Abstract:A monitoring sample site with 2 hm2 of karst mixed evergreen and deciduous broadleaf forest in Mulun National Nature Reserve was established according to the standard of CTFS (i.e., the Center of Tropical Forest Sciences) by the Institute of Subtropical Agriculture. 22 indexes covering plant community, soil property, and topography were chosen to study the relationships between the three factors through the analysis of traditional statistics, principal component analysis (PCA), cluster analysis and canonical correlation analysis (CCA). The results indicated that the Mulun karst forest had high landscape heterogeneity, high content of soil nutrients, abundant woody species, and rational community structure. Variations were observed based on the indexes except elevation, Simpson index, evenness, and pH. Soil indexes were the primary factor to the ecosystem, plant community diversity and structure indexes were the second, topographical indexes were the last. Based on the differences from the community types, soil fertility and topography, the forest was divided into four groups through the cluster analysis. The third group with best optimized characteristics was mostly distributed in the middle and upper part of shady slope with larger percentage of exposed rock, larger slope degree and greater elevation. High correlation existed between vegetation, soil and topography. Between vegetation and soil factors, organic matter, nitrogen and phosphorus primarily affected plant community structure. Between vegetation and topography factors, slope direction and percentage of exposed rock were the major factors affected plant community structure and diversity. Between soil and topography factors, slope direction and position directly resulted in the variance of soil organic matter, total nitrogen, available phosphorus, and pH.