Spatio-temporal variability of richness estimators: coastal marine fish communities as examples.

We assessed the performance of two estimators of species richness, the Chao2 and the Coleman 'random placement curve'. Using a dataset of intertidal fish from the Norwegian Skagerrak coast, we found that Chao2 was effective for low sampling intensity, often reaching asymptotic values for few samples, but for higher sampling intensity the performance deteriorated. For large samples, the Coleman random placement curve was more effective than the Chao2 estimates when comparing spatio-temporal patterns of species richness. Spatial patterns were clearly and consistently identified by both methods, whereas the coastal fish communities displayed too much variability in the early summer for any sensible measure of temporal patterns of fish-species richness to be made. To control for spurious results due to systematic differences in mean abundance of the samples the analyses were performed also on data standardised by the number of individuals in the samples, without any significant change in the results. We conclude that modest sampling effort is sufficient to characterise spatial patterns of coastal fish-species richness, while a detailed and high-precision description of seasonal patterns could not be obtained with any reasonable sampling effort.