Sample size and quality of indication--a case study using ground-dwelling arthropods as indicators in agricultural ecosystems

A frequently cited but rarely used procedure for estimating sample size was tested on the level of both species and community parameters. The analysed field data resulted from a pitfall trap study dealing with the effects of changing agricultural land use on ground-dwelling arthropods. The relation between level of precision (LOP) and required sample size (RSS) showed significant differences between species and community parameters. For most species parameters a high precision level (5– 10%) would require an unaffordable large sample size. For species that are habitat generalists realistic sample sizes around 10 allow an LOP between 25 and 50%, whereas the same sample sizes leads to a much lower LOP between 50 and 90% for selected habitat specialists. In general, the LOP–RSS relation was more favourable for the two tested community parameters (richness and evenness) than for the mean number of individuals of a species. For purposes of bioindication or comparisons of communities, the results presented herein favour the use of community parameters above species parameters. Finally, as a rule of thumb the RSSs for precision levels of 5, 10, 25 and 50% were given for both, level of species and for community parameters.