Niche and neutral theories emphasize different processes contributing to the maintenance of species diversity. In this study, we calculated the local contribution to beta diversity (LCBD) of every cell, using variation partitioning in combination with spatial distance and environmental variables of the 25-ha Badagongshan plot (BDGS), to determine the contribution of environmentally-related variation versus pure spatial variation. We used topography and soil characteristics as environmental variables, distance-based Moran’s eigenvectors maps (dbMEM) to describe spatial relationships among cells and redundancy analysis (RDA) to apportion the variation in beta diversity into three components: pure environmental, spatially-structured environmental and pure spatial. Results showed LCBD values were negatively related to number of common species and positively related to number of rare species. Environment and space jointly explained ~60% of the variation in species composition; soil variables alone explained 21.6%, slightly more than the topographic variables that explained 15.7%; topography and soil together explained 27%, slightly inferior to spatial variables that explained 34%. The BDGS forest was controlled both by the spatial and environmental variables and the results were consistent across different life forms and life stages.

• This study aimed to reveal species–genetic diversity correlations (SGDCs) and their underlying mechanisms in natural and disturbed forests. • A community survey and molecular analyses were carried out to compare species diversity (SD), the genetic diversity of the dominant tree species Euptelea pleiospermum (GD), the altitudinal patterns of SD and GD, SGDC, genetic differentiation (F_ST), community divergence (F_ST-C), effective population size (N_e), and recent migration rate between mountain riparian forests along the Yandu (natural) and Nan (disturbed) rivers. • In natural forests, both SD and GD showed a unimodal altitudinal pattern and GD was positively correlated with SD, whereas a unimodal pattern and positive SGDC were not found in the disturbed forests. SD and F_ST at the natural sites were higher than those at the disturbed sites. However, there were no significant differences in GD, F_ST-C, N_e or recent migration rate between the natural and disturbed sites. • A correlation between the patterns of SD and GD along a geographical gradient (e.g. altitude) is an important driver of positive SGDC. The absence of positive SGDC in the disturbed forests may result from reduced SD but unaffected GD, indicating nonparallel changes in SD and GD. This study furthermore cautions against generalizations about changes in SD and GD following disturbance.