## Introduction

At its Sixteenth Meeting of the Conference of the Parties in March 2013 in Bangkok, the Convention on International Trade in Endangered Species of Wild Fauna and Flora (CITES) adopted a series of historic proposals, including listing seven species of commercially exploited elasmobranch fish under Appendix II of the Convention. The Appendix-II listings, based on the CITES treaty, means that although they are not necessarily now threatened with extinction, they may become so unless trade is strictly regulated; the listing aims to ensure that the international trade will not threaten the survival of the wild population of the species and to provide a method of tracing and monitoring their trade. The effective date of listings of sharks and rays under CITES Appendix II was delayed for 18 months until September 2014 to give governments time to resolve technical and administrative issues. Some of the challenges include scientific issues such as: limited knowledge for the identification of products in trade, such as processed shark fins, meat, cartilage and gill rakers; insufficient information about wild populations; lack of fishery information; and insufficient data and information on utilization, marketing and trade of sharks and rays1,2,3.

Among newly listed elasmobranchs, Manta rays are iconic species due to the largest size in rays and the significant rates of decline in population sizes in recent years4,5,6. According to recent reports, the increase in the trade in ray gill rakers in the Asian markets, especially in China, has posed a threat to the survival of Manta rays2,5,6,7,8,9,10. However, trading specimens of Manta ray gill rakers were always in mixed shipment with other Mobulid rays9, including two recognized Manta species and nine species of its sister genus, Mobula6,11,12,13. Despite China’s large population, the demand for gill plates, or Peng Yu Sai (the Chinese name for the Mobulid ray gill rakers), is not from the majority of the Chinese, but from the people of Hong Kong, Guangdong and Guangxi14,15.

Since CITES may be the only international treaty about biodiversity conservation that has legal mechanisms, it is considered to be an ideal complement to national and regional management of marine fish. CITES still lacks experience in fish management and in determining the relationship with other international agreements about marine fish16, however, the success of the implementation of the listing of Mantas depends on a well-designed strategy to address surveillance, compliance and enforcement. The substantial implementation of the international cooperation among the source country, the transit country, and the consumer country must be immediately started. In time the paper prepared, a proposal for the Convention of Parties of CITES was submitted to listing all species of Mobula into the Appendix II of CITES. One of reasons of the listing suggested that the specimens of Mobula spps. in trade, mainly the dried gill plates, were unlikely to be able to distinguish by enforcement officers17. Thus, some information such as user friendly guides of species identification for experts and non-experts are urgently needed1 and must be shared among all parties.

In the present study, we sampled at the mainland Chinese markets, adopted DNA barcoding methods for species identification and identified the species components of ray gill rakers. Moreover, the DNA identity of the gill rakers was compared to the morphology of the dried gills. This was used to determine morphological differences between the dried materials, which could be used to classify the dried specimens into species, without the use of DNA data when conducting management or enforcement.

Visual morphological differences of dried ray gill rakers were subsequently obtained, and classification errors of gill rakers characteristics in each species were tested. We consequently presented a rapid practical identification method based on macroscopic morphology of specimens. This method is particularly beneficial for management and enforcement officers as well as for researchers who intend to study the biodiversity conservation of Manta rays in both source and consumer countries. Some recommendations for the global protection of Manta and Mobula rays were also provided in the article. If the Mobula rays were to be included in the CITES Appendices, we also hope that our work can be of assistance in identification and implementation of CITES for these species as well.

## Results

### Species identification and market component

Not all specimens were successfully amplified because of the low DNA quality in some dried gills. In this study we successfully sequenced 761 bp of the COI gene and 1033 bp of the NADH2 gene resulting in 35 and 58 haplotypes respectively. All of the haplotype sequences generated were deposited in GenBank under the accession numbers KT626464-KT626550. Based on DNA barcodes of COI and NADH2 genes, 188 individuals were successfully identified to the corresponding species with 99–100% identical values compared with those deposited sequences in NCBI. Five monophyletic clusters of individuals (with 99–100% supports) were recovered in Neighbor-joining (NJ) trees under the Kimura 2-parameter (K2P) model in both genes, corresponding to five species of Mobulid rays (Appendix I and II). Although the detail calculation was not shown in the tables, the intra species distances were 0–1.1% in COI gene and 0.3–0.6% in NADH2 gene, while those inter species distances were 3.7–12.5% and 4.4–17.1%, respectively. Finally five species of rays were identified (Table 1), namely, Mobula japanica (Müller & Henle, 1841), Mobula tarapacana (Philippi, 1892), Mobula kuhlii (Müller & Henle, 1841), Mobula thurstoni (Lloyd, 1908) and Manta birostris (Walbaum 1792), representing 54.8%, 14.4%, 13.3, 6.4% and 11.2% of the sample set, respectively (Table 1).

### The price changes among various species or in different localities

Dried gills of five species were sold at an average price of 220 ± 98 US$per kilogram in China (Table 1). Regardless of whether it is loose or tight (F = 0.150, P = 0.861) or in any colors (F = 0.566, P = 0.638), a dried ray gill is likely more expensive when the average length of filaments is long (F = 19.486, P < 0.01). Excluding the influence of length of the gill filaments on price, the prices of dried ray gill rakers are significantly different among sampling cities (F = 30.188, P < 0.01). The average price of dried gill rakers were 201 ± 82 US$ in Guangzhou, 229 ± 47 US$in Zhanjiang, 290 ± 25 US$ in Beihai, 294 US$in Zhapo and 445 ± 159 US$ in Shanwei, respectively. Regardless of the abovementioned influences, the price of dried gill rakers of Manta rays is significantly different to that of Mobula rays (AIC = 2773.8, t = 8.986, P < 0.01, Fig. 1).

### Visual discriminations and identification

Tukey’s multiple comparisons of the mean filament length of each species revealed that the filament lengths of the dried gill of each species are significantly different from one another (P < 0.01), except for the differentiation between Mo. thurstoni and Mo. kuhlii (P = 0.54) (Table 2). Gill rakers of Ma. birostris have lengths ranging from about 30 mm to almost 100 mm, and those longer than 70 mm are usually from this species (Fig. 1). Specimens however, cannot be distinguished into species only based on the filament lengths or sizes of the gill rakers (Fig. 2).

Unlike the dried gill rakers of other Mobula rays, the specimens of Mo. tarapacana and Manta birostris can be reliably distinguished via the visual method (Table 2). The dried gill rakers of Ma. birostris and Mo. tarapacana are tight arranged with large-terminal lobes that are joined together. Finger-like projections were observed at the lower edge of each lobe in Manta species, but these projection structures are invisible in several small or poorly preserved specimens. The dried gill rakers of Mo. tarapacana are mostly dark in color with a light base and the so-called variegated gills (Hua Sai in Chinese). Moreover, cilia can be seen under a hand loupe in some well-preserved specimens.

Most terminal lobes of dried gill of Mo. japanica have a sharp apex and are independent from each other, and thus the lobes could be used in species discrimination. Mo. kuhlii and Mo. thurstoni both have oval-shaped gill lobes and their gill rakers are similar in size; thus, discriminating the two species from each other is difficult (Table 2). Overall, when all types of visual traits are used in the random forests formula, most specimens of dried gill rakers could be discriminated to the DNA-identified species level with classification errors of 3.88% in Mo. japanica, less than 0.1% in Ma. birostris and Mo. tarapacana, 16% in Mo. kuhlii and more than 99.9% in Mo. thurstoni (Table 2). A total estimate of error rate is 10.64%. The characteristics of middle lobes are the most useful trait in visual classification followed by the characteristics of the terminal lobes.

For frontline enforcement at customs or border inspection, we develop a three-step process for identifying the gill rakers of Manta rays. The process involves weeding out specimens at the first step, inspecting whether or not terminal lobes are enlarged and fused. The variegated gills are then abandoned, and the finger-like projections are finally observed at the edge of middle lobes through a loupe (Fig. 3).

## Discussion

The fishing of Mobulids for their dried gill rakers had not been given particular attention until the turn of the century7,18. Ma. birostris was listed on Appendices I and II of the Convention on Migratory Species (CMS) in 2011. Currently, increasing concern has been given to this activity and culminated in an adoption of the proposal to amend CITES Appendix II in March, 2013. In early 2014, Indonesia, which was reported to be the largest landing country of Manta rays19, has declared a ban on fishing Manta rays20,21. All international treaties need effective local and national implementation and national fisheries management within relevant parties14. The synergies among different treaties rely on the parties’ willingness and capacity of mainstreaming the administrative and enforcement power of its ministries and sectors, as well as those in the provinces. If so done, it is believed that the increasing threat of gill rakers trade will be curbed by following the reinforcement of international treaties, implementation and enforcement of national laws and regulations, regulation of the demand of side markets, regional cooperation, and national fishery management to ensure the survival of Manta rays, as well as Mobula rays. As a follow-up to the solution to the identification of Mantas, it is therefore imperative that further field studies are carried out around the world to quantify the extent and impact of Manta fishery and trade and to monitor the products flow in order to ensure the preservation of Manta and Mobula ray species and populations worldwide.

In this study, five species, Mobula japanica, Mo. tarapacana, Mo. kuhlii, Mo. thurstoni, and Manta birostris, have been firstly verified to be traded within China. The taxonomy of the genus Manta has been questionable and convoluted, having previously been considered a monotypic genus possibly with three species Manta birostris, Ma. alfredi and Ma. sp. cf. birostris, but further examination of specimens is necessary to clarify the taxonomic status of Manta species11. Aslo, Manta alfredi and Manta birostris are indistinguishable based on both the CO1 and NADH2 genes22,23. Much more sensitive molecular marker should be found and used in the further study because accurate species identification will help to highlight the specific threats and will ultimately assist in the correct assessment of the conservation status in each different species of Manta. Only Mobula japanica and Mo. tarapacana could be distinguished from other Mobula species by the results of this study (Table 2). The morphological approach of gill rakers does not appear robust to distinguish the remaining Mobula rays at the present stage. Given that Mobula are likely as threatened as Manta, and are also slated for inclusion under CITES in October, 2016, we continue doing some works on the morphology of the surface of gill lamellae by using SEM. However, for conservation management and customs enforcement, establishing a kind of voucher sequences for each Mobulid species that can be used as the baseline for comparison moving forward is urgent needed. Ideally, these will be tied to voucher specimens, or alternatively, sequences linked to photographs of whole specimens and gill raker, morphological diagnosis characteristics, as well as meristic data. Such approach is also applicable to all CITES listing species, and need the great effort and collaboration between source countries and consumed counties.

In the IUCN Red List of Threatened Species, Ma. birostris and Ma. alfredi (Krefft, 1868) are listed as “Vulnerable”24 and the Mobula rays involved in this research are all listed as “Near Threatened” or “Data Deficient”25,26,27,28. The population trends of Ma. birostris, Ma. alfredi and Mo. kuhlii are decreasing24,25, but those of other species are unknown. Mobulids may have a similar fecundity rate, where one adult female gives birth to one pup annually. However, the life history parameters of four Mobula species in the present research, namely, age at maturity, longevity, reproductive age, and natural mortality, and even population genetics are mostly unknown6,26,27,28. Thus, scientists and fishery organizations must promote biological, ecological and molecular studies on Mobulid rays, especially on those species largely used for their gill rakers, such as Mo. japanica.

Mo. kuhlii are found in the Indian Ocean and in the Western Central Pacific. On the other hand, Manta birostris, Mo. japanica, Mo. tarapacana, and Mo. thurstoni are circumglobally found in tropical to temperate seas. Compared with the five species involved in this research, Manta alfredi and the rest of the Mobula rays are mainly regionally distributed in the Eastern Pacific, Atlantic, and Mediterranean Sea and in surrounding waters4,6. Moreover, the diver project database reported that Mobulids are fished and then sold in local markets in Sri Lanka, Indonesia, and East Africa16. Therefore, it is presumed that dried gill rakers in Chinese markets are mainly from rays fished and traded from the Indian Ocean and the Western Central Pacific. The trade in dried gill rakers in China may affect coastal countries around the Indian Ocean and Western Central Pacific, which have been identified as one of the main hotspots where sharks and rays are seriously threatened2. More analyses are urgently needed on the genetic backgrounds of local populations, on tracing the origin of Ma. birostris, location of harvests, populations size, and distribution and trends from the harvest sites to determine key resource countries that would receive international assistance and cooperation for sustainable fishery management, implementation of international treaties, and enforcement measures.

From the size distribution and the morphological characters of dry gill raker, 2 empirical rules can be presented for enforcement officers. The results of filaments size and price investigation reveal the filament lengths of Ma. birostris to be from about 30 mm to almost 100 mm with an average of 62.6 mm (Fig. 1), showing a largest mean value and a widest distribution among all species and indicating a variety of sizes of this species consumed for its gill rakers. The gill rakers of Ma. birostris are also generally larger than those of Mobula rays (Fig. 1). Among all species of Mobula rays, adult fish rakers of Mo. tarapacana are large and can even rival Manta rays in size5. Although some of the rakers of Ma. birostris are similar in size compared with those of Mo. tarapacana, many specimens from Ma. birostris are significantly larger than those of Mo. tarapacana. The filaments longer than 70 mm are usually from Ma. birostris (Fig. 1). Hence, a larger size of gill rakers whose filaments are longer than 70 mm would be an important criterion for enforcement officers. Based on the results of this study, we also develop a three-step process for identifying the gill rakers of Mobulid rays which can be result in a classification errors of 3.88% in Mo. japanica, and less than 0.1% in Ma. birostris and Mo. Tarapacana (Table 2, Fig. 3). These three species have higher commercial value and occupy more than 80% of the market component (Table 1). The three-step process of identifying gill rakers of Manta rays has already been trained and greatly improved the efficiency of law enforcement of frontline inspections, customs, and fishery management administrations in China.

### Species-level molecular identification

Whole-genomic DNA was extracted using the TIANamp Genomic DNA Kit (TIANGEN BioSci. and Tech. Co., Beijing, China) following the manufacturer’s recommendations and assessed with 1.0% agarose gel. The mitochondrial cytochrome oxidase subunit I (COI) and NADH2 gene analyses are currently the most effective strategies for species identification and recognition in elasmobranches (589 and 574 of the 1221 recognized elasmobranch species were recorded by COI in FISHBOL and by NADH2 in NCBI, respectively)22, which provide an appropriate genetic resource for comparison and for finding similarities between our results and deposited ones. Thus, we chose the double molecular markers for species identification. Two dedicated primers for each gene were designed using OLIGO 633.

COI-F0: TAACTCTCAGCCATCTTACC,

COI-R823: GCCGATAGCTATTATTGCTCA,

ND2-F2: GCCCATACCCCAACCATGTT and

ND2-R2: TGCCTATCTAGAAGGTTTTAGCT

DNA amplification by PCR was carried out in a final volume of 30 μL, containing 15 μL of 2 × Taq PCR MasterMix, 1 μL of each primer, 1 μL to 2 μL genomic DNA, and ddH2O. The PCR thermocycling conditions for COI gene were 95 °C for 5 min, followed by 35 cycles of 95 °C for 30 s, 52 °C for 30 s, 72 °C for 50 s, and a final extension of 72 °C for 10 min. The conditions for the ND2 gene were the same as those for the COI gene, except that the annealing temperature was 54 °C. The PCR amplification products were sequenced at Sino Geno Max (Beijing) with the forward and reverse primers used for the amplification. Sequences were assembled and edited using CLUSTAL W34 and DNAStar 7.10 ((DNASTAR Inc.). Consensus sequences were blasted through NCBI or BOLD web-based systems. The hits with the highest coverage and maximum identical values were chosen to assign an unknown sample to a known species. Furthermore, a neighbour-joining analysis was used to explore sequence differences at inter or intra-species level.

### Morphological measurements

A small number of museum collections revealed that the filter lobe morphology and ultrastructures of Mobulid rays can be used as a key diagnostic parameter among some species5,35. For each filament plate of gill rakers, the filament lengths were measured (Fig. 4, FL, mm) using a digital vernier caliper at five points: two from edges of the gill raker plate, one from the midpoint, and two between the edge and the midpoint. The average was calculated as the filament length and was used in further analyses.

Based on the combined information of the filament and lobe morphology5,28, four diagnostic characteristics, namely, arrangements and colors of filaments, shapes or structures of middle lobes, and shapes of terminal lobes, were chosen to evaluate the discriminating availabilities of each species. For the arrangement of filaments, three levels were used for intuitionistic observation, namely, loose, general, and tight. A loose arrangement means that each filament has clearance at the edge. A tight arrangement means that the filaments almost overlap at the edge with adjacent ones, and the general arrangement is the midpoint of a loose and tight arrangement (Figs 3 and 4).

The samples included four kinds of independent traits in the middle lobes of dried gill rakers, which are three kinds of subtle structures viewed through a 30X-21 mm magnifying loupe and an oval lobe shape. Finger-like projections are macroscopic dentils extending down from the ventral edge of a lobe, and the lobe surface is smooth in appearance. Cilia are tiny hairs on the surface of the lobe, and veins are dark, thin, vein-like lines on the lobe surface. Except for the three abovementioned subtle structures and the oval shape, several specimens have no remarkable characteristic on the middle lobes (Figs 3 and 4).

Three types of shape of terminal lobes could be used in species identification. Several specimens have enlarged lobes on the terminal end where lobes are joined or fused together. In several specimens, the apex of the lobes is sharp and does not come into contact with the terminal lobes of adjacent filaments. The third kind of terminal lobe has a similar size with the middle lobes and also has an oval shape.

### Analysis and statistics

R version 3.3.1 was used in statistical analysis and figure constructions. Shapiro–Wilk test was performed in the normality tests of data. One-way ANOVA was performed to compare the differences in price of gill rakers among three types of arrangement among different colors and among cities. We also used a general linear model to analyze the relationship between the price and the size of dried gill rakers, treating species, site, arrangement, and color as cofactors. Tukey’s multiple comparisons were performed to test the differentiations among the filament lengths of each species identified via DNA barcoding. We used Breiman and Cutler’s random forests for classification and regression36,37 based on a forest of 1000 trees by using random inputs in testing the availability of each characteristic in visual discrimination. Multidimensional scaling is used to visualize the dissimilarities among specimens based on their sizes38.