Modelling post-absorptive protein and amino acid metabolism in the ruminant.

A goal of animal scientists has been to generate information capable of improving the productivity of farm animals. Historically, improvements in productivity have been achieved by enhancements in performance, i.e. increased rates of growth, lactation, etc. However, maximum performance may not always equate to maximum profit, particularly if one considers environmental costs associated with the animal enterprise (Tamminga, 1992). In addition, as generally each of the ingredients that can be fed has a different nutritive value and financial cost, various combinations of ingredients will yield different levels of performance and costs of production for the producer. In order to give due consideration to all these factors, mathematical equations adequate to describe the relationships between inputs and outputs must be formulated. Failure to formulate such relationships prevents one from truly maximizing profitability except in a serendipitous manner. Mathematical systems describing the energetics of most ruminant farm animals have been formulated and parametrized (Janige, 1978; Van Es, 1978; Agricultural Research Council, 1980; National Research Council, 1989; Agricultural and Food Research Council, 1990). These systems have been refined over time and have served the industry well. However, the ruminant protein evaluation systems (Viriti et al. 1987; National Research Council, 1989; Agricultural and Food Research Council, 1992; Tamminga et al. 1994) have been less fruitful (Beever & Cottrill, 1994). One common feature of these various systems is that they are based largely on empirical relationships derived from whole-animal observations. A major advantage to this approach is that the resulting model will usually be fairly simple and, thereby, easily deployed by the industry. However, if the problem of study is fairly complicated, as the description of protein requirements of animals appears to be, such an approach becomes cumbersome due to the inability to directly utilize data such as trans-organ fluxes or enzyme expression levels to describe and parametrize the model. An alternative approach is to construct sub-models describing organ or cellular function and to aggregate and consolidate these sub-models into a system describing the whole animal. This approach lends itself to direct parametrization from organ, cellular, and subcellular data. It also allows the compartmentalization of the problem, which can be useful when dealing with very complicated systems. In theory, one would identify these two approaches as discrete. However, in practice they represent the two ends of a continuous spectrum. Examples of empirical, mixed, and sub-model approaches for ruminant protein systems include the Cornell net carbohydrate and protein system (CNCPS; Fox et al. 1992; Russell et al. 1992; Sniffen et al. 1992; O'Connor et al. 1993), the models of Baldwin et al. (1987) and Gerrits et al. (1996); and the model of Gill et al. (1989) respectively. Although the objectives of these various