Pitfalls in Small-Scale Species-Area Sampling and Analysis

Analyses of the dependency of species richness (S) on area (A), the so-called species-area relationships (SARs), are widespread approaches to characterize and compare biodiversity patterns. This article highlights - with a focus on small-scale SARs of plants in continuous ecosystems - how inappropriate sampling methods or theoretical misconceptions can create artifacts and thus may lead to wrong conclusions. Most of these problems have been recognized before but continue to appear regularly in the scientific literature. The following main points are reviewed and discussed: i) Species richness values and SARs depend on the measurement method (any-part vs. grid-point system); ii) Species-richness values depend on the shape of the analyzed plots; iii) Many published SARs are not true SARs but instead represent species sampling curves or their data points consist of richness totals for incontiguous subplots; iv) Nested-plot design is the preferred sampling method for SARs (the claim that this approach would cause pseudoreplication is erroneous); v) SARs should be constructed using mean values of several counts for the smaller areas; vi) SAR functions can be fitted and selected both in the S- and the log S-space but this must be done consistently for all compared function types. It turns out that the finding of non-power function SARs in many studies is due to a lack of awareness of one or several of the named points. Thus, power-function SARs are even more widespread than a recent review would suggest. I therefore propose to use the power law as a universal model for all types of SARs but to test whether the slope z varies with spatial scale. Finally, I urge readers to be aware of the many pitfalls in SAR studies, to fully disclose methodology, and to apply a meaningful and consistent terminology, especially by restricting the terms "species-area relationship" and "species density" to situations in which each data point represents a contiguous area.