Anticipating invasion of parasite with graph theory: Dryocosmus kuriphilus, a threat for Castanea sativa

The increase in world trade contributed to a rise in the number of insect biological invasions. The colonisation success of herbivorous pests depends mostly on the abundance of vegetal resource across the newly invaded territory. The explicit spatial representation of plant distribution offers a cost-effective approach to anticipate the spread of herbivorous insects over areas with fragmented and heterogeneous vegetal populations. In this article, we focus on the case of the recent invasion of chestnut trees (Castanea sativa) of the French Eastern Pyrenees by its most virulent pest, the gall-forming parasite Dryocosmus kuriphilus. Using tools from graph theory and available public data on the mesh size and distribution of chestnut trees, we model the spatial distribution of the pest resource across natural forests. In this framework, a graph provides a mapping of 1 km2 quadrats that constitute the territory. Quadrats are grouped to form communities, which are areas with homogeneous chestnut tree distribution, and in which the risk of parasite infestation and propagation is similar. Graph traversal algorithms then measure the vulnerability and the dangerousness of each patch, defined as their susceptibility to become infected and to contribute to the pest propagation. Such a spatial assessment of the invasion risk provides unique insights into potential propagation scenarios, improving the early detection and spread monitoring of these insect invaders. Author summary