Species–area relationships arise from interaction of habitat heterogeneity and species pool

Species–area relationships (SARs) represent a ubiquitous and useful empirical regularity characterizing biodiversity. The rate of species accumulation, captured by the value of the exponent, z, varies substantially and for many reasons. We hypothesized that one of the major contributors to this variation is heterogeneity and its change with scale. To test this hypothesis, we used an array of natural microcosms for which we had invertebrate species composition and physical properties of habitat. Using GIS and cluster analysis, we organized the species data into four sets: communities grouped by spatial proximity in the field, randomly, by similarity of their physical habitat and by dissimilarity of their physical habitat. These groupings produced varying levels of heterogeneity at different scales. We fitted species–area and species–volume relationships to the four groups of communities, and obtained z-values for each group or a portion of the group if the slope of the relationship varied. As predicted, we recovered a number of properties reported by others. More interestingly, we found that small- and large-scale habitat heterogeneity produced scale-dependent z-values while the random grouping of pool habitats produced z-values more robust across scales but also susceptible to initial values of habitat richness. Habitat area affected rate at which species accumulated much less than the mean degree of inter-habitat differences: increasing area that is heterogeneous at broader scales produces higher z-values than increasing an area that shows heterogeneity at small scale only. Our results, while from a microcosm system, rely on logic transferable to larger scale data sets.