The management of lakes requires the definition of physico-chemical thresholds to be used for ecosystem preservation or restoration. According to the European Water Framework Directive, the limits between physico-chemical quality classes must be set consistently with biological quality elements. One way to do this consists in analyzing the response of aquatic communities to environmental gradients across monitoring sites and in identifying ecological community thresholds, i.e. zones in the gradients where the species turnover is the highest. In this study, fish data from 196 lakes in France were considered to derive ecological thresholds using the multivariate method of gradient forest. The analysis was performed on 25 species and 36 environmental parameters. The results revealed the highest importance of maximal water temperature in the distribution of fish species. Other important parameters included geographical factors, dissolved organic carbon concentration and water transparency, while nutrients appeared to have low influence. In spite of the diversity of species responses to the gradients, community thresholds were detected in the gradients of the most important physico-chemical parameters and of total phosphorus and nitrate concentrations as well. The thresholds identified in such macroecological study may highlight new patterns of species natural distribution and improve niche characterization. Moreover, when factors that may be influenced by human activities are involved, the thresholds could be used to set environmental standards for lake preservation.