(1996). They isolated the water-soluble
and water-insoluble fractions of lucerne
(Medicago sativa) and bromegrass (Bromus
inermis) and recorded gas curves during in
vitrodigestion of the whole forage and of
the two fractions. All curves were fitted
using a two-pool logistic model (Equation
10.6) and gas volumes corresponding to the
faster- and slower-digesting components
were derived from these data. The authors
demonstrated a close correspondence
between the total faster pool volume (from
the whole forage) and the sum of the faster
pool sizes in the water-soluble and water-
insoluble fractions. They also compared
the total gas curve from the whole forage
with that created by adding the individual
curves from the two component fractions.
Discrepancies of up to ±10% were found,
but the overall agreement was quite good.
These results suggest that pool allocations
under the logistic model are not unreason-
able.
Cone et al. (1997) did a similar
experiment but analysed the results differ-
ently. They used a three-pool model to fit
digestion curves from grasses, clover,
maize silage, corn cob mix and chopped
ear corn silage. They hypothesized that
pools 1 and 2 corresponded to the water-
soluble and water-insoluble fractions of
these feeds (pool 3 was equated to micro-
bial turnover – see Appropriate Use of
Blanks, p. 215). They compared the
separately measured digestion curves of
these isolated fractions with the curves
corresponding to pools 1 and 2 in the
intact feed and found some similarities
among these sets of curves but noted that
the match was not exact. More work is
needed to explore the consistency of pool
allocations using this model. We conclude
that caution is appropriate in using multi-
pool analysis of gas data. While it is clear
that plants contain carbohydrate fractions
that are digested at different rates, the
‘pools’ depicted in Fig. 10.7 should be
viewed as purely mathematical constructs
that may or may not correspond to
chemical entities. The actual existence of
such fractions should be documented by
independent evidence.
Plant Carbohydrate Fractions and
Nutritional Models
In vitro digestion rates from gas curves can
be used in nutritional models. The Cornell
Net Carbohydrate and Protein System
(CNCPS) predicts nutrient supply based on
the competition between ruminal digestion
and passage. The carbohydrate composi-
tion of each feed ingredient is described by
the following four fractions: (A) sugars and
organic acids; (B1) starch and pectic sub-
stances; (B2) digestible fibre; and (C)
indigestible residue (Sniffen et al., 1992)
(Fig. 10.8). Organic acids are treated, for
convenience, as carbohydrates in this
system.
The CNCPS model thus requires rate
and pool size information on these four
fractions. The B2 (digestible fibre) fraction
is relatively easy to deal with because this
fraction can be isolated chemically and its
digestion behaviour measured in vitro
(Schofield and Pell, 1995). If the digestion
curve from NDF (or B2) is subtracted from
the digestion curve for an equivalent
amount of the intact forage, one obtains a
curve corresponding to the digestion of the
neutral detergent-soluble (NDS) fraction of
the forage (Schofield and Pell, 1995a). This
fraction contains both the A and B1 CNCPS
fractions (Fig. 10.8).
The A fraction (sugars, organic acids)
can be removed from a forage by treatment
with 80% aqueous ethanol (Smith, 1981).
The ethanol-insoluble residue (EIR) can
also be fermented in vitro. Thus, if we fer-
ment separately the whole forage (WF) and
the EIR and NDF fractions (Fig. 10.8), curve
subtraction will allow us to derive separate
digestion curves for:● A fraction = WF EIR
● B1 fraction = EIR NDF
● B2 fraction = NDF.There are three main assumptions underly-
ing this approach:1.That the chemical treatments used to
prepare the EIR and NDF fractions do not
change the digestion behaviour of these
fractions.Gas Production Methods 223