Agriculture Reference
In-Depth Information
Models of scientific theories
identify those factors that systematically influ-
enced the results of the experiments. In poultry
science this approach was pioneered espe-
cially by G.D. Rosen and was explained by him
in promoting the term holo- rather than meta-
analysis (Rosen, 2006) for this type of model-
ling. In this paper Rosen eloquently summar-
izes the objectives and applications of holo- (or
meta-) analysis as follows:
Explicit statements of theoretical ideas have
not played a very prominent part in poultry
research, perhaps because of the availability
and cost-effectiveness of empirical trials.
One notable exception is the hypothesis
proposed by Fraps (1955) to explain the
ovulatory cycle of the laying hen. The model
of this theory was presented diagrammatic-
ally and described in the accompanying
text. Etches and Schoch (1984) formulated
the theory in mathematical terms, work
which was later extended by Johnston and
Gous (2006). This theory or model under-
pinned decades of experimental and physio-
logical studies, work that was summarized
by Etches (1996).
The other theory that has had a pro-
found effect on nutritional modelling in
both poultry and pigs was promulgated by
Emmans (1981); this is the idea that animals
have a definable purpose and that they try
to eat enough food to fulfil that purpose.
This leads eventually to a quantifiable the-
ory of food intake (Emmans, 1997) and has
formed the basis of several models for mono-
gastric animals (Emmans, 1981; Ferguson
et al
., 1994).
These two theories have been brought
together in work by R.M. Gous to model the
productivity of laying hens and broiler
breeders (Johnston and Gous, 2006; Gous and
Nonis, 2010).
(i) prediction of responses with confidence
limits; (ii) provision of software to quantify
responses to pronutrients and nutrients
unique in time, place, and economics;
(iii) translation of research conditions to
praxis (field) conditions, as in processed vs
mash feed, as-hatched vs sexed birds, floor
pen vs cage housing, presence vs absence
of diagnosed or endemic disease, practical
vs purified diet, and optimal vs suboptimal
dosages; (iv) exposure of key missing
variables, e.g. temperature and nutrient
contents; (v) discovery of theoretically
unpredictable independent variables and
interactions; and (vi) definition of topics
and priorities for future research.
The methods and applications of meta-analyses
in animal nutrition are described and dis-
cussed by Sauvant
et al
. (2008) and will not
be discussed further here.
The following list demonstrates several
significant contributions that the application
of meta- or holo-analysis has made to ap-
plied poultry science.
• Morris (1968) combined the results of
34
experiments showing the relationship
between feed intake in laying hens and
dietary energy level. A general rule for
predicting food intake was derived for
birds of different body size (characteristic
food intake). This analysis has not been
updated using more recent data.
• Fisher and Wilson (1974) combined the
results of
160
estimates of broiler re-
sponse to dietary nutrient density (energy
level at constant nutrient:energy ra-
tios). Linear regressions were derived
to summarize the effects of sex, age, breed
type, energy:protein ratio and feed form
on growth and feed intake responses.
This analysis has not been updated
using more recent data.
Models to extend and increase
the value of pen trials
Possibly the earliest method used to combine
the results of repeated experiments was to
use a committee of experts to review the
experiments and to reduce the multiple find-
ings to simple (and ostensibly useful) summa-
ries. This method was extensively used in
nutrition to resolve the issue of nutrient re-
quirements (Agricultural Research Council,
1975; National Research Council, 1994). A fur-
ther extension was to fit regression models
simultaneously to the results of several experi-
ments, both to summarize findings and to
