in Fig. 10.3 are fitted using Equation 10.8,
only one (wheat straw) gives a high Fvalue
combined with well-defined parameters
(see Appendix: Non-linear curve fitting).
Do Multiple Pools Really Exist?
In forage analysis, we are used to the
concept of multiple carbohydrate fractions.
The classic detergent system of Goering and
Van Soest (1970) divides carbohydrates into
fractions soluble and insoluble in neutral
and acid detergent (NDS, NDF and ADF,
respectively). These fractions can be charac-
terized further in terms of their content of
soluble sugars, starch, pectin, hemicellulose
and cellulose. In vitrostudies on individual
components have shown that the rates of
fermentation of these fractions differ
widely. Soluble sugars and pectin ferment
more quickly than most forms of starch and
much more quickly than cellulose and
hemicellulose (Sniffen et al., 1992).
The multi-pool kinetic analysis of gas
curves also predicts up to three pools with
varying sizes, digestion rates and (some-
times) lag terms. The relationship, if any,
between these mathematically derived
pools and the chemical fractions in a given
forage is potentially of interest.
Models with different mathematical
structures can yield quite disparate results
from the same gas curve. This point is
illustrated in Fig. 10.7 where the data for
wheat straw and soy hulls (Fig. 10.3) have
been fitted using either the multi-pool
model of Groot et al. (1996) (Equation
10.7) or the logistic model of Schofield et
al. (1994) (Equation 10.6). The Groot
model requires two pools for soy hulls,
one for wheat straw; the logistic model
requires exactly the reverse. For purely
descriptive applications, the choice of
model will be dictated by the closeness
of fit. In a more detailed analysis, the
validity of pool assignments may become
important.Gas Production Methods 221Fig. 10.6.Curves produced by the empirical equation of Groot et al. (1996). Different values of the
parameters b and c produce quite different shapes (b5c3 means b = 5 h and c = 3).