ABSTRACT: Understanding how environmental change alters the composition of plant assemblages is a major challenge in the face of global climate change. Researches accounting for site-specific trait values within forest communities help bridge plant economics theory and functional biogeography to better evaluate and predict relationships between environment and ecosystem functioning. Here, by measuring six functional traits (specific leaf area, leaf dry matter content, leaf nitrogen, and phosphorus concentration, leaf nitrogen/phosphorus, wood density) for 292 woody plant species (48,680 individuals) from 250 established permanent forest dynamics plots in five locations across the subtropical evergreen broadleaved forests (SEBLF) in China, we quantified functional compositions of communities by calculating four trait moments, i.e., community-weighted mean, variance, skewness, and kurtosis. The geographical (latitudinal, longitudinal, and elevational) patterns of functional trait moments and their environmental drivers were examined. Results showed that functional trait moments shifted significantly along the geographical gradients, and trait moments varied in different ways across different gradients. Plants generally showed coordinated trait shifts toward more conservative growth strategies (lower specific leaf area, leaf N and P concentration while higher leaf nitrogen/phosphorus and wood density) along increasing latitude and longitude. However, trends opposite to the latitudinal and longitudinal patterns appeared in trait mean values along elevation. The three sets of environmental variables (climate, soil and topography) explained 35.0-69.0%, 21.0-56.0%, 14.0-31.0%, and 16.0-30.0% of the variations in mean, variance, skewness, and kurtosis across the six functional traits, respectively. Patterns of shifts in functional trait moments along geographical gradients in the subtropical region were mainly determined by the joint effects of climatic and edaphic conditions. Climate regimes, especially climate variability, were the strongest driving force, followed by soil nutrients, while topography played the least role. Moreover, the relationship of variance, skewness and kurtosis with climate and their geographical patterns suggested that rare phenotypes at edges of trait space were selected in harsher environments. Our study suggested that environmental filtering (especially climate variability) was the dominant process of functional assembly for forest communities in the subtropical region along geographical gradients.