Genetic hypervariability of a Northeastern Atlantic venomous rockfish

Background Understanding the interplay between climate and current and historical factors shaping genetic diversity is pivotal to infer changes in marine species range and communities’ composition. A phylogeographical break between the Atlantic and the Mediterranean has been documented for several marine organisms, translating into limited dispersal between the two basins. Methods In this study, we screened the intraspecific diversity of 150 individuals of the Madeira rockfish (Scorpaena maderensis) across its distributional range (seven sampling locations in the Atlantic and Mediterranean basins) using the mitochondrial control region and the nuclear S7 first intron. Results The present work is the most comprehensive study done for this species, yielding no genetic structure across sampled locations and no detectable Atlantic-Mediterranean break in connectivity. Our results reveal deep and hyper-diverse bush-like genealogies with large numbers of singletons and very few shared haplotypes. The genetic hyper-diversity found for the Madeira rockfish is relatively uncommon in rocky coastal species, whose dispersal capability is limited by local oceanographic patterns. The effect of climate warming on the distribution of the species is discussed.

One of the evident effects of climate change is the increase in seawater temperature, which translates into a global tropicalization trend (e.g., Bianchi & Morri, 2003;Wernberg et al., 2016) that also affects the Northeastern Atlantic. Along the southern and western coasts of the Iberian Peninsula, several organisms are expanding their poleward distribution, some with great impact in community composition (e.g., Lourenço et al., 2012;Nicastro et al., 2013;Bode et al., 2020;Robalo et al., 2020). One of the fish species recently reported off southwestern continental Portugal is the Madeira rockfish, Scorpaena maderensis Valenciennes 1833 (Encarnação et al., 2019), an estimated moderately vulnerable species (36/100) according to the model by Cheung, Pitcher & Pauly (2005). The IUCN Red List highlights the unknown current population trend of this least concerned species (Nunoo et al., 2015). Its genetic characterization and phylogeography have therefore become imperative.
The Madeira rockfish, Scorpaena maderensis, is distributed in the eastern Atlantic, including the islands of Azores, Madeira, Canaries and Cape Verde and in the Mediterranean Sea (Hureau & Litvinenko, 1986;Eschmeyer et al., 1990;Froese & Pauly, 2019). The species is a benthic sedentary species, mostly occupying shallow coastal areas with rocky bottoms and estuaries (usually underneath boulders or in crevices). Congeneric species, S. scrofa and S. porcus, have high residency and narrow home ranges (Özgül et al., 2019). Based on the present-day known distribution, mainly of subtropical nature, the estimated seawater temperature range for the species is 16-25 C (Kaschner et al., 2019), but there are no studies on the thermal tolerance of the species. The Madeira rockfish is a generalized and opportunistic feeder of benthic or epibenthic crustaceans and, occasionally, algae, gastropods, polychaetes and small fishes (La Mesa, La Mesa & Tomassetti, 2007), and shows sexual dimorphism in growth rate, maximum size and longevity, with differences registered between the Mediterranean and Azorean populations (Morato et al., 2001;La Mesa, La Mesa & Micalizzi, 2005). A specialized mode of oviparity is described for the genus and the eggs are deposited as a whole in a protective gelatinous matrix that facilitates spawning cohesiveness and floatation (Wourms & Lombardi, 1992). The spawning season takes place from December to February in the Mediterranean (La Mesa, La Mesa & Micalizzi, 2005) and from March to June in the Azores (Costa, 2007). However, structural features of its biology are yet to be clarified, particularly the ones related to reproduction and early-life traits.
The present work is the first population study for this species, comprising a wide sampling coverage of its distribution range (seven locations from the Atlantic and the Mediterranean Sea), and two molecular markers (mitochondrial and nuclear) to screen the genetic diversity of S. maderensis with the following objectives: (1) evaluate the genetic diversity within and among locations; (2) assess the population genetic structure of the Madeira rockfish; and (3) evaluate the putative existence of a soft barrier between the Atlantic and the Mediterranean populations.

MATERIALS AND METHODS Sampling
Specimens of S. maderensis were collected from seven locations across its distributional range in the Atlantic and Mediterranean: Cyprus (CYP), Greece (GRE-Euboea), Sicily (SIC-Messina, Riposto and Siracusa; Italy), Azores (AZO-Faial; Portugal), Madeira (MAD-Funchal; Portugal), Selvagens (SEL; Portugal) and Canaries (CAN-Tenerife; Spain) ( Fig.1 and Table 1). Specimens were provided by fishers as the species is a frequent by-catch in coastal short-range artisanal fisheries and fins were clipped after assessing the species identification for each individual. Samples were preserved in 96% ethanol and deposited in ISPA-IU/MARE tissue collection.

DNA extraction, amplification and sequencing
Total genomic DNA was extracted with the REDExtract-N-Amp Kit (Sigma-Aldrich, St. Louis, MO, USA) following the manufacturer's instructions. The mitochondrial control region (CR) and the first intron of the nuclear S7 ribosomal protein gene (S7) were amplified, in a Bio-Rad Mycycler thermal cycler, using primers L-pro1 and H-DL1 (Ostellari et al., 1996), and S7RPEX1F and S7RPEX2R (Chow & Hazama, 1998). The PCR protocol was performed in a 20 ml total reaction volume with 10 ml of REDExtract-N-ampl PCR mix (Sigma-Aldrich, St. Louis, MO, USA), 0.8 ml of each primer (10 mM), 4.4 ml of Sigma water and 4 ml of template DNA using the following PCR conditions: initial denaturation at 94 C for 7′, followed by 35/30 cycles (denaturation at 94 C for 30/45″, annealing at 55 C for 30/45″, and extension at 72 C for 1′; values CR/S7, respectively) and a final extension at 72 C for 7′. The forward primers (L-pro1 and S7RPEX1F) were used for the sequencing reaction, and the PCR products were purified and sequenced in STABVIDA (http://www.stabvida.net/). Chromatograms were manually checked, edited with Codon Code Aligner (Codon Code Corporation, http://www.codoncode.com/index.htm) and sequences were aligned with Clustal X 2.1 (Larkin et al., 2007). For S7, chromatograms were checked for double peaks (see Fig. S1 in Supplemental Materials) and, whenever possible, both strands of the same specimen were recovered following the approach of Sousa-Santos et al. (2005). All sequences were deposited in GenBank (Accession numbers MN716857-MN717002; and MN717003-MN717124, respectively for CR and S7) (Table S1 in Supplementary Materials).

Molecular data analyses
The genetic diversity and population structure of S. maderensis were assessed using several packages developed for R v.4.0.2 (R Core Team, 2020), in RStudio (RStudio Team, 2020). We used haplotypes (Aktas, 2020) and pegas (Paradis, 2010) R-packages to estimate standard descriptive measures of genetic diversity, including number of haplotypes and private haplotypes, haplotype diversity (h, Nei, 1987) and nucleotide diversity (π, Nei, 1987) and respective standard deviations. The software HP-Rare (Kalinowski, 2005) was used to estimate allelic richness (R) and private allelic richness (pR), using rarefaction to correct for sample-size bias associated with the relative abundance or easiness to collect samples of this species. For the S7 gene fragment the programme ARLEQUIN v3.5 (Excoffier & Lischer, 2010) was used to reconstruct the haplotypes with the ELB algorithm (Excoffier, Laval & Balding, 2003), and to perform the exact probability tests for deviations from the Hardy-Weinberg equilibrium (HWE) (Guo & Thompson, 1992). The same software was used to assess population structure, performing analyses of molecular   (Weir & Cockerham, 1984), Nei's G ST (Nei, 1987), Hedrick's G' ST (Hedrick, 2005)) and allelic differentiation (Jost's D (Jost, 2009)) measures. For both fragments, the PopART software (Leigh & Bryant, 2015) was used to build TCS haplotype networks (Clement, Posada & Crandall, 2000) based on the parsimony methodology by Templeton, Crandall & Sing (1992).

RESULTS
For the CR, a fragment of 354 bp was amplified and the 146 sequences obtained defined 105 haplotypes, with a total of 80 polymorphic sites found. Differences among haplotypes corresponded to 80 transitions, 12 transversions and 1 indel. For the S7 the 220 sequences (110 individuals) defined a total of 177 haplotypes. For this marker, the fragment obtained was 517 bp long and the differences among haplotypes corresponded to 206 mutations (70 transitions, 80 transversions and 50 indels). The S. maderensis S7 dataset, as a whole, conformed to the HWE (p = 0.998), although 36 out of the 171 polymorphic sites were in heterozygote deficit. For both fragments, the genetic diversity indices were generally very high, with little variation among collection sites (Table 1). The proportion of private haplotypes was high for all the locations (Table 1), with only 9.52% and 2.26% being shared between the Atlantic and the Mediterranean, for the CR and the S7, respectively. The obtained haplotype networks revealed deep hyper-diverse bush-like genealogies, with a large number of singletons, few shared haplotypes and no evidence for geographic structure (CR: Fig. 2, S7: Fig. 3; see also Table 1, details on haplotype composition are given in Table S1 in Supplementals Materials).
The divergence parameters yielded significant values for the overall CR (Table 2). In both markers, results from the pairwise comparisons were equivocal, with F ST showing non-significant values, Nei's G ST revealing significant values for some of the comparisons, and Hedrick's G' ST and Jost's D yielding all comparisons statistically significant (Table 2), i.e., all pairs of sampling sites usually have distinct haplotypes. In fact, eight (CR and S7) out of 22 pairwise comparisons revealed complete haplotypic differentiation (D = 1), including between some of the geographically closest sampling sites (Table 2). This high structuring was supported by the AMOVA results (CR: F ST = 0.031, p = 0.004; S7: F ST = 0.016, p = 0.005), which also revealed that variation among sampling sites accounted for only 3.08% (CR) and 1.58% (S7) of the total variation (Table S2 in Supplemental Materials).

DISCUSSION
The present work comprises a wide geographic sampling coverage of Scorpaena maderensis with locations from the Atlantic and the Mediterranean Sea and a molecular dataset with two markers. Our results highlight two main features in the population genetics of the Madeira rockfish: (1) deep hyper-diverse bush-like genealogies, characterised by large numbers of singletons and few shared haplotypes; and (2) absence of genetic structure across sampled locations, with no detectable Atlantic-Mediterranean break in connectivity. Before discussing these findings in detail, we address the main caveats concerning this study: the sampling strategy and the molecular markers used. Although most locations are represented by numbers of individuals in line with previous phylogeographic studies in marine species, one can a posteriori posit that the high number of singleton haplotypes found is biased by insufficient sampling. In fact, a recent study published by our team recorded even higher genetic diversity in a coastal fish species, revealing that it would be necessary to sample a total of 700 individuals for the sampling to be representative of the population (Robalo et al., 2020). Additionally, we have no samples from intermediate locations between the Atlantic archipelagos and the  . Another caveat is using only one mitochondrial and one nuclear marker in a day and age where next-generation sequencing producing thousands of markers are being increasingly used. This study is in line with previous research in the pursuit for patterns and processes involved in the phylogeography of the species from the North-East Atlantic (e.g., Bargelloni et al., 2005;Debes, Zachos & Hanel, 2008;Francisco et al., 2011;Robalo et al., 2013a). These previous studies used the same set of markers, allowing across species comparisons and multi-species approaches (e.g., Robalo et al., 2012;Robalo et al., 2013b;Francisco et al., 2014;Almada et al., 2017;Castilho et al., 2017) while revealing very distinct patterns.
Genetic hyper-diversity of the Madeira rockfish  Robalo, 2020;Robalo et al., 2020). Furthermore, the CR sequence hypervariability may alternatively or concomitantly be explained by the mutation rate of the fragment, the evolutionary-rates hypothesis, or the metabolic rate theory as discussed in Robalo et al. (2020).

Genetic structure of the Madeira rockfish
The present results reveal no evidence for genetic structure, geographically associated or not, and therefore we posit that the Madeira rockfish is not composed of discernible groups within the Atlantic and the Mediterranean, nor these two basins are clearly differentiated. This hypothesis is strongly dependent on the genetic markers used in the study. In studies with other species, the CR region has yielded equivocal results regarding the detection of genetic structure. We can find in the literature examples of findings of hypervariability and absence of genetic structure (e.g., Francisco et al., 2011;Mehraban et al., 2020;Song et al., 2020), and studies presenting hypervariability and significant genetic structure (e.g., Cunha et al., 2014;Castilho et al., 2017;Robalo et al., 2020). North-Eastern Atlantic past recolonization processes and historical and present dispersal movements are influenced by species-specific life-history traits, favourable oceanographic conditions, such as sea surface temperatures, and suitable recruitment habitat (e.g., Wares & Cunningham, 2001;Pappalardo et al., 2015).
The results also do not reveal any phylogeographic break between the Atlantic and the Mediterranean locations for S. maderensis (Fig. 2, Table 2), similarly to what has been previously recorded in other species (e.g., Trachurus trachurus (Comesaña, Martínez-Areal & Sanjuan, 2008), Diplodus sargus (Stefanni et al., 2015) and Dentex dentex (Viret et al., 2018)). Although many factors can explain this outcome, there are two biological characteristics that may play a relevant role: large mean pelagic larval duration and high adult dispersal capability, features common to all these species. The observed discrepancy across statistics can be attributed to their different nature (Bird et al., 2011). In cases where the geographic distribution of haplotypes is uncorrelated with the relationship among alleles, which is S. maderensis case (Figs. 2 and 3), the fixation indices will not accurately depict the structure, and the differences found among the different measures can often be uninformative to the underlying biology of population structuring.
The mean pelagic larval duration (PLD) influences on a certain degree a species dispersal potential before reaching the juvenile stages. Although it is recognized the PLD is not a universal driver of range size and therefore a promoter of connectivity in many fish (e.g., Weersing & Toonen, 2009;Selkoe & Toonen, 2011), in certain situations it seems to have some influence (Lester & Ruttenberg, 2005). To our knowledge there is no no data on the PLD of S. maderensis, but congenerics are known to spend 29 (S. porcus in Macpherson & Raventos (2006)) and 30 days (S. guttata in Carr & Reed (1993)) in the plankton, which is not a short duration. The hydrographic regime in this stretch of the North-East Atlantic is dominated by the Azores Current and its south-eastward branch, the Canary Current, a complex system of eddies (Stramma, 1984;Hernández-Guerra et al., 2001). At the Atlantic-Mediterranean transition, the eastward flowing Atlantic water describes a quasi-permanent anticyclonic gyre (Millot, 1999). The PLD and the circulation regime of the area would thus contribute to the unconstrained gene flow between the two basins and among the Macaronesian archipelagos.
Adult rockfish of the genus Scorpaena display a low active dispersal capacity (Özgül et al., 2019). Nevertheless, adults of the Madeira rockfish may perform short-distance movements. Short dispersal movements following suitable habitat may have happened, in the past decade, with individuals of this species being recorded for the first time in the Gorringe seamount (Abecasis et al., 2009) andin South Portugal (Encarnação et al., 2019), near the entrance of the Gibraltar Strait. Although a certain degree of connectivity is expected from the results of this study it would be interesting to investigate the origin of these newcomers, mainly because adult dispersal is one of the essential life-history patterns influencing connectivity and population structure (e.g., Francisco, Pereira & Robalo, 2019).
In conclusion, although no specific information is available regarding S. maderensis, its putative life-history patterns (i.e., dispersal mostly through the larval stage given the more sedentary nature of adults) is conducive to the lack of genetic structure. This lack in structure is shared by other fish groups in the same geographical areas, like gobids. A recent work on Gobius cruentatus (Čekovská et al., 2020, for additional species see references within), a species with a similar life-history pattern, has also revealed high genetic variability and no geographic structure with an estimated migration route following the main currents of the distribution area.
A meta-analysis to tackle whether or not climate change influences marine ecological phenomena found that over 80% of all observations were coherent with the expected impacts of climate change. Moreover, the rates of geographic distribution shifts were, on average, consistent with those needed to track ocean surface temperature changes (Poloczanska et al., 2013). It is expected that S. maderensis will similarly follow a trajectory compatible with its optimal physiological temperature, and therefore it may extend its geographic distribution towards north. S. maderensis is a species with both a commercial and a biotechnological interest, it would be of importance to conduct fishery census to detect the arrival of this species to new locations.

ADDITIONAL INFORMATION AND DECLARATIONS Funding
This study was funded by Fundação para a Ciencia e Tecnologia (FCT) Portugal, through the strategic projects MARE/UIDB/MAR/04292/2020 and MARE/UIDP/MAR/04292/ 2020 granted to MARE (MARE-ISPA), and UID/Multi/04326/2019 and UIDB/04326/ 2020 granted to CCMAR. This study was also supported by the University of Catania through the "PIA.CE.RI." grant 2020. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Grant Disclosures
The following grant information was disclosed by the authors:

Competing Interests
Rita Castilho is an Academic Editor for PeerJ.

Author Contributions
Sara M. Francisco conceived and designed the experiments, analyzed the data, prepared figures and/or tables, authored or reviewed drafts of the paper, and approved the final draft. Rita Castilho conceived and designed the experiments, analyzed the data, prepared figures and/or tables, authored or reviewed drafts of the paper, and approved the final draft. Cristina S. Lima performed the experiments, authored or reviewed drafts of the paper, and approved the final draft. Frederico Almada conceived and designed the experiments, performed the experiments, authored or reviewed drafts of the paper, and approved the final draft. Francisca Rodrigues performed the experiments, authored or reviewed drafts of the paper, and approved the final draft.
Radek Šanda performed the experiments, authored or reviewed drafts of the paper, and approved the final draft. Jasna Vukić performed the experiments, authored or reviewed drafts of the paper, and approved the final draft. Anna Maria Pappalardo performed the experiments, authored or reviewed drafts of the paper, and approved the final draft. Venera Ferrito performed the experiments, authored or reviewed drafts of the paper, and approved the final draft. Joana I. Robalo conceived and designed the experiments, authored or reviewed drafts of the paper, and approved the final draft.

DNA Deposition
The following information was supplied regarding the deposition of DNA sequences: All sequences are available at GenBank: MN716857 -MN717002 and MN717003 -MN717124.

Data Availability
The following information was supplied regarding data availability: The FASTA files with the alignments for CR and S7 markers are available as Supplemental Files.

Supplemental Information
Supplemental information for this article can be found online at http://dx.doi.org/10.7717/ peerj.11730#supplemental-information.