Introduction Near infrared (NIR) spectroscopy is an analytical technique measuring light absorption in the 780–2500 nm region which is closely related to important chemical bonds (OH, NH and CH). NIR can be used to measure many nutritionally important constituents of concentrate and forage feedstuffs (Roberts et al. Reference Roberts, Workman and Reeves2004; Andres et al. Reference Andres, Murray, Calleja and Giraldez2005).
Approach NIR spectroscopy depends on the development, in representative sets of samples, of mathematical relationships (calibration equations) between spectra and constituents or attributes of the samples measured by conventional chemistry. These calibrations are then applied to the spectra of unknown samples to estimate constituents or attributes of interest (Williams and Norris, Reference Williams and Norris2001). NIR calibration equations tend to be specific to the circumstances of the data used for their development.
A major advantage of NIRS is that application allows rapid, routine and economical analysis of feedstuffs where appropriate calibrations are available for constituents of interest. Also, only one determination of the NIR spectrum of the feedstuff is required to estimate a large range of constituents. NIR can also be used to measure some functional properties of feedstuffs. Disadvantages include the need to acquire and analyse large (hundreds or sometimes thousands) sets of samples representing the range of NIR spectral diversity associated with various classes of the feedstuff (e.g. plant cultivars, agronomic, soil and seasonal conditions) to develop reliable and robust calibration equations. Conventional analysis is required for each constituent or attribute to determine the reference values necessary to develop the calibration equations. In addition, ongoing conventional analysis of subgroups of samples is needed indefinitely to validate and adjust calibration equations. Other constraints are that NIR instrumentation requires substantial capital investment, and considerable technical skills are required to develop and maintain calibration equations.
In a similar manner to the analysis of forages, NIR analysis of faeces allows estimation of many attributes of the diet of ruminants. Such estimation of diet from faecal NIR spectra depends on the similarity of the NIR spectra of forages and matching faeces (diet-faecal pairs) despite effects of digestion in the gastro-intestinal tract. Prediction of diet from faecal NIR spectra appears most reliable for forage diets. Thus many attributes of the diet selected by grazing ruminants (e.g. nitrogen content, digestibility, non-grass content) can be estimated when appropriate calibration equations are available.
In general NIRS is appropriate for organic constituents that comprise greater than about 1% of the feedstuff. NIRS is not generally suitable for mineral analysis although there are exceptions. Some functional as well as chemical properties of feedstuffs can be measured using NIRS. For example NIRS is often more satisfactory than the established laboratory procedures to estimate in vitro digestibility of forages. Also the voluntary intake of forage dry matter and digestible energy by ruminants can often be estimated more accurately from NIRS measurements of the forage than from constituents such as fibre or lignin.
Conclusions NIRS can be used to analyse many chemical and functional properties of animal feedstuffs. Reliable analysis depends on development and maintenance of appropriate calibration equations, and these require appreciable resources and skills.
