Abstract
To upgrade grazing management recommendations in the Mediterranean area, one needs to assess the nutritive value of woody foliages including their changes over time and with location. Using a wide range of native and cultivated foliages, our objective was to evaluate the potential of near infrared (NIR) reflectance spectroscopy to determine foliage chemistry and in vitro digestibility for ruminants. The samples, representative of small ruminant eating bites, were divided into the different plant parts. Samples were carefully conditioned, being air-dried at 60°C; drying times were individually varied to ensure complete dehydration without excessive heating. Samples were analysed for organic matter (OM), nitrogen (N), neutral detergent fibre (NDF), acid detergent fibre (ADF), acid detergent lignin (ADL) and in vitro digestibility with a pepsin–cellulase method developed for forages (IVDMD). NIR scans were made with an NIRSystems 5000 instrument and data analysis was performed using ISI software. Partial least squares (PLS) regression equations were developed for IVDMD and each constituent. Nine mathematical treatments, with and without scatter correction, were compared. The database of woody foliage spectra and a reference fodder database were compared in two ways: the first involved a comparison of the spectral variation in each collection while the second measured the Mahalanobis distance of each spectrum in one database from the average spectrum in the other. In the case of N, a broad-based calibration was compared with others derived from various sample sub-sets; these latter were formed either according to sample type or following a proximity analysis of five spectral groups by principal component analysis. NIR predictions of IVDMD were applied to describe the effect of phenological changes in the edible parts of Quercus pubescens Willd. The foliage base is very heterogeneous and wider than the reference fodder base. The fodder base variation does not include the foliage samples. The lowest standard errors of calibration (SEC) and cross-validation (SECV) are comparable to literature results on forest leaves, but better for lignin (SECV of 1.5 compared to SEC values of 2.5–2.9). SECV for IVDMD is satisfactory at almost 2.0 because this value is similar to precision data normally reported for in vitro digestibility estimations on narrower sample sets than the current one. The broad-based calibration for N gave similar or lower standard errors to those obtained using sample sub-sets. One can show that IVDMD values for Quercus pubescens leaves decrease from 65 to 15% with increasing maturity; leaf IVDMD values are approximately 15% higher than the previous year's stem value from May until October. NIR spectroscopy appears to be an adequate technique for the prediction of the nutritive value of Mediterranean foliages from trees and shrubs, with a reliability similar to that obtained from classical fodder analysis procedures. This study shows that broad-based calibrations with PLS regression could be made on extremely diverse sets of data (IVDMD ranging from 28 to 94%), grouping distinct edible plant parts within the same data base.
© 1993 NIR Publications
PDF Article
More Like This
Cited By
You do not have subscription access to this journal. Cited by links are available to subscribers only. You may subscribe either as an Optica member, or as an authorized user of your institution.
Contact your librarian or system administrator
or
Login to access Optica Member Subscription