Prediction equations derived from in situ and in vitro analytical techniques to determine in vivo organic matter digestibility (OMD) are useful tools to estimate the quality of livestock feed. Most derived equations are aimed at groups of feedstuffs (forages or concentrates) or feeds separately. In this study of OMD, the prediction equations of the modified two-stage Tilley & Terry in vitro technique (MT) and pepsin-acid multi-enzymatic technique (PME) are compared, validated, and improved in relation to verified in vivo results using feedstuffs and complete diets. Initial comparison with in vivo data showed that the combined dataset and that of single feedstuffs achieved acceptable R2 values for both MT and PME (0.88 and 0.92, 0.87 and 0.89, respectively). The validation with the second dataset established that the initial equations were valid with R2 values of 0.96 for MT and 0.91 for PME on the combined feeds dataset. The establishment of a prediction equation using both datasets resulted in improved R2 values over the initial equation. With combined feeds using MT it was 0.94, compared with 0.88, and using PME, it was 0.91 compared with 0.87. No significant decrease occurred in the variation of OMD between the datasets, as explained by the model when omitting on separate slope and intercept, thus confirming the same population assumption. The data sets could be combined for a new prediction equation. The R2 values were 0.94 and 0.91 for MT and PME methods for combined feeds, respectively. The new improved in vivo prediction equation in each instance was thus valid and a true improvement on the initial prediction equations. The PME method can be used for predicting OMD as it negates the use of rumen liquor and confidently replaces MT OMD determinations.
"Experientia docet" - Experience is the best teacher