www.biodiversity-science.net 研究报告 indirectly. Therefore, it is a critical issue of ecology. The aim of this study is to quantitatively explore the relationship between tree diversity and biomass or productivity of subtropical natural forest ecosystems in late successional stages. Methods: Based on long-term community surveys of topographies and soil nutrients from permanent forest ecosystems in the subtropical evergreen broad-leaved forest, we analyzed the relationship between tree diversity and productivity/biomass and their influencing factors. In this analysis, the correlations between tree diversity and each factor were evaluated using a Pearson correlation analysis. The single and shared effects of each factor were quantified by variance partitioning analysis (VPA). In addition, the relationships between soil nutrients and topographies and their effects on productivity and biomass were further evaluated, either directly or indirectly, through species and structural diversity by using a structural equation model (SEM). Result: Species diversity displayed a negative correlation with biomass and a positive correlation with productivity. Contrarily, structural diversity was positively correlated with biomass and negatively correlated with productivity. The effects of environment factors on tree diversity and biomass/productivity were varied. Specifically, soil moisture had a significant effect on productivity, species diversity was correlated with soil nutrients and topographies, and structural diversity was strongly related to soil nutrients. The variance partitioning analysis results indicated that the single effect of structural diversity explained the largest portion of variance in biomass (35.39%) and productivity (5.21%), followed by the shared effect of structural and species diversity on biomass (13.66%) and productivity (3.53%). Soil nutrients and topographies explained less variation in productivity and biomass. The structural equation results analysis indicated that structural diversity had a direct positive effect on biomass, and biomass had a strong direct negative effect on productivity. Structural diversity indirectly reduced productivity by increasing biomass. Soil nutrients and topographies were mainly affected by biomass, while productivity indirectly affected tree species and structural diversity. Conclusions: These results indicate that (1) species diversity had the greatest direct effect on productivity, and structural diversity had the greatest direct effect on biomass in southern subtropical evergreen broad-leaved forests; (2) species diversity explained productivity better than structural diversity, while structural diversity explained biomass accumulation better than species diversity; and (3) both species diversity and structural diversity can be affected by soil nutrients and topographies.