Does TNFD Require a Biodiversity Footprint? Metrics, Models and False Precision

Technical status. TNFD’s metric architecture is current, while methods for state-of-nature measurement and biodiversity footprinting continue to develop. TNFD’s 2023 footprinting paper focuses primarily on financial institutions and is a discussion paper, not a requirement. The 2026 state-of-nature paper is also open for consultation and should be labelled as non-final.

Limitation. This article does not endorse a particular commercial model or determine ecological significance. Specialist methods require appropriate licences, data governance, scientific expertise and validation against the organisation’s purpose and locations.

Boards, investors and reporting teams often ask for a biodiversity equivalent of tonnes of carbon dioxide. The attraction is understandable: one score appears easy to aggregate, compare and track. Nature is more difficult because ecological condition, species, water, land, pollution, ecosystem function and cultural values are not interchangeable in the way a common physical unit can suggest.

A single footprint can hide critical distinctions:

one hectare of intact habitat is not equivalent to one hectare of heavily modified land;

impacts in different ecosystems or countries may not be substitutable;

a small area can be highly important for a threatened species;

water withdrawal depends on basin condition and competing users;

pollution pressure and habitat conversion create different pathways;

ecological recovery can take decades and may fail; and

model outputs may reflect assumptions rather than measured outcomes.

The practical question is therefore not “Which score is the TNFD score?” It is “Which metric or model is fit for the decision, and what context must remain visible?”

TNFD Metrics and Targets A-C ask organisations to disclose metrics used to assess and manage material nature-related risks and opportunities, metrics used to assess and manage material dependencies and impacts, and targets and performance. The architecture includes:

core global disclosure metrics;

core sector disclosure metrics;

additional global and sector metrics; and

assessment metrics used internally during LEAP.

Core metrics provide a cross-sector starting point, but they do not create one composite biodiversity score. TNFD’s current resources state that no single state-of-nature metric captures all relevant dimensions. Ecosystem condition and species extinction risk therefore remain core global placeholder indicators while consensus and methods continue to develop.

An organisation can use a footprint model as one input, but it still needs to disclose the metrics that represent its material pathways and explain the basis used.

A biodiversity footprint converts multiple pressures or activities into an estimated effect on biodiversity or ecosystem condition. Depending on the method, inputs may include land occupation, land-use change, water use, pollution, climate pressures, commodity sourcing, sector averages and geographic data.

The model may produce outputs such as:

potentially disappeared fraction of species;

mean species abundance loss;

habitat-condition-adjusted area;

species-risk indicators;

ecosystem-integrity measures;

biodiversity-impact units; or

a proprietary composite score.

These outputs are not automatically comparable. They can represent different endpoints, spatial scales, time horizons, baselines and assumptions. The unit name alone does not explain what was measured.

Figure 1. Biodiversity measurement has several layers: business activity, pressure, location and ecosystem context, modelled change, observed outcomes and decision use. No single layer replaces the others.

1. Location can disappear in aggregation

Nature-related effects depend on the receiving ecosystem. A portfolio-level score may indicate relative hotspots, but it can conceal the difference between a moderate score in a highly sensitive location and a larger score in a heavily modified area.

The organisation should retain the ability to trace the aggregate back to countries, basins, landscapes, sites, commodities or assets. If location is based on sector averages rather than actual coordinates, that limitation should be explicit.

2. Different dimensions are combined through value choices

Composite models frequently weight species, habitats, ecosystem condition or pressures. The weighting can embed scientific and normative choices. A result may change materially when the model changes the reference state, characterisation factor, spatial resolution or treatment of uncertainty.

These choices are not reasons to reject modelling. They are reasons to disclose the methodology and avoid presenting the output as a direct observation.

3. Pressure is not the same as state or outcome

A model may estimate the biodiversity consequence of land occupation or water use, but the organisation may not have observed the local ecological outcome. Conversely, site monitoring may reveal an outcome that a generic model misses.

A controlled metric register should distinguish:

activity data;

pressure or impact-driver metrics;

modelled impact;

state-of-nature indicators;

response metrics; and

verified outcome evidence.

Calling a reduction in modelled pressure “ecosystem recovery” overstates the evidence.

4. Time and reversibility are simplified

Some methods annualise or normalise impacts. Nature loss may be cumulative, irreversible or delayed, while restoration may have long time lags and uncertain success. A single annual score can therefore make damage and recovery appear more symmetrical than they are.

The organisation should explain the time horizon, recovery assumptions, permanence and treatment of past versus current impacts.

5. Data confidence varies across the model

A footprint can combine measured site data, supplier declarations, satellite information, life-cycle databases, sector averages and proxies. The final number may look precise even when major inputs are estimated.

Confidence should be reported by data layer and material hotspot. Useful controls include data-quality scoring, sensitivity analysis, comparison with local evidence and thresholds for specialist review.

TNFD’s emphasis on locations is a safeguard against false aggregation. For a material issue, the organisation should be able to explain:

the location or geographic proxy used;

the ecosystem or biome;

ecological condition and sensitivity;

relevant species or ecosystem services;

the activity and pressure pathway;

affected people and rights where relevant;

the model or metric applied;

the result and confidence; and

the decision or disclosure it supports.

A group-level footprint is stronger when it acts as a navigation tool into this evidence rather than a replacement for it.

Footprinting can add value when the purpose is clear and the model is fit for that purpose.

Screening and prioritisation

A financial institution or diversified group may use a model to identify sectors, commodities, geographies or counterparties for deeper assessment. The score is a risk-screening signal, not a conclusion that the highest number is always the most material issue.

Portfolio and value-chain analysis

Where direct data are limited, footprinting can provide a consistent first-pass estimate across a large portfolio or supply chain. The organisation should distinguish modelled exposure from actual impact and use engagement or traceability to improve high-priority areas.

Scenario analysis

A model can help explore how policy, ecosystem decline, commodity demand or business change might alter exposure. Scenario outputs should be treated as conditional on assumptions, not predictions.

Target design and progress tracking

A footprint may help set a portfolio or value-chain direction, but targets should retain the material locations and pathways. A reduction in the aggregate can result from portfolio mix changes rather than improvement in affected ecosystems.

Product or strategic comparison

Models can support comparison of alternatives when boundaries, functional units and data quality are consistent. They should not be used to make broad “biodiversity friendly” claims without lifecycle, location and rights review.

Specialist ecological methods are particularly important when:

an activity affects a sensitive or irreplaceable ecosystem;

threatened species or critical habitat may be present;

legal permitting or lender standards require site assessment;

restoration success must be demonstrated;

Indigenous Peoples or Local Communities may be affected;

the decision could cause irreversible impact;

offsets or biodiversity-credit claims are proposed; or

a model result conflicts with local evidence.

In these situations, remote or portfolio models can support scoping, but they should not replace field data, ecological expertise, rights assessment or engagement.

Figure 2. The decision to use a biodiversity footprint should start with purpose and materiality, then test location, data, model fit, confidence and the need for specialist assessment.

Step 1 - define the decision

State whether the model supports screening, materiality, capital allocation, engagement, target setting, disclosure or a public claim. A model appropriate for portfolio screening may be unsuitable for site approval.

Step 2 - define the material pathways

Identify the dependencies, impacts, risks or opportunities and the relevant locations, sectors, commodities or assets. Avoid selecting a model simply because it is available.

Step 3 - compare methods

Assess:

endpoint and unit;

pressures covered;

spatial resolution;

ecosystems and species represented;

baseline or reference state;

time horizon and recovery assumptions;

value-chain and portfolio coverage;

input-data requirements;

uncertainty and sensitivity capability;

transparency, licence and reproducibility; and

suitability for the intended claim.

Step 4 - test the data

Map measured, calculated, modelled and proxy inputs. Identify where generic databases substitute for actual locations or activities. Set improvement actions for material gaps.

Step 5 - validate hotspots

Compare model results with site data, satellite evidence, supplier information, incidents, stakeholder knowledge and specialist assessment. Investigate material contradictions rather than averaging them away.

Step 6 - preserve disaggregation

Retain location, ecosystem, activity, value-chain node and confidence. Prevent unrelated positive and negative outcomes from being netted through a single score.

Step 7 - control changes

Version-control the model, database, factors, boundary and assumptions. Assess whether comparatives require recalculation when the methodology changes.

Step 8 - disclose proportionately

Explain why the model was used, what it measures, what it excludes, the boundary, data quality, material assumptions, uncertainty, specialist validation and how the result affected decisions.

A biodiversity-footprint calculation should have controls comparable to other material estimates. The evidence pack should include:

model and version;

licence and permitted use;

methodology and technical documentation;

reporting boundary and functional unit;

source-data register;

geographic mapping and proxies;

factor and database versions;

calculation file and review;

uncertainty and sensitivity analysis;

specialist validation;