Hundreds of experiments conducted over the last decades demonstrate a positive relationship between species diversity and ecosystem functions. Following good experimental practice, most of these studies have manipulated species richness artificially by assembling communities randomly. Yet, natural communities along ecological gradients often show niche-based responses to selection gradients and species extinction order is generally not a random process. Such responses are commonly trait-mediated and the effects of communities on ecosystems’ functions also depend on species traits. In an effort to disentangle the relationship of trait mean value and diversity with ecosystem functioning, we revisited a community assembly mesocosm experiment that simulated habitat heterogeneity and a typical gradient of productivity to test how body size diversity and composition of cladocerans responded to such gradients and whether and how such trait responses impacted top-down control of unicellular algae, a key ecosystem function in aquatic systems. Nutrient addition and habitat heterogeneity lead to an increase in community average size (CAS), which ultimately resulted in an increased zooplankton resource use efficiency (RUE_ZP). This increase in CAS acted as a buffer for ecosystems functions despite the decline in species richness under high phosphorus levels. Habitat heterogeneity also increased size diversity (SD) and species richness (S) and modified CAS and SD responses along the nutrient gradient. CAS proved to be more powerful in explaining variation in RUE_ZP than SD and S. Our results suggest that exploring the response of the mean value of a key trait, such as zooplankton body size, to eutrophication might offer a much better understanding of variations in the top-down control of algae compared to relying solely on taxonomic and functional diversity metrics.