Timely poacher detection and localization using sentinel animal movement

Area of Science:

• Wildlife conservation technology
• Animal behavior and ecology
• Machine learning in environmental science

Background:

• Wildlife crime is a highly profitable illegal industry, driving endangered species declines.
• Existing anti-poaching measures are insufficient to curb illegal activities and protect wildlife.
• There is a critical need for innovative solutions to detect and deter poachers effectively.

Purpose of the Study:

• To propose and evaluate an early warning system for poacher detection.
• To utilize the natural threat-response movements of sentinel animals for detection.
• To assess the accuracy and localization capabilities of the proposed system.

Main Methods:

• Analysis of human-evasive movement patterns in 135 savanna herbivores across four species.
• Implementation of an internet-of-things (IoT) architecture with wearable sensors and wireless data transmission.
• Application of machine learning algorithms to identify characteristic behavioral signatures of human presence.

Main Results:

• Accurate detection of human intruders with 86.1% accuracy.
• Localization of intruders with less than 500m error in 54.2% of staged intrusions.
• Identified behavioral signatures include increased speed, energy expenditure, directional persistence, herd coherence, and suboptimal habitat selection.

Conclusions:

• Sentinel animal behavior, analyzed via wireless biologging and predictive analytics, offers a viable method for early poacher detection.
• The system effectively identifies deviations from normal behavior, indicating potential threats.
• This approach can enhance wildlife conservation, surveillance, and safety.