Background and aimsThe importance of niche processes and neutral processes to community assembly has been affirmed by most studies, although their relative importance needs to be determined in many systems. Moreover, as the spatial scale changes, the ecological processes that determine the community pattern may differ. We tested these ideas in subtropical karst forest in southwestern China in order to aid efforts of community reconstruction.MethodsTo test the importance of niche-based and neutral mechanisms we compared the fit six models to the observed SAD of the plot at three different sampling scales (10 m × 10 m, 20 m × 20 m, 50 m × 50 m). We also used spatial autocorrelation and distance-based Moran's eigenvector maps (dbMEM) combined with variation partitioning to further determine the relative contribution of the niche process and the neutral process under different sampling scales.ResultsThe neutral theoretical and statistical models fit the observed species abundance distribution curve best at each sampling scale. And variation partitioning showed that although the contribution of spatial structure was lower at larger sampling scales, it was still important, suggesting that neutral processes drive community structure at all sampling scales. In contrast, habitat filtering and interspecies competition may lead to a net weakening of the contribution of the niche process to the species abundance pattern of the community because they act in opposite directions. ConclusionsIn the restoration and reconstruction of local karst forest communities, environmental heterogeneity, inter-species relationships, and geographic spatial differences should all be considered.