Product Authentication Using Two Mitochondrial Markers Reveals Inconsistent Labeling and Substitution of Canned Tuna Products in the Taiwanese Market

Fish of the tribe Thunnini represent a significant proportion of the stock caught by the fishing industry, with many of these fishes being collectively called tuna. However, only certain species can be used legally as an ingredient in canned tuna products, depending on regional food regulations. In Taiwan, only Thunnus species or Katsuwonus pelamis can be used as canned tuna. Here, we authenticated 90 canned tuna products, including 25 cat food samples, by sequencing two mitochondrial regions, 16S rRNA (16S) and the control region (CR). BLAST analysis revealed that Sarda orientalis, Euthynnus affinis, Auxis rochei, and Auxis thazard are all used as substitutes for legitimate tuna products. We found that 63.33% of investigated samples are true canned tuna, i.e., contain Thunnus species or skipjack tuna. We advocate that the Taiwanese government publishes an official standardized list of fishes, especially so that scientific, Chinese and vernacular names can be assigned unambiguously based on a “one species-one name policy”, thereby clarifying which species can be used in seafood products such as tuna. Furthermore, we feel that the large-scale and long-term monitoring of canned tuna products is warranted to fully assess the extent of tuna product adulteration in Taiwan.


Introduction
Approximately 17% of the global human population's intake of animal protein in 2017 constituted fish [1]. Although aquaculture satisfied about half of that consumption, wild-capture from oceans, lakes and rivers remains a mainstay of the global fishing industry. Among these wild-caught fishes, scombrids are particularly important fishery resources, especially species in the tribe Thunnini that constitute~10% of the international seafood market [2,3]. In 2018, the global catch of Thunnini species represented~7.9 million tons, 58% of which can be attributed to skipjack tuna (Katsuwonus pelamis) (in Chinese: 正鰹) and yellowfin tuna (Thunnus albacares) (in Chinese: 黃鰭鮪) [1]. A large proportion of the Thunnini catch is destined for the canning industry [4,5].
The tribe Thunnini comprises five genera: Thunnus (in Chinese: 鮪屬), Katsuwonus (in Chinese: 正鰹屬), Auxis (in Chinese: 花鰹屬), Euthynnus (in Chinese: 巴鰹屬), and Allothunnus (in Chinese: 細鰹屬). Fishes of this tribe can be generally termed "tuna". For example, Auxis rochei is called bullet tuna (in Chinese: 圓花鰹), Euthynnus alleteratus is little tunny (in Chinese: 小巴鰹), and Allothunnus fallai is slender tuna (in Chinese: 細鰹). However, the Chinese translation of tuna is 鮪 in Taiwan or 金槍魚 in Mainland China, which specifically refers solely to Thunnus spp. Previously, many different scombrids were used as an ingredient in "canned tuna", even if they did not belong to the tribe Thunnini. For instance, Sarda (in Chinese: 齒鰆屬) spp. were once widely used in canned tuna because they possess a similar taste and texture to it [6]. Importantly, a species of the tribe Thunnini may not always be used legally as a canned tuna product ingredient. Various legislative bodies have developed regulations that clearly define which species can be used in canned tuna products (Table 1). The Food and Agriculture Organization (FAO) and the federal government of the United States allow spotted tuna to be used as canned tuna, but that species is prohibited by Taiwanese and Japanese regulations. In general, fishes of the genus Thunnus and skipjack tuna are widely recognized as legal canned tuna species. To align with international standards, the Taiwan Food and Drug Administration allows skipjack tuna to be used in 鮪 (for tuna)-labeled canned tuna products, even though it does not belong to the genus Thunnus, but other "pseudo-tunas" can no longer be used legally as a canned tuna ingredient.

Sample Collection
We purchased a total of 90 canned tuna products, belonging to 59 brands, from grocery stores or online, encompassing all major brands in Taiwan. Twenty-five of the collected samples represented canned cat food. Cans were selected if their label showed the Chinese word 鮪 (for tuna), if the company website claimed the product was made from Thunnus fishes, if the ingredients list contained Thunnus spp. or skipjack tuna, or if an image on the label indicated the can harbored Thunnus fishes. We recorded information, typically written in Chinese, on brand, manufacturer or importer, place of manufacture, labeling, and ingredients. If the cans were imported from the USA or Japan, the respective English or Japanese labels were also recorded ( Table 2). The sampled cans were first photographed using a smartphone (Supplementary Information S1), and then a small quantity of the contents of each can was removed using autoclaved dissection tools, washed with 95% ethanol, before being preserved in 99.5% ethanol at −20 • C until DNA extraction. Some of the canned cat food products contained more than one type of meat, so potential Thunnus meat was selected based on color and texture.

Molecular Identification
DNA was extracted from each of the 90 tissue samples using DNA Extraction Kit S (Cat No./ID: GS100, Geneaid). PCR amplifications of the mitochondrial 16S rRNA fragment (16S) (85 bp) were performed in a mixture containing 5 ng template DNA, 12.5 µL of 2× Taq PCR MasterMix (GN-PCR201-01, Genomix), and 12.5 µmol of each forward and reverse primer. We used primers designed by Horreo et al. [47] and modified them by adding M13 primer to facilitate sequencing: Forward, M13F(-20)16S-HF (5 -GTA AAA CGA CGG CCA GTA TAA CAC GAG AAG ACC CT-3 ); Reverse, M13R(-24)16S-HR1+2 (5 -AACAGCTATGACCATGCCCRCGGTCGCCCCA AC-3 ). These primers were made up to a final volume of 25 µL using distilled water. If BLAST analysis (in the NCBI basic local alignment search tool) indicated that a sequenced 16S fragment belonged to Thunnus spp., we PCR-amplified a fragment of the mitochondrial control region (CR, approximately 236 bp) from the same DNA sample. CR amplification was conducted in a mixture containing 5 ng template DNA, 12.5 µL of 2× Taq PCR MasterMix (GN-PCR201-01, Genomix), and 12.5 µmol of each forward and reverse primer-Forward, Tuna-CR_F; Reverse, Tuna-CR_R1 or Tuna-CR_R2 [48]-made up to a final volume of 25 µL using distilled water. Thermal cycling began with one cycle at 95 • C for 4 min, followed by 35 cycles of denaturation at 95 • C for 30 s, 47-55 • C for 30 s, and 72 • C for 30 s and, finally, a single extension step at 72 • C for 7 min. PCR products were purified using a PCR DNA Fragment Extraction Kit (Geneaid, Taipei, Taiwan). The amplified mitochondrial fragments were subjected to Sanger sequencing, performed by Mission Biotech. (Taipei, Taiwan) using M13 sequencing primers. Primer sequences linked to the amplified fragments were trimmed before constructing the contigs using CodonCode Aligner. The mitochondrial sequences we generated in this study have not been submitted to GenBank as they do not come from voucher specimens.

Data Analysis
Edited 16S sequences were first aligned using ClustalW in MEGA11 [49], and then the haplotypes were determined in DnaSP 6. Species identity for each 16S haplotype was achieved by comparing them (by BLAST) to reference sequences in the NCBI GenBank database. Following the approach of both Armani et al. [50] and Horreo et al. [51], only matches displaying full sequence coverage and 100% similarity, and with unambiguous species-level scientific names, were considered positive fish identifications. If more than one fish species was shown as a positive match, all of them were considered potential candidates ( Table 2).
All CR sequences used in the study of Mitchell and Hellberg (2016) were downloaded to serve as reference sequences, and then our CR fragments and reference sequences were aligned in ClustalW. A neighbor-joining (NJ) analysis was then conducted based on Kimura two-parameter (K2P) distances and 1 × 10 3 bootstrapping replicates in MEGA11 [49]. According to the phylogenetic species concept [52], monophyly is a prerequisite for species recognition, so our specimens were authenticated based on the reference species with which they clustered and formed a monophyletic group (with high statistical support, i.e., bootstrapping value ≥ 70) in the NJ phenogram.

Comparison of Analytical Results and Product Labels
We compared the molecular identification of each sample to the ingredient list of the sampled cans. Since a Taiwanese government-approved standard for common names of fishes does not exist, the English names of labeled Chinese names were ascertained from the fish database of Taiwan (https://fishdb.sinica.edu.tw/ (accessed on 26 October 2021)). Although the Chinese symbols 鰹魚could broadly refer to any species from the genera Auxis (in Chinese: 花鰹屬), Euthynnus (in Chinese: 巴鰹屬), and Katsuwonus (in Chinese: 正 鰹屬), skipjack tuna is the only species generally termed 鰹魚 and that can be used legally in Taiwan to make canned tuna (Table 1). Therefore, we assumed that if 鰹魚 appeared on the ingredient list of a canned tuna product, it specifically represented skipjack tuna. Many of the imported products displayed labeling in Chinese and the language of source, but in those cases we exclusively relied on the Chinese label since Chinese is the only official language in Taiwan.
A sample was judged as displaying inconsistent labeling if the fish name in the ingredient list could not be linked unambiguously to a Thunnus species or skipjack tuna. It was then deemed mislabeled if the molecularly authenticated species it contained did not match the ingredient list on the label. Where a vernacular name used in an ingredient list refers to more than one species, a case of mislabeling was assigned when the molecularly authenticated species did not correspond to any fishes bearing that vernacular name. Finally, we determined a product as being true canned tuna if it contained Thunnus species or skipjack tuna.

Results
We observed inconsistent labeling in 11 of 65 canned tuna products destined for human consumption, but no such problem with cat food products. Inconsistent labeling reflected canned tuna products also claiming to be made from oriental bonito (Sarda orientalis) (in Chinese: 東方齒鰆) or displaying the ambiguous vernacular name 煙仔虎, which can refer to either skipjack tuna or oriental bonito (Figure 1, Table 2).

Results
We observed inconsistent labeling in 11 of 65 canned tuna products destined for human consumption, but no such problem with cat food products. Inconsistent labeling reflected canned tuna products also claiming to be made from oriental bonito (Sarda orientalis) (in Chinese: 東方齒鰆) or displaying the ambiguous vernacular name 煙仔虎, which can refer to either skipjack tuna or oriental bonito (Figure 1, Table 2). We successfully amplified the 16S fragment from all 90 samples, resulting in eight haplotypes ( Supplementary Information 2). All haplotypes could be identified to specieslevel by BLAST analysis, but only haplotypes Hap_A and Hap_E specifically relate to oriental bonito and skipjack tuna, respectively. More than one species was identified by BLAST analysis for the remaining six haplotypes, but based on the number of BLAST hits we assume Hap_B represents kawakawa (Euthynnus affinis) (in Chinese: 巴鰹), Hap_C is skipjack tuna, Hap_D and Hap_F are Thunnus species, Hap_H is either longtail tuna (T. tonggol) (in Chinese: 長腰鮪) or bigeye tuna (T. obesus), and Hap_G is either bullet tuna or frigate tuna (Auxis thazard) (in Chinese: 扁花鰹).
Our BLAST analysis of 16S sequences revealed that 31 of our samples contained Thunnus fishes. However, the success rate of CR amplification from those samples was We successfully amplified the 16S fragment from all 90 samples, resulting in eight haplotypes (Supplementary Information S2). All haplotypes could be identified to specieslevel by BLAST analysis, but only haplotypes Hap_A and Hap_E specifically relate to oriental bonito and skipjack tuna, respectively. More than one species was identified by BLAST analysis for the remaining six haplotypes, but based on the number of BLAST hits we assume Hap_B represents kawakawa (Euthynnus affinis) (in Chinese: 巴鰹), Hap_C is skipjack tuna, Hap_D and Hap_F are Thunnus species, Hap_H is either longtail tuna (T. tonggol) (in Chinese: 長腰鮪) or bigeye tuna (T. obesus), and Hap_G is either bullet tuna or frigate tuna (Auxis thazard) (in Chinese: 扁花鰹).
Our BLAST analysis of 16S sequences revealed that 31 of our samples contained Thunnus fishes. However, the success rate of CR amplification from those samples was quite low (5/31; 16%). The aligned CR dataset (Supplementary Information S3) for NJ analysis is 256 bp in length and contains 47 taxa, including 159 variable sites and 131 parsimonyinformative sites. The NJ analysis of CR sequences supports that samples T1, T7, and T56 are yellowfin tuna, and that sample T34 is longtail tuna, but we could not authenticate sample D2G based on its CR sequence ( Figure 2).
Excluding the canned cat food samples that were all accurately labeled, 25 of the remaining 65 canned tuna products were mislabeled and a further three were potential mislabeling cases. Our BLAST analysis confirmed that sample T36 contained Thunnus fish, but did not reveal which species. We found the labeling of sample T48 to be misleading. In Chinese, the symbol "鮪" (for tuna) is never associated with oriental bonito, so it is unreasonable for the symbol for oriental bonito to be placed in parentheses following "鮪 魚" (representing "tuna fish") on the label for this sample. We observed a similar issue for sample T4, since neither skipjack tuna nor oriental bonito can be regarded as a type of "鮪" (Thunnus spp.). Since the ingredient statement on the label fails to clearly indicate which species is contained in the can, it is difficult to judge whether or not these two samples are mislabeled. Notably, many of the products identified as exhibiting inconsistent labeling were also found to be mislabeled. The mislabeling rate of canned products for human consumption was 38% (25/65). Mislabeling was even more pervasive among cat food products, with a rate of 68% (17/25). The main reason for this high mislabeling rate of cat food products is that many claim to contain Thunnus fishes but are in actual fact made from skipjack tuna. The overall mislabeling rate for the 90 tested products of this study is~47% (42/90).

Discussion
According to Article 28 of the Act Governing Food Safety and Sanitation in Taiwan, public labeling, promotion and advertisement of foods or food additives, cleansers, utensils, containers or packaging designated by the central competent authority shall not be false, exaggerated or misleading. The 11 cases of inconsistent labeling we identified among our 90 samples, which display "鮪" (for tuna) on their labels but also list scombrids other than Thunnus species or skipjack tuna as an ingredient, obviously mislead customers   Of the 65 human food products we tested, 37 (57%) legitimately contained either Thunnus fishes or skipjack tuna, and 20 out of 25 cat food products are true canned tuna. Overall, the proportion of true canned tuna products is about 63.33% (57/90) in this study.

Discussion
According to Article 28 of the Act Governing Food Safety and Sanitation in Taiwan, public labeling, promotion and advertisement of foods or food additives, cleansers, utensils, containers or packaging designated by the central competent authority shall not be false, exaggerated or misleading. The 11 cases of inconsistent labeling we identified among our 90 samples, which display "鮪" (for tuna) on their labels but also list scombrids other than Thunnus species or skipjack tuna as an ingredient, obviously mislead customers into believing these products contain true canned tuna. For this study, we solely relied on the information on Chinese labeling, but we also noted conflicting information between Chinese and source-language labeling of imported products. For example, the ingredient statement in Japanese of sample T42 clearly declares that it is made from albacore tuna (T. alalunga), but its Chinese label only states that it contains Thunnus fishes (in Chinese: 鮪 魚). Similarly, the Japanese label of sample T59 indicates skipjack tuna as an ingredient (in Japanese: かつお), but its Chinese label specifies yellowfin tuna (in Chinese: 黃鰭鮪) ( Table 2). Such conflicting labeling of imported products not only confuses consumers but may also circumvent legal controls.
The "one species-one name" policy is critical to the authentication of fishery products [38]. Clearly, usage of scientific names could enable investigators to easily judge if a product is mislabeled. Under European Union labeling regulations, including the species' scientific name on fishery product labels is mandatory [53]. However, scientific names are not required on Taiwanese fishery products nor are such names familiar to the public. Xiong et al. [54] and Chang et al. [22] advocated the adoption of the Chinese-Latin Dictionary of Fish Names (https://fishdb.sinica.edu.tw/chi/chinesequer1.php (accessed on 26 October 2021)) as a standard list of fishes in Chinese corresponding to scientific nomenclature. This Dictionary indeed clarifies that the Chinese symbols 東方齒鰆 (in English: oriental bonito) correspond to Sarda orientalis, but it does not include other Chinese vernacular names. Thus, any official "one species-one name" standard should not only contain scientific and Chinese names, but also incorporate vernacular names.
Notably, our success rates for amplifying the two mitochondrial DNA fragments differed considerably-100% for 16S, but only 16% for CR. The canning process is known to damage DNA molecules, with Quinteiro et al. [43] documenting that most DNA segments extracted from canned tuna are <100 bp in length. Therefore, it is not surprising that amplification of the 85-bp 16S region was more successful than the 236-bp CR fragment (Binominal Generalized Logical Model, p < 0.01).
Apart from haplotypes Hap_A and Hap_C, a single species was not identified by BLAST for the other 16S haplotypes. There are a number of possible reasons for that outcome. First, DNA degradation through the canning process limits molecular authentication based on longer sequences, such as via conventional DNA barcoding. Accordingly, shorter DNA segments must be targeted, but they contain less information and so are less likely to unambiguously assign a specific species [55]. Second, molecular identification based on mitochondrial sequences is very sensitive to gene flow and incomplete lineage sorting [25,56,57]. The tribe Thunnus comprises very closely related species, some of which display genetic introgression [58,59]. Consequently, though Hap_D, Hap_E, and Hap_H are all clearly form Thunnus fishes, their exact species identity remains unclear. Though conventional DNA barcoding can distinguish Thunnus fishes [60][61][62], it would be problematic to amplify the~650 bp barcode from the degraded DNA of canned samples. Third, a reliable database is crucial to accurate DNA-based identification [63]. GenBank does not guarantee that deposited sequences display correct species names. For example, the BLAST result for Hap_C matches multiple sequences for skipjack tuna sequences and one for yellowfin tuna (GenBank accession number: KM055376), implying that accession KM055376 is very likely misidentified. Hence, as highlighted in a number of studies [22,64,65], a reliable and complete DNA reference database for authenticating seafood resources is sorely needed.
In this study, we found that many canned tuna products in Taiwan are made from oriental bonito, kawakawa, bullet tuna, or frigate tuna instead of legitimate Thunnus fishes or skipjack tuna. Although oriental bonito was never found in canned cat food products, the other three substituted fishes were identified in both human and cat food samples. These same four species have also been reported as illegitimate tuna substitutes in other studies [44,46,48,50,66,67]. Though istiophorid fishes have been reported as mislabeled Thunnus products in other studies [46,68], we did not detect them in this study.
Our NJ analysis of CR sequences, including five haplotypes generated in this study, further revealed that both yellowfin tuna and longtail tuna are used in canned tuna products. Yellowfin tuna is one of the commonest canned tunas [5,69], so it is not surprising that three of our five CR haplotypes clustered with yellowfin tuna sequences (sample T34 was identified as longtail tuna, and sample D2G could not be identified to species level). Our difficulties with amplifying the CR region mean that the specific Thunnus composition of canned tuna products in Taiwan remains unclear. Identifying canned tuna products to species level is important because certain Thunnus fishes have higher mercury levels [70], posing a human health risk. Therefore, mitochondrial regions other than CR, such as ATP synthase membrane subunit 8 (ATP8), ATP6, and COIII could be considered [71], or smaller CR fragments could be targeted.
We observed the terms 白身鮪魚 or 鮪魚白肉 commonly in the ingredient statements of our cat food samples (Table 2), reflecting the high mislabeling rate (17/25) among cat food products. However, most of the cat food samples (20/25) still represented true canned tuna, albeit not the species that might be expected. To date, there is no official definition for either of these two Chinese terms. They may be translated as "light tuna", which often refers to yellowfin tuna or skipjack tuna, but could actually be any fishes mentioned in the Code of Federal Regulation Title 21 (CFR 161.190) and with flesh color in the Munsell color system ≥5.3 [48]. If those terms were to be officially recognized as translations of light tuna, then the mislabeling rate of cat food we report herein would be much lower (down to 8/25) (Binominal Generalized Logical Model, p < 0.01). Accordingly, we implore the responsible authorities to clearly define the terms for use in canned product labeling.
Although we found that 63.33% of our samples are true canned tuna, this outcome may not reflect the actual adulteration level of canned tuna products in Taiwan. First, we selected only one small piece of tissue from each can, but the mixing of tuna species in cans has been found in the European market [4]. An assessment of how prevalent the mixing of tuna species is in cans in the Taiwanese market would be needed to determine how close our calculated adulteration level is to the real scenario. Second, we solely sampled major brands, so there are some that remain to be assessed, especially of cat food. Moreover, seasonality in scombrid catch may alter the species composition of adulterated canned tuna products. Thus, more comprehensive and long-term monitoring of the species composition of canned tuna products is needed.

Conclusions
We report an overall mislabeling rate of 46.67% among the 90 samples of this study, with 63.33% of sampled canned products being true canned tuna legitimately made from Thunnus fish or skipjack tuna. In many cases, the labels of the sampled canned tuna products would confuse customers as to what species they contain. Either they contain species such as oriental bonito that do not conform to Taiwanese legislation, or ill-defined terms such as 白身鮪魚 or Chinese vernacular names are used. We assert that a standard list of scientific names and their corresponding Chinese and vernacular names conforming to the "one species-one name" principle, as well as clear definitions of terms for use in canned tuna labeling, is crucial to tackling fish product adulteration. We found that 37% of investigated canned tuna products contain illegitimate species. In many cases, the manufacturers have substituted so-called "pseudo-tunas", such as oriental bonito, kawakawa, bullet tuna, and frigate tuna, for legal species, i.e., Thunnus species and skipjack tuna. Our study demonstrates that a pair of primers targeting a short segment (85 bp) of 16S performs well in amplifying DNA extracted from canned food samples. However, the limited information content provided by this short sequence hampered molecular identification to species level, especially given the close phylogenetic relationships and potential for gene flow among Thunnus species. Moreover, the CR fragment we targeted largely proved uninformative, likely owing to the extreme DNA fragmentation caused by high heat treatment during the canning process. Our previous study of seafood adulteration in conveyor-belt sushi restaurants revealed no case of tuna fraud in such establishments [22], so such adulteration appears to be more common in canned products. A large-scale and long-term monitoring study would help fully establish the extent of canned tuna fraud in the Taiwanese market.

Data Availability Statement:
The data presented in this study are available on request from the first author.