Optimal allocation of invasive species surveillance with the maximum expected coverage concept

AIM: We address the problem of geographically allocating scarce survey resources to detect pests in their pathways of introduction given information about their likelihood of movement between origins and destinations. We introduce a model for selecting destination sites for survey that departs from the aim of reducing propagule pressure (PP) in pest destinations and instead aims to increase monitoring of pest origins. The model is a maximum expected coverage problem (MECP), which maximizes the expected number of origins that are covered by the survey system, where an origin is covered if at least one of its transmission pathways connects to a surveyed destination. For comparison, we present two models that aim to reduce PP in destination sites. One model maximizes the expected number of transmission pathways that are covered by survey locations and the other maximizes the expected number of survey locations that have one or more pest introductions. LOCATION: United States, Canada. METHODS: We demonstrate the models by analysing the human‐mediated spread of the emerald ash borer (Agrilus planipennis Fairmaire), a major pest of ash trees in North America, by visitors to campgrounds in central Canada and the US Midwest. The models incorporate estimates of spread rates from a network of campers travelling from approximately 6500 invaded domains to 266 uninvaded campgrounds in three Canadian provinces (Ontario, Quebec and Manitoba) and three US states (Michigan, Minnesota and Wisconsin). RESULTS: The MECP and PP‐based model solutions agreed for large surveillance budgets but exhibited differences when the budgets were small. These results stem from differences between the coverage‐based objective in MECP and the PP‐based metrics in the PP models. MAIN CONCLUSIONS: Our comparison of MECP and PP‐based models reveals the trade‐offs between objectives. Overall, the MECP is generic and can be adapted to survey species that are spread via other human‐mediated vectors.