Decision support for managing an invasive pathogen through efficient clean seed systems: Cassava mosaic disease in Southeast Asia

CONTEXT Effective seed systems must both distribute high-performing varieties efficiently and slow or stop the spread of pathogens and pests. Epidemics increasingly threaten crops around the world, endangering the incomes and livelihoods of smallholder farmers. Responding to these food and economic security challenges requires stakeholders to act quickly and decisively during the early stages of invasions, typically with very limited resources. The recent introduction of cassava mosaic virus into southeast Asia threatens cassava production in the region. OBJECTIVES Our goal in this study is to provide a decision-support framework for efficient management of healthy seed systems, applied to cassava mosaic disease. The specific objectives are to (1) evaluate disease risk in disease-free parts of Cambodia, Lao PDR, Thailand, and Vietnam by integrating disease occurrence, climate, topology, and land use, using machine learning; (2) incorporate this predicted environmental risk with seed exchange survey data and whitefly spread in the landscape to model epidemic spread in a network meta-population model; and (3) use scenario analysis to identify candidate regions to be prioritized in seed system management. RESULTS AND CONCLUSIONS The analyses allow stakeholders to evaluate strategy options for allocating their resources in the field, guiding the implementation of seed system programs and responses. Fixed rather than adaptive deployment of clean seed produced more favorable outcomes in this model, as did prioritization of a higher number of districts through the deployment of smaller volumes of clean seed. SIGNIFICANCE The decision-support framework presented here can be applied widely to seed systems challenged by the dual goals of distributing seed efficiently and reducing disease risk. Data-driven approaches support evidence-based identification of optimized surveillance and mitigation areas in an iterative fashion, providing guidance early in an epidemic, and revising them as data accrue over time.