Sieve analysis is used in feed analysis, and studies of digestive physiology with various approaches to describe an average value of particle size which can serve to compare different samples. To demonstrate the effects of such different approaches, we compared five particle size indicators to demonstrate advantages and disadvantages of each method, the modulus of fineness (MOF), the discrete mean (dMEAN) and median (dMED), and the continuous mean (cMEAN) and median (cMED), well aware of the fact that a gold standard for this procedure is lacking. Data were obtained from 580 individual faecal samples of different herbivore species by wet sieving over a cascade of nine sieves with mesh sizes ranging from 0.063 to 16 mm. MOF, dMEAN and dMED can be calculated directly from the results of sieve analysis, but cMEAN and cMED require a curve-fitting procedure. Across the whole sample size, dMEAN and cMEAN showed the highest correlation. The correlation between the respective MEAN and MED was higher for d than for c. As expected, MOF deviated most from the other measurements. Simulating different sieve sets resulted in a poor correlation between the results from the different sets in MOF and cMED, but a good correlation in dMEAN and cMEAN, suggesting that these latter measures can also be compared between studies that do not use identical sieve sets. As the calculation of dMEAN is comparatively simpler and less time-consuming than that of cMEAN, we propose the dMEAN as a standard for the description of a mean particle size value obtained from sieve analysis. For practical application, the good correlation of different simulated sieve sets indicates that sets with fewer sieves could be used in large-scale studies to reduce analytical workload.
© 2011 Blackwell Verlag GmbH.