DNA Barcoding Herbal Supplements: Promise and Pitfalls

Headlines about herbal supplements containing rice, wheat and houseplants rather than the herb on the label were, for a period, a regular feature of consumer journalism. They rested on DNA barcoding: sequencing a short stretch of genetic material from a product and matching it against a reference database to identify which species it came from.

The technique is genuine and powerful. Applied to raw plant material it can distinguish species that look identical once dried and ground, and it does so without any expertise in botanical morphology. Applied to a bottle of capsules containing a standardised extract, it can produce a result that is confidently and entirely wrong, because the manufacturing process that produced the extract also destroyed the DNA the test depends on.

This is not a small technical caveat. It is the central issue in interpreting any DNA-based finding about a supplement, and it explains why some widely publicised results were later disputed by scientists who had no interest in defending the supplement industry.

How Barcoding Identifies Plant Material

The idea behind DNA barcoding is that a short, standardised region of the genome can act as a species identifier, much as a printed barcode identifies a product. The region must vary enough between species to distinguish them and vary little enough within a species that all members share it, and it must be flanked by sequences conserved enough that a single pair of primers can amplify it across a wide range of organisms.

The workflow has four steps. DNA is extracted from the sample, breaking open cells and separating nucleic acid from proteins, polysaccharides and the pigments and phenolics that plants carry in abundance. The marker region is amplified by the polymerase chain reaction, using primers that bind the conserved flanking sequences. The amplified product is sequenced. And the resulting sequence is compared against a reference database to find its closest match.

Each step has failure modes. Extraction from plants is harder than from animal tissue, because cell walls resist disruption and because the compounds that make herbs medicinally interesting inhibit the enzymes used downstream. Amplification needs template long enough to span the primer sites, which is where degraded material fails. Sequencing needs a reasonably pure product, so a mixture of species yields an unreadable overlapping trace. And database comparison can only return what has been deposited, so a species absent from the reference set matches whatever relative is present.

That last point deserves emphasis. A barcoding result is a statement about similarity to database entries, not an absolute identification. Reference databases contain misidentified entries deposited in good faith, and coverage of medicinal plants is uneven. A confident-looking match is only as good as the reference behind it.

Marker Regions Used for Plants

Animal barcoding settled quickly on a single region of the mitochondrial genome, and it works well across most animal groups. Plants have no equivalent, and the reason is biological rather than technical.

Plant mitochondrial genomes evolve slowly in their coding sequences and rearrange their structure readily, a combination that makes them poor barcodes. Attention therefore moved to the chloroplast genome and to nuclear ribosomal regions, and the outcome was not a single marker but a small set used in combination.

Two chloroplast genes form the usual core. One is a segment of the gene encoding the large subunit of the enzyme responsible for carbon fixation, which amplifies reliably across almost all plants but discriminates poorly between close relatives. The other encodes a maturase enzyme, which discriminates better but amplifies less consistently. Together they trade coverage against resolution.

Two further regions are widely added. A spacer between two chloroplast genes offers good discrimination in many groups, though it contains repeats that complicate sequencing. And the internal transcribed spacer of the nuclear ribosomal cluster, particularly its second segment, has become the most used single marker for medicinal plants specifically. It is short, present in very many copies per cell, and variable enough to separate many species that the chloroplast markers cannot.

The high copy number matters more than it might appear. A degraded sample contains few intact stretches of any given sequence, and a region present in hundreds or thousands of copies per cell has a far better chance of yielding one amplifiable fragment than a single-copy region does. Short markers also survive fragmentation better than long ones, which is why mini-barcodes of a hundred or so bases have become the practical choice for processed material.

Marker region	Genome	Main strength	Main limitation
Large subunit gene segment	Chloroplast	Amplifies across nearly all plants	Poor at separating close relatives
Maturase gene	Chloroplast	Better species discrimination	Less reliable amplification
Intergenic spacer	Chloroplast	Good resolution in many groups	Repeat regions complicate sequencing
Internal transcribed spacer	Nuclear ribosomal	High copy number, good resolution	Multiple variants within one plant
Mini-barcodes	Various	Works on fragmented DNA	Less discriminating, needs good reference data

Even with the full set, some genera remain difficult. Recently diverged species, hybrids, and groups that reproduce without sexual recombination can share barcode sequences despite differing substantially in chemistry, which matters when one is medicinal and another is not.

Why Extracts Defeat DNA Methods

This is the pitfall that turns a valid method into a misleading one, and it follows from what DNA is and what manufacturing does to it.

DNA is a long polymer held together by phosphodiester bonds, and those bonds are broken by heat, by acid, and by prolonged exposure to certain solvents. Enzymes released when plant tissue is damaged also degrade it. Every one of these conditions occurs during the production of a herbal extract.

Consider what happens to a root that becomes a capsule of standardised extract. It is dried, often with heat. It is milled. It is macerated or percolated with a solvent, commonly ethanol and water, sometimes at elevated temperature, for hours. The liquid is separated from the plant solids, which are discarded, and the liquid is concentrated by evaporation, again with heat. The concentrate may be spray-dried onto a carrier, which involves brief exposure to a hot gas stream. The powder is blended with excipients and encapsulated.

Two things have happened. The DNA that survived the process has been fragmented into pieces too short to span primer binding sites. And, more fundamentally, the solid plant material in which most of the DNA resided was filtered out and thrown away, since the point of an extract is to carry compounds that dissolve in the solvent, and DNA largely does not partition into an ethanolic extract in useful quantity.

So a DNA test on a genuine, high-quality extract of the correct species will frequently return no amplifiable plant DNA at all, or will return DNA from whatever excipient was added rather than from the herb. Rice flour, maltodextrin and cellulose are common carriers, and they carry their own intact genomes.

This makes negative and unexpected results ambiguous in a specific way. Finding rice DNA in an extract capsule may mean the product was adulterated with rice flour as a cheap filler. It may equally mean rice starch was used as a declared carrier and its DNA survived while the herb’s did not. Distinguishing these requires knowing the formulation, and a laboratory testing a retail product blind does not.

The gradient of testability runs roughly from whole dried herb, where DNA is abundant and intact, through cut and sifted material and simple milled powders where it is usually recoverable, to concentrated and standardised extracts where it is often absent, to purified single-compound preparations where the question is meaningless.

Filler and Substitute Species Findings

None of the above means substitution is imaginary. Botanical adulteration is a genuine and long-standing problem with several distinct forms, and DNA methods have contributed real evidence about it.

Substitution occurs when a different species is supplied in place of the labelled one. Sometimes it is deliberate fraud, exploiting the fact that dried powdered material of one species looks identical to another. Sometimes it reflects genuine confusion, since common names are shared between unrelated plants and regional pharmacopoeias assign the same name to different species. Sometimes it happens at harvest, when collectors of wild material take a look-alike growing alongside the target.

Dilution with inert filler is a separate practice, in which cheap bulk material extends an expensive herb. Contamination is different again, involving unintended material from shared equipment, storage or the field.

The consequences vary from none to serious. A capsule diluted with rice powder underdelivers but rarely harms. A substituted species may be pharmacologically inert, or may carry compounds with real toxicity, and there have been well-documented episodes in which a toxic species substituted for an intended one caused significant injury to numbers of people. This is why identity testing is not a purely commercial concern.

DNA methods are particularly good at detecting these problems in raw material, which is where they belong in a quality system. A supplier delivering dried root can be checked before it enters production, when the material is at its most testable and when rejecting a batch is still straightforward. Testing the finished capsule is testing at the point where the method works worst and where remediation is most expensive.

Criticism of Early Barcoding Studies

Several high-profile investigations reported that a large fraction of tested herbal products did not contain the labelled species, and the results were widely reported. Scientists working in botanical authentication raised substantial objections, and the objections were technical rather than defensive.

The central criticism was the one described above: applying DNA methods to extracts and interpreting the absence of a detectable barcode as absence of the herb. A product can contain the correct botanical, at the correct concentration, with the correct chemistry, and yield no amplifiable DNA. Reporting that as failure to detect the species is defensible; reporting it as evidence the product does not contain the species is not.

Related objections followed. Some studies used a single marker region rather than a set, which reduces the chance of amplification and the reliability of identification. Some drew conclusions from database matches without confirming them by other means. Some did not report positive and negative controls in a way that let readers judge whether extraction and amplification had worked at all, which is essential when a negative result is the finding. And some did not distinguish between DNA from a declared excipient and DNA from an undeclared adulterant.

The methodological response has been constructive. Consensus has grown around several principles: use multiple markers, use mini-barcodes for processed material, run controls that demonstrate the method could have detected the target had it been present, characterise the product form before choosing a method, and treat DNA evidence as one input alongside chemical analysis rather than as a verdict.

The reasonable conclusion is neither that the early studies were worthless nor that their headline figures should be repeated. They drew attention to a real problem and they overstated it in a specific and identifiable way. Later work using appropriate methods has continued to find substitution and adulteration, generally at lower rates than the most alarming reports, and concentrated in particular product categories and supply chains rather than spread evenly across the market.

Chemical Fingerprinting as a Complement

DNA and chemistry answer different questions, and the difference is the key to using either sensibly. DNA tells you what organism the material came from. Chemistry tells you what compounds are present now. A product can pass one and fail the other in both directions.

The chemical toolkit has several layers. Thin layer chromatography separates an extract on a coated plate and produces a pattern of spots compared against a reference; it is cheap, quick, surprisingly discriminating, and still a workhorse in pharmacopoeial monographs. High performance liquid chromatography resolves compounds far better and quantifies markers against standards, which is how a standardised extract’s claimed content is verified. Mass spectrometry coupled to chromatography identifies compounds by mass and fragmentation, catching species-specific markers and undeclared additions. Infrared spectroscopy compares whole spectra against reference libraries fast enough for use at goods-in.

Each has limits. Marker compounds are frequently shared between related species, so chemistry alone may not distinguish a genuine article from a close relative. Chemical profiles vary with growing conditions, harvest time, plant part and processing, so a legitimate batch can look unusual. And measuring one marker compound says nothing about the rest of the extract, which is why adding a purified compound to a weak extract is a known form of adulteration that a single-marker assay will not catch.

Microscopy deserves mention alongside these. For powdered raw material, examining cell structures, starch grains, trichomes and crystals under a microscope remains a valid pharmacopoeial identity method, cheap and requiring no instrumentation beyond the microscope, though it requires an experienced analyst.

The sensible arrangement uses each where it is strongest: DNA or microscopy on incoming raw material to confirm botanical identity, chemistry through processing and on the finished product to confirm composition and concentration, and heavy metal, pesticide and microbiological testing alongside both.

Choosing the Right Test for the Product Form

The practical question for anyone reading a test report, or deciding what testing to ask for, is whether the method suits the material.

For whole or cut dried herb, DNA barcoding is well suited and so is microscopy. Material is unprocessed, DNA is intact, and identity is the main question. For a milled powder of the whole plant part, DNA usually still works, though the extraction step becomes harder as surface area and exposure increase. For a simple encapsulated powder, the same applies, with the caveat that excipients contribute their own DNA.

For a concentrated or standardised extract, DNA is the wrong primary tool. Chemistry is the appropriate method, verifying both the fingerprint of the whole extract and the concentration of marker compounds. Where botanical identity of the source material matters, the place to establish it is upstream, at the raw material stage, and a manufacturer with a proper quality system will have done so and will be able to say so.

For tinctures and liquid preparations, chemistry again. For a purified isolated compound, identity of the compound is the only meaningful question and both botanical methods are irrelevant.

For a consumer, the actionable version is short. A certificate of analysis is more informative than a testing logo, and the questions worth asking of one are which tests were run, on which batch, and by whom. Independent third-party verification programmes exist and generally test finished products against their labels using chemical methods. And a product whose marketing leans on DNA verification of an extract is making a claim that deserves a question rather than confidence.

The useful mental model is that a supplement has a history, and different tests read different parts of it. DNA reads the beginning, telling you what was harvested. Chemistry reads the end, telling you what is in the capsule now. Manufacturing sits between them, and it is precisely the step that erases the first record while creating the second. Any testing claim that ignores which part of the history it is reading should be treated with the same caution as a label claim it was meant to verify.