PREDICCION DE LA CONTRIBUCION ESPECIFICA EN MEZCLAS ELABORADAS CON MUESTRAS PURAS DE TREBOL BLANCO (Trifolium repens) y BALLICA INGLESA (Lolium perenne), MEDIANTE ESPECTROSCOPIA DE REFLECTANCIA CERCANA AL INFRARROJO (NIRS)

Abstract

A study was conducted in order to evaluate the potential of near infrared reflectance spectroscopy (NIRS) for predicting the percentage of white clover and perennial ryegrass in mixtures made up from pure samples of each species. Samples were obtained in five consecutive weekly cuts and 55 mixtures with 0 to 100% of each species (fresh basis) were prepared. NIR spectra were taken in the fresh and dry ground mixtures and calibration models were developed for predicting percentage of each species, as contribution in dry weight (reference data), applying different treatments to the spectral data. Equations were tested by cross-validation. Spectra from pure clover and ryegrass samples showed differences in specific absorption bands. Calibrations with good predictive potential could be developed, according to the cross-validation process, with coefficients of determination (1-VR) above 0.9 and standard errors of ca.6 percentage units for fresh and 4 percentage units for dry samples of either species. It is concluded that NIR can be used as a fast methodology to predict clover and ryegrass composition in mixtures made up from pure samples, without the drawbacks of the tedious and time-consuming traditional hand separation method.