OPTIMIZING NEAR INFRARED REFLECTANCE SPECTROSCOPY TO PREDICT NUTRITIONAL QUALITY IN CHICKPEA HAULMS FOR LIVESTOCK FEED

The near infrared reflectance spectroscopy (NIRS) was used to develop calibration equations to predict chickpea haulm (Cicer arietinum) feed quality traits and mineral constituents. A total of 1348 cultivars of chickpea representing a nation-wide range of environments in Ethiopia and genotypic diversity (113 cultivars and 7 landraces) used in the framework of the Ethiopian National Chickpea Breeding and Genetics Program were scanned using a FOSS 5000 spectrophotometer. 130 samples representing the spectral characteristics of the chickpea haulms, selected using WinISI II software V.1.50, were chemically analyzed for the development of the calibration equations. A modified partial least-squares (MPLS) regression with cross validation was used to confirm the equations and identify possible spectral outliers (GH-value>3, where GH is the Mahalanobis distance). Values for coefficient of determination (R), standard error of prediction SEP(C) and ratio of performance deviation (RPD) were used for validation of the equations. Results showed ash (r =0.97; RPD=3.64), crude protein (r2= 0.99; RPD = 8.09), acid detergent fiber (r2 = 0.99; RPD = 6.43), neutral detergent fiber (r2=0.99; RPD = 6.65), lignin (r2 = 0.99; RPD =5), ME (r=0.99; RPD=24.3), IVOMD (r=0.99; RPD=26). These results show that the calibration equations can accurately predict nutritional quality traits of chickpea haulms. The use of the NIRS method can facilitate cost-effective and rapid decision making by researchers and farmers.