Cover crop-based cropping systems have gained considerable attention in recent years because of the recognized benefits of increasing agroecosystem diversity to enhance ecological functions and agroecosystem services through positive effects between species. While agronomic tools and methods to assess these services in multispecies plant communities barely exist, trait-based approaches, originally developed in the field of comparative functional ecology, provide an appropriate framework to study the effects of plant diversity on agroecosystems. We present our theoretical and applied research conducted over recent years using a trait-based approach to help improve cover crop choice to sustain agroecosystem services in banana agroecosystems of the French West Indies. We first provide a conceptual framework for assessing expected agroecosystem services from cover crops in banana agroecosystems, i.e., weed control, plant-parasitic nematode control, improvement of the nutrient cycling and avoidance of interference for resource acquisition in banana (dis-service), based on plant effect traits. Then, we experimentally characterized a collection of cover crop species to identify their resource acquisition strategies and the potential trade-offs they reveal. More practically, we characterized the cover crop species according to their potential to deliver four agroecosystem services expected from banana/cover crop systems. We constructed functional profiles of the cover crop species, and compared these profiles to cover crop ideotypes to select the best-suited species for each service. Finally, we used "engineering traits" to help design and optimize design and management of cover crop species mixtures. These examples from our research demonstrate that functional, trait-based approaches have strong potential to tackle agronomic issues and to resolve fundamental and applied questions related to the design of multispecies cropping systems that optimize multiple services.