Abstract Aim Three fundamental and inter-related concepts have accrued debates: ecological communities, beta diversity (β), and spatial scale. Spatial scaling of β informs the community concept because the scale of maximal β corresponds to the most apparent size of an ecological community (without invoking external features of habitat, etc.). Here we test five alternative hypotheses about spatial scaling of β for ants, birds, diatoms, and trees across the contiguous USA, using spatial grains from 1 to 10 6 km 2 . We compare β scaling among clades and test hypotheses about repeatability where data permit for: (a) summer and winter bird β in six consecutive years; (b) trees through time (4 years, spaced 5 years apart). Finally, we compare different forms of β (i.e., observed and deviations from null models based on spatial heterogeneity and spatial homogeneity). Location The contiguous United States of America Time Period Recent but varying with clade Taxa Studied ants, birds, diatoms, and trees Methods We obtained data from publicly-available sources and assigned point locations to hexagonal grids ranging from 1 to 10 6 km 2 . At each spatial grain, we calculated mean pairwise β between each hexagon and its neighboring grids. We also compared alternative β measures and evaluated potential confounding effects of neighborhood size and species richness on results. Results Spatial scaling of β repeatedly supported the regional community concept among clades, though with different spatial scales per clade. Based on peak mean β, community size for trees (∼300 km 2 ) < winter birds (∼500 km 2 ) < summer birds (∼2000 km 2 ) ≈ ants (∼2000 km 2 ) < diatoms (∼11,000 km 2 ). We note that community scales represent peaks on gradients rather than definitive one-size-fits-all scales. Spatial scaling of β was sensitive to seasonality (birds) and consistent among years for both birds and trees. Also, β deviation from a null model based on spatial heterogeneity adjusted observed β but was less sensitive to neighborhood size and species richness than β deviation based on spatial homogeneity. Main conclusions Results here indicate that: (a) similar patterns should occur across the tree of life; (b) local ecological and evolutionary forces scale up to form repeatable regional community patterns in ways not yet fully understood; (c) local biodiversity conservation efforts need to be coordinated at biogeographical scales to best achieve goals; and (d) a recent method to calculate β deviation from a null model based on spatial heterogeneity improves β research.