How to account for behavioral states in step-selection analysis: a model comparison

Area of Science:

• Ecology
• Movement Ecology
• Wildlife Biology

Background:

• Step-selection models analyze animal movement data for habitat selection.
• Unobserved behavioral states (e.g., foraging, resting) influence habitat preferences and movement.
• Standard step-selection analyses often ignore these behavioral states, potentially biasing results.

Purpose of the Study:

• To evaluate the performance of combining integrated step-selection analysis (iSSA) with hidden Markov models (HMMs).
• To compare the HMM-iSSA approach against standard iSSA and a classification-based iSSA.
• To quantify the consequences of ignoring behavioral states in step-selection analyses.

Main Methods:

• Developed and applied a Hidden Markov Model integrated Step-Selection Analysis (HMM-iSSA).
• Conducted an extensive simulation study to test model performance.
• Utilized a case study involving simultaneously tracked bank voles (Myodes glareolus) for empirical comparison.

Main Results:

• The HMM-iSSA approach allows joint estimation of behavioral states and state-dependent habitat selection.
• Empirical comparison showed HMM-iSSA outperformed standard and classification-based iSSA.
• The study quantified the impact of unobserved states on habitat selection inference.

Conclusions:

• Combining HMMs with iSSA provides a robust framework for analyzing state-dependent habitat selection.
• Ignoring behavioral states in step-selection models can lead to inaccurate conclusions.
• The HMM-iSSA method is implemented in the R package HMMiSSA for broader accessibility.