An investigation of the feasibility of predicting nitrogen-corrected true metabolizable energy (TMEn) content in barley from chemical composition and physical characteristics

A study utilizing 91 barley samples was conducted to investigate the feasibility of predicting the nitrogen-corrected true metabolizable energy (TMEn) value of barley for poultry. The samples included three barley types (six-row malting and feed and two-row feed) that were grown at 12 different locations in Manitoba from 1986 to 1988. Substantial variation in chemical and physical measurements was evident among the barley samples studied, and data indicated TMEn of 13.10–14.59 with a mean of 13.97 (MJ kg−1 on dry matter). It was found that TMEn was negatively correlated with neutral detergent fibre (NDF) level (r = −0.784) and positively correlated with test weight (r = 0.627). Significant relationships also existed between TMEn and contents of starch and fat, but only 7–8% of the variation in TMEn was accounted for by those variables. Inclusion of NDF, gross energy (GE), fat, protein and starch terms in a multiple-regression equation resulted in a slight decrease in residual standard deviation and an increase in R2. The best combinations of predictor variables were, in decreasing order, NDF, GE and fat selected from stepwise regression and NDF, fat and starch selected from the Mallows Cp values. With the exception of the equation using test weight as sole predictor, the errors estimated from the simple- (NDF as sole predictor), multiple- and stepwise-regression equations or the equation selected according to the Mallows Cp value were similar to the error approached for the biological measure, TMEn.