Dispersal limitation and environmental structure interact to restrict the occupation of optimal habitat.

Whether plant distributions are governed more by neutral-based distance effects or niche-based environmental responses remains elusive. A lack of habitat matching, where species distributions do not correspond to environmental variability, suggests neutrality but can also be explained by niche models through the interactions of dispersal limitation, spatial autocorrelation of the environment, species interactions, and spatial scale. We untangle these effects in a field study with multiscale statistical analyses. We demonstrate that despite significant niche-based environmental responses by a savanna plant, we still see weak habitat matching, with the mechanisms responsible differing by spatial scale. At the coarse scale (100-200 m), dispersal limitation restricted the occupation of optimal habitat. At the fine scale (<30 m), dispersal was not limiting, but a lack of autocorrelation of environmental variables prevented the aggregation of reproductively active plants in optimal microsites. Species associations were largely unimportant at all scales. Extending our analysis to the entire community revealed similar scale-dependent limitations of distance and the environment, indicating weak habitat matching for all species. This work supports predictions that environmental specializations do not necessarily produce deterministic distributions in plant communities. It also provides a mechanistic explanation for why co-occurring plant species can have largely undifferentiated distributions.