Near Real-Time Calving Detection in Grazing Cows Using GNSS-Derived Behavioral Anomalies

Area of Science:

• Animal Science
• Agricultural Technology
• Behavioral Ecology

Background:

• Calving detection in grazing systems is challenging due to remote locations.
• Timely calving detection is crucial for animal welfare and herd management.
• Existing methods often lack real-time capabilities in extensive grazing environments.

Purpose of the Study:

• To develop and evaluate a near real-time calving alert system using Global Navigation Satellite System (GNSS) collar data.
• To identify behavioral anomalies indicative of calving in grazing cows.
• To assess the system's performance across different farms and calving events.

Main Methods:

• Collected GNSS collar data from 149 cows across three farms in Spain and Australia, including 76 calving events.
• Calculated individual, social, and herd-relative behavioral indicators every 30 minutes using 24-hour rolling windows.
• Defined anomalies as deviations from recent behavior using delta-self and Z-score transformations, evaluated at 30-minute intervals or accumulated over 3-hour windows.

Main Results:

• The calving alert system achieved an F1-score of 0.58 using a complete set of behavioral indicators.
• Anomaly detection and alerts were concentrated in the final two days before calving.
• Increased spatial separation and reduced movement relative to the group were key indicators preceding calving.

Conclusions:

• GNSS-based behavioral anomaly accumulation offers a transparent and flexible approach for calving detection in grazing systems.
• The developed system demonstrates potential for improving calving management in extensive livestock operations.
• Further refinement of anomaly thresholds and indicator sets could enhance system precision and recall.