Niche processes and other spatial processes, such as dispersal, may simultaneously control beta diversity, yet their relative importance may shift across spatial and temporal scales. Although disentangling the relative importance of these processes has been a continuing methodological challenge, recent developments in multi-scale spatial and temporal modeling can now help ecologists estimate their scale-specific contributions. Here we present a statistical approach to (1) detect the presence of a space–time interaction on community composition and (2) estimate the scale-specific importance of environmental and spatial factors on beta diversity. To illustrate the applicability of this approach, we use a case study from a temperate forest understory where tree seedling abundances were monitored during a 9-year period at 40 permanent plots. We found no significant space–time interaction on tree seedling composition, which means that the spatial abundance patterns did not vary over the study period. However, for a given year the relative importance of niche processes and other spatial processes was found to be scale-specific. Tree seedling abundances were primarily controlled by a broad-scale environmental gradient, but within the confines of this gradient the finer scale patchiness was largely due to other spatial processes. This case study illustrates that these two sets of processes are not mutually exclusive and can affect abundance patterns in a scale-dependent manner. More importantly, the use of our methodology for future empirical studies should help in the merging of niche and neutral perspectives on beta diversity, an obvious next step for community ecology.