Abstract
Species–area relationships (SAR) and species-abundance distributions (SAD) are among the most studied patterns in ecology, due to their application to both theoretical and conservation issues. One problem with these general patterns is that different theories can generate the same predictions, and for this reason, they cannot be used to detect different mechanisms of community assembly. A solution is to search for more sensitive patterns, for example by extending the SAR to the whole SAD. A generalized dimension (Dq) approach has been proposed to study the scaling of SAD, but to date, there has been no evaluation of the ability of this pattern to detect different mechanisms. An equivalent way to express SAD is the rank-abundance distribution (RAD). Here I introduce a new way to study SAD scaling using a spatial version of RAD: the species-rank surface (SRS), which can be analysed using Dq. Thus, there is an old Dq based on SAR Dq(SAR), and a new one based on SRS Dq(SRS). I perform spatial simulations to examine the relationship of Dq with SAD, spatial patterns and number of species. Finally, I compare the power of both Dq, SAD, SAR exponent and the fractal information dimension to detect different community patterns using a continuum of hierarchical and neutral spatially explicit models. The SAD, Dq(SAD) and Dq(SRS) all had good performance in detecting models with contrasting mechanisms. Dq(SRS), however, had a better fit to data and allowed comparisons between hierarchical communities where the other methods failed. The SAR exponent and information dimension had low power and should not be used. SRS and Dq(SRS) could be interesting methods to study community or macroecological patterns.