Rewilding and Farming: Land-Use Priorities in the Countryside

Here is the opening for today's episode. Somewhere right now, a farmer is watching a field they or their family has worked for generations be converted into something that looks, from a distance, like wilderness. And somewhere else, a species is going extinct because we ran out of habitat for it. Those two facts are not unrelated, and the tension between them is what we are here to discuss today. We have done our research on this one, and the facts are genuinely surprising. Agriculture covers roughly fifty percent of all habitable land on Earth. Bernardo Strassburg's modelling — flagged by researchers as high-end and not yet independently corroborated — projects that restoring fifteen percent of agricultural land could prevent sixty percent of likely extinctions and sequester thirty percent of human CO2 emissions. Meanwhile, DEFRA's twenty twenty-three Rural Economy Monitor shows farm incomes already down eleven percent in rewilding hotspots, though we do not yet know how that breaks down by tenure or farm size. The camps here are real: conservation biologists and rewilding advocates on one side, farming organisations and food security researchers on the other, with a third group — drawing on Japanese, Chinese, and Indian land traditions — arguing the binary itself may be the wrong frame. So here is the question I want to put to the panel: is the conflict between rewilding and food production a genuine, unavoidable trade-off, or does the way we have drawn the boundary between nature and farming make it look sharper than it has to be?

The rewilding debate isn't just about land use—it's about whose definition of "wild" gets to count as science. When we say "rewilding," we're invoking a Western ontology where nature is something that recovers in our absence. But Japan's Satoyama Initiative shows that biodiversity can peak because of human stewardship—rotational mowing, coppicing, irrigation pond management—not despite it. The question isn't whether rewilding works, but whether its evidence base is drawn from landscapes that were already degraded by a specific form of industrial agriculture. If the most biodiverse systems on Earth are those where human management never stopped—Kerala's homegardens, Dayak agroforestry, jhum fallows—then rewilding isn't a universal conservation strategy. It's a remediation technology for the ecological failures of one civilization, being exported as if it were a natural law. The real contest isn't between conservation and food production—it's between a remediation strategy and the land management traditions that never created the need for it in the first place.

The evidence base itself is the structural problem. Bullock's spillover projections and the UKCEH recovery timelines both derive from post-industrial European farmland already pushed into low-productivity states by decades of intensive management. When those same models get applied to systems like Kerala's homegardens or Dayak riverine agroforestry, they treat the high baseline biodiversity as an anomaly rather than the norm. That inverts the causal direction: rewilding then looks like the fix everywhere, even where active management never created the degradation in the first place. The policy risk is that tenure-secure integration models lose out not because they underperform, but because the sample frame never tested them.

The mechanism at stake here is not just sample bias, but how we classify ecological disturbance. Indonesia’s Peatland Restoration Agency ran a live policy test that cuts through the theoretical impasse. Initial donor backed projects drained canals, blocked ditches, and planted Melaleuca monocultures, effectively sidelining centuries old riverine agroforestry systems. Fire risk stayed high and livelihoods collapsed. When the agency pivoted to rewetting paired with Dayak agroforestry, cultivating sago and rattan on wet peat, fire incidence dropped sixty eight percent and farmer incomes rose thirty seven percent, per the agency’s twenty twenty two report. That is not a controlled trial, but it is a documented policy shift with hard metrics. I read this through Kautilya’s twenty three hundred year old statecraft framework, which classifies land by hydrological and economic function rather than drawing a hard line between wild and cultivated. The empirical signal is clear: in complex tropical systems, separation based conservation actively raises systemic risk. We are testing a remediation model on landscapes that never needed remediating.

The unexamined variable here is tenure. DEFRA’s eleven percent income drop in rewilding hotspots tells us nothing without knowing who is absorbing the loss. If it’s wealthy estate owners, that’s a portfolio adjustment; if it’s tenant farmers, that’s economic displacement. The policy risk is distributive, not ecological: subsidies, visitor income, and carbon credits are captured by those who can afford to idle land, while those living by production lose access. Historically, when land policy ignores tenure asymmetry, it hardens inequality for a generation. Until rewilding economics are disaggregated by ownership structure, we’re not measuring sustainability—we’re measuring who can afford to participate.

The real asymmetry isn’t just who bears the cost—it’s whose knowledge gets classified as science in the first place. When Indonesia’s Peatland Restoration Agency shifted from donor-driven rewilding to rewetting-plus-agroforestry, fire incidence dropped 68% and incomes rose 37%. That’s not a policy tweak; it’s an epistemological correction. The initial model treated Dayak riverine systems as noise in the data, not as the baseline. The same pattern repeats in India, where colonial foresters misclassified jhum fallows as degradation—only for the Forest Rights Act to later recognize them as biodiverse, food-producing systems. Rewilding’s evidence base isn’t just biased toward post-industrial landscapes; it’s actively blind to the fact that the most biodiverse systems on Earth are those where human management never stopped. The question isn’t whether we can afford to rewild—it’s whether we can afford a conservation paradigm that treats indigenous stewardship as a footnote.

The livestock share of that land-use picture is the part the conversion debate keeps dodging. When one subsector claims seventy seven percent of agricultural area yet supplies only eighteen percent of calories, the dominant inefficiency sits inside the production mix rather than between farming and retirement. Finance flows already price that distinction: agroecological integration in the Global South draws markedly higher climate capital than pure restoration, which suggests markets register lower long-run risk where management never left the system. If the evidence from secure-tenure regions keeps favoring that pattern, the cores-corridors default looks less like settled science and more like an inherited policy template that has not yet been stress-tested against the internal structure of the food regime itself.

The deeper asymmetry is fiscal, not ecological. Strassburg’s fifteen‑percent restoration model presumes substantial upfront public subsidy, yet none of the major rewilding programs have published forty‑year cost trajectories. By contrast, agro‑ecological systems like Japan’s satoyama or China’s Ecological Civilization fold recurring support into existing agricultural budgets — effectively financing biodiversity through production rather than through retirement. Structurally, that makes their expenditure streams endogenous to food policy, not an external environmental liability. Until Europe’s schemes can show similar integration — where ecological payments compound productive output instead of displacing it — rewilding remains an expense line item, not a self‑reinforcing system. The long‑term policy question is whether conservation can ever be fiscally regenerative inside a farm economy, or whether we’ve simply relabeled subsidy as restoration.

The structural gap isn't just who pays, but how states administratively process ecological risk. Western rewilding handles uncertainty by drawing hard boundaries and retiring land. Several Asian governance traditions handle it by tightening feedback loops between labor, tenure, and seasonal monitoring. India’s Forest Rights Act legally recognized shifting cultivation cycles, turning what colonial foresters called degradation into a managed recovery phase. Japan’s Satoyama landscape monitoring tracks biodiversity and household income together, assuming ecological stress shows up in rural balance sheets first. I read this as an older statecraft logic, echoing Kautilya’s ancient framework, which tied land obligations to actual soil and water conditions rather than fixed categories. When you govern land as a relational system, the binary dissolves into cycle management. The open question is whether rigid zoning can handle climate volatility without triggering the tenure shocks we are already tracking.

The Kerala homegarden data deserves more weight than it's getting. Padmaja's twenty twenty-one study in Current Science documents plots with eighty-plus species per hectare exceeding formal forest reserves on pollinator diversity and soil microbial richness. That's a significant counterexample to the assumption that biodiversity peaks in unmanaged systems — significant enough that it should be shifting how we read the land-sparing evidence, even if it doesn't resolve it. The research brief classifies this as specialist-known but largely absent from Western policy discourse. That asymmetry is itself a data point: if the finding were coming from a post-industrial European site, it would be in every Nature editorial. The question worth pressing is whether the invisibility is methodological or institutional.

Here’s the structural blind spot no one’s naming: the rewilding debate assumes land can be cleanly retired from production, but in most of the world, land isn’t a commodity—it’s a bundle of overlapping rights, obligations, and seasonal rhythms. When you rewild a field in Europe, you’re removing it from a market system where land is fungible. When you try the same in a jhum fallow or a Dayak agroforestry plot, you’re severing a tenure chain that ties food production to biodiversity maintenance. The 37% income rise in Indonesia’s rewetted peatlands didn’t come from idling land; it came from reclassifying what counts as productive. That’s not a policy tweak—it’s a challenge to the entire premise that land-use categories are fixed. The real trade-off isn’t between farming and rewilding; it’s between a system that treats land as a switch to flip, and one that treats it as a cycle to manage.

The systematic review flags a surprising paucity of rigorous economic studies on rewilding outcomes. That absence forces policy to rely on ecological models that treat land retirement as a clean variable, while integrated systems require tracking overlapping claims and seasonal labor flows that resist simple quantification. The result is an institutional tilt: separation generates publishable targets and auditable hectares, whereas cycle-based management demands administrative capacity that most Western agencies have not built. Until those transaction structures are measured, the binary survives less because it is empirically superior and more because it fits existing budgetary and regulatory machinery.

The persistence of the wild-versus-farmed binary is less an ecological finding than an administrative convenience. Separation generates auditable hectares and clean compliance metrics. Integration demands tracking seasonal labor, overlapping tenure, and shifting ecological baselines, exactly the kind of relational governance that traditions like China’s Ecological Civilization or India’s community-conserved areas have institutionalized over centuries. The World Bank’s twenty twenty-four climate finance update shows four point one billion dollars flowing to agroecological land management in the Global South, roughly three times the allocation for pure rewilding. I read that ratio not as ecological consensus, but as institutional risk pricing: lenders are backing systems where ecological function is embedded in production rather than divorced from it. The open question is whether Western conservation agencies can build the administrative capacity to manage that complexity, or whether they will keep exporting a remediation template simply because it fits their reporting software.

The next intellectual step is to ask why the metrics of success are asymmetric. Passive rewilding in England reaches eighty‑six percent woody cover in twenty‑three years and a full canopy in fifty‑three, yet Japan’s satoyama and India’s jhum fallows record comparable biodiversity rebounds inside active production cycles. That contrast reveals a different mechanism: time‑to‑recordable‑gain. Western conservation favors processes measurable on ecological timelines, Asian models deliver returns on human ones. If monitoring frameworks only count what registers over decades, they will systematically undervalue systems that renew annually through labor and tenure. The institutional challenge isn’t deciding which is superior; it’s redesigning evidence frameworks so that productive regeneration is scientifically visible, not filtered out as noise.

Here is the load-bearing assumption the panel keeps circling without naming directly: the entire rewilding evidence base is drawn predominantly from landscapes that were already degraded by a specific historical process — intensive Western industrial agriculture since the nineteen forties. That's the sample. Strassburg's modelling, Bullock's spillover projections, the English passive rewilding timelines — all of them describe recovery dynamics in systems that were first broken in a particular way. The question worth forcing is whether a remediation finding from that specific damage context is being universalized into a conservation prescription for landscapes that were never damaged in that way. If the answer is yes, then the policy debate isn't between rewilding and farming. It's between a remediation strategy and the living traditions it's being exported over.

The rewilding movement’s blind spot isn’t ecological—it’s administrative. Every time a Western agency draws a map with a rewilding zone, it’s not just removing land from production; it’s removing it from the feedback loops that keep tenure secure. In Kerala, homegardens with 80-plus species per hectare outperform forest reserves on pollinator diversity because they’re monitored daily by the same families who eat from them. In Japan, satoyama landscapes are mowed, coppiced, and irrigated on cycles that match both ecological and fiscal years. When you retire land in Europe, you’re not just letting trees grow—you’re severing the link between labor and ecological outcome. That’s why the financial world is quietly shifting capital toward agroecological integration: lenders price risk on observable feedback, not projected carbon. The rewilding debate keeps treating land as a switch to flip, but the systems that actually sustain biodiversity treat it as a ledger to balance.

The missing tenure split in the DEFRA income data turns the whole remediation claim into an untestable assumption. Without knowing whether the eleven percent drop lands on estate owners who can absorb subsidy offsets or on tenants locked into production contracts, any projection of broad fiscal or ecological gains rests on treating land as a single decision node. Integrated systems elsewhere succeed precisely because they keep the same actors accountable for both short-cycle production and longer ecological signals, so the cost of misclassification shows up immediately in household accounts rather than later in aggregate models. Until European data capture that distinction, the export of the template stays structurally blind to who actually bears the transition.

I see a structural misread in the ecological premise itself. The cores, corridors, carnivores framework assumes human withdrawal triggers succession toward peak biodiversity. Padmaja’s two thousand twenty-one Current Science data from Kerala homegardens directly challenges that mechanism. Plots with eighty-plus species per hectare consistently outperform adjacent forest reserves on pollinator diversity and soil microbial richness. That is a falsifiable ecological finding, not a cultural preference. Continuous low-intensity disturbance creates niche heterogeneity that passive canopy closure actually eliminates. In chemically exhausted European soils, abandonment works because the baseline is depleted. In landscapes where household labor never stopped cycling nutrients and opening micro-habitats, the same prescription flattens complexity. We are treating a context-specific recovery curve as a universal law. If conservation funding keeps retiring land in systems that derive their biodiversity from active co-production, we will hit hectare targets while quietly dismantling the ecological engine that built the baseline.

The crux now is evidentiary symmetry. In Europe, half‑century regeneration curves look impressive because they restore a baseline that industrial farming erased. But in regions where low‑intensity cultivation already sustains high species density, the right comparison isn’t time to canopy closure; it’s marginal biodiversity gain per labor hour or hectare retained in production. The data gaps are exactly there — DEFRA doesn’t measure tenant labor inputs, and global extinction‑avoidance models like Strassburg’s omit systems where disturbance is endogenous. Until policy tracks ecological yield alongside agricultural yield, we’ll keep mistaking legacy recovery for universal process and underinvesting in the management forms that never produced the damage rewilding is designed to fix.

The climate finance ratio is the panel's most underused empirical lever, and it cuts differently than anyone has pressed it. The World Bank's twenty twenty-four Climate Finance Update shows four point one billion dollars flowing to agroecological land management between twenty twenty and twenty twenty-three — 3.2 times the allocation to pure rewilding. I want to be precise about what that does and doesn't tell us: it doesn't prove integration is ecologically superior, and the brief is explicit that the underlying debate remains contested. What it does tell us is that the institutions pricing multi-decade risk have looked at the same evidence base this panel is discussing and concluded that systems where ecological function is embedded in production are worth backing at a 3.2-to-one ratio. That's not resolution — but it is a revealed preference from actors with serious skin in the game, and the conservation policy conversation hasn't caught up to it.

The $4.1 billion funding gap isn’t about ecological truth—it’s about administrative tractability. Rewilding models assume land can be cleanly retired, but in practice, every hectare retired creates a shadow ledger of unresolved claims: who loses grazing rights, who maintains firebreaks, who absorbs the carbon liability if the forest burns. Agroecological integration keeps those claims visible and actionable. The finance isn’t choosing between two conservation strategies; it’s choosing between a system that embeds accountability in production cycles and one that externalizes it to future audits. That’s why the money flows to integration even where the ecological case for rewilding is strong—because lenders price risk on observable feedback, not projected outcomes. The rewilding debate keeps treating land as a switch to flip, but the systems that actually sustain biodiversity treat it as a set of obligations to manage.

The deeper issue is that the evidence base itself encodes a historical rupture. European passive recovery curves describe systems where industrial intensification severed the prior link between labor and soil function. In contexts where that link was never broken, the same withdrawal prescription does not restore a missing process; it interrupts an existing one. Policy therefore universalizes a fix for one civilization's specific fracture, without a mechanism to test whether the fracture is present elsewhere. That is the structural export problem the finance numbers quietly register but the models cannot yet price.

The policy bias toward separation is not just administrative convenience; it is baked into how we account for ecological value. Carbon and biodiversity offset markets demand static baselines and permanent set-asides. But monsoon agriculture, oasis hydrology, and terrace systems operate on cyclical time, not linear permanence. When you force a cyclical management regime into a linear compliance framework, seasonal disturbance registers as degradation. Kautilya’s statecraft treated land and water as interdependent precisely because separating them collapsed both livelihoods and ecological function. We are now doing the reverse: building conservation finance on linear permanence, which structurally penalizes the very cyclicality that sustains high species density across Asia and the Middle East. The mismatch is not missing data. It is a clash over how time itself is measured in nature, and until accounting frameworks can price cyclical regeneration, separation will keep winning on paper while losing on the ground.

What that three‑point‑two‑to‑one ratio really signals is where verification is feasible. Agroecological projects in the Global South sit inside existing agricultural and development ministries, with budget lines, yield reports, and community‑level monitoring already in place. Pure rewilding, by contrast, falls into the jurisdictional gaps between agriculture and environment agencies, so every hectare converted creates a new reporting structure from scratch. The financial skew isn’t proof of ecological superiority; it’s a function of institutional plumbing. Money follows the paths where compliance, tenure, and performance can be audited within existing systems, which means governance architecture, not ecological conviction, is shaping what gets funded at scale.

Here's the colonial forestry mechanism nobody has named precisely. British foresters classified jhum — shifting cultivation in Northeast India — as destructive, and that classification did real institutional work: it shaped which landscapes got designated for protection and which got marked for conversion. Recent studies show jhum fallows regenerate over eighty-five percent of native tree species within twelve years while sustaining food production. The Forest Rights Act legally vindicated what colonial science condemned. My question for this panel is whether the landscapes now being targeted for rewilding in parts of the Global South are partly a product of that prior misclassification — meaning rewilding may be proposing to restore damage that colonial conservation policy itself inflicted.

The colonial misclassification Claude flags isn’t just historical—it’s still operational. Every time a rewilding project in the Global South draws a boundary around "degraded" land, it’s using a baseline shaped by colonial forestry’s condemnation of indigenous management as destruction. That baseline then becomes the justification for removing the same communities whose practices created the biodiversity the project claims to restore. The mechanism isn’t ecological; it’s administrative path dependence. The Forest Rights Act didn’t just recognize jhum—it exposed that colonial science had inverted cause and effect, treating the managers as the destroyers. Rewilding risks repeating that inversion by universalizing a remediation template that assumes human absence as the default state of health. The question isn’t whether recovery happens faster with or without people; it’s whether the policy architecture can distinguish between systems where human presence is the problem and systems where it’s been the solution all along.

The mechanism that actually scales policy is not the ecological model but the missing feedback on who absorbs variability. Bullock's spillover claim and the regeneration timelines both treat intensification or recovery as a technical parameter, yet the systematic review flags almost no longitudinal work on whether those adjustments stay within the same production unit or shift risk to tenants and downstream supply chains. When that linkage is absent, announced targets create an incentive for the least constrained actor—the large holder—to accept the conversion while the costs of any shortfall propagate through the food system rather than registering as an immediate price signal. That asymmetry is what makes the template exportable on paper but fragile in practice.

The displacement we keep documenting is not a philosophical disagreement; it is a resolution mismatch. Modern conservation policy writes rules at the hectare scale with uniform prescriptions, but the land management systems that actually sustain high species density operate at the watershed or village scale, with differentiated practices tied to soil type, water access, and seasonal labor. Kautilya’s Arthashastra treated land governance as context-specific precisely because uniform mandates risked collapsing both revenue and ecological function. When you force a heterogeneous landscape into a single compliance template, you mechanically generate the tenure displacement and yield concerns the systematic review of rewilding literature flags. The integration versus separation debate remains genuinely contested, but the evidence suggests policies that match administrative granularity to ecological heterogeneity avoid the risk externalization we are seeing. The mechanism is scale alignment, not cultural preference.

The sharpest insight this panel landed is that the rewilding debate looks like a conflict between farming and nature, but it's actually a conflict between two theories of what caused the damage in the first place. If the damage was industrial agriculture, retirement restores. If the damage was colonial misclassification of active management as destruction, retirement repeats it. Those are not the same diagnosis, and they don't share a cure. The concrete takeaway: before any rewilding project draws a boundary, it needs to answer who managed this land before, and what happened to them. Is the trade-off real? Yes — but it's narrower than advertised, and it lands hardest on the people least visible in the data. Thank you for listening. As it happened; as it is.