Mitocondrial COI and 16S rDNA sequences support morphological identification and biogeography of deep-sea red crabs of the genus Chaceon (Crustacea, Decapoda, Geryonidae) in the Eastern Central and South Atlantic Ocean

The geographical spreading of new fishing activities and the increasingly deeper locations of these activities have shown the worldwide distribution of gerionid crabs and new descriptions of Chaceon taxa. However, incomplete penetrance, variable expressivity, and phenotypic overlap make the morphometric identification of these species difficult. In this study, partial sequences of the cytochrome c oxidase subunit 1 (COI) and 16S mitochondrial ribosomal RNA (16S rRNA) genes have been analyzed in Chaceon species from the Eastern Central and South Atlantic and compared with sequences of species from Western Atlantic. Our results corroborate the proposed morphological species and highlight the significant separation of the Eastern Atlantic species and those from Atlantic coasts of South America for both markers (97% Bayesian posterior probability, BPP / 83% Bootstrap replicates, BT). Interestingly, Chaceon sanctaehelenae shows a closer relationship with the species of the American coast than with those from the Eastern Atlantic. On the other hand, while COI marker clearly separates Chaceon atopus and Chaceon erytheiae species (99 BPP / 91% BT), these species share haplotypes for the 16S rRNA marker, pointing to a recent speciation process. Moreover, a close relationship was observed between Chaceon maritae and Chaceon affinis (94% BPP / 77% BT). The topologies of the trees obtained indicate that the ancestor of this genus was closer related to those species from South America than to those from the Eastern Atlantic.


Introduction
Morphological identification of species belonging to the Geryonidae family has been a difficult task in the past for people studying deep-sea resources and marine biodiversity. Interspecific similarities and intraspecific morphological variability have contributed to former and potential current misclassification [1]. Thus, other considerations such as the geographical site of collections have also been an element used in the species identification, regarding previous species citations in the same geographic zone [1,2,3].
Until the early eighties most species along the East Atlantic Ocean and also off the African coast of the Indian Ocean were misidentified as Geryon quinquedens, now a species restricted to the Northwest Atlantic. In 1981, Manning and Holthuis [2] described Geryon maritae as a new species in the Atlantic coast of Africa. These authors reported that most records of Geryon affinis or Geryon quinquedens from West Africa pertained to G. maritae.
Later, the geographical spreading of new exploratory fishing activities and associated studies, as well as the increasingly deeper locations of these fishing activities, provided new biological material that showed the worldwide distribution of gerionids and new species descriptions. Through this process, Manning and Holthuis [3] recorded three species of Geryon from South West Africa-Namibia and South Africa: Geryon chuni, Geryon macphersoni and G. maritae. Afterward, they faced the restructuration of the Geryonidae family moving most of the species in the genus Geryon into a new genus, Chaceon and describing nine new species [1].
Thus, since the eighties the number of new descriptions of Chaceon genus has considerably increased reaching the amount of 34 species [4]. However, genera of the Geryonidae family and some species assignment to these genus have been in controversy up to now. Thus, Ng et al. [5] listed 31 species of Chaceon where C. sanctaehelenae is dropped and Chaceon chuni and Chaceon inghami are considering as Geryon chuni and Geryon inghami.
In some cases, the multigenic inheritance, the reduced penetrance and variable expressivity of morphological traits, as well as the observer subjectivity could make the morphometric identification difficult. In this context, where the presence of cryptic species has high possibility to occur, molecular analysis could shed light on morphological identification and biogeography of the genus Chaceon.
In this sense, the use of 16S mitochondrial ribosomal RNA (16S rRNA) and cytochrome c oxidase subunit 1 (COI) allowed reconstructing the initial phylogenies in crustacean [6,7]. Specific studies on Chaceon species have been conducted by Weinberg and col. [8] in the North Atlantic, and Mantelatto and col. [9] in the South West Atlantic. However, there are no studies providing a broader vision of phylogenetic relationships among species of Chaceon genus from Eastern Central and South Atlantic.
Therefore, the objective of this study is to review the morphology-based taxonomic classificaton of deep-sea red crabs of the genus Chaceon using mitocondrial COI and 16S rRNA sequences.

Material and methods
Forty-three samples of muscle tissue from individuals of deep-sea red crabs (Chaceon spp.) were collected from different places in the Eastern Central and South Atlantic (Fig 1). Species names assigned on morphological basis [1,3,10] and sampling areas are reported in Table 1. The presence and potential geographic distribution of species of Chaceon genus in the Central East and South Atlantic is shown in Fig 1. Five species of Chaceon were morphologically identified. C. erytheiae from Valdivia Bank was identified following Macpherson [10]. C. atopus [1] and C. sanctaehelenae [1] specimens collected in the Vavilov Ridge (Gulf of Guinea) were identified using Manning and Holthuis [1] description. Finally C. affinis [11] from the Canary Islands and Fantasma Bank (north Canary Islands) and C. maritae [2] from Western Sahara waters were identified through Manning and Holthuis [2]. Also, nine samples of unidentified specimens from Namibian Exclusive Economic Zone (EEZ) and the Walvis Ridge (Valdivia Bank) were incorporated to the analysis (Table 1; Figs 1 and 2). Table 1 also includes the serial code; the ID Code used on labelling the samples; name of research cruises or fishing trips in which samples were obtained; collectors; places of harvesting and depths; and measurements of individuals and sex. All collected specimens belonged to commercial exploited species so there is no prohibition for their fishing. The fishing vessels and fishing research vessels where they were captured have the legal permits to carry out their fishing and research activities. This study did not involve endangered or protected species.
Amplifications of PCR were performed in a Bio-Rad S1000 Thermal Cycler with dual bloc in a total volume of 25 μl including 1x buffer (GeneAll Biotechnology, South Korea), 150 μM  Reaction efficiencies were estimated by electrophoresis in agarose real-safe stained gels (Real Biotech Corporation, Taiwan). Subsequently, PCR products were enzymatically purified with Illustra ExoProStar 1-Step (GE Healthcare, UK) according to the manufacturer's instructions. Sequencing reactions were performed for both strands with the same primers used in the amplification at the Genomic Service of the University of La Laguna (SEGAI). Sequences of DNA were edited and assembled using MEGA v. 6 [15]. Sequence alignment was performed using CLUSTAL W [16] as implemented in MEGA. The absence of stop codons in the COI alignment was confirmed using the software MEGA 6 [15]. In order to establish the phylogenetic relationships within the Chaceon genus, we used published sequences from other two Chaceon species, C. notialis and C. ramosae [9] from the Western Central Atlantic, as well as those from Liocarcinus depurator and Carcinus aestuarii used as outgroups in the analysis. Accession numbers to GenBank data base of all sequences used are reported in Table 2.
Site saturation was evaluated in DAMBE v.5.5.18 software [17] by plotting the numbers of observed transitions and transversions against pairwise genetic distance.
Phylogenetic relationships were constructed from both individual mitochondrial COI and 16S rDNA, as well as concatenated sequences. The best-fit nucleotide substitution model was determined using jModelTest 2 [18] according to the Bayesian Information Criterion. HKY + G and HKY + I, were the chosen models for COI and 16S rDNA, respectively and so used in the Maximum Likelihood (ML) and Bayesian Inference (BI) analyses. In the combined data set, substitution models, parameters of nucleotide frequencies, substitution rates, gamma shape, and invariant sites proportion were unlinked across partitions.
ML was performed with Treefinder version March 2011 [19] always with 1,000 Bootstrap replicates (BT). BI analysis was performed using MrBayes v.3.2.1 [20] on the Mobyle SNAP Workbench [21]. Two parallel runs were performed in MrBayes using four Markov chain Monte Carlo (MCMC) chains. Ten million MCMC generations were run, with a sampling frequency every 10 2 generations. Of the resulting trees, the first 10,001 trees were discarded as burn in after checking for stationarity with TRACER v1.3 [22], and the following 90,000 trees were used to estimate the topology and tree parameters. Percentage of times a node occurred within those 90,000 trees was interpreted as the Bayesian posterior probability (BPP) of the node. Trees generated from both analyses were edited by Treegraph 2 [23].

Results
From all 43 samples of deep-sea red crabs (Chaceon spp.), a fragment of 468 base pairs (bp) of mtDNA COI gene was amplified and sequenced from 38 samples. The absence of stop codons seems to discard the presence of nuclear mitochondrial insertions (Numts) in the sequences. Among the 468 nucleotide sites, 142 sites were variable and 87 were phylogenetically informative. From 42 samples, the amplification of the corresponding fragment of 16S rRNA resulted in a single band of 429-430 bp. The alignment of the sequences generated a 432 bp matrix because of the need to include "gaps" (variants insertion / deletion, "indels") for optimal alignment. Of the 432 sites, 104 showed variation and of these, 47 were informative. Indels were discarded in subsequent analyses. The absence of double bands in PCR and double peaks in the electropherogram of the sequences as well as the similarity among them and with those originals deposited in Genbank seems to discard the presence of nuclear mitochondrial DNA sequences (Numts) in these sequences too.
The detected variation between sequences showed no deviation from neutrality in either case (D = -1.23 and D = -0.31 with P> 0.10 for 16S rRNA and COI, respectively). The saturation plots for COI and 16S rRNA (Fig 3) show that transitions and transversions are roughly linearly correlated, with the Jukes-Cantor genetic distance [24] with no obvious tendency to level off. In addition, Xia's index for substitution saturation in COI produced values of 0.024 (first and second codon positions) and 0.104 (third codon position), and 0.067 in the case of the 16S rRNA, significantly lower (P < 0.001 in all cases) than the critical values for symmetric  [11], male carapace width 177 mm; (B) C. maritae [11] male carapace width 117 mm: (C) C. atopus [1], male carapace width 153 mm; (D) C. erytheiae [10], male carapace width 112 mm; and (E) C. sanctaehelenae [1], male carapace width 150 mm. The first point that emerges from these relationships is the significant separation of Eastern Atlantic species, excluding C. sanctaehelenae, from those who inhabit the Atlantic coasts of the American continent when we take into account the phylogenetic tree for the concatenated sequences (97% BPP / 83% BT) (Fig 6). Curiously, C. sanctaehelenae species shows a closer relationship with the species of the American coast, C. ramosae and C. notialis, than with those from the Eastern Atlantic as C. atopus, although the distribution areas of both species overlap. On the other hand, all analyses strongly support the close relationship between C. atopus and C. erytheiae species (100% PPB / 99% BT) (Fig 6). However, while COI marker clearly separates both species (99% BPP / 91% BT) (Fig 4), individuals from C. atopus and C. erytheiae share haplotypes for the 16S rRNA marker (Fig 5). Finally, it is worth noting that the sequences of 9 samples from Namibia which had not been morphologically identified turned out to belong to the species C. erytheiae (S1-S5) and C. maritae (N1-N4) ( Table 1) as it was expected, given the geographical location of their capture     Table 1 except for C. notialis and C. ramosae. Numbers at nodes represent Bayesian posterior probability and Bootstrap values of maximum likelihood analysis, respectively. (Fig 1). A close relationship was observed between C. maritae and C. affinis (94% BPP / 77% BT). Specimens of C. maritae from the Sahara Coast and Namibia waters did not show any differences between them and formed their own clade with 100% Bayesian posterior probability (Fig 6). On the other hand, C. affinis specimens were also grouped in a different clade with a relatively high branch support value (100% BPP / 90% BT) (Fig 6).

Discussion
Mitochondrial DNA (mtDNA) is the marker commonly used for genetic reconstruction of the history of populations, population demography, biogeography and speciation, and recommended for taxonomic studies [25]. Despite the fact that mtDNA is not exempt from some disadvantages as the presence of Numts or pseudogenes [26], the presence of nuclear sequences is highly unlikely in our data.
Nowadays, only two phylogenetic studies have been found regarding Chaceon Atlantic species [8,9] from the Northwest and Southwest Atlantic. Thus, the present work adds a new piece to the puzzle for the Southeast Atlantic.
Our results show that Chaceon species from the Central and South East Atlantic seem to have shared a common immediate ancestor with the Western Atlantic species.
Although it is difficult to contrast our results with those from other authors when different species are studied, some ideas can be extracted expanding the geographic scope. Weinberg et al. [8] analyzed samples from the Northwest Atlantic, Chaceon quinquedens (Southern New England and Gulf of Mexico) and C. fenneri (Eastern Florida). Three sequences of C. affinis, from the Eastern Atlantic (Madeira Islands and Canary Islands) were also included in the analysis. The authors detected genetic-based subdivision between C. quinquedens from the New England region of the Atlantic Ocean and C. quinquedens from the Gulf of Mexico by using sequence data from two genetic loci (16S and ITS). However, they found little or no genetic difference (16S data) between C. quinquedens from the Gulf of Mexico, C. fenneri from Eastern Florida, and C. affinis from the Eastern Atlantic, suggesting that these trans-Atlantic taxa share a more recent common history than the two populations of "C. quinquedens".
Trying to compare our results with those of Weinberg et al. [8] we added all their sequences to our 16S rRNA data matrix. The output tree (S1 Fig) identified the samples of New England C. quinquedens as C. erytheiae, being their haplotypes identical. Also, the sequences of C. quinquedens from the Gulf of Mexico corresponded to sequences of C. affinis from the East Atlantic. In spite of the absence of COI sequences for those specimens, the coincidence of the phylogenetic trees for 16S RNAr and COI in our study would be in agreement with a misclassification.
In their article, Mantelatto et al. [9] showed that species like C. ramosae and C. notialis form a differentiated clade, aside from C. affinis and C. maritae clade and that the estimated genetic distance of COI gene among Chaceon specimens can be used to separate the species C. notialis and C. ramosae. Once this paper was published, we realized the possible incorrect geographical location of their C. maritae identification and we personally confirmed that these samples really belong to the Valdivia Bank. In this sense, given our research experience in this area we wonder whether the morphological identification of these specimens is correct. Although two species were initially identified for this region [27,28], current reanalysis demonstrated the presence of one single species in all samples (C. erytheiae) belonging to the Valdivia Bank.
In this case, we also tried to understand similarities and differences between their analyses and our findings, adding their sequences to our data matrix and reanalyzing the outputs. The phylogenetic tree for COI (S2 Fig) showed that their C. maritae sequences have 100% similarity with our samples of C. erytheiae, reaffirming the possible morphological misidentification of this species by Mantelatto et al. [9]. It was not possible to carry out the same analysis with their sequences for 16S RNAr since these only overlap with our sequences 200 nts. Anyway, their sequences are identical to ours of C. erytheiae for the region shared. All the samples we analyzed from Valdivia Bank belong to C. erytheiae. However, we also have doubts about the COI sequence they obtained from GenBank for C. affinis, finally not published, and different to the ones that we obtained for this species (S2 Fig) which could explain the contradictory results they found between markers.
C. maritae is the main Geryonidae species in the West African coast, from the Western Sahara coast to the Namibian waters; it is located in the sandy-muddy continental slope at depths from 200 to 900 meters. This species is part of important fisheries (as by-catch in trawling deep-sea shrimp fisheries or target species in trap fisheries) along its distribution area from which Mauritania, Senegal, Angola or Namibia can be highlighted [29,30,31,32,33]. Results indicate that analyzed samples from Western Sahara coast and Namibian waters, both extremes of its broad geographic distribution, belong to the same species (BT and BPP = 100%), which constitutes a quite relevant fact taking into account the physical separation. However, this homogeneity could be explained by horizontal migratory processes of this species along the African coast. Records of mark and recapture individuals of this species reveal that mature females covered considerable distances of up to 380 km, but commonly between 100 and 180 km [28].
Despite being morphologically different, C. maritae forms a clade with C. affinis. C. affinis is a species that has colonized oceanic archipelagos from Cape Verde to Azores [34,35,36] and seamounts off the Central African coast and also the deepest continental slope (700-1900 m) off Morocco and Western Sahara (C. Hernández-González, IEO, pers. comm.), off the Atlantic coast of the Iberian Peninsula [37], Northeast Atlantic southern 64˚N (Iceland) [38] and the Mid-Atlantic Ridge hydrothermal vents [39]. There are few records of the presence of these two species in the same archipelago or seamount, only in the Canary Islands and Cape Verde [40,41,42]. When more than one species settle in any geographic place, the depth range of distribution of each one tends to be different. This is the case of C. affinis, which distribution is deeper (500-1900 m) than C. maritae (100-950 m). Both species also show different patterns of size and sex population structure in relation to depth [35].
Similarly, C. atopus and C. sanctaehelenae seem to inhabit different depths in the same area of Vavilov Ridge, 500-800 m and 300-400 m, respectively. In relation to these two species it is interesting to remark that the ancestor of each one seems to have a different geographic origin, being that of C. sanctaehelenae in the Southwest Atlantic (C. notialis and C. ramosae); while the ancestor of C. atopus appears to have existed in the eastern Atlantic.
C. atopus and C. erytheiae share a preference for habitats deeper than 500 m, reaching depths near to 2000 m in the case of C. erytheiae. For these species, the molecular variation detected for COI, the marker with a higher substitution rate, allowed to distinguish haplotypes from both species, while for 16S rRNA marker individuals of both species shared haplotypes. This speaks in favour of a process of incipient speciation between these species.
Photographic records of Chaceon specimens caught along the Walvis Ridge, in the southwest Valdivia Bank, suggest the presence of C. erytheiae (size<140 mm of carapace width, CW) at those latitudes, until the Wust Seamount. Size records of individuals from this region and the southernmost Discovery Seamounts also suggest the presence of another bigger species that seems to be Chaceon gordonae (size<180 mm of CW), that is one of the cited species in the northern part of the region which we missed in our genetic analysis.
The basal position of species from the continental slope off Brazil and Uruguay seem to indicate a closer relationship with the ancestor of the genus. Consistent with our results, the presence of fossil Chaceon records (C. peruvianus) in South America [43] could represent the origin of current species in the Eastern Central and South Atlantic, in a parallel process of migration-dispersal and speciation northward and eastward. The most suitable path to cross the Atlantic would be the equatorial region, with a complex system of currents and countercurrents [44] at different depths that would help pelagic larvae transport [45] to new suitable habitats in the intermediate seamounts until reaching the west coast of Africa, spreading out northward and south by migrations and reaching new oceanic features by larvae transport. In this last mechanism, local oceanographic events as the Angola Gyre could have played an important role. Also features as the Mid-Atlantic Ridge, Walvis Ridge and Rio Grande Rise comprise potential paths that could explain the distribution diversity of Chaceon species in those regions.