I. Calderón*, G. Giribet+, X. Turon

Two markers and one history: phylogeography of the edible common sea urchin Paracentrotus lividus in the Lusitanian region

I. Calderón*, G. Giribet+, X. Turon#

*Department of Animal Biology, Faculty of Biology, University of Barcelona, 645 Diagonal Ave, 08028 Barcelona, Spain. +Department of Organismic and Evolutionary Biology & Museum of Comparative Zoology, Harvard University, 26 Oxford Street, Cambridge, MA 02138, USA. #Center for Advanced Studies of Blanes (CSIC), C. d’Accés a la Cala S. Francesc 14, 17300 Blanes (Girona), Spain.

Corresponding author:

I. Calderón

Department of Animal Biology, Faculty of Biology, University of Barcelona, 645 Diagonal Ave, 08028 Barcelona, Spain.

Telephone: +34934021441

Fax: +34934035740

E-mail:

Abstract

Benthic marine invertebrates with long-lived larvae are believed to have dispersal capabilities that contribute to maintaining genetic uniformity among populations over large geographical scales. However, both hydrological and biological factors may limit the actual dispersal of such larvae. We studied the population genetic structure of the edible common sea urchin Paracentrotus lividus (Lamarck, 1816) to explore its dispersal patterns in the Atlanto-Mediterranean region and, more specifically, to ascertain the role of the Strait of Gibraltar in shaping the genetic structure of this species. For this purpose, we analysed 158 individuals for the mitochondrial 16S rRNA gene and 151 of these for the nuclear single-copy intron ANT (Adenine Nucleotide Transporter) from 16 localities from the Atlantic and Mediterranean basins, spanning over 4000 km. Mitochondrial 16S rRNA shows higher genetic diversity in the Mediterranean than in the Atlantic and reveals a sharp break between the populations of both basins, probably as a consequence of the barrier imposed by the Almería-Orán hydrological front, situated east of the Strait of Gibraltar. Both markers suggest that a recent population expansion has taken place in both basins, most probably following the Messinian salinity crisis.

Introduction

The last decades have witnessed an ever-increasing interest in discerning the role of historical and current processes in shaping observed genetic structure at the intraspecific level (reviewed in Avise 2000). In marine benthic invertebrates, gene flow between populations is mainly driven by dispersal of larvae that, in some cases, can remain in the water column for weeks or even months. Nevertheless, recent studies in several benthic invertebrates have shown little correlation between larval lifespan and dispersal ability (e.g., Benzie 2000; Hellberg et al. 2002; Palumbi 2004). Indeed, barriers to gene flow are not always conspicuous, especially in marine habitats where geographical barriers, currents, temporal and spatial spawning patterns (Hart and Scheibling 1998), physical and behavioural properties of larvae (Thomas 1994), juvenile mortality (Gosselin and Qian 1996, 1997; Hunt and Scheibling 1997) and many other abiotic or biotic factors can ultimately determine population structure over space and time (Moberg and Burton 2000; Sponaugle et al. 2002).

Current and historical barriers to gene flow may leave a strong footprint on population structure. A good model to study the role of such barriers on marine organisms is the Strait of Gibraltar, which constitutes the limit between two marine biogeographical regions, the north-eastern Atlantic Ocean and the Mediterranean Sea. Historically, the exchange of waters between both basins was interrupted during the Messinian salinity crisis (Maldonado 1985; Pérès 1989), which constituted one of the most dramatic events during the Cenozoic era (Duggen et al. 2003). The opening of a new connection through the Strait of Gibraltar re-established water exchanges between basins, leading to the recolonisation of the Mediterranean by organisms from the Atlantic. Besides, fluctuations in sea level during the Quaternary also produced sporadic separations between both basins (Nilsson 1982; Waelbroeck et al. 2002). Population differentiation across the Atlantic-Mediterranean divide has been described in a number of marine species (e.g., Borsa et al. 1997; Bargelloni et al. 2003; Baus et al. 2005). The concordance in intraspecific patterns in species with diverse life history traits points to the importance of the historical processes that occurred in this area. Nowadays, the so-called Almería-Orán hydrological front, situated east of the Strait (Tintore et al. 1998) still hinders migration between basins for numerous marine species (Patarnello et al. 2007 and references therein).

Sea urchins (Echinodermata, Echinoidea) are a diverse group of marine deuterostomes (Smith et al. 2006) that play an important role in structuring benthic communities (e.g., Palacín et al. 1998; Sala et al. 1998; Sivertsen 2006). The common edible sea urchin Paracentrotus lividus (Lamarck, 1816) is a commercially exploited species found in the north-eastern Atlantic and throughout the Mediterranean (Boudouresque and Verlaque 2001). Larval lifespan has been estimated between 20 and 40 days (Fenaux et al. 1985; Pedrotti 1993) and thus, individuals of this species are potentially able to disperse over long distances. Nonetheless, high spatial, bathymetric and temporal variability in settlement suggests that biotic or abiotic factors may affect dispersal (Hereu et al. 2004; Tomas et al. 2004).

Notwithstanding the vast amount of literature on the ecology and biology of Paracentrotus lividus (e.g., Savy 1987; Turon et al. 1995; López et al. 1998; Tomas et al. 2006), little is still known about its genetic structure. Iuri et al. (2007), in a study based on two mitochondrial and one nuclear markers, stated that P. lividus presents no genetic differentiation within the Gulf of Naples. At a larger scale, Duran et al. (2004) observed panmixia within Atlantic and Mediterranean basins using cytochrome c oxidase subunit I (hereafter COI) DNA sequences, but detected also a slight but significant pattern of genetic differentiation between the two basins.

The aim of our study was to obtain a more detailed picture of the population genetic structure of Paracentrotus lividus throughout the Atlanto-Mediterranean region and, especially, to ascertain the role of the Strait of Gibraltar in shaping the genetic structure of this species. In order to achieve such a goal, we sampled the common sea urchin in16 locations from the Atlantic and Mediterranean basins and analysed two molecular markers of different characteristics: a mitochondrial ribosomal gene and a nuclear intron.

Material and methods

Sixteen locations were sampled for Paracentrotus lividus by scuba along the Atlanto-Mediterranean arch (Figure 1). Distances between sampling sites ranged from 20 to around 4400 km. The gonads were dissected from live specimens, fixed in 96% ethanol and stored at –80ºC until processing.

Genomic DNA was extracted using the REALPURE extraction kit (Durviz, Spain) and two molecular markers were analysed. A fragment of the mitochondrial 16S rRNA gene was amplified for 158 individuals with universal primers 16Sa and 16Sb (Kessing et al. 1989). The single-copy intron of the nuclear Adenine Nucleotide Transporter (ANT) gene was amplified for 151 individuals by EPIC-PCR, using degenerate universal primers designed by Jarman et al. (2002). In both cases, amplifications were performed in a final volume of 25 μL using 2.5 mM of MgCl2, 1 mM of dNTPs, 0.5 μM of each primer and 1 U of Taq polymerase. For the mitochondrial 16S rRNA, 1 μL of DMSO was added per sample. PCR amplicons were vacuum-cleaned and labelled using BigDye® Terminator v.3.1 (Applied Biosystems, Branchburg, New Jersey, USA). Sequences were obtained on an ABI 3730 and 3100 Genetic Analyzer (Applied Biosystems) for 16S and ANT, respectively.

In order to reconstruct the allelic phase from the ANT genotypic data we used the program PHASE v2.1 (Stephens et al. 2001; Stephens and Scheet 2005). To confirm the existence of only two alleles per individual and to check the results provided by PHASE, PCR products of six individuals from different populations were cloned with the pGEM-Easy Vector cloning kit (Promega, Wisconsin, USA) following manufacturer’s instructions. Four to six colonies per individual were sequenced.

Population genetics analyses

Haplotype and nucleotide diversity values were calculated with DnaSP v.4.10.3 (Rozas et al. 2003). Genetix v 4.05.2 (Belkhir et al. 2004) was used to calculate haplotype frequencies and inbreeding coefficients from the data obtained for ANT with PHASE. Pairwise genetic distances (FST) for both markers were calculated with Arlequin ver. 3.1 (Excoffier et al. 2005) and their significance was assessed by performing 10,000 permutations. A Multidimensional Scaling (MDS) was performed with Systat 11 (SPSS) to graphically visualise these results.

SAMOVA 1.0 (Dupanloup et al. 2002) was used to define groups of populations that are geographically homogeneous and with the highest differentiation among each other. Analyses were performed for K=2 groups with 10,000 simulated annealing procedures. AMOVA, as implemented in Arlequin, was performed to further examine hierarchical population structure. Finally, the correlation of genetic and geographical distances was tested with the Mantel test procedure available in Arlequin.

Haplotype network

For 16S rRNA, the complete data set was used to build a median-joining network using Network v4.2.0.1 (Bandelt et al. 1999). For ANT, however, due to the high number of haplotypes found, we limited the network to populations surrounding the Strait of Gibraltar. The loops observed in the networks were solved using criteria derived from coalescent theory (Templeton et al. 1987; Templeton and Sing 1993).

Demographic analyses

Neutrality tests and mismatch distribution analyses can provide hints to infer population demographic events. Tajima’s D (Tajima 1989a), Fu’s FS (Fu 1997) and R2 (Ramos-Onsins and Rozas 2002) were calculated with DnaSP. Mismatch distributions (Rogers and Harpending 1992; Harpending 1994), as well as goodness-of-fit tests for demographic and spatial expansions, were calculated with Arlequin.

Results

Diversity and population structure

For the mitochondrial 16S rRNA, a fragment of 582 bp was sequenced from 158 individuals. A multi-T region was observed in the amplified fragment, containing from 6 to 9 Ts. Except for these, all other changes observed were substitutions. Thirty-one polymorphic sites and 38 haplotypes were observed. Of these, 16 haplotypes were present in Atlantic samples (9 of these were exclusive to this basin) whereas Mediterranean samples comprised 29 (22 haplotypes exclusive to this basin; Table 1 in Appendix I). Haplotype diversity is thus higher in the Mediterranean than in the Atlantic basin, and so is nucleotide diversity (Table 1).

For the ANT intron, 323 bp were sequenced for 151 out of the 158 individuals sequenced for 16S rRNA. Fifty-seven variable sites were observed. Cloning of six randomly chosen individuals proved that ANT was a single-copy marker and confirmed in every case the haplotype assignment provided by PHASE. The allelic reconstruction estimated 142 haplotypes (Tables 2 and 3 in Appendix I), 86 in the Atlantic (60 haplotypes exclusive to this basin) and 82 in the Mediterranean (56 of which were exclusive to this basin). Haplotype and nucleotide diversity did not differ between basins (Table 1). Twenty-nine out of 151 individuals appeared to be homozygotes, the number of homozygote individuals being evenly distributed in both basins. Eight populations showed a significant departure from Hardy-Weinberg equilibrium (Table 1).

FST for 16S rRNA showed higher levels of population differentiation between basins (mean FST = 0.109) than within the Atlantic and Mediterranean basins (means of 0.021 and –0.004 respectively; Table 4 in Appendix I). As a consequence, a sharp separation between Atlantic and Mediterranean populations was observed in the MDS representation (Figure 2a). On the contrary, differences between basins as revealed by FST measures were not as clear for the ANT data, mostly due to the high variability observed in the sampled populations (Table 4 in Appendix I). Likewise, MDS did not show a clear separation between basins (Figure 2b).

For 16S rRNA, SAMOVA for K=2 clustered Atlantic populations, plus Ceuta and Tarifa, in one group, and the remaining Mediterranean populations in another group. Ceuta and Tarifa, although located in the Mediterranean Sea, are placed west of the Almería-Orán hydrological front. Results from AMOVA showed a significant variance component associated with the differentiation between basins (11.20%, considering Ceuta and Tarifa as Atlantic populations). The variance between populations within basins was low and not significant (0.08%), but increased to 6.39% and became significant when no groups were specified (Table 2a). The Mantel test showed a significant correlation coefficient between genetic and geographical distances (r=0.395, p=0.009) for the whole sample. However, neither the correlation coefficients estimated within basins (r=0.05 and r=0.225 for the Atlantic and the Mediterranean, respectively) nor the coefficient for between-basins pairs alone (r=0.184) were significant, indicating that the overall significant results may, in fact, be an artefact. Indeed, this outcome may stem from the fact that the global analysis lumped together comparisons within basins, which corresponded to populations poorly differentiated and geographically close, with between-basin comparisons, which relied on populations that tended to be more divergent and widely separated. This suggests that there is no isolation by distance between our basins, but only a sharp genetic break at the Almería-Orán front.

As for ANT, SAMOVA showed a clear differentiation between Nao and all other populations when K=2 groups. When K>2, populations were separated one by one from the main group and the sharp distinction between the Atlantic and Mediterranean basins observed for 16S rRNA was not observed. For the sake of comparison, we computed AMOVA (and subsequent analyses) with the same Atlantic and Mediterranean groups as for 16S rRNA. A much higher variability within populations was observed, accounting for more than 97% of the overall variation (Table 2b). Variation among populations within basins was in general small but significant (between 2 and 3%) whereas variation between groups (basins) was not significant. As expected, the Mantel test provided a smaller, non-significant correlation coefficient (r=0.137, p>0.05).

Haplotype network

The network obtained for the 16S rRNA data suggests that haplotype 1 is the ancestral haplotype due to its high frequency, its wide geographical distribution and its central position in the network. All haplotypes are separated by a few mutational steps (Figure 3). For ANT, the haplotype network was only built for the 4 populations around the Strait of Gibraltar (Cádiz, Ceuta, Tarifa and Gata), comprising 78 individuals representing 51 haplotypes. The single network obtained presented a high amount of loops that were almost always unambiguously resolved (Figure S1).

Demographic analyses

Neutrality tests for the 16S rRNA detected a population expansion for the whole sample set, with significant values for all three tests. However, only the Fs and R2 statistics detected significant expansions within each basin (Table 3). In the case of ANT, neutrality tests provided the same results observed for 16S rRNA, although R2 did not detect a significant expansion for the Atlantic basin (Table 3).