Description: Aim: To develop and test theory based on connectivity to identify optimal networks of marine protected areas (MPAs) that protect multiple species with a range of dispersal strategies. Location: The eastern North Sea in the Atlantic Ocean. Methods: Theory of finding optimal MPA network is based on eigenvalue perturbation theory applied to population connectivity. Previous theory is here extended to the persistence of multiple species by solving a maximization problem with constraints, which identifies an optimal consensus network of MPAs. The theory is applied to two test cases within a 120,000 km2 area in the North Sea where connectivity was estimated with a biophysical model. In a realistic case, the theory is applied to the protection of rocky-reef habitats, where the biophysical model is parameterized with realistic dispersal traits for key species. Theoretical predictions of optimal networks were validated with a simple metapopulation model. Persistence of optimal consensus MPA networks is compared to randomly selected networks as well as to the existing MPA network. Results: Despite few overlapping MPA sites for the optimal networks based on single dispersal strategies, the consensus network for multiple dispersal strategies performed well for 3 of 4 contrasting strategies even without user-defined constraints. In the test with five realistic dispersal strategies, representing a community on threatened rocky reefs, the consensus network performed equally well compared to solutions for single species. Different dispersal strategies were also protected jointly across the MPA network (93% of sites), in contrast to simulations of the existing MPA network (2% of sites). Consensus networks based on connectivity were significantly more efficient compared to existing MPAs. Main conclusions: Our findings suggest that the new theoretic framework can identify a consensus MPA network that protects a whole community containing species with multiple dispersal strategies.