The metacommunity concept has been widely used to explain the biodiversity patterns at various scales. It considers the influences of both local (e.g., environmental filtering and biotic interactions) and regional processes (e.g., dispersal limitation) in shaping community structures. Compared to environmental filtering and spatial processes, the influence of biotic interactions on biodiversity patterns in streams has received limited attention. We investigated the relative importance of three ecological processes, namely environmental filtering (including local environmental and geo-climatic factors), spatial processes and biotic interactions (represented by interactions of macroinvertebrates and diatom), in shaping different traits of macroinvertebrate communities in subtropical streams, Eastern China. We applied variance partitioning to uncover the pure and shared effects of different ecological processes in explaining community variation. The results showed that environmental filtering, spatial processes, and biotic interactions jointly determined taxonomic and trait compositions of stream macroinvertebrates. Spatial processes showed a stronger influence in shaping stream macroinvertebrate communities than environmental filtering. The contribution of biotic interactions to explain variables was, albeit significant, rather small, which was likely a result of insufficient representation (by diatom traits) of trophic interactions associated with macroinvertebrates. Moreover, the impact of three ecological processes on macroinvertebrate communities depends on different traits, especially in terms of environmental filtering and spatial processes. For example, spatial processes and environmental filtering have the strongest effect on strong dispersal ability groups; spatial processes have a greater effect on scrapers than other functional feeding groups. Overall, our results showed that the integration of metacommunity theory and functional traits provides a valuable framework for understanding the drivers of community structuring in streams, which will facilitate the development of effective bioassessment and management strategies.