Death is an everyday—yet vital—process in ecosystems. The cause of death says a lot about what is happening in animal communities and their habitats. As well as natural causes, deaths can also indicate that disease has broken out, environmental toxins have spread or animals have been killed illegally. Getting timely data about these deaths can help scientists and conservationists to protect habitats and wildlife. But how does one go about getting reliable information about such deaths? The GAIA initiative came up with a cleaver solution: vultures.
Who could know more about death than a scavenger who is constantly on the lookout for carcasses? Vultures have an outstanding sense of sight and sophisticated communication. They can reliably and quickly track down carcasses together with their conspecifics and thus “monitor” large land areas.
White-backed vultures and a jackal have discovered a carcass. i³: Animal, artificial and human intelligence
The researchers in the GAIA project are equipping African white-backed vultures with a transmitter that uses a camera to record images from the vulture’s perspective. An AI algorithm runs directly on the transmitter and analyses the movement data. The researchers classify this data and thus obtain information about animal behaviour and the environment from a distance in near real-time. “We therefore call this concept i³ – the combination of three intelligences. Animal, artificial and human intelligence,” says Jan Zwilling, with whom RESET spoke shortly after his return from Uganda. Zwilling is involved in the project on behalf of the Leibniz-IZW. Other project partners are Fraunhofer IIS, Rapid Cubes GmbH and Zoo Berlin.
The information obtained with the help of the vultures has practical benefits for environmental protection and environmental research. The early warning system can be used to recognise problems and initiate measures quickly, such as checking a national park for an outbreak of wildlife disease. At the same time, the information is an important database for research into ecosystems.
A white-backed vulture is fitted with a transmitter. Challenges in the development of a transmitter the weight of a chocolate bar
In November 2024, after three years of development, the team tested a transmitter prototype on Ugandan vultures. Getting there was not easy.
As the transmitters for vultures require more functions than those available on the market, technical hurdles had to be overcome. Another key challenge in the development of the transmitters was the miniaturisation of the technology. To avoid placing unnecessary strain on the animals, there is a strict upper limit for the total weight of the transmitter. “Everything has to fit into the weight of a bar of chocolate. The housing, the lens, the camera sensor, the battery, a solar panel and the microprocessors that perform the artificial intelligence calculations,” reports Dr Jörg Melzheimer, project manager of the GAIA initiative at the Leibniz-IZW. This limits the computing power, the AI must be lean enough to run on the transmitter. “This is anything but trivial, as it means that only a fraction of the computing power of a PC or smartphone is available.”
In addition, two very different disciplines—engineers and technicians on the one hand and wildlife biologists and veterinarians on the other—had to be brought together. “Only together, as biologists with an affinity for technology and engineers who are now familiar with vultures, were we able to get this far,” reports Melzheimer.
Data scientists and biologists worked together on the project. Reality check on wild scavengers
The technology also had to withstand a tough reality check. There is a world of difference between laboratory conditions and continuous use on wild scavengers. When travelling with African vultures, the transmitters are exposed to extreme solar radiation as well as heavy rain. And the animals behave unpredictably, so that, for example, the camera got dirtier than expected. In addition, the radio infrastructure is not always reliable. This often makes it difficult to localise the transmitters. “These are just a few examples of the infinite number of details that can mean that a transmitter designed with the best knowledge and the greatest possible care does not fulfil its function perfectly in reality,” says Jan Zwilling.
However, the GAIA initiative team was able to confirm that the system works in initial tests. The detection of carcasses has already achieved a hit rate of 92 percent and is therefore very accurate.
Soon after the system was established in Namibia, vultures in other African countries such as Mozambique, Zambia and Uganda were equipped with transmitters and early warning systems were set up in partnership with other organisations.
The prototype of the transmitter is tracked on site. Better protection for vultures with new findings
The GAIA initiative has fitted more than 130 vultures with transmitters so far. The work of the research group goes beyond the establishment of an early warning system. At the same time, new insights are also being gained into the birds, which are threatened with extinction.
In recent decades, the populations of many vulture species have declined sharply. The main reasons for this are the loss of habitat and food in landscapes characterised by humans. In addition, there is a high number of direct or indirect poisonings. The population of the white-backed vulture, for example, has declined by around 90 per cent in just three generations.
“Due to their ecological importance and their rapid decline, it is necessary to significantly improve our knowledge and understanding of vultures in order to better protect them,” says Dr Ortwin Aschenborn, GAIA project manager at the Leibniz-IZW. “Our research using AI-based analysis methods will not only give us novel insights into ecosystems, it will also expand our knowledge of how vultures communicate, interact and cooperate, forage for food, hatch and rear young and how they pass on knowledge from one generation to the next.”
Will ravens soon be “telling” us about animal diseases?
What the researchers of the GAIA initiative have developed as a blueprint with vultures in Africa is certainly transferable to animals in Europe. Corvids and birds of prey also feed, at least in part, on carrion. This also makes them reliable indicators of animal carcasses. “We have already developed an AI for ravens to a very advanced stage and can already classify the behaviour of the birds with great accuracy and thus identify suspected carcasses,” says Melzheimer.
The technology could be used, for example, to monitor African swine fever in Europe. “The ravens ‘patrol’ the forests and indicate dead wild boars with enormous reliability and efficiency.” The information processed with the help of AI could then recognise local outbreaks, be forwarded directly to the responsible authorities and enable animal disease to be monitored with comparatively little effort. According to Melzheimer, the first steps have been taken, but such a system has not yet been implemented.
Animal tracking in environmental research
Other projects also rely on monitoring with the help of animal transmitters. For example, researchers have found that tracking turtles leads to seagrass beds more reliably than satellite images. And the Global Ocean Observing System (GOOS) relies on the support of elephant seals to gain new insights into the conditions in the Antarctic Ocean. As the mammals dive to depths of up to 2,000 metres several times a day, researchers have unprecedented access to the ocean without the need for expensive submersibles or research vessels. Read more here: How animal experts of the oceans are helping with Antarctic climate research.
The GAIA initiative still has a lot to come
The team was surprised by the great response to the project. In addition to the partnerships with various organisations, the project team has been able to establish close relationships with members of the German parliament, ministries and representatives such as the Federal Government Coordinator for German Aerospace over the past two years. “The Bundestag Committee on the Environment even came to visit us in Namibia. This shows us that we have struck a chord with politicians and that the potential of modern approaches in wildlife research and species conservation is recognised,” says Zwilling.
Nevertheless, the two core projects of the GAIA initiative will expire at the end of 2024. However, the team’s project work has not yet come to an end. The transmitter prototype is to be further developed and a satellite communication component of the system is still to come. Then there is still the transfer to other animal species, ecosystems and issues.
The view from above from a vulture’s perspective. The funding applications for this have already been submitted to various funding organisations. For a long time, things looked very good for the direct connection of the funding, reports Jan Zwilling. However, since 6 November 2024, financing from federal funds has been somewhat difficult and the project has been put on hold. “Of course, we hope that a new perspective will emerge at federal level next year. Until then, we will have to rely more heavily on support from NGOs or the private sector in order to continue our work, at least to a limited extent.”
