What Counts as Proof in Biodiversity Monitoring?

Proof in biodiversity monitoring depends on the claim. A species was detected, a species is present, a population grew, a habitat improved and an intervention caused the improvement are five different statements, and each needs a different kind of evidence. The first rule for anyone writing about monitoring results is to define the claim before judging the proof.

Detection is the first rung

The narrowest claim is that a signal linked to a species turned up in a sample. It might be a DNA sequence, a call on an acoustic recorder or a shape in an image. Under the method's assumptions and quality controls, that is good evidence the organism, or material from it, was there. It says little on its own about how many individuals there are, how healthy the habitat is or why the species was present.

Airborne eDNA shows how the first rung gets built. In early 2022 two research teams published studies in Current Biology testing whether DNA filtered from air could identify animals. One team, led from Queen Mary University of London, reported on measuring biodiversity from DNA in the air. The other, from the University of Copenhagen, tested airborne environmental DNA for terrestrial vertebrate community monitoring. Both teams ran their tests at zoos. A zoo keeps a known list of the animals it holds, which gives researchers something to check their detections against. That design answers the detection question well. It leaves the ecological questions for later work.

The later work followed. The Copenhagen group went on to report that airborne environmental DNA captures terrestrial vertebrate diversity in nature, moving the method from a controlled setting into the field. Companies that commercialise these methods inherit the evidence and the limits of each step.

Why a non-detection is weak evidence of absence

Presence is usually easier to support than absence. A species can be missed because of sampling effort, season, weather, the sensitivity of the method or the part of the site that was sampled. Ecologists treat this as imperfect detection, and a whole family of statistical methods exists to handle it. Occupancy models, for example, estimate the chance that a species is present while allowing for the chance it was missed. Researchers have used them to validate metabarcoding-based biodiversity assessments using coastal marine eDNA.

For a company, the practical point is about wording. "Not detected" is a statement about the sample. "Not present" is a statement about the site, and it needs repeat sampling and a model to back it. Commercial material that slides from one to the other gives a sceptical reviewer an easy target, and in planning or disclosure work a false absence can carry a real cost.

Abundance and change need more than a signal

A stronger signal does not always mean more organisms. The amount of DNA in a sample, or the number of calls on a recorder, depends on behaviour, distance, weather and how the material moves and decays. Some methods can support relative abundance under calibrated conditions. Others suit detection or occupancy better. If a chart implies abundance, the page should say how abundance was estimated, or the chart should show something else.

A claim that biodiversity changed needs a comparison. The monitoring programme has to have measured before, used a comparable method, controlled for season and place and kept sampling effort steady. A baseline turns a snapshot into a time series. A weak or late baseline can make every later measurement hard to read, which is why baseline design deserves as much care as the sensor.

What does it take to attribute change to an intervention?

Suppose detections rise after a restoration project. That is an observed association. Claiming the project caused the rise needs stronger evidence, such as a comparison with similar sites that were left alone, measurements before and after at both, and a plan to account for weather, surrounding land use, migration and changes in sampling. Without that design, the honest statement is that change was observed alongside the intervention. Good monitoring communication keeps what changed apart from why it changed.

Outcome claims also need a named outcome. Species richness, habitat structure, soil function, water quality and connectivity are separate things, and an improvement in one says little about the others. "Biodiversity improved" is usually too broad for a reviewer to test. A project that names the outcome it cares about, and why, gives its monitoring something specific to prove.

A claim ladder for biodiversity evidence

Every piece of biodiversity communication sits somewhere on a ladder. The higher the rung, the more the evidence has to carry.

1. Detection. A signal linked to a species was observed in a sample.
2. Presence. The organism is likely present under stated conditions.
3. Pattern. Observations differ across places or over time.
4. Abundance. The pattern supports an estimate of population size or density.
5. Outcome. A named ecological condition changed.
6. Attribution. A specific intervention caused some or all of that change.

The common failure sits between rungs three and six. A company with strong evidence at rung two writes a headline at rung five. The companion piece on evidence-based storytelling covers how to make the stronger case without climbing past the evidence.

DNAir, which captures DNA from the air, illustrates the ladder in practice. Detecting biological signals supports monitoring at the lower rungs. Change over time, attribution and ecological outcome are separate claims above them, and the single word "biodiversity" tends to fold them together. The studio repositioned DNAir around biological intelligence with a concrete use for each audience, set out in the DNAir case.

Match the evidence to the decision

A method can be scientifically sound and still poorly matched to a buyer's decision. The customer may need results at a different frequency, spatial scale, taxonomic breadth or turnaround time. Decision-grade evidence answers two questions. The method has to suit the claim, and the output has to arrive in a form and at a time the decision can use. The article on turning biodiversity data into a decision someone will pay for works through the second question.

The evidential bar should rise with the stakes. A land manager deciding where to send a survey team can act on a lighter signal than a regulator assessing compliance or an investor underwriting a nature-credit project. Independent review, peer-reviewed methods and external verification raise confidence, but each covers a specific object. Be precise about what was reviewed, whether the laboratory method, the sampling protocol, the model, the project outcome or the reporting process. A peer-reviewed method supports claims about the method and leaves claims about a particular project to be proven separately.

Uncertainty can stay visible without making the page unreadable. Put the decision-level conclusion first. Link it to the method, the limits and the validation, so that a reviewer can reach them in one step. Readers who need the detail find it, and readers who do not can still follow the argument. Disclosure is raising this bar across the market. TNFD's 2025 recommendations for upgrading the nature data value chain put data quality alongside discoverability and access, which means buyers who report will ask how each figure was produced.

How to test the strongest claim on your website

Take the boldest biodiversity statement your company makes in public and place it on the ladder. Write down what was directly observed, what was inferred, what comparison exists and what uncertainty remains. Note whether the sentence claims causation and, if it does, what design supports that. Then name the decision the claim is meant to support. If the evidence sits lower on the ladder than the headline, rewrite the headline to match the evidence and put the ambition in the roadmap.

For the wider map of where monitoring fits in the market, read what nature tech is. If your website makes claims your evidence cannot yet carry, our web service starts there.

Sources and further reading

• Measuring biodiversity from DNA in the air, Current Biology, 2022
• Airborne environmental DNA for terrestrial vertebrate community monitoring, Current Biology, 2022
• Airborne environmental DNA captures terrestrial vertebrate diversity in nature, Molecular Ecology Resources, 2024
• Validating metabarcoding-based biodiversity assessments with multi-species occupancy models, PLOS ONE, 2020