Final Report Summary - ECOLOGYOFMOVEMENT (New approach to animal movement modeling combining ecological and cognitive sciences, with application to wildlife responses to infrastructure) The transport of goods and people plays a central role in today’s society and its economy. Roads for vehicles represent the lion’s share of transportation routes on the land. These roads can markedly affect animal movements due to habitat alteration during construction, movement facilitation or hindrance (e.g. barrier-effects), and collision-related mortality. As movement allows animals to move between areas when conditions change, it is a crucial adaptation to spatio-temporal variability in resource distribution and living conditions. Therefore, significant efforts are made to preserve and restore movement between habitats in the “Natura 2000” framework in Europe (Art. 10, Council Directive 94/43/EEC). Potential movement hindrance and barrier effects caused by infrastructures are given particular attention in this context, as these can have negative consequences for gene-flow and meta-population dynamics. When animals actually do cross roads and railways, collisions represent an important treat to human safety (not to mention the issue of animal welfare), and cause 300 human fatalities and 30,000 injuries in Europe every year. Ungulate-vehicle collisions cause high costs for the society, which have been conservatively estimated at € 1 billion in Europe. In this project, we set out to investigate the ecological drivers and mechanisms for ungulate space use and movements, and how these relate to collision risks. Movements are animals’ primary behavioral adaptation to spatiotemporal heterogeneity in resource availability. We developed and tested a framework relating the length and frequency of animal movements to the spatiotemporal scale of environmental changes. In general, the length and frequency of animal movements are determined by the interaction between how quickly resource availability changes (i.e. its temporal autocorrelation) and over which spatial area it changes (i.e. its spatial autocorrelation). In support of the proposed theory we found in moose (Alces alces) that frequent, smaller scale movements (e.g. foraging movements) were triggered by fast, small-scale ripples of changes (e.g. local depletion of food by the feeding animal itself), whereas infrequent, larger scale movements (e.g. migratory movements) matched slow, large-scale waves of change in resource availability (e.g. due to spring green-up, or snowfall). In another study, we found that also red deer (Cervus elaphus) respond with migratory movements to large-scale spring green-up. Although, they are not always closely tracking this green-up.In subsequent studies we linked these spatial and temporal changes in the location of animals to the risk of collisions with vehicles. It is obvious that for a collision between a vehicle and an ungulate to occur both the vehicle and the animal need to be at the same time in the same place. The occurrence of cars is spatially very predictable as they are bound to roads. Thus, in two more detailed studies on moose and red deer we focused on the spatial and temporal determinants of animal road crossings based upon high resolution GPS-tracking. We found strong yearly variation in crossing frequencies. Most road crossings in both species occur during fall and winter. During winter the animals move to lower altitudes, and thus closer to roads. We have shown that these large-scale migratory movements are a response to large-scale changes in snow fall during autumn. We found for both moose and red deer that their crossing frequency is highest during dusk and early in the night, in locations where most forage is available. In two studies at the Norwegian scale, we studied the determinants of actual moose vehicle collisions. In a long term study, with data from 14 Norwegian counties over 31 years, we found that local moose density was the most important factor explaining the variation in moose vehicle collisions. Also, traffic volume (private car mileage) and snow depth did affect collision risk. The relationship between traffic volume and temporal variation in moose vehicle collisions was stronger in counties with general low traffic volume, possibly because high traffic volume can act as a barrier to moose road crossings. Likewise, the temporal effect of snow depth was mainly present in counties with on average deep snow, i.e. where it constitutes a constraint on moose movement and space use during winter. This effect is caused by the migration in fall to areas with less snow at lower altitude. This migration leads to an increased density of animals locally around roads, which are also at lower altitudes, resulting in an increased collision risk. In another study, we found that latitudinal variation in moose vehicle collisions are at least partly explained by a coincidence between moose activity periods and peaks in traffic volume. Notably during fall and winter, the peak in moose activity occurs during the morning and evening traffic peaks leading to an increased risk in moose vehicle collisions.These results indicate mitigation actions of ungulate vehicle collisions at both large and small scales. Large-scale mitigation actions should focus on the central role of ungulate population density for collisions. Our study on moose found an isometric relationship between collisions and moose density. Thus, a 25% reduction in moose population size would lead to a similar reduction in collisions. Therefore, population reduction seems the most effective way towards decreasing collisions. Especially in areas where high ungulate density are causing other types of damage (e.g. damage to forestry or agriculture), a reduction in population size should be given serious consideration to decrease ungulate vehicle collisions. Where such reduction in population size is impossible or undesirable, some smaller scale interventions may prove helpful. These interventions focus mostly on raising driver awareness. Our results show that moose road crossings are highly variable in time, which indicate that mitigation actions should include a strong temporal component. Instead of having static warning signals, which lead to fast habituation in drivers, it would be more fruitful to have signaling that varies with the risk of ungulate crossings. We find an increased risk of crossings during dusk, dawn and night time. It would therefore be most effective to have signals of high ungulate prevalence (i.e. in high moose density areas close to feeding sites) during those times of the day. Similarly, the increased probability of moose collisions related to snow depth suggests that warnings of increased collision risks may be given based on meteorological information – e.g. warning signs being activated based on weather forecasts. However, further studies are needed to establish the best predictors, such as accumulated snow depth on the ground versus expected snow fall. Our study does not prove strong evidence for small scale spatial clustering of ungulate crossings in Norway. Both moose and red deer road crossings occur near feeding areas. The relatively uniform landscape with many feeding opportunities for these ungulates does not seem to indicate clear “hot spots” for ungulate crossings or collisions with vehicles. Therefore, the spatial small-scale mitigation actions – such as wildlife over- or under-passages – seem not a promising approach in Norway (except, in combined with wildlife fences). However, more targeted studies are underway to address specifically the issue of spatial clustering of collisions.