In silico sampling reveals the effect of clustering and shows that the log-normal rank abundance curve is an artefact

The impact of clustering on rank abundance, species-individual (S-N)and species-area curves was investigated using a computer programme for in silico sampling. In a rank abundance curve the abundances of species are plotted on log-scale against species sequence. In an S-N curve the number of species (S) is plotted against the log of the total number of individuals (N) in the sample, in a species-area curve S is plotted against log-area. The results from in silico sampling confirm the general shape of S-N and speciesarea curves for communities with clustering, i.e., a curve that starts with a smaller slope but that later is temporarily steeper than the curve expected for Poisson-distributed species. Extrapolation of S-N and species-area curves could therefore be misleading. The output furthermore shows that sigmoid rank abundance curves (curves of the type of a log-normal or broken stick) can be an artefact of the standard procedure of first sorting the species in sequence of abundance in combination with clustering in the low abundant and rare species. This makes the usual explanation given to the log-normal rank abundance curve dubious. An extension of the negative-binomial rank abundance curve-fit model is discussed to make it suitable for also fitting sigmoid rank abundance curves.