Rewilding
in a Hurry

The Whole Thing in One Page

Rewilding is usually pictured as a dramatic release: wolves step from crates, bison cross a plain, fences come down and nature resumes. The picture is attractive and incomplete. Rewilding is less about returning animals than returning work. It asks whether water, fire, grazing, predation, decomposition, migration and succession can again shape a place without people prescribing every result.

That changes the object of conservation. A landscape can look green while depending on mowing, draining, planting, feeding, culling and repeated rescue. It may protect rare species and still be ecologically dependent. Rewilding asks a different question: what processes are missing, blocked or weakened, and what would become possible if they returned? Its target is not a preferred photograph. It is a living system with more capacity to choose its own next state.

Sometimes the strongest action is subtraction. Stop ploughing a marginal field and succession begins. Block drains and a peatland can hold water again. Remove an obsolete weir and fish regain a migration route. Sometimes restraint is insufficient because the ecological worker has disappeared. A beaver moves water, wood and sediment. Large herbivores open vegetation and carry nutrients. Predators alter prey numbers, behaviour, carcass supply and competition. Animals are processes with legs.

None of this produces a guaranteed route back. There is no correct historical date to restore. Climate has shifted, species have vanished, land has owners, and many landscapes valued as natural were shaped by people for centuries or millennia. Rewilding uses the past to diagnose losses, then restores functions and room for change under present conditions. The result may resemble no previous moment exactly.

Room matters. A wolf population cannot persist in a fenced field. A river cannot behave like a river if every flood is treated as failure. Populations need movement, disturbances need space, and ecological effects cross property lines. Small projects can still restore useful processes, but the less room a system has, the more human control remains hidden inside it.

The famous Yellowstone story shows both the promise and the danger. Returning wolves restored predation and altered a network of interactions. It did not create one clean chain in which wolves fixed elk, trees, beavers and rivers by themselves. Hunting, bears, climate, forage and hydrology mattered too. Rewilding works through networks, not a row of dominoes.

The human part cannot be removed. A beaver wetland may reduce risk downstream and flood one field. A predator may enrich a region and kill one farmer's sheep. A restored fish population may require limits on catches. Rewilding that transfers ecological costs to people with no voice will be reversed by politics, however sound its biology.

The central paradox follows. Rewilding aims to reduce human control, yet reaching that point usually requires deliberate human action: change law, remove a barrier, restore a species, compensate a neighbour, monitor welfare, share authority and decide when to stop intervening. Better governance is what makes less ecological management possible. The destination is more self-directed nature. The route is better management of our own management.

That is the book.

Why You Should Care

In February 2025, England opened a legal route for licensed releases of Eurasian beavers into the wild. Its importance lay in the work the animal performs. A beaver fells trees, digs canals and builds dams. Water slows. Ponds appear. Sediment settles. Wet ground spreads sideways. A stream managed as a drainage pipe begins to make choices again.

Some of those choices are useful to people. Beaver wetlands can hold water, create habitat and change the timing of downstream flows. Some are inconvenient. A dam may flood a field, block a culvert or drown a tree somebody valued. The act can be ecological recovery from one viewpoint and property damage from another. That is rewilding in miniature: the return of a process means the return of consequences that no longer stop at the reserve boundary.

Most modern landscapes hide how much work people do on nature's behalf. Pumps keep peat dry. Embankments keep rivers in channels. Mowers hold grass at one height. Culls substitute for absent predators. Hatcheries replace failed reproduction. Trees are planted where seeds no longer arrive. Fire is suppressed until fuel accumulates. These interventions may be necessary, and some support biodiversity. Yet a system that loses its ecological functions when the annual management budget disappears has not recovered much agency.

Rewilding gives you a way to see that distinction. It separates abundance from autonomy. A place can contain many species while depending on continual human instruction. Another can look rough, ordinary or unfinished while water, vegetation and animals are beginning to regulate one another again. This does not make the second place automatically better. It makes the question sharper: are we preserving a result, or restoring the capacity to produce results?

That question matters because governments and organisations are committing to restoration at scale. The global biodiversity framework calls for at least 30 per cent of degraded terrestrial, inland-water, coastal and marine ecosystems to be under effective restoration by 2030. The European Union's restoration law sets an overarching objective for measures across at least 20 per cent of EU land and sea by 2030. Rewilding is only one approach within that larger effort, but its language will be used to justify land purchases, species releases, river works, public spending and changes to livelihoods.

The idea is unusually easy to romanticise. Wolves, bison and wild horses photograph well. A blocked drain, an oyster reef, a compensation scheme and a twenty-year monitoring programme do not. The visible animal can therefore become the whole story while the conditions that determine success remain off camera. Reintroducing a species into the causes that removed it is not recovery. It is another failure with better publicity.

Rewilding also forces an uncomfortable argument about what counts as nature. Should a chalk grassland be allowed to become scrub if grazing stops? Is a cattle breed an acceptable substitute for an extinct wild herbivore? Does cultural burning make a landscape less wild, or does excluding it erase the disturbance regime that kept the ecosystem alive? Which historical baseline deserves authority when climate is moving species beyond old ranges? There is no universal answer, which is why process, scale and governance matter more than a single visual ideal.

The promise of rewilding is not that nature will solve every environmental problem once people leave. People are not leaving, and many systems cannot recover unaided. The promise is that some ecological work can be returned to living systems, making recovery more dynamic and less dependent on endless prescription.

After this book, a green landscape will no longer look complete merely because it is green. You will notice who is doing the work, what is prevented from moving, which disturbance has been suppressed, who bears the cost of recovery and whether the system could continue without another meeting. You stop asking whether a place looks wild and begin asking what it can do for itself.

The Core Ideas

Nature Is a Set of Processes, Not a Picture

Imagine two grasslands. The first is clipped, flower-rich and full of insects. It is cut on schedule, scrub is removed, drainage is maintained and grazing is adjusted to preserve a chosen community. The second is rougher. Some patches are grazed hard, others are tall, thorn scrub is spreading, dead wood lies where it fell and wet hollows appear after rain. Which is wilder?

The question has no answer until you say what wildness means. The first may support rarer species and have greater conservation value. The second may contain more self-directed ecological change. Rewilding begins by separating those qualities. Biodiversity asks what lives there. Ecological autonomy asks how much of the system's future is generated by its own processes.

The word rewilding emerged through several lineages. North American advocates in the 1990s stressed large protected cores, connections between them and the return of large carnivores. The formula was memorable because fragmentation and predator loss were obvious failures across a continent divided by roads, farms and settlements. European practice developed in denser cultural landscapes, where restoring river movement, natural grazing, woodland regeneration and wildlife passage often mattered more than drawing a border around an empty wilderness. Later scientific frameworks broadened the idea towards ecological complexity, connectivity, dynamic change and reduced continuing control.

This process-centred definition matters because appearances can mislead. A plantation is green, yet every row may have been selected, planted and maintained to specification. A wetland can contain water because pumps keep it at a fixed level. A reserve can protect birds through annual mowing that substitutes for lost grazing. None of those methods is wrong. They are forms of management with known objectives. Rewilding asks whether some of the ecological work can be restored instead of permanently imitated.

Disturbance belongs inside that model. Rivers migrate. Trees fall. Storms open canopies. Fires burn. Herbivores browse young growth. Predators kill. Drought removes weak individuals. Carcasses feed scavengers and microbes. Succession changes communities after a field, fire or flood. Nature is not a collection of stable objects. It is a set of recurring operations whose combined effects produce habitat.

A serious project therefore starts with a functional diagnosis. If a floodplain is cut off by embankments, the missing process is periodic inundation. If deer prevent every tree from reaching adulthood, the missing pressure may be predation, hunting or another check on herbivory. If a forest is isolated by roads, the missing process is movement. If fish cannot reach spawning ground above an obsolete weir, stocking more fish treats the symptom while leaving the broken route untouched.

This is also why tree planting is not a synonym for rewilding. Planting can be excellent restoration. Yet where seed sources, soils and dispersers remain, removing grazing or repeated burning may allow woodland to regenerate with less prescription and more variation. In other places, planting is necessary because the ecological memory has gone. The method follows the bottleneck.

The distinction can be expressed as composition, structure and function. Composition is which species are present. Structure is how vegetation, water and habitat are arranged. Function is what the system does. Rewilding gives unusual weight to function because restored processes can keep generating new composition and structure after the project team has stopped choosing them.

The first change, then, is a change of target. Rewilding does not begin by asking what the landscape should look like. It asks what the landscape should be able to do.

The First Tool Is Often Restraint

Conservation sounds active. Restore, plant, breed, move, fence, manage, control. Rewilding added an awkward verb: stop.

Passive recovery is the least theatrical form of the idea and can be the most powerful. Stop ploughing a marginal field and seeds arrive from hedges, soil banks and passing animals. Stop draining a peatland and water can return. Stop mowing a riverbank and vegetation develops structure. Remove an obsolete barrier and fish regain kilometres of river. Reduce repeated disturbance and succession resumes.

This is not the same as neglect. Abandonment is what happens when management disappears without a diagnosis, responsibility or plan. Deliberate restraint begins by asking what pressure prevents recovery, what risks might follow its removal and what evidence would show whether the system is responding. Choosing not to intervene can be a demanding intervention because the manager must tolerate uncertainty and resist tidying every unfamiliar change.

Natural regeneration shows both the promise and the limit. Where soils remain functional, native seed sources are nearby and browsing is moderate, woodland can return without mass planting. Pioneer vegetation changes light and temperature. Birds and mammals move seeds. Shrubs protect young trees. Dead material accumulates. Later species establish through pathways no planting plan could fully prescribe. Assisted natural regeneration targets the obstacles: control repeated fire, reduce grazing at a critical stage, suppress a dominant weed or add scattered seed sources, then let succession do most of the work.

The phrase where conditions allow carries the entire argument. A mine spoil may be toxic. A field surrounded by intensive agriculture may receive few seeds. Heavy deer browsing may remove every sapling. An invasive plant may capture the released space. A degraded peatland may continue drying until drains are physically blocked. Restraint works when the system retains enough memory and connectivity to respond. Where those have been erased, the first phase may require engineering, planting or translocation.

Restraint can also damage valued habitats. Many European grasslands, heaths and wood pastures contain species that prospered under long histories of grazing, cutting or burning. Stop all management and scrub may close over them. Calling the change natural does not save the orchids or butterflies that disappear. Rewilding must decide whether its objective is process autonomy, particular species, cultural continuity or some negotiated combination. Those goals can conflict.

Knepp in West Sussex became the best-known British example because the transformation was visible. Intensive arable and dairy farming on difficult clay soils had become financially strained. From around 2000, the estate shifted roughly 3,500 acres towards process-led recovery. Cropping declined, water was restored, thorny scrub spread, dead wood remained and free-roaming cattle, ponies, pigs and deer created varied grazing and disturbance. Nightingales, turtle doves and purple emperor butterflies became part of the public story.

Knepp is a case, not a universal recipe. Its ownership, soils, surrounding landscape, finances and English ecology are specific. Its large herbivores are fenced, monitored and partly domestic proxies. Decisions about stocking, welfare and income still occur. The achievement is not human withdrawal. It is a shift from prescribing the vegetation map towards creating conditions in which ecological interactions produce more of it.

Restraint also changes the manager's relationship with surprise. Conventional management often treats deviation from the plan as failure. Process-led recovery expects some outcomes to be unplanned. That does not excuse harm. It means the project needs thresholds for intervention without turning every unexpected species, flood or vegetation change into a reason to reclaim full control.

Restraint is therefore a tool rather than a creed. Rewilding asks which controls can be removed, which must remain for now and whether the need for them declines through time.

Animals Are Infrastructure

A beaver is not merely an animal living beside water. It is moving hydrology.

A dam slows a channel. Water spreads into side pools. Sediment settles. Canals extend wet ground. Trees are felled and light reaches the bank. Dead wood enters the stream. Some organisms gain habitat and others lose it. The animal changes the physical conditions in which a community develops. Ecologists call this ecosystem engineering. Rewilding treats it as infrastructure performed by a living body.

The effects are not identical everywhere. A beaver dam in a narrow headwater catchment behaves differently from one beside a road or on low farmland. Evidence that beavers can attenuate some flood peaks is promising and dependent on location, dam configuration and scale. Calling them natural flood defences can therefore mislead. They are agents with many effects, some welcome and some requiring management.

Large herbivores build landscapes by subtraction and movement. Grazing suppresses some plants and releases others. Browsing keeps saplings short. Hooves break turf and expose soil. Dung carries nutrients and feeds invertebrates. Wallowing creates wet hollows. Hair, seeds and microbes travel on bodies. Carcasses support scavengers and decomposers. When herds move unevenly through a large area, they can help create mosaics of short grass, scrub, woodland gaps and dense cover.

Frans Vera made this idea influential in Europe by challenging the assumption that postglacial lowland landscapes would have been almost entirely closed forest without people. He argued that wild cattle, horses and deer maintained more open, park-like mosaics. The historical extent of openness remains debated. Rewilding does not require the strongest version of Vera's reconstruction. It requires the well-supported fact that herbivore abundance and behaviour can reshape vegetation, and that their loss or confinement changes the system.

Predators perform work through several routes. They kill prey. They alter where and when prey feed. They suppress or compete with smaller predators. Their carcasses and kills feed ravens, eagles, bears, beetles and microbes. The effects depend on productivity, habitat, alternative prey, human hunting and the ability of animals to move. A predator has no single ecological job.

The same principle extends below the glamour line. Earthworms mix soil. Parrotfish grind reef material. Salmon carry marine nutrients inland when they die after spawning. Seabirds move nutrients from sea to island. Oysters filter water and build three-dimensional reefs. Dung beetles bury waste. Pollinators connect plants that cannot move. Remove enough ecological workers and a landscape may retain its scenery while losing the operations that maintain it.

Work is distributed, which is why the word keystone needs care. Some species have effects far beyond their abundance, but no organism runs an ecosystem alone. If the habitat, prey, water or movement route is gone, returning one celebrated worker cannot rebuild the missing workplace. Rewilding succeeds by restoring relationships among workers, not by appointing an ecological chief executive.

This makes species reintroduction more demanding than historical presence plus public enthusiasm. The cause of the original loss must have changed. Habitat, food, genetics, disease, welfare, source populations, road risk and social acceptance all need scrutiny. A species placed back into the conditions that removed it has not been restored. It has been scheduled for another failure.

Animals are infrastructure because they perform repeated work across time and space. Rewilding becomes powerful when it restores workers and relationships rather than collecting specimens.

Predators Matter, but Trophic Cascades Are Not Magic

Few ideas have sold rewilding as effectively as the trophic cascade. Restore a predator, reduce or move the herbivore, release the plants, rebuild habitat. The mechanism is real, repeatedly observed in some systems and dangerously easy to turn into a universal story.

A trophic cascade occurs when change at one level of a food web produces effects at other levels. Sea otters eat sea urchins; fewer urchins can allow kelp to persist. Large carnivores can reduce herbivore abundance or change where prey feed, with consequences for vegetation. They can suppress medium-sized predators, changing smaller prey. The pattern depends on the strength of each link and on all the other forces acting at the same time.

Yellowstone is useful because it refuses to become a clean parable. Wolves were exterminated from the park by the early twentieth century and reintroduced in 1995 and 1996. Elk numbers and distribution changed afterwards. Predation returned. Coyotes were affected. Carcasses became available to scavengers. Aspen and willow recovered in some locations, and beaver colonies later increased.

The compressed popular version gives wolves sole credit for a valley-wide chain that regenerated trees and changed rivers. That claim is too neat. Elk were also affected by bears, cougars, hunting outside the park, severe weather, forage and migration. Vegetation responses differed among sites. Hydrology influenced whether willow or aspen could escape browsing. Recent studies have strengthened the case for substantial aspen recovery after carnivore restoration, while other researchers continue to dispute estimates of the cascade's strength and the share attributable to wolves.

The disagreement improves the lesson. Rewilding does not insert one keystone and receive a predictable result. It restores interactions inside a network. The network may produce large indirect effects, weak effects, delayed effects or outcomes concentrated in particular places. A viral story recruits support, then creates an easy target when its single cause fails to explain the whole system.

The cascade is strongest as a hypothesis that names links to test. Did browsing fall where predation risk or prey mortality rose? Did plants escape into taller size classes? Did the effect persist across wet and dry years? Which rival explanations remain? The mechanism becomes more useful when it produces measurements rather than a slogan.

Predators also expose the political boundary of rewilding. A wolf does not recognise a park boundary. A lynx does not understand livestock insurance. Large carnivores need territories, prey and movement routes that cross ownerships and jurisdictions. Ecological success can therefore create political failure unless coexistence is designed before and after release.

Prevention matters: guarding dogs, shepherding, night enclosures, carcass management, fencing and rapid response can reduce losses. Compensation and insurance can distribute the remaining cost. None removes conflict. Nor should the cost be hidden. If a wider public wants predators for ecological and cultural reasons, the small number of people living beside them cannot be expected to finance the policy through dead livestock and anxiety.

Predators restore a missing relationship: being eaten. That relationship can change ecosystems in ways culling and fencing cannot fully imitate. The correct model is a network of causes, not a line of dominoes. Good science becomes more necessary when the story is exciting.

Space and Connectivity Decide What Can Become Wild

A nature reserve can protect a rare plant in a few hectares. It cannot contain every animal, flood, fire or migration that shaped the plant's ecosystem.

Rewilding becomes harder as processes range farther. Wolves can travel tens of kilometres. Rivers connect mountains to estuaries. Migratory fish require passage through whole catchments. Wildfire crosses property lines. Young animals disperse beyond their parents' territories. Genetic exchange requires populations to meet. A protected patch can be excellent habitat and still function as an ecological island.

Connectivity therefore sits near the centre of rewilding. It can mean a continuous corridor, a removed fence, an unfenced river, road crossings, stepping-stone wetlands, hedges through farmland or a working landscape permeable enough for movement. The purpose is not to draw green lines on a map. It is to restore movement as an ecological process.

Movement performs several jobs. Individuals find mates. Populations recolonise after local loss. Species track climate and seasonal food. Genes move between groups, reducing isolation. Predators and herbivores can shift rather than exhausting one enclosed patch. A corridor that looks connected to a person may still fail for a dormouse, salmon or wolf, so useful connectivity is measured by movement rather than appearance.

Space is also part of population regulation. In an open system, animals can migrate when food declines. In a fenced reserve, the same population encounters a hard boundary. A policy of minimal intervention can then cause suffering because the system is too constrained to regulate itself through movement, dispersal or predation.

Oostvaardersplassen in the Netherlands became the warning. The reserve lies on reclaimed land and became an influential experiment in low-intervention management with Heck cattle, Konik horses and red deer. Large herbivore numbers rose inside a fenced, predator-poor system. During the severe winter of 2017 to 2018, 3,226 animals died, with most killed by managers when their condition became poor. Public anger and official review led to a new policy with lower target numbers, earlier intervention and stronger welfare safeguards.

The episode is often used as a verdict on rewilding. Its more useful lesson concerns geometry and responsibility. Natural mortality exists, but a human-created enclosure is not an open ecosystem merely because feeding and culling are reduced. Managers still chose the fence, the species, the absence of predators and the limits on dispersal. Non-intervention inside artificial boundaries does not remove human responsibility for those boundaries.

Autonomy is therefore scale-dependent. A garden can increase wildness, yet it cannot support a viable population of large carnivores. A restored river reach can improve habitat, while an upstream dam still controls sediment and migration. A farm can become a rich stepping stone, while surrounding roads continue killing dispersing animals. Small projects matter when their claims match the processes their scale can support.

Connectivity can also carry harm. Disease, invasive species and fire move through connected systems. That does not make fragmentation desirable. It means connection must be designed with the relevant process in mind. A fish passage, wildlife bridge or permeable farming matrix is a hypothesis about movement, and its benefits and risks require monitoring.

Rewilding has a geometry. Cores provide room. Corridors provide routes. Permeable farms, towns and coasts connect them. The test is whether organisms and disturbances can complete their work across the landscape, rather than remain trapped inside attractive demonstrations.

There Is No Correct Historical Baseline

Putting nature back raises an immediate question: back to when?

Before industrial farming? Before European settlement? Before agriculture? Before humans? Before the last ice age? Each date describes a different climate, species mixture and pattern of human influence. Choose one without argument and rewilding becomes historical theatre with animals.

Restoration often uses reference ecosystems: surviving or reconstructed examples that indicate what composition and processes might be expected under particular conditions. Rewilding also needs history. It tells us whether beavers were hunted out, rivers straightened, predators exterminated, wetlands drained or grazing intensified. Present conditions are not neutral merely because they are familiar.

History cannot provide a single permanent target for three reasons. First, ecosystems have always changed. Climate shifted, seas rose, species migrated, disease arrived, storms and fires reset vegetation, and human communities altered land and water long before industrial society. The image of one untouched state waiting behind disturbance is often false.

Second, extinction has removed options. Europe cannot return the aurochs because the species is gone. Some projects use robust cattle as ecological proxies for parts of its grazing function. Proxy rewilding separates function from identity: a living substitute may browse, trample and move nutrients without recreating the extinct animal. That can restore useful work, but it also introduces choices about resemblance, risk and welfare. Calling a proxy ancient does not make the decision objective.

Third, climate change is moving the target. A plant community suited to the nineteenth century may not persist under late twenty-first-century heat, drought or fire. Holding historical composition in place can demand increasing control. A process-based strategy may favour connected habitat, structural variety and room for species to move, even when the future community has no exact precedent.

The past remains indispensable. It identifies lost relationships and prevents recent degradation becoming the baseline against which improvement is judged. Fisheries scientists call this shifting baseline syndrome: each generation accepts the depleted abundance of its youth as normal. Rewilding resists that amnesia by recovering evidence of what once moved, bred, flooded, burned and fed there.

A useful baseline is therefore plural. Archaeology, old maps, pollen, bones, written records, living reference sites and local knowledge each reveal part of the range. The project then adds a future test: can diverse life and key processes persist under changing conditions with less continuing control? Authenticity becomes evidence, not a cage.

Yet historical reference can carry political danger. Landscapes described as wilderness may have long Indigenous histories of burning, hunting, harvesting, planting and stewardship. Excluding people to create an imagined pristine state can repeat fortress conservation under a newer name. In many regions, recovery depends on Indigenous authority and knowledge because the desired processes include human-set fire, seasonal harvest or relationships with animals that colonial policy tried to break.

The Iinnii Initiative led by Blackfoot nations makes the point. Returning buffalo to traditional lands concerns grazing and nutrient movement, but also food, ceremony, sovereignty, education and cross-border governance. Reducing the project to a species release would miss the institution that makes ecological return possible.

Rewilding uses history diagnostically rather than devotionally. The past reveals losses and possibilities. It does not supply a photograph that living systems are obliged to reproduce.

The End Point Is Coexistence, Not Human Absence

The first Core Idea changed the target from a picture to a process. Follow that logic to its end and rewilding becomes a negotiation about control.

Humans dominate most accessible ecosystems through farming, fishing, forestry, dams, roads, drainage, hunting, fire suppression, pollution and deliberate species movement. Rewilding cannot erase that history. It asks whether some control can be surrendered so that ecological processes regain room to act.

Surrendering control is a social choice. A beaver dam creates wetland and a claimant. A returning jaguar restores predation and changes how livestock owners judge risk. Scrub regeneration creates cover and closes a familiar view. A river allowed onto its floodplain can protect a town downstream while taking one field out of production upstream. Ecological benefits and local costs are rarely distributed to the same people.

That is why modern rewilding guidance includes coexistence, rights, knowledge and adaptive management. These are not polite additions to the biology. A predator population that neighbours continue to kill cannot become self-sustaining. A wetland drained after a political backlash is not restored. A corridor lost when one landowner changes policy was never secure. Social legitimacy is part of ecological persistence.

Iberá in Argentina shows the institutional work. Jaguars had disappeared from Corrientes for about seventy years. Recovery required protected habitat, prey, captive breeding, release methods, monitoring and public acceptance. It also became linked to local tourism and a wider regional story about wildlife. By the end of 2024, Rewilding Argentina reported 33 free-living jaguars in Iberá and twelve cubs born during that year. The number matters less than the transition it represents: a breeding population is no longer a collection of released animals. It is a new neighbour whose range, conflicts and economic effects extend beyond a project team.

The Blackfoot-led return of buffalo offers a different model. Ecological restoration is tied to tribal authority, cultural renewal, food sovereignty and the repair of relationships damaged by colonial removal. Here people are not stakeholders invited into somebody else's wild project. They are authors of the restoration. That distinction matters wherever conservation has treated local presence as the problem to be removed.

Economic design matters too. Rewilded landscapes may support tourism, lower-input grazing, fisheries recovery, flood storage or payments for ecosystem services. None is guaranteed. Tourism can be seasonal. Carbon finance can favour tree cover over open habitat. Land values can rise and displace residents. A philanthropic demonstration may not provide a durable income model for neighbours. Benefits must be credible to those carrying costs.

Authority needs the same attention as income. Participation can mean receiving information, being consulted, sharing a decision or holding the right to decide. Those are not equivalent. Projects that use the language of community while retaining every decisive power reproduce control in a social form even as they claim to surrender it ecologically.

Good governance therefore prices conflict early. It establishes who decides, who pays, what loss triggers compensation, which risks remain unacceptable and how rules change when animals disperse or rivers move. It gives communities authority to shape the project rather than presenting consultation after the central choices have been made.

The causal loop now closes. Excessive human control can simplify ecosystems until they require permanent management. Restoring autonomy demands deliberate institutions that protect movement, distribute costs and define boundaries, allowing ecological control to decline later. Human agency creates the conditions for greater non-human agency.

The wildest future available on a crowded planet will not be empty of people. It will be a future in which people remain present but stop insisting on being the author of every ecological result.

How It Actually Works

Diagnose the missing process

A serious rewilding project starts with causes, not a shopping list of species. What has changed? Is the land drained, fragmented, polluted, repeatedly burned, overgrazed, stripped of herbivores, stripped of predators, cut off from seed sources, blocked by dams or locked into a management regime that prevents succession?

The same visible symptom can have several causes. A treeless hillside might be naturally open, maintained by grazing, repeatedly burned, too wet for woodland, too exposed, distant from seed or browsed by deer. Planting trees without identifying the cause can create a new problem while looking impressively active. A river without salmon may lack clean gravel, passage, cool water or adults. More hatchery fish do not repair all four.

The diagnosis joins ecology to social reality. Who owns and uses the land? What happens upstream and downstream? Which roads, fences and laws constrain movement? Which species and processes have been removed? What risks are unacceptable? Who could lose income, access or safety? What evidence would count as recovery?

The answer should produce an intervention hierarchy. Remove a pressure where possible. Repair physical processes where necessary. Reconnect habitat. Allow recolonisation and regeneration. Move a species only when it cannot return unaided and the causes of its loss have changed. The sequence is a discipline against starting with the photogenic part.

Objectives must also be stated in verbs rather than scenery. Restore seasonal flooding. Increase safe dispersal. Reduce browsing enough for trees to recruit. Recover natural age structure in a fish population. Such goals can be tested. A promise to make a valley wild contains no mechanism, no threshold and no way to learn from failure.

Remove the barriers

Many projects begin by taking something away. Drains are blocked. Fences are opened or moved. Obsolete weirs are removed. Intensive grazing stops. Repeated mowing ends. Fertiliser and pesticide use falls. Hunting pressure changes. A road crossing is built so animals can survive movement between habitats.

Removal can release a process quickly. Water returns to a peatland as soon as drainage loses its grip, though vegetation and carbon dynamics may take much longer to recover. Fish can use reopened river kilometres within a season if suitable habitat remains above the barrier. A former field can move into succession as seeds arrive and competition changes.

Other legacies persist. Decades of fertiliser can keep soils nutrient-rich, favouring a few competitive plants. Peat can have subsided or oxidised. Seed banks may be depleted. Contaminants remain after the factory closes. A road continues fragmenting movement even when traffic is reduced. Rewilding therefore distinguishes removing the current pressure from repairing the damage it left.

The best intervention may be modest and temporary. A fence can protect young woodland until trees escape browsing, then come down. Scattered planting can provide seed sources without fixing every future tree. An invasive plant may need several years of control before native succession can proceed. Success means the scaffold becomes less necessary, not that every action must be absent from the first day.

The sequence also protects what survives. Removing a pressure across a whole site can damage remnant populations adapted to the old regime. Managers may retain refuge patches, phase change or move slowly enough to observe response. Rewilding values self-direction, but it should not sacrifice irreplaceable species merely to make the intervention look purer.

Give water room

Water reveals the method because it ignores administrative convenience. Rivers transport sediment, erode banks, deposit gravel, flood side channels and connect upstream decisions to downstream consequences. Straightening, embanking and draining simplify those operations so land can be farmed or built upon. The river becomes faster, narrower and more predictable until a flood exceeds the design.

Process recovery may involve removing a weir, setting embankments back, reconnecting an old channel, blocking ditches, restoring woody material or allowing erosion within a defined corridor. None guarantees a return to a historic river. It restores degrees of freedom. Water can occupy more than one path, sediment can move, and flood energy can spread across land chosen to receive it.

Beavers add a mobile version of the same work. Their dams can create wetlands and local water storage without a construction plan. They also create disputes when water crosses a property boundary. Wild release therefore requires more than finding suitable habitat. It needs local management groups, clear licensing, rapid advice, lawful options for protecting infrastructure and a way to decide when a dam remains or is altered.

Hard limits remain. Houses, roads and high-value farmland may require protected banks, culverts or exclusion zones. Rewilding a river rarely means accepting uncontrolled flooding everywhere. It means deciding where the river can regain movement rather than forcing it to behave like a pipe along its entire course.

Catchment order matters. Work downstream may fail while erosion, pollution or rapid runoff continue upstream. One restored reach can still provide habitat, but system recovery follows water across ownerships. Agreements over farming, forestry, drainage and abstraction may therefore do more than a beautiful engineering scheme confined to one parcel.

Allow succession, then read it correctly

On former farmland, stopping production rarely produces instant woodland. Fertile soils can first yield dense grasses, docks and thistles. Thorny scrub follows where seeds arrive and browsing allows it. Young trees may establish under the protection of bramble or hawthorn. Paths become rough, familiar views close and dead material accumulates. The site can look neglected during the stage when ecological structure is returning.

Managers then face choices. A designed plantation reaches tree cover quickly. Spontaneous regeneration can produce greater variation in age, spacing, scrub, gaps and genetics, but may be slow or fail. Assisted regeneration targets bottlenecks: reduce deer, introduce scattered native seed, control invasive plants or protect wet ground. If open habitat is part of the objective, grazing can interrupt succession and create a shifting mosaic.

Time is a management problem. Grants and political terms run for a few years; succession can take decades. Projects that promise immediate spectacle often lose support before the system matures. Good communication explains what change should look like at each stage, which outcomes remain uncertain and what would trigger intervention. A sign saying nature is taking over explains little. A sign explaining why thorn scrub protects saplings gives people a mechanism to watch.

The correct reading is therefore neither every change is success nor every untidy patch is failure. Monitoring asks whether desired processes are strengthening: seeds arriving, age structure diversifying, water persisting, browsing changing, dead wood accumulating, animals moving. Appearance is evidence only when linked to a mechanism.

Herbivores can redirect this sequence. Too few may allow rapid canopy closure where open patches are desired. Too many can stop recruitment across the site. Variable numbers and movement matter more than a fixed stocking recipe. Managers often need several years of observation before they know whether grazing is restoring heterogeneity or replacing one rigid regime with another.

Restore connectivity before spectacle

A release site is not a landscape. A population needs room to disperse, find mates, recolonise after local losses and track changing resources. Corridors, stepping stones and permeable working land are useful only if organisms use them. A hedge can connect woodland for one species and remain a barrier to another. A fish pass that most fish cannot ascend is green architecture rather than connectivity.

Planning therefore follows the animal or process beyond the project boundary. Where will a young lynx go? Which roads will it cross? Can a salmon reach cold tributaries during a heatwave? Does a wetland network remain connected in drought? Can grazing animals shift when forage falls? Which neighbouring landowner controls the narrowest link?

This is often less glamorous than release. It involves easements, road design, catchment agreements, planning rules and relationships among owners. Yet movement decides whether a restored population becomes resilient or remains a managed island. A corridor lost to one development can undo years of breeding work.

Connectivity also permits climate adaptation. As temperature and rainfall change, species need routes towards suitable conditions. Historical restoration without future movement can trap organisms in a faithful reconstruction of a climate that is disappearing. Rewilding gains durability when the landscape offers options rather than one designated refuge.

A network is only as strong as its narrowest repeated failure. One busy road, poisoned stream, impermeable fence or hostile district can sever movement across a much larger area. Mapping should therefore combine habitat with mortality and behaviour. The cheapest hectare to protect may be less important than the awkward crossing every dispersing animal must survive.

Reintroduce with an exit strategy

When a species cannot return on its own, translocation may restore a missing ecological actor. The release is late in the process, after questions that publicity often skips. Why did the species disappear? Is habitat sufficient? Are prey or food secure? Are founders genetically suitable and healthy? What disease risk exists? How will animals be transported, released and monitored? What happens if they disperse into danger or conflict becomes severe?

Iberá's jaguar recovery shows the sequence. The species had been absent from Corrientes for about seventy years. Work began with protected habitat, prey recovery, breeding, local engagement and the legal and practical machinery for release. Founders and offspring were prepared for wild life, then tracked after release. Reproduction in the wild changed the project from returning individual animals to living with a growing population.

That transition matters. A small founder group is a conservation operation. A breeding population becomes a regional institution. Animals cross the administrative map. Some enter unsuitable areas. Livestock owners require credible prevention and compensation. Roads and persecution become continuing risks. Monitoring shifts from whether released animals survive to whether the population disperses, reproduces and interacts without permanent intensive control.

Every translocation needs an exit strategy. Managers must know what failure would look like, when supplementary releases stop, when an animal may be removed and which interventions are temporary. Rewilding does not fail because people remain responsible. It fails when a supposedly wild population can exist only through endless control that prevents it behaving as a population.

Founder choice has long consequences. Too few or closely related animals can leave genetic weakness. Animals raised without appropriate behaviour may struggle after release. Disease screening protects both the released population and resident wildlife. Welfare must be assessed during capture, transport, acclimatisation and the period when dispersal exposes animals to roads, hunger and conflict.

Restore relationships at sea

Marine rewilding exposes the weakness of the fence-and-release image. Currents connect sites. Fish move across jurisdictions. Water quality, extraction and temperature can overwhelm work at one reef or meadow. An oyster reef, seagrass bed or kelp forest depends on physical habitat, propagules, grazers, predators, fishing pressure and the chemistry moving through it.

The first intervention may be pressure removal: stop destructive dredging, improve sewage treatment, protect a spawning area or reduce harvesting long enough for age structure to recover. Where local propagules are gone, active reseeding or translocation may be needed. Restoring native oysters can rebuild three-dimensional habitat and filtration. Seagrass planting can start recovery where sediment and water quality are suitable. Kelp restoration may fail if urchin grazing remains intense or marine heatwaves repeatedly remove canopy.

The ocean makes outcomes harder to contain and predict. Benefits and costs travel. A no-take area may export adults or larvae to surrounding fisheries, but it also concentrates short-term restriction on particular fishers. Coastal communities need authority in design, enforcement and benefit sharing. Social inclusion is therefore prominent in recent definitions of marine rewilding rather than an afterthought imported from land.

The operating logic still holds. Diagnose the broken relationship, reduce the pressure, restore missing structure or species where needed, reconnect the wider system and allow feedbacks to develop. The absence of wolves does not make the work less wild. An oyster reef filtering water and building habitat can return more ecological function than a spectacular mammal release.

Recovery at sea also depends on what happens outside the restoration plot. Larvae can arrive from distant populations or be swept away. Nutrient pollution comes from land. Mobile predators and fishing fleets cross boundaries. Marine projects therefore need governance at the scale of currents and use, even when the first intervention is a few hectares of seagrass or shellfish reef.

Restore disturbance without denying risk

Fire creates the hardest collision between process and safety. Many ecosystems developed with recurring fire, including fires deliberately set by Indigenous peoples. Suppression can allow fuels to accumulate, while climate change increases extreme fire weather in many regions. Walking away can therefore produce neither historical conditions nor tolerable risk.

Recovery may require prescribed burning, managed wildfire, grazing or mechanical fuel reduction near settlements. Cultural burning adds knowledge about timing, patch size, species and purpose that was often suppressed by colonial law. Human-set fire can restore a disturbance regime rather than disqualify the landscape from wildness.

Grazing presents the same problem in slower motion. Projects use cattle, ponies or other robust herbivores where original wild grazers are extinct or impractical. Fixed agricultural stocking reproduces livestock management under a new label. Total non-intervention inside small fences can create welfare crises. The difficult middle allows variable grazing and behaviour while retaining responsibility for winter food, disease, boundaries and population size.

The animal should be judged by function rather than romance. Does grazing create structural variation? Does browsing alter regeneration? Are nutrients moving? Are bare patches and dung communities increasing? A proxy need not resemble an extinct ancestor perfectly to perform useful work. A fashionable breed does not prove that the work is occurring.

Restoring cultural burning demonstrates that human action and ecological autonomy are not opposites. A carefully timed patch burn can renew foods, reduce fuels and create habitat variation while returning decision-making to communities whose practices were prohibited. The process becomes less controlled by distant institutions because knowledgeable local management has been restored.

Measure declining dependence

Rewilding lacks one universal score because projects restore different processes at different scales. Counting species, hectares or released animals can help, but none alone shows whether autonomy is increasing. Monitoring must connect actions to mechanisms.

Did blocking drains raise water tables through dry months? Did removing a barrier increase fish passage and spawning? Did predator return change browsing, carcass supply or smaller predators? Did herbivore introduction create patchiness without unacceptable welfare costs? Did animals use the corridor? Did local conflict rise or fall? Did the amount of routine management decline?

Good monitoring records failure. A released species that repeatedly leaves apparently suitable habitat may reveal that the habitat model is wrong. Regeneration stalled for ten years may indicate browsing, soil or seed limitation. A wetland captured by one invasive plant may expose a missing disturbance. Adaptive management uses those signals to change the hypothesis rather than protect the original plan.

Where possible, monitoring needs a comparison. A nearby untreated site, measurements from before intervention or variation in treatment can distinguish recovery from weather and wider trends. Perfect experiments are rare in working landscapes, but a project that records only its preferred outcome cannot tell whether its story caused the change.

Finance becomes an ecological variable because it determines which processes can continue. Tourism may reward visible animals. Carbon payments may favour trees over grassland or wetland. Agricultural subsidies may reward intervention. A short grant may pay for release but not coexistence. Mature projects plan institutional succession: who owns the animals after the founder leaves, who manages a flood dispute, and whether a corridor survives a change of government.

One final test clarifies the goal. Imagine the project loses half its management budget. Which ecological functions stop? If water levels collapse without pumps, herbivore numbers immediately become dangerous and every valued species needs annual rescue, the system remains dependent. It may still be excellent conservation. It is not far along the rewilding gradient.

If rivers still flood, vegetation still regenerates, animals still move and reproduce, and managers focus mainly on boundaries, welfare, monitoring and coexistence, more agency has returned. Rewilding is working when human absence from one meeting no longer produces ecological paralysis.

How we know

Rewilding is young as a named field, so its evidence comes from several older literatures: restoration ecology, population biology, trophic ecology, landscape connectivity, species translocation, natural regeneration and human-wildlife coexistence. Strong evidence supports particular mechanisms, including beaver engineering, natural forest recovery, predator-prey effects and the value of movement between populations. Evidence for whole projects is less uniform because goals, scales and baselines differ.

Attribution is the largest difficulty. Ecosystems change for many reasons at once. Yellowstone shows both the power and the problem: predation returned and vegetation changed, but the contribution of wolves to each local outcome remains debated. Claims are strongest when they name a mechanism, measure change through time, compare plausible alternatives and report uncertainty without turning it into paralysis. Evidence is also uneven. Dramatic releases are monitored and publicised more often than quiet pressure removal, failed projects can disappear from view, and social outcomes are measured less consistently than animal survival or vegetation. The final test, whether management dependence declines, requires long time series that many young projects do not yet possess.

What People Get Wrong

“Rewilding means leaving land alone”

Sometimes leaving land alone is the right move. It is not the definition.

A degraded system may be unable to recover because the process it needs has been physically removed. A drained peatland will not rewet while the drains remain open. Salmon cannot reach spawning habitat above an impassable barrier. A locally extinct beaver cannot build a dam by spontaneous generation. An invasive plant can hold succession in place for years.

Rewilding therefore ranges from passive recovery to intense initial intervention. The distinguishing direction is towards stronger ecological processes and less continuing prescription. A project may block drains, remove a weir, control an invasive species or move an animal before restraint becomes useful.

The myth became persuasive because wildness sounds like withdrawal. It offers a moral story in which people stop interfering and nature repairs everything. Past interference has altered soils, water, species pools and landscape geometry. Walking away can preserve the damage.

The correction matters because neglect can be sold as ambition. Good rewilding has a diagnosis, monitoring and thresholds. Non-intervention is a decision supported by evidence, not an excuse to avoid one. Restraint earns the name only when a system has a plausible route to recovery.

“Rewilding is just bringing back wolves”

Wolves are excellent publicity. They are not a programme.

Large carnivores matter where their ecological role is missing and where viable populations can coexist with people. Rewilding can also mean removing a river barrier, restoring floodplain movement, allowing natural regeneration, reconnecting habitat, returning beavers or recovering oyster reefs. None requires an apex predator.

The predator-centred image comes partly from early North American formulations built around large cores, corridors and carnivores. Yellowstone then supplied an irresistible case. Later practice broadened the idea towards ecological function, dynamic change and reduced management dependence.

Even where wolves return, they cannot reconnect a river, restore soil structure or move a road. They influence a network whose other broken parts still matter.

The correction changes planning. Predator symbolism can turn a functional question into a culture war. A project should ask which process is missing in this place. Sometimes the answer is wolf. Often it is water, movement, grazing, decomposition or restraint. The animal follows the diagnosis rather than supplying it.

“Nature has a balance that rewilding restores”

The balance of nature survives because it is comforting. Healthy ecosystems fluctuate.

Populations rise and fall. Fire resets vegetation. Rivers erode and deposit. Storms open gaps. Disease changes abundance. Predators and prey vary. Succession alters communities. Climate moves ranges. A system can persist without returning to one fixed number or appearance.

Rewilding aims to recover processes that let systems respond to change, not freeze them at harmony. This is why an ecologically active landscape may look messy, contain death and change rapidly.

The myth becomes dangerous when every fluctuation is treated as either proof of success or a reason intervention is forbidden. Oostvaardersplassen showed the ethical cost of treating starvation inside a fenced, human-created system as automatically acceptable because mortality occurs in nature. People remain responsible for boundaries they built.

The correction matters because autonomy is not indifference. Adaptive management can permit surprise while retaining welfare, legal and safety limits. A dynamic ecosystem needs room to change and honest rules for the changes society will not accept. Fluctuation is expected; preventable suffering inside artificial constraints remains a human decision.

“Yellowstone proved that wolves changed the rivers”

Yellowstone proved something more useful: restoring a predator can reorganise ecological interactions, and causal stories about those interactions require care.

Wolves returned in 1995 and 1996. Predation resumed. Elk abundance and distribution changed. Coyotes, scavengers and carcass supply were affected. Aspen and willow recovered in some places, and beaver colonies later increased.

The claim about rivers depends on several links: fewer or warier elk, less browsing beside streams, more willow and aspen, more beavers, stronger banks and altered channels. Evidence is not equally strong for every link or across the whole park. Bears, cougars, human hunting, severe weather, groundwater and geomorphology complicate attribution. Later work shows substantial young-aspen recovery in parts of Yellowstone, while researchers still contest the size and cause of the wider cascade.

The myth spread because one dramatic cause is easier to remember than a network. The correction does not diminish the restoration. It removes a claim the evidence cannot yet carry.

The lesson is to trace each proposed link and compare rival causes. When one species is said to repair an ecosystem through a guaranteed chain reaction, the story has outrun the ecology. A stronger account keeps the predator's importance without granting it control over every result.

“Rewilding means recreating the world before humans”

There is no single pre-human world to recreate, and in many landscapes humans have been ecological actors for millennia.

Climate changed repeatedly before industry. Species migrated and disappeared. Indigenous peoples burned, hunted, planted, harvested and engineered landscapes. Agricultural systems created habitats now valued for biodiversity. A date chosen as natural is often a cultural choice wearing scientific clothing.

Rewilding uses history to identify lost species and processes. It does not require a museum reconstruction. Climate change may produce future communities with no exact historical equivalent. Extinction has removed actors that cannot be returned.

The misconception is especially harmful where conservation displaced people in the name of wilderness. Treating inhabited land as empty can erase rights and knowledge. A new label does not make that exclusion benign.

The correction matters because the past should guide diagnosis rather than dictate destiny. The useful question is what ecological integrity, movement and autonomy can be restored now, under conditions already changing. Historical evidence widens ambition; it does not remove present responsibility.

“More trees always means more rewilding”

Forests are powerful ecosystems. Tree cover is not a universal score for naturalness.

Some places are naturally open. Others have long been maintained as species-rich grassland, heath or wood pasture through soils, climate, herbivores and human practice. Dense planting can reduce biodiversity, change water and erase the habitat a project meant to recover.

Even where woodland is desirable, spontaneous regeneration may produce greater variation than a uniform plantation. A planted stand can store carbon while remaining structurally poor, ecologically isolated and dependent on repeated care. Planting may be necessary where seed sources or soils are badly damaged, but a hectare count does not show whether a functioning forest is developing.

The myth grew because trees are visible, countable and easy to finance. Processes are harder to photograph. Climate policy also turned planting into a familiar unit of action, sometimes without enough attention to biome or permanence.

The correction makes rewilding place-specific. A wetland should become wetter. A grassland may need grazers. A river needs room. A forest may need dead wood, regeneration and disturbance. Nature is not a synonym for trees.

“Rewilding pushes people out”

It can, and that is a failure mode rather than a requirement.

Some conservation traditions created protected areas by excluding local communities. Rewilding can repeat the pattern when land is acquired, access changes and livelihoods are treated as obstacles to purer nature. It can also support tourism, fisheries recovery, lower-input farming, flood storage, restoration employment and cultural renewal.

The difference is governance. Who sets the goals? Who owns land and animals? Who receives benefits? Who bears predator losses, flooding or limits on harvest? Are Indigenous and local rights recognised? Is compensation prompt? Can people change management when conditions change?

Modern guidance treats coexistence and local knowledge as part of the ecological project. That is practical. A predator population facing continued local killing is not secure. A wetland drained after backlash is not restored. Social legitimacy affects persistence.

The correction rejects the false choice between people and nature. Humans are already inside the system. The task is to return ecological agency while distributing authority, costs and benefits fairly enough for the arrangement and local trust to last. Where people hold legitimate rights, durability begins with consent and shared power rather than permission granted from outside.

Use It

Ask what work the system is doing for itself

When you look at a landscape, stop counting only what is present. Ask which processes operate without continual human replacement.

Who moves seeds? What controls herbivores? Where does dead material go? Can water spread onto a floodplain? Can animals move between habitat patches? What creates bare ground? What happens after a tree falls? Which jobs are performed by tractors, pumps, chainsaws, culls or concrete because the ecological worker has disappeared?

This lens changes how tidiness looks. A carefully maintained reserve may contain high biodiversity and low autonomy. A rough scrubland may contain ordinary species and rapidly recovering processes. Neither is automatically better. The distinction tells you what kind of conservation is occurring and what it costs to sustain.

It also works at small scales. A city park can leave dead wood, connect ponds and reduce mowing. A farm can retain hedges and wet corners that make the wider landscape permeable. A river project can remove one barrier and restore more function than years of stocking fish above it. The useful question is not whether management exists. It is which management substitutes for lost ecological work and whether that substitution can shrink.

Distinguish the bottleneck from the symptom

Ecological decline is often treated at the visible end. A bird is rare, so breed more birds. A river has few fish, so release fish. A hill has few trees, so plant trees.

Rewilding directs attention upstream in the causal chain. Why is the bird failing to reproduce? Which route prevents the fish reaching spawning ground? Is browsing, fire, seed isolation, soil or water stopping tree establishment? The same symptom can arise from several mechanisms, and the visible intervention may leave all of them intact.

Use the lens by asking what would happen after the project stopped. If released fish disappear each generation, stocking is replacing reproduction rather than restoring it. If planted trees die without watering, the site has not regained woodland dynamics. If a rare meadow survives only through annual cutting, that may be worthwhile conservation, but its dependence should be named.

The bottleneck can be social as well as ecological. A compensation scheme that pays too slowly can become the main threat to a carnivore population. A corridor may fail because one planning decision closes the narrowest link. A wetland may be technically restored and politically doomed because neighbours had no role in the choice.

Match freedom to scale and reversibility

The word wild can encourage absolutes: remove all fences, stop all culling, release the animal, accept whatever follows. Better decisions match freedom to the size of the system and the reversibility of the risk.

A large open population can regulate itself through movement, predation and variable resources. A small enclosure cannot. Fire in a remote fire-adapted landscape differs from fire beside homes. Beaver activity in a broad floodplain differs from a dam beside a railway culvert. The same process can be desirable at one scale and dangerous at another.

Ask three questions. How far can the process travel? What boundary will it meet? Can harm be reversed if the model is wrong? A removable fence, temporary grazing change or trial release may permit learning. Introducing a novel proxy species, allowing a population to exceed the carrying capacity of a closed reserve, or changing a river beside critical infrastructure has different stakes.

This is not an argument for timidity. It is an argument for proportionate freedom. Wildness grows when organisms and physical processes receive enough room to surprise us, while decisions with large irreversible consequences receive more evidence and stronger consent.

Price the conflict before celebrating the benefit

Rewilding produces public benefits and local costs with unusual frequency.

A restored floodplain may protect a town while inundating one field. A carnivore may enrich a national park and kill one household's livestock. A protected nursery area may improve a fishery later and restrict catches now. Scrub recovery may support wildlife and change fire risk or a familiar view.

Before describing a project as a win for everyone, identify who can lose. Then ask whether prevention, compensation, insurance, revenue sharing, land purchase or different design can make the arrangement durable. A benefit spread across millions of people does not pay one farmer's immediate bill.

The lens also catches unequal authority. Consultation after a release plan is fixed is weaker than shared decisions over goals, monitoring and response. Indigenous-led restoration begins from a different position again: the people whose relationships and sovereignty were damaged by species removal direct the return.

Conflict is not evidence that recovery is misguided. It is evidence that ecology has re-entered a social landscape. Projects become durable when the beneficiaries help carry the cost and those exposed to risk can alter the rules.

Measure declining dependence, not resemblance

A project can look wild while remaining management-heavy. It can also look unfinished while ecological functions become stronger.

Choose measures tied to the process. Water tables and hydroperiods for wetland recovery. Fish passage and spawning for river connectivity. Dispersal and genetic exchange for corridors. Browsing, carcass supply and prey behaviour for predator return. Vegetation structure, movement and welfare for large herbivores. Local loss, trust and payment speed for coexistence.

Then add one institutional measure: how much routine intervention is required now compared with five years ago? A successful reintroduction may initially demand intense monitoring. A restored river may need earthworks. The aim is not low effort on day one. It is a trajectory in which more functions continue without repeated replacement.

This lens prevents attractive proxies taking over. Number of trees planted, animals released and hectares labelled are inputs. They do not prove that a system can regenerate, move, regulate and adapt. The more useful result is a widening range of choices available to the ecosystem.

The limits

Rewilding is one conservation approach, not the entire conservation toolkit. Some species survive because people actively maintain the habitats they need. Small reserves may remain dependent on mowing, grazing, water control or predator exclusion. Pollution, invasive species, altered soils, climate change and isolation can prevent recovery after pressure is removed.

Reintroductions carry disease, genetic, welfare and conflict risks. Large animals inside fences remain partly a human responsibility. Natural death does not settle the ethics when people designed the enclosure. Predators may require continuing prevention and compensation. Fire restoration cannot ignore homes. Rivers cannot be given unlimited space through cities.

Nor is rewilding automatic climate policy. Restoring forests, peatlands, wetlands and marine vegetation can store carbon or reduce risk, but outcomes differ by ecosystem and timescale. Tree planting in open or carbon-rich habitats can cause harm. Carbon accounting can narrow biodiversity recovery into one metric.

Land is the largest constraint. Food, housing, energy, transport and nature compete for space, although the categories can overlap. The practical future is a mosaic: some areas with high ecological autonomy, some farmed or fished with greater permeability, some intensively used, and all connected where possible. Claims must match the room available.

The one thing to keep

Keep one question: what can this system do for itself?

It cuts through imagery. A green landscape may depend on constant labour. A rough landscape may be recovering functions. A successful species return may restore a relationship rather than a number. A river may need room more than planting. A forest may need herbivores, dead wood and time more than another million seedlings.

Nature never disappeared. What disappeared were choices available to living systems: where water could go, where animals could move, what could eat what, what could regenerate and which disturbances could occur without immediate suppression.

Putting nature back means returning some of those choices. People still set laws, boundaries and tolerable risks. The achievement is to make fewer ecological decisions on nature's behalf, while making the human decisions around them better.

Terms

Rewilding. Restoration aimed at recovering ecological processes, interactions, connectivity and increasing self-direction, with less continuing human control over ecological outcomes. It describes a direction of change rather than one fixed method, species list or visual end point. A project can move along that gradient through time.

Ecological process. A recurring biological or physical operation, such as predation, grazing, decomposition, flooding, fire, succession or migration, that shapes an ecosystem through time.

Ecological integrity. The degree to which an ecosystem retains characteristic composition, structure, functions and capacity for self-organisation under its environmental conditions. It does not require a static state.

Autonomy. The capacity of an ecological system to develop and respond with reduced need for continuing human prescription. It is a gradient, not the total absence of people, law, monitoring or responsibility.

Passive rewilding. Recovery driven mainly by removing human pressures and allowing natural succession, recolonisation and disturbance to resume. It still requires diagnosis, responsibility, monitoring and a credible reason to expect recovery.

Assisted natural regeneration. Targeted removal of barriers to recovery, such as heavy grazing, repeated fire, weeds or seed isolation, while relying mainly on succession rather than mass planting.

Succession. Change in biological communities after disturbance or release from management. It can move towards woodland, grassland, wetland or another state depending on conditions and feedbacks.

Trophic level. A broad position in a food web defined by how an organism obtains energy, such as producer, herbivore, predator or decomposer. Real diets often cross levels, so the categories are useful simplifications rather than rigid boxes.

Trophic cascade. Effects that pass through a food web when change at one trophic level alters organisms at others, sometimes reaching vegetation, habitat structure or physical processes.

Keystone species. A species whose ecological effects are large relative to its abundance. The term concerns influence, not charisma, rarity or a guarantee of ecosystem-wide control in every place.

Ecosystem engineer. An organism that changes physical conditions and thereby alters habitat for other species. Beavers, reef-building oysters and burrowing animals are common examples.

Megafauna. Large animals. Thresholds vary, but rewilding discussions usually mean large herbivores and carnivores whose movement, feeding, engineering and death can have landscape-scale effects.

Herbivory. Consumption of plants by animals. At sufficient intensity it changes vegetation structure, regeneration, nutrient movement and habitat for other organisms.

Browsing. Feeding on leaves, shoots and woody vegetation, contrasted loosely with grazing on grasses and low plants. Browsing often determines whether young trees escape into the canopy.

Predation. Killing and consuming other animals. Predators can affect prey abundance, behaviour, distribution, carcass supply and competition among predators.

Mesopredator release. An increase in medium-sized predators after larger predators decline or disappear, potentially changing populations of smaller prey and other competitors.

Reintroduction. Deliberate release of a species into part of its historical range from which it has disappeared. It is one form of conservation translocation.

Conservation translocation. Deliberate movement and release of organisms for conservation benefit. It includes reintroductions, reinforcements and introductions outside former range under tightly defined conditions, each with different risks.

Ecological proxy. A living substitute used to perform some functions of an extinct or unavailable species, such as robust cattle providing aspects of large wild herbivore grazing.

Connectivity. The degree to which organisms, genes, water, sediment or disturbance can move through a landscape or seascape. It depends on the organism or process being measured and cannot be inferred from a map alone.

Wildlife corridor. Habitat intended to support movement between larger areas. A mapped strip becomes a functional corridor only when the relevant organisms can and do use it.

Core area. A relatively large habitat area able to support breeding, feeding or disturbance with less edge pressure. Its necessary size depends on species and process.

Permeability. How easily organisms move through land or water outside protected cores and corridors. Farms, towns, roads and fisheries can be more or less permeable.

Reference ecosystem. A real or reconstructed example used to guide restoration goals and expected conditions. It informs judgement without supplying a single mandatory historical picture or overriding future climate constraints.

Novel ecosystem. A system with new combinations of species or conditions that cannot readily be returned to a previous state. The term describes change rather than endorsing it, and such systems can still contain high value.

Shifting baseline syndrome. The tendency for each generation to accept the depleted conditions of its youth as normal, causing long-term loss to disappear from memory and targets.

Natural regeneration. Renewal of vegetation through existing seed banks, nearby seed sources, sprouts and dispersal rather than planted stock. Success depends on soils, browsing and connectivity.

Adaptive management. Structured learning in which managers monitor outcomes, compare them with expectations and alter action when evidence shows that the working model is wrong or incomplete.

Coexistence. The continuing social and practical arrangements that allow people and wildlife to share landscapes despite costs and risk. It does not mean conflict disappears; it means conflict is anticipated, reduced, shared and governed without removing either people or wildlife.

Oostvaardersplassen. A Dutch reserve whose low-intervention large-herbivore experiment became a major welfare controversy. It demonstrates that autonomy claims must account for fences, space, welfare, movement and continuing collective human responsibility for designed constraints.

Go Deeper

Paul Jepson and Cain Blythe, Rewilding: The Radical New Science of Ecological Recovery (2022). Start here for the accessible overview. Jepson and Blythe explain how the movement developed from large-carnivore conservation into a broader approach based on ecological function, dynamic systems and institutional change. The book is strongest on the variety of modern projects and on rewilding as a practical shift in conservation thinking. Its enthusiasm is part of its appeal, so keep the cautions about evidence, welfare and local power from this book beside it. Read it to see how one label now contains projects ranging from passive forest recovery to predator return and river restoration.

Isabella Tree, Wilding: The Return of Nature to a British Farm (2018). Read this for one landscape experienced through time. Tree tells the story of Knepp's move away from intensive farming towards process-led recovery, with scrub, water, free-roaming herbivores and unexpected species at the centre. It is inviting, concrete and unusually good at showing how untidiness becomes ecological structure. It is also one privately owned English estate, not a universal model for climates, land systems or communities elsewhere. Its value lies in the observed sequence and the honest friction between ecological surprise, farm economics, neighbours and animal management.

Andrea Perino and colleagues, “Rewilding Complex Ecosystems”, Science 364 (2019). Read this for the scientific model in compact form. The paper frames rewilding around trophic complexity, stochastic disturbance and dispersal, and treats ecological function as the organising target rather than a fixed historical composition. It is short and dense. A reader new to ecology may need to pause over the diagrams and terminology, but the article shows how a broad movement can be translated into mechanisms that can be studied and compared. Use it to test whether a project restores dispersal, disturbance and trophic complexity rather than relying on atmosphere or branding.

IUCN Commission on Ecosystem Management, Guidelines for Rewilding (2025). Read this when the question shifts from inspiration to responsibility. The guidelines place ecological integrity, dynamic processes, scale, local knowledge, participation, adaptive management and coexistence inside one framework. They are written for practitioners and decision-makers rather than as narrative science. That makes them the useful counterweight to stories built around dramatic releases: they show how much work must be done before, around and after an animal or process is returned. The document is also the clearest current statement that social legitimacy and ecological autonomy belong in the same design.

Notes and Sources

The Whole Thing in One Page and Why You Should Care

Working definition. The book follows the process-centred account developed by Perino and colleagues in 2019, the ten principles assembled by Carver and colleagues in 2021, and the IUCN Commission on Ecosystem Management's 2025 guidelines. These sources do not make rewilding one fixed technique. They describe a direction towards stronger ecological integrity, trophic complexity, disturbance, dispersal, connectivity and greater capacity for self-organisation, while retaining participation, monitoring and adaptive management. The distinction between restoring an ecological process and preserving a chosen visual state is the book's synthesis of that literature.

Early lineages. The North American shorthand of cores, corridors and carnivores comes from Michael Soulé and Reed Noss's 1998 essay. It captured the need for large protected areas, movement between them and the return of ecologically influential large animals. Later work broadened the term because projects in densely inhabited landscapes, fresh water and the sea often depend on processes other than large-carnivore return. The book treats the early model as one important lineage rather than the permanent definition.

Global restoration target. Target 2 of the Kunming-Montreal Global Biodiversity Framework calls for at least 30 per cent of areas of degraded terrestrial, inland-water, coastal and marine ecosystems to be under effective restoration by 2030. Rewilding is one possible approach within that wider commitment. The target is not a requirement to rewild 30 per cent of the planet, and the book avoids treating protected-area coverage and ecosystem restoration as the same measure.

European restoration law. Regulation (EU) 2024/1991 entered into force in August 2024. Its overarching objective is for restoration measures to cover at least 20 per cent of EU land and 20 per cent of EU sea by 2030, followed by measures for ecosystems in need of restoration by 2050. The law contains ecosystem-specific obligations and national planning requirements. The narrative uses only the overarching figure because the detailed legal machinery belongs to a broader conservation or law treatment.

Beavers in England. Natural England and Defra published the wild-release policy and licensing route on 28 February 2025. A project must demonstrate benefits, assess risks and provide monitoring, management and exit arrangements. The book describes this as opening a legal route for licensed releases rather than claiming that any applicant may release beavers. Puttock and colleagues' multi-site study supports the claim that beaver dams can attenuate flows, while also showing spatial and seasonal variation. The wording therefore avoids presenting beavers as universal flood protection.

Sources for the Core Ideas

Process, composition and autonomy. Restoration ecology commonly distinguishes composition, structure and function. Rewilding literature gives unusual weight to processes and reduced continuing control because restored functions can generate later changes that managers did not prescribe. This does not make autonomy the only conservation value. A highly managed grassland can remain more important for particular threatened species than a more self-directed scrubland. The book keeps biodiversity value and ecological autonomy separate so that neither is smuggled in as a synonym for the other.

Natural and assisted regeneration. FAO's 2019 practical manual sets out the conditions under which assisted natural regeneration can work, including retained soils, nearby seed sources, control of repeated pressure and the capacity of vegetation to resprout or recolonise. Cook-Patton and colleagues mapped the carbon accumulation potential of natural forest regrowth, while Di Sacco and colleagues stressed protecting existing ecosystems, choosing suitable places and using natural regeneration where possible. These sources support restraint as a serious method without implying that every degraded site will recover unaided or that all restoration should create forest.

Knepp. Isabella Tree's account is the principal source for the sequence at Knepp: the financial difficulty of intensive farming on heavy clay, the shift beginning around 2000, the roughly 3,500-acre scale, the restoration of water and the use of free-roaming cattle, ponies, pigs and deer. The book uses Knepp as a project narrative and not as controlled proof that the same recipe will work elsewhere. The estate remains fenced, privately owned and actively governed, which is why it illustrates a reduction in prescription rather than human disappearance.

Beavers as ecosystem engineers. The physical effects described in the text are standard features of beaver ecology: dam building, canal digging, wood movement, sediment retention and the creation of wetland habitat. Puttock and colleagues provide direct hydrological evidence from English sites. Effects vary with catchment form, dam sequence, season and nearby infrastructure. This variability is material because the same engineering that stores water can create local flooding or obstruct a culvert.

Large herbivores and Vera. Frans Vera's 2000 book challenged the view that temperate lowland Europe would have been uniformly closed forest without people, arguing for a larger role for grazing animals in maintaining open mosaics. The historical extent of openness and the weight assigned to herbivores remain disputed. The manuscript retains only the narrower claim needed for rewilding: large herbivores can strongly alter regeneration, vegetation structure, nutrient movement and disturbance. Rewilding projects do not need Vera's strongest palaeoecological reconstruction to justify restoring grazing processes.

Animals as ecological workers. The examples of salmon moving marine nutrients, oysters building reefs, seabirds transferring nutrients and dung beetles burying waste are used to show that function is distributed across many organisms. The term keystone species is retained with a warning because strong effects are context-dependent. The book rejects the idea that one charismatic species can replace missing habitat, water, movement and social permission.

Predators and trophic cascades. Kauffman, Brodie and Jules tested whether wolf-related changes in elk behaviour explained aspen recovery across Yellowstone and found that the simple behavioural-cascade account was insufficient at landscape scale. Mech warned against turning the wolf into a sanctified single cause. Barber-Meyer challenged a proposed wolf-to-grizzly-bear cascade by examining changing foods and bear abundance. These papers support the manuscript's caution without denying predator effects.

Yellowstone after 2025. New work has strengthened parts of the case while keeping the argument open. Painter, Beschta and Ripple reported major changes in the height structure of young aspen after large-carnivore restoration. Ripple and colleagues described a strong trophic cascade. MacNulty and colleagues then challenged the analytical basis of that estimate. The book therefore states that aspen recovered in some places and that predation returned, while leaving the size, distribution and attribution of the cascade as an active scientific dispute. It does not repeat the popular claim that wolves alone changed the course of Yellowstone's rivers.

Sea otters and kelp. Estes and Duggins's work across Alaskan sites is a classic demonstration that sea otter predation can reduce sea urchins and support kelp, with important regional variation in recovery. It supplies a marine example of a trophic cascade whose mechanism is strong but still conditioned by recruitment, local history and physical disturbance.

Connectivity. Hilty and colleagues' IUCN guidelines distinguish ecological networks from corridors and stress that connectivity concerns movements of organisms and ecological processes at multiple scales. A continuous strip is not automatically functional habitat for every species. The book therefore defines connectivity through observed movement, gene flow, hydrology or dispersal rather than the appearance of a green line on a map.

Oostvaardersplassen. Staatsbosbeheer's monitoring report for 1 May 2017 to 30 April 2018 records mortality of 3,226 large herbivores: 2,684 red deer, 467 Konik horses and 75 Heck cattle. About 89 per cent were shot under the reactive management protocol when condition had deteriorated. Bert Theunissen analysed the history, expert assumptions and public conflict surrounding the reserve. In July 2018, Flevoland adopted a policy based on the Van Geel committee's advice, including a reduction towards 1,100 large grazers and a changed welfare and monitoring regime. The case is used to show the interaction of fences, scale, movement, welfare and public legitimacy. It is not offered as evidence that all low-intervention projects fail.

Historical baselines. Historical reference can reveal missing species, reduced abundance and altered disturbance, but it cannot provide one universal date to reproduce. Climate, extinction and long human histories make the choice of baseline partly normative. The concept of shifting baselines remains useful because depleted abundance can become normal within one generation. The book uses history to widen ambition and diagnose loss, then judges future goals against present and expected conditions.

Indigenous stewardship and fire. Lake and colleagues document the ecological knowledge, cross-jurisdictional barriers and institutional conditions involved in returning Indigenous fire stewardship in the United States. Their account supports the distinction between indiscriminate fire and culturally governed burning with specific timing, patch size, species and purposes. The narrative does not present cultural burning as a generic tool transferable without the authority and knowledge of the communities concerned.

Buffalo and the Iinnii Initiative. The Buffalo Treaty was first signed on 24 September 2014 by eight Indigenous Nations and has since expanded. Its articles treat buffalo return as ecological, cultural, spiritual, economic, health and educational renewal. Wildlife Conservation Society material describes the Iinnii Initiative as Blackfoot-led buffalo restoration across Blackfoot Confederacy territory. The book uses it to correct the assumption that human authority is external to rewilding. It does not reduce the treaty to a species-management project.

Iberá jaguars. Rewilding Argentina's 2024 annual report recorded 33 free-living jaguars in Iberá and 12 cubs born during the year. The project followed habitat and prey recovery, captive breeding, preparation for release, monitoring and local work around coexistence. The figures are a dated project report rather than an independent population census and should be updated before any later edition. They support the narrower point that a release programme changes into a long-term governance problem once animals breed and disperse.

Sources for How It Works

The intervention sequence. Diagnose pressure, remove barriers, permit recovery, restore connectivity and translocate only where necessary is a synthesis rather than a universal official protocol. It draws on the IUCN rewilding and translocation guidelines, FAO regeneration guidance and process-based scientific frameworks. Some projects will change the order. A species release can be needed before a process resumes; urgent invasive-species control can continue for years. The purpose of the sequence is to prevent a dramatic release from replacing diagnosis.

Water. The treatment of rivers and wetlands rests on the distinction between restoring water movement and maintaining water at a chosen level. Drain blocking, floodplain reconnection, barrier removal and beaver activity can all restore processes, but each can shift risk across boundaries. The book deliberately avoids universal claims about flood reduction or carbon because outcomes depend on soil, topography, infrastructure, climate and timescale.

Translocation safeguards. The IUCN Species Survival Commission's 2013 guidelines remain the central source for justification, feasibility, source populations, genetics, disease, welfare, release methods, monitoring and response to failure. The 2025 rewilding guidelines do not replace those safeguards. The phrase exit strategy is used broadly for explicit decisions about when releases stop, when removal may be necessary and who retains responsibility.

Marine rewilding. Brooker and colleagues' 2025 review defines marine rewilding around ecological and trophic complexity, habitat and species interventions, pressure reduction and social inclusion. Estes and Duggins supply the sea otter and kelp mechanism. The book gives oyster reefs, seagrass and kelp as examples of structures or relationships that can be restored, while stressing that currents, water quality, fishing and heat can overwhelm a small site. It does not claim that marine rewilding has one accepted universal method or mature evidence base.

Fire. Lake and colleagues support the claims about cultural burning and the damage caused when colonial and agency systems excluded Indigenous knowledge. The book also keeps safety limits explicit. Restoring a fire regime can involve prescribed burning, managed wildfire, grazing and fuel treatment near settlements. The correct mixture is place-specific and cannot be inferred from the fact that fire occurred historically.

Monitoring and declining dependence. No universal rewilding metric is accepted. Process indicators, biodiversity measures, connectivity, intervention intensity, welfare and social outcomes all appear in current frameworks. The proposed question about what stops if the budget is halved is an editorial lens, not a scientific index. It is intended to expose continuing dependence, not to suggest that low spending proves ecological success.

Attribution. Whole-project evidence is difficult because landscapes lack clean controls, several drivers change together and projects choose different goals. Before-and-after data, untreated comparisons, spatial variation and explicit rival explanations strengthen inference. Yellowstone is retained as the central example because it contains strong evidence of change and unusually visible disagreement over attribution.

What People Get Wrong and Use It

Leaving land alone. Passive recovery is one method, not the definition. The FAO manual and IUCN guidance both require attention to barriers and risk. Drained peat, blocked migration routes, lost seed sources, invasive species and locally extinct ecological workers can prevent spontaneous recovery. The text therefore distinguishes deliberate restraint from unmanaged abandonment.

Balance of nature. Modern ecology does not assume one stable equilibrium for every system. Disturbance, succession, demographic fluctuation and alternative states are ordinary features. The book uses this correction narrowly: dynamic change does not remove human responsibility for artificial boundaries, welfare or public safety.

Trees. Di Sacco and colleagues warn that tree planting can damage native ecosystems when the place, species and social design are wrong. FAO guidance favours natural regeneration where conditions permit. The book therefore treats tree cover as one possible result rather than a universal measure of rewilding.

People and rights. Perino and colleagues, Carver and colleagues, the IUCN guidelines and the Kunming-Montreal framework all place social constraints, local knowledge, participation or Indigenous and local rights inside restoration practice. These sources do not guarantee fair implementation. The book's stronger claim is causal: projects that concentrate costs, erase legitimate authority or lack durable conflict arrangements are less likely to persist ecologically.

Climate. Restoration can contribute to mitigation and adaptation, but the effect varies among forests, peatlands, wetlands, grasslands and marine systems. Cook-Patton and colleagues address natural forest regrowth rather than all rewilding. The book therefore rejects a universal carbon number and keeps biodiversity, ecological function and climate accounting distinct.

Current-source verification. Current legal, policy and project-status claims were checked on 11 August 2026. The England beaver policy, EU implementation timetable, Iberá jaguar figures and IUCN 2025 guidance should be rechecked before a later edition because they can change.

Bibliography

Scientific and scholarly works

Barber-Meyer, Shannon M. "Trophic Cascades from Wolves to Grizzly Bears or Changing Abundance of Bears and Alternate Foods?" Journal of Animal Ecology 84, no. 3 (2015): 647-651. doi:10.1111/1365-2656.12338.

Brooker, Esther E., Gerald Midgley, Neil Burns, Charlotte E. Trotman, Amanda Gregory and Charlotte Rachael Hopkins. "Defining Marine Rewilding Can Help Guide Theory and Practice in Marine Conservation." Communications Earth & Environment 6 (2025): 241. doi:10.1038/s43247-025-02155-x.

Carver, Steve, Ian Convery, Sally Hawkins, et al. "Guiding Principles for Rewilding." Conservation Biology 35, no. 6 (2021): 1882-1893. doi:10.1111/cobi.13730.

Cook-Patton, Susan C., Sara M. Leavitt, David Gibbs, et al. "Mapping Carbon Accumulation Potential from Global Natural Forest Regrowth." Nature 585 (2020): 545-550. doi:10.1038/s41586-020-2686-x.

Di Sacco, Alice, Kate A. Hardwick, David Blakesley, et al. "Ten Golden Rules for Reforestation to Optimize Carbon Sequestration, Biodiversity Recovery and Livelihood Benefits." Global Change Biology 27, no. 7 (2021): 1328-1348. doi:10.1111/gcb.15498.

Estes, James A., and David O. Duggins. "Sea Otters and Kelp Forests in Alaska: Generality and Variation in a Community Ecological Paradigm." Ecological Monographs 65, no. 1 (1995): 75-100. doi:10.2307/2937159.

Kauffman, Matthew J., Jedediah F. Brodie and Erik S. Jules. "Are Wolves Saving Yellowstone's Aspen? A Landscape-Level Test of a Behaviorally Mediated Trophic Cascade." Ecology 91, no. 9 (2010): 2742-2755. doi:10.1890/09-1949.1.

Lake, Frank K., Vita Wright, Penelope Morgan, Mary McFadzen, Dave McWethy and Camille Stevens-Rumann. "Returning Fire to the Land: Celebrating Traditional Knowledge and Fire." Journal of Forestry 115, no. 5 (2017): 343-353. doi:10.5849/jof.2016-043R2.

MacNulty, Daniel R., David Cooper, Michael Procko and T. J. Clark-Wolf. "Flawed Analysis Invalidates Claim of a Strong Yellowstone Trophic Cascade after Wolf Reintroduction: A Comment on Ripple et al. (2025)." Global Ecology and Conservation 63 (2025): e03899. doi:10.1016/j.gecco.2025.e03899.

Mech, L. David. "Is Science in Danger of Sanctifying the Wolf?" Biological Conservation 150, no. 1 (2012): 143-149. doi:10.1016/j.biocon.2012.03.003.

Painter, Luke E., Robert L. Beschta and William J. Ripple. "Changing Aspen Stand Structure Following Large Carnivore Restoration in Yellowstone." Forest Ecology and Management 594 (2025): 122941. doi:10.1016/j.foreco.2025.122941.

Perino, Andrea, Henrique M. Pereira, Laetitia M. Navarro, et al. "Rewilding Complex Ecosystems." Science 364, no. 6438 (2019): eaav5570. doi:10.1126/science.aav5570.

Puttock, Alan, Hugh A. Graham, Josie Ashe, David J. Luscombe and Richard E. Brazier. "Beaver Dams Attenuate Flow: A Multi-Site Study." Hydrological Processes 35, no. 2 (2021): e14017. doi:10.1002/hyp.14017.

Ripple, William J., Robert L. Beschta, Christopher Wolf, Luke E. Painter and Aaron J. Wirsing. "The Strength of the Yellowstone Trophic Cascade after Wolf Reintroduction." Global Ecology and Conservation 58 (2025): e03428. doi:10.1016/j.gecco.2025.e03428.

Soulé, Michael E., and Reed Noss. "Rewilding and Biodiversity: Complementary Goals for Continental Conservation." Wild Earth 8, no. 3 (1998): 18-28.

Theunissen, Bert. "The Oostvaardersplassen Fiasco." Isis 110, no. 2 (2019): 341-345. doi:10.1086/703338.

Guidance, law and institutional evidence

Buffalo Treaty. Articles of the Buffalo Treaty. First signed at the Blackfeet Reservation, Montana, 24 September 2014. Current treaty and history consulted 2026.

Carver, Steve, Ian Convery, Sally Hawkins, Sarah Hertel, Jessica Fallon, Krista Lyons, Rene Beyers, Alexandra Locquet, Tristan Derham and Zoltan Kun, eds. Guidelines for Rewilding. Gland: International Union for Conservation of Nature, 2025. doi:10.2305/MTYK9384.

Convention on Biological Diversity. Kunming-Montreal Global Biodiversity Framework. Montreal: Secretariat of the Convention on Biological Diversity, 2022.

Cornelissen, P., N. Beemster and H. E. Kuypers. Vegetatie, vogels, grote herbivoren en recreatie in de Oostvaardersplassen: Verslag monitoring periode 1 mei 2017 t/m 30 april 2018. Zeist: Staatsbosbeheer, 2018.

European Parliament and Council. Regulation (EU) 2024/1991 of 24 June 2024 on nature restoration and amending Regulation (EU) 2022/869. Official Journal of the European Union, 29 July 2024.

Food and Agriculture Organization of the United Nations. Restoring Forest Landscapes through Assisted Natural Regeneration: A Practical Manual. Bangkok: FAO, 2019.

Hilty, Jodi, Graeme L. Worboys, Annika Keeley, et al. Guidelines for Conserving Connectivity through Ecological Networks and Corridors. Best Practice Protected Area Guidelines Series No. 30. Gland: International Union for Conservation of Nature, 2020.

IUCN Species Survival Commission. Guidelines for Reintroductions and Other Conservation Translocations. Version 1.0. Gland: International Union for Conservation of Nature, 2013.

Natural England and Department for Environment, Food and Rural Affairs. Wild Release and Management of Beavers in England. Published 28 February 2025; guidance consulted 2026.

Province of Flevoland. Beleidskader Beheer Oostvaardersplassen. Flevoland, 2018; implementation material consulted 2026.

Rewilding Argentina. Annual Report 2024. Buenos Aires: Fundación Rewilding Argentina, 2025.

Wildlife Conservation Society. "WCS Transitions Programs in the Rocky Mountains to Indigenous Community-Driven Local Organizations." 21 February 2022.

Books and project accounts

Jepson, Paul, and Cain Blythe. Rewilding: The Radical New Science of Ecological Recovery. Cambridge, MA: MIT Press, 2022.

Tree, Isabella. Wilding: The Return of Nature to a British Farm. London: Picador, 2018.

Vera, Frans W. M. Grazing Ecology and Forest History. Wallingford: CABI, 2000. doi:10.1079/9780851994420.0000.