DNA barcode reveals occurrence of threatened species and hidden diversity on Teleost fish trade in the Coastal Amazon

This study aimed to identify the teleost fish species sold in Bragança, a major fishing hub on the north coast of Brazil. The COI gene analysis was performed for the identification of fish species. The local market uses common names that are not accurate and do not reflect the diversity of the species. 204 sequences were obtained, with 119 haplotypes. 83 species were identified by comparing with public databases and constructing phylogenetic trees, with Carangidae being the most prevalent family. The study also found Haemulon atlanticus, Menticirrhus cuiaranensis and Hoplias misioneira, a newly described species from the Amazon basin, among the samples. Additionally, 73 commercial names were recorded, including 10 categories, and the illegal trade of Epinephelus itajara was detected. The DNA Barcode method proved to be effective for discriminating the species. The study highlights that common and commercial names are vague and underestimate the fish diversity, and that Brazil needs to revise its regulations for commercial and scientific names.


Teleost diversity traded using DNA barcoding
We analyzed 500 base pairs of DNA from 204 fish samples and found 258 polymorphic sites.The final alignment did not contain any deletions, insertions or stop codons.We obtained 119 haplotypes and compared their sequences to public databases.Table 1 shows the haplotypes, their molecular identification based on genetic similarity, and the commercial names of the fish samples.The sequences are publicly available, codes OR459502-OR459617 and OR515260-OR515262.
We identified 82 species of teleost fish, belonging to 15 orders, 31 families and 58 genera, from the 73 commercial names previously recorded.We used both genetic similarity and phylogeny for molecular identification (Fig. 1).Many taxa did not match the commercial name assigned to them.The most diverse families were: Carangidae (12 species), Sciaenidae (10 species) and Ariidae (9 species) (Supplementary Table S1).
Most of the recorded species (68) were from marine and/or estuarine habitats (Fig. 2), while 14 were from freshwater (Fig. 3).We also report the first record of commercialization of Haemulon atlanticus (formerly known as H. steindachneri) (Family Haemulidae), Menticirrhus cuiaranensis (Sciaenidae) and Hoplias misioneira (Erythrinidae) species on the north coast of Brazil.

Molecular identification
Using DNA Barcoding, we discriminated 82 species of fish from the Amazon region.We compared their genetic sequences with public databases and found 81 matches at the species level.The exceptions were Aspistor quadriscutis and Batrachoides surinamensis, which we identified by morphology and phylogenetic analysis.These two species had no reference sequences in public databases, so our sequences will serve as the first references for them (Table 1; Figs. 2, 3).
We also identified eight species by similarity only using the BOLD platform, but we did not include their sequences in our analyses because they came from a private source.They were Trachinotus cayennensis, Cetengraulis edentulus, Hoplerythrinus unitaeniatus, Astyanax bimaculatus, Notarius grandicassis, Sciades parkeri, Sciades proops and Sciades herzbergii.
Some samples had ambiguous identification results, such as the ones labeled as "cangatã", which matched Aspistor luniscutis by similarity and A. quadriscutis by morphology.Other samples had high similarity with more than one species, such as "caica" 01 with Mugil curema and Mugil rubrioculus, "caica" 02 with Mugil hospes and Mugil brevirostris, "caica" 03 with Mugil curema and Mugil trichodon; the "gurijuba" sample with Netuma sp. and S. parkeri, "urubaiana" with Elops smithi and Elops saurus (Table 1).Another controversial case was found for the sequences of "bragalhão", "bagre" and "uricica branca", which the comparisons returned them ash Sciades couma, based on different sequences (ITAPE024 and ITAPE351).The clusters resulting from the NJ tree showed that "bragalhão" and the ITAPE024 sequence form a cluster and that the "bagre" and "uricica branca" form another clustering with the ITAPE351 sequence (Fig. 2).The two groups differ from each other with a divergence of 5.20%.
Mean genetic distances increased according to taxonomic level, with average of 0.13% within species, 11.55% within genera (between species) and 18.28% within families (between genera) respectively (Table 2).Intraspecific values ranged from 0.0% to 1.42%.The species showed barcode gaps, with a minimum distance between congeners of 4.16%, between S. couma and S. props (Table 2).1).Among the designations, 10 were considered categories (Fig. 4), as they presented more than one species being sold by the same name.
As an example, we have the "pampo" category, which had the highest number of species (n = 5), including Chloroscombrus chrysurus, Hemicaranx amblyrhynchus, Trachinotus carolinus, Trachinotus goodei and Peprilus crenulatus (Fig. 4).Beside from this, 10 cases in which different trade names were used for the same species were also observed, such as Cynoscion acoupa, which has been sold as "pescada amarela", "pescada branca" and "garoupa" (Table 1).On the other hand, the species Haemulon parra, H. atlanticus and M. cuiaranensis were found being sold without presenting a commercial name, therefore they were called "without commercial designation" (SDC).
The commercial designations, when compared with the correlation of common names and respective scientific names provided by Normative Instruction No. 53 of September 1, 2020 (MAPA), showed that there is compatibility for many of the identified species (47.6%), however, for other species, the names are different (23.4%), in addition to species found being commercialized, but which do not have a name on the MAPA list (28%) (Supplementary Table S1).

Discussion
This work represents the most comprehensive molecular analysis with Teleost fish traded in the coastal Amazon, with more than 200 individuals collected over four years of study.DNA barcoding tool was used to identify and www.nature.com/scientificreports/validated the real commercialized diversity, masked due to the use of categories, and revealed an important trade of threatened species, in addition to species that its commercialization was first recorded in this study.

DNA barcoding for ichthyodiversity identification
The DNA Barcoding tool was used to identify the fish diversity traded in Braganca.Comparisons were made with public database and phylogenetic trees, comprising 82 fish species, corresponding to the highest ichthyodiversity recorded to date, which was higher than those found by Braga et al. 9 and Freire et al. 29 and like the ones found by Martins et al. 12 , when considering only teleost fish.The identifications carried out in previous studies were based on vernacular nomenclatures and taxonomic keys, while in this study, we identified a large number of species commercialized in Bragança and in the North region through the DNA Barcoding approach, confirming the efficiency of the molecular tool to discriminate taxa, as observed in other studies with ichthyodiversity 24,27 , as well as to identification of processed products by the fishing industry 28 .Within the identified species, the family Carangidae was the most representative with 12 species, contrary with previous studies that positioned the Sciaenidae family as the most representative 9,12,29 .Carangidae is constantly identified as one of the main families that composes the ichthyofauna of the Brazilian north coast 3,30 .
The entry of large number of Carangidae family species into local trade can be attributed to the emergence of a new market, during the closed season for "pargo" (L.purpureus) and other species of greater commercial value from the north coast of Brazil where the vessels are licensed for various fish species and many carangids, popularly known as 'black fish' , are caught (personal communication).Another important fact is that this study is the first implementing the molecular approach on the diversity of teleost fish commercialized in Bragança, when compared to previous research that only used taxonomic keys 9,12,29 , leading the authors not to reach the identity of the evaluated species, thereby underestimating the group of Carangidae species sold.
In this study, the commercialization of H. misioneira in the North of Brazil was also observed for the first time.This species was described from the Hoplias malabaricus species complex, in the Uruguay, Paraguay and Paraná basins 31 and according to Guimaraes et al. 24 , this species has a disjunct distribution, also occurring in the Amazon Basin.This is the second record outside its natural range, which shows that this species is probably distributed in other areas, since the H. malabaricus complex has a wide distribution 32 .
In addition, we recorded the trade of newly described species already being sold, and without having a commercial name, such as M. cuiaranensis, and H. atlanticus.This scenarios shows how the diversity of fish in the coastal Amazon is underestimated and misunderstood, since the capture and commercialization of this species was already happening even before we were aware of its presence.This is worrying, because while part of the biodiversity remains unknown, natural resources are being exploited at an increasingly accelerated pace 5 .Probably, when it comes to fish, many taxa can be extinct even before being formally described due to the intense dynamics of capture and commercialization, with diverse and non-standard nomenclatures, associated with inefficient and/or non-existent inspection which strongly collaborate to reduce the biodiversity.

Inconsistencies between morphological and molecular identifications
Ambiguities in the identifications were found for the "cangatã" fish, which molecular identification presented them as A. luniscutis, however, the species that is found in the Brazilian north region is A. quadriscutis.In addition, a study with a molecular and morphological approach showed that Aspistor species found on the Brazilian coast have morphological differences, but do not present significant genetic distances in mitochondrial genes such as cytochrome b (Cyt b) and subunits 8 and 6 of ATP synthase (ATPase 8/6) 33 .It may be that the same happens for the COI gene, causing the samples of A. quadriscutis to show great similarity with A. luniscutis and forming a clade in the phylogenetic tree (NJ).
For some species of the Ariidae family, the inconsistencies in the identifications probably occurred due to identification errors and consequent erroneous deposits in public database, as observed for the designations "bragalhão", "bagre" and "uricica branca", identified as S. couma, but which formed two distinct groups with a genetic distance of 5.20%, in the NJ tree, being "bragalhão" (clade 1), bagre" and "uricica branca" (clade 2), that was identified based on morphology as S. couma and S. herzbergii respectively.
Comparisons with public database showed ambiguity in identification.Cases like these were observed for the Mugilidae family, the species referring to "caica" 02, was 100% similar to M. hospes and M. brevirostris in the NCBI and BOLD.However, in the South Atlantic only M. brevirostris occurs 34 , therefore, the sequences deposited in Brazil as M. hospes are, considered M. brevirostris 35 .For the other members of the Mugilidae family, the identification was confirmed from the study conducted by Durand et al. 35 , where a new identification of sequences from the Mugilidae family deposited in Genbank was carried out, correcting erroneous deposits, therefore we identified "caica" 01 as M. rubrioculus, and "caica" 03 as M. curema.
Another case of incongruity was observed for the "urubaiana" fish, identified as Elops smithi and Elops saurus in public banks.However, Sousa et al. revealed the occurrence of only E. smithi on the Brazilian coast.Thus, the sequence assigned to E. saurus in Brazil is possibly a taxonomic error, which has already been reported for these two species in the literature 36,37 .
The samples considered to be H. atlanticus were identified in public database as H. steindachneri, before the description of H. atlanticus for the Western Atlantic, both belonging to the H. steindachneri complex 38 .Despite the study by Carvalho et al. 38 used the genetic tool to confirm the existence of the two species, these sequences could not be used in this work, as they are not available in public databases.
Records of incompatibilities due to inaccurate or erroneous deposits in public database have been reported in the literature for both the BOLD system 27 and the NCBI 34 .Faced with these shortcomings, researchers must be carefully when carrying out identifications by consulting specialized literature or specialists in each group or using reference database to resolve ambiguous cases, so that identification errors are mitigated and not perpetuated and reliability in the data deposited in public database is maintained.
Despite some ambiguities in species identifications, the DNA Barcoding tool was efficient in discriminating most of the taxa found in this study, with expressive Barcode gap.We recovered as the greatest intraspecific distance, (1.42%), found for the species Mylossoma duriventre, and the smallest interspecific distance (4.16%) between the species Sciades couma and Sciades proops.

Commercial name and hidden diversity
82 species were found out of 73 trade names sampled, showing that there is no correspondence between the number of trade names and the number of traded species, since in some cases the designations act as a category and in others the same species receives different trade names.Marketing by category, as in Bragança and as it happens in most places, ends up with underestimating the fish diversity offered, mainly due to the difficulty of differentiating the taxa of some families with similar morphology, as observed for Centropomidae and its congeners sold as "camurim", the C. undecimalis, C. parallelus and C. ensiferus; and Mugilidae sold as "caica", M. rubrioculus, M. brevirostris, M. curema, and M. incilis.
Marketing through generalist names can pose a threat to fish conservation, as several species can be sold through categories, including endangered species 14,16,17 , as the case of E. itajara sold under the designation/ category "garoupa".It is important to note that the commercialization of E. itajara in Brazil has been prohibited since 2002 40 , therefore, the commercialization of this species is taking place illegally.The trade of endangered species is worrying, basically when it is facilitated by the non-standardization of the commercial nomenclature that masks this market.
One case of substitution occurred for fish sold under the name "pescada branca", which according to normative instruction MAPA Nº 53 of 2020, should only be used for the species Cynoscion leiarchus and Plagioscion squamosissimus.However, all samples collected under this designation were identified as Cynoscion acoupa, a species normally sold as "pescada amarela" and which is of great commercial importance 7,12 .This replacement probably occurred accidentally, since the individuals collected were juveniles and many species of the Sciaenidae family are morphologically similar in the early stages and have a sympatric distribution, which can lead to difficulties in the correct identification of taxa, as already reported in other works 18 .
Although the MAPA normative tries to establish and standardize the relationship between the common names and respective scientific names for the main commercialized species, it still has redundancies, as it provides several common names for a single species and in some cases displays nomenclatures for fish down to the genus level, such as "canguiro" and "pampo" for Trachinotus sp., opening space for permanence of categories.When comparing the Feira Livre designations with Normative Instruction No. 53 of September 1, 2020, we noticed that a range of taxa does not have a similar name in the normative instruction, as well as several species are not present in the list.This reveals that we have a document that needs to be revised to establish in a coherent and specific way the trade name and corresponding species in Brazil.The alternative to reduce the gaps left by categorization is the creation of lists by region, since the nomenclatures vary a lot, even in nearby places.

The importance of knowing the diversity of fish in the trade
Trade in Bragança is predominantly carried out with marine species, but some freshwater species are also sold, including fish from fish farming such as "tilápia" Oreochromis niloticus and "Tambaqui" Colossoma macropomum.
The results reveal that trade in Bragança is quite dynamic, with changes in the composition of species offered over the year (Table 3).Certainly, there are species that were not sampled, as the landings and commercialization of fish in Bragança occur daily 6 , however, this study presents the most complete data regarding the diversity of teleost fish commercialized in the Bragança region (Table 3).Our results shows the commercialization of species that were hidden by popular nomenclature and imprecise taxonomic identification.These data raise an alert about the capture and sale of species that already have low stocks, allowing the competent authorities to manage and supervise this market.
The measures for conservation and fisheries management be effective, it is first necessary to know the really diversity.We present here a list (Supplementary Table S1) with the correspondence between the commercial and biological designation for the species commercialized in Bragança, Amazonian coastal region, the first obtained through molecular identification and which will be an important tool for ordering the commercialization of fish in the region, considering all fish collected and molecular identifications carried out.

Final considerations
In the present research, the DNA Barcoding tool proved to be extremely efficient for the discrimination and correct identification of the species sampled in Bragança.The results showed cases of replacement, trade of endangered species and unrecorded species diversity.Our results confirm that common and commercial names are inaccurate which underestimate ichthyodiversity and may favour replacements and trade of endangered species.Although we have a regulation to establish the relationship of commercial and specific names, it is incomplete, inefficient and needs a reformulation, which considers the diversity of names and the different Brazilian regions, to propose a standard name for each species.We therefore present a list of correspondence between trade name and referent species, considering the trade of Teleosts in the coastal portion of the Brazilian Amazon (Supplementary Table S1).

Ethics statement
All individuals were obtained from points of sale, they were already dead.There was no need to apply the guidelines of the Institution's Ethics Committee.In the same way, it was not necessary to obtain a collection license, since individuals were purchased during the commercialization process, or donated by traders.Table 3. List of fish families and species sold at the Bragança Free Market over the years.With species collection location, "Mercado" (M) and/or "Feirinha" (F), biennium of sample collection 2016/2017 and/ or 2018/2019 and comparison of the species in the present study with previous work carried out in the same locations: Freire et al. 29 and Martins et al. 12 .

Figure 1 .
Figure 1.Neighbor joining tree (NJ) for the 119 haplotypes with reference bank, NCBI, and BOLD sequences.The numbers over the branches indicate Bootstrap's statistical support.The coloring of taxa was organized by family and '*' indicates threatened species.

Figure 2 .
Figure 2. Bayesian inference tree for commercialized marine fish species at the Street Market of Bragança-PA.The numbers over the branches indicate statistical support.The coloring of taxa was organized by Family.

Figure 3 .
Figure 3. Bayesian inference tree for commercialized freshwater fish species at the Street Market of Bragança-PA.The numbers over the branches indicate statistical support.The coloring of taxa was organized by Family.

Figure 4 .
Figure 4. Alluvial diagram representing the 10 commercialization categories and the corresponding species.On the left side are the categories and on the right are the species.

Table 1 .
Genetic similarity results of the 119 haplotypes of the 82 fish species traded on the Amazon coast compared to sequences from public databases.In parentheses, the number of individuals sharing each haplotype.

Table 2 .
Minimum, maximum, and average values of genetic divergence, using the K2P evolutionary model, among the sampled species, genera, and families.