Feed management is one of the principal levers by which the production and composition of milk by dairy cows can be modulated in the short term. The response of milk yield and milk composition to variations in either energy or protein supplies is well known. However, in practice, dietary supplies of energy and protein vary simultaneously, and their interaction is still not well understood. The objective of this trial was to determine whether energy and protein interacted in their effects on milk production and milk composition and whether the response to changes in the diets depended on the parity and potential production of cows. From the results, a model was built to predict the response of milk yield and milk composition to simultaneous variations in energy and protein supplies relative to requirements of cows. Nine treatments, defined by their energy and protein supplies, were applied to 48 cows divided into 4 homogeneous groups (primiparous or multiparous x high or low milk potential) over three 4-wk periods. The control treatment was calculated to cover the predicted requirements of the group of cows in the middle of the trial and was applied to each cow. The other 8 treatments corresponded to fixed supplies of energy and protein, higher or lower than those of the control treatment. The results highlighted a significant energy x protein interaction not only on milk yield but also on protein content and yield. The response of milk yield to energy supply was zero with a negative protein balance and increased with protein supply equal to or higher than requirements. The response of milk yield to changes in the diet was greater for cows with high production potential than for those with low production potential, and the response of milk protein content was higher for primiparous cows than for multiparous cows. The model for the response of milk yield, protein yield, and protein content obtained in this trial made it possible to predict more accurately the variations in production and composition of milk relative to the potential of the cow because of changes in diet composition. In addition, the interaction obtained was in line with a response corresponding to the more limiting of 2 factors: energy or protein.

In recent years, it has become increasingly clear that understanding nutrient partitioning is central to a much broader range of issues than just being able to predict productive outputs. The extent to which nutrients are partitioned to other functions such as health and reproduction is clearly important, as are the efficiency consequences of nutrient partitioning. Further, with increasing environmental variability, there is a greater need to be able to predict the ability of an animal to respond to the nutritional limitations that arise from the environment in which it is placed. How the animal partitions its nutrients when resources are limited, or imbalanced, is a major component of its ability to cope, that is, its robustness. There is mounting evidence that reliance on body reserves is increased and that robustness of dairy cows is reduced by selection for increased milk production. A key element for predicting the partition of nutrients in this wider context is to incorporate the priorities of the animal, that is, an explicit recognition of the role of both the cow's genotype (genetic make-up), and the expression of this genotype through time on nutrient partitioning. Accordingly, there has been a growing recognition of the need to incorporate in nutritional models these innate driving forces that alter nutrient partitioning according to physiological state, the genetically driven trajectories. This paper summarizes some of the work carried out to extend nutritional models to incorporate these trajectories, the genetic effects on them, as well as how these factors affect the homeostatic capacity of the animal. At present, there are models capable of predicting the partition of nutrients throughout lactation for cows of differing milk production potentials. Information concerning genotype and stage of lactation effects on homeostatic capacity has not yet been explicitly included in metabolic models that predict nutrient partition, although recent results suggest that this is achievable. These developments have greatly extended the generality of nutrient partitioning models with respect to the type of animal and its physiological state. However, these models remain very largely focussed on predicting partition between productive outputs and body reserves and, for the most part, remain research models, although substantial progress has been made towards developing models that can be applied in the field. The challenge of linking prediction of nutrient partitioning to its consequences on health, reproduction and longevity, although widely recognized, is only now beginning to be addressed. This is an important perspective for future work on nutrient partitioning.