Cyanobacterial species of the Microcystis aeruginosa complex (MAC) generate harmful blooms in multiple aquatic ecosystems, from freshwater to estuaries. The intracomplex genetic diversity and its phenotypic plasticity emerge as the main hypothesis of ecological success of the group in eutrophic ecosystems. This intra-complex diversity is composed by ecotypes, phylogenetically closely related organisms differing in their ecological preferences. Since MAC ecotype classification method is based on mcyJ genotyping, the different salinity-resistant ecotypes will display different patterns of the toxin microcystins (MC) variants. Under the hypothesis of a rapid selection of ecotypes caused by salinity, here, we experimentally evaluated the role of salinity (0, 5, 10 and 25) in the genotypic community configuration and toxicity of MAC. To test it, we combined qPCR (quantification of MC-producing cells and mcyE transcripts), highresolution melting analysis (genotyping), mcyJ amplicon sequencing (ecotypes identification), HPLC (MC variants concentration) and machine learning for data analysis. We found that abundance of toxic MAC cells, mcyE transcription and MC production were negatively affected by increasing salinity. Salinity above 5 selected MC-producing MAC ecotypes which showed different MC variant profiles when exposed to different salinities. We suggest that when MAC species are transported from a freshwater ecosystem to brackish water, they face a salt-induced stress (salinities between 5 and 10) with a positive selection of MC production. At 25 salinities, the selected ecotypes maintain their abundance but with decreased toxin production. This compositional change implies a large health risk in estuarine regions that deserves further research.