The Habitat Amount Hypothesis (HAH) posits that site-scale species richness should be correlated with the total amount rather than the configuration of habitat in the local landscape. This hypothesis is difficult to test empirically because habitat amount and configuration are typically confounded (e.g., if habitat fragmentation occurs as a result of habitat loss) and both influence patch-scale properties such as patch size and shape (the perimeter-to-area ratio, PAR) that may influence site-scale richness more directly. We thus investigated the relative effects of habitat amount and configuration on local-site richness of arthropods in an experimental model landscape system, in which the total amount and overall fragmentation of habitat (red clover) were independently controlled at a ‘landscape’ scale (plot = 256 m2). Arthropods were surveyed at two local scales (1-m2 and 4-m2) and at the whole-plot scale six times over a three-year period.
Results/Conclusions
Neither patch size, habitat amount, nor fragmentation were strongly correlated with local-scale richness (generalized linear models; all P > 0.05). Instead, habitat amount had an overwhelmingly positive effect on plot-scale richness (all P < 0.001), although early on there was also a significant interaction with fragmentation. Model selection using information criteria (AICc) reinforced that local-scale richness was best modeled as a constant, independent of patch or landscape attributes. Nevertheless, patch attributes (size and PAR) were more often in the top-ranked model set (ΔAICc < 2.0) than landscape attributes (total amount and fragmentation of habitat). At the 'landscape' scale, habitat amount alone was the top-ranked model in all surveys, although there was support for the interactive effects of habitat amount and fragmentation on species richness at this scale. In sum, we find experimental support for the HAH, but only for species richness at the landscape scale. The effects of fragmentation were more subtle, being transient (occurring early in the experiment), non-linear (important only at some habitat amounts), or indirect (by increasing the PAR of patches). Thus, habitat configuration still plays an important, if secondary, role in the response of communities to the amount of habitat on the landscape.