Growing crops in more diverse crop systems (i.e. intercropping) is one way to produce food more sustainably. Even though intercropping, compared to average monocultures, is generally more productive, the full yield potential of intercropping might not yet have been achieved as modern crop cultivars are bred to be grown in monoculture. Breeding plants for more familiarity in mixtures, i.e. plants that are adapted to more diverse communities (i.e. adaptation) or even to coexist with each other (i.e. coadaptation) might have the potential to sustainably enhance productivity. In this study, the productivity benefits of familiarity through evolutionary adaptation, where one species adapts to its neighbourhood, and coevolutionary coadaptation, where two or more species adapt to each other, were disentangled in a crop system through an extensive common garden experiment. Furthermore, evolutionary and coevolutionary effects on species-level and community-level productivity were linked to corresponding changes in functional traits. We found evidence for higher productivity and trait convergence with increasing familiarity of the plants composing the community. Furthermore, our results provide evidence for coevolution of plants in mixtures leading to higher productivity of coadapted species. However, with the functional traits measured in our study we could not fully explain the productivity benefits found upon coevolution. Our study is, to our knowledge, the first study that investigated coevolution among randomly interacting plants and was able to demonstrate that coadaptation through coevolution of coexisting species in mixtures promote ecosystem functioning (i.e. higher productivity). This result is particularly relevant for the diversification of agricultural and forest ecosystems, demonstrating the added value of artificially selecting plants for the communities they are familiar with.