Loss of biodiversity may cause significant changes to ecosystem structure and functioning. Evidence from long-term in situ removal experiments is rare but important in determining the effects of biodiversity loss against a background of environmental variation. Limpets and mussels are thought to be important in controlling community structure on wave-exposed shores in the UK: limpets as key grazers, mussels as ecosystem engineers. A long-term factorial removal experiment revealed interactive effects that varied between 2 shores in SW England. At one site (Harlyn), removing limpets caused a significant shift in community structure, but where limpets were lost, the presence or absence of mussels made little difference. Where limpets were present, however, the removal of mussels changed the structure and variability of the community. At the other site (Polzeath), the loss of mussels caused significant changes in community structure, and limpets played a less important role. At Harlyn, fucoid algae were abundant throughout the year. There were fewer algae at Polzeath, and cover was dominated by the summer bloom of ephemerals. At Harlyn, the limpets played a major role in controlling algae, but their effects were mediated by the presence of mussels. Other grazers were not able to fulfil their role. At Polzeath, mussels were far more important, and ephemeral algae grew on them regardless of the presence or loss of limpets. These findings emphasise the need to assess spatial and temporal variation in the effects of biodiversity loss and the importance of interactive effects of loss of multiple species from different functional groups.