Linking resource selection and step selection models for habitat preferences in animals.

Area of Science:

• Ecology
• Movement Ecology
• Computational Biology

Background:

• Discrepancies exist between individual-based (step selection functions) and population-based (resource selection functions [RSFs]) models of animal species-habitat associations.
• This incompatibility hinders a unified understanding of habitat use patterns.

Purpose of the Study:

• To resolve the fundamental incompatibility between individual and population viewpoints in animal movement analysis.
• To introduce a novel application of Markov chain Monte Carlo (MCMC) algorithms to animal movement studies.

Main Methods:

• Proposed an analogy between animal movement and MCMC sampler movement to ensure convergence of step selection rules to population utilization distribution parameters.
• Introduced a rejection-free MCMC algorithm, the local Gibbs sampler, mimicking realistic animal movement.
• Accommodated a wide range of biological assumptions within the MCMC framework.

Main Results:

• Demonstrated theoretically and empirically that locations simulated using the local Gibbs sampler yield correct RSFs.
• Showed that the proposed MCMC approach guarantees convergence to the target utilization distribution.
• Successfully estimated resource selection and movement parameters using simulated data.

Conclusions:

• The local Gibbs sampler provides a unified framework for analyzing animal species-habitat associations.
• This MCMC-based approach reconciles conflicting results from individual and population-level models.
• Offers a flexible method for estimating movement and resource selection parameters under diverse biological assumptions.