Formulation of Hay-Grain Diets for Ruminants I. Evaluation of Multiple Regression Equations for Prediction of Bermudagrass Hay Quality from Laboratory Analyses2
Thirty-two bermudagrass [Cynodon dactylon (L.) Pers.] samples with known in vivo dry matter intake (DMI) and/or dry matter digestibility (DMD) were analzyed for crude protein and Van Soest fiber fractions. Digestible dry matter intake (DDMI), DMI and DMD were regressed on various preselected combinations of chemical composition data using least-squares multiple regression procedures. In addition, DMD was regressed on in vitro organic matter digestion. Of the resulting prediction equations, several were selected on the basis of theoretical cause and effect relationships and least-squares fit (R2 and sy · x). These equations were tested on four additional hays by comparing their predicted DDMI, DMI and DMD values with those achieved when the hays were fed to sheep. Predicted DMI was lower (P<.05) than achieved for three of the four hays. Predicted DMD was higher (P<.05) than achieved for one hay. However, predicted and achieved DDMI values ranked the hays in the same order. The simple regression equation generated by regressing DMD on in vitro organic matter digestion gave more accurate DMD predictions than did a multiple regression equation.