AbstractPhytoplankton communities are increasingly subject to multiple stressors of natural or anthropogenic origin. The cumulative effect of these stressors, however, may vary considerably from the sum of impacts from individual stressors. Nonlinear effects, such as changes in community traits can either boost up (synergistic) or weaken (antagonistic) single stressors. Despite previous empirical studies and meta analyses on the interaction types of various multiple stressors, a more fundamental understanding of cumulative effects is lacking. To fill this gap, we here propose a new theoretical framework that is centered on the concept of interaction traits and their trade-offs. The framework is applied to a novel size-based plankton model resolving multi-species phytoplankton-nutrients-detritus-zooplankton dynamics within the upper mixed layer. The model is validated using data from a series of outdoor mesocosm experiments. In the direct aftermath of single perturbations that increase net growth rate, here nutrient enrichment and grazer removal, the simulated phytoplankton community undergoes structural changes as visible in altered community traits. These temporal variations explain why the multiple stressor interaction switches from antagonistic to synergistic as compensatory trait variations reduce over time of the experiment. This finding can be generalized within our trait-based explanatory framework to mechanistically assess and predict effects of other stressor combinations and for other organism groups.