Unexpected Absence of Population Structure and High Genetic Diversity of the Western Atlantic Hermit Crab Clibanarius antillensis Stimpson, 1859 (Decapoda: Diogenidae) Based on Mitochondrial Markers and Morphological Data

: Recent studies on genetic variability have revealed different patterns of genetic structure among populations of marine decapod species with wide geographical distribution. The hermit crab Clibanarius antillensis has a broad distribution along the western Atlantic Ocean, from south Florida (United States) to Santa Catarina (Brazil). This factor, in addition to differences in larval morphology and in adult coloration, makes this species a good model for studies on intraspecific variations. Therefore, we evaluated the molecular and morphological variability of C. antillensis along its distribution in order to check the levels of population structure. The results were based on the morphological analyses of 187 individuals and 38 partial sequences of the mitochondrial gene 16S rRNA and 46 of cytochrome c oxidase subunit I (COI) from specimens whose locations covered the whole species distribution. The molecular analyses did not show any apparent population structure of C. antillensis . This result was corroborated by the morphological analyses since the characters analyzed did not show any pattern of variation. Our results may be explained by a set of factors, such as the dispersive potential of the species and the absence of barriers that could prevent gene flow. In addition, high genetic diversity was observed, mainly for COI, which may be explained by the historical processes of the species, which seem to be in almost constant expansion in the last 700,000 years and experienced no genetic bottleneck. Apparently, this species was little affected by the climate fluctuations of Pleistocene. Additionally, our morphological analyses al-lowed us to present herein a redescription of the studied species since we noted differences from the characters in the diagnosis.


Introduction
Species and their populations are constantly changing. Their history, as well as details from their current stage of genetic structure, are a combination of different past events [1], which may be understood by investigating their genetic processes [2]. Gene flow, for example, is essential to maintain genetic homogeneity or heterogeneity among populations of a species [3,4].
For most marine invertebrate species, planktonic larvae and their life span influence their dispersion process which allows them to interconnect populations by reaching long distances [5,6]. Therefore, long larval stages are usually related to high dispersal capacity and levels of gene flow and reduced population genetic structure [7][8][9]. However, some studies revealed that high levels of connectivity and genetic homogeneity were not nec- Some morphological differences were found between larvae from Brazil and Panama and Mexico populations, such as the number of antennular aesthetascs, the number of denticles of crista dentate of the third maxilliped, the development of the external lobe of the maxillule and endopod of the maxilla [34][35][36]. Additionally, distinct coloration patterns were found among adults from different localities [37].
Additional investigations on genetic variability may contribute to a better comprehension of biogeographic processes, population differentiation and biodiversity along groups/families of hermit crabs in the western Atlantic. Additionally, it may allow checking if there is an evolutive signal among them. Based on the reported scenario, the wide distribution of C. antillensis, its larval stage duration and the context previously described, this hermit crab is a suitable species for investigations on genetic variability and morphological analyses. Therefore, the aim of this study was to: (1) check the levels of population structure along C. antillensis distribution; (2) analyze morphological and molecular variations, and (3) analyze, preliminarily, demographic factors related to its current diversification pattern.

Sample Collection
Most individuals were obtained from the Crustacean Collection of the Department of Biology, University of São Paulo, Brazil (CCDB). In order to cover most part of the species distribution (Figure 2), we also analyzed specimens obtained by means of loans or donations from the following collections: University of Louisiana at Lafayette Zoological Collection, LA, United States (ULLZ-recently transferred to the National Museum of Natural History, Smithsonian Institution, Washington, D.C. USNM; both catalog numbers are used, as specimens are now permanently cross-referenced under both numbers at the USNM); Florida Museum of Natural History, University of Florida, FL, United States (UF); United States National Museum, Smithsonian National Museum of Natural History, Washington, DC, United States (USNM); American Museum of Natural History, NY, United States (AMNH); Natural History Museum of Los Angeles County, CA, United States (NHMLA); Colección Nacional de Crustáceos, Universidad Autónoma de Mexico, Mexico (CNCR). Before the analyses, the identification of specimens was confirmed based on previous morphological characters established in the literature [28,30,31,38,39].

DNA Extraction, Amplification and Sequencing
For DNA extraction, we used muscle tissue from pereiopods or abdomen and followed saline protocols described by Schubart et al. [40], with modifications from Mantelatto et al. [41], and Chelex ® resin [42]. Some adaptations were made to suit our material.
The fragments were amplified by polymerase chain reaction (PCR) [43] in a Veriti 96-Well Thermal Cycler ® (Applied Biosystems, Foster City, CA, USA). The molecular markers 16S rRNA and cytochrome c oxidase subunit I (COI) were chosen, since these mitochondrial genes have been widely used and effective on studies that contribute to our comprehension of Decapod diversity [40,41,44,45]. Of all primers used in this study (Table 1), we designed one pair of each marker in Primer-Blast (National Center for Biotechnology Information, Bethesda, MD, USA) [46]: 16SLClib and 16SHClib; COILClib and COIHClib, due to amplification difficulties. For this purpose, we based the design on the alignment of two 16S rRNA GenBank sequences (KF182529 and DQ369941) and new sequences of 16S rRNA and COI.  ; annealing for 40 cycles of 1 min at 95 °C, 1 min at 38-48 °C and 75 s at 72  °C; final extension for 6 min at 72 °C. PCR products were electrophoresed on 1.5% agarose gel for confirmation, purified using the SureClean Plus ® kit (Bioline, Tauton, MA,  USA), following the manufacturer's instructions, and sequenced with the ABI BigDye Terminator Mix (Applied Biosystems, Foster City, CA, USA) in an ABI 3730 XL DNA Analyzer (Applied Biosystems automated sequencer, Foster City, CA, USA), following the manufacturers' protocol.
The forward and reverse obtained sequences were edited and used to construct a consensus sequence in BioEdit 7.2.5 (Ibis Therapeutics, Carlsbad, CA, USA) [49]. The identity of the consensus was confirmed with BLAST (Basic Local Alignment Search Tool) [50] by comparisons to accessioned sequences of GenBank database. COI consensus were checked for the occurrence of pseudogenes at the online Translate tool on SIB ExPASy [51]. Multiple sequences were aligned for each gene using MUSCLE (Multiple Sequence Comparison by Log-Expectation, European Molecular Biology Laboratory-The European Bioinformatics Institute, Hinxton, UK) [52].
Besides the sequences we obtained, which were all submitted to GenBank (National Center for Biotechnology Information, Bethesda, MD, USA), we also included two 16S rRNA sequences of C. antillensis retrieved from GenBank (Table 2). For genetic distance and phylogenetic analyses, we added five 16S rRNA and 13 COI sequences of other species of the genus Clibanarius (Table 3)

Genetic Distance Analyses
Genetic distances were calculated to determine intra and interspecific variation rates with the software MEGA 6.06 [54], using the Kimura 2-parameters substitution model [55]. Two genetic distances histograms were constructed in Microsoft Excel 2010, with interval ranges of 0.2%.

Phylogenetic Analyses
Maximum likelihood (ML) analyses [56] were conducted in RAxML-HPC Black Box 8.2.4 (Randomized Axelerated Maximum Likelihood, Heidelberg Institute for Theoretical Studies, Heidelberg, Germany) [57], implemented at the online platform Cyber Infrastructure for Phylogenetic Research (CIPRES). We used the default parameters for RAxML and the evolution model GTR+Γ+I [General Time Reversible [58] + Gama + In-variables sites] and the consistency of the topologies was measured by bootstrap method (1000 replicates). The topologies were visualized and edited using FigTree 1.4.2 (University of Edinburgh, Edinburgh, UK) [59]; only values > 50% were reported.

Genetic Variability Analyses
The genetic variability analyses were conducted for both 16S rRNA and COI. The genetic diversity indexes, such as number of haplotypes (H), haplotype diversity (Hd), nucleotide diversity (π) and average number of nucleotide differences (K), were calculated in DnaSP 5.10.1 [60]. Haplotype networks were constructed using statistical parsimony method with TCS 1.21 [61]. In case of ambiguous connections, the criteria proposed by Excoffier and Langaney [62] were considered. Analyses of Molecular Variance (AMOVA) were conducted using the software Arlequin 3.5.2.2 (University of Bern, Bern, Switzerland) [63] to calculate the variance within and between previously established groups and the fixation index values (FST).
The Bayesian skyline plot (BSP) [69] analyses was conduct only for COI and it was used to infer the demographic history of the species under coalescent model. First, the substitution model HKY+I+G [Hasegawa-Kishino-Yano [70] + Invariable sites + Gama] was selected using jModelTest 2.1.10 (Free Software Foundation, Inc., Boston, MA, USA) [71] with Bayesian information criterion (BIC). Afterwards, some parameters were selected in BEAUti (Bayesian Evolutionary Analysis Utility, University of Auckland, Auckland, New Zealand) to create the input file in BEAST 1.8.4 (Bayesian Evolutionary Analysis Sampling Trees, University of Auckland, Auckland, New Zealand). The divergence rate was 1.4% per million years [72], the number of Markov chain Monte Carlo interactions was 10 million, at every 1000 chains, with a 10% burn-in. Then, the output was analyzed using Tracer [69], and a graphic was created.

Morphological Assessment
Morphological data was accessed to compare specimens of C. antillensis from different localities. We adopted all diagnostic characters found in the taxonomic literature [28,30,31,38,39]. Therefore, we measured length of shield (sl), rostrum, lateral projections, left ocular peduncle, right chelae, dactyl, propodus, carpus, merus and ischium of the left second pereiopod; width of front and right chelae. We also analyzed shape and disposition of tufts of setae of shield; shape of rostrum, front and telson lobes; shape and number of spines of ocular acicle; number and disposition of spines of antennal acicle; number and disposition of spines and tufts of setae of right cheliped; coloration and number and disposition of spines of second and third pair of pereiopods. A redescription of the species was made, since we noted differences between some characters observed in this study in comparison to literature descriptions.

Genetic Distance Analyses
The automated sequencing protocols to obtain two fragments of mitochondrial genes resulted in ~1170 base pairs (bp). The alignment of 16S rRNA with 530 bp included 38 sequences of C. antillensis and 12 sequences from other species of Diogenidae. The intraspecific divergence for C. antillensis varied from 0-0.99%, whereas interspecific values ranged from 1.48-24.98%, with the first value corresponding to the divergence between sequences of C. vittatus and C. symmetricus (Figure 3a). An interspecific gap was not evident for this marker. The alignment of COI with 640 bp included 46 sequences of C. antillensis and 16 sequences from other species of Diogenidae. In this case, the interspecific gap was evident, since the intraspecific divergence for C. antillensis varied from 0-2.90% and the interspecific values ranged from 5.80-22.80% (Figure 3b).

Phylogenetic Analyses
Both phylogenetic trees, generated by ML analyses, indicated the monophyly of C. antillensis in clades with bootstrap values of 87% for 16S rRNA ( Figure 4) and 79% for COI ( Figure 5). There were no pattern dividing groups that could reveal genetic structure. Additionally, in both trees, C. tricolor was closer to C. antillensis than other congeneric species.

Genetic Variability Analyses
The population alignment of 16S rRNA consisted of 396 bp with 38 specimens from 20 localities. Seven haplotypes were detected, haplotype diversity was 0.417, total nucleotide diversity was 0.00241, the average number of nucleotide differences was 0.506, and the number of polymorphic sites was five.
The COI alignment had 524 bp with 46 specimens from 17 localities. Forty-two haplotypes were detected, haplotype diversity was 0.995, total nucleotide diversity was 0.01253, the average number of nucleotide differences was 6.564, and the number of polymorphic sites was 63. Genetic diversity index for each locality are in Table 4.  For 16S rRNA, a central haplotype (H1) was shared by 29 individuals from 18 localities, two haplotypes were shared by two (H2) and three (H3) individuals from different localities, and four were singletons. This network did not show any genetic structure ( Figure 6a). For COI, two of 41 detected haplotypes were shared by two individuals from different localities of Brazil (H2 and H3); one (H1) by two specimens from Venezuela and one from Costa Rica; the others were singletons. The network did not show any genetic structure; however, there was high genetic diversity for this gene (Figure 6b). For 16S rRNA, AMOVA revealed that the variance component within localities (102.02%) exceeded the variance component among localities (−2.02%), with negative and no significant FST-value (FST = −0.0202; p > 0.05), which suggested low or absence of genetic differentiation between localities. For COI, even though FST-value was positive, moderate (0.05 < FST < 0.15) [73] and significant (FST = 0.1231; p < 0.05), which suggested genetic differentiation among localities, variance component within localities (87.69%) was higher than that found among localities (12.31%).

Demographic Analyses
Tajima's D and and Fu's Fs values were significant and negative for both 16S rRNA (D = −1.64246, p < 0.05; Fs = −4.64238, p < 0.02) and COI (D = −1.91472, p < 0.05; Fs = −25.14949, p < 0.02) genes, which indicated the rejection of the null hypothesis of population neutrality. Mismatch distribution graphics revealed a unimodal distribution pattern for both genes, which were compatible with the sudden population expansion model (p values for SSD and HRI statistics > 0.05) (Figure 7). Therefore, the null hypothesis of population expansion may not be rejected. The BSP for COI gene showed an increase in effective population size, suggesting that the species had expanded over the past 700,000 years, with a period of stabilization between 450,000 and 250,000 years ago, yet there was no evidence of genetic bottleneck (Figure 8).

Morphological Assessment
We analyzed 187 specimens of C. antillensis (121 males, 33 females and 33 ovigerous females) from 17 localities (covering the entire distribution) and with sl ranging from 1.40 to 5.65 mm. We found some variations on the number of spines of ocular acicles, antennal acicles and dorsaldistal surface of carpus of second pereiopod (Table 5). These variations did not show any pattern of morphological distinction between geographic groups; however, they differed from literature descriptions. Therefore, we made a redescription of the species, as follows.

Diagnosis
Shield subrectangular. Second and third pair of pereiopods with dactyl shorter than propodus, lateral surface of merus with a dark stripe on light background and lateral surfaces of carpus, propodus and dactyl with a light stripe on dark background; dactyl with orange distal region.

Description
Shield (Figures 1,9a) subrectangular, longer than broad, with cervical suture and linea transversalis well developed; anterior margin between rostrum and lateral projections straight; lateral margins slightly sloping; dorsal surface plain, lateral region with 2-5 tufts of long setae and anterior region with few scattered setae. Rostrum triangular, twice as long as lateral projections.
Ocular peduncles (Figures 1,9a) as long as frontal width, cylindrical, slightly broader at the base, left slightly longer than right; dorsal surface with scattered tufts of short setae. Corneas slightly dilated. Ocular acicles (Figures 9a,b)  Antennal peduncles (Figures 1,9a,c) barely reaching distal margin of cornea. Fifth segment dorsal surface with tufts of short setae, lateral margin with tufts of long setae. Fourth segment dorsolateral region of distal margin with one spine and setae. Third segment ventrodistal margin with one spine and setae. Second segment dorsodistal and laterodistal margins with tufts of setae; laterodistal margin with one spine; lateral margins occasionally with projections. First segment unarmed. Flagella long, slender, reaching to dactyl of first pair of pereiopods, with short setae. Antennal acicle lateral and dorsal surfaces with long, scattered setae; lateral and dorsal surfaces with 59 spines.
Chelipeds subequal, right slightly larger than left. Chela (Figures 1,9d) twice as long as broad; dorsal surface with short spines; ventral surface with tubercles and tufts of setae; palm and fixed finger with scattered setae; fixed finger lateral surface with tufts of short setae; fixed finger and dactyl ending in spoon-shaped corneous tip. Carpus short, lateral and mesial surfaces with scattered tubercles and long setae, similar to chela; dorsal surface mesial angle with row of spines and long setae; few dorsodistal spines; ventral surface unarmed. Merus long, dorsal surface with small tubercles, long setae and some dorsodistal spines; ventromesial margin with row of short spines; ventral surface with few tufts of setae and few lines. Ischium unarmed.
Fifth pereiopod (Figures 1,9g) chelate, with scattered tufts of long setae. Propodal rasp well developed, covering about one third of propodus lateral surface.
Uropods asymmetrical, left larger than right. Endopodal and exopodal rasps well developed, dorsolateral margins with setae.
Telson (Figure 9h) asymmetrical, left lobe larger than right. Distal margin of posterior lobes rounded, with row of short spines and long setae; lobes separated by distinct median cleft; lateral margins with long setae and indentations distinct.
Color (fresh specimen) Shield with small white spots and darker anterior region. Ocular peduncles greenish-blue with a brown area on dorsal surface. Antennular peduncles orange with a bluish color on distal region of the segment; antennular flagella orange. Antennal peduncles orange with a yellowish color on first two segments; Antennal flagella orange. Chelipeds olive to rusty brown with white spines and white tubercles; chela with a lighter color. Second and third pair of pereiopods with a dark stripe on light background on lateral surface of merus; a light stripe on dark background on lateral surfaces of carpus, propodus and dactyl; dactyl with orange distal region. Figure 1 shows a preserved specimen with the original color pattern and supplements the above description.

Remarks
The similarity between C. brasiliensis and C. antillensis was first noted by Stimpson [38]; however, Forest and Saint Laurent [28] later stated that the description of C. brasiliensis and the original figure of Dana [75] corresponded to C. antillensis. Although the name C. brasiliensis had priority, Forest and Saint Laurent [28] did not reestablish it, because it was not mentioned since Moreira [80]; therefore, the valid name is C. antillensis. The holotype of C. antillensis was collected by Theo Gill and it should be at the National Museum of Natural History, Smithsonian Institution, at United States of America; however, it seems to be lost, according to Provenzano [39]; according to the database of WoRMS edited by Lemaitre and McLaughlin [89], the syntype is deposited in the Naturhistorisches Museum, Switzerland (catalogue NHM 61.44), but not checked by us.
Clibanarius tricolor and C. antillensis are very close morphologically and it is hard to distinguish them when they are preserved and lost their original color. C. antillensis is found from the USA (Florida) to the south of Brazil (Santa Catarina) and C. tricolor is found from the USA (Florida) to the southeast of Brazil (Espírito Santo) [29,30,32]. Both are the only species of the genus Clibanarius from the Western Atlantic that have dactyls of second and third pair of pereiopods shorter than propodi. They are easily distinguishable by their second and third pair of pereiopods original color pattern, once C. tricolor has transverse orange bands on proximal margins of segments, which, except for white or yellow background dactyl, is otherwise blue with dark punctae; C. antillensis has broad longitudinal stripes on dark background, as described above. When preserved in alcohol, the blue on pereiopods of C. tricolor fades, remaining only orange bands and punctae; on C. antillensis, they become orange with lighter stripes [39].

Genetic Structure
Based on our analyses for 16S rRNA and COI genes, we found no genetic structure for C. antillensis along its distribution. The intraspecific divergence was lower than the interspecific variability for both genes, without an evident interspecific gap for 16S rRNA (Figure 3). This occurred due to the proximity between the intraspecific divergence of C. antillensis (0.99%) and the interspecific divergence between Clibanarius vitattus and C. symmetricus (1.48%), two species that possibly went through recent divergence processes [44]. Besides that, there was no gap within intraspecific variability of C. antillensis, which indicated the absence of population structure. Additionally, the phylogenetic trees, as well as the haplotype networks, did not show any grouping pattern that may indicate genetic structure (Figures 4-6). This was also evidenced by AMOVA, as within localities variance components were higher than among localities.
Many marine species have populations widely distributed with low genetic differentiation and habitats interconnected by gene flow [8]. Some examples among decapods distributed along the western Atlantic can be mentioned: the slipper lobsters Scyllarides brasiliensis Rathbun, 1906 [90], the mangrove crab Ucides cordatus Linnaeus, 1763 [24,91], the swimming crab Callinectes danae Smith, 1869 [22] and the congeneric species Clibanarius sclopetarius [27]. The absence of genetic structure within these species, as well as among specimens of C. antillensis from different localities, may be explained by the lack of physical barriers restricting gene flow and by their larval dispersive capacity [92].
In general, many marine species have planktonic larval stages, and their wide dispersal may happen during their first development weeks. In this period, a large number of larvae are released and passively transported by marine currents system, through which individuals might reach long distances and promote genetic and demographic connectivity among populations [18,[93][94][95]. Along the western Atlantic, there is the South Equatorial Current, which reaches the Brazilian coast (9-15 °S) and bifurcates into north (Northern Brazilian Current) and south (Brazil Current) [96]. This bifurcation has different effects on the genetic structure of many marine species, acting as a barrier to gene flow [97,98] or not [45,90,91]. These currents may not prevent gene flow of C. antillensis, in fact, they may facilitate the dispersion of its larvae. Current systems are associated with long-distance connectivity and long duration of larval stages [3,99]. C. antillensis larvae go through five to six stages of development, which altogether take at least 43 days [34,35].
Salinity is another feature limiting the dispersal of species. It is relevant especially in estuarine areas and other coastal environments, since it presents high and constant variations, which affect the physiology and ecology of organisms [19]. It may influence biochemical composition, growth, survival and development of larvae [100,101], feeding activity [19], carbon accumulation rates [102], as well as osmoregulatory activities [103]. In fact, salinity has been described as a barrier for dispersion and gene flow of some decapod's species [22,26,104,105]. In those studies, the absence of gene flow between populations resulted from the incapacity of their larvae to traverse the Amazon River plume at the Atlantic Ocean, where the volume of water discharged changes the local salinity.
On the other hand, the outflow of the Amazon River has not been a barrier for the dispersal of other decapods [22,91] since the larvae may be more tolerant to low salinity. This might be the case of C. antillensis larvae, which can develop at salinity levels of 29-35 ppt [34,35]. However, adults probably are not able to establish on conditions where the salinity is reduced, since there is a gap along their distribution [28,30,31], which corresponds to the north region of Brazil, where the Amazon river ends and promote salinity influence.
Genetic connectivity may also be influenced by behavioral site fidelity and local retention of larvae [11,20]. Hence, even if the larvae present high dispersal potential, if they are retained next to their natal populations for many generations, populations might undergo through enough differentiation, resulting on genetic structure [11]. In addition, there are many other features interfering on genetic structure, such as the biology and life cycle, habitat, local oceanic conditions, local adaptation, ecological and geographic limitations, past geological events, and recent history. Together, they may influence gene flow at specific directions or moments [8,10,14,[106][107][108]. Therefore, even if different species have similar dispersal capacity, it is not easy to establish genetic structure patterns, since they are influenced by different factors at the same time [8,23,109].

Genetic Diversity
In addition to the lack of genetic structure in C. antillensis, a high genetic diversity was found, especially for the COI gene, which presented a total nucleotide diversity of 0.01253 and haplotype diversity of 0.995. The former value is considered high when π > 0.005 [110] and the closeness of the latter to 1 indicates high number of singletons, which corresponds to an individual sequence of certain gene [111]-as observed in the haplotype network for COI (Figure 6b). The high number of low-frequency haplotypes, as well as high values of nucleotide and haplotype diversities might be related to large and stable populations with long evolutionary history and high mutation rates or with secondary contact between different lineages [110,112]. High diversity indexes using the same gene were also found on studies of Opecarcinus hypostegus (Hd = 0.9994, π = 0.02558) [113], U. cordatus (Hd = 0.9820, π = 0.005862) [91] and Callinectes ornatus (Hd = 0.9570, π = 0.01360) [22], which may indicate that their populations have been stable through time or undergone through a slightly recent expansion [22,91]. For C. antillensis, the BSP recovered a long demographic history from 700,000 years ago, with periods of stabilization and small population expansion (Figure 8). Demographic expansion was also evidenced by significant and negative values of neutrality tests for both genes [114] and by mismatch distribution, with a unimodal distribution pattern and non-significant SSD and HRI values [66] (Figure 7).
These results may reflect historical processes, such as past geological events and the demographic history, which could influence current geographical distribution and genetic variation of marine individuals. The high genetic diversity is common to many marine species [115][116][117]. It might be preserved by long-distance dispersal during expansion [116], or many migrations among close areas, generating a higher number of new haplotypes than others that are lost [118,119].
The demographic history and distribution of many marine species were influenced by climatic fluctuations that occurred during Pleistocene (~ 2.6 million-10,000 years ago) [120,121]. During glacial periods, many populations of marine species used to refuge on low latitude regions, possibly resulting in genetic drift. When the climate became warmer, they would recolonize other areas and reestablish populations that had disappeared during the previous glacial event [122][123][124][125]. Consequently, genetic diversity would be considerably higher in areas where colonizers came from different refuges compared to those originated from a single population source [126,127]. Some haplotypes were exclusive because they may have not participated in recolonization [124].
Nonetheless, each species had a unique response to climatic oscillations during glaciation [17]. There are many marine species that were not strongly affected by glacial periods, consequently, their populations probably continued to expand during these periods, resulting in lack of genetic structure [89,125,128,129]. This might be the case of C. antillensis. The almost continuous expansion with periods of stabilization indicated by the BSP analyses for the last 700,000 years, associated with high values of haplotype and nucleotide diversities, do not indicate that this species went through any genetic bottleneck followed by expansion. Therefore, if the species had refuged during Pleistocene, there would not have been sufficient isolation to cause a reduction on genetic diversity or its populations would have isolated themselves in many refuges, maintaining the diversity on periods when population growth was more stable.

Morphological Variations
Morphological analyzes have also corroborated the absence of genetic structure. Although some features presented variability, they have not shown any pattern related to geographic groups, and in some cases, specimens from the same locality presented differences in characters. Intraspecific morphological variations without any pattern have been reported for other decapods with wide distribution [130][131][132]. These differences, as well as the distinct coloration patterns found among adults [37], may be related to environmental conditions (habitat, wave action, food supply, and salinity), local selection pressures and intra or interspecific interactions that may affect each organism differently [133][134][135].
Among the characters analyzed, the number of spines of the ocular acicles, antennal acicles and second pereiopod carpus presented the largest variation. They also differed from the literature descriptions (Table 5).
In the present study, ocular acicles had three to nine spines on dorsodistal margin, in which seven to nine spines were found on 30 specimens, mainly on male, followed by ovigerous females and females from different localities. According to the literature, the number of spines was: three or four [30,74], six [39], and up to six [31].
In the present study, antennal acicles had five to nine spines on lateral and dorsal surfaces, in which eight or nine spines were found on 34 specimens, mainly on male, followed by female and ovigerous females from different localities. According to the literature, the number of spines was up to seven [31].
In the present study, the carpus of the second pereiopod had one to four spines, in which three or four spines were found only on five males and one ovigerous female from Mexico and four Brazilian states. According to the literature, the number of spines was one or two [39].
The importance of including detailed variations on a redescription of the species is to assure that some traits are not neglected and to facilitate the differentiation of closely related taxon [130]. Hermit crabs are usually hard to be distinguished by a unique character, especially if they have lost their original color [130]. C. vittatus and C. symmetricus, for example, only differ by the color pattern of their pereiopods [44], as well as C. antillensis and C. tricolor. Therefore, if preserved specimens have lost their color, the availability of a set of characters is required to facilitate their distinction.
It is important to define the genetic diversity of marine species once it allows us to understand how historical processes and contemporary environmental conditions have influenced their populations along their distribution. In addition, they may reveal aspects of gene flow, evolution, genetic differentiation, and spatial population boundaries [136]. Such studies, consequently, provide information about biodiversity and conservation strategies of species [137,138]. The present study enables the understanding of marine phylogeographic patterns along the western Atlantic Ocean. Overall, our mitochondrial data for both 16S rRNA and COI genes and morphological comparisons did not reveal structure patterns, related or not to geographical patterns, among populations of C. antillensis. These results may be explained by a set of factors including planktonic larval duration of the species and the absence of effective barriers to gene flow. Besides, there were high genetic diversity for COI gene and signs of population expansion in neutrality tests, mismatch distribution and Bayesian skyline plot. This last analysis revealed small population effective size expansion in the last 700,000 years, with some periods of stabilization, and no evidence of bottleneck effect. Therefore, the species might not have been strongly influenced by Pleistocene climatic oscillations.

Funding:
The present study is part of a long-term project to evaluate the taxonomy and genetic variability of decapods in the Western Atlantic, and was supported by scientific grants provided to FLM. This research is part of a Bachelor's thesis by KSN, supported by scientific fellowships from Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES -Ciências do Mar II) and