Page Banner

United States Department of Agriculture

Agricultural Research Service

Title: Towards Truly "global" Near Infrared Calibrations for Protein and Neutral Detergent Fiber in Dried Ground Forages

Authors
item Undersander, D. - UNIV. OF WISCONSIN-MADISO
item Berzaghi, P. - UNIV. OF PADOVA, ITALY
item Dardenne, P. - AGRICOLES, GEMBLOU, BELGI
item Flinn, P. - KELSPEC SERVICES PTY LTD
item Martin, Neal
item Paul, C. - FAL, BRAUNSCHWEIG, GERMAN
item Buchman, B. - FOSS, DENMARK
item Mazeris, F. - DELAVAL INTL., SWEDEN
item Lagerholm, M. - FOSS, SWEDEN
item Cowe, I. - FOSS, SWEDEN

Submitted to: Meeting Abstract
Publication Type: Abstract Only
Publication Acceptance Date: May 1, 2005
Publication Date: June 26, 2005
Citation: Undersander, D.J., Berzaghi, P., Dardenne, P., Flinn, P., Martin, N.P., Paul, C., Buchman, B., Mazeris, F., Lagerholm, M., Cowe, I.A. 2005. Towards truly "global" near infrared calibrations for protein and neutral detergent fiber in dried ground forages [abstract]. XX International Grassland Congress. p. 258.

Technical Abstract: Over the past five years, Foss and DeLaval have sponsored the activities of a group of forage analysts with the aim of developing "global" Near Infrared (NIR) calibrations for parameters that are important in ruminant nutrition. The approach adopted has been based on the amalgamation of historical databases from centres worldwide and calibrations for protein and neutral detergent fibre (NDF) in dried ground forages have been developed based on databases that currently comprise approximately 30,000 records. Protein and NDF, while not the most important parameters in ruminant nutrition, were chosen for the initial calibration development exercise because of the amount of data available and because the methodologies adopted by different laboratories worldwide were relatively uniform. Over the past two years, several trials have been carried out worldwide comparing the performance of "global" calibrations with the performance of locally developed calibrations for indigenous forages and based on reference values from local laboratories. The global models were based on artificial neural network (ANN) technology using a database that included spectra from all parts of the world and representing harvests for the last twenty years. Some tropical forages, and all the major types of temperate forages including silages, were included in the database. Standard methods were used to generate reference values for protein and NDF. The global calibrations would require slope and/or bias corrections before being used. This was because Scandinavian forages were not represented in the calibration database and local laboratories generated the reference values. The global models had poorer performance than the locally developed regional calibrations, but were able to handle a much wider range of samples with acceptable performance than could the regional calibrations. The benefits of having to maintain a single universal calibration rather than many species-specific calibrations are important to the economics of managing forage networks

Last Modified: 12/18/2014
Footer Content Back to Top of Page