Temperature seasonality has been proposed to be an important determinant of bat community structure in Atlantic Forest of South America and elsewhere. I characterized spatial and environmental patterns of species richness of phyllostomid bats in Atlantic Forest of South America and examined the degree to which temperature seasonality relative to other important environmental gradients was related to patterns of diversity across 60 heterogeneously sampled assemblages. I compared efficacy of different methods of accounting for inventory completeness (rarefaction, species estimation, minimum number of individuals and raw data) in characterizing gradients of bat diversity. I also compared levels of spatial autocorrelation generated by these methods and the degree to which such autocorrelation influenced results of regression analyses. Temperature seasonality was consistently the most important predictor of species richness across all methods of accounting for inventory completeness as well as between spatial and non-spatial regression methods. Spatial autocorrelation of both species richness and regression residuals was lowest for species richness estimators and highest for data rarefied to the highest sample sizes. Strong and characteristic environmental gradients typifying Atlantic Forest of South America translate into coincident gradients of richness of bat assemblages. Although data were characterized by heterogeneous sampling efforts, relationships of species richness with seasonality of temperature were strong and easily detectable. Methods of accounting for inventory completeness improve assessment with species richness estimators performing better than rarefaction methods.