DNA-based identification reveals illegal trade of threatened shark species in a global elasmobranch conservation hotspot

Here, we report trading of endangered shark species in a world hotspot for elasmobranch conservation in Brazil. Data on shark fisheries are scarce in Brazil, although the northern and northeastern regions have the highest indices of shark bycatch. Harvest is made primarily with processed carcasses lacking head and fins, which hampers reliable species identification and law enforcement on illegal catches. We used partial sequences of two mitochondrial genes (COI and/or NADH2) to identify 17 shark species from 427 samples being harvested and marketed on the northern coast of Brazil. Nine species (53%) are listed under some extinction threat category according to Brazilian law and international authorities (IUCN – International Union for Conservation of Nature; CITES – Convention on International Trade of Endangered Species of Wild Fauna and Flora). The number increases to 13 (76%) if we also consider the Near Threatened category. Hammerhead sharks are under threat worldwide, and composed 18.7% of samples, with Sphyrna mokarran being the fourth most common species among samples. As illegal trade of threatened shark species is a worldwide conservation problem, molecular identification of processed meat or specimens lacking diagnostic body parts is a highly effective tool for species identification and law enforcement.

Rhizoprionodon porosus and Carcharhinus acronotus were the most abundant species, contributing 33.10% and 15.88% of all samples, respectively (Table 1). Nine species are included at least in one of the categories of extinction threat ( Table 1). None of the species was recorded at all localities, and only one (R. lalandii) was landed in a unique site (Raposa). Twelve species were identified among samples from Bragança, eleven from Tutoia, nine in both Belém and Raposa, six from Amapá state's coast, three from Carutapera, and two from Vigia. One sample from Tutoia could not be identified to the species level because of a 98% similarity to both Squalus brevirostris and Squalus megalops on NADH2. However, whatever the species, this is the first record of Squalus from Brazil's North Coast.  (Table 2). Average interspecific K2P divergence was 11.3%, with pairwise differences ranging from 3.7% for the two Rhizoprionodon species to 23.2% between Squalus sp. and G. cirratum (Table 2).

Discussion
Given the lack of morphological traits for species identification of processed shark samples, we used molecular markers to assess the species composition of the sharks landed and traded on Brazil's North Coast. DNA-based identification revealed seventeen species, nine (53%) of which are listed in some category of threat (IUCN, CITES or Ordinance 445). If the species under the Near Threatened category are also considered, the number increases to 13 (76%). We also provide the first record of the family Squalidae for the BNC. The present study is so far the largest one focusing exclusively on the shark trade monitoring using molecular techniques in Brazil.
The trade of endangered shark species is a worldwide phenomenon 20,23,35 . Results herein were not surprising, given Brazil's status as one of the seven leading producers and consumers of shark meat 2,4,5 , and the number of vulnerable or endangered endemic species found in the country's coast 13,19 , which now corresponds to 33% 11 . However, the presence of so many species at risk of extinction in one of the most important fishing zones and elasmobranch conservation hotspots is a major concern both for conservation and the long-term sustainability of local fisheries. In Guyana's coast, the western part of the elasmobranch conservation hotspot where BNC is classified 18 , thirteen shark species were identified using COI 36  Maranhão state's coast is not included in the hotspot zone. Nevertheless, here we show a significant number of threatened species and specimens being landed and commercialized in the state, mainly in Tutoia (area of the Parnaíba river delta). Thus, we recommend that the 640 km coastline of Maranhão should be included in the hotspot 3 proposed by Dulvy et al. 18 as part of this major elasmobranch conservation area.
We provide evidence of an ongoing trade of threatened shark species on Brazil's North Coast, thus indicating the practice of environmental crimes by local fisheries. Previous studies using molecular data show that the Brazilian shark trade involves a smaller diversity than that recorded here, including endangered species such as Sphyrna lewini 26,[30][31][32]39,40 . Based on these findings, we reinforce the need to use molecular identification tools for monitoring shark fisheries and law enforcement, especially when accurate morphological identification is difficult or even impossible.
Forensic genetic studies have been contributing to the more effective monitoring of the fish trade worldwide 20,37 . Most are based on Hebert et al. 34 proposal of using a single molecular marker (COI) for animal species identification, among other reasons, due to the ongoing decline in the numbers of taxonomists worldwide. In fact, COI is an effective tool for shark species identification 41 , because, as in the case of the BNC sharks, it exhibits interspecific genetic distances above the 2% threshold used as criterion for species delimitation. However, other markers that evolve at different rates may be important to resolve some taxonomic uncertainties that COI is unable to deal with 37,42 , as we show here for NADH2-based species identification.
The NADH2 marker has also been used to identify elasmobranch species 37 . It was decisive for the identification of some species in our samples, such as Mustelus higmani which does not have COI sequences deposited in any public database. Furthermore, NADH2 was more effective to identify Rhizoprionodon species, because COI sequences had 100% match with both R. porosus and R. terranovae, while NADH2 had 100% match with R.  porosus and 97% with R. terranovae. Despite this, both markers produced reliable species delimitation in the NJ trees (Figs 1 and 2). That raises the suspicion that some sequences from this genus deposited in the public databases might belong to misidentified specimens, as suggested by Kolmann et al. 36 , which reinforces the importance of using multiple markers for species identification. Studies carried out in the BNC between the 1980s and the late 2000s demonstrate that C. porosus used to be the most commonly harvested species 26,43 . Furthermore, Sphyrna tiburo, R. porosus, R. lalandii, S. lewini, and Isogomphodon oxyrhynchus used to be the following most abundant elasmobranch fishing resources in Maranhão state, in that order 43 . Since then, C. porosus, S. tiburo, S. lewini, and I. oxyrhynchus have been included in Brazil's endangered species list due to severe population decline throughout their ranges. The most evident case is of I. oxyrhynchus, which is considered to be on the brink of extinction due to overfishing 44 .
Some studies reveal the impact that fisheries have in changing species abundances in other regions of the globe 45,46 . Thus, we argue that a shift in species abundance has occurred in the BNC, based on our results, the scientific studies carried out in the BNC in the last 30 years, and the current species conservation statuses. R. porosus is now the most abundant species, while the once previously most common one (C. porosus) has suffered a considerable catch reduction, possibly due to population declines. Nevertheless, further fisheries independent studies are required to test if this pattern reflects changes in the species abundance in the assemblage and if the major cause for these declines is indeed overfishing.
As the most species rich family among sharks, the requiem sharks (family Carcharhinidae) correspond to the majority of the shark trade worldwide 20 and are considered a priority group for conservation due to its common harvest in unregulated and unreported fisheries 18 . Carcharhinidae species compiled the vast majority of the samples identified here (418 specimens, ~74%), but neither R. porosus (33.1%) nor C. acronotus (15.88%), the most abundant species, are threatened with extinction. Nevertheless, C. acronotus is classified as Near Threatened (Table 1), and its populations are known to be declining worldwide 47 . A three-fold smaller proportion (5.8% of the identified samples) of C. acronotus is reported in Guyana's coast 36 . This species reproduces biennially with females reaching sexual maturity at around 4.5 years, and a mean of 3.5 offspring per gestation 48 , resulting in low rates of intrinsic population growth 49 . Taking these into account, the relatively large number of C. acronotus specimens (n = 68) identified in the BNC may be misleading. Further population analyses, including assessments of intra-population genetic diversity, are important to establish its current conservation status and to develop effective conservation measures for the species at the BNC.
The relatively high abundance of R. porosus may reflect its more favorable life history traits, especially its continuous reproduction 49,50 , which may guarantee a recruitment rate higher than fishing pressure. Nonetheless, there are no demographic data on the species in the area, which impedes a reliable assessment of the impacts of an increasing fishing pressure on its populations. In Guyana's coast, Rhizoprionodon species were also the most abundant sharks caught by artisanal fisheries 36 . However, R. lalandii (18.9%) was the most common species, while R. porosus composed only 3.8% of the samples. Differently, we found an extremely low number of R. lalandii specimens in the BNC samples (0.23%), which is puzzling considering that R. porosus and R. lalandii were both abundant in Maranhão state during the 1990s 51 . There is no reason to assume that fishing gear used by local fleets would target R. porosus selectively over R. lalandii. This is the first study to confirm the occurrence of R. lalandii in the BNC based on genetic evidence, but there are no data on the local habitat use or migration patterns of either species. This reinforces the need of fine scale population data for a better assessment of the conservation status of R. lalandii.
The smalltail shark, Carcharhinus porosus, was the third most abundant species in BNC (9.81%), although it was rarer in comparison with other studies in both Maranhão state and Guyana's coast 36,43 . In fact, it was the most abundant shark species harvested in the area during the 1990s, representing 43% of the total catch, whereas R. porosus was only the third (10% of shark catches) 51 . Carcharhinus porosus was also the most abundant shark species fished in Pará State in the early 2000s 26 and the second most caught species in Guyana in 2015 36 . An 85% decrease of the total biomass landed in 2004 led to its classification as critically endangered in Brazil since 2014 and its harvest has been illegal ever since 16,52 . Nevertheless, the species is considered to be data deficient by IUCN, due to the lack of information from most of its range, although BNC is considered its center of abundance 52 . Data on habitat use and population genetic diversity will be essential for the development of effective conservation measures.
Other requiem sharks (C. leucas, C. falciformis, G. cuvier and C. limbatus) also identified here are all classified as Near Threatened although C. falciformis is included in Appendix II of CITES. These species comprised 11.4% of the samples, whereas they represented 16.7% of the Guyana shark fisheries 36 . Both studies had a small proportion of these large bodied species in the samples. Despite regulations, C. falciformis is one of most fished sharks worldwide, primarily as bycatch in tuna fisheries 5 . In the southern Atlantic Ocean, its population is suggested to be in decline 10 . While its harvest is not prohibited in Brazil, there are no data on catches in the BNC, and its capture by different types of fishery may be an indication of overfishing. The regular monitoring of catches in both cases will be necessary to evaluate the conservation status of this species.
One interesting finding when using both COI and NADH2 sequences was the placement of I. oxyrhynchus in the Carcharhinus clade together with C. acronotus, C. porosus, C. limbatus, and C. leucas (Figs 1 and 2), which is consistent with previous studies 26,42 . Given this, we suggest that a thorough taxonomic and phylogenetic revision of the genus Carcharhinus is performed, including I. oxyrhynchus, to test its monophyly. Perhaps this is the species that demands more immediate conservation actions since it is Critically Endangered worldwide 44,53 . Our records indicate the capture of a surprising number of specimens (n = 14) for such a depleted species. However, it was not recorded among 132 samples from Guyana 36 . This species is endemic to South America's North Coast, thus presenting a highly restricted range 44 , and is considered to be on the brink of extinction due to its low resilience in facing the slightest fishing pressure 44 . Further studies of genetic diversity are urgently required, together with the monitoring of landings to evaluate the status of the remaining populations. Over the short term, however, harvesting should be suspended, and law enforcement strengthened by Brazilian environmental authorities and other countries included in the conservation hotspot area.
Hammerhead sharks (genus Sphyrna) represented almost a fifth of all samples (80 specimens, ~19%). An almost two-fold higher composition of Sphyrna is reported for Guyana (~37% of the identified samples), with S. lewini being the most abundant hammerhead shark in the area 36 . In the BNC, S. mokarran was the fourth most common species of all samples (9.36%) and the most common from its genus. These sharks are in extreme extinction threat worldwide, and most species are listed as Critically Endangered in Brazil. While they are not targeted specifically by fishermen, harvesting is a common practice at the BNC 39,54 . Their morphology, in particular the large body and the laterally expanded head, facilitate its capture by nets 55 . In addition, Sphyrnids are highly vulnerable to stress, and often die after capture, even if they are returned to the water alive 55 . As most of the region's fisheries are based on trawls and gill-netting, there is considerable potential for increasingly bycatch levels for hammerheads. Our data combined with those from Kolmann et al. 36 regarding hammerhead shark fishing and trading grounds suggests more effective inspections and severe law enforcement at landing and trading sites are urgent in both the BNC and Guyana to ensure the protection of such a remarkable species.
Together, other species recorded (Mustelus canis, Mustelus higmani, Squalus sp., and Ginglymostoma cirratum) corresponded to over 7% of samples. We report the genus Mustelus on Brazil's North Coast for the first time using molecular data. In particular, M. higmani is an elusive species, for which few data are available. Pregnant females were recorded in the samples as bycatch of shrimp trawl fisheries. M. higmani seems to reproduce continually and thus overfishing does not seem to be a great concern for its conservation 56 , even though it composed around 40% of the shark landings in Venezuela 57 . However, the lack of data on its actual distribution in the world, abundance patterns, and the impact of trawl fisheries in its populations are important factors that need to be addressed to provide a robust conservation status conclusion.
Harvesting of G. cirratum is prohibited under Brazilian law, and its presence in the samples further reinforces the need for more effective fisheries inspections. Brazil's North Coast represents one of the last regions of Brazil where G. cirratum captures are still common, although stocks may collapse in the near future. The reason for this potential collapse is that, even though fecundity is somewhat high with an average number of 34 embryos per litter, reproduction is biennial and a long time is necessary to reach sexual maturity 58 . Since G. cirratum is considered to be Vulnerable in Brazil and several local extirpations have occurred throughout Brazil's coast 59 , systematic studies evaluating catch rates and size composition of captures are needed to support any conclusion on its population status in the BNC.
The first record of the genus Squalus for Brazil's North Coast is intriguing, given that these sharks are thought to be restricted to deep waters 60 . Most of the 31 known Squalus species occur in the Indian and Pacific Oceans [60][61][62] , and only eight occur in the southwestern Atlantic Ocean [61][62][63] . The specimen was landed at Tutoia, in Maranhão state, and was probably caught in a gill net. While this may have been a random event, it shows that the deep water shark fauna of the Brazilian North Coast is poorly known. In fact, our data points that this could be a new species of Squalus since there was a similar match (98%) with two species that comprise a species complex just now being unraveled 60 . Comprehensive systematic surveys off the Amazon estuary and adjacent areas are urgently needed, especially given the presence of mesophotic biogenic reefs in the area 64 . In fact, the whole BNC has an enormous potential for the discovery of new species of deep water sharks, rays, and skates.
Overall, we show that threatened shark species are being harvested throughout Brazil's North Coast, which could result in stocks collapses, local extinctions, and possibly even global extinction of some species with restricted range, like Isogomphodon oxyrhynchus. This prediction may be offset by the implementation of measures at a regional level to reduce shark bycatch. These could include nets with a tighter mesh attached to larger buoys, smaller soak time, and the release of live specimens. These are powerful strategies to reduce bycatch, increase the potential of releasing non-target fishes alive, and to contribute to the establishment of sustainable shark fisheries. Over the long term, more detailed research on population dynamics, genetic diversity, demographic patterns, and habitat use are crucial to subsidize the development of measures such as the establishment of priority areas for conservation. There is also an urgent need for the implementation of more effective monitoring and law enforcement for which molecular identification can be a fundamentally effective tool.

DNA-based species identification.
Samples were stored in 100% ethanol and maintained at −20 °C until DNA extraction. DNA was extracted using a saline protocol 65 . Samples that did not yield DNA were extracted using the Wizard Genomics Purification kit (Promega) following the manufacturer's Mouse Tail protocol. DNA was quantified in Nanodrop 2000 (ThermoFisher) and visualized in 1% agarose gel electrophoresis stained with GelRed (Biotium, Inc).
We used partial sequences of the mitochondrial DNA genes cytochrome oxidase subunit I (COI) and/or NADH dehydrogenase subunit 2 (NADH2) to identify species. Samples were identified using either one of the markers. The COI amplification by PCR (Polymerase Chain Reaction) used the universal FISH F1/R1 primers 41 . NADH2 was amplified using the ILEM/ASNM universal primers 42 . Reactions were carried out using 50 to 100 ηg of DNA template, 1× GoPro buffer, 1.5 mM MgCl 2 , 200 µM dNTP, 0.2 µM of each primer (COI) and 0.3 µM (NADH2), 1U of Taq polymerase (Promega), and ultrapure water to complete for 15 µL of final volume. Samples that did not yield adequate amplicons were amplified by adding 1% DMSO (dimethyl sulfoxide) to the PCR mix. Amplification conditions for COI were: denaturation with 94 °C for 2 minutes followed by 35 cycles of 94 °C for 30 seconds, 54 °C for 30 seconds, and 72 °C for 1 minute, with a final extension phase at 72 °C for 10 minutes. Conditions for NADH2 were: denaturation with 94 °C for 3 minutes followed by 39 cycles of 94 °C 30 seconds, 48 °C for 30 seconds and 72 °C for 90 seconds, and an extension phase with 72 °C for 5 minutes. Amplicons were purified using Illustra Exo Prostar (GE Healthcare Life Sciences) or PEG 8000. Sequencing was carried out using BigDye terminator v3.1 kit (Applied Biosystems) in an ABI XL 3500 (Applied Biosystems).
The quality of sequences was assessed using Geneious Pro version 9 66 . Alignments were performed using MUSCLE 67 with the default settings. Both final alignments resulted from trimming the 5' and 3' ends to avoid poor base quality. Intra and inter-specific distances were calculated using Kimura-2-parameters (K2P) 68 in MEGA 6 69 . Samples were identified using the Basic Local Alignment Research Tool (BLAST) 70 , which compared our sequences with those deposited in the Barcode of Life Database systems (BOLD) and in GenBank (National Center for Biotechnology Information -NCBI), producing similarity percentages.
Neighbor-joining trees were built for both COI and NADH2 sequences using MEGA 6 with 1000 bootstraps 71 using the K2P distances. Sequences from Aetobatus narinari, Hydrolagus africanus and Callorhinchus capensis were used as outgroup for COI, and Aetobatus narinari and Hydrolagus novazelandiae for NADH2. Additional sequences of both markers from all identified species and sister-groups were retrieved from NCBI for comparison (Supplementary Table S1). Our sequences were deposited in GenBank under the following access numbers: COI: MF686569 -MF686584; NADH2: MF740888 -MF740919).