Soil eDNA and Metabarcoding for Biodiversity Monitoring

Sequence the DNA organisms shed into a soil sample to inventory the biological community, so a biodiversity claim rests on a detection record with a stated baseline and stated limits.

Also known as: environmental DNA, soil eDNA, metabarcoding, DNA barcoding for soil biodiversity.

Every organism in a handful of soil leaves traces of itself behind: shed cells, mucus, root exudate, fragments of the dead. That shed material carries DNA, and DNA carries names. Pull a soil core, extract the DNA it holds, sequence it, and match the fragments against a library of known organisms, and you can list much of what lives in that soil without ever seeing a single bug, worm, or fungal thread. That list is what a biodiversity claim has lacked: a count you can repeat, baseline, and audit instead of assert. The catch is that the list is only as honest as the marker you amplified, the library you matched against, and the baseline you measured it from.

Definition

Soil environmental DNA (eDNA) monitoring extracts the DNA that organisms shed into a soil sample and sequences it to inventory the biological community without observing the organisms directly. The usual method is metabarcoding: amplify a short, taxonomically informative marker region, sequence the amplicons, and match the reads against a reference library to assign each one a name. A single soil core becomes a multitrophic species list spanning bacteria, fungi, protists, nematodes and other microfauna, arthropods, and earthworms.

The phrase carries a precise operating meaning that the marketing version obscures. The interesting question isn’t “sequence the soil and get a biodiversity score.” It is: which marker and which reference library, against which baseline and which control sites, sampled how and how often, at what taxonomic resolution and what detection limit, so the resulting index supports the specific claim a buyer, lender, or credit registry will underwrite. The method is mature enough that the cost and turnaround are no longer the binding constraint; the discipline around baseline and controls is.

The term sits inside a settled vocabulary. eDNA is the shed genetic material in an environmental sample. Metabarcoding is the multi-species sequencing-and-matching workflow applied to that sample. A marker or barcode is the short region amplified (16S for bacteria, ITS for fungi, COI or 18S for animals). A reference library is the curated set of named sequences the reads are matched against; the European Commission’s Joint Research Centre runs a dedicated soil DNA-barcode program precisely because library coverage is the gate on how many reads can be named at all.

Why It Matters

Biodiversity is the outcome this field most often asserts and least often measures on the page. Cover cropping, hedgerows, silvopasture, reduced tillage, and managed grazing all claim it. Most of the time the claim rests on practice adoption or a surface count, not on the soil community where the bulk of agroecosystem biodiversity actually lives.

Soil eDNA closes part of that gap. A 2024 agroecosystem study (Xing, Lu et al., Environmental Research) sampled four systems, detected 1,146 species, identified 48 organismal indicators sensitive to management, and built an eDNA-based index that tracked conventional soil-quality indices at R² = 0.58. That is not a perfect correlation, and the authors don’t claim it is. It is enough to say the soil community can be read as a multitrophic signal of ecological health rather than inferred from the practices applied above it.

For the financier and the standards reader, the method forces a harder framing than a score. A biodiversity credit or nature-market claim is only as good as its baseline and its control sites. The npj Biodiversity 2024 standards work on biodiversity-credit baselining and the forest-carbon-market eDNA study (Communications Earth & Environment, 2024) both argue that shed-DNA monitoring is what could make biodiversity co-benefit claims standardized and auditable rather than asserted. The same logic applies on a farm: eDNA gives the financier a diligence question (what is the baseline, what species array, what control sites, what sampling cadence), the operator a way to evidence a biodiversity claim without a resident taxonomist, and the standards reader the additionality-and-baseline frame that separates a defensible credit from a marketing one.

It matters as a measurement, not as an advocacy point. The honest version states the limits in the same breath as the capability: a species list is a detection record, not a population count or a function measurement, and the reads can only be named as far as the reference library reaches.

How It Shows Up

A commercial monitoring platform. NatureMetrics built a business on shed-DNA biodiversity monitoring across soil, water, and air, reporting more than 600 corporate customers across 110 countries and a USD 25M Series B in January 2025. The commercial signal matters for the reader making a diligence call: the sampling kits, lab turnaround, and reporting are productized rather than research-bench one-offs, which is what moves eDNA from a study method to an operating one. The diligence questions don’t change because a vendor is involved; they get sharper. Which marker, which library, which baseline, which controls.

A soil-microbiome service for agronomy. Trace Genomics’ BeCrop platform sequences the soil microbiome and reports functional and community readouts to growers and their advisors. This is the agronomic, rather than the credit-market, use: the operator is reading the soil community to inform management, not (yet) to underwrite a tradable claim. It shows the same data serving two different jobs, and the claim discipline differs by job. A management readout can tolerate looser baselines than a credit can.

A standardized credit-monitoring method. The 2024 npj Biodiversity standards paper and the Communications Earth & Environment forest-carbon eDNA work both treat metabarcoding as the candidate measurement layer under a baseline-monitoring-validation structure for nature markets. The pattern they describe is the one this concept is built to support: baseline the site, define control sites, sample on a fixed cadence with a named marker and library, and report a detection record against the baseline rather than a bare score. A 2025 two-step soil-microbiome metabarcoding method (Scientific Reports) is one example of the workflow being refined toward that standard.

Caveats and Open Questions

The limits are specific, and they bound what the claim can say:

• Reference-library gaps cap taxonomic resolution. Reads can only be named as far as the library reaches. Bacteria and fungi are comparatively well covered; soil fauna and many invertebrates are not. A read with no library match is a detection without a name, and a thin library systematically undercounts exactly the groups a farm biodiversity claim most wants to show. The EU JRC soil DNA-barcode program exists to widen that gate.
• Read abundance is not a clean population count. Amplification bias, variation in DNA shed per organism, and extraction differences mean read counts track presence far better than they track biomass or numbers. Treat the output as a detection record — what was present in the sample — not as a census.
• A species list is not a function measurement. eDNA tells you who is there, not what they are doing or how much carbon they cycle. The reads correlate with soil function and with soil organic carbon; they do not measure either directly. A biodiversity detection and a carbon stock are different claims, and a sound MRV design keeps them separate even when one sample feeds both.
• The claim is only as good as the baseline and controls. Without a baseline sample and unmanaged or differently managed control sites, a species list is a snapshot, not evidence of change attributable to management. This is the additionality problem in biological clothing: a rich community may have predated the practice. The control-site discipline is what turns a sequencing receipt into a defensible claim.
• Sampling design drives the result. Where, how deep, how many cores, what season, and how often all move the species list. Soil is spatially patchy at the scale of a single core, so a monitoring file has to fix its sampling protocol before its numbers mean anything across years.
• Geographic and crop-system narrowness. Much of the validating literature comes from temperate systems and a small set of agroecosystems. The index strength and indicator set observed in one region and cropping system should not be assumed to transfer unchanged to another.

The open question the field has not closed is governance: which marker, which library version, which baseline rule, and which validation standard a registry will accept, so that two eDNA reports of the same land produce comparable claims. Until that converges, an eDNA result is a strong detection record and a weak commodity — credible to a diligence reader, not yet fungible across registries.

Related Articles

Sources

• Xing, Lu, and colleagues’ 2024 Environmental Research study reports soil eDNA biomonitoring across four agroecosystems: 1,146 species detected, 48 management-sensitive indicators, and an eDNA index tracking soil-quality indices at R² = 0.58.
• The European Commission Joint Research Centre’s DNA barcodes for soil biodiversity program documents the reference-library and barcoding work that gates how many soil reads can be named.
• The 2024 Communications Earth & Environment study examines eDNA monitoring as a measurement layer for biodiversity co-benefits in forest carbon markets.
• The 2024 npj Biodiversity paper sets out baseline, monitoring, and validation standards for biodiversity credits, the integrity frame an eDNA result has to sit inside.
• A 2025 Scientific Reports method paper describes a two-step soil-microbiome metabarcoding workflow, an example of the protocol being refined toward standardized monitoring.
• NatureMetrics is a commercial shed-DNA biodiversity-monitoring platform, cited here for market-scale and operating signal rather than as authority on first principles.