Abstract
After its disappearance from France around the 1950s, the grey wolf (Canis lupus) has returned and is now settled throughout the French Alpine Arc. Its predations on livestock, called depredations, have increased across time since its recolonisation of the region. Depredations mostly concern sheep within the context of pastoralism. The depredation process creates conflicts between wolf conservation and pastoral activities. Besides financial compensations of depredation losses and subsidies for non-lethal protective measures of flocks, France has added another tool of mitigation: lethal removals of wolves. Uncertainty remains about the efficiency of lethal measures to reduce the depredations levels, whether it is in France or elsewhere where these measures are applied. Two opposite assumptions are currently made. The first assumption supports that lethal measures are efficient through population reduction and selection of wolves less likely to depredate. The second assumption supports that lethal measures are counter-productive because of pack disruption which in turn increase the needs of wolves to rely on livestock. The reason for the uncertainty about the effects of lethal measures is the combination of a low number of studies on the subject and a weak scientific inference, thus producing contradictory results. Moreover, most studies have focused on the North American situation. My work consisted in reducing the uncertainty about the effects of lethal measures on the recorded successful depredations on sheep in France. We adopted two approaches. First, we developed an individual-based modelling approach to study the whole dynamic induced by lethal measures on the wolf population structure and on depredations. We integrated biological mechanisms as pack dissolution that were never integrated before. We tested different scenarios of depredation behaviour of wolves. Our model supported that the modelled lethal measures were efficient to reduce depredations through population size reduction 1) when the individual probability of depredation was increasing for wolves in pack because of increasing food needs or 2) when, in addition, it was also increasing for adult wolves and with pack size, to mimic hunting abilities. On the other hand, our model did not allow us to conclude on the effects of lethal measures when the individual probability of depredation was decreasing for adults or with pack size. Second, we analysed the effect of the deaths of wolves on the depredation levels in France with data provided by the French administration. We developed a methodology based on kernel calculations. We concluded that the effect of culling in France was highly variable according to contexts. The majority of the results involved a reduction of the depredations, but could also involve no effect or an increase of the depredations. Finally, we added a retrospective analysis of the depredation spatial pattern in France by adjusting for spatial and temporal sheep distributions and flock sizes. We concluded that hotspots likely did not result from the unique factor of wolf territoriality. We also observed that depredations tended to homogenise across time in the French Alpine Arc. The method finally allowed us to spot hotspots encompassing small flocks or grazing over a short period, which would have remained unspotted with the current hotspot determination of the French administration. Thus, this work shows that the local context, linked to environmental, wolf and pastoral variables, needs to be taken into consideration for the management of depredations, because lethal measures are unlikely to have a unique effect on wolves and their depredations.