Agriculture Reference
In-Depth Information
Birkett and de Lange, 2001; Green and
Whittemore, 2003; Wellock
et al
., 2003a;
van Milgen
et al
., 2008; NRC, 2012). The
successful application of these models in
practice has varied owing to a number of
factors including complexity, ease of use,
ability to integrate into existing business
software and the robustness of their scien-
tific theory under commercial application.
Despite the limited degree of success, there
is no doubt that the integration of a bio-
logical growth model with a dynamic feed
formulator and optimization capabilities
significantly enhances the ability to make
well informed decisions in a highly vola-
tile market and a changing production en-
vironment. It is for this reason that an
integrated pig management system, called
Watson
®
, was developed and applied in
commercial practice. The theoretical frame-
work and associated quantitative biology
contained within Watson
®
is based on sci-
entific evidence published over the last
30
years and can be reviewed in previously
published papers (Wellock
et al
., 2003a,c;
Ferguson, 2006). Traditionally, pig growth
models have been characterized by their
ability to partition nutrients (energy and
protein or amino acids) into protein and fat
tissue, with a strong emphasis on predict-
ing static nutrient requirements and growth
responses. However, they have had limited
capacity to accommodate dynamic inter-
actions between voluntary feed intake, ani-
mal performance and production economics,
including raw material costs and pig
prices. Without the ability to predict the
effects of interactions between the animal,
the feed and the physical and social envir-
onment on voluntary feed intake and sub-
sequent body tissue growth, pig growth
models will continue to receive minimal
commercial attention. The current market
and economic challenges faced by pork
producers are unprecedented and there-
fore there is a constant need to determine
the economically optimal nutrition and
management solutions. In addition to fi-
nancial sustainability, there is increasing
demand for socially responsible pork pro-
duction including improved animal welfare
or social interactions, and reductions in
carbon footprint, eutrophication and acid-
ification. With this in mind, the expect-
ation and role of simulation models in
commercial practice is evolving such that
they are required to: (i) simultaneously
focus on nutrient, economic and environ-
mental sustainability responses through
optimization procedures integrating ani-
mal biology, least cost feed formulations
and economics; (ii) be more context orien-
tated by providing solutions to rapidly
changing market conditions and improve
problem solving capabilities; (iii) incorpor-
ate intelligent user interface processes to
improve the accuracy and reliability of
critical input data as well as simplifying
the process, such as describing the genotype
or quantifying the health status; (iv) pro-
vide a more diverse user base such as
business leaders, sales managers, technical
advisors and not just nutritionists; (v) be-
come part of the company's 'DNA' or an
integral part of the value proposition pro-
vided to their customers; and (vi) provide
accurate predictions across a wide range of
commercial conditions. This chapter will
focus on certain key components identified
as being important in the process of suc-
cessfully applying an integrated pig model
to commercial practice.
Animal Biology
For a detailed description of the biological
theory refer to papers by Emmans (1981)
and Ferguson (2006). However, one of the
key biological components necessary for
commercialization of a growth simulation
model is the prediction of voluntary feed
intake. Without the ability to predict feed
intake, it is not possible to incorporate op-
timization and particularly stochastic opti-
mization because it is the link between
animal performance and economics (Gous
and Berhe, 2006). However, predicting feed
intake is also very challenging because of
the complexity of feed intake regulation. It
is for this reason that in most models feed
intake is either an input (Pomar
et al
.,
1991; TMV, 1994; Moughan, 1995; Birkett
and de Lange, 2001), or empirical feed
