Population structure and spatial patterns are important characteristics of plant communities and they can reveal ecological characteristics of the species. This study was carried out in the evergreen broad-leaved forest of Tiantong (29°48′N， 121°47′E) with Castanopsis carlesii and Schima superba as the most important dominant species. The floristic composition and community structure were investigated using quadrat method and every-tree measuring method, and then population structure, spatial distribution and spatial association of individuals in different layers for 10 main tree species were analyzed.
The total relative basic area(RBA) of the 10 species, including the 5 dominant species of the community, was over 81.9%. The size distribution index(SDI) was negative and relative low except for Symplocos heishanensis which had a positive value. The species occupied different vertical space, however, C.fargesii which distributed in its northern boundary lacked the sub-tree layer to some extent, while Lithocarpus harlandii, Cyclobalanopsis gracilis and C. nubium whose most suitable sites located in the mid-mountains showed obvious stunting characters in the studied area.
The spatial distribution varied among intra- and inter-specific cohorts in the different layers and showed complex multilevel distribution patterns, but most species had an aggregated distribution or had colony as the basic units considering the total number of individuals. There were differences in spatial associations among intra- and inter-specific cohorts in the different layers and spatial scales examined. Distribution patterns of Castanopsis carlesii and Schima superba indicated that they were more adaptive in the whole section, while C. fargesii was distributed in limited habitats. The spatial associations indicated that C. carlesii and S. superba were strong-positively associated, but they were strong-negatively associated with C. fargesii at the scale less than 100m2. The positive associations among cohorts increased and the negative associations decreased as the scales increasing. When the scales analyzed were 25m2, 100m2 and 400m2, the number of pairs with strongly positive associations were 5, 26 and 60, respectively, while the number of pairs with strongly negative associations were 110, 73 and 47. The result implied that the critical factors forming the observed spatial associations varied with the spatial scales. By occupying different horizontal and vertical space and having different life history strategies, the main species of the community could exist, resulting in rich biodiversity in the studied community.