Agriculture Reference
In-Depth Information
in the feed ingredient and meat markets and
the demonstrable benefits of biological
models, newfound interest has emerged in
the poultry industry in improving their op-
erational efficiency and decision making.
Some modellers mentioned that the
sense that the nutritionist will be replaced by
a model is one of the most detrimental factors
for the use of models. However, there is so
much more that the nutritionist does, that this
is impossible. The prescription for feeding an
animal includes the levels of energy, protein,
individual amino acids, minerals, vitamins
and medications that the animal needs at all
stages of growth. Models can help to deter-
mine the levels of some of these nutrient
categories, but not all. The experience and
judgement of the nutritionist in making deci-
sions about ingredients, quality of ingredients,
additives, minerals, vitamin sources and
levels, medications, etc. cannot be replaced
by a model or by a less educated person.
Other professionals using or developing
models indicated that the 'failure', relative
to applications in other species, in applying
poultry models may partially be explained
by the fact that poultry nutritionists expect
strong accuracy, while they are not capable
of providing the amount of data that would
be needed to generate such model accuracy.
In other words, commercial application re-
quires a lot of understanding of and quantita-
tive information about the growing conditions
(i.e. environmental conditions, accurate FI
per bird, BW gain at different ages) of every
flock that is to be simulated. However, in
reality, producers only have mean values
that in themselves may not be accurate. Also,
because feeding programmes are imple-
mented in several flocks at the same time ra-
ther than only one, or one animal as it can be
in the dairy or swine industries, it is almost
impossible to adapt a feeding programme to
the environment and genetic potential for
every single flock. In order to at least opti-
mize parameters by groups of farms with
similar traits, it is necessary to have all the
data about those farms.
Furthermore, poultry model application
will be more successful when model devel-
opers/scientists and professors have suc-
ceeded in teaching a new approach to
nutrition. This approach consists of knowing
growth, tissue and chemical development of
the avian species or strain to work with, the
environmental conditions during grow out,
and the feed with which one is dealing. It is
also necessary to understand and interpret
model outputs that show how the bird be-
haves in such conditions, so that one can fi-
nally decide the appropriate nutritional or
production strategy to be used while simultan-
eously understanding the model's limitations
and considering individual flock variability
and feed quality. Moreover, econometric tech-
niques should be applied in poultry nutrition
and production. Applying these techniques
should help to understand the impacts of
changes in feedstuff costs and prices of
poultry products to determine the most ad-
equate nutrient levels for each specific market
condition and obtain the maximum profit-
ability.
Another explanation for the lack of use
of models in poultry research and produc-
tion was pointed out as the relatively low
experimental cost associated with poultry
research, compared with other species such
as ruminants. Consequently, very few efforts
have been devoted to optimizing the experi-
mental design, with no formal request to
make explicit the underlying assumptions
tested, leading to a lack of conceptual
models. This might change due to the in-
creasing costs and legal requirements at-
tached currently to avian experimentation.
Finally, poultry production chains in-
volve stakeholders who are direct competi-
tors on the market. Therefore, any research
that might lead to an economical advantage
is, by definition, not shared. As a conse-
quence, systemic modelling, which is a long-
term action relying on substantial economical
and intellectual supports, is difficult to im-
plement due to the lack of access to some
data and to the weakness of human resources
available within a given firm. Ways to over-
come these problems may be to create con-
sortiums between networks of firms and
research institutes to define and develop a
shared model, or public research and exten-
sion networks aiming at developing models
and at building databases usable for meta-
analysis or other modelling techniques.
