Determination of quality parameters by near infrared reflectance spectroscopy in whole-plant corn silage

The prediction of quality parameters by near infrared reflectance analysis (NIRA) in whole-plant corn (Zea mays L.) was studied. Quality parameters included percent protein and in vitro dry matter digestibility (IVDMD). Calibrations were developed using two types of NIRA instruments: (a) a Neotec 51A, a six-tilting-filters type instrument and (b) a Technicon InfraAlyzer 400R, a 19-fixed-filters type instrument. Plant samples were collected from four locations across Ontario (Brucefield, London, Guelph and Elora). Forty samples were used in the calibration sets (C) for each quality parameter and for each instrument. A second group of samples, a prediction set (P) consisting of 200 samples, was used to validate the calibration equations. Regression analysis between NIRA predicted and IVDMD, indicated that this parameter was well predicted in both C and P sets and with the two instruments. Coefficients of determination (r 2 ) for C and P sets were 0.91 and 0.85 for the InfraAlyzer 400R and 0.92, and 0.81 for the Neotec 51A, respectively. A standard error of the estimate (SEE) of 1.70 was observed for the prediction of IVDMD in both C and P sets with the InfraAlyzer 400R. Values of SEE for IVDMD using the Neotec 51A were 1.76 and 1.73 for C and P sets, respectively. NIRA predictions of percent protein showed differences between instruments. The r 2 for C and P sets were 0.95 and 0.81 for the InfraAlyzer 400R and 0.90 and 0.58 for the Neotec 51A, respectively. The low r 2 value for percent protein in the P set might be related to the mathematical treatment of the reflectance data chosen for the analysis. The SEE for the prediction of percent protein varied between 0.25 and 0.56.Key words: Corn (whole-plant), quality, infrared reflectance