What AI cannot do in a biodiversity net gain assessment

No desk-based tool can produce a biodiversity net gain baseline, and understanding why sets the ai biodiversity net gain limits for any scheme caught by the gain condition. The statutory metric runs on two inputs that only exist once a competent person has walked the site: the habitat type under the UK Habitat Classification, and the condition of each parcel scored against published criteria. Aerial imagery, mapped designations and a postcode cannot supply either.

What software of any kind can do is everything around the calculation. The policy narrative, the structure of the assessment, the management and monitoring commitments, the explanation of the trading rules, the list of information still outstanding. That is useful work. It is not the baseline.

The statutory metric and who completes it

Mandatory biodiversity net gain comes from the Environment Act 2021, which inserted the biodiversity gain condition into the Town and Country Planning Act 1990. Most planning permissions in England are now granted subject to a condition requiring a biodiversity gain plan to be approved before development begins, with a minimum 10 per cent gain measured in biodiversity units and secured for at least 30 years. The duty applied to major development from 12 February 2024 and to small sites from 2 April 2024.

The gain is calculated using the statutory biodiversity metric published by Defra. The metric is a spreadsheet with rules, and the regime expects it to be completed by a competent person. That is a legal expectation rather than a professional courtesy, and a Biodiversity Net Gain Assessment submitted with a metric completed by someone without the relevant competence is the kind of document a local planning authority returns.

Competence here means ecological competence in habitat survey and condition assessment, normally evidenced through membership of a professional body and relevant field experience. The authority is entitled to ask who completed the metric and on what basis.

Habitat classification needs someone on the site

The metric's first column is habitat type, and the classification is finer grained than a map can support. Modified grassland and other neutral grassland are different habitat types with different distinctiveness scores. Lowland meadow is different again. The difference between them is the species present, which you establish by looking at the sward in the growing season.

The same problem applies across the site. A line of trees on an aerial photograph could be a hedgerow with trees, a line of trees, or scrub, and the metric treats hedgerows and watercourses in separate modules with their own units. A hard surface may be sealed or may be a sparsely vegetated land parcel colonised since the site fell out of use.

Season matters as much as presence on site. Botanical survey work is generally done between April and September, and a February visit will miss the species that decide the classification. That constraint, not the drafting, is what sets the programme on most sites, which is why the survey calendar belongs in the project plan before anything else.

A Preliminary Ecological Appraisal is usually the first commission, and it establishes both the habitat baseline and whether protected species survey work is needed. The two pieces of work use the same site visits where they are programmed together.

Condition assessment and its criteria

Each habitat parcel is scored for condition against criteria published with the metric, and the score moves the unit count substantially. Those criteria are observational. They ask about sward height and structure, the proportion of undesirable species, the presence of bare ground, signs of disturbance, hedge gappiness, canopy continuity and management evidence.

None of that can be read from imagery, and none of it can be assumed from habitat type. Two parcels of the same grassland type on the same site routinely score differently because one has been grazed and the other cut for amenity.

Condition assessment is also where professional judgement is exercised and recorded. The assessor notes which criteria pass and which fail, and that record is what an authority's ecologist checks when they disagree with a result. An assessment that reports a condition score without the criteria sheet behind it cannot be audited, and it will not survive scrutiny.

How a baseline error travels through the calculation

The metric multiplies. Distinctiveness, condition, strategic significance and, for proposed habitats, difficulty of creation and time to target condition all feed into the unit figures, so a baseline error does not stay small.

Understate the baseline and the scheme appears to deliver a gain it has not delivered. The authority's ecologist then corrects the habitat type or condition at the gain plan stage, the percentage falls below 10 per cent, and the design has to find more habitat on a site where the layout is already fixed. Overstate the baseline and you buy off-site units or statutory credits you never needed.

Either way the correction arrives after permission, because the gain plan is approved post-decision. That is the structural reason to get the baseline right at application stage even though the plan is approved later. The mechanics of how units are counted and traded are set out in habitat units as the currency of biodiversity net gain.

One further trap sits in the baseline date. Where habitat has been cleared or degraded after 30 January 2020 without permission, the metric baseline is taken as the pre-degradation condition. A site tidied up before purchase can therefore carry a baseline nobody on the current team has seen, and establishing it requires historic imagery, records and judgement rather than a site visit alone.

The gain plan and the condition on permission

The biodiversity gain plan is submitted after permission and approved by the authority before development may begin. It sets out the pre-development biodiversity value, the post-development value, the steps taken to minimise habitat loss, the on-site and off-site habitat proposed, and any statutory credits purchased.

Off-site units have to be registered on the biodiversity gain sites register and legally secured, usually by a conservation covenant or planning obligation, for at least 30 years. Monitoring and management obligations run for the same period. These are long-term commitments with cost attached, and they are assessed against the habitat the metric says the scheme needs.

Nothing in that process is a drafting exercise. The numbers come from survey, the securing comes from legal agreements, and the approval comes from the authority's own ecological advice.

Where a draft earns its place

A structured draft does real work on the parts of the submission that are narrative rather than numerical. It can set out the policy framework and the statutory basis, describe the mitigation hierarchy and how the layout responded to it, outline the 30 year management and monitoring regime, explain the trading rules and why the proposed habitat satisfies them, and mark every missing input clearly so nobody mistakes an assumption for a measurement.

PlanScope inserts "[TO BE CONFIRMED]" flags where site-specific facts are absent rather than filling the gap, which is the behaviour you want in a document that will be checked by an ecologist. The wider pattern of what drafting gets right and wrong is covered in what drafts get right about structure and wrong about evidence.

The sequence that works is ecologist first, draft second. Commission the habitat survey and condition assessment, get the metric completed by the competent person, then build the narrative document around their outputs and send it back to them for review and sign-off. You can request that review through the professional directory.

Running it the other way round produces a document that reads well and contains a baseline nobody can defend, which is the single most expensive mistake available on a biodiversity net gain submission.

Biodiversity net gain assessment requires a competent person under the Environment Act 2021 regime, and a draft is a starting point for that professional rather than a substitute for them. For specialist technical reports, qualified professional input may be required. This is general guidance, not legal advice; check the position with your local planning authority or a qualified professional before relying on it. PlanScope provides AI-assisted draft reports for planning applications. Reports should be reviewed before submission. We do not guarantee planning approval.