The mammoth steppe was the largest terrestrial biome on Earth during the late Pleistocene, stretching from western Europe across Siberia and into North America. At its peak, it covered millions of square kilometers of what is now tundra, boreal forest, and temperate grassland, supporting an astonishing density of large herbivores and their predators in conditions that seem, by modern standards, impossibly harsh. This ecosystem vanished not because the land itself disappeared, but because the web of relationships among its plants, animals, soils, and climate unraveled. Understanding how it worked, and how it fell apart, has become unexpectedly relevant to modern concerns about permafrost thaw and carbon emissions.
A Grassland Where You Would Not Expect One
If you visited northern Siberia or interior Alaska today, you would find soggy tundra, mossy bogs, and sparse larch forest. During the last glacial period, those same latitudes looked more like the modern Serengeti or the high plains of central Asia. The mammoth steppe was dominated by grasses and, critically, by forbs, the broad-leaved flowering herbs that are often overlooked in popular descriptions of ancient grasslands. Recent evidence suggests that forbs were not just minor players but a major component of the steppe’s plant community, contributing substantially to its richness and productivity.1Frontiers in Ecology and the Environment. The paradox of forbs in grasslands and the legacy of the mammoth steppe Ancient environmental DNA extracted from ground squirrel coprolites up to 700,000 years old confirms this picture, with Pleistocene-aged samples dominated by graminoids and forbs.2Nature Communications. Ground squirrel coprolites preserve complex archives of ancient environmental DNA over 700,000 years
This plant mix mattered. Forbs have deeper root systems and higher nutritional content than many grasses, and they fix nitrogen through associations with soil microbes. A steppe rich in forbs would have been far more nutritious for herbivores than a simple grass monoculture, helping explain how the ecosystem fed such enormous concentrations of large animals in a cold, dry climate with limited growing seasons.
The Productivity Paradox
For decades, one of the deepest puzzles about the mammoth steppe was how it produced enough food. Climate models of the Last Glacial Maximum (roughly 26,000 to 19,000 years ago) consistently showed low temperatures, low atmospheric CO₂, and weak solar input at high latitudes. On paper, plant productivity should have been dismal. Yet the fossil record shows that the steppe supported woolly mammoths, woolly rhinos, steppe bison, horses, saiga antelope, musk oxen, reindeer, and many other large herbivores in numbers that dwarf what modern Arctic ecosystems sustain. Something was not adding up.
A key part of the answer turns out to be the herbivores themselves. Modeling work has shown that the large average body size of Pleistocene grazers allowed them to exploit grass production far more efficiently than smaller modern herbivores do. The grazers did not just consume the steppe; they amplified it. With grazers factored in, simulated total net primary productivity was about 17% higher than without them, because grazing recycles nutrients, prevents litter buildup, and keeps the vegetation in a young, actively growing state.3PubMed Central. The large mean body size of mammalian herbivores explains the productivity paradox during the last glacial maximum In other words, the animals were not just living off the land. They were co-producing it.
A Crowded Landscape of Herbivores
The mammoth steppe was not simply a grassland with mammoths on it. It supported a guild of herbivores so diverse that different species carved out distinct feeding niches, reducing direct competition and allowing more total biomass to coexist. Woolly mammoths appear to have occupied a particularly unusual niche. Isotopic analysis of individual amino acids from mammoth collagen revealed that mammoths were more enriched in the heavy nitrogen isotope than almost any other herbivore or carnivore living alongside them. That signal points to a distinct food source, possibly plants growing on extremely arid ground, vegetation fertilized by accumulated dung, or a specific plant selection strategy that set mammoths apart from other grazers.4PubMed Central. Solving the woolly mammoth conundrum: amino acid ¹⁵N-enrichment suggests a distinct forage or habitat
Steppe bison and horses, by contrast, were primarily grass consumers. Reindeer browsed on lichens and shrubs at the steppe margins. Saiga antelope favored open, flat terrain and fast-growing herbs. Musk oxen consumed coarser vegetation that other species avoided. This kind of niche partitioning is familiar from modern African savannas, where zebras, wildebeest, and gazelles graze the same grassland but eat different parts of the plant community at different heights and growth stages. The mammoth steppe ran on the same principle, just at higher latitudes and with heavier animals.
Predators and the Top-Down Pressure
Where large herbivores gathered, large carnivores followed. The mammoth steppe’s predator guild was considerably richer and included larger individuals than anything alive today in the same regions. Cave lions, steppe lions, cave hyenas, wolves, and several species of large bear all overlapped in range and competed intensely for prey.
Isotopic evidence from cave lion bones shows that their diets were flexible but could become specialized. Before the Last Glacial Maximum, individual cave lions showed wide dietary variation, with some focused heavily on reindeer and others possibly targeting juvenile cave bears. After the glacial peak, the data converge toward reindeer as the dominant prey item, suggesting that dietary options narrowed as conditions changed.5Quaternary International. Isotopic evidence for dietary ecology of cave lion (Panthera spelaea) in North-Western Europe: Prey choice, competition and implications for extinction The fossil record from over a hundred European cave and open-air sites documents that lions and spotted hyenas competed directly, much as their relatives do in modern Africa. Lions focused on cervids and bovids, while hyenas had partly overlapping prey preferences, and both species raided each other’s kills. At many fossil sites, a small percentage of lion remains show signs of hyena scavenging, and the competition between these predators, wolves, and cave bears in den sites sometimes turned lethal.6Paleontology Journal. Palaeopopulations of Late Pleistocene Top Predators in Europe: Ice Age Spotted Hyenas and Steppe Lions in Battle and Competition about Prey
This predator community had real consequences for herbivore populations. Analysis of predator-prey body mass relationships suggests that Pleistocene carnivores, especially when hunting in groups, could take prey much larger than what modern predators typically tackle. Young proboscideans at their most vulnerable ages fell squarely within the predicted prey size ranges of several Pleistocene carnivores. Because elephants and mammoths have exceptionally long intervals between births, even moderate predation on juveniles could meaningfully limit population sizes.7PubMed Central. The impact of large terrestrial carnivores on Pleistocene ecosystems The mammoth steppe, in short, was regulated from both the bottom up (plant productivity shaped by grazing) and the top down (predators constraining herbivore numbers).
How Animals Engineered the Ecosystem
Modern ecologists increasingly recognize that the mammoth steppe was not simply a habitat that large animals happened to live in. The animals actively maintained it. Herds of grazers trampled moss, broke up shrubs, compacted snow in winter, and recycled nutrients through their dung. These processes kept the vegetation in a grassland state and prevented the slow creep of mosses, shrubs, and eventually trees that would otherwise shade out the grasses and forbs.
This matters enormously for the ground beneath the steppe. Grasslands reflect more sunlight than dark shrubland or forest, keeping surface temperatures lower. In winter, herds trampling and compacting snow reduce its insulating effect, allowing cold air to penetrate the soil and keep permafrost frozen. When the megafauna disappeared, these feedback loops broke down. The tundra and boreal forest that replaced the steppe are warmer at ground level, insulate permafrost with thick moss and undisturbed snow, and accelerate thaw. The implication is that the mammoth steppe was not simply a product of glacial climate; it was a co-creation of climate and megafauna, and it could not survive the loss of either one.8PubMed Central. Pleistocene Arctic megafaunal ecological engineering as a natural climate solution?
Permafrost and a Vast Carbon Store
Beneath the mammoth steppe, millions of years of plant growth and loess deposition built up enormous stores of organic carbon in frozen ground. The most carbon-rich of these deposits are called Yedoma, ice-rich permafrost sediments that accumulated during the late Pleistocene across unglaciated regions of Siberia, Alaska, and parts of northwestern Canada. Yedoma formed through a combination of wind-blown dust, slope wash, snowmelt runoff, and simultaneous ground ice growth, locking plant material into a frozen matrix before it could fully decompose. Estimates suggest these deposits contain up to 130 gigatons of organic carbon, much of it well-preserved and vulnerable to rapid decomposition once thawed.9Earth-Science Reviews. Deep Yedoma permafrost: A synthesis of depositional characteristics and carbon vulnerability
That number is striking in context. The total carbon dioxide currently in the atmosphere represents roughly 900 gigatons of carbon. Releasing even a fraction of Yedoma’s stored carbon through microbial decomposition after thaw would be a significant addition to global greenhouse gas budgets. The carbon in these deposits is not coal-like residue; it includes relatively fresh organic matter, sometimes identifiable as ancient roots and plant fragments, that microbes can begin breaking down quickly once the ice melts.10Geophysical Research Letters. Simulating soil organic carbon in yedoma deposits during the Last Glacial Maximum in a land surface model The mammoth steppe did not just support life above ground; it built a carbon reservoir below ground that we are only now beginning to reckon with.
Reading the Fossil Record Through Unexpected Clues
Reconstructing an ecosystem that vanished thousands of years ago requires creative proxies. Bones and teeth provide isotopic and genetic information, but some of the most revealing evidence comes from sources you might not expect: fossilized dung, ancient coprolites, and the spores of fungi that grow exclusively on herbivore feces.
Sporormiella, a fungus that completes its life cycle on the dung of large herbivores, leaves spores that preserve well in lake sediments. When megaherbivore populations were large and healthy, Sporormiella spore counts in nearby sediment cores are high. When those populations declined, spore counts drop sharply, often before other proxies like pollen or charcoal register any environmental change. This pattern has been documented in Madagascar, where Sporormiella declined steeply around 1,720 radiocarbon years ago, predating the widespread vegetation shifts and increased fire activity that followed megaherbivore loss.11PubMed Central. Sporormiella and the late Holocene extinctions in Madagascar The same approach has been applied across Pleistocene landscapes, providing a timeline for when large herbivore populations collapsed.12Quaternary Science Reviews. Using dung fungi to interpret decline and extinction of megaherbivores: problems and solutions
Ancient DNA from coprolites has become another powerful tool. Ground squirrel coprolites recovered from permafrost deposits in northwestern Canada have yielded complex environmental DNA archives spanning over 700,000 years, with Pleistocene samples dominated by mammoth-steppe flora: grasses, forbs, and a range of other plants that paint a detailed picture of what grew in the immediate area.2Nature Communications. Ground squirrel coprolites preserve complex archives of ancient environmental DNA over 700,000 years These coprolites act as tiny time capsules, preserving not just the squirrel’s own diet but a snapshot of the entire local plant community.
Collapse, Bottlenecks, and the Last Mammoths
The mammoth steppe did not disappear overnight. Its unraveling played out over several thousand years, accelerating around the Bølling-Allerød warm period beginning about 14,700 years ago and continuing through the Pleistocene-Holocene transition. As temperatures rose and precipitation patterns shifted, shrubs and mosses invaded former grassland. The deep, cold, dry conditions that sustained the steppe gave way to wetter, warmer summers and milder winters. Simultaneously, human hunters spread across the steppe’s range.
The genetic record of woolly mammoths tracks this collapse in detail. Microsatellite DNA analysis of northeastern Siberian mammoths reveals two genetically distinct groups separated in time: one before about 12,000 radiocarbon years ago and one after about 9,000 years ago. The transition between them involved a roughly 30% decrease in individual heterozygosity, meaning each animal carried significantly less genetic diversity. Mitochondrial DNA tells a parallel story, with about 65% of mitochondrial diversity lost.13PubMed. Microsatellite genotyping reveals end-Pleistocene decline in mammoth autosomal genetic variation Bayesian analysis of mitogenome data shows the effective population size dropping roughly 15-fold, from around 26,000 individuals before the Bølling-Allerød to around 1,700 by the early Holocene.14PubMed Central. Mitogenome evolution in the last surviving woolly mammoth population reveals neutral and functional consequences of small population size
Continental mammoth populations were gone by roughly 10,000 years ago, but small relict groups held on in island refugia for thousands of years longer. Mammoths survived on St. Paul Island in Alaska until about 5,600 years ago and on Wrangel Island in the Arctic Ocean until about 4,000 years ago, well into the era of Egyptian pyramids.15PubMed Central. Functional Architecture of Deleterious Genetic Variants in the Genome of a Wrangel Island Mammoth These final populations were profoundly inbred, accumulating harmful mutations at an accelerating rate. They were not thriving holdouts; they were genetic ghosts, the last echoes of an ecosystem that had already functionally disappeared.
Collateral Extinctions You Have Never Heard Of
When people talk about Ice Age extinctions, the focus falls on charismatic megafauna: mammoths, saber-toothed cats, giant ground sloths. But the collapse of the mammoth steppe sent shockwaves through far less glamorous corners of the food web. Dung beetles, for instance, co-evolved with large-bodied herbivores over millions of years and depend on appropriately sized feces for reproduction. When the megafauna vanished, so did the dung supply.
European dung beetle assemblages show a significant reduction in average body size over the last 53,000 years, tracking the progressive loss of large herbivores. The down-sizing is consistent with megafauna loss rather than climate warming, because warming alone would not select against larger beetles.16Oikos. Down‐sizing of dung beetle assemblages over the last 53 000 years is consistent with a dominant effect of megafauna losses In South America, the pattern went further: a newly described extinct species of Scybalophagus dung beetle from Chile appears to have gone extinct at the Pleistocene-Holocene boundary, likely as a direct collateral consequence of megaherbivore disappearance during a period of drastic environmental change.17Journal of Quaternary Science. A new extinct species of Scybalophagus dung beetle supports the collateral extinction hypothesis at the Chilean South American Pleistocene–Holocene boundary
These losses matter beyond the beetles themselves. Dung beetles bury and process enormous quantities of feces, aerating soil, recycling nutrients, and reducing parasite loads on remaining herbivores. The loss of the largest dung beetles likely impaired nutrient cycling in post-Pleistocene grasslands, creating a subtler but real degradation of ecosystem function that compounded the more visible disappearance of the megafauna.
Fragments of the Steppe That Still Exist
The mammoth steppe is often described as entirely vanished, but that is not quite accurate. In a few isolated pockets of southern Siberia, remnants of steppe-tundra vegetation persist alongside species considered glacial relicts. The most studied of these refugia is in the Russian Altai Mountains, particularly the Chuya Basin and the Ukok Plateau in the southeast. Habitat modeling of the Last Glacial Maximum shows that desert-steppe dominated most of the non-glaciated Altai during peak cold conditions. When the Holocene brought warmer, wetter climates, most of that grassland converted to forest or forest-steppe, but the southeastern Altai experienced very little habitat change and retained its open, arid vegetation.18Palaeogeography, Palaeoclimatology, Palaeoecology. Modelling the Last Glacial Maximum environments for a refugium of Pleistocene biota in the Russian Altai Mountains, Siberia
Botanical and zoological surveys confirm that both plant and snail species considered glacial relicts still occur across most habitats in Altaian steppe-tundra, though the driest steppe types are avoided by the snail species.19Boreas. A modern analogue of the Pleistocene steppe‐tundra ecosystem in southern Siberia These are not perfect replicas of the mammoth steppe. They lack the megafauna, and their species composition has inevitably shifted over ten thousand years of isolation. But they provide a living laboratory for understanding what mammoth-steppe vegetation looks like when the soil, altitude, and aridity happen to resist the Holocene’s encroaching forests.
Pleistocene Park and the Rewilding Hypothesis
The idea that large herbivores maintained the mammoth steppe has an obvious modern corollary: could reintroducing herds of grazing animals to the Arctic re-create some of the steppe’s effects? That is the premise behind Pleistocene Park, an experimental reserve in northeastern Siberia where bison, horses, musk oxen, yaks, and other large herbivores have been released into a fenced area of former tundra. The goal is not nostalgia. It is to test whether herbivore activity can convert mossy, shrub-dominated tundra back into grassland, and whether that grassland can slow permafrost thaw.
The theoretical case is surprisingly strong. Modeling indicates that increasing herbivore density in northern ecosystems would compact snow in winter, reducing its insulation and allowing cold air to reach the soil. Under this scenario, roughly 80% of permafrost in the top ten meters could persist until 2100, even under aggressive warming projections, with average permafrost temperatures remaining below −4°C.20Nature / Scientific Reports. Protection of Permafrost Soils from Thawing by Increasing Herbivore Density Grassland-dominated systems could further help by capturing more carbon through the deep root systems of grasses and forbs, while reducing the methane emissions associated with waterlogged tundra.8PubMed Central. Pleistocene Arctic megafaunal ecological engineering as a natural climate solution?
The practical obstacles are immense. Stocking herbivores at the densities needed to transform vegetation across millions of square kilometers of Arctic and sub-Arctic land would require breeding and transporting animals on a scale that has no precedent. Predators would need to be part of the system eventually, or herbivore populations would overshoot and crash. And there is no guarantee that modern proxy species can replicate the ecological effects of extinct ones: a European bison is not a steppe bison, and a Yakutian horse is not a Pleistocene horse, even if they fill broadly similar grazing roles. Still, early results from Pleistocene Park and similar experiments show measurable changes in vegetation composition and snow compaction within the fenced areas. The mammoth steppe may be lost, but the principles that sustained it are being tested on the ground where it once stood.