Increasing threats to freshwater biodiversity from environmental changes and human activities highlight the need to understand the linkages between biological communities and their environment. Species richness has dominated our view of biodiversity patterns for over a century, but it is increasingly recognized that a traitbased, causal view of biodiversity may be more meaningful than species richness or taxonomic composition. This rationale has led to the exploration of functional diversity (FD) indices to quantify variation in traits that mediate species' contributions to ecosystem processes. In the present study, we quantified FD of fish communities in two large shallow lakes in China with different disturbances level using long-term monitoring data sets. Random-Forests regression was applied to examine how changes in FD were related to natural and human-related environmental variables. Fish stocking, water quality, climate, and hydrological changes were identified as the most important predictors of FD long-term trends. However, the major drivers of FD differed between two lakes, i.e., human activities explaining a greater proportion of FD variance in Lake Taihu, whereas physicochemical environmental factors prominently explained FD variance in Lake Hulun. Moreover, FD indices appeared more sensitive than species richness to multiple disturbances, suggesting that functional traits can be used to detect ecosystem alterations. This study offers insight into how FD can improve our understanding of the associations between fish communities and environmental changes of relevance also for lake and fisheries management. (C) 2020 Elsevier B.V. All rights reserved.