Species richness is a scale-dependent property, as reflected in the ubiquitously described pattern of species area relationship (SPAR).  The number of species occupying any given area is affected by processes operating at various spatiotemporal scales.  Barring evolutionary mechanisms operating at extremely largescales, three main mechanisms have been proposed for explaining SPARs: sampling effects, habitat heterogeneity, and extinction-colonization balance.  The qualitative distinctions and the quantification of the contributions of these mechanisms to observed SPARs are challenging tasks that have been approached using various combinations of statistical approaches, sampling designs and, rarely, experiments.  We used a hierarchical multi-grain sampling design to sample scale-dependent plant species density in habitat islands of natural vegetation within an agricultural landscape in the southern Judean Lowlands in Israel.  Our main goal was to test the combined effect of the position along a sharp climatic gradient, patch size, patch isolation, and habitat heterogeneity on plant species density in light of SPAR.  More specifically, we used a hierarchical partitioning of total variance to determine whether the relative contribution of the above mentioned scale-dependent explanatory factors to the explained variance in species density vary with the sampling grain.

Results/Conclusions

We found that overall both geographical position, habitat heterogeneity, and patch size (as a proxy for colonization-extinction balance), but not patch isolation significantly affected species density in at least one sampling grain.  Furthermore, the significance and the relative contribution of each of these factors to the explained variance in species density varied with the sampling grain.  While the geographical position explained most of the variance in species density at small sampling grain (0.06 m²), microhabitat heterogeneity explained most of the variance in species density at larger sampling grain (225 m²).  Patch size contributed mostly to the explained variance in species density at intermediate sampling grain (1 m²) and less so at smaller or larger grains.  Standardized sampling unit are often used to test and infer the effects of various scale-dependent mechanisms on species richness patterns including SPARs.  Our results indicate that the very choice of sampling grain may influence the observed patterns and the inference regarding the hypothesized mechanisms that generate these patterns.  We showed that the combination of a hierarchical sampling design and a hierarchical partitioning of variance provide a more comprehensive picture of the effects of scale-dependent factors on species density patterns and SPARs.