Biodiversity credits

A carbon credit works as a tradable unit because of a physical accident: carbon dioxide mixes globally within a couple of years, so a tonne avoided in Kalimantan and a tonne avoided in Ohio have the same effect on radiative forcing. The unit is defined by the molecule, and location is irrelevant to the claim. Nothing equivalent exists for biodiversity, and the difference is not a market-design problem waiting for better rules. It is the subject matter.

The unit has to be built

There is no natural scalar for biological diversity. Species richness, abundance, genetic diversity, functional diversity and habitat condition move independently and sometimes in opposite directions; a restored grassland can gain species while losing the specialists that made it valuable. So any credit scheme must construct a metric, and in practice they converge on the same shape: extent multiplied by condition, scored against a reference state.

England’s statutory biodiversity net gain, mandatory for most development from February 2024 under the Environment Act 2021, is the clearest worked example. Its metric multiplies habitat area by a distinctiveness score, a condition score and a strategic-significance factor to give habitat units, and requires a 10 per cent uplift secured for 30 years. The instructive part is what surrounds the arithmetic: rules against trading down from higher to lower distinctiveness, spatial multipliers that discount off-site compensation, and time discounting for the years a created habitat spends below target condition. Every one of those adjustments exists because the units are not actually interchangeable. A fungible commodity does not need rules forbidding certain exchanges.

Comparison between sites is the hard part

Condition scoring is site- and habitat-specific by construction: the indicators for chalk grassland are not the indicators for a mangrove. Two credits can therefore be arithmetically equal and ecologically incommensurable. This is why biodiversity credits tend to stay within a jurisdiction and a habitat type, and why proposals for a global fungible nature unit should be read sceptically — the fungibility would have to be asserted, not derived.

The reference state adds a second judgement. Uplift is measured against what the site would otherwise have been, so the scheme inherits the counterfactual problem of carbon baselines. It adds one of its own: which reference counts as intact, given that most landscapes have been modified for centuries.

What the monitoring technologies actually deliver

Environmental DNA metabarcoding and passive acoustic monitoring have made surveying far cheaper, and they are genuinely informative — but about presence, not amount. A metabarcoding result reports which taxa left detectable DNA in a sample; primer affinity varies between taxa, so read counts are not abundance, and DNA persists after the organism has gone or arrived by water flow from elsewhere. Acoustic indices only see taxa that vocalise, and detection depends on vegetation, weather and distance.

Both are excellent at showing that something is present, weak at showing how much, and largely blind to the plants, fungi and soil structure that underpin habitat condition. Since condition is half the metric, the cheap remote methods verify the half of the credit that is easier to measure, and expert field survey still carries the rest.