Estimation of nutritional value of livestock feeds is essential for improving animal productivity. Assessing feed quality is best achieved by feeding animals under controlled conditions and measuring livestock gain or milk yield. However, livestock feeding trials are expensive and time consuming. The next best option is to use “wet chemistry”, in other words to use conventional laboratory analysis to measure key metrics such as digestibility and nitrogen concentrations. But this is also time consuming and requires infrastructure and trained lab personal. In this context Near-infrared Reflectance Spectroscopy (NIRS) is a useful tool to estimate feed quality cheaply and rapidly.  While a small conventional feed lab might analyse 50 feed samples a month, with NIRS it would be feasible to predict feed quality of about 3000 samples over the same period. NIRS is a promising technology in crop improvement programmes for quick screening of large numbers of samples in a short time for quality traits. It works on the principle of correlating optical spectral signatures with physico-chemical properties of the sample matrix. NIRS is used globally in feed evaluation studies and for rapid crop improvement/ breeding programs. 
 

ILRI uses a range of desktop FOSS NIRS instruments including the Forage Analyzer 5000, 6500 and XDS (the latest is the FOSS DS 2500) machines with the WinISI-II & IV software package. ILRI is also experimenting with use of mobile hand-held (Phazir/Brimrose) and pocket type (TellSpec/SCiO) NIRS machines for prediction of nutritional value of feeds on site. The nutritional parameters estimated are dry matter content (DM,%), ash content (%), crude protein (CP, %), neutral detergent fiber, (NDF, %), acid detergent fiber, (ADF, %), acid detergent lignin, (ADL, %), in vitro organic matter digestibility (IVOMD, %) and metabolizable energy (ME, MJ/kg dm). ILRI has also developed equations to estimate minerals, amino acids, fatty acids and antinutritional factors such as tannins, phenols and HCN. After scanning ground samples (1 mm mesh size) for their NIR spectra, calibration validation procedures are developed.  Generally, half of a given set of samples is used for developing a calibration model with the reference chemical data generated by laboratory analysis. NIRS calibration equations are validated by predicting laboratory analysis of the second half of actually measured values (blind predictions). Agreement between NIRS predicted and actually measured laboratory values is assessed by the coefficient of determination (R2) and standard error of validation. Outlier samples are re-analyzed.  

Globally, anyone having any FOSS NIRS machine with WinISI calibration software can make use of more than 70 global equations developed by ILRI by standardising their machine with ILRI’s Master NIRS at ILRI, Hyderabad, India.

ILRI NIRS Calibration Equations

ILRI has developed over 80 NIRS calibration equations for different feed types and different constituents. These equations are available for sharing on request and on condition that they are properly acknowledged in any resulting outputs. A list of calibration equations is provided here.