Early-detection surveillance for stem rust of wheat: insights from a global epidemic network based on airborne connectivity and host phenology

Abstract

Stem rust of wheat, caused by the airborne pathogen Puccinia graminis, is a re-emerging crop disease representing a major concern to global food security. Potential long-distance transport by wind over a worldwide distributed host represents a challenge to effective surveillance and control of this disease. To monitor this disease, we have created a global epidemic network for stem rust of wheat combining (a) Lagrangian simulations of air-mass trajectories computed with the NOAA's HYSPLIT model; (b) land use from the Map Spatial Production Allocation Model and (c) meteorological and environmental conditions that are known to affect bio-physical processes involved in the biology of P. graminis spores. Our findings are in agreement with the well known north-American 'Puccinia pathway' and suggest the existence of other sub-continental pathways at the global scale. We used network theory to conceive surveillance strategies aimed at early detection of outbreaks while minimizing the number of nodes to be surveilled (also referred to as sentinels). We found that the set cover algorithm, due the high average connectivity of the network (density = 0.4%), performs better than a number of other network metrics and permits us to identify an optimal sentinel set (1% of the network nodes) to surveil 50% of the network. Our results also show that effective surveillance plans for stem rust of wheat can be designed, but that they need to account for the actual geographical scale of the underlying epidemiological process and call for an international and trans-boundary approach.