What you see is where you go? Modeling dispersal in mountainous landscapes

Abstract

Inter-patch connectivity can be strongly influenced by topography and matrix heterogeneity, particularly when dealing with species with high cognitive abilities. To estimate dispersal in such systems, simulation models need to incorporate a behavioral component of matrix effects to result in more realistic connectivity measures. Inter-patch dispersal is important for the persistence of capercaillie (Tetrao urogallus) in central Europe, where this endangered grouse species lives in patchy populations embedded in a mountainous landscape. We simulated capercaillie movements with an individual-based, spatially explicit dispersal model (IBM) and compared the resulting connectivity measure with distance and an expert estimation. We used a landscape comprising discrete habitat patches, temporary habitat, non-habitat forests, and non-habitat open land. First, we assumed that dispersing individuals have perfect knowledge of habitat cells within the perceptual range (null model). Then, we included constraints to perception and accessibility, i.e., mountain chains, open area and valleys (three sub-models). In a full model, all sub-models were included at once. Correlations between the different connectivity measures were high (Spearman’s ρ > 0.7) and connectivity based on the full IBM was closer to expert estimation than distance. For selected cases, simple distance differed strongly from the full IBM measure and the expert estimation. Connectivity based on the IBM was strongly sensitive to the size of perceptual range with higher sensitivity for the null model compared to the full model that included context dependent perceptual ranges. Our heuristic approach is adequate for simulating movements of species with high cognitive abilities in strongly structured landscapes that influence perception and permeability.