Context : Biodiversity models postulate that the relationship between richness and the proportion of natural land cover (pNLC i.e., not dominated by human uses) is: 1) monotonic positive, 2) reasonably strong , 3) consistent in different geographic areas . Earlier work examining 100-km 2 human-dominated landcover in Ontario, Canada, observed that surveyed avian species richness is a peaked function of pNLC.
Objective : We tested whether the same relationship between avian species richness and pNLC is also observed in an independent geographic area that has similar biomes. We also tested whether the peaked relations might be caused by temperature gradients, gradients in the size of species pools, grain size in the analyses, and landscape heterogeneity.
Methods : Using breeding bird atlases of Ontario (Canada) and New York State (USA), we estimated species richness in landscapes varying from 25 to 900 km 2 . We related richness to the pNLC in each landscape and examined the same relationships after controlling for temperature, habitat heterogeneity, and species pool size. Results : Landscape-level species richness is a peaked, and not very strong function of pNLC. The relationship is not an artefact of temperature gradients, species pool size, or land cover heterogeneity.
Conclusions : The proposition that increased amounts of natural land cover promotes species richness is clearly true at the limit, in landscapes with relatively little (<30%) natural cover. In landscapes with higher amounts of natural cover, a certain amount of human-modified land covers can provide habitat for species that prefer open habitats.