Abstract
Large carnivores are recovering in Europe, due to an increasing forest cover, ungulate population and conservation measures. Tthis return poses challenges as carnivores can interact with livestock farming. Assessing their distributions can help to predict and mitigate conflicts with human activities. Because large carnivores are highly mobile, elusive and live at very low density, modeling their distributions presents several challenges due to 1) their imperfect detectability, 2) their dynamic ranges over time and 3) their monitoring at large scales consisting of opportunistic data without a formal measure of the sampling effort. In this thesis, we focused on two carnivore species, wolves (Canis lupus) and Eurasian lynx (Lynx lynx), to develop the methodological aspects related to the modelling of species distributions. We considered the application of occupancy models in the context of monitoring large carnivores in Europe. These models allow the establishment of a link between the species’ presence and environmental covariates while accounting for imperfect detectability, in order to establish the proportion of a study area occupied by the species.We first assessed wolf range dynamics in France from 1994 to 2016, while accounting for species imperfect detection and showed the importance of accounting for time- and space-varying sampling effort using dynamic site-occupancy models.Second, acknowledging that false positives may occur when monitoring rare species, we showcased a dynamic occupancy model that simultaneously accounts for false negatives and positives to jointly analyze data that include both unambiguous detections and ambiguous detections. The analysis of data on the Eurasian lynx in Alpine countries suggested that incorporating ambiguous detections produced more precise estimates of the ecological parameters.Third, we developed a model accounting for heterogeneity in detection while dealing with false positives. Applying our new approach to a case study with grey wolves in France, we demonstrated that heterogeneity in wolf detection was due to a heterogeneous sampling effort across space.Fourth, to deal with multiple data sources, we developed a Poisson point process approach which allows the inclusion of different data sources when building SDMs. By doing so, we also answered the question about which source(s) of information would provide most of the information when monitoring the lynx in Norway.Fifth and finally, to understand the underlying mechanisms of the colonization of wolves in France, we developed a statistical framework for estimating spatiotemporal occupancy and abundance dynamics using the ecological diffusion framework. We demonstrated the potential of our approach to predict the potential future distribution of wolves in the short term, an element that could contribute to target management areas or focus on areas of potential conflict.Overall our work shows that opportunistic data can be analyzed with species distribution models that control for issues linked to the type of monitoring used to produce the data. Our approaches have the potential for being used by decision-makers to optimize the monitoring of large carnivores and to target sites where carnivores are likely to occur and mitigate conflicts.