The sensitivity of n-alkane analysis to measurement error: implications for use in the study of diet composition

Abstract

It is possible to estimate diet composition from an analysis of n-alkanes in the faeces of ruminant animals. The method requires the estimation of the concentrations of n-alkanes in the plants and faeces and then the solving of a system of simultaneous equations. There are at least three places in which significant measurement error may be introduced. First, there may be error in the determination of the concentrations of the n-alkanes in the herbage. This error may be the result of analytical error in the chemical analysis, or in the gathering of the representative sample of herbage. In either case, error in this estimate may be particularly important, since this estimate is not independently repeated for each animal in the study, but is conducted once and used throughout the study. Error may also be introduced in the estimates of digestibility of the n-alkanes themselves. The n-alkane method might be ideal if in fact the n-alkanes were completely indigestible – they are not and, furthermore, they are differentially digestible. Lastly, there may be measurement error in the estimate of the n-alkane concentrations in the faeces, which utilize the same analytical procedures that are used on the herbage. That is, if measurement error exists in the herbage estimates, it is quite possible that it also exists in the faeces estimates. We address these issues through the use of Monte Carlo simulation to investigate the likely effects of measurement error on diet composition and digestibility estimates obtained using the n-alkane method. Our results suggest the following conclusions: (1) in the face of any sort of measurement error, estimates of digestibility are likely to be unreliable; (2) when measurement error exists, one of the diet components will usually be under-estimated and the other will usually be over-estimated; (3) any sort of progressive bias in the n-alkane recovery estimates will probably have large and very significant effects on the results; and (4) if measurement error in the estimates of the n-alkane concentrations in the herbage and in the faeces are similar in expectation, then their effects tend to cancel each other out.