Farm Animal Metabolism and Nutrition

sample, it is necessary to develop a calibra-
tion for the determination of solid feed
composition by NIRS. However, the two
procedures are as different as the samples
in question. In the case of protein or P in
solution, for all intents and purposes one
has a single analyte (protein or P content)
changing in an otherwise constant media
(water, buffers, etc.). Also, the analyte at
best represents a very small percentage of
the total system. When using NIRS to
determine feed composition (i.e. neutral
and acid detergent fibre (NDF and ADF),
crude protein (CP), lignin, etc. (Van Soest,
1994)), one has a solid medium (feed itself)
in which the analyte (NDF, etc.) is not only
the object of the determination, but also
constitutes a significant fraction of the
medium itself. Also, unlike the earlier
calibrations for protein or P in which the
media (water, buffers, etc.) are generally
non-absorbing at the wavelength chosen,

with feeds the media consists of other com-

ponents of the feedstuff which also have

absorption bands in the NIR, and generally

at the same wavelengths as the component

of interest. This can be seen in Fig. 9.3,

where the NIR spectra of casein and cellu-

lose are shown. For a hay or silage, the NIR

spectrum thus consists of overlapping

spectra of the many different constituents

present, each one different, but, like cellu-

lose and casein, often having absorptions

in the same or similar regions. Except for

pure and simple materials such as acetone,

spectra do not consist of sharp, individual

peaks. Thus, developing a calibration

becomes a very complicated process. As

a result, calibration development has

resulted in NIRS becoming almost synony-

mous with the terms multivariate statistics

and chemometrics.

Multivariate simply means that there

are multiple variables (wavelengths in the

Use of Near Infrared Reflectance Spectroscopy 187

Fig. 9.1.Diffuse reflectance for a ground solid showing: (a) reflection from the cell cover; (b) specular
reflection from the sample; (c) absorption and diffuse reflection from the sample; and (d) total absorption
by the sample.