Practicable methods for delimiting a plant invasion

AIM: Invasive populations are often irregularly distributed due to sporadic dispersal events and patchy distribution of suitable habitat, making it difficult to recognize the population boundary and effectively target management. We designed a survey prioritization that addresses these irregularities and tested it on an invasive King Devil Hawkweed (Hieracium praealtum) population. LOCATION: Bogong High Plains, Victoria, Australia. METHODS: Our optimal design prioritized discovery of infestations beyond the known population boundary, while accounting for the diminishing likelihood of occurrence, plant detectability and the search resources available. Hawkweed management agencies implemented this design in the field. We evaluated its practicality and their progress towards delimitation. RESULTS: The survey design favoured locations outside the known population boundary with a relatively high likelihood of species occurrence and high capacity to detect individuals. Difficult‐to‐search sites were selected for intensive survey when the likelihood of occurrence was sufficiently high. The GPS tracks recorded by managing agencies during the survey corresponded well with the priority rankings, but did not include search effort information. In subsequent surveys, managing agencies have improved recording protocols and expanded their performance metrics. This will facilitate future evaluation of progress towards delimitation and eradication. MAIN CONCLUSIONS: Our optimization method effectively identified priorities for delimiting an invading population, accounting for the agencies' budget and the risk of detection failures. Probabilistic maps proved useful for communicating search priorities, and GPS track logs provided valuable spatial data for inferring where the species is likely to be absent. When comprehensive spatio‐temporal data are available, survey findings can be used to update species occurrence models and thus our quantitative understanding of the population's extent. A probabilistic map provides a more nuanced view of delimitation than drawing a strict boundary around known infestations. Burgman et al.'s extended delimitation score is a promising summary statistic for reporting.