Agriculture Reference
In-Depth Information
Another aspect that is not represented
appropriately in the current version of In-
raPorc is the fact that animals have different
mechanisms to cope with an amino acid de-
ficiency. Currently, an amino acid defi-
ciency results in a lower supply of available
amino acids, but feed intake will not be af-
fected. It is known that feed intake can be
reduced when pigs are offered a diet with a
limiting amino acid content. This has been
observed for certain amino acids (e.g. Trp,
Val, Ile and Met) but not for others (e.g. Lys
and Thr). This may be an explanation for
the lower maximum efficiencies of these
amino acids relative to Lys. In a series of ex-
periments, Gloaguen
et al
. (2011, 2013) ob-
served that the growth response of piglets to
a limiting amino supply varied among
amino acids. A deficiency in Val or Ile re-
sulted in a strong reduction in growth,
whereas a deficiency in His or Leu resulted
in more moderate reductions in perform-
ance. The response to an amino acid defi-
ciency may thus be driven by a reduction in
feed intake, changes in amino acid compos-
ition of body protein, differences in max-
imum efficiency of amino acid utilization or
by a combination of these factors.
estimate of the requirement of the herd in
the current version of InraPorc. However, the
optimum nutrient supply relative to the
requirement of the average animal depends
on the economic context. When nutrients
are expensive, a level greater than 110% may
be economically more beneficial (Quiniou
et al
., 2013).
To answer the question of 'which ani-
mal in the herd do we want to feed?', we
have to know how animals within a herd
differ. During the past few years, we have
studied the variation in the five main model
parameters because they do not vary in an
independent way. An animal that eats a lot
probably also grows faster (or the other way
around, if you prefer). It is important to
consider the covariance among model
parameters because ignoring it results in an
overestimation of the observed variation.
Relatively little is known about the covari-
ance structure among model parameters
and this problem is not specific to InraPorc
and, indeed, concerns all simulation models
(Knap, 1996).
Vautier (2013) obtained data on feed in-
take and body weight for 1288 individual
pigs originating from different cross-breeds,
and sexes (barrows and gilts), which were
raised in different batches for a total of
40
subpopulations. Feed intake was deter-
mined daily using an automatic feed dis-
penser and animals were weighed every
2-3
weeks. With these data, each of the in-
dividual pigs could be characterized so that
the covariance structure of model param-
eters could be determined, with possible ef-
fect of cross-breed, sex and batch.
Figure 2.5
shows the correlation structure of the five
main model parameters indicated by the
solid lines and calculated traits indicated
by the dashed lines. Overall, more than
50% of the variation was shared among the
40
subpopulations. The two parameters de-
scribing the feed intake curve on an NE
basis (DFI50 and DFI100) were correlated
and, as anticipated, average daily feed in-
take (ADFI) was positioned in-between.
Likewise, the parameter describing the
mean protein deposition (Pm) was correl-
ated with average daily gain and, to a lesser
extent, with the parameters describing the
Accounting for Variation
Among Animals
InraPorc is a (somewhat) mechanistic and
deterministic model and thus does not ac-
count for variation among animals. The
consequence of this is that certain predic-
tions have to be used with caution. For ex-
ample, one could use InraPorc and define
an animal profile based on the average ani-
mal in the herd. However, feeding the herd
based on the nutritional requirements of
the average animal would result in nutri-
ent supplies that would be sufficient for
50% of the pigs in the herd, and deficient
for the other 50%. Preliminary studies in-
dicated that providing nutrients at a level
of 110% of the requirement of the average
pig covered the requirement for a majority
of animals in the herd (Brossard
et al
.,
2009) and this level is given as a rough
