An individual-based model for white storks (Ciconia ciconia) in Germany during breeding season

Jannatul Ferdous, Ronny Peters, Uta Berger, Elina Takola 

Understanding how habitat selection influences individual fitness is essential for predicting species responses to environmental change. Resource Selection Functions (RSFs) are widely used to quantify habitat preferences, but they often overlook individual variation and rarely link habitat use to demographic outcomes. We combined empirical habitat-selection modelling with a spatially explicit individual-based model (IBM) to investigate how habitat use influences movement, energetic performance, breeding success, and survival of white storks (Ciconia ciconia) in Germany. Using breeding-season GPS data from 33 individuals, we fitted matched used-available conditional logistic regression models incorporating vegetation productivity (NDVI), landscape structure, and land-cover classes. The selected model indicated negative selection for NDVI (β = -0.42) and positive selection for edge density (β = 0.19), suggesting a preference for open, structurally heterogeneous landscapes. Strong intraspecific variation was observed, with several individuals exhibiting habitat-selection responses that differed from population-level trends. Cross-validation indicated moderate predictive performance, with used locations ranked among the three highest-scoring alternatives in 37-39% of choice sets. Later, we integrated the selected RSF into an IBM simulation of breeding-season movement across five selected years, which includes resource acquisition through movement choices, energetic dynamics, breeding decisions, mortality risk, and breeding-site fidelity. Simulated outcome showed a breeding success ranging from 78.8% to 92.6%, while annual survival ranged from 87.1% to 100%. Individuals occupying more suitable habitats achieved higher cumulative energy balances and higher breeding success. Sensitivity analyses showed that model outcomes were robust to variation in movement and relocation parameters but sensitive to energetic breeding thresholds. Our study demonstrates how RSF-informed IBMs can translate habitat-selection patterns into fitness-related outcomes and improve predictions of species responses to landscape change.