Quantitative analysis of chlorophyll and protein in alfalfa leaves using fibre-optic spectroscopy and PLS regression

Abstract

The performance ofnear-infrared (NIR) spectroscopy as a rapid technique for the estimation ofchlorophyIl and protein contents in alfalfa (Medicago saliva, L.) was investigated. Leafsamples from 33 cultivars wex:e taken from three plarlt positions (apical, middle arrd basai)iri two different harvests made in an experimental plot located in the province of Salamanca (Spain). A fibre-optic probe was employed directly on fresh leaves to measure near infrared spectra between 1100-2200 nm. Chemical analyses were performed using analytical reference methods; chlorophyIl was measured by spectrophotometry of dimethyl sulphoxide extracts and protein content was quantified using the Kjeldahl method. Partialleast squares (PLS) regression models were developed with a calibration set of 120 samples spanning a concentration range ofO.39-4.60 mg il (fresh weight) for chlorophyll and 9.92-45.32% (dry matter) for protein content. The models obtained were validated with independent test samples. The statistics ofvalidation provided the best accuracy using a first derivative transformation ofthe data in both parameters. Thus, we achieved standard errors ofprediction ofO.24 mg g-I and 3.27% for chlorophyll and protein contents, respectively. These results encourage the use ofNIRS equipped with a fibre-optic probe to monitor and assess the composition and quality of forages in a non-destructive way.