01 The question, as it was asked
You cannot get nature to sign a contract with you.
The question arrives in supplier meetings again and again, and it never fits the form we hand people. What can we do more on nature — land, forest, watershed? And immediately the form asks for a KPI: how much carbon back per dollar in, by when, verified how.
But a forest is a system, not a counterparty. It has fire years and drought years. It can be restored and still lose canopy. It can hold carbon and lose species. The impact elements are many, and some of them move in the opposite direction to each other. So the honest answer to “how many tonnes for how many dollars” is: it depends on the place, the people and the decade.
“Is that the reason we just make it a conference room topic, year after year, and never put it into action?”
This station does not close the question. It gathers what is known around it — the scale, the regulation, the standards, the arithmetic, the failure patterns, and the companies already spending real money — so that the next time it comes up, it comes up with evidence attached, and one piece of it can actually start.
02 Small issue or big issue — data talks
The land question is not a side topic. It is roughly a third of the problem.
Every figure below links to its primary source. Check the current edition before you quote it — these datasets update annually.
Forest cleared each year, 2015–2020. Net forest loss ran at about 4.7 million hectares a year over 2010–2020 once regrowth is counted.
FAO FRA 2020 →Driven by agricultural expansion — cropland and grazing. This is a supply chain problem before it is an environmental one.
FAO remote sensing survey →Roughly ten football pitches a minute. In 2024 the figure rose sharply again, fire-driven, to the worst year in the record.
WRI / Global Forest Watch →Agriculture, forestry and other land use. The sector you have not scoped is comparable in size to the electricity work you have already done.
IPCC AR6 WGIII →Available from natural climate solutions — protection, improved management, restoration. They receive a small fraction of climate finance.
Griscom et al., PNAS 2017 →Economic value generation moderately or highly dependent on nature and its services. Nature loss is a business continuity risk, not a CSR line.
WEF Nature Risk Rising →Significantly altered by human action; around one million species face extinction. The baseline you are restoring against has already moved.
IPBES Global Assessment →Target materials needed for renewable energy. Around 8% of mining areas overlap protected areas and 7% Key Biodiversity Areas — the clean transition has a land footprint.
Sonter et al., Nature Comms 2020 →Still lack deforestation commitments covering all their forest-risk commodities, and far fewer can evidence implementation. The gap is execution, not intent.
Forest 500, Global Canopy →03 Why it resists measurement
Carbon is a single number. Land is a place.
Energy work was tractable because a megawatt-hour is a megawatt-hour anywhere on earth, and the counterparty signs a PPA. Land refuses all four of those conveniences:
A hectare is not a hectare
Two hectares of restored land with identical carbon can differ by an order of magnitude in biodiversity, water regulation and community value. There is no single unit, so there is no single price.
Effort in ≠ outcome out
Fire, drought, pest, tenure conflict and a neighbour's decision all sit between your money and the outcome. Some forests decline while under protection. This is real, not a reporting failure.
Permanence is a social question
Carbon stored in trees can be released in an afternoon. Whether it stays depends on land tenure, livelihoods and governance — things a procurement contract does not control.
Ecological time vs contract time
Meaningful recovery runs 10–30 years. Supplier contracts run 1–3. That mismatch, not the science, is usually what kills the investment case.
“If it cannot be measured, it is hard to invest.”
The counter-argument, plainly: we have never required precision before acting on the things that mattered. Scope 3 was un-measurable in 2015 and is a disclosure requirement now. Water risk was site-specific and got a shared tool. What changed was not the science — it was that a market rule made a rough number mandatory, and rough numbers improved fast once they were being used. Land is at exactly that point.
04 The regulatory floor · market access
If your product enters the EU, this stopped being voluntary.
This is the strongest argument you have inside the company: it is no longer a nature programme, it is a condition of sale. Verify current application dates — the EU timetable has been amended more than once.
Deforestation-free products
Cattle, cocoa, coffee, oil palm, rubber, soya and wood — plus derived products, including leather, tyres, furniture, paper and packaging — may only be placed on the EU market if they are deforestation-free against a 31 December 2020 cut-off, legally produced, and covered by a due diligence statement.
The operative demand: geolocation coordinates of every plot of land where the commodity was produced, with polygons required above 4 ha. That is the clause that reaches all the way down your chain.
Regulation textDue diligence with teeth
Requires in-scope companies to identify, prevent and mitigate adverse human rights and environmental impacts across the chain of activities, and to adopt a climate transition plan. Environmental impacts include ecosystem degradation and protected-area harm — not just emissions.
Directive textBiodiversity becomes a disclosure
ESRS E4 asks for material impacts on biodiversity and ecosystems, including land-use change, sites in or near biodiversity-sensitive areas, and transition plans. It effectively forces the scoping exercise you are describing.
ESRS delegated actIllegal deforestation ban
Prohibits use of forest-risk commodities produced on land occupied illegally, with due diligence and reporting duties. Secondary legislation has been slow — but design for it, because it is narrower than EUDR and easy to bolt on once EUDR data exists.
Schedule 17Illegal plant & timber trade
Already law: it is unlawful to import plants or plant products harvested in violation of foreign law, with declaration requirements that have been phased across product categories including composite wood and paper. Enforcement is criminal, and the burden falls on the importer.
Lacey Act overviewWhat is coming
The FOREST Act (illegal-deforestation import restrictions on a defined commodity list) has been reintroduced repeatedly without passing. Treat North America as a Lacey Act floor today, with EUDR-grade data as the design target — you will not want to build two systems.
Nature Restoration Regulation
Binding restoration targets for Member States across ecosystems. Relevant to your factories because it changes permitting, land availability and expectations for industrial sites inside the EU.
Regulation textA working precedent for “pay it back”
Most new development must deliver at least 10% measurable biodiversity gain, secured for 30 years, using a statutory metric. Whatever you think of the metric, it proves the model your question is reaching for: convert land, owe habitat, measure it in defined units, hold it for decades.
BNG guidance05 Standards · do not invent your own
The definitions already exist. Adopt them and skip two years of debate.
Accountability Framework (AFi)
The reference for what “deforestation-free”, “conversion-free”, cut-off date, and credible implementation actually mean. If your policy language does not match AFi, expect it to be challenged.
accountability-framework.orgSBTN — Science Based Targets for Nature
Five steps: assess, interpret & prioritise, measure/set/disclose, act, track. Its AR3T framing — Avoid, Reduce, Restore & Regenerate, Transform — is the sequencing your question needs.
sciencebasedtargetsnetwork.orgSBTi FLAG
For companies with material land-based emissions: separate FLAG targets plus a no-deforestation commitment with a cut-off no later than 2025. This is where land stops being optional inside a carbon target.
sciencebasedtargets.orgGHG Protocol Land Sector & Removals
How to account for land-use change emissions, biogenic carbon and removals in the inventory — so restoration and conversion both land in the same book instead of a slide.
ghgprotocol.orgTNFD & the LEAP approach
Locate the interface with nature, Evaluate dependencies and impacts, Assess risks, Prepare to respond. LEAP is the most practical answer to “where do we even start scoping?”
tnfd.globalIBAT · WDPA · KBA
Drop your site and supplier coordinates into a protected-area and Key Biodiversity Area screen. One afternoon of work gives you a defensible risk-ranked site list — the cheapest first step in this whole subject.
ibat-alliance.orgIRMA
Independently audited mine-site standard covering biodiversity, no-go areas, water, tailings and community consent. The only widely accepted way to say something true about a mine you do not own.
responsiblemining.netResponsible Minerals Initiative
Smelter and refiner assessment infrastructure already used for conflict minerals. It is the existing pipe through which land and biodiversity questions can reach refiners — reuse it rather than building new.
responsiblemineralsinitiative.orgTrase · Global Forest Watch · SPOTT
Trase links commodity trade flows to sub-national deforestation risk. GFW gives near-real-time alerts. SPOTT scores producer transparency. All three are free — you can risk-map your top commodities before asking anyone for budget.
trase.earth06 The simple arithmetic you asked for
A land ledger for one factory.
You asked: if a factory occupies X hectares, and before that it was something — how do we scope what to give back, and where? Here is a minimum viable method. It is deliberately crude. Crude and used beats precise and pending.
- Step 01Draw the fence lineTotal site footprint in hectares from the lease or title — including car parks, ponds, worker housing, on-site quarry, access road and any land cleared for expansion but not yet built. Most site data understates this by a third.
- Step 02Recover the “before”Take the site centroid and pull historical land cover — Hansen tree cover 2000 via Global Forest Watch, ESA WorldCover, national wetland inventories. Classify what was there: natural forest, wetland, peat, native grassland, mangrove, already-degraded, or already-industrial.
- Step 03Fix one cut-off date and never move itUse 31 December 2020 to align with EUDR and SBTi FLAG. Conversion before it is baseline; conversion after it is a liability with a remediation obligation attached.
- Step 04Add the footprint that is not inside the fenceWater abstracted (and the catchment it comes from), tailings and overburden for materials you buy, land under biomass or hydro you count as renewable, and the haul road that opened the frontier. This is where the real number lives.
- Step 05Place the payback in the same watershedSame ecoregion, same catchment, ideally buffering or connecting an existing Key Biodiversity Area. Restoration two countries away is a purchase, not a repair. Confirm land tenure and community consent before a single tree is bought.
Nobody will accept this as science. That is fine — it is not science, it is a ledger. Its job is to convert “good to do” into a number with a unit, an owner, a location and a date, so it can be argued with. Every measurement system that works started as something this rough.
07 Sub-tier · the example you raised
Follow the battery down until you hit soil.
Most programmes stop at the cell maker. The land impact is four tiers further down, and visibility collapses exactly where it starts to matter.
- Tier 0Pack & module assemblyKnown, audited, in the system. Land impact: the site itself.Visible
- Tier 1Cell manufacturingNamed, contracted, energy and water KPIs in place. Where most “battery sustainability” programmes end.Visible
- Tier 2Cathode & anode active material, separator, foil, electrolyteOften known by name, rarely by site. Energy-intensive; frequently co-located with captive coal power in industrial parks — an energy and land footprint at once.Partly visible
- Tier 3Refining — nickel sulphate, lithium hydroxide, cobalt sulphate, graphiteThe real chokepoint and the real emissions. Smelting parks bring roads, worker settlements, port expansion and tailings. Land change here is mostly outside anyone's supplier list.Mostly dark
- Tier 4Mine & concentratorOpen-pit stripping, overburden, tailings storage, haul roads, water drawdown. Lithium brine operations change hydrology across a whole salar; laterite nickel operations clear forest by design.Mostly dark
- Tier 5The land itself, and the people on itThe frontier the access road opened. Induced settlement, downstream sediment, fisheries, tenure conflict, loss of a Key Biodiversity Area. Almost never in any scope, and the largest impact in the chain.Invisible
The road, not the pit
Mine footprints look small on a map. The access infrastructure that reaches them is what opens forest to settlement, logging and fire. Measuring only the licence area systematically understates impact.
The transition's own footprint
82% of mining areas target materials needed for renewable energy production; a meaningful share overlaps protected areas, KBAs and wilderness. Decarbonisation buys us land pressure unless it is designed against.
Sonter et al. 2020Chain of custody breaks at the smelter
Refined metal is fungible; that is the point of a refinery. Once mixed, provenance is gone. Which is why the only workable lever is refiner-level sourcing rules plus mine-site assurance — the RMI model, extended from conflict to land.
08 Failure patterns
What usually goes wrong — worth knowing before you spend.
A mill list is not traceability
Publishing a palm mill list says where oil was crushed, not where trees fell. Plantation-level polygons are the unit that matters; the same gap exists between a smelter list and a mine.
Indirect suppliers absorb the risk
In cattle, animals move from a cleared property to a clean one before sale. Monitoring only direct suppliers produces a clean dashboard and an unchanged forest. Any tier-1-only programme has this hole.
Single coordinates are unverifiable
A point tells you nothing about the 300 ha around it. EUDR asks for polygons above 4 ha for exactly this reason. Collect polygons from day one or collect twice.
Credits that did not represent real reductions
Independent analysis of voluntary forest-carbon projects found a large share of credits overstated impact relative to what would have happened anyway. Buying credits instead of fixing sourcing is the most expensive way to fail.
West et al., Science 2023Over-crediting through baseline choice
Analysis of a major compliance forest-offset programme estimated roughly 30% over-crediting from crediting rules alone. Method design, not intent, decides whether the tonne is real.
CarbonPlan analysisPlantation is not forest
Fast-growing monoculture on former grassland can lower water tables, reduce biodiversity and burn. “Trees planted” is the weakest metric in this field. Count hectares under improved management and survival at year 5 instead.
No tenure, no permanence
Restoration without secure land rights and free, prior and informed consent gets reversed by the next economic cycle. The social clause is not the soft part of the programme; it is the durability mechanism.
Cut-and-run instead of remediation
Terminating a supplier caught converting land moves the volume to a buyer who asks nothing. Default to a time-bound, verified remediation plan, with exit as the consequence of refusal — not of disclosure.
Waiting for the perfect metric
The most common failure of all, and the one your question names. Five years of conference-room debate cost more forest than an imperfect programme would have. Start with one commodity, one country, one cut-off date.
09 Practice · money, data, still working
Who is actually doing it — and what they ask of suppliers.
Chosen for published data and supplier requirements, not for pledges. Read their latest report before citing — everything here moves year to year, and none of these companies is finished.
Unilever
Publishes deforestation-free progress by commodity with volume percentages, works to plot-level geolocation and polygon data with satellite monitoring, and pairs it with regenerative agriculture requirements and a nature fund. The useful thing to copy: they report the share of volume verified, not the share of suppliers who signed a policy.
unilever.com/sustainability/natureNestlé
Assesses primary supply chains for palm oil, soy, pulp & paper, meat and sugar as deforestation-free using supply chain mapping plus satellite verification, and funds landscape-level “forest positive” work beyond its own volume — the honest acknowledgement that a supply chain sits inside a landscape it does not own.
nestle.com/sustainabilityMars
Deliberately simplified its palm supply chain — cutting from well over a thousand mills to a small, deeply traceable set — on the logic that you cannot govern what you cannot see. The most transferable idea on this page: reduce supplier count to buy visibility.
mars.comKering
Runs an Environmental Profit & Loss account that prices land use, water and emissions across all tiers including raw material production, has a dedicated biodiversity strategy with a regenerative agriculture target measured in hectares, and funds conversion of supplier land. Monetising tier-4 land use is what makes it a board conversation.
kering.com/sustainabilityApple
Its Restore Fund puts equity-scale capital into working forests and restoration to generate both carbon removal and financial return, alongside deforestation-free requirements on fibre packaging. The lesson: structured as an investment vehicle, it survives budget cycles that a CSR line item would not.
apple.com/environmentBMW Group, Volvo Cars, Ford
All three engage IRMA-based assessment of mine sites for battery raw materials, and several have piloted blockchain-backed traceability for cobalt from mine to cell. This is the clearest existing answer to “how do we say something true about a mine four tiers away”.
IRMAInter IKEA / IKEA
Certified wood sourcing at very high volume share, plus direct forest land ownership and management — vertical integration as a traceability strategy. Also a reminder that certification percentage is a floor, not the finish line.
about.ikea.com/sustainabilityThe Amazon Soy Moratorium
Not a brand — a buyers' agreement. After traders committed not to buy soy from land deforested after a cut-off date, the share of soy expansion coming from newly cleared forest in the Brazilian Amazon fell dramatically. A cut-off date plus collective refusal to buy changed a landscape. That is the precedent.
Gibbs et al., Science 201510 Make it a must, not a nice-to-have
Eight clauses that travel upstream.
Written to be pasted into a supplier code, an RFQ scorecard or a contract annex. Each one is enforceable, evidence-backed, and already used somewhere.
- Clause 1Polygons, not pointsSupplier provides WGS84 polygon geolocation for every production site and, for forest-risk commodities, every plot or mine licence area feeding it. Point coordinates accepted only below 4 ha.
- Clause 2One cut-off dateNo conversion of natural ecosystems after 31 December 2020 — aligned with EUDR, AFi and SBTi FLAG so the supplier is not asked three different questions by three different customers.
- Clause 3All ecosystems, not just legal forestScope covers natural forest, peatland, wetland, mangrove, native grassland and savanna. “Legally permitted clearance” is not an exemption.
- Clause 4Flow-down to the sourceSupplier contractually imposes the same terms on sub-tier suppliers to the farm, plantation, forest concession or mine, and provides a named tier-n list refreshed twice a year.
- Clause 5Independent evidenceSatellite monitoring of supplied plots with alert response within 30 days; third-party assurance for high-risk nodes (IRMA for mine sites, credible certification plus verification for commodities).
- Clause 6No-go listNo sourcing from World Heritage sites, Ramsar wetlands, IUCN category I–IV protected areas, primary or High Carbon Stock forest, or land taken without free, prior and informed consent.
- Clause 7Remediation before terminationNon-compliance found after the cut-off triggers a time-bound, verified remediation and restoration plan. Exit follows refusal to remediate — never disclosure itself, or you will never be told again.
- Clause 8Report the land ledgerAnnual site footprint, pre-conversion land cover, water catchment and restoration hectares — reported as m²·yr per unit shipped, so land enters the product scorecard beside energy and recycled content.
Start with one commodity, one region, one tier deeper than you currently see. Publish the number even when it is embarrassing. An embarrassing number that exists moves faster than a perfect number that is still being designed.
11 Sources · go to the raw material
Every link, so you can argue with the original.
If reading this raised a question, that is the point. Open the primary source, check whether the edition has moved, benchmark one of your own sites against it, and bring back what you find.
Keep it open
Do you sometimes think about this question too?
It was not posted here for an answer. It was posted so it does not get lost between the energy programme and the packaging programme — so that next quarter someone opens it again and finds it has grown a little more evidence.
Take any part of it into your meeting. Argue with the multipliers. Replace the battery example with your own commodity. Run the ledger on one site and see how wrong it is. Then tell someone what you learned.
Not just a good-to-do. A must, measured — even if the first measurement is rough.
Open station · still practising · still learning