Estimating animal resource selection from telemetry data using point process models.

Area of Science:

• Ecology
• Wildlife Biology
• Statistical Modeling

Background:

• Animal resource selection functions (RSF) analyses using telemetry data are common but face statistical challenges due to data autocorrelation.
• Existing weighted distribution methods for RSF analysis are computationally intensive and cannot be aggregated over time for space-only inference.

Purpose of the Study:

• To develop a novel statistical approach for analyzing animal telemetry data to overcome limitations of current methods.
• To propose a space-time point process model that generalizes existing weighted distribution models for RSF inference.

Main Methods:

• The study treats telemetry data as a realization of a space-time point process, considering relocation times as random.
• Developed point process models generalize weighted distribution models, enabling the use of common statistical software for analysis.
• Models can be aggregated over time by marginalizing the temporal component, allowing for flexible RSF analysis from individual to study area levels.

Main Results:

• The proposed point process models offer computational advantages and are readily implementable by ecologists.
• Analysis of northern fur seal telemetry data demonstrated the model's applicability.
• While individual-level analysis showed little habitat selection, study area-level analysis revealed strong selection relative to covariates.

Conclusions:

• The space-time point process framework provides a powerful and flexible alternative for analyzing animal telemetry data.
• This approach simplifies computation and enhances the ability to infer resource selection at various spatial scales.
• The findings highlight the importance of considering different spatial scales when assessing animal habitat selection.