In this paper (open access) the authors trained a Machine Learning model to differentiate between Top, Bottom and Spontaneous fermented bottled beers.  Data were collected using a non-invasive hand held NIR scanner pointed directly through the unopened bottle using a customised foam attachment.  The model was trained on 25 samples of major brands purchased online, rather than reference samples of verified traceability, but the training samples covered a wide range of beer types from stouts to light ales, and a wide range of bottle types and colours.

The authors report good classification based on fermentation method.  They consider that evidence of a wrong fermentation method could be one quick and easy check that could flag counterfeits.  They also correlated the NIR data with sensory panel assessments and SPME-GC-MS data and concluded that non-invasive NIR has the potential to classify beers based on their aroma profiles.

Image from the paper

There has been a lot posted recently about honey authenticity and test methods.  This blog from the FSA pulls it all together in one concise and systematic page.  It includes

• Honey sampling guidelines
• The weight of evidence approach to interpreting test results
• The UK AMWG review of the EU “From the Hives” report
• New testing methods developed under FSA-funded research

The results of the EC 2021-2022 honey sampling and analysis co-ordinated action, following the  From the Hives report, were concerning.  This 2023 report concluded that all 10 of the sampled honeys imported from the UK were “suspicious”. 

This finding prompted further investigation by the UK Department of Environment, Food and Rural Affairs (Defra).

Defra have now published an independent expert review into the analytical methods used in the survey.  There is a lot of technical content in the review.  It re-emphasises that no single honey authenticity test is likely to be definitive, and that a weight of evidence approach should be used with some tests being weighted higher than others.  When the total weight of evidence is not strong then phrasing such as “warrants further investigation” would be a fairer conclusion than “suspicious”.

One specific learning from the review is that laboratories must take care with the selection of authenticity markers, depending on the analytical question being asked.  The example given is oligosaccharides.  Some of these markers are known to vary between honey that has had moisture mechanically removed compared to honey that has not.  Moisture removal may be a production necessity (in humid climates where honey will not evaporate naturally) or a commercial choice to speed the harvest cycle (as is commonly used in China).  Moisture-removed honey is common within UK blends of Chinese origin honeys  but is not permitted in some EU countries.  Thus a test based on oligosaccharide markers could differentiate UK honey from EU for reasons that are already understood.  It might not provide any new insight, for example, on sugar or syrup adulteration.

Photo by Art Rachen on Unsplash

This peer-reviewed pre-print (open access) reports a classification model for different Greek olive oil cultivars using combined data from two analytical techniques: volatile component analysis (6 marker compounds) by solid phase microextraction – gas chromatography (SPME-GC-MS) and spectral analysis by attenuated total reflectance-Fourier transform infrared spectroscopy (ATR-FTIR)

The model was built to differentiate Greek oils from 3 cultivars: Koroneiki, Megaritiki and Amfissis.  The reference database was constructed from samples collected over 3 harvest seasons.  The authors report that application of the supervised methods of linear and quadratic discriminant cross-validation analysis, based on volatile component data, provided a correct classification score of 97.4 and 100.0%, respectively. The corresponding statistical analyses were used in the mid-infrared spectra where the 96.1% of samples were discriminated correctly.

The authors conclude that ATR-FTIR and SPME-GC-MS, in conjunction with the appropriate feature selection algorithm and classification methods, are powerful tools for the authentication of Greek olive oil. They consider that the proposed methodology could be used in industrial settings for the determination of Greek olive oil botanical origin.

Shrimp surimi-based products (SSPs) are composed of minced shrimp meat and are highly susceptible to fraudulent substitution by cheaper fish surimi.

This study (open access) employed a double-gene metabarcoding approach to authenticate SSPs sold in bulk (business-to-business) on Chinese e-commerce platforms. 16S rRNA and 12S rRNA genes were amplified and sequenced from 24 SSPs. Mislabeling was evaluated based on the correspondence between the ingredients (only those of animal origin) reported on the products’ labels and the molecular results.

The authors found that 21 of the 24 products were mislabeled. The replacement of Penaeus vannamei with other shrimp species was particularly noteworthy. In some samples the primary species detected in terms of sequence abundance were not shrimp but fish, pork, chicken, and cephalopods. The 12S rRNA sequencing results revealed that fish species like Gadus chalcogrammus, Evynnis tumifrons, and Priacanthus arenatus were added to some SSPs in significant proportions, with certain products relying on fish priced from “Low” to “High” levels to substitute higher-cost shrimp. Notably, many fish species in SSPs were highly vulnerable to fishing, raising sustainability concerns.

The authors conclude that the high mislabeling rate, as well as the detection of endangered fish species (Pangasianodon hypophthalmus), underscores significant quality control and supply chain integrity issues.

Photo by Fernando Andrade on Unsplash

The Joint Research Centre of the European Commission have published their monthly collation of food fraud reports, combined for November and December 2024, here.  Thanks again to FAN member Bruno Sechet who has turned these into an infographic.  The original infographic (in much better resolution!), along with his commentary, is on Bruno's LinkedIn feed.

The source for the JRC collation are global media reports, and these always gives a slightly different picture than collating official reports.  Best practice in horizonscanning is to look at multiple collations of fraud incidents/suspicions along with their commentaries, both official and media, and make an intelligent assessment of their complementary scopes and limitations when drawing insight.  FAN's annual aggregated report gives a high-level overview of food fraud incidents in 2023 as recorded in official reports. The 2024 version is in preparation and will be published on our website soon.   

The  aim of this proof-of-principle study (open access) was to design a universal DNA microarray (“DNA Chip”) to distinguish all edible fish species by comparing hybridization signal patterns from samples with patterns obtained from reference specimens.

The researchers designed a universal set of 96 DNA probes that cover all fish species of food interest.  These were narrowed down by virtual modelling experiments from a long-list of 28,000 candidates which they had generated experimentally. They also included 4 control probes (sequences not present in edible fish).  All probes were based on sequences from either 16S ribosomal RNA or cytochrome b.

DNA was isolated with either a CTAB method or with commercial DNA extraction kits. The gene markers cytb (approx. 464 bp) and 16S rDNA (approx. 600 bp) and an additional pUC57 vector DNA region (542 bp) were amplified in triplex PCRs. The DNA probes were spotted contactless using piezoelectric dispensing technology as 19 × 19 arrays. For hybridization of the generated PCR amplicons on the prepared microarrays the INTER-ARRAY Hybridization Kit was used according to manufacturer's specifications. The arrays were measured directly after staining and then processed using the INTER-VISION GENOTYPING 1.2.0 software.

The authors tested 86 fish fillets sourced from verified suppliers and were able to correctly identify all species by hierarchical clustering analysis of the results.  The entire process takes a few hours.  They conclude that the method is ready for further validation and ruggedness testing. More replicates and species should be analyzed to confirm current results. Likewise, the robustness of the DNA array should be determined, e. g. by using different thermocycler or users and laboratories.

Graphical abstract from the paper

In this paper (open access) two optical spectroscopic techniques,  Laser-Induced Breakdown Spectroscopy (LIBS) and UV-Vis-NIR absorption spectroscopy, are assessed for EVOO adulteration detection, using the same reference database of olive oil samples. In total, 184 samples were studied, including 40 EVOOs and 144 binary mixtures with pomace, soybean, corn, and sunflower oils, at various concentrations (ranging from 10 to 90% w/w). The reference class of “pure” EVOOs were limited to oils from a specific geographic region (either Crete, Lesvos, Kalamata or Achaia, with a different model built for each case).

The emission data from LIBS, related to the elemental composition of the samples, and the UV-Vis-NIR absorption spectra, related to the organic ingredients content, were analyzed, both separately and combined (i.e., fused), by Linear Discriminant Analysis (LDA), Support Vector Machines (SVMs), and Logistic Regression (LR). In all cases, very highly predictive accuracies were achieved, attaining, in some cases, 100%.

The authors conclude that both techniques have the potential for efficient and accurate olive oil verification test protocols, with the LIBS technique being better suited as it can operate much faster.

This paper (purchase required) reports the use of a portable optical sensor (Multi-Spectral Imaging) to build a classification model for detecting milk adulteration. This encompassed mixtures of milk from different species (cow, goat, and sheep), as well as dilution of cow’s milk with water. The study's scope also included milk with diverse heat treatments, fat content, and commercial brands.

The authors report that discriminant analyses provided reliable predictive models, with Accuracy and Cohen's Kappa values ranging between 0.80 and 1. In quantitative studies, the quantification of milk mixtures at a minimum percentage interval of 10% was detected with Mean Absolute Error (MAE) values between 0.14 and 0.05, and 0.03 for cow's milk adulterated with water at adulteration levels of 5%.

The authors conclude that the portability of these instruments adds a significant advantage by enabling on-site and real-time determination and quantification of milk adulteration.

Cocoa is high on many companies’ current risk radar for authenticity threats, due to recent supply pressures and price increases. Carob has legitimate uses as a cocoa replacement, and carob flour has been cited as a potential cocoa adulterant.

This paper (purchase required) reports the use of direct analysis in real time mass spectrometry (DART-MS) as a rapid laboratory-based authentication test with the potential for a portable device. Reference samples of cocoa powders, carob flours, and mixtures of the two were extracted with buffer and interrogated by DART-MS. The mass spectra were used to develop classification models. A principal component-linear discriminant analysis (PCA-LDA) model was used to discriminate between cocoa powder and cocoa powder amended with 15% carob flour. The accuracy using internal validation was 100%. Using an external validation dataset, the accuracy, precision, and recall were 96.0%, 94.8%, and 97.3%, respectively.

For a descriptor of DART-MS see FAN’s analytical method explainers.

An electronic nose (“e-nose”) is a sensor used to selectively measure volatile organic compounds.  Although e-noses have advantages in terms of cost and ease of use, they also have inherent limitations in terms of sensitivity to detect subtle variations in compound concentrations, leading to inconsistent results if not properly managed. The data generated by e noses generally require advanced processing techniques for interpretation of complex signal patterns. This is why e-nose food classification applications tend to use Deep Learning techniques such as Recurrent Neural Networks.

In this publication (open access) the authors used an array of 7 sensors to build a model to differentiate pork, bovine and fish gelatin.  The model was based on a commercial sample of each, dissolved in water as a 1% solution and warmed.  The model was then applied to different in-house mixtures of the gelatins at different time-points after preparation.  The authors do no report if it was validated with orthogonal samples of verified origin.  The sensors had selective sensitivity to a range of volatiles including ethanol, methane, propane, butane, ammonia and hydrogen sulfide.

The authors report that classification efficiency, as measured by the AUC (Area Under the ROC Curve), was variable when considering one sensor in isolation but was good when all 7 sensors were multiplexed.  The AUC increased with time from sample preparation, rising to over 98% at 2-hours from the samples being prepared.  The authors conclude that this makes the technique a promising candidate for constructing a routine instrument to check the species of commercial gelatin.

FSA-funded project: Review of current and emerging analytical methods for the testing of oil for authenticity (Project FS900520)

With funding from the UK Food Standards Agency, Fera Science Limited (Fera) in York, UK is currently undertaking a project to review the current and emerging analytical methods for testing edible oils and support the further development of analytical methods which will underpin and uphold the authenticity of edible oils in the supply chain. 

As part of the project’s evidence gathering, Fera would like to invite parties involved in sourcing, processing, and/or testing edible oils to participate in an online questionnaire. 

The fundamental mission of the FSA is food you can trust. The FSA strategy sets out FSA’s vision to ensure that the UK food system is safe, and that food is what it says it is. This involves building scientific capability in Public Analyst (PA) Official Laboratories (OLs) and working with Defra’s food authenticity programme to conduct research and development for analytical methods. Suitable analytical methods are required to ensure that food is what it says it is and to manage risk around food authenticity.

 As key stakeholders, your insight will help to inform FSA regarding issues in oil authenticity and future-proofed analytical tools to support both industry and regulators, while maintaining consumer confidence in our food. 

 Your participation will be very much appreciated and your views and insight will be invaluable to the project aims.

 A summary of key findings from the questionnaire will be included in the final report, but no sensitive information will be published.

Please complete the questionnaire here. If you have any questions, please contact info@fera.co.uk.

Your kind participation will be very much appreciated and your views and insight will be invaluable to the project aims.

Photo by Stephanie Sarlos on Unsplash

This pre-print (open access) reviews recent advances in electroanalytical methods.  These have the advantage, for food authenticity applications, that they are generally cost-effective and adaptable to field conditions. This review covers the application of these techniques across various food matrices, including olive oil, honey, milk, and alcoholic beverages.

The author reports that, by leveraging methodologies such as voltammetry and chemometric data processing, significant advancements have been achieved in identifying both specific and non-specific adulterants.

The review highlights novel electrode materials, such as carbon-based nanostructures and ionic liquids, which enhance sensitivity and selectivity. Additionally, electronic tongues employing multivariate analysis have shown promise in distinguishing authentic products from adulterated ones.

The integration of machine learning and miniaturization offers potential for on-site testing, making these techniques accessible to non-experts. Despite challenges such as matrix complexity and the need for robust validation, the author concludes that electroanalytical methods represent a transformative approach to food authentication.

Halloumi produced in Cyprus has a transitional exception until 2029 from the EU PDO regulations which stipulate that >50% of the milk content must be from sheep or goats.  This is because of the relatively low sheep/goat milk production on the island.  However, national Cypriot law still stipulates that the sheep/goat milk content must be >19% during this transition.  Major dairy companies on Cyprus have lobbied against this transitional law, arguing that it is unachievable without large scale import of sheep/goat milk powder.

It has been reported that a 2024 survey of one of the largest halloumi brands on sale in Cyprus found sheep/goat milk content at only 5%.  The same newspaper also reports that the regulators are working with Bureau Veritas on building a reference database of compositional parameters, to address longstanding analytical challenges in verifying the sheep/goat milk content of imported milk powder.

Photo by Ambitious Studio* | Rick Barrett on Unsplash

This study (open-access author’s link available until February 14, with thanks to Michele Suman for sharing) reports the development and validation of a non-targeted classification method for authenticity of dried oregano leaves by atmospheric pressure matrix-assisted laser desorption ionization mass spectrometry (AP-MALDI-MS).

The model was trained on 23 authentic oregano samples (sourced from a reputable company with full supply chain traceability - originated from Italy, France, Turkey, or Albania, harvested between 2019 and 2022) along with five pure adulterants (dried leaves of savory (Satureja montana), myrtle (Lagerstroemia indica), sumac leaves (Rhus coriaria), strawberry tree (Arbutus unedo), and olive tree (Olea europaea)), plus sixteen adulterated oregano samples, intentionally mixed with the above mentioned adulterants at ranges between 5 % and 60 %.

The most abundant signals were characterized by collision induced dissociation and library search, the spectral data were submitted to statistical analysis. A basal inquiry of the data by partial least squared discriminant analysis (PLS-DA) was carried out for the simple assessment of the discrimination capabilities of the ± AP-MALDI-MS signatures. The researchers then constructed two distinct random forest (RF) classifiers using the positive and negative most informative ions teased out by recursive feature elimination from the training sets. The aforementioned most significant variables (m/z values) were also merged by mid-level data fusion and used to build a third RF classifier.

They report that the cross-validations of the three RF classifiers achieved good outcomes as demonstrated by the satisfactory values of overall accuracy (84.9 %, 92.1 %, and 92.8 %, respectively). The three RF classifiers were tested on the hold-out data, which revealed reliable classifier performances (accuracy 80.1 %, 87.0 %, and 85.4 %).

Photo by 360floralflaves on Unsplash

This paper (purchase required) reports a method to differentiate pork gelatin from beef gelatin (down to 0.01% cross-contamination levels) based on the LC-MSMS analysis of 13 peptide marker ions (8 for bovine, 5 for porcine).  The authors report that their method was validated at three concentration levels and accurately identified the gelatin species in pharmaceutical capsules and gels.

LC-MSMS analysis of peptides provides an alternative approach to DNA testing, which has known difficulties in application to highly processed products like gelatin due to the low amount of viable DNA or distinctive fragments.  LC-MSMS is the approach described in a recent Defra research report which is referenced on the FAN research pages (scroll down table to FA0177).

A recent FAN blog described non-destructive impedance sensors as a tool to classify meat freshness.

In this paper (open access) the authors have used the same principle and developed a classification model for potato varieties based on the effect of their dry matter content on an electrical impedance sensor.  The test is destructive as the potato must be sliced.  The authors built a reference database from data from 9 cultivars (Actrice, Ambra, Constance, El Mundo, Fontane, Gaudi, Jelly, Monalisa and Universa) sourced directly from the grower.  These cultivars were chosen because they cover a wide range of dry matter content.  The authors collected multivariate analytical data from the impedance sensor; impedance magnitude and phase data along with derived parameters such as the minimum phase point of each spectrum, the ratio between the low- and high-frequency values of the impedance magnitude,  the dissipation factor, the distance between the zero and the maximum value of the Nyquist plot, and  the Cole model equivalent circuit parameters.

They conclude that machine learning methods for predicting potato dry matter and varieties, based on impedance data, can achieve an equivalent (sub-optimal) performance to conventional methods and that they hold promise for future improvement to surpass conventional methods. An improved deeper analysis could aim to reduce the root-mean-squared error and increase the coefficient of determination value, thereby enhancing the accuracy of dry matter data predictions. To achieve this, various techniques such as feature engineering, hyperparameter tuning, and advanced modelling approaches (e.g. convolutional neural networks) could be explored. The authors consider that alternate chemometric methods like the Kennard-Stone algorithm, which selects representative samples based on distance criteria, could lead to more robust dataset partitioning. Additionally, incorporating data fusion with results obtained through infrared spectroscopy could further improve the model’s performance.

Photo by Rodrigo dos Reis on Unsplash

This study (open access) investigated species substitution, mislabeling, and the sustainability of seafood products in the seafood markets of South China. 478 samples were purchased from retail markets in 11 cities across three provinces (Guangxi, Guangdong, and Hainan) between May 2021 and December 2023. Cytochrome c oxidase subunit I (COI) gene amplification was used to identify 156 fish species across 105 genera and 60 families. The researchers have published the correlation between genetic and taxonomical details.

The researchers used a combination of morphological and DNA barcoding methods to produce an atlas guide for these 156 economically important fish species.

Molecular identification revealed that 9.6 % (15/156) of fish species were mislabelled, with commercial fraud detected in three processed species: Hilsa kelee, Chelidonichthys kumu, and Argyrosomus japonicus. Some substitutions may have been unintentional.  3.8 % (6/156) of species identified were classified as threatened by the International Union for Conservation of Nature. The study also uncovered an example of illicit cross-border sales of fish products.

The authors conclude that their findings provide a technical reference for effective fish species identification and offer valuable insights into seafood market monitoring.

Photo by Dan Gold on Unsplash

Impedance is a complex Cartesian function describing the difference between an inputting and exiting sinusoidal electrical signal.  It can be depicted graphically as a plot (vector) of resistance vs reactance.  The linearity of this plot, and the angle of the vector, are distinctive.  In a sample of meat or fish, impedance is affected by the cell structure and the water content.  Both of which are an indicator of freshness.  An impedance sensor, comparing the result with a “normal” database, can therefore be used to detect unfresh meat or meat that has been prior frozen and defrosted without declaration.

This review (open access) describes published applications, comparing the technique with other approaches such as HADH Enzyme measurement (see FAN method explainers).  It concludes that the development of Impedance Sensor methods is now at a stage where the technique is ideal as a cheap, non-destructive inline check in the food industry, particularly if coupled with machine learning to spot unusual or anomalous samples.

Photo by Victoria Shes on Unsplash

The Soil Association (the UK's leading organic certification body) has issued an alert about fraudulent certificates from two named companies.  The alert notification includes a list of links to certification bodies and official websites where you can cross-check the veracity of certificates, in EU, US, GB and a number of other countries