Given a set of potential species and a replicator dynamic model of their interaction, the community assembly problem seeks the maximal set of species that can co-exist indefinitely without extinction. In this study we compare a standard model, which assembles a community one species at a time, with an evolutionary algorithm that selects sets of species directly. The comparison is performed using a standard competition model. The system is tested with three different available species pools of one hundred species. The diver- sity of communities located with the evolutionary algorithm substantially exceeds that of those located by serial addition of single species. In agreement with past research, the serial species addition algorithm located communities that, while not the largest, were highly resistant to invasion by a single additional species. A comparison of the diversity between the communities located by the two algorithms demonstrated that the evolutionary algorithm located a very much larger variety of community types. For all three species pools, the communities found in different runs of the serial species addition algorithm shared large common cores of species.