Aplicación de la técnica NIRS para el análisis de materia seca y proteína en hojas de alfalfa

Abstract

In the present work we developed calibration equations by near-infrared reflectancespectroscopy in the visible/near-infrared (vis-NIR) region to estimate dry matter and proteincontents in leaves of alfalfa. A total of 26 cultivars were grown under irrigation in the provinceof Salamanca, and leaf samples were collected from three plant positions: basal, middle andapical. A total of 156 plant samples from two consecutive harvests (July and August, 2005),were considered. The calibration equations were obtained by means of the MPLS (ModifiedPartial Least Squares) regression, using several mathematical transformations to the spectraldata. The statistics obtained in the calibration and external validation processes allowed toconclude that the determination of protein was slightly more accurate than the determination ofdry matter. The second derivative of the spectra, with or without the previous application of theMSC (Multiplicative Scatter Correction) treatment was the mathematical transformation whichled to the lowest standard error of prediction for dry matter (SEP= 0.90%) and protein (SEP=1.13%). The mathematical models obtained show that the NIRS technique is satisfactory for thedetermination of both parameters.