Using metacommunity theory to understand the mechanisms shaping community structure is a promising framework that has been widely applied to ecological research. In lakes, the spatial pattern of phytoplankton assemblages depends on the relative importance of environmental conditions, spatial processes, and biotic interactions (e.g., grazing pressure), but the inclusion of the latter two factors was often overlooked. We tested how these three factors contributed to phytoplankton community composition in a shallow lake by separating the responses of taxonomic and trait compositions (i.e., nine species traits groups) of phytoplankton in Lake Changhu, China. Our results indicated that the taxonomic composition of phytoplankton assemblages in Lake Changhu are mainly determined by environmental factors (7.6 +/- 1.3%), followed by spatial processes (4.7 +/- 1.0%) and grazing pressure (2.9 +/- 0.5%). However, for the nine species traits groups, relative influences of environmental, spatial and grazing factors were trait specific, suggesting that different mechanisms were responsible for community composition supporting the potential advantages of using traits in water quality assessment. More specifically, some traits (e.g., large cell size and filamentous) may be excellent candidates for biomonitoring in lakes as they are predominantly driven by environmental factors (12.4% and 17.2% for large cell size and filamentous respectively), while other traits (e.g., unicellular and non-motile) are controlled largely by spatial processes or grazing and may not be suitable as bio-indicators. We also advocate inclusion of biotic factors (e.g., grazing pressure) in community studies, since we have found relatively weak but unneglectable effects of grazing on structuring phytoplankton community (2.9 +/- 0.5% for taxonomic composition while 3.1 +/- 4.1% for trait composition). In general, our findings suggest that a combination of metacommunity theory and the use of traits provide a useful framework for assessing driving factors structuring phytoplankton community in lakes, and such framework can be very useful for future lake bioassessment and management efforts.