Improper sample preparation negatively affects near infrared reflectance spectroscopy (NIRS) nutrient analysis of ground corn

Abstract

Near infrared reflectance spectroscopy (NIRS) is widely used in the feed industry for the determination of nutrient composition of feed ingredients and finished feeds. This technology is a useful tool that can provide instant nutrient values for nutritionists to make sure that feeds are manufactured to specification. However, to achieve proper nutrient estimations from NIRS, samples must be ground to a specific particle size for each evaluation. Previous studies have demonstrated that particle size (PS) of the sample may affect NIRS nutrient estimates of corn and other feedstuffs. Therefore, an experiment was conducted to evaluate the effect of corn PS on the accuracy of NIRS proximate nutrient estimations. Corn was ground using a 2-pair roller mill set at 8 different roller pair settings. Roller pair spacing ranged from 50.8 µm at setting 2 to 152.4 μm 152.4 µm at setting 6. Roller mill settings 3–2, 3–3, 4–3, 4–4, 5–4, 5–5, 6–5, 6–6 generated samples with mean PS of 693, 882, 968, 1750, 1877, 2862, 2911, and 3343 µm, respectively. A rotor grinder with a 500-mesh screen was used to grind corn to 267 µm, which served as a control for the study as it is the recommended method for feed sample preparation per NIRS equipment recommendations. An NIR spectrometer and Adisseo Precise Nutrition Evaluation software with a calibration for ground corn were used to determine dry matter, ash, crude fat, crude protein, and crude fiber from 3 replicate samples from each PS treatment. Data were analyzed as a 1-way ANOVA using the GLIMMIX procedure of SAS ver. 9.4 and least square means were separated using the PDIFF option at P ≤ 0.05. Samples ground with only the roller mill with PS larger than the control treatment were lower in dry matter, ash, and crude fat. Crude protein content of ground corn was also lower in samples with PS larger than the control except for samples ground to 3343 µm (6−6). In contrast, crude fiber was greater in samples with PS greater than the control treatment. Overall, the results of this study underscore the importance of proper sample preparation to obtain accurate nutrient content estimations from NIRS analysis.