Fecal near-infrared reflectance spectroscopy calibrations for predicting diet quality and intake of donkeys

Date

2006-08-16

Journal Title

Journal ISSN

Volume Title

Publisher

Texas A&M University

Abstract

The objective of these studies was to develop near-infrared reflectance spectroscopy calibration equations from diet-fecal pair datasets to predict the diet quality and intake of donkeys. One hundred-forty diet-fecal pair samples were generated from two independent in vivo feeding trials conducted in the United States (N = 100) and Africa (N = 40). At each site, ten female donkeys were fed mixed diets blended from 25 forage and crop residues. The modified partial least square model (MPLS) was used to develop calibration equations for crude protein (CP), digestible organic matter (DOM), dry matter digestibility (DDM) and organic matter digestibility (OMD), for the US, Africa and US/Africa combined datasets, and dry matter (DM) and organic matter (OM) intake calibrations from the US datasets. Crude protein (CP) equations were developed with standard error of calibration (SEC) < 1.0 and coefficient of determination (R2) > 0.90, (SEL = 0.5). The US, US/Africa and Africa CP equations had SEC value of 0.77, 0.97 and 0.88 with corresponding R2 of 0.97, 0.95 and 0.88, respectively. Validation of the US CP equation resulted in a standard error of prediction (SEP) of 1.79 with corresponding coefficient of correlation (r2) of 0.82 and slope of 0.84 indicating high accuracy of prediction. In vivo derived DOM equations were also developed for the US, Africa and US/Africa datasets with SEC values of 2.58, 4.91 and 3.52, and R2 of 0.60, 0.81 and 0.84, respectively. In addition, the SEC and R2 values were 3.25 and 0.72 for US OMD, 3.28 and 0.79 for US DDM, and 4.2 and 0.85 for US/Africa OMD, and 4.3 and 0.87 for US/Africa DDM equation, respectively. Calibration equations for predicting DMI and OMI have resulted in SEC values of 3.45 and 3.21 (g/kgw0.75) and R2 values of 0.89 and 0.84, respectively. The present study explored the relationship between DMI and diet quality attributes. Crude protein and digestible organic matter to crude protein ration (DOM/CP) with r2 values of 0.60 and 0.39, respectively, have shown good correlations with intake. The present studies have confirmed the potential for the fecal NIRS profiling for predicting CP, DOM, DDM, OMD, DMI and OMI of donkeys. Both calibration and validation results have indicated that the present donkey equations were comparable to previously developed equations for ruminants; they have the capability for accurate prediction of diet quality and intake, and can be a useful tool for monitoring the nutritional well-being of donkeys with acceptable accuracy. Research works to further expand the present calibration equations with additional diet-fecal samples particularly from Africa that did not meet the required accuracy level is recommended.

Description

Citation