Insight, Nature Restoration
Twenty-seven draft restoration plans were submitted yesterday. The ecological question of which habitats to restore is largely settled. The unsettled question is whose land changes use, and the people who answer it have every reason to organise while the beneficiaries do not.
Member States were required to submit draft National Restoration Plans to the European Commission by 1 September 2026, under Regulation (EU) 2024/1991. The Commission assesses each draft within six months, final plans follow around September 2027, and the first revision is due by 30 June 2032.
The plans have to set out, in the words of the coalition of environmental organisations tracking them, where restoration will happen, what measures will be taken, who is responsible, and how it will be financed. The Commission and the European Environment Agency have supplied a uniform format, data models and reporting templates, and the guidance to Member States includes an instruction that is easy to skim past: allow sufficient time for stakeholder participation, because it is key to success and it takes time.
The ecological science underneath this is in reasonable shape. Which habitats are degraded, what restoration measures work on them and what the aggregate benefit looks like are all reasonably well established. More than 80 per cent of European habitats are in poor condition, and the case for acting is not seriously contested by anybody who has looked at it.
The unsettled question is a different kind of question, and no amount of ecological modelling answers it. Restoration happens on land. Land belongs to somebody, is farmed by somebody, and is often farmed by somebody other than the person who owns it. Deciding where restoration happens is deciding whose land changes use, and that is an allocation decision dressed as a technical one.
Mancur Olson's argument in The Logic of Collective Action (1965) is usually cited for the case where benefits are concentrated and costs are diffuse, which is why a small industry lobbies effectively against a large public. The mechanism is about concentration: a small group with a large individual stake organises, and a large group with a small individual stake does not, because each member gains too little to justify the effort and can free-ride on others.
Restoration is the mirror image, and the mechanism runs the same way with the sign reversed.
The costs are concentrated. A farmer whose field is rewetted, a landowner whose drainage is blocked, a forestry business whose rotation changes, a municipality whose development land is reclassified. Each bears a large, identifiable, personal loss. They know exactly who they are, they can find each other, and they have every reason to organise.
The benefits are diffuse. Flood attenuation for a town downstream, water quality for a catchment, carbon storage for everybody, biodiversity for the future. Each beneficiary gains something real and small, most do not know they are beneficiaries, and none will ever attribute a flood that did not happen to a wetland restored fifteen years earlier.
That asymmetry predicts, without any bad faith on anyone's part, that opposition will be organised and articulate while support will be broad and passive. It also predicts where restoration will actually be sited: on land whose users are least able to object. Public land, marginal land, land held by absent owners, land farmed by tenants rather than owners, and regions with less political weight.
Which means the distributional question is not whether restoration is worth doing. It is whether the sites chosen reflect ecological priority or the path of least resistance, and those two criteria produce different maps.
That is checkable. It requires comparing the ecological priority ranking against the sites actually selected, and looking at the tenure, ownership and political characteristics of both sets. Nobody appears to be planning to do it, and the draft plans submitted yesterday are the baseline that would make it possible.
The compensation instruments available are mostly annual payments for foregone production, and they are poorly matched to what is being asked.
Restoration is frequently irreversible in practice. Rewetting peat, removing drainage, breaching an embankment or restoring a floodplain changes what the land can do more or less permanently. So a farmer is being asked to accept a permanent change to an asset in exchange for a payment stream that lasts as long as a funding programme, in a policy environment that has changed direction twice in a decade.
Read from the landowner's side, declining is not obstruction. It is a rational assessment of counterparty risk against an irreversible commitment. The same reasoning applies as in the trough farmers face during a practice change: the instrument on offer does not match the shape of what is being asked.
The instruments that would match are different in kind. A capital payment reflecting the permanent change in land value rather than an annual income substitute. Purchase or a permanent easement, which converts an ongoing negotiation into a completed transaction. Or a genuine role in the resulting asset, so that a landowner whose field becomes a flood margin has a share in what it produces rather than a receipt for what it stopped producing.
And there is a tenure problem underneath all of it. A tenant cannot consent to a permanent change in land they do not own, and an owner has little reason to accept one on land somebody else farms. The right that matters is not the one the scheme addresses, which is the finding that ran through the evidence on tree planting and applies with more force here.
The reporting architecture is already built. Member States will report hectares under restoration measures, indicator trends and expenditure, through a common format and a shared platform. That is a genuine achievement and it measures implementation.
Four things would measure whether it works.
The gap between the ecological priority map and the selected sites. The single most informative analysis available, obtainable from documents that now exist, and it directly tests whether allocation followed evidence or resistance.
Consent quality, not consultation counts. How many landowners were engaged is an activity measure. Whether the people whose land changed understood what they were agreeing to, felt able to decline, and would agree again three years later is the outcome, and it requires asking them.
Who bore the cost, and whether they were compensated for the right thing. Owner or tenant, and whether the payment matched a permanent change or an annual loss.
Persistence and maintenance at ten years. Restoration reverses. Blocked drains get unblocked, structures fail, management lapses when a payment ends, and the first revision cycle in 2032 is the natural point to establish what survived. The regulation's own ten-year revision rhythm makes proper longitudinal measurement possible in a way that most programmes never allow, and that opportunity will be wasted by default unless a baseline is taken now.
The draft plans submitted yesterday are that baseline. They state intentions, sites and mechanisms before implementation has confounded anything, and there is a narrow window, roughly until the final plans in September 2027, in which the comparison can be set up rather than reconstructed.
The Lab works on this in climate and ecosystems and regenerative agriculture, through field research with the people whose land a plan lands on rather than with the plan.
If you are working on restoration planning and want the allocation question examined while the drafts are still drafts, tell us what you need to know.
This is an independent insight piece by Transitions Lab. For the Lab's applied work, see Climate & Ecosystems. See also The Right That Matters Is to the Tree on why the specific right, not ownership, predicts investment in land, The Trough Before the Dividend on instruments that do not match the shape of what is being asked, and The Municipality Is the Instrument on why implementation capacity decides what a plan delivers. To discuss a study, see Contact.