A central challenge in community ecology is to understand the connection between biodiversity and the functioning of ecosystems. While traditional approaches have largely focused on species-level diversity, increasing evidence indicates that there exists substantial ecological diversity among individuals within species. By far, the largest source of this intraspecific diversity stems from variation among individuals in ontogenetic stage and size. Although such ontogenetic shifts are ubiquitous in natural communities, whether and how they scale up to influence the structure and functioning of complex ecosystems is largely unknown. Here we take an experimental approach to examine the consequences of ontogenetic niche shifts for the structure of communities and ecosystem processes. In particular we experimentally manipulated the stage structure in a keystone predator, larvae of the dragonfly Anax junius, in complex experimental pond communities to test whether changes in the population stage or size structure of a keystone species scale up to alter community structure and ecosystem processes, and how functional differences scale with relative differences in size among stages. We found that the functional role of A. junius was stage-specific. Altering what stages were present in a pond led to concurrent changes in community structure, primary producer biomass (periphyton and phytoplankton), and ultimately altered ecosystem processes (respiration and net primary productivity), indicating a strong, but stage-specific, trophic cascade. Interestingly, the stage-specific effects did not simply scale with size or biomass of the predator, but instead indicated clear ontogenetic niche shifts in ecological interactions. Thus, functional differences among stages within a keystone species scaled up to alter the functioning of entire ecosystems. Therefore, our results indicate that the classical approach of assuming an average functional role of a species can be misleading because functional roles are dynamic and will change with shifts in the stage structure of the species. In general this emphasizes the importance of accounting for functional diversity below the species level to predict how natural and anthropogenic changes alter the functioning of natural ecosystems.