Theories and concepts were investigated that have been applied in thre
e extant models of rumen function by Baldwin et al., Danfaer and Dijks
tra et al., as a preliminary step to investigate theories of intake re
gulation. These models are the most detailed at present and differed p
articularly in the description of microbial metabolism. Simulations we
re performed on inputs derived from seven experimental diets with very
complete observations available of rumen dynamics. Comparison between
models indicated that their simulation results differed markedly. In
addition to daily feed intake and feed composition as model inputs, ea
ch model requires its own set of parameter inputs. However, some param
eter inputs could not be estimated accurately from the available obser
vations. The role of these unknown inputs on simulation results was st
udied by manipulating their estimated value. It was concluded that, in
particular, parameter inputs whose concepts do not correspond to rume
n observations have a large impact on model behaviour. Therefore, mode
ls need to be developed further to versions that use parameter inputs
that can be readily estimated from rumen observations. The goal of the
investigated rumen models is to simulate the chemical transactions in
side and the nutrient release from the rumen. Nutrients absorbed after
digestion seem to be important factors for regulation of daily intake
. However, the models take daily feed intake as a known input instead
of describing the regulation of intake, although it is highly unpredic
table in practice. Thus, extant rumen models need to be developed and
evaluated further before they can be used to investigate theories that
go beyond the rumen compartment. (C) 1997 Academic Press Limited.