VARIANCE IN SPECIES RICHNESS, SPECIES ASSOCIATION, AND NICHE LIMITATION

A low variance in species richness, relative to a null model, has been proposed as evidence for niche limitation in ecological communities. We demonstrate that niche limitation does not inevitably lead to low v ariance in species richness. Furthermore, low variance in species rich ness can be caused by physical limits to the number of individuals in a sample, and by spatial dependence. Two methods (the rotation/reflect ion method and the random shifts method) for creating spatially-explic it null models are proposed to correct for spatial dependence in trans ect-based and grid-based sampling schemes. These methods can also be u sed to correct for measures of association between species. The rotati on/reflection method is not as powerful as the random shifts method, b ut the Latter requires some simplifying assumptions about the nature o f spatial patterns. A new sampling design, the trainsect, is proposed to eliminate the need for spatial assumptions.