Main landscape and road related variables describing ungulate vehicle collisions hazardous locations in Catalonia

Animal-vehicle collisions are a traffic safety and socio-economic topic for many European countries. In Catalonia (NE Spain) the number of accidents involving wildlife is increasing despite to the mitigation strategies applied. Wild ungulates which show rising population sizes are the animals most frequently involved in the accidents.

Ungulate-vehicle collisions (UVC) commonly show an aggregated pattern along the road network and clusters could be identified as hotspots for traffic safety. Assessing the effects of road or landscape variables and distinguishing the main explanatory variables associated to this clustered pattern could help road managers in designing more effective mitigation measures.

During a five-year period (2007-2011), data from 2,320 accidents involving wild ungulates were registered by traffic police, road management teams and the wildlife management department along 12,124 km of the Catalan road network. In contrast to other European regions where deer are the most frequent species causing traffic accidents, wild boar (Sus scrofa) is the main species involved in our study area being responsible for 85% of the overall accidents. 308 significant UVC clusters were identified using a modified kernel density estimation technique (KDE+) and from those 124 shown the highest-frequency of accidents (≥ 3 accidents/km in the studied period). 600 UVC events were selected randomly, 300 located within the highest-frequency UVC clusters and 300 outside them. 12 landscape variables (road junctions, water and ecotone crossings, distances to urban and vegetation cover patches, proportions of different land uses cover types and landscape diversity) and 9 road-related variables (traffic volume, speed limit, road straightness, road-cross section, presence of barriers and medians, roadside vegetation and presence of garbage containers nearby the road) were selected for characterizing the UVC occurrences. Using multiple logistic regressions, several models were fitted assessing the relationship between the explanatory variables and the probability of UVC clustering. Straighter roads, higher speed limits and dense road verge vegetation were consistently associated with increase probabilities of UVC clustering, whereas built-up areas surrounding roads and the presence of road cuttings or embankments decreased this probability. The presence of garbage containers nearby the road were also positively correlated with UVC clustering probably due to wild boar attraction for garbage.

According to the results, UVC clustering in road hotspots is prompted by both landscape and road related factors. Landscape features are relevant for understanding the accidents aggregation, but managing road-related features such as clearing the roadside vegetation might be valuable for reducing wildlife-vehicle collisions in high hazard hotspots.

accident cluster; kernel density estimation; traffic safety; wild boar; wildlife-vehicle collisions