Artificial Intelligence Identification of Wildlife Vehicle Collision Hotspots

Abstract

Transportation networks increase in size and extend into forested, wetland and rural areas, wildlife–vehicle collisions (WVCs) are becoming a more significant road ecology issue. The collisions result in direct loss of life and property, loss of wildlife viability, habitat fragmentation, and safety and economic costs to motorists. Therefore, collision hotspot identification – the road segments where collisions disproportionately occur – is a key component of prioritizing conservation and mitigation resources; however, traditional methods of hotspot identification using field surveys and accident records are incomplete at a spatial scale, inconsistently recorded and hard to maintain in large road networks. The conceptual framework described in this paper focuses on the AI-assisted identification of wildlife–vehicle collision hotspots by structuring and analysing wildlife observations, road design features, and environmental conditions simultaneously to identify patterns of association between landscape conditions and collision presence. Road segments are divided into low, moderate and high-risk zones to aid in targeted conservation and infrastructure planning. Predicted results show that hotspots are located close to habitat borders, water sources, and corridors for wildlife and that there will be significant interatomic and intraseasonal variation in collision risk. The method is expected to provide more continuous spatial and regularly updated hotspot maps, than the manual survey-based evaluation alone, and requires field-verified ecological data for validation. Implications include conservation planning (information on the siting of crossing structures, fencing, and seasonal monitoring) and road planning (maps of hotspots can be used to manage speed limits and design of road infrastructure). AI is not intended to replace ecological field observations and expert judgement, but to be used as a decision-support tool to help increase the safety of roads, enhance connectivity, and help plan transport in ways that are good for wildlife.