Accounting for behaviour in fine-scale habitat selection: A case study highlighting methodological intricacies

Area of Science:

• Ecology
• Animal Behavior
• Spatial Ecology

Background:

• Habitat selection is fundamental to ecology, influenced by foraging, predator avoidance, and thermoregulation.
• Step-selection functions (SSFs) analyze fine-scale habitat selection based on movement and environmental conditions.
• Quantifying behavior-specific habitat selection using SSFs requires careful consideration of the availability domain.

Purpose of the Study:

• To evaluate the impact of behavior-specific availability domains and models on SSF structure, coefficients, and predictive performance.
• To compare behavior-specific SSFs against behavior-unspecific approaches using muskoxen as a case study.
• To determine the optimal approach for robust inference in habitat selection studies.

Main Methods:

• Hidden Markov Models (HMMs) were used to classify behavioral states (resting, foraging, relocating) from GPS data.
• Step-selection functions (SSFs) were fitted separately for each behavior and compared to a general model.
• Habitat selection was analyzed in relation to terrain, vegetation, and snow conditions.

Main Results:

• Defining behavior-specific availability domains primarily affected model structure (variable selection).
• Fitting separate behavior-specific models mainly influenced selection strength.
• Behavior-specific domains improved predictive performance for foraging and relocating, but decreased it for resting.

Conclusions:

• Accounting for behavior in SSFs impacts model structure, coefficients, and predictive performance.
• Both behavior-specific availability domains and model fitting should be explored for robust ecological inference.
• This approach is critical for informing wildlife management and conservation initiatives.