Capturing spatiotemporal dynamics of Alaskan groundfish catch using signed-rank estimation for varying coefficient...

Area of Science:

• Ecology
• Statistical Modeling
• Fisheries Science

Background:

• Varying coefficient models (VCMs) offer flexibility in modeling complex systems.
• Fisheries research commonly uses VCMs to analyze spatial variation in marine fish populations.
• Current VCM applications rely on penalized least squares, which is sensitive to non-normal data and outliers.

Purpose of the Study:

• To introduce and evaluate a robust signed-rank-based estimation method for VCMs.
• To address the limitations of classical estimation methods in ecological data analysis.
• To improve the accuracy and reliability of VCMs in fisheries applications.

Main Methods:

• Application of a signed-rank-based procedure for robust estimation in VCMs.
• Comparison of robust estimates with classical likelihood-based VCM fits.
• Validation through simulations and analysis of a North Pacific Ocean fisheries dataset.

Main Results:

• The signed-rank-based estimation method demonstrated a superior fit compared to classical VCM fits.
• Improved prediction accuracy was observed with the robust method, particularly for non-normal or misspecified distributions.
• The method proved effective in handling ecological data characterized by non-normality and outliers.

Conclusions:

• Rank-based estimation is a valuable tool for VCMs in ecological modeling.
• Robust methods enhance the reliability of inferences from fisheries data.
• The signed-rank-based approach provides a more suitable alternative for noisy and non-normal ecological datasets.