Effect of freezing and grinding method on near-infrared reflectance (NIR) spectra variation and chemical composition of fresh silage

In order to evaluate possible effects of rapid freezing and grinding methods on near-infrared reflectance (NIR) spectra variability and chemical composition, samples from 18 pasture silages, previously pre-homogenized, were frozen either with liquid nitrogen (LN) or dry ice (DI) and, subsequently, ground with a Tecator® homogenizer (Tec) or a Moulinex® food processor (Mou). NIR spectra were taken (monochromator NIRSystems 6500) on four fresh subsamples per silage and root mean squares (RMS) computed on derivatized spectra (software NIRS 3, ISI, 1992) as a means of estimating spectra dissimilarities among subsamples. Principal components were computed and samples plotted according to the first three scores to visualize potential effects of treatments. Seven silages from the four treatments plus a control (pre-homogenized only) were freeze-dried and analyzed for toluene dry matter (DMTol), crude protein (CP) and acid detergent fiber (ADF). Average RMS value for silages was 423.97 and no effect of freezing or grinding method, nor any interaction between them, was found (P > 0.05), although DI-Mou and DI-Tec treatments showed lower (402.8) and higher (459.4) mean RMS values, respectively. Plotting of samples according to the main principal component scores of spectra did not show any effect of treatments, as samples tended to cluster by silage instead as by treatment. Chemical composition of treated, and control samples, was not different (P > 0.05). On average, composition was as follows: DMTol, 269.2 g kg−1; CP and ADF (oven DM basis), 126.0 and 403.2 g kg−1 DM, respectively. It is concluded that any sample handling method could be selected for calibration and subsequent routine analysis, but if a method is chosen, it is suggested to consistently adhere to that method.