drying on chemical composition and NIR spectra of pasture silage

In order to assess effects of drying method on chemical composition and NIR spectra of pasture silage, 20 samples of silages from permanent pastures o f different qualities were subsampled and dried either by forced draught ov en at 65 degrees C or freeze-dried. Samples were analysed for crude protein (CP), crude fibre (CF), neutral detergent fibre (NDF), acid detergent fibr e (ADF), gross energy (GE) and in vitro digestible organic matter in the dr y matter (DOMD) and composition expressed on a dry-matter (DM) basis, eithe r obtained as oven 105 degrees C DM (DM105 degrees) Or toluene DM (DMtol). Effects of the drying method on chemical composition were estimated by pair ed comparisons. Near-infrared reflectance (NIR) spectra were taken and the difference spectra between treatments for each sample were plotted in order to visually inspect effects of treatments, either as log 1/R or transforme d by their first derivative. Principal components explaining spectral varia bility were computed and samples graphically displayed according to the eig envalues of the first (1 and 2), or second (2 and 3) pairs of principal com ponents. Oven-drying resulted in a reduction in CP (p<0.02) content and an increase in CF (p<0.1), NDF (p<0.001) and ADF (p<0.06) content of silages, when expressed on a DM105 degrees basis. When expressed on a DMtol basis, a reduction in CP (p<0.02) and DOMD (p<0.06) content was observed with oven drying, but fibre fractions were not affected (p>0.1). Effects of the dryin g treatment were visually apparent on NIR spectra when plotted as log 1/R a gainst wavelengths, mainly as a baseline shift. Difference spectra (OD - FD ), both as log 1/R and its first derivative, showed consistent absorption b ands, indicative of different molecular responses of OD and FD silages to i ncident light. Plotting samples according to eigenvalues of the first two o r second two principal components suggests that drying methods affect the d istribution of samples. This is also indicative of an effect of drying trea tment on spectral features and reinforces the general advice that sample pr ocessing should be consistent in order to avoid adding errors to NIR analys is. (C) 1999 Elsevier Science B.V. All rights reserved.