African Elephants' movement pattern analyzer

Human activities and their impact influence animals behavior, be it for adaptation to the change in the ecosystem as well as for threats escaping. These footprints are the biggest drivers for the change in natural movement. Efforts are being made to mitigate these uncontrolled changes. Data Mining and Artificial Intelligence has been used by eco-informaticians to model this change, and their work is valuable to evaluate, understand and predict change, as well as to develop mitigation strategies and adaptation catalogue. This is an ongoing process in the intersection of many domains: ecology awareness, societal adaptation, climate science, policy development and law enforcement.

Since a few decades, many organizations, zoos and individuals have tagged animals to collect geo-locations data for biodiversity research as well as for management purpose. In conservation particularly, location datasets are used to map species’ home ranges, and, when used in near real-time, to enhance law enforcement, mitigate human-wildlife conflicts. Most African elephants movement data is collected to map home range for animals and to map these positions comparing to poaching indicators for law enforcement.

The purpose of this research is to detect outstanding Movement patterns from animal location timeseries as factors that trace external influence on the tranquility or disturbance in the ecosystem using Bayesian Networks. This will give a probabilistic approach evaluating the influence of biotop and abiotop data to elephants movement.

Animal movement patterns are indicators of many factors. The grazing capacity of the landscape, and the animal grazing habits can be determined by the time spent in an area. Change in local weather, and ecological conditions cause a change in these habits, and human presence causes uncommon change in this behaviour.

Real-time or near-real-time poaching indicators with location timeseries to evaluate the change in animal behaviour, we used Garamba’s African Elephants data to implement an Animal Movement Patterns analyser using Bayesian Networks. Local biotope data comprises climate, landcover and topography, as well as infrastructures (road, villages, rangers’ installations…). Animal behaviour will be calculated based on variables considered from our literature review. A Bayesian Network AI will be implemented to alert of any Outstanding patterns.