How many wildebeest live in the Serengeti ecosystem is a question that, until recently, had a single confident answer repeated across guidebooks, tourism boards and safari operators for nearly fifty years: 1.3 million. That figure came from aerial surveys conducted since the 1970s and had barely moved in any published estimate since. In September 2025, a study using satellite imagery and artificial intelligence produced a very different number, and the gap between the two has become one of the more genuinely unsettled questions in East African wildlife science.
The long standing figure of 1.3 million
The 1.3 million estimate traces back to aerial point sampling surveys, in which a fixed wing aircraft flies a series of straight, evenly spaced transects across the ecosystem and observers count and photograph the wildebeest visible within each transect, with the results then extrapolated across the full survey area using statistical methods. The most recent aerial survey of this kind, conducted by the Tanzania Wildlife Research Institute in 2023, covered close to 4,800 square kilometres of the southern Serengeti during the period after calving and produced a figure consistent with the long standing 1.3 million estimate. This method has been the basis of wildebeest population science since the 1950s, when the migratory population was estimated at around 250,000 individuals, before a rinderpest vaccination programme for domestic cattle in the surrounding areas removed a disease that had been suppressing wildebeest numbers for decades and allowed the population to grow through the following twenty years to the figure that has held, with some drought related fluctuation, ever since.
The rinderpest recovery that shaped today’s population
The scale of the modern wildebeest population only makes sense against the backdrop of the disease that suppressed it for the first half of the twentieth century. Rinderpest, a viral cattle disease introduced to East Africa in the 1890s, spread into wild ungulate populations including wildebeest and buffalo, keeping numbers artificially low across the region for decades. A large scale vaccination campaign targeting domestic cattle herds around the Serengeti in the late 1950s and early 1960s removed the disease reservoir that had been reinfecting wild populations, and wildebeest numbers grew rapidly through the following two decades as a result, climbing from an estimated 250,000 individuals to over a million by the late 1970s. This growth phase is the origin of the 1.3 million figure that then remained largely static, with periodic drought related dips, for the next five decades, which is itself part of why a sudden and substantial downward revision has prompted such scrutiny from researchers who have spent their careers working with the older methodology.
The 2025 satellite count and why it produced a different number
A team led by researchers at the University of Oxford, in collaboration with international partners, applied deep learning models to high resolution satellite imagery covering more than 4,000 square kilometres of the Serengeti Mara ecosystem, captured during August 2022 and August 2023 by Maxar Technologies satellites orbiting between 617 and 770 kilometres above the earth’s surface. Two separate AI models were used to detect and count individual wildebeest within the imagery, and both produced closely aligned results: between roughly 324,000 and 338,000 individuals in 2022, and between roughly 503,000 and 533,000 in 2023. Both figures sit at under half the widely cited 1.3 million estimate, a shortfall the researchers describe as at least 700,000 animals against the traditional aerial figure.
The study’s lead researcher, Dr Isla Duporge, has been open about the uncertainty this creates rather than presenting the satellite figure as a straightforward replacement for the aerial count. Some individual wildebeest may have been obscured by tree cover and missed by the imagery, and the resolution of the satellite photographs, with each wildebeest appearing as a shape only six to twelve pixels across, means the AI models cannot always reliably distinguish a wildebeest from a similarly sized zebra or eland standing nearby, which introduces its own margin of error into the lower figure.
Why the two counting methods disagree
Part of the disagreement is methodological rather than a sign that the population itself has genuinely halved. The satellite survey covered a defined area during a specific two week window in August of two consecutive years, while the herd’s distribution across the wider ecosystem shifts throughout the year in response to rainfall, and any survey covering a fixed area at a fixed time risks missing animals that have moved outside the surveyed zone during that particular week. Grant Hopcraft, a conservation biologist at the University of Glasgow who has spent years working on the Serengeti’s wildebeest census using traditional methods, has been publicly sceptical of the new figure, arguing that the satellite study covered only part of the herd’s range at that time of year and most likely missed a substantial portion of the population as a result, and describing the study as unconvincing as a true abundance estimate on its own.
The researchers behind the satellite study have themselves been careful not to claim that hundreds of thousands of wildebeest have died or vanished. Instead, they suggest the herds may have shifted their migration routes and distribution patterns in ways that the historical aerial survey methodology, built around fixed assumptions from the 1970s, has not fully captured, and that resolving the discrepancy will likely require further seasons of parallel counting using both methods before either figure can be treated as definitively more accurate than the other.
What this means for a visitor to the Serengeti
For anyone planning a safari, the practical answer to how many wildebeest are in the Serengeti ecosystem is that the true figure is now genuinely under active scientific dispute, sitting somewhere between just under 600,000 and 1.3 million depending on which survey method is trusted, and that this uncertainty has no bearing on what a visitor will actually see. Even at the lower end of the range, the concentration of wildebeest moving through the Serengeti Mara ecosystem during peak season remains the largest single mammal migration on earth, with herd density in the most concentrated stretches reaching up to a thousand animals per square kilometre, dense enough that the columns are visible from satellite imagery in the first place. Whichever figure eventually proves closer to correct, the spectacle a visitor encounters at a river crossing or on the open plains during calving season is unaffected by which counting method produced the underlying population estimate.
Wildebeest numbers within the wider migration
Wildebeest are the dominant species within the Great Migration but travel alongside substantial numbers of other grazers whose own populations carry separate estimates. Zebra numbers within the migration are most commonly cited at around 200,000 to 300,000, Thomson’s and Grant’s gazelles together are estimated at roughly 300,000 to 400,000, and around 12,000 eland complete the migratory herd. These figures have not been subject to the same satellite recount as the wildebeest population, and the established aerial survey numbers remain the standard reference for these species for the time being. Together with the wildebeest, whichever total proves accurate, the combined migratory herd remains the largest concentration of large mammals moving as a single ecological unit anywhere on the planet.
Why the wildebeest population matters beyond the number itself
The precise headline figure matters less to the ecosystem than the ecological function the population performs, and that function holds regardless of whether the true number sits closer to 600,000 or 1.3 million. Wildebeest grazing pressure shapes the structure and composition of the Serengeti’s grasslands, keeping the plains open rather than allowing them to revert to woodland, and their dung and trampling cycle nutrients back into soil that supports the growth new grass depends on for the following season. The sheer scale of the herd also sustains the ecosystem’s predator populations directly, with lion, hyena, cheetah, leopard and African wild dog numbers across the Serengeti Mara ecosystem all tied closely to the density and movement patterns of the migratory herds they follow. A meaningful long term decline in wildebeest numbers, if the newer satellite figures do eventually prove accurate rather than a product of survey limitations, would carry consequences for every other species in the food chain, not simply for the migration spectacle itself.
Accurate population data therefore has consequences well beyond scientific curiosity. Conservation planning across the Serengeti Mara ecosystem, including decisions about where to permit new lodge development, how to manage livestock grazing along the migration’s edges, and where to prioritise anti poaching patrols, depends on having a reliable sense of how many animals actually depend on a given stretch of habitat at a given time of year. If the true population has genuinely been overestimated for decades, some of that planning may need to be revisited.
Pressures facing the migratory population
Regardless of which population estimate proves more accurate, wildebeest in the Serengeti Mara ecosystem face a set of pressures that were far less significant when the original aerial survey methodology was developed in the 1970s. Habitat fragmentation driven by agricultural expansion, new roads and infrastructure development, and fencing along the migration’s traditional corridors has reduced the total area available to the herds as they move between Tanzania and Kenya each year. Rainfall patterns, which dictate the timing and direction of the entire migration, are also becoming less predictable as the regional climate shifts, and researchers involved in both the aerial and satellite survey work have pointed to this changing rainfall pattern as one plausible explanation for why the herd’s distribution during the 2022 and 2023 satellite survey windows may not have matched the assumptions built into the older aerial methodology.
How population counts are likely to change going forward
The 2025 study is the first time AI powered satellite analysis has been applied to a live count of a migratory ungulate population at this scale, building on earlier work by some of the same researchers using similar techniques to count elephant populations from space. Given the scale of the discrepancy the study has revealed, it is likely that both survey methods will continue to run in parallel for several more years, with researchers seeking to reconcile the two approaches by refining satellite resolution, expanding the surveyed area to better match the herd’s full seasonal range, and cross checking results against GPS collar data from tracked individual wildebeest, which can help establish whether the bulk of the population genuinely fell within the area the satellites covered during the survey window.
Until that reconciliation happens, both figures are likely to continue circulating in parallel, with most tourism marketing and guidebooks retaining the long familiar 1.3 million figure for the time being, even as the scientific literature increasingly treats the true number as an open question rather than a settled fact. Visitors researching the migration in guidebooks, on tour operator websites or in older wildlife documentaries should expect to see the 1.3 million figure for some years yet, simply because it takes time for a figure repeated for half a century to work its way out of common use, even once the scientific consensus behind it has genuinely shifted.
