A probabilistic pathway model of forest insect dispersal via recreational firewood transport

Because human-mediated dispersal of invasive species is typically vector-based (e.g., via transportation corridors), spatio-temporal models that omit such vectors may poorly predict long-distance dispersal events. This issue is critical in modeling the spread of invasive forest insects by means such as recreational firewood transport, where long-distance dispersal is rare but may lead to significant impacts. This study used data from the US National Recreation Reservation Service (NRRS) documenting >7 million visitor reservations at >2500 campgrounds and recreational facilities nationwide. The distribution of visitors’ travel distances to these facilities is strongly leptokurtic but well fit to theoretical distributions such as the lognormal. Importantly, the distance distribution varies according to particular regional travel patterns. Given this variability, we analyzed the NRRS data in a network setting. We represented visitor home and campground locations as two sets of linked nodes, with the importance of each link defined by the number of campers traveling along it. We applied a probabilistic pathway model to the network to identify major vectors of forest insect spread via recreational firewood movement. We estimated the key probability of an individual camper transporting a viable forest insect based on firewood inspections and usage surveys conducted at various state and national parks. Repeated model simulations yielded probabilistic estimates of the most likely pathways and destinations for a forest insect introduced at any origin node. Furthermore, the results provide probabilistic estimates of the most likely origins for any destination node found to be invaded. Such outputs may substantially improve early detection efforts.