Comparison of methods used to predict the in vivo digestibility of feeds in ruminants

Digestibility is a very useful index of the energy content of ruminant feed s, but cheaper and quicker laboratory methods are required as an alternativ e to the ultimate measure of in vivo digestibility using animals. These met hods involve either prediction of digestibility from chemical composition o r in vitro and in situ simulation of the digestion process. This review pre sents a range of chemical and in vitro techniques for predicting digestibil ity, together with an assessment of their advantages and limitations, parti cularly the degree to which they account for the sources of variation in in vivo digestibility in ruminants. In situ digestion of feed samples in the actual rumen environment is probably the most accurate of the non in vivo p rocedures, but is not suitable for routine application. The in vitro gas pr oduction technique offers no advantages in prediction of total tract digest ibility, but is useful for screening cereal grains for rate of starch hydro lysis in the rumen. The preferred procedure for routine laboratory predicti on of digestibility is the pepsin-cellulase technique, provided amylase is included or high temperature digestion is used for samples high in starch c ontent. Prediction of digestibility from chemical composition is not recomm ended. The optical technique of near infrared reflectance spectroscopy can be calibrated against any of the methods outlined in this review, and is un surpassed in terms of speed and repeatability. Direct NIR prediction of in vivo digestibility is also possible, but is limited by the lack of adequate numbers of feed samples with known in vivo values. Future work should be a imed at filling this gap and also improving the accuracy of laboratory meth ods for predicting the digestibility of low quality feeds.