Fertilizers, pesticides and global warming are threatening freshwater aquatic ecosystems. Most of these are shallow ponds or slow-flowing streams or ditches dominated by submerged macrophytes, periphyton or phytoplankton. Regime shifts between the dominance of these primary producers can occur along a gradient of nutrient loading, possibly triggered by specific disturbances influencing their competitive interactions. However, phytoplankton dominance is less desirable due to lower biodiversity and poorer ecosystem function and services. In this study, we combined a microcosm experiment with a process-based model to test three hypotheses: 1) agricultural run-off (ARO), consisting of nitrate and a mixture of organic pesticides and copper, differentially affects primary producers and enhances the risk of regime shifts, 2) warming increases the risk of an ARO-induced regime shift to phytoplankton dominance and 3) custom-tailored process-based models support mechanistic understanding of experimental results through scenario comparison. Experimentally exposing primary producers to a gradient of nitrate and pesticides at 22°C and 26°C supported the first two hypotheses. ARO had direct negative effects on macrophytes, while phytoplankton gained from warming and indirect effects of ARO like a reduction in the competitive pressure exerted by other groups. We used the process-based model to test eight different scenarios. The best qualitative fit between modeled and observed responses was reached only when taking community adaptation and organism acclimation into account. Our results highlight the importance of considering such processes when attempting to predict the effects of multiple stressors on natural ecosystems.