Resilience through diversity: The potential of modelling species and variety interactions to enhance resilience of production systems

Societal Impact Statement Agricultural production systems in the global North combine monocultures of specialised varieties and breeds with external interventions and inputs. Increasing the diversity of varieties, breeds and species may increase the system's resilience to external pressures through beneficial interactions. However, more diverse agricultural systems actually show large variation in their response to stresses. We propose to use functional‐structural modelling, alternating with field experiments, to optimise the design of biodiverse systems and validate predicted interactions under stresses. This information can then be used to design biodiverse production systems that are more resilient and guide crop variety and livestock breeding for them. Summary Many arable, horticultural, livestock and forestry production systems are optimised for productivity in monocultures, with environmental factors managed primarily through external interventions and inputs (fertilizers, herbicides, pesticides, antibiotics, irrigation). Enhancing system resilience may be achieved by (re)introducing diversity within and between species and breeds of crops, livestock and trees. While the ecological concepts underlying the effect of biodiversity on resilience are known, in practice the response of more biodiverse agricultural production systems to external pressures shows large variation. Hence, resilient production systems do not emerge automatically from greater diversity but must be designed. Studying the specific combining advantages of different species/varieties/breeds at the production system level may provide the required information, but the number of possible combinations and interactions is too large to screen with controlled field experiments only. We argue that modelling provides a solution to this combinatorial problem by allowing for an in silico exhaustive search of possible interactions. We describe various crop models and discuss the use of functional‐structural models for (re‐)designing production systems, specifying the functional traits to be selected for in crop variety and livestock breeding. While process‐based functional‐structural models can predict how genotypes will respond to environmental stressors, experimental trials measure these responses in the field. An iterative process– where models inform experiments and experiments, in turn, refine models– may lead to a nuanced understanding of resilience mechanisms and a robust set of tools for designing diversified production systems. We discuss opportunities and pitfalls of this combined approach.