The restoration and the preservation of aquatics ecosystems is a critical issue in our contemporary society. In lake ecosystems, phytoplankton taxonomic-based indicators have been developed to evaluate water quality, but suffer of limited ecosystem ecological value. The recent development of functional approaches may allow to evaluate other aspects of ecosystem quality, and to develop new trait-based indicators responding to different environmental conditions. Here, our aim was to analyze the response of phytoplankton traits to numerous environmental variables and to identify relevant traits for the development of future indicator metrics. We used a French national database of 469 lakes, consisting in phytoplankton biovolumes and physicochemical values. The response of 84 morpho-functional traits towards environmental variables was tested with Machine Learning models, taking into account lake typology. We identified 21 traits significantly related to environmental variables. Dissolved Organic Carbon, Nitrates and Total Suspend Solids were the physiochemical parameters which had the higher influence on our traits selection. However the response of phytoplanktonic traits to environmental variables did not change according to lake typology, advocating for a consistent response at the whole national scale contrary to the classical taxonomic approach. We finally identified several candidate traits that could be used for the development of new metrics for French lakes in a context of bio-assessment programs.