Energy requirements of animals are most readily expressed in terms of net energy (NE), while the energy yield of feed is, at least initially, expressed in terms of metabolisable energy (ME). Energy evaluation systems ‘translate’ NE requirements into ME requirements (ME systems) or assign NE values to feeds (NE systems). Efficiency of ME utilisation is higher for maintenance than for production and the NE yield of a feed varies, therefore, with ME intake. In addition, energetic efficiency for maintenance and production is thought to be different for lactating and non-lactating animals and to be affected by diet quality. As a result, there are currently many national energy evaluation systems that are complex, differ in their approach and are, as a result, difficult to compare. As ruminants in most production systems are fed ad libitum, this is also the most appropriate intake level at which to estimate energetic efficiency. Analyses of older as well as more recent data suggest that ad libitum feeding (i) abolishes the effects of diet quality on energetic efficiency (almost) completely, (ii) abolishes the differences between lactating and non-lactating animals (almost) entirely and (iii) results in overall energetic efficiencies that are always close to 0.6. The paper argues that there is now sufficient information to develop an international energy evaluation system for ad libitum fed ruminants. Such a system should (i) unify ME and NE systems, (ii) avoid the systematic bias and large errors that can be associated with current systems (iii) be simpler than current systems and (iv) have as a starting point a constant efficiency of ME utilisation, with a value of around 0.6. The remarkably constant efficiency of ME utilisation in ad libitum fed ruminants could be the result of energetic efficiency as well as feed intake regulation being affected by the same variables or of a direct role of energetic efficiency in feed intake regulation. Models to predict intake on the basis of the latter hypothesis are already available for non-reproducing ruminants but remain to be developed for reproducing animals.

Energy evaluation systems translate an animal’s net energy (NE) requirements into feed metabolisable energy requirements (MER). The Feed into Milk (FiM) project (Agnew RE, Yan T, France J, Kebreab E and Thomas C 2004. Energy requirement and supply. In Feed into Milk. A new applied feeding system for dairy cows (ed. C Thomas), pp. 11–20. Nottingham University Press, Nottingham, UK) proposed a new system to predict MER of dairy cows that is, in contrast to previous energy evaluation systems for cattle, independent of feed quality. The FiM system shares this characteristic with an energy evaluation system for ad libitum-fed cattle proposed in 1994 by Tolkamp and Ketelaars (T&K). The FiM system requires nine parameters to translate requirements for NE into MER for dairy cows, while the T&K system for cattle requires only two for the same purpose. This paper analyses the contribution of each of the parameters to the final MER predictions, the differences in MER prediction between the two systems and the underlying causes of these differences. The systems differ considerably in their estimates of the NE that is required for maintenance and in their (implicit) assumptions about the partial efficiency of ME utilisation for lactation. The T&K system is based on a constant partial efficiency of ME utilisation, but in the FiM system this efficiency changes with milk yield (MY) and shows a sharp discontinuity that is at odds with the underlying biology. These are the two main causes of the differences in MER predictions. Nevertheless, over a range of MYs between 10 and 40 kg, and for cows maintaining, gaining or losing weight, the MER predictions of the two systems are very similar with maximum differences of up to ±2% only. FiM predictions of MER are systematically higher than T&K predictions for cows with very low and very high MY. It is concluded that the FiM system could reduce parameter requirements with negligible effects on MER predictions. The combination of a very high maintenance NE parameter and a curvilinear model with two subsequent corrections leads to internal inconsistencies in the FiM system. The T&K system is much simpler but it might benefit from including more recent information for the estimation of its parameters.