Epsilon Archive for Student Projects

Abstract

European ungulate populations are increasing both in number and distributional range, resulting in more ungulate-vehicle collisions (UVC). These UVC cause socio-economic losses and are a growing problem in Sweden. Since 2010, drivers in Sweden are legally obliged to report UVC-accidents to
the police. The police usually call upon specially contracted hunters to take care of the killed or injured animal and produce a report. With this information, from police and hunters, it is possible to map the occurrence of
UVC and derive predictions on where and when the likelihood for accidents is especially high.
The purpose of this study was to build on already existing data and research on UVC in Sweden and develop predictive models for the spatial occurrence of accident hotspots. I explored and analysed which road, traffic, landscape,
ecological and behavioural related attributes correlate with the aggregation of UVC involving roe deer, moose, wild boar, reindeer, and fallow deer respectively and collectively. Using these variables, I created models by
logistic regression to predict UVC hotspots that I believe will potentially assist in future management and preventive actions
My results indicate that a combination of road and landscape variables were good predictors of the occurrence of hotspots in all species except in fallow deer. Nevertheless, road characteristics proved to be the most important
parameters for predicting the occurrence of hotspots. Three road parameters i.e., Traffic Volume, Speed and Proportion of Unfenced Road had positive correlation to the occurrence of hotspots in all the species’ models. Other common variables that were present in at least 50% of the models included areas of open land, exploited land, arable land, minor and major roads and the distance to built-up areas.