A new study has used satellite imagery alongside artificial intelligence (AI) to count wildebeest across the vast Serengeti-Mara ecosystem in East Africa for the first time.
The findings, published in PNAS Nexus, reveal fewer than 600,000 animals, far below the widely cited figure of 1.3 million that has remained largely unchanged since the 1970s.
“The field of wildlife conservation relies on having accurate data on wildlife population numbers,” says lead researcher Dr Isla Duporge, who believes the study not only helps our understanding of migratory wildebeest numbers, but also offers new opportunities for surveying other animal species.
- Mesmerising footage shows the great wildebeest migration in action
- 10 greatest wildlife spectacles on the planet
Counting wildebeest from space
The 'great wildebeest migration' is one of the world's most incredible natural spectacles, with vast herds crossing Tanzania’s Serengeti and Kenya’s Masai Mara each year.
The migration sustains predators such as lions, crocodiles and spotted hyaenas, while drawing tourists from across the globe and generating essential revenue for both countries.
Until now, population counts have relied on aerial surveys using manned aircrafts to photograph herds along fixed transects. Scientists then extrapolate these counts to estimate total numbers. But this method risks inaccuracies, says Duporge, especially if herds move between survey areas or cluster unevenly.
Satellite technology offers a safer and more accurate alternative. A single high-resolution image can cover hundreds of thousands of square kilometres, reducing errors and minimising disturbance to wildlife. However, manually counting animals from these images is impractical given the sheer data volume.
In the new study, the University of Oxford-led team devised what they believe is a far more accurate method of counting by combining satellite imagery with AI.
- Why do female wildebeest have horns?
- Waterhole attack: Incredible slow-mo footage shows moment huge crocodile strikes
- Scientists spent 6 years tracking Yellowstone's great bison migration. What they found is remarkable
The new method involved training two deep-learning models (which take in information from multiple data sources and analyse that data without human intervention) to identify wildebeest using a dataset of 70,417 manually labelled wildebeest.
They then applied these models to over 4,000km² of high-resolution satellite imagery of the Masai Mara National Reserve in Tanzania and Kenya. These were captured from August 2022 and 2023 by satellites between 383 miles (617km) and 478 miles (770 kilometres) above the Earth’s surface.
Both models produced strikingly consistent results, estimating between 324,202 and 337,926 wildebeest in 2022, and between 502,917 and 533,137 in 2023.
“The sheer difference between traditional estimates and our new results raises questions about where the ‘missing’ wildebeest might be,” says Duporge, who highlights that even the higher figure (from 2023) falls hundreds of thousands short of the longstanding 1.3 million estimate.
“Based on data from GPS tracking surveys, we are confident that most of the herd were contained within the surveyed area," adds Duporge. "And whilst some individuals may have been obscured by tree cover, it seems unlikely that such a large number – on the order of half a million – would have been concealed in this way.”
What's does this mean for wildebeest?
Researchers say the results do not necessarily mean wildebeest numbers have collapsed. Changes in migration routes or survey coverage may explain the discrepancy.
Still, pressures on wildebeest are mounting. Expanding agriculture, infrastructure and fencing are fragmenting migration corridors, while shifting rainfall patterns caused by climate change affect grazing grounds. Reliable population data is vital for conservation planning, say the researchers.
“The most basic fact to know as a foundation for conserving any species is how many of them there are,” says Professor David Macdonald, co-author of the study and founder of Oxford’s Wildlife Conservation Research Unit.
"The technological breakthrough of our study - satellite-based wildlife monitoring, powered by AI - potentially revolutionises the answer for wildebeest, besides opening up incredible possibilities for monitoring other large species.”
This is the first time AI satellite analysis has been used to count a widely dispersed wild mammal population, building on earlier work that tracked elephants from space. The technique could help monitor other herd species such as zebra, reindeer and camels. Researchers are now working on adapting it to count rhinos.
Top image: Northward migration of wildebeest. Credit: WorldView-3 satellite image | Tiejun Wang
More amazing wildlife stories from around the world
- Research vessel spends 45 days in remote Arctic – and finds living thing moving inside the ice
- In 2020, humans suddenly vanished from a popular bay in Hawai‘i. What happened next was remarkable
- Scientists spent 6 years tracking Yellowstone's great bison migration. What they found is remarkable
- More than 1,000 rattlesnakes caught in Michigan wetlands. What scientists find in their DNA is concerning
