Developed by the GAIA Initiative, a cutting-edge synthetic intelligence algorithm utilizes animal tag data to consistently classify the behaviors of white-backed vultures, remotely monitoring distant places to document and assess wildlife and environmental situations? As opportunistic feeders, vultures are constantly on the lookout for the next available carcass. Utilizing advanced technologies, including tagged animals and a secondary artificial intelligence algorithm, researchers are now capable of consistently locating carcasses across vast geographical areas. The algorithms presented in a recently published article are thus pivotal components of an early warning system capable of swiftly and accurately detecting critical changes or incidents in the environment, such as droughts, disease outbreaks, or poaching of wildlife?
The GAIA Initiative is a collaborative effort among analysis institutes, conservation organizations, and private companies aimed at developing a cutting-edge early warning system to detect and respond to environmental changes and significant ecological events. Developed by a collaborative effort between the Leibniz Institute for Zoo and Wildlife Research, the Fraunhofer Institute for Integrated Systems, and Tierpark Berlin, innovative AI algorithms have been successfully created.
The decline of wildlife has far-reaching consequences for ecosystems – whether it’s the targeted hunting of a predator or the unforeseen impact of disease outbreaks, environmental pollution, or poaching. In order to comprehensively investigate mammalian species communities and ecosystems, it is crucial to systematically document and analyze the common and distinct patterns of mortality that govern these populations. To harness its full potential, the GAIA Initiative leverages the exceptional abilities of white-backed vultures in tandem with cutting-edge bio-logging technologies and artificial intelligence. According to Dr. Jörg Melzheimer, lead of the GAIA challenge and a scientist at the Leibniz-IZW, this integration of animal, human, and synthetic intelligence forms the foundation of our innovative approach, allowing us to harness the remarkable ecological insights that wildlife has to offer.
With their keen senses honed over countless millennia, vultures have evolved to swiftly and accurately locate carrion across vast expanses. Their exceptional visual acuity and sophisticated cooperative communication enable a collective group to survey vast territories with ease. Vultures play a crucial ecological role by removing decaying animal matter from the environment, thereby preventing the spread of diseases that can affect various species. “For wildlife conservation scientists like ourselves, the sentry abilities of vultures are invaluable assets in identifying and responding to mortality-related issues promptly.” “To leverage vulture data effectively, we require an interface, which at GAIA, we achieve through the integration of animal tags with sophisticated artificial intelligence.”
The satellite tracking data provided by GAIA’s initiative to outfit white-backed vultures in Namibia yields a dual harvest of insights, comprising two distinct categories of information. The GPS sensor provides the exact location of the tracked individual at any given moment in time. With its ability to capture intricate motion patterns, the ACC sensor records comprehensive velocity profiles of the tagged animal along all three spatial axes simultaneously. Teams of knowledge are leveraged by sophisticated synthetic intelligence algorithms developed at the Leibniz Institute of Zoology and Wildlife Research. According to Dr. Wanja Rast, a wildlife biologist and AI expert at the Leibniz-IZW, specific behaviors are characterized by distinct acceleration patterns that generate unique signatures within the animal’s brain, as detected by sensors in the anterior cingulate cortex (ACC). “To accurately identify and attribute specific behaviors to their corresponding signatures, we trained an AI using relevant reference knowledge.” Data were obtained from two white-backed vultures equipped with tags at Tierpark Berlin, as well as 27 wild vultures tagged in Namibia. In conjunction with data gathered through tag readings (ACC), researchers documented behavioral observations of the animals – within the zoo via video recordings and in the wild by observing the birds after tagging.
“Our results show that we collected approximately 15,000 accurate knowledge factors of ACC signatures associated with a distinct and verified vulture behavior.” These actions comprised energetic flight, gliding, mendacious behavior, feeding, and standing still. According to Rast, the knowledge set allowed us to develop an AI algorithm that accurately assigns ACC knowledge to specific behaviors with a high degree of reliability.
By combining the categorization of the animals’ behavior with GPS data from the tracking tags, By applying spatial clustering algorithms, researchers identified regions where specific behaviors exhibited a higher frequency of occurrence. They thereby acquired precisely detailed locations where vultures foraged. According to Aschenborn, the GAIA subject scientists and their collaborators within the subject have successfully confirmed over 500 suspected carcass locations based on sensor data, as well as more than 1,300 clusters of diverse non-carcass behaviors. The sector-verified carcass data conclusively indicate the location of vulture feeding sites, as determined by the researchers’ final AI training dataset – this algorithm accurately pinpoints where an animal has likely perished, with a corresponding carcass at ground level. Aschenborn notes that their method, combining vulture behavior, animal tags, and AI, can accurately predict carcass locations with a 92% success rate, showcasing the potential of this system for large-scale monitoring of animal mortality.
The GAIA early warning system relies on AI-driven capabilities to identify critical changes or incidents in atmospheric conditions, with three core components: behaviour classification, carcass detection, and carcass localisation. According to Melzheimer, the methodological step has been successfully executed thus far within the GAIA knowledge lab at the Leibniz Institute of Zoology and Wildlife Research in Berlin. “With the recent development of advanced animal tags by our consortium, AI-powered analytics are seamlessly integrated into each device.” Real-time information on an animal carcass’s location can be provided without prior knowledge, ensuring timely transmission without lag.
Utilizing a satellite TV connection rather than terrestrial GSM networks enables secure connectivity in remote wilderness areas, unaffected by local infrastructure or limitations. On even the remotest locations, sudden and critical changes in the environment – such as disease outbreaks, prolonged droughts, or the illegal slaughter of wildlife – can be detected instantly.
Over the past several decades, populations of numerous vulture species have plummeted dramatically, putting them at an acute risk of extinction. Primary causes include the scarcity of natural habitats and food sources in human-modified environments, as well as an overwhelming array of intentional or unintentional poisoning incidents. According to records, the population of white-backed vultures plummeted by approximately 90% within just three generations, translating to an alarming average annual decline of 4%. Due to the urgent need to conserve these ecologically crucial birds, it is essential that we significantly expand our knowledge of vultures and their habitats in order to effectively protect and preserve them. Our innovative application of AI-driven analytical approaches will not only provide us with valuable insights into ecosystems but also. By tracking the habits and interactions of over 130 tagged vultures across various African regions, particularly in Namibia, GAIA can significantly enhance our understanding of these birds’ communication, social dynamics, foraging patterns, reproductive cycles, parental care, and knowledge transmission from one generation to the next. Until now, researchers have analyzed more than 95 million GPS data points and 13 billion data points from advanced computer control systems.