Authors: Planillo, Aimara; Louvrier, Julie;Wenzler-Meya, Moritz;Reinhardt, Ilka;Kluth, Gesa;Michler, Frank-Uwe;Stier, Norman;Knauer, Felix;Kuemmerle, Tobias;Kramer-Schadt, Stephanie;

After centuries of extirpation, wolves are now recolonizing the human-dominated landscape of Germany. This represents a major challenge for wildlife management and conservation, including increasing human-wildlife conflicts. Limited data and non-stationarity responses to habitat selection in expanding populations pose significant challenges to reliably predict wolves’ range expansion and suitable areas for conservation. We used the natural experiment of the recent wolf expansion into Germany from the past 20 years, combining telemetry with territory monitoring data. We compared predictions from models representing spatio-temporal colonization steps with a global ensemble model that used all data. Models for different colonization steps differed substantially from the global model, showing individuals in newly colonized areas select for areas with lower human disturbance. This cautions against extrapolating habitat models in space based on data from initial colonization phases, when species are out of equilibrium with their environment. To further understand the expansion process, and based on the global suitability map, we developed an individual-based model accounting for population demographic, dispersal and settlement processes, which shows bottlenecks in range expansion and gives a temporal horizon of recolonization. Our modelling approach highlights the importance of explicitly accounting for non-stationarity in habitat selection and demography when modeling range-expanding carnivores.