Influence of landscape composition and structure on habitat selection of a large herbivore in managed forests along a latitudinal gradient

Large herbivores are keystone species in forest ecosystems, influencing forest structure and biodiversity through their selective browsing. Therefore, understanding herbivore habitat selection across spatiotemporal scales in managed forest landscapes is crucial for wildlife and forest management. This study examines how landscape structure (patch size and contiguity, distance to the nearest road) and composition (habitat type and heterogeneity) influence the seasonal habitat selection of moose (Alces alces) across five ecological zones. Using GPS data from 392 adult moose across 21 study sites (56-670N) in Sweden, we combined Hidden Markov Models and Integrated Step-Selection Analysis to apply a patch-landscape approach that considers animals' behavior-specific responses at the scale of individual habitat patches and the broader landscape matrix. This approach allowed us to assess the role of small-scale habitat features and their spatial arrangement within the larger landscape context for moose movement and patch selection, thereby considering both landscape structure and composition. We found a dominance of landscape composition (i.e. habitat type) shaping moose selection at the patch scale, but also context-specific relevance of landscape structure (e.g. distance to the nearest road, patch size and contiguity). Moose preferred deciduous-mixed and young forests and generally avoided proximity to roads. Individuals occasionally selected for large and well-connected forest patches. Our findings highlight that forest management should prioritize preserving and connecting young and mixed-deciduous forest patches to facilitate moose access to their preferred habitats, thereby helping to distribute moose (and potentially browsing pressure) across forest patches within the managed landscape.