Multi-scale habitat suitability models for the Eurasian otter Lutra lutra

The Eurasian otter (Lutra lutra L.) underwent a strong decline in Europe between the 1960s and the 1980s. The decrease in the concentration of harmful pollutants in the environment due to more stringent regulations and the enactment of legal protection have allowed otter populations to gradually recover since the 1980s in several European countries. Compared to other populations in Europe, the Italian population has recovered rather slowly, and signs of the species expanding its range have only recently started to become apparent. The Italian range of the otter is confined to the southern part of the Italian peninsula. The residual population is relatively small and it is geographically isolated separeted from other European populations. Furthermore, this population is currently separated into two isolated subpopulations. Given the small size and the current expansion trend of the south-central subpopulation, it is important to identify areas that can potentially host otters and also to identify the areas through which the species could. The habitat suitability models represent important tool to produce maps of the potential distribution of the species and to identify dispersion ways. The aims of the research project was to identify the factors that influence otter distribution and habitat suitability models at different scale, regional and European. The regional model aimed to identify the distribution, quality and connectivity of habitats of seven river catchments located in the northern portion of the current otter range in Italy. An expert-based Habitat Suitability (HS) model was developed. Connectivity was assessed within and between river basins through landscape algorithms by taking into account variables that could influence otter dispersal. In the same study area we developed inferential models to assess the capacity identifying areas of species expansion. We used data before recolonization events to produce HSMs, one using presence-only data (Ecological Niche Factor Analyses, ENFA) and a second using presence-absence data (Generalized Linear Model, GLM). We used data from the recolonization event to validate these models. We also compared the spatial predictions of these models to a second set of models, using the recolonization data. Our results demonstrated that taking into account absence data can produce wrong predictions of the areas suitable for the recovery of the species. The presence-absence model built with data before recolonization disagrees with the other models about the environmental factors important for that species and the location of the suitable areas. As a conclusion, in non-equilibrium situations, the prediction process could be fooled by misleading absence data. If the problem is ignored, it may misinform wildlife conservation efforts to the point that management actions are non- or counter-effective. At European scale we attempted to determine which factors influence the otter distribution and use them to predict the potential distribution of the species in Europe, under current and future climate. The environmental variables used are related to water availability, food supply, resting site and human disturbance using six different modelling approaches. Future projections are derived by running the CCM3 climate model under a 2xCO2 increase scenario. At the European scale, the otter is mostly influenced by water availability. The current potential distribution reveals large gaps of unsuitable habitats limiting connectivity between otter populations in Europe. Climate change would have different effects on otter habitat suitability in Europe. In the Western part, the model predicts losses of suitable habitats, whereas gains are predicted in central Europe and Eastern Europe shows equal rates of losses and increases of suitable habitat. Our results are important in helping setting up conservation actions and promote otter recovery in Europe.