Prediction of the digestible and metabolizable energy content of wheat milling by-products for growing pigs from chemical composition

• We determine the chemical composition of 30 samples of wheat milling by-products (wheat bran, wheat middlings, wheat shorts, wheat red dog, wheat feed flour).
• We use chemical components to establish prediction models to estimate DE and ME content.
• Nutrient composition, DE and ME content varied substantially and NDF can be used to accurately predict the DE and ME content.

Thirty samples of wheat milling by-products (wheat bran, wheat middlings, wheat shorts, wheat red dog, wheat feed flour), collected from 11 flour mills, were fed to growing pigs to determine their digestible energy (DE) and metabolizable energy (ME) content and to establish equations for predicting their DE and ME content based on chemical analysis. The basal diet was based on corn and soybean meal while the other 30 experimental diets contained 290.4 g/kg wheat milling by-products added at the expense of corn and soybean meal. The 31 diets were fed to 96 growing pigs (BW = 61.9 ± 3.2 kg) according to a completely randomized design during two successive periods. During each period, the 30 experimental diets were fed to three pigs and the basal diet was fed to six pigs, resulting in 6 replications per experimental diet and 12 replications for the basal diet over the two periods. The chemical composition of the 30 samples was variable, and starch and fiber content had a strong negative correlation (r = −0.96 to −0.99 for CF and ADF, respectively). The DE content of wheat feed flour, wheat red dog, wheat shorts, wheat middlings and wheat bran averaged 17.4, 16.9, 15.2, 12.5 and 12.0 MJ/kg DM, respectively. From the stepwise regression analysis, a series of DE and ME prediction equations were generated. The best fit equations for wheat milling by-products were: DE (MJ/kg DM) = 19.2 − (0.016×aNDF) with R2 = 0.94, RSD = 0.58 and P<0.01; and ME (MJ/kg DM) = 16.9 – (0.0136×aNDF) with R2 = 0.94, RSD = 0.50 and P<0.01. Few additional variables came into the equation models because of the strongly correlated relationships among chemical components. The results indicate that DE and ME content varied substantially and various correlated single predictors (aNDF, ash, CF, starch, etc.) can be used to accurately predict the DE and ME content when fed to growing pigs.