Species diversity and community structure indices (richness, diversity and evenness) are typically derived from contemporary biological survey atlases to identify conservation areas or to determine macroecological-environment relationships. Surprisingly, there are few studies based on national taxa surveys that describe and measure the influence human disturbance patterns may have on these ecological measures. This study uses various spatial statistical methods to examine and model large-scale spatial structure in diversity, structure indices derived from a comprehensive bird atlas, and composite variables of environment and human land transformation. Data were derived from an atlas of South African birds and segregated into five vegetation biomes. For each sample location, environmental and land-use data were used to calculate composite environmental gradients in climate, topography and human land transformation. Semivariograms were used to detect large-scale trends and spatial scale, and Moran's I statistic correlograms to test for large-scale spatial autocorrelation in detrended diversity indices. Ordinary least squares regressions for all biomes indicated a significant positive relationship between high levels of human transformation and species diversity, whereas high transformation had a significant negative influence on evenness. These same relationships held for the majority of biomes after the effects of environmental gradients were removed. The analysis also shows that in areas with favourable environmental resource levels, diversity for birds increases. These areas also tend to contain a large proportion of highly transformed land, and evenness decreases in the woodland, grassland, Karoo and fynbos biomes. This suggests that high-intensity transformation may bring in novel resources for birds not normally found in some of the biomes, with some species adept at exploiting these changes and reaching high densities with a tendency for assemblage structure to drift towards single species dominance. The results imply that species data derived from contemporary atlases may begin to demonstrate the effects of human influence on ecological measures rather than only indicating the effects of environmental variation on community structure.