Microcystis blooms are a global contemporary problem and the mechanisms underlying strain-level ecology (e.g. toxigenic fraction) and toxin (microcystin, MC) production are not sufficiently understood. Recent research suggests that MC synthesis depends on the availability of nitrogen and light, and that they protect toxigenic cells against damage by H2O2. The non-toxigenic strains employ the alternative strategy of enzymatic degradation of H2O2. Thus, MC-producing cells may have an advantage at high nitrogen and light availability. A model based on this mechanism was able to reproduce the observed patterns of toxigenic fraction and MC concentration in Lake Erie. However, it is unclear if this mechanism also applies to other systems. We investigated this by modeling nine different cases (i.e. lakes, stations, years). The model can reproduce observed patterns (toxigenic fraction, concentration of MC, biomass, nutrients and if available H2O2) for all cases, which constitutes support for the proposed mechanism. To explore lake management, we simulated single and dual nutrient (i.e. nitrogen and/or phosphorus) reduction, which predicts two types of outcomes. For lakes with nitrogen limitation at some time during the blooming season (e.g. Lake Erie) a phosphorus only reduction does not reduce MC concentration proportionally and may even increase it. Reducing phosphorus lowers biomass, which increases nitrogen and light availability and raises MC production and toxigenic fraction. For lakes with replete nitrogen (e.g. Lake Taihu) MC concentration is predicted to decrease. Here, further nitrogen availability will not increase MC production. These results advance mechanistic understanding of Microcystis strain ecology and toxin production and provide guidance for management.