Flexible spatial modelling improves population estimates for elusive carnivores in fragmented landscapes

Le résumé fourni par la source

The brown bear (Ursus arctos) is a large carnivore of conservation concern whose populations are recovering across much of their European range, although in many regions, including Spain, they remain fragmented. To support large-scale conservation planning based on empirical data, in 2020 we conducted a coordinated non-invasive genetic sampling effort across the species’ entire range in the Cantabrian Mountains and applied a Bayesian spatial capture–recapture (SCR) model integrating generalized additive models (GAMs) and finite mixture structures. This framework allowed us to estimate a population of 367 individuals (95% BCI: 327–408), revealing two main population cores and localized high-density patches exceeding 4.5 bears per 100 km². Density was modeled as an inhomogeneous Poisson process using spatial smoothing splines, capturing broad-scale heterogeneity not explained by environmental covariates alone, such as elevation or forest cover. To account for latent movement classes and improve inference on dispersal and behavioral heterogeneity, we incorporated sex-specific finite mixture models. Our results demonstrate that spatial structure in recovering populations may reflect historical fragmentation more than current habitat features. By incorporating latent movement classes, our modelling approach enhances ecological realism and inference, offering a scalable tool for monitoring elusive species. These findings provide actionable insights for national recovery strategies and regional management plans, highlighting the importance of accounting for spatial heterogeneity and movement variability in conservation policy.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.