Fraud in highly appreciated fish detected from DNA in Europe may undermine the Development Goal of sustainable fishing in Africa

Despite high effort for food traceability to ensure safe and sustainable consumption, mislabeling persists on seafood markets. Determining what drives deliberate fraud is necessary to improve food authenticity and sustainability. In this study, the relationship between consumer’s appreciation and fraudulent mislabeling was assessed through a combination of a survey on consumer’s preferences (N = 1608) and molecular tools applied to fish samples commercialized by European companies. We analyzed 401 samples of fish highly consumed in Europe and worldwide (i.e. tuna, hake, anchovy, and blue whiting) through PCR-amplification and sequencing of a suite of DNA markers. Results revealed low mislabeling rate (1.9%), with a higher mislabeling risk in non-recognizable products and significant mediation of fish price between consumer´s appreciation and mislabeling risk of a species. Furthermore, the use of endangered species (e.g. Thunnus thynnus), tuna juveniles for anchovy, and still not regulated Merluccius polli hake as substitutes, points towards illegal, unreported and/or unregulated fishing from African waters. These findings reveal a worrying intentional fraud that hampers the goal of sustainable seafood production and consumption, and suggest to prioritize control efforts on highly appreciated species.

Scientific Reports | (2021) 11:11423 | https://doi.org/10.1038/s41598-021-91020-w www.nature.com/scientificreports/ those cases the defrauder benefits from selling inadvertently species that are often poorly managed and wild caught 15 . Examples are the covered exploitation of protected species, (e.g., high levels of mislabeling concealing endangered species have reported in sharks 16,17 , or selling individuals caught over quota as suggested in Miller et al. 18 . This is especially pernicious not only for species conservation but also for the society, because selling IUU catch undermines consumer's awareness and conscious efforts of sustainable consumption 19 .
In the control of deliberate mislabeling, a priority is to determine the point of the commercial chain and the type of product where it happens more frequently 20 ; that is to know mislabeling niches for control to be reinforced therein 21 . Is there a role of consumer's appreciation in fish mislabeling? Apparently, it is especially frequent in highly appreciated products, like sushi 22 , processed cod products 23 , scarce and expensive Bluefin tuna Thunnus thynnus that is commonly replaced by other tuna species 24,25 . In other words, defrauders sell what the consumer wants whether the product is real or not. However, fraudulent mislabeling having an important economic component, defrauders will substitute preferentially, if not only, expensive species.
In this study we will focus on the relationship between consumer's appreciation and fraudulent mislabeling in Spain. This country was chosen for having been identified as one of the countries with highest seafood consumption (42.4 kg/person/year) and also seafood mislabeling in the world 14 . We considered two species pairs of blue and white fish, one cheap and one expensive from each type. We analyzed consumer's appreciation from self-reported consumption habits, and, using a suite of DNA markers to authenticate seafood samples, we determined mislabeling in recognizable and unrecognizable fish products.
Expectations were: First, in deliberate mislabeling a higher proportion in not recognizable versus recognizable products and expensive versus cheap species (Hypothesis a). Second, and central to this study, price will mediate between consumer's appreciation and % of substitution, more expensive species being more mislabeled regardless consumer's appreciation (Hypothesis b).

Results
Fish mislabeling. In  albacares, 3 T. obesus) and 90 as blue whiting (Micromessistius poutassou). The sequences of the genetic barcodes were aligned within species, and the haplotypes (sequence variants) submitted to GenBank, where they are available with the accession numbers MW557502-MW557516 for control region barcodes and MW549508-MW549521 for COI barcodes.
Mislabeling was found in seven samples, 1.9% of the total samples analyzed ( Table 1): two of anchovy (one product per sample), two of hake (four products, one in one sample and three in other) and three of tuna (five products in total: one, two and two per sample). The total proportion of mislabeling was 0% in blue whiting, 2.4% in anchovy, 4.17% in hake and 5.2% in tuna. From the label, five of the 11 mislabeled products were caught in front of African coasts (Center-NE Atlantic, Central E Atlantic, SE Atlantic), four from European waters (NE Atlantic, Aegean Sea) and two had no geographic information in the label (Table 2).
Mislabeling was significantly more frequent in not recognizable products (10 out of 11 mislabeled products, Table 2), and in expensive species that are tuna and hake in this region (only two anchovies and none blue whiting). Mislabeling risk ratio RR of non recognizable versus recognizable products was 1.0369, 95% confidence [1.005-1.07], being significantly higher than 1 with z = 2.292 and p = 0.022. Similarly, mislabeling RR of expensive versus cheap species was 1.0371, 95% confidence [1.001-1.074], also significant (z = 2.022 with p = 0.043).
A deeper view at the mislabeled products (Table 2) suggests that the mislabeling found in this study could be compatible with IUU, at least in some cases. Mislabeled anchovies were about 12-13 cm long, and substitute species were small tuna fry or juveniles, one identified from DNA as yellowfin Thunnus albacares (the capture region stated on the label was NE Atlantic), and the other as Thunnus alalunga (captured from the Aegean Sea in the Mediterranean basin). The first one was sold whole, in fresh, eviscerated but with head and tail thus it was recognizable; the other was frozen headless, tailless and eviscerated. Under European (Regulation EU 850/98 of the European Council of 30 March, 1998) and Spanish laws (Royal Decree 560/1995 of 7 April, 1995), fish juveniles are protected and these tuna juveniles cannot be caught nor commercialized, thus the origin of these substitute fish was unequivocally IUU (Table 2). Small tuna fry and adult anchovies are indeed morphologically different but could be confounded by non-expert consumers, especially when sold headless. Engraulis encrasicolus distribution overlaps with spawning areas of Thunnus albacares in northeast African waters, and with those of Thunnus alalunga in the Mediterranean Sea, thus these substitutions are compatible with capturing tuna juveniles while fishing anchovy.
All the mislabeled hakes were sold frozen, in fillets or slices, labeled as Merluccius capensis/M. paradoxus that are South African hakes ( Table 2). Three of them were substituted by M. polli that is an African species living from tropical and subtropical East Atlantic waters, and overlaps with M. capensis in the southern part of its distribution. South African M. capensis/paradoxus stocks are fished in a sustainable way certified by the Blue label of Marine Stewardship Council (https:// www. msc. org/ home/ meet-the-wild-ones/ from-sea-to-shelf-wildsouth-afric an-hake, accessed January 2021), and they should not be caught together with other species. Thus the cause of this particular substitution is compatible with IUU (without discarding substitution in a later step). The other mislabeled hake was substituted by north Atlantic cod Gadus morhua which distribution does not overlap with south African hakes, thus substitution during fishing or landings was not possible in this case and must have occurred later in the commercial chain (Table 2).
Mislabeled tuna products were all labeled as "bonito del Norte" Thunnus alalunga (longfin or albacore tuna), which is the most frequently sold tuna species in the region. The three products that were substituted by northern   (Table 2) were all unrecognizable (slices), but the fish body (without the head) was visible and slices were cut fresh in front of the customer. They were sold in two different stores. The other two mislabeled samples did not report the capture region on the label. They were frozen slices, and were identified from DNA as yellowfin tuna Thunnus albacares and bigeye tuna Thunnus obesus, the last one with lower score and identity % with the best match in BLAST ( Fish consumption and its relation with mislabeling. In the social survey (N = 1608 participants aged between 14 and 63 years) the average frequency of fish consumption was 3.13 (SD 0.923), that is, surveyed participants declared to eat any fish with a frequency between monthly and weekly (1 being the highest and 5 the lowest frequency, 3 being weekly), that is, at least monthly. The raw dataset can be found in Rodriguez et al. 27 . A clear difference between species was found regarding consumption levels ( Fig. 1). Tuna and hake were consumed weekly and monthly by a much larger proportion of participants than anchovy and whiting. Moreover, 49% and 61.6% of the participants declared not eating anchovy and blue whiting, respectively, being this proportion 20.7% for hake and 14.3% for tuna. The difference between species for the pattern of consumption frequency was highly significant ( χ 2 4,4 = 1154.8 with p < < 0.001). Consumer's appreciation of a species was estimated from the proportion of participants consuming it at a frequency at least equal to the average frequency of fish consumption, this meaning that the species accounts for an important proportion of the fish consumed. Since the average fish consumption frequency in the region was 3.13 (= between monthly and weekly), we added the classes "Weekly" and "Monthly" (i.e. at least monthly) consumption of a species as a proxy of its appreciation. Mislabeling was significantly correlated with both price (Fig. 2 above) and the consumer's appreciation proxy (Fig. 2 below): r 2 = 0.945 and 0.91 respectively, both with p < 0.05.
The relationships between these variables are summarized in Fig. 3, with price as a mediator between consumer's appreciation and fraudulent mislabeling. Zero-order unstandardized regression for predicting price from consumer's appreciation (measured from consumption here) had B = 5.945 (SE = 0.806). Partial unstandardized regression coefficient for predicting mislabeling from fish price holding consumer's appreciation constant was B = 0.015 with SE = 0.02 (Table 3). The distribution of products statistics Z α *Z β was statistically significant being 5.376 > 2.18, p < 0.05. Mediation effect of price between consumer appreciation and mislabeling was thus significant in spite of limited number of species analyzed.

Discussion
Combining a suite of molecular tools (three DNA markers and four different protocols) and a social survey amongst consumers, this work showed several evidences of fraud in appreciated species that are important for both food sciences and conservation. The mislabeling found in this study was likely fraudulent, since, according to hypothesis a, its risk was significantly higher in expensive than in cheap species, and in not recognizable than  www.nature.com/scientificreports/ in recognizable products. Moreover, according to Hypothesis b it affected principally the species most appreciated by the consumer and amongst them those with higher prices, since price appeared to be a mediator between consumer's appreciation (or, directly, consumption of a species) and mislabeling. This result points at the need of focusing on highly consumed highly priced species as priority mislabeling niche. Inadvertent mislabeling due to simple morphological confusion between species, frequently reported in previous studies in Spain 10 , does not seem the main cause of mislabeling here. As in other studies, mislabeling occurred in tuna and hake. The level found was comparable to that published by other authors in Spain and elsewhere. Sotelo et al. 25 found 6.7% mislabeling in fresh and frozen tuna in European markets of six countries between 2013 and 2014, similar to the 5.2% found here, although in Spain the level they found (20% in 20 samples) was clearly higher than in our study in which sampling was done six or five years later. Gordoa et al. 24 also found higher levels of tuna mislabeling in Spanish supermarkets in samples of 2014-2015. It is possible that mislabeling is decreasing in Spain in the last years, following the current trend of improvement of market operations pointed out by Mariani et al. 6 at European level. Thunnus albacares and T. obesus were the main substitutes in Sotelo et al. 25 study, while T. thynnus was the most frequent substitute here. This reverse substitution (expensive T. thynnus for a cheaper species) has been rare, and when happening it was interpreted as indicative of illegal origin 24 . All together our results would indicate a change in tuna mislabeling trends in Spain.
Decrease of hake mislabeling seems to be happening in Europe as well, and the increase of the last years found in Blanco-Fernandez et al. 7 review would be going down in Spain, since the proportion of mislabeled hake products found here (less than 5%) was low again. A difference with past mislabeling reports was the main substitute species found here, the Benguela hake Merluccius polli, rarely found as a substitute in other studies 7 . This would imply a change in hake substitutes and the unreported introduction of this species in European markets.
Anchovy mislabeling in Europe has been reported before 28 . It affects especially semipreserved products where it can as high as 16% being Engraulis anchoita the substitute species 29 . To our knowledge this is the first time that Thunnus is reported as a substitute of anchovy products. As in hake and tuna, it seems that the mislabeling trends are changing in this country towards different substitute species.
Although from our data and other studies 6 fish mislabeling seems to be decreasing in Spain, the substitute species of this study are of concern for conservation because they point at the involvement of IUU, one of the causes of fraud identified by Fox et al. 20 , in current mislabeling in Spain. Finding small juveniles of near-threatened Thunnus albacares and T. alalunga sold as anchovy may reveal a serious problem in anchovy fisheries. If they catch juveniles of other species as by-catch, they are not sufficiently selective of the target species. Being true that the proportion was small, about 2.4% of the individuals analyzed, multiplying it by the tons of anchovy catch of Asturias fleet (1 566 059.4 kg in 2019 from official records; https:// temat ico. astur ias. es/ dgpes ca/ fich/ 992019_ Z. pdf) it would represent about 37,585 kg of unreported tuna fry, more than 35 tons. On the other hand, the use of endangered Thunnus thynnus (and perhaps the vulnerable Thunnus obesus too) in substitution of less threatened tuna species is, definitively, a call of attention for the increase of control in European tuna markets, as demanded by other researchers 25 . Since the catch areas of label and substitute species overlap, these substitutions would suggest that Bluefin tuna caught over quota is being sold as another species.
Finally, Merluccius polli -main substitute of M. capensis/paradoxus in hake products here-is now the subject of sustainable management plan designs in its northern distribution 30 , while in the center and south the stocks are largely unknown 32 . Introducing the species as a substitute in European markets impedes traceability of its catch. It could be introduced at industrial level (for example to complete a batch of slices or fillets), or be another example of the use of unregulated fishing products. Selling M. polli from IUU would hamper the efforts of sustainable management of African hakes, both the currently MSC-certified South African Merluccius capensis/paradoxus and the future of Benguela hake stocks. introduction of the species as a substitute dustrial.
In conclusion, these results reaffirm a general trend of decreasing mislabeling rates, which is to be expected as efforts in food sciences for better traceability continue. However, while inadvertent substitutions decrease, special attention must be directed towards the detection of deliberate fraud. The found substitutes are an indication for IUUs as a source for fraud. This is a cause for concern and calls for a focus on products that are especially susceptible to these actions; i.e. highly consumed, expensive products. Monitoring of this type of products must be reinforced in order to prevent that fraudulent practices reach the consumer's plate. Moreover, the whole SDG 12 is undermined if conscious consumers cannot be sure of eating really their choice of sustainably caught fish. Table 3. Regression analysis of the variables mislabeling (dependent), price (expected mediator) and consumer's appreciation (independent). Unstandardized regression coefficients (B) and their standard error (SE), t statistics and its p-value, and R 2 of the partial correlation. Zero-order regression predicting price from consumer's appreciation above; below, multiple regression results of the three variables.

Methods
Ethics statement. This study adheres to good research practices based on the European code of conduct for research integrity 31 . Since part of the research required the application of a population survey, the Committee of Research Ethics of Asturias Principality approved this study with reference 166/19. All methods were performed in accordance with relevant guidelines and regulations. Prior to the application of the survey, all participants were provided with an information sheet detailing the objectives of the research and that no personal data, except age and place of application, would be collected. Informed consent was obtained from all the participants. Likewise, their willingness to participate in the research was guaranteed. They were invited to verify their answers, to have access to the study results, and were informed about their right to withdraw from the process at any time. From these regulations, seafood labels must contain at least the following information: common and scientific names of the species; geographic origin (country of production, or FAO area if from marine fishing); production method (extractive fisheries or aquaculture).
Products selected were big enough or contained the tail or head, to be sure that each sample corresponded to one individual. When the product was packed (e.g. fillets or slices) we sampled only one piece by pack. Products were classed as recognizable (i.e. the head is present and the whole body contains morphological characteristics that serve to identify the species) or unrecognizable (the species cannot be ascertained visually). Muscle fragments from each sample were cut, preserved in 100% ethanol and stored in the laboratory at 4 °C until analysis. Molecular analysis. DNA was extracted with Chelex-100 Resine (10%) and 4.5μL proteinase K (20 U/ml, by EURx) from frozen tissue. Different tools were employed for identification of the different fishes (Table 4). Cytochrome Oxidase Subunit I gene (COI) was employed for the first identification of blue whiting, anchovy and hake samples using primers described by Ward et al. 33 . Tuna was identified using the primer combination of L15998 with H-strand primer CSBDH, described by Alvarado-Bremer 34 and used by Viñas and Tudela 35 to distinguish between the eight Thunnus species from any kind of processed tissue, that amplify a fragment of the mitochondrial control region.
In some cases, a second marker was necessary to distinguish between different closely related species belonging to the same genus. For anchovy, mitochondrial control region was amplified using the primers described by Alvarado-Bremer 34 . Mitochondrial control region was also amplified for the identification of hake samples, but using primers MmerHK01 and MmerHK02, described by Lundy et al. 36 .
PCR amplifications were carried out using 10 pmol of each primer, 1.5-2.5 mM MgCl 2 , 0.25 mM dNTPs, 1 × Buffer GoTaq Promega, 0.15 µl of GoTaq Polymerase (5u/µL) and 2-4µL of DNA in a final volume of 20µL, with little differences among primers (Table 4). www.nature.com/scientificreports/ PCR amplifications were carried out in a thermo cycler ABI PCR 2700 (Applied Biosystems) with the following conditions: an initial denaturing step of 90 °C for 5 min, followed by 35 cycles consisting of 30 s of denaturing at 90 °C, 30 s of annealing at 55 °C for hake-control region amplification and 57 °C for COI, an elongation step for 45 s at 72 °C, and a final elongation step at 72 °C for 15 min after the cycles for hake-control region, and for 10 min for COI. And an initial denaturing step of 94 °C for 5 min, followed by 35 cycles consisting of 45 s of denaturing at 94 °C, 45 s of annealing at 50 °C, an elongation step for 1 min at 72 °C, and a final elongation step at 72 °C for 5 min for anchovy and tuna control region. Resulting amplicons were sent for Sanger sequencing to Macrogen Inc. (Spain).
Sequences were edited with BioEdit v.7.0.5.3. and Lasergene Seqman software by DNASTAR 37 . Sequences obtained from forward and reverse primers separately were combined whenever possible to obtain a more robust barcode. Edited sequences were compared to GenBank sequence database, using BLAST (Basic Tool Alignment Search Tool) (https:// blast. ncbi. nlm. nih. gov/ Blast. cgi) optimized for highly similar sequences (megablast), with word size = 28, match/mismatch scores of 1 and -2, linear gap costs, threshold 0.05. The best match hit was chosen as putative species of the sample from DNA.

Survey to investigate consumers' preferences.
To know the habits and preferences regarding fish consumption, a survey was conducted in urban and rural geographical areas of Asturias. This Spanish province of 1,000,000 inhabitants has a rich tradition of fishing and seafood is important in its gastronomy. In a face-toface format, the researchers went to secondary education centers, university schools/faculties and consumer centers for its application. Participants were asked to indicate their age, gender and place of residence. Two questions about fish consumption followed. The first question (five possible choices) was: "How frequently do you consume fish normally? Please choose one of the following options: (1) More than 5 days a week; (2) Between 2 and 5 days a week; (3) One or two days a week; (4) Monthly; (5) Less than monthly". Then they were asked for the frequency of consumption of each of the four target species using the following question: "How frequently do you eat the following species? Please choose the option that fits best your consumption habits", with four possible choices: (1) Weekly; (2) Monthly; (3) Occasionally; (4) Rarely/Never".

Statistics.
To test hypothesis a about fraudulent mislabeling, comparison between different types of samples (e.g. recognizable versus not recognizable, cheap versus expensive species, etc.) regarding the proportion of mislabeled samples was done employing the risk ratio (RR), with z tests and their p H o being RR = 1.
Differences in consumption between the four fish species based on the number of informers consuming a species weekly, monthly, occasionally or [rarely + never] (four classes, four species) were tested using Chi-square of contingency with nine degrees of freedom.
The variable Consumer's appreciation of a species was estimated based on the average of total fish consumption. The proportion of participants that consume a fish species (anchovy, hake, tuna, blue whiting) at a frequency at least equal to the average frequency of any fish consumption (first question) was taken as a proxy of regional consumer's appreciation of that species. Pairwise correlations between this variable, price and proportion of mislabeled products were calculated using Pearson's r. The hypothesis of price as a mediator between consumer's appreciation and mislabeling was tested from the distribution of products following MacKinnon et al. 38 and Wuensch 39 , taking into account the small number of fish species where substitution was measured. Briefly, β = unstandardized regression coefficient for predicting the mediator (price) from the independent variable (consumer's appreciation), and α = partial unstandardized regression coefficient for predicting mislabeling from price holding constant the consumer's appreciation were divided by the standard errors to calculate Z β and Z α scores respectively. For a 0.05 non-directional test the critical value of the statistics Z α *Z β is 2.18. Thus, higher values can be considered significant.
Significance threshold was p < 0.05. Statistics was done with free software PAST version 2.17c 40 .