Soybean meal (SBM) is a widespread constituent of poultry diets internationally. However, variations in agronomic conditions, cultivar variety, and processing techniques can alter digestible lysine, methionine, threonine, and nitrogen-corrected true metabolize energy content (TMEN) of SBM. To detect these deviances in nutritional value, broiler or rooster bioassays with subsequent laboratory analyses can be completed to determine digestible amino acid and TME content. But, these bioassay determinations are expensive and time-consuming. In contrast, near-infrared reflectance spectroscopy (NIRS) has the potential to enable the rapid prediction of nutritional component values of feed ingredients upon delivery and prior to incorporation into poultry diets. The goal of the current research was to design NIRS calibration curves that accurately predicted the TMEN and digestible lysine, methionine and threonine content of solvent-extracted and mechanically processed SBM. The TME and digestible amino acid content for over 100 SBM samples was determined by the intact and cecectomized rooster bioassays, respectively. Each SBM sample was uniformly ground prior to obtaining a complete near-infrared spectral analysis using a Bruker MPA: FT-NIR Spectrometer equipped with OPUS software. Roughly half of the samples were used to construct each calibration curve, while the remaining half were used to validate each calibration curve. For the validation samples, the relationship between the predicted values obtained from the NIRS calibration curve versus the rooster bioassay determined values had correlation coefficients (R2) values ranging from 0.83 to 0.90 for digestible lysine, methionine, and threonine content and 0.98 for TMEN. However, for all three amino acids and TMEN, the validation samples had predicted values that deviated less than plus or minus 2.5% of their bioassayed determined value, except for two samples in the methionine validation and one sample in the threonine validation, which had NIRS predicted values that deviated more than 2.5%, but less than 5% of their bioassayed values.
This research indicates that NIRS can be used to accurately predict TMEN and digestible lysine, methionine, threonine content of the SBM samples for poultry diets.