Successional chronosequences provide a unique opportunity to study the effects of multiple ecological processes on plant community assembly. Using a series of 0.5 × 0.5 m2 plots (n = 30) from five successional sub-alpine meadow plant communities (ages 3, 5, 9, 12, and undisturbed) in the Qinghai-Tibetan Plateau, we investigated whether community assembly is stochastic or deterministic for species and functional traits. We tested directional change in species composition, functional trait composition, and then functional trait diversity measured by Rao's quadratic entropy for four traits – plant height, leaf dry matter content, specific leaf area, and seed mass – along two comparable successional chronosequences. We then evaluated the importance of species interactions, habitat filtering and stochasticity by comparing with random communities and partitioning the environmental and spatial components of Rao's quadratic entropy. We found no directional change in species composition, but clear directionality in functional trait composition. None of the abiotic environmental variables (except P) showed linear change with successional age, but soil moisture and nitrogen were positively related to functional diversity within meadows. Functional trait diversity increased significantly with the increase in successional age. Comparison with random communities showed a significant shift from trait divergence in early stages of succession (3- and 5-yr) to convergence in the later stages of succession 9-, 12-yr and undisturbed). The relative importance of abiotic variables and spatial structure for functional trait diversity changed in a predictable manner with successional age. Stochasticity at the species level may indicate dispersal limitation, but deterministic effects on functional trait distributions show the role of both habitat effects and biotic interactions.