A guide to pre-processing high-throughput animal tracking data

Area of Science:

• Movement Ecology
• Bioinformatics
• Data Science

Background:

• Modern animal tracking generates massive datasets with high temporal resolution.
• Location errors in tracking data can exceed animal step size, leading to misinterpretation of behaviors.
• Standardized, automated methods for cleaning and processing this data are scarce.

Purpose of the Study:

• To provide guidance on building automated pipelines for pre-processing high-throughput animal tracking data.
• To introduce the R package atlastools for efficient data cleaning and analysis.
• To demonstrate the utility of the pipeline and package with simulated and real-world animal tracking datasets.

Main Methods:

• Development of a pre-processing pipeline balancing ease of use and computational efficiency.
• Application of the pipeline to simulated data with introduced location errors.
• Utilizing the pipeline to segment and cluster location data into 'residence patches' for space use analysis.
• Demonstration using tracking data from the Wadden Sea ATLAS system (WATLAS) and Egyptian fruit bats (Rousettus aegyptiacus).

Main Results:

• The proposed pipeline effectively reduces location errors while preserving valid animal movements.
• The R package atlastools facilitates standardized and reproducible data pre-processing.
• The residence patch method, enabled by the pipeline, provides biologically meaningful insights into animal space use.
• Pre-processing significantly improves the quality of tracking data and enhances analytical outcomes.

Conclusions:

• Automated pre-processing pipelines are essential for handling big data in animal tracking.
• The atlastools package offers a user-friendly and robust solution for movement data analysis.
• Standardized methods in movement ecology lead to more reliable inferences from tracking data.