A key challenge in research linking biodiversity and ecosystem functioning is to incorporate the trophic interactions that characterise natural systems. There is a particular shortage of studies investigating consumer species richness and composition (identity) effects in the context of ecosystem development (or succession). We manipulated the richness and composition of an assemblage of molluscan grazers (Patella ulyssiponensis, Gibbula umbilicalis and Littorina littorea) added to rock pools denuded of existing biota. We created monocultures and all possible multispecies combinations in a substitutive design, and ran a field experiment for 13 mo. We used 2 separate nested analyses to isolate the roles of species richness, species composition nested within levels of species richness and the specific effect of the limpet P. ulyssiponensis, a putative key species. We found no evidence that the biomass or productivity of the developing macroalgal assemblage was affected by grazer richness or species composition nested within richness levels. Rather, the presence of P. ulyssiponensis, irrespective of the presence of other grazer species, acted to suppress mean values of these response variables. Biomass and productivity were not strongly related, showing that they provide unique information on ecosystem functioning in this system. Macroalgal species richness was also reduced by P. ulyssiponensis, and correlated positively with macroalgal biomass, indicating a link between these response variables. Macroalgal species composition was largely insensitive to either species richness or the presence of P. ulyssiponensis, but responded to particular combinations of species within levels of these factors. The key role of P. ulyssiponensis in determining ecosystem functioning is apparent from our results, but we note that consumer species richness may play an important role under more heterogeneous conditions.