Part of the food fraud pillar of the TAAG Learning Center. If PCR is new to you, start with what PCR is and how it works.
Table of contents
There is no universal authenticity test
In our introduction to food fraud we saw that the problem can take many forms: one species swapped for another, a diluted ingredient, a false origin, a substance added to fake quality. Each of these leaves a different fingerprint in the food, which is why no single analysis answers the question "is this product authentic?".
The practical way to choose a method is to start from the specific question you need to answer:
- What species is it? Meat, fish, shellfish: the answer is usually in the DNA or the proteins.
- Has something cheaper been added? Honey with syrups, blended oils: this is where stable isotopes, chromatography and mass spectrometry come in.
- Where does it come from? Geographic origin: isotopes and trace element profiles.
- Does it resemble the authentic product? Rapid screening of many samples: spectroscopy and nuclear magnetic resonance (NMR) with statistical models.
A review published in Trends in Analytical Chemistry (Danezis and colleagues, 2016) groups authentication techniques into these same families and stresses that they are increasingly combined rather than used in isolation.
DNA-based methods: the tool for identifying species
DNA is the natural answer when the question is the species, because its sequence does not change with the commercial presentation: a fillet, a burger or a canned product still contains the genetic material of the animal it came from, though more or less fragmented depending on the process. Before any analysis, the nucleic acids must be extracted from the matrix, a step that in cooked, fatty or highly processed products conditions everything that follows.
Species-specific real-time PCR
This is a targeted analysis: primers and probes are designed to amplify only the DNA of one particular species, and the result answers a question such as "is there horse in this meat declared as beef?". It was the central tool after the 2013 horsemeat scandal: European Commission Recommendation 2013/99/EU called for controls using a real-time PCR method able to detect horsemeat down to a level of 1% by weight, compared against a standardized control sample from the EU reference laboratory. That 1% is still used as a pragmatic limit in official controls of meat products. Several species can be searched for in parallel using formats such as multiplex PCR.
DNA barcoding: when you don't know what to look for
If the question is open-ended ("what species is this fish?"), a standardized fragment of the genome is sequenced and compared against a reference library. In animals, the most widely used fragment is a region of about 655 base pairs of the mitochondrial gene cytochrome c oxidase I (COI). The FDA maintains its Reference Standard Sequence Library (RSSL) for this purpose, with sequences from more than 1,040 seafood specimens linked to their scientific name and their accepted market name.
Full-length barcoding works well on fresh tissue but fails when the DNA is degraded by cooking or sterilization. Mini-barcodes were developed for those cases: shorter fragments of the same region. In a study published in Scientific Reports (Shokralla and colleagues, 2015), six primer pairs targeting fragments of 127 to 314 base pairs identified 41 of 44 processed fish products to species or genus level (93.2%). The best single pair reached an 88.6% success rate, compared with 20.5% for the full barcode primers.
The limit of DNA methods: quantification
Detecting a species is not the same as saying how much of it there is. The number of DNA copies per gram of tissue varies between species and tissue types, so converting "DNA copies" into "percentage by weight" is not straightforward. A 2024 study comparing real-time PCR and digital PCR in homogenized foods found that bird species had on average 3.31 times more genome copies than mammals, with variability so wide that this ratio could not be used as a simple correction factor. Digital PCR showed fewer differences between matrices than real-time PCR, but the underlying problem remains. That is why many control schemes work with thresholds, such as the 1% for horsemeat, rather than exact figures.
Massively parallel sequencing expands these possibilities, because it can identify several species at once in a mixture without knowing in advance which ones to look for.
Protein-based methods
Before PCR became widespread, fish species identification relied on protein profiles, such as isoelectric focusing (IEF). It is still used: in the EU coordinated plan on white fish substitution in 2015, national laboratories applied IEF alongside PCR-RFLP, DNA barcoding and real-time PCR. Proteins are denatured by heat, so their usefulness drops in cooked products, where DNA is usually more informative.
Stable isotopes: the chemical fingerprint of origin
Elements such as carbon, hydrogen, oxygen or nitrogen exist in isotopic forms that are distributed differently depending on the plant that fixed the carbon, the water of the region or the animal's diet. Isotope ratio mass spectrometry (IRMS) measures those proportions with great precision.
The honey case, and why a classic method stopped being enough
The ratio of carbon-13 to carbon-12 separates C4-metabolism plants, such as maize and sugarcane, from most of the plants from which bees collect nectar. For years, AOAC method 991.41, based on this measurement, made it possible to detect honey extended with corn or cane syrups.
The EU coordinated action "From the Hives" (2021-2022) showed the limit of that approach: of 320 samples of imported honey, 147 (46%) were suspected of not complying with the EU honey directive, compared with 14% in the earlier 2015-2017 plan. According to the Commission, AOAC method 991.41 was not effective at detecting them, which indicates that corn or cane syrups were replaced, wholly or partly, by rice, wheat or beet syrups. These are C3 plants, with a carbon isotope signature similar to that of nectar. The Joint Research Centre (JRC) had to combine several methods to identify syrup markers, and even then the result was qualitative: presence or absence of markers, not a percentage of adulteration.
The lesson applies beyond honey: once a method becomes standard, whoever commits fraud learns to evade it.
Spectroscopy and NMR: rapid fingerprint screening
Near-infrared (NIR) spectroscopy, Fourier-transform infrared (FTIR), Raman and NMR do not look for one particular compound: they generate a full spectrum that works as a fingerprint of the product. That fingerprint is compared, using statistical models (chemometrics), with a database of authentic samples.
- Advantages: they are fast, need little sample preparation, and several can be taken to the plant or the receiving point.
- Limitations: they depend entirely on the quality and representativeness of the reference database, and their result is a statistical deviation ("this sample does not look like the authentic ones"), not the identification of what was added.
That is why they work best as a first barrier: they allow many lots to be checked and only the suspicious ones sent to a confirmatory method.
Chromatography and mass spectrometry
Liquid or gas chromatography coupled to mass spectrometry identifies and quantifies specific compounds. It is used in two ways:
Targeted analysis
It looks for a known substance, such as melamine in dairy products or Sudan dyes in palm oil and spices. It is the way to confirm adulteration with an unauthorized substance and to generate a result with regulatory value.
Non-targeted analysis
With high-resolution mass spectrometry, everything the sample contains is recorded and differences against authentic samples are sought. It is useful for discovering new adulterants but, like spectroscopy, it depends on good references.
How to choose: validation and combination
- Define the question before the method. Species, origin, dilution or added substance lead to different techniques.
- Ask for the method validated in your matrix. An analysis that works on a fresh fillet may fail on a canned product, just as happens with the validation of microbiological methods.
- Combine screening and confirmation. A fast, broad method to prioritize, a specific one to confirm.
- Interpret with thresholds. Traces of another species may come from cross-contamination on the line, not from deliberate substitution.
Knowing which method to use is half the work. The other half is deciding which raw materials to test and how often, which falls under the fraud vulnerability assessment.
Conclusion
A food's authenticity is verified with different tools depending on the question: DNA for species, isotopes for origin and certain added sugars, spectroscopy for rapid screening and mass spectrometry to confirm specific substances. No method is definitive on its own, and the honey case shows that adulterators adapt to the best-known analyses. That is why it pays to combine screening and confirmation, to require methods validated in the real matrix and to interpret results using thresholds.
About TAAG
Discover how TAAG's experience in nucleic acid extraction and real-time PCR applied to complex food matrices fits into industry quality control programs.
Frequently asked questions
Which method tells me whether a fish is the declared species?
DNA methods. If a specific species is being sought, species-specific real-time PCR; if the question is open, DNA barcoding of the COI gene compared against a reference library, such as the FDA's RSSL. In highly processed products, mini-barcodes are used: shorter fragments that better withstand DNA degradation.
Can PCR tell what percentage of another species a meat product contains?
Only approximately. The number of DNA copies per gram varies between species and tissues, which makes it hard to convert the result into a percentage by weight. That is why official controls usually work with thresholds, such as the 1% by weight used in the EU for horsemeat since 2013.
Why is isotope analysis no longer enough to detect adulterated honey?
Because the classic AOAC 991.41 method detects sugars from C4 plants such as maize or cane. In the 2021-2022 EU coordinated action it was not effective, which indicates a shift toward rice, wheat or beet syrups, whose carbon isotope signature is similar to that of nectar.
Can an NIR spectrum prove fraud?
On its own, rarely. It indicates that a sample deviates statistically from the authentic ones in the reference database, but it does not identify what was added. It is a screening tool that should be complemented with a confirmatory method.