Predictive mapping of wild carrot ( Daucus carota ssp. carota ) in central Oregon

Abstract

This study addresses the distribution of wild carrot [ Daucus carota (L.) ssp . carota ], in the hybrid carrot seed [ Daucus carota (L.) ssp. sativus (Hoffm.) Arcang.; hereafter Daucus carota ] production region of central Oregon. Field surveys conducted in 2022 and 2023 in collaboration with regional partners provided observational data to train a relative risk model (random forest). Integrating 75 known D. carota occurrences, more than 2,000 known absences, and 1,000 randomly generated background points with data on topography, climate, soil, road proximity, and land cover, the model effectively distinguished potential D. carota establishment sites, achieving a high predictive accuracy (AUC = 0.911) against a background dataset. We identified four model-probability cutoff values (using precision–recall F -measure, varying alpha parameter), yielding five risk categories, from very low to very high risk. The moderate to very high risk categories highlight priority areas for monitoring and management to mitigate the risk of D. carota establishment, a concern especially relevant for local hybrid carrot seed production given the potential for cross-pollination and genetic contamination. The resulting risk maps provide a spatially explicit framework for identifying environmentally suitable habitats and potential invasion corridors, supporting early detection and rapid response efforts. By focusing surveillance and management activities within high-risk areas, stakeholders can improve monitoring efficiency, reduce the likelihood of D. carota establishment and spread, and help maintain the genetic integrity of central Oregon’s hybrid carrot seed production system.