An AI-based platform to investigate African large carnivore dispersal and demography across broad landscapes: A case study and future directions using African wild dogs

Abstract

Understanding dispersal patterns and demographic processes is crucial for the development of evidence-based conservation practices. Obtaining such information relies on the ability to identify and track individuals across spatial and temporal scales relevant to the life-history events under investigation. This knowledge can be achieved by combining photographic and sighting data collected by various sources with a high accuracy automated individual identification platform. Here, we present the African Carnivore Wildbook (ACW), an AI-based graphical user interface tool capable of identifying individuals of several African carnivore species and specifically developed to accommodate the above outlined needs. We showcase the ACW functionality using the endangered African wild dog as an example. Pictures collected over an area >56,000 km² and submitted to ACW allowed inferences on movement patterns and dispersal at regional and international scales; for instance, transboundary dispersal events >200 km were documented. ACW furthermore enabled monitoring some individuals for >4 years; such information is invaluable for reliable survival analyses. We discuss how the ACW can contribute to data collection at appropriate spatial and temporal scales to support population monitoring, scientific research and management of African wild dogs and other apex carnivores and to the conservation of these charismatic species.