The relationship between species diversity and ecosystem functioning is a central topic in ecology today. Classical approaches to studying ecosystem responses to nutrient enrichment have considered linear food chains. To what extent ecosystem structure, that is, the network of species interactions, affects such responses is currently unknown. This severely limits our ability to predict which species or functional groups will benefit or suffer from nutrient enrichment and to understand the underlying mechanisms. Here our approach takes ecosystem complexity into account by considering functional diversity at each trophic level. We conducted a mesocosm experiment to test the effects of nutrient enrichment in a lake ecosystem. We developed a model of intermediate complexity, which separates trophic levels into functional groups according to size and diet. This model successfully predicted the experimental results, whereas linear food-chain models did not. Our model shows the importance of functional diversity and indirect interactions in the response of ecosystems to perturbations, and indicates that new approaches are needed for the management of freshwater ecosystems subject to eutrophication.