Near infrared reflectance spectroscopy (NIRS) is a rapid method of analysis that enables a multi- purpose analyzer to be calibrated to predict the nutritional component values of feed ingredients by analyzing their reflectance properties in the near infrared spectrum. The goal of the current research was to create a NIRS calibration curve that accurately predicting the nitrogen corrected true metabolizable energy (TMEN) value for individual feed ingredients. Adult Single Comb White Leghorn roosters were used to determine the TMEN values of feed ingredients. Roosters (8 to 10 per feed ingredient tested) were fasted for 24 hours prior to being precision fed a fixed amount of each feed ingredient. Excreta was then collected for the following 42 hours from the fed roosters and from a group of unfed roosters, which served as controls to account for endogenous losses. Gross energy and total nitrogen content of the feed ingredients and dried collected excreta were determined to calculate TMEN values for each tested feed ingredient. A wide range of determined TMEN values (658 – 5,588 Kcal/kg), and sample variety, from ingredients commonly used in an industry setting like corn, meat and bone meal, and soybean meal, to experimental/alternative ingredients such as black soldier fly larva meal, Jatropha meal, and ground cassava leaves, were obtained. A full near infrared spectral analysis was completed on a ground sample of each feed ingredient using a Bruker MPA: FT-NIR Spectrometer equipped with OPUS Version 7.5 software. A calibration was made for TMEN using 49 feed ingredient samples which resulted in a prediction equation with a correlation coefficient (R2) of 0.97. This prediction equation was then validated with 316 different feed ingredient samples, and the R2 between the NIRS and rooster bioassay TMEN values was 0.92, but only 60% of the predicted values were within plus or minus 5% of their actual bioassay determined value. In contrast, subsequent calibration curves based on individual ingredient type, such as corn or bakery meal, resulted in validations in which 100 and 93% of samples, respectively, were within plus or minus 5% of their bioassay determined value.
The results indicate that NIRS has the capability of accurately predict the TMEN values of poultry feed ingredients, which would reduce the need for expensive and time consuming rooster bioassays for determining TMEN values on feed ingredients, and enable poultry nutritionists to formulate more precise diets based on TMEN values of the ingredients actually delivered to the feed mill.