Farm Animal Metabolism and Nutrition

Data analysis requires, at minimum,
a spreadsheet program with macro and
good graphics capability. We currently are
using Microsoft Excel 97© (Microsoft,
Redmond, Washington). For numerical
analysis of gas curves, a non-linear curve-
fitting program such as SAS (SAS Inst.,
Inc., Cary, North Carolina) or TableCurve©
from Jandel/SPSS (Chicago, Illinois) will
be required (see Appendix). These pro-
grams will run much faster on a machine
with a faster processor.

Hardware costs

Pressure sensors are relatively expensive
(US$40–100 each) and have a limited life-
time. They are usually designed for
measurement of dry gases; the water-
saturated atmosphere of an in vitrofermenta-
tion, coupled with traces of hydrogen
sulphide, restricts their lifetime to between 6
months and 2 years. Multiple place systems
such as those of Cone et al.(1996) (12 place)
or Pell and Schofield (1993) (16 place) thus
represent a significant investment in sensors.
A/D cards are also relatively expensive
(US$400–800 each, 16 channel) because the
market for this hardware is more restricted
than that for computers.

Availability

Both of the open systems discussed above
are now available commercially. (Enquiries
should be addressed to: Dr M.K.
Theodorou, IGER, Plas Gogerddan,
Aberystwyth, Ceredigion, SY23 3EB, UK or
to Dr John W. Cone, ID-DLO, Department of
Ruminant Nutrition, PO Box 65, NL-
8200 AB Lelystad, The Netherlands.)
Detailed directions for construction of the
closed system have been published (Pell
and Schofield, 1993; Schofield, 1996).

Models for Curve Fitting

In Fig. 10.3 are shown typical gas curves
from the in vitrodigestion of four different

types of feed. These data illustrate the

range of shapes commonly seen for such

curves. At a qualitative level, one might

conclude that:

(i) On a dry matter basis, soy hulls are

the most, lucerne the least digestible

of these feeds;

(ii) rates of digestion vary widely, both

with feed and with time; and

(iii) maize silage and wheat straw produce

more sigmoidal curves than soy hulls

and lucerne.

For quantitative analysis, there would

be a clear advantage if a mathematical

model were available to analyse the detailed

information hidden within these gas curves.

There is no general agreement on a single

model among workers in this field, and we

briefly review proposed models and identify

underlying assumptions.

Models, an initial distinction

Most animal feeds contain a complex mix-

ture of carbohydrates that are digested at

different rates. For kinetic analysis, we can

treat the substrate either as a single pool,

digesting at a fractional rate that varies

with time, or as multiple pools, each of

which has a characteristic fractional diges-

tion rate. The latter multiple pool approach

is appealing because chemical assays have

been devised to measure separate carbo-

hydrate pools (such as NDF, ADF, NSC)

and these pools are digested at different

rates in vitro. Multiple pools are also

conceptually simpler to understand. We

consider them first.

Simple exponential model

The in vitrofibre digestion method of Tilley

and Terry (1963), as modified by Goering

and Van Soest (1970), has been widely used

to measure insoluble fibre digestion rates.

The data from such experiments are usually

fitted using a discontinuous exponential

function (discontinuous because of the lag

term) of the form:

Gas Production Methods 217