Extreme mitochondrial variation in the Atlantic gall crab Opecarcinus hypostegus (Decapoda: Cryptochiridae) reveals adaptive genetic divergence over Agaricia coral hosts

The effectiveness of migration in marine species exhibiting a pelagic larval stage is determined by various factors, such as ocean currents, pelagic larval stage duration and active habitat selection. Direct measurement of larval movements is difficult and, consequently, factors determining the gene flow patterns remain poorly understood for many species. Patterns of gene flow play a key role in maintaining genetic homogeneity in a species by dampening the effects of local adaptation. Coral-dwelling gall crabs (Cryptochiridae) are obligate symbionts of stony corals (Scleractinia). Preliminary data showed high genetic diversity on the COI gene for 19 Opecarcinus hypostegus specimens collected off Curaçao. In this study, an additional 176 specimens were sequenced and used to characterize the population structure along the leeward side of Curaçao. Extremely high COI genetic variation was observed, with 146 polymorphic sites and 187 unique haplotypes. To determine the cause of this high genetic diversity, various gene flow scenarios (geographical distance along the coast, genetic partitioning over depth, and genetic differentiation by coral host) were examined. Adaptive genetic divergence across Agariciidae host species is suggested to be the main cause for the observed high intra-specific variance, hypothesised as early signs of speciation in O. hypostegus.


Patterns of polymorphism.
A 675 base pairs long fragment of the COI region was sequenced for a total of 195 individuals (Table S1). Across all collection sites, 146 nucleotide sites were polymorphic, yielding 187 unique haplotypes (h = 0.9994). Of these, 123 were third codon position changes, along with 23 first codon position changes and zero second codon position changes. Overall nucleotide diversity (π) = 0.02558 ( Table 1). Translation of the sequences to amino acid data revealed only five polymorphisms in five individuals (RMNH. Crus.D.57581, 57456, 57557, 57559 and 57476; Table S1), all at different positions of the sequence. An Automatic Barcode Gap Discovery (ABGD) analysis shows that only one Molecular Operational Taxonomic Unit is present in O. hypostegus.
Depth differentiation. A mantel test was used to test for a relationship between genetic differentiation (Φ st) and the difference in depth of collection (Table S1) between each sample, but no significant relationship was found (r = 0.1063, P = 0.1426). Hence, there is no statistical evidence for genetic isolation over depth ( Fig. 2A).
Partitioning the IBD analysis into individuals collected from the same host coral species had no effect on the outcome. No significant relationship was found between genetic differentiation (Φ st) and depth for individuals sampled from A. agaricites (r = − 0.1604, P = 0.8599) (Fig. 2B), nor for individuals sampled from A. lamarcki (r = 0.0465, P = 0.3372) (Fig. 2C). For the individuals sampled from the remaining host coral species, A. humilis, A. grahamae and A. fragilis, population sample sizes were too small or too few populations were sampled to perform a valid IBD analysis (Table S1).  Table 3). No significant genetic differentiation was measured between individuals collected from A. grahamae (n = 4) and A. humilis (n = 7), possibly due to the small samples sizes. These results are in concordance with the median joining network (Fig. 3) that reveals a large mutation distance between individuals inhabiting A. humilis or A. agaricites and individuals collected from the other host corals. The largest mutation distance was found between O. hypostegus individuals inhabiting the coral hosts A. humilis and A. agaricites (Fig. 3). The

Discussion
Patterns of polymorphism. In the COI sequence data of the 195 Opecarcinus hypostegus specimens collected from Curaçao, 146 COI polymorphic sites were found and 187 unique haplotypes (Table 1). Strikingly, in Cryptochiridae collected from various locations in the Indo-Pacific hardly any polymorphic sites are present on the COI gene, even over distances as large as between the Indo-Malayan region and New Caledonia 24 , or the Red Sea and Japan 36 . Generally, COI sequence data shows high resolution at species level, and work well as a barcoding marker. First and second COI codon positions are highly conserved, whereas third codon positions can evolve rapidly, making this locus a common choice for population genetics and phylogeography 38,39 . As expected, almost all variation in O. hypostegus was found on the third codon position. When translated to amino acids, however, only five polymorphisms at five different positions were retrieved, hence there are no cryptic species present in O. hypostegus. This result was confirmed by the ABGD analysis. Extreme levels of genetic variation have been reported within natural populations in, for example, planktonic chaetognaths (arrow worms) 40 and mesopelagic shrimp 41 . Many instances of high mitochondrial diversity have directly or indirectly been interpreted as evidence of cryptic speciation, but some of these cases may need to be subjected to re-evaluation when investigated using nuclear loci 40,42 .
The genetic diversity obtained in this study (mean h = 0.9994, mean π = 0.02558), from a very small geographic area, can be classified as an extreme level of intra-specific variance compared to the reported mean and median values for haplotype (0.63388 and 0.70130) and nucleotide diversity (0.00388 and 0.00356) for 23 animal species in a meta-analysis 43 , which showed a positive, non-linear relationship between the population-level estimates of h and π. The values obtained in our study strongly deviate from their values, with π being much higher. A combination of high nucleotide and haplotype diversities has been linked to large stable populations with a long evolutionary history and possible secondary contact between differentiated lineages 44 . In contrast, the negative indexes of neutrality Fu and Li's F, and Tajima's D indicate a departure from neutral processes, which can be caused by demographic changes or selective events. Due to the non-significance of these values the hypothesis of neutrality can, however, not be rejected.
Small scale geographical genetic differentiation. The leeward side of Curaçao is about 65 km long from southeast to northwest. For the seaweed Sargassum polyceratium Montagne, 1837, fine-scale differentiation was retrieved around Curaçao, with bays showing significant differentiation from each other 21 . A Mantel test revealed a relationship between the genetic similarity of certain individuals and geographical distance for O. hypostegus (Fig. 1A). Although the relationship was weak and not highly significant, this suggests a genetic structure within O. hypostegus individuals living in close spatial proximity being more genetically similar than expected under a random distribution of genotypes. Splitting the sample into groups of individuals collected from the same host coral species increased both the magnitude and significance of the isolation-by-distance pattern for individuals inhabiting Agaricia agaricites (Fig. 1B). For A. lamarcki, no statistical evidence for isolation-by-distance was retrieved (Fig. 1C). This difference may be explained by the higher abundance of A. agaricites corals off Curaçao, compared to A. lamarcki 34 , providing suitable habitat closer to the natal site of the O. hypostegus larvae settling on A. agaricites. Furthermore, A. lamarcki has a wider depth distribution than A. agaricites 19 , which may influence the isolation-by-distance results.
Genetic partitioning over depth. No statistical evidence was found for genetic differentiation over depth in O. hypostegus ( Fig. 2A), at least not within the studied depth range of this study (5-38 m). Partitioning the analysis into groups of individuals collected from the same host coral had no effect on the outcome, and individuals inhabiting A. agaricites or A. lamarcki did not show any significant genetic differentiation over depth (Fig. 2B,C).
Due to the technical limitations of scientific diving, our sampling was restricted to a maximum of 38 m depth. The depth distribution of O. hypostegus is, however, known to extend to the mesophotic zone (ca. 60 m), where an O. hypostegus individual was observed inhabiting an A. lamarcki coral 33 . We may argue that sampling over a depth range that is at least twice as large as in the present study, would reflect the total O. hypostegus distribution more completely. This could increase the likelihood of revealing genetic differentiation over depth, because sampled individuals living over a larger depth distribution face more variable environmental conditions. Genetic differentiation over coral hosts. Under an ecology-driven gene flow scenario, gene flow may be strongest among similar environments 6,[45][46][47][48][49] . This pattern may arise through mechanisms such as selection and local adaptation that will disrupt the patterns of isolation-by-distance 50 or as a consequence of selection against maladapted immigrants from different environments 6 . In our study, significant genetic subdivisions between individuals inhabiting different host coral species were observed (Tables 2 and 3; Figs 3 and 4). There was statistical evidence for diversification across host coral species (i.e. environment) in O. hypostegus, which is expected to be an important alternative strategy to direct competition for the same host in Cryptochiridae 23 .
Cryptochirid males "visit" females inhabiting separate galls or pits, dubbed the "visiting" mating system 29,51 . It is unclear how far a male gall crab can travel to find a female partner. Many male and female gall crabs can inhabit the same coral colonies, especially if these are large in size. If a male O. hypostegus mates with females on the same coral colony, a genetic preference for that coral host might end up getting fixed in a population. The study on  (Table S1), the colours indicate the coral host species and correspond with those in Fig. 3. gall crab occurrence rates on the leeward side of Curaçao revealed significant higher O. hypostegus prevalence in A. lamarcki compared to A. agaricites and A. humilis, suggesting a preference for inhabiting A. lamarcki 34 .
Phylogenetic results showed in the median joining network (Fig. 3) and phylogenetic tree (Fig. 4) support the genetic differentiation across host species in O. hypostegus, with distinct clustering of individuals inhabiting the hosts A. lamarcki, A. agaricites and A. humilis. In the haplotype network, the observed groupings show different patterns. A star-shaped burst pattern (interlinked haplotypes with few mutation steps between them) can be observed for the individuals inhabiting A. agaricites (Fig. 3) 44 . These patterns appear due to high numbers of low frequency alleles with small average pair-wise distances between them, and may be evidence of a recent expansion from a small number of ancestors. In addition, the nucleotide diversity of specimens inhabiting A. agaricites is lower in comparison to other coral hosts (Table 1). In an expanding population, haplotype diversity and number of polymorphic sites can quickly increase, while nucleotide diversity usually lags behind. Indeed, high levels of h with moderate to low levels of π have frequently been attributed to recent divergence in marine species 52 . Over time, when a population stops expanding and starts to stabilize, nucleotide diversity will increase. Newly created low frequency haplotypes either increase in the population or are lost, which increases the average number of segregating sites between haplotypes over time 44 . In the A. lamarcki grouping, a higher number of segregating sites is observed between the haplotypes (Fig. 3), suggesting that this population is stabilizing.
A study on the historical evolutionary patterns of cryptochirids has indicated that the phylogeny of coral gall crabs is directed by the evolution of their scleractinian hosts 23 . For yet unknown reasons, Indo-Pacific gall crab species show stricter host-specificity patterns than their Atlantic counterparts 33 . The congeners of O. hypostegus in the Indo-Pacific are highly host specific and are often associated with one or several closely related coral species 53,54 . Presumably, the current genetic diversification across host corals found in O. hypostegus may not only result in a stronger local-adaptation to ecological differences between coral hosts over time, but might even be strong enough to eventually foster speciation.

Concluding remarks
The main objective of the present study was to examine which spatial and ecological factors influence Opecarcinus hypostegus gene flow off Curaçao and can explain the observed high genetic diversity at the COI gene. Factors that were expected to influence the Opecarcinus hypostegus gene flow patterns were examined; we found a weak relationship for geographical genetic differentiation (mostly for the host coral A. agaricites) and no evidence for genetic differentiation over depth. The observed clustering in the haplotype network and phylogenetic tree, however, suggests that adaptive divergence over the coral hosts is present. We hypothesise that this divergence is an early sign of (sympatric) speciation. This divergence might result in several closely related species of Opecarcinus inhabiting Agaricia corals in the Caribbean, in a similar way to its congeners inhabiting various closely related Agariciidae corals in the Indo-Pacific 53,54 . To further test this hypothesis, data from additional markers and localities is needed.

Materials and Methods
Field sampling. During field surveys in 2013 (16 Oct-9 Nov) and 2014 (12 Mar-28 Apr), specimens of Opecarcinus hypostegus were collected by chiselling off a small piece of their agariciid coral host from depths between 5 and 38 m at 29 localities on the leeward side of Curaçao (Dutch Caribbean, southern part of the Caribbean Sea). Four samples were collected at ca. 20 m depth from the island Klein Curaçao, located approximately 10 kilometres southeast of Curaçao (Fig. S1, Table 4, S1). The corals were visually identified to species level during the surveys using field guides, Coralpedia (http://coralpedia.bio.warwick.ac.uk) and the Coral IDC tool (http://www.researchstationcarmabi.org) 55 . The leeward side of Curaçao stretches some 65 km from southeast to northwest with an almost continuous coral reef, providing uninterrupted suitable habitat for gall crab larvae settlement in which no clear geographical barriers appear to occur. In total, 210 O. hypostegus gall crab samples were collected from five Agaricia host coral species. Crabs were preserved in ethanol (80% in 2013, 96% in 2014).    (Table S1).  Table 4. Detailed sampling locality data including: locality codes, site name, collection site coordinates, number of collected samples (N), number of polymorphic sites (P) and nucleotide diversity (π) per locality.

Molecular analyses.
Scientific RepoRts | 7:39461 | DOI: 10.1038/srep39461 differentiation and geographical distance can detect an isolation-by-distance pattern. A Mantel's test was performed in the program IBDWS v. 3.23 68 using Φ st as a measure of genetic differentiation, which incorporates sequence distance information. Significance was determined by permuting the data 30,000 times. The population structure was described with an analysis of molecular variance method (AMOVA) 69 implemented in ARLEQUIN 70 . Significance was determined with 10,000 random permutations of the data. Arlequin was also used to calculate the pairwise Fst values between individuals collected from different host coral species. The web version of ABGD 71 was used to estimate the genetic distance corresponding to the difference between a speciation process versus intra-specific variation in O. hypostegus. Runs were performed using the default range of priors (pmin = 0.001, pmax = 0.10) using the JC69 Jukes-Cantor measure of distance. The analysis involved 195 sequences with a total of 675 positions in the final dataset.