For the RNA-world hypothesis to be ecologically feasible, selection mechanisms acting on replicator communities need to be invoked and the corresponding scenarios of molecular evolution specified. Complementing our previous models of chemical evolution on mineral surfaces, in which selection was the consequence of the limited mobility of macromolecules attached to the surface, here we offer an alternative realization of prebiotic grouplevel selection: the physical encapsulation of local replicator communities into the pores of the mineral substrate. Based on cellular automaton simulations we argue that the effect of group selection in a mineral honeycomb could have been efficient enough to keep prebiotic ribozymes of different specificities and replication rates coexistent, and their metabolic cooperation protected from extensive molecular parasitism. We suggest that mutants of the mild parasites persistent in the metabolic system can acquire useful functions such as replicase activity or the production of membrane components, thus opening the way for the evolution of the first autonomous protocells on Earth.