Patterns of biodiversity affect soil properties at different scales, conversely, soil characteristics and landscape features influence biodiversity. It is important to determine these relationships for understanding ecosystem processes. Many studies have carried out during the last few years mainly concentrated on factors that influence plant diversity in grassland or shrubland. Focused on the topography and forest heterogeneity for a warm temperate-zone deciduous broad-leaved forest in the Donglingshan Mountains near Beijing, detailed plant diversity, topography and soil features of 76 plots were investigated in a small watershed. To discern the complex relationships, multivariate statistical analysis techniques (Principal Component Analysis (PCA), Cluster Analysis (CA) and Canonical Correlation Analysis (CCA)) were employed. The results of PCA and CA showed that soil organic matter (SOM) is an important indicator to soil fertility. The coverage, richness and α-diversity index of three layers of plants (tree, shrub and herb) have unique features under different soil fertilities. High fertility plots often exist on south-facing slopes, in upper slope positions, and have gentle slope gradients. The coverage, richness and α-diversity index (Shannon index) of tree and shrub layers are the highest in mid-fertility plots, which have the highest available phosphorus (AP) and potassium (AK) contents, but those same summary descriptors for herbs are the least. CCA analysis elucidated the relationships of three different index groups (topography, soil and plant). Elevation and aspect have a close relationship with shrub richness and α-diversity. Elevation is also an important factor influencing SOM. SOM and total nitrogen have the greatest effect on plant characteristics (mainly shrub coverage) among all soil factors.