Relationships between bioavailable energy estimates made with pigs and cockerels

Abstract

Samples of 84 feedstuffs, comprising both single ingredients and mixed diets, were assayed for apparent and true metabolizable energy corrected to zero nitrogen balance using pigs (PAME n ) and cockerels (CTME n ), respectively. The purpose was to assess the feasibility of using data derived with birds to predict energy available to pigs. Linear regression equations were obtained for various classes of feedstuffs using PAME n as the dependent variable and CTME n , with and without crude fibre (CF), as the independent variable. The overall relationship between PAME n and CTME n was weak ( R 2 = 0.697) and was not improved ( P>0.05) by including CF in the model. Samples of four meat and bone meals had atypical relationships; their removal from the data set elevated R 2 values to 0.775 and 0.898 for models without and with CF, respectively. An equation based on data from cereals, cereal by-products, mixed diets and miscellaneous materials had the form PAME n=6.686(±0.838)+0.5960(±0.0528) CTME n−0.01190(±0.002146) CF with an R 2 value of 0.923 and a residual mean square of 0.3030. The best equation descriptive of animal + vegetable protein sources was PAME n=9.891(±1.434)+0.4274(±0.1053) CTME n−0.01612(±0.00367) CF with an R 2 value of 0.833 and a residual mean square of 0.9934. The equations were tested by examining the differences between actual and predicted PAME n values using data from which the equations were derived. A second test involved applying the equations to independent CTME n data and comparing predicted PAME n results with published values. It became apparent that for cereals and mixed diets PAME n and CTME n values are not different. Causes of variation and opportunities for future work are discussed.