Instead of predicting the ME value of silage from the MADF measurement using a regression equation (see Chapter 4, 'Fibre, lignin, cellulose,
© 2002 CAB International. Methods in Agricultural Chemical Analysis: a Practical 167
Handbook (N.T. Faithfull)
nitrogen-free extract and starch', ADAS is recommending the discontinuation of the above method in favour of one using NIRS methodology (Barber et al., 1990). The procedure involves three stages:
1. Prediction of organic matter digestibility (OMD) from NIRS data;
2. Conversion of OMD to DOMD (digestibility of organic matter in oven dried matter) using ash values, and then correction to true dry matter basis (allowing for volatiles) to give DOMDc;
This procedure, however, has the built-in problems of the difficulties in standardizing the NIRS methodology and then using two subsequent regressions. The advantage of NIRS over MADF has been summarized by Offer (1993). Basically, different analytical methods and regression equations have produced various ME values from the same silage sample. ADAS and other advisors, including the animal feed trade (UKASTA - United Kingdom Agricultural Supply Trade Association), met and agreed to the newly proposed method in an attempt to remedy the situation. The reason for preferring the new method is that extensive in vivo feeding trials of 200 silages showed that MADF gave poor correlation with OMD whether carried out by wet chemistry or predicted by NIR. OMD was best predicted by NIR using an eight-term linear regression. This was, however, very sensitive to interference from moisture and particle size effects. Further mathematical treatment to remedy this defect led to the recommended 76-term partial mean square model regression equation. The resulting ME values, however, were found to be 0.4 MJ kg-1 (or 1.2 MJ kg-1 CDM -corrected dry matter) greater than previous predictions using DOMDo X 0.16. The reduction of concentrate allowances to dairy cows by 20% as suggested by these figures was judged unacceptable and the feed allowance should therefore be adjusted to allow for this underestimate. The NIR-OMD-ME route, however, remains the most precise of the available methods.
Was this article helpful?