A masters thesis project sampled 106 retail packs of frozen shrimp (prawns) on sale in California and tested them for declared weight (i.e. excluding glaze), declared species and declared country of origin.  26% of samples had >20% glaze and 37% were legally underweight, rising to 57% in the super/extra colossal shrimp category. Ambiguous or incorrect species labelling was also observed in 37% of samples.  The vast majority (98%) of country of origin labels were verified as correct.

Photo by Fernando Andrade on Unsplash

The purpose of this review (purchase required) was to identify the most susceptible cheese type for fraud and the most commonly reported methods for evaluating fraud in all types of cheeses. The authors conducted a systematic review of the scientific literature. They conclude that Mozzarella cheese was most reported to be adulterated or at risk of adulteration.  The methods that were most used in detecting fraud were PCR and spectrometry methods, with less use being made of stable isotope, image analysis, electrophoretic, ELISA, sensors, near-infrared and NMR. The least used method was sensory evaluation.

Photo by Waldemar on Unsplash

The authors of this study (here – purchase required) recommend a protocol for authenticity testing of processed meats that combines PCR with histology.  PCR indicates species substitution whilst histology indicates substitution with cheaper organs from the same species.  They reported that, for a market survey of 105 sausages and beef hams in Iran, neither technique on its own gave a true picture; in fact, by PCR alone, all of the sausage samples would have been reported as unadulterated.  It was only by combining the techniques that they found over 55% of the hams and 65% of the sausages to be adulterated, mainly with cheaper tissues from the same species.  It proves the continued value of histology, which requires experienced microscopists to interpret slides; an increasingly rare skill within routine analytical laboratories.

Photo by Harpreet Singh on Unsplash

A recent publication (here – open access) describes the development and validation of a smartphone reader to combat the previously reported fraud of adding red dye to cheaper species of tuna and passing them off as Thunnus thynnus (red tuna).  The red colour is associated with freshness in the minds of consumers.  The authors built a reference database of colourimetric readings (as CE XYZ data) using a spectroradiometer.  This classified tuna by species and/or by adulteration with beetroot extract.  They then used the pro camera on a commercial smartphone (Galaxy 7) to read test samples.  Reading with the smartphone required a dark room.  They used statistical functions to convert the camera’s RGB reading to CEXYZ and an achromatic reference sample to convert to CIELAB colour space to then match samples against the classification model.  They reported good matching, with a 0.6% false negative rate and 10% false positive rate for validation samples classified as adulterated or mis-labelled.

The Insider Business US TV documentary “11 of the most faked foods in the world” has now been posted on YouTube and is generating interest (2M views in 3 days).  They cover truffles, maple syrup, wasabi, Parmesan cheese, vanilla, caviar, honey, olive oil, Wagyu beef, coffee and saffron.  For each, they describe why it is a premium product and how it could be adulterated or substituted.  They do not discuss why they chose these foods as the “most faked” or justify the conclusion that consumer purchases in the US are “probably faked” but they do discuss examples and case studies.

Laser Photo-Acoustic Spectroscopy (LPAS) can be used to obtain a spectrum of dried herbs and then classification models can be built to detect adulteration.  It is analogous in this way to conventional Infra-Red spectroscopy.  The advantage of LPAS is the power of the source; a laser, vs a lamp.  In this publication (a preprint, not yet peer-reviewed) the authors report the development of an LPAS system trained and calibrated to detect olive leaf adulteration in oregano down to 20% and ideal for in-line use in an industrial setting.

A recent paper (purchase required) describes a feasibility study to use non-destructive ultrasonic inspection to detect olive oil adulteration with two other edible vegetable oils (sunflower and corn). Pulsed ultrasonic signals with a frequency of 2.25 MHz were used. Test samples were adulterated in variable percentages between 20% and 80%. The viscosity and density values were shown to correlate with acoustic parameters (ultrasound pulse velocity, frequency variables obtained from the Fast Fourier Transform, and attenuation) when measured at both 24 °C and 30 °C.  Acoustic parameters were able to discriminate adulteration at all of the percentages and mixes studies. The responses obtained through the parameters related to the components of velocity, attenuation, and frequency of the ultrasonic waves are complementary to each other. The authors concluded that classification of pure and adulterated oil samples is possible through non-destructive ultrasonic inspection.

Photo by Roberta Sorge on Unsplash

This paper (purchase required) describes the development of an array of 14 sensors based on colourimetric reactants immobilised on a paper or plastic support.  The advantage of this approach, rather than traditional “e-tongue” systems based on reactions in liquid solution, is that it enabled the development of a “dipstick” test that could be taken into the field rather than having to send test samples to a laboratory.  The authors report that they successfully used Machine Learning to train the system to discriminate between different botanical and local geographic origins of Iranian honey and also to discriminate when honeybees within these specific classifications had been illicitly fed with sugar syrups.  It shows the potential of a cheap field-based test which would be trained and used for verification testing at the beginning of honey supply chains where very specific "authentic" classification references are available, rather than relying on testing further up the chain when chemometric classification becomes much more diffuse and difficult.

An Australian survey has revealed a serious problem with mislabelling of seafood products.  More than 10% of retail and food-service samples tested were found to be mislabelled as to their species.  Shark and ray products had the highest occurrence of mislabelling at 35.9%. A specific issue was the mislabelling of protected species.  DNA barcoding revealed that in 18 instances, products incorrectly labelled as "flake" – or to a lesser extent as "shark" – were in fact holocephalians (i.e. chimaera fishes).  1 sample of spotback skate was mislabelled as "stingray" and 1 sample of school shark mislabelled as "gummy shark", both of which are critically endangered.  In total, 672 samples from supermarkets, fishmongers and restaurants were sampled across seven Australian states.  The study, by the Perth-based Minderoo Foundation (a privately-funded sustainability campaign group) was published in Nature and has been picked up by the media (example here)

A recent article describes a novel approach using Machine Learning to detect large-scale fraudulent online sellers by recognising give-away signs in how they conduct their digital marketing. The strategy is scalable. The authors employ a supervised learning approach to classify postings as fraudulent or real based on past data from buyer and seller behaviours and transactions on a popular online marketplace platform. They combine bespoke data preprocessing procedures, feature selection methods, and state-of-the-art class asymmetry resolution techniques to search for aligned classification algorithms capable of discriminating between fraudulent and legitimate listings. Their best detection model obtained a recall score of 0.97 on the holdout set and 0.94 on the out-of-sample testing data set, based on 45 selected features.

The use of calcium carbide to artificially ripen fruit such as bananas and mangoes is suspected to be prevalent in many countries.  It is cheap and effective.  It is also banned in most countries; calcium carbide is a carcinogen.  The authors of this paper (open access) have developed a cheap and simple test for calcium carbide that can be used a point-of-sale screening tool by enforcement officials or even consumers.  It is based on an instant colour change when free sulfahydryl groups (a ubiquitous impurity in calcium carbide) reacts with 5,5′-7 dithiobis-(2-nitrobenzoic acid.  This colour change is visible to the naked eye at calcium carbide concentrations typically used to ripen fruit.

It is 10 years since the Global Food Safety Initiative (GFSI) launched an interdisciplinary approach to detect food fraud and to understand and reduce the root causes.  This review article by Prof John Spink (purchase required) reflects on its evolution and impact.  He concludes that GFSI achieved momentum by establishing a food industry-wide definition and scope of the problem. Focusing on all types of fraud and for all products – not just incoming goods and adulterant-substances – created a holistic approach. Prof Spink describes this evolution in terms of the “hype cycle”, and concludes that food fraud prevention is is now in the ‘Scope of Enlightenment’ (e.g., processes and systems are simplified and optimized) and moving to the ‘Plateau of Productivity’ (e.g., standard operating procedures are adopted). GFSI, itself, is maturing and evolving through the ‘hype cycle.’ He predicts that the next ten years will include a more rigorous and thorough adoption of management systems with a continuous improvement process.

Saffron is one of the highest value food products and so often a target for adulteration.   Carthamus tinctorius L. (safflower stamens), dyed stamens of Nelumbo nucifera Gaertn. (lotus stamens) and stigmas of Zea mays L. (corn stigmas) have all been reported as bulking agents.  In this paper (purchase required)  the authors have developed a cheap and quick fluorescence probe which is selective to saffron and can be calibrated to detect dilution or bulking.  The three-dimensional (3D) fluorescence uses a hydrophilic hydrazine-naphthalimide functionalized chitosan (HN-chitosan) polymer probe. The amino functional group in the HN-chitosan probe react specifically with the oxygen-containing group of active ingredients in saffron. The authors applied four advanced chemometrics methods for the classification of saffron and adulterated saffron, and reported good performance in both training and prediction sets. They successfully applied the PLS regression model to the prediction of adulteration levels in saffron. They conclude that this strategy provides a new solution for rapid identification and quantification of adulteration in saffron.

Photo by Vera De on Unsplash

Different sources of animal feed have different profiles of natural stable isotopes.  For example, C3 plant-driven feeds (including natural pasture, grass silage, soybean silage) have a significantly different carbon-isotope ratio to C4 plant-driven feeds (including maize silage).  These differences can underpin tests to verify premium claims such as “grass fed beef”.  Similar price premiums apply to yak meat.  In this paper (purchase required) the authors bult a multivariate model which discriminated between grass-fed and maize-fed yak meat by stable-isotope ratio analysis (C, N, S, O, N) of the muscle and stable-isotope ratio analysis (C) of six selectively extracted fatty acids.  As with all such models, its use is restricted to the species, geographic regions and husbandry practices included in the reference database but it proves the concept.

Photo by Jasper Garratt on Unsplash

The authors of this study (purchase required) used an untargeted chemometric screening approach,  LC-Q-Orbitrap MS analysis,  to identify discriminatory markers between organic and conventional milks.  Putative markers were then verified by targeted analysis.  In the targeted strategy, they used UPLC-MS/MS to quantify three markers. They propose the peptide of Thr-Ala-Val and D-biotin as distinguishing markers for organic, vs conventional, Jersey milk.  They propose trimethylamine N-oxide as a metabolite marker for organic, vs conventional, Yak milk.  They do not state which aspect of husbandry differences might be responsible for differences in these markers.

Photo by Kim Gorga on Unsplash

The latest blog post by Prof John Spink describes the output of an EU Joint Research Centre exercise to map the honey supply chain.  He concludes that it is an exemplar model of how this should be done; it is a systematic picture of nodes, links, vulnerabilities at each point, "guardians" at each point, vulnerability "hot spots" and countermeasures.  John also lists some training resources on how to conduct supply chain mapping and vulnerability assessment.  You can sign up to receive e-mail notifications of his blogs through the same link, above.

This paper (purchase required) describes the development of an easy-to-read point of use DNA test for beef, pork and chicken.  It is based on a Nucleic Acid Lateral Flow Strip (LFS) coupled with multiplex PCR. The read-lines on the LFS develop within 10 minutes.  The coloured lines are produced by triplex colour latex microsphere labels, enabling a different colour for each species.  The authors report that, if the lines are read by a smartphone imager, the quantitation compares favourably with electrophoresis and detection limits of 0.1% w/w in adulterated meats can be achieved. Specificity analysis revealed no cross-reactivity with meat derived from beef, pork, chicken, duck, mutton, rabbit, goose, ostrich, camel and horse. The assay was successfully employed for adulteration identification in 28 commercial beef and lamb products.

Photo by John Cameron on Unsplash (And, yes, this is a Covid LFD.  Just for illustration.)

Indicator Displacement Array (IDA) is a rapid on-site test technique based upon a panel of different colourimetric indicators (e.g. different types of pH indicators, and/or different colour-labelled receptor binding sites).  It gives a multi-component data set which can be read using a smartphone.

In this study (here – purchase required) the authors applied IDA to authenticating Longjing tea.  Longjing is a legal Protected Designation of Origin (PDO) in China and has a premium retail value.  The name cannot be used for teas from the same botanical species, using the same production methods, if they originate from other regions.  There has been a history of fraudulent “Longjing” labelling.  The authors report that they have developed and validated a chemometric identification model for Longjing tea using IDA.  They have also designed a panel of IDA sensors that are food-contact safe, meaning that their design is suitable for routine use within the tea supply chain.  They conclude that IDA sensors could be similarly developed and applied to other applications where the origin of food or ingredients needs to be verified at point-of-use.

Photo by Bluesea Tea on Unsplash

The authors of a recent study (here – purchase required) report on the use of biogenic amines to characterise honeys from different regions of Brazil.  They found that the pattern of amines in honey, including serotonin, melatonin, dopamine, histamine, putrescine and cadaverine, varied depending upon the geographic region, foraging flora and production year.  The study did not report that this variation was distinctive enough be a basis for identity verification but it does give another set of data parameters that could be added into a multi component analysis panel of identification characteristics.  The authors propose that amines could be used as a quality specification parameter for some honeys, as many of these compounds have perceived health benefits for consumers.

Photo by Art Rachen on Unsplash

A customs operation in Greece led to the siezure of 21 tonnes of untaxed ethanol intended for the production of contrabrand drinks.  The consignment was carried by a Serbian distribution company and had originated in Bulgaria.  This was an intelligence-led operation started by the Bulgarian customs authorities.  The cover load was nominally windscreen wash but investigation revealed that the destination company was fictitious.

Read a report here