The Atlantic–Mediterranean transition: Discordant genetic patterns in two seabream species, Diplodus puntazzo (Cetti) and Diplodus sargus (L.)

Molecular Phylogenetics and Evolution 36 (2005) 523–535 www.elsevier.com/locate/ympev

The Atlantic–Mediterranean transition: Discordant genetic patterns in two seabream species, Diplodus puntazzo (Cetti) and Diplodus sargus (L.) L. Bargelloni a, J.A. Alarcon b, M.C. Alvarez b, E. Penzo a, A. Magoulas c, J. Palma d, T. Patarnello a,e,¤ a Dipartimento di Biologia, Università di Padova, Via G. Colombo, 3, I-35121 Padova, Italy Departamento de Genética, Facultad de Ciencias, Universidad de Málaga, 29071, Málaga, Spain c Institute of Marine Biology of Crete, P.O. Box 2214, 71003 Iraklio, Greece d CCmar, Universidade do Algarve, UCTRA, Campus de Gambelas, 8000-810 Faro, Portugal e Facoltà di Medicina Veterinaria-Agripolis, Università di Padova, Via Romea 16 I-35020 Legnaro, Italy b

Received 29 June 2004; revised 5 April 2005 Available online 4 June 2005

Abstract Sparids are a group of demersal perciform Wsh of high commercial value, which have experienced an extensive radiation, particularly in the Mediterranean, where they occupy a variety of diVerent niches. The present study focuses on two species: Diplodus sargus and D. puntazzo, presenting a wide distribution from the Mediterranean to the eastern Atlantic coasts. They display similar ecological behaviour and are evolutionary closely related. Both are highly appreciated in Wsheries and D. puntazzo is currently under domestication process. However, little is know on their population structure and it is an open question whether any genetic diVerentiation exists at the geographic level. To address this issue we examined sequence variation of a portion of the mitochondrial DNA (mtDNA) control region in population samples of each of the two species collected over a wide geographic range. In addition to the mtDNA, analysis of nuclear loci (allozymes) was included in the study to compare patterns revealed by nuclear and mitochondrial markers. The studied samples covered an area from the eastern Mediterranean to the Portuguese coasts immediately outside the Gibraltar Strait. The two species revealed a level of sequence polymorphism remarkably diVerent for the control region with the D. puntazzo and D. sargus showing 111 and 28 haplotypes, respectively. Such a diVerence was not detected with allozyme markers. The two species also showed large diVerences in their population structure. While D. puntazzo presented a marked genetic divergence between the Atlantic and Mediterranean samples, D. sargus showed little intraspeciWc diVerentiation. These results were supported using both mtDNA and allozyme markers, and were interpreted as the consequence of diVerences in the history of the two species such as Xuctuations in the eVective population size due to bottlenecks and expansions, possibly combined with present-day diVerences in levels of gene Xow.  2005 Elsevier Inc. All rights reserved. Keywords: Sparid Wsh; Phylogeography; Atlantic–Mediterranean transition; mtDNA; Allozymes

1. Introduction

*

Corresponding author. Fax: +39 0498276209. E-mail addresses: [email protected] (M.C. Alvarez), magoulas@ ns0.imbc.gr (A. Magoulas), [email protected] (J. Palma), [email protected] (T. Patarnello). 1055-7903/$ - see front matter  2005 Elsevier Inc. All rights reserved. doi:10.1016/j.ympev.2005.04.017

The perciform family Sparidae comprises more than 100 species worldwide, with a peak of diversity in the Northeast Atlantic and the Mediterranean, where 24 species, belonging to 11 genera, have been described (Bauchot and Hureau, 1986). Sparids, or seabreams, are demersal Wshes living in coastal waters and occupying a

524

L. Bargelloni et al. / Molecular Phylogenetics and Evolution 36 (2005) 523–535

variety of trophic niches; they are generally gregarious, though adults might be solitary in some species. In the Mediterranean area, sparids are of great interest for Wsheries and aquaculture. Most sparid species are, in fact, excellent foodWshes, with high commercial value. In the recent years, seabreams have also gained considerable importance for aquaculture. For instance, the gilthead seabream, Sparus aurata, has become one of the most important cultured species in the Mediterranean region. Several other sparids are cultivated in Wsh farms, or potential candidates for aquaculture. Despite the growing interest in this group of Wshes, however, many aspects of their biology remain unknown. This is particularly worrying if we consider that a correct management of biological resources should be grounded on the most complete information about the natural genetic diversity of the species involved. While interspeciWc relationships within the family Sparidae have been investigated also at the molecular level (Hanel and Sturmbauer, 2000; Orrell and Carpenter, 2004), little is known about the partition of DNA polymorphism among geographic populations within the same species. The present work is aimed at investigating the population structure of two sparid species of the genus Diplodus with relevance both in Wshery and aquaculture,

namely D. puntazzo (the sharpsnout seabream) and D. sargus (the white seabream). Population samples were collected at four geographic locations, three into the Mediterranean Sea (Greece, Italy, Spain) and one into Atlantic waters immediately outside the Strait of Gibraltar strait (Faro, Portugal) roughly covering three geographic areas : the Northeast Atlantic just outside the strait of Gibraltar, the Western and the Eastern Mediterranean Sea (Fig. 1). These samples span only partially the natural distribution range of the species. D. puntazzo is present in the Mediterranean Sea, the Black Sea, and Eastern Atlantic (oV the African costs). D. sargus was described as a species complex which includes D. s. sargus in the Mediterranean and Black Sea, D. s. cadenati in the Eastern Atlantic (from Bay of Biscay to Senegal including Canary and Azores Islands), D. s. capensis (from Angola to Mozambique and Southern Madagascar) and D. s. lineatus endemic to the Cape Verde (De la Paz et al., 1973). However, to avoid confusion, we will use in the present work the species name D. sargus rather than the sub-species one since, on the base of the geographic origin of the investigated populations, we should name D. s. sargus the Mediterranean samples whereas the Atlantic one (Faro) should be considered as part of the sub-spececies D. sargus cadenati (Summer et al., 2001).

Almeria

Fig. 1. Sampling locations for D. sargus and D. puntazzo. Samples were collected directly or through local Wshermen in Faro (Portugal, FAR), Alicante (Spain, ALI), Otranto (Italy, OTR), Mesolongi (Greece, MSL, D. puntazzo), Iraklion (Greece, IRK, D. sargus). Sample sizes are indicated within squares for D. sargus, within circles for D. puntazzo. The circulation in the Alboran Sea and location of the Orian-Almeria Front (OAF) are also shown.

L. Bargelloni et al. / Molecular Phylogenetics and Evolution 36 (2005) 523–535

Despite the limited sampling scheme, that covers only the northernmost distribution of the species, it was particularly important to have obtained samples from both sides of the Gibraltar Straits. In fact, this divide has been hypothesised to represent the boundary between the two biogeographic provinces, the north-east Atlantic and the Mediterranean (Quignard, 1978), while studies on intraspeciWc genetic variation in a variety of organisms have showed, for several of the examined species, a reduction of gene Xow in relation to the transition between the Mediterranean and the Atlantic (reviewed in Borsa et al., 1997; but see also Bargelloni et al., 2003; Naciri et al., 1999; Pannacciulli et al., 1997; Perez-Losada et al., 1999; Zane et al., 2000). We report here data on allozyme and mtDNA variation in D. puntazzo and D. sargus: both species are very common throughout the Mediterranean Sea, and are quite similar in terms of habitat preference and life history. We show that, despite the aforementioned similarities, when genetic variation is examined at the intraspeciWc level, both nuclear and mitochondrial markers suggest a marked diVerence between the two seabream species. D. puntazzo show a strong genetic divergence between Atlantic and Mediterranean samples, while for D. sargus there is no evidence for diVerentiation when individuals from these two geographic regions are compared. Such discordant results are discussed in light of the historical and present-day features that characterise the Atlantic–Mediterranean biogeographic region(s), we also considered the Wne diVerences in ecology between the two species, that might have an impact on microevolutionary processes.

2. Materials and methods 2.1. Collection of samples Adult specimens of both species were collected from the wild at Wve diVerent locations (see Fig. 1). Individuals were shipped in dry ice to the lab, where they were kept at ¡40 °C until analysis. Eye, liver, and muscle tissue were removed for protein and DNA analyses. Due to practical and technical reasons, data on genetic variation from the two species diVer in terms of sample size, number of scored allozyme loci, and length of sequenced mtDNA region. To allow for a better comparison between the two species, for D. puntazzo, a reduced data set was obtained as follows: sample size was made equal to that of D. sargus by randomly selecting the appropriate number of individuals, allozyme loci were reduced to the same ones scored for D. sargus, and sequence length was shortened to match length and position of the fragment analysed by SSCP in D. sargus.

525

2.2. Allozymes scoring Frozen samples were subjected to horizontal starch gel electrophoresis. Electrophoresis protocols, staining procedures, genetic interpretation of zymogram patterns, and locus designation, were done according to Reina et al. (1994). Scored loci for D. sargus were: adenilate kinase (AK*), alcohol dehydrogenase (ADH*), endopeptidase (ENDOP), esterase (EST), glucosephosphate isomerase (two loci: GPI-1*, GPI-2*), glycerol-3-phosphate dehydrogenase (G3PDH*), lactate dehydrogenase (three loci: LDH-1*, LDH-2*, LDH-3*), malate dehydrogenase (three loci: MDH-1*, MDH-2*, MDH-3*), malic enzyme (two loci: MEP-1*, MEP-2*), phosphoglucomutase (PGM*), 6-phosphogluconate dehydrogenase (PGDH*), superoxide dismutase (two loci: SOD-1*, SOD-2*). For D. puntazzo, in addition to the above mentioned loci, two other allozyme systems were scored: adenosine deaminase (ADA*) and iditol dehydrogenase (IDDH*). Raw allozyme data are reported in Appendix A. 2.3. MtDNA scoring Total genomic DNA was extracted for each individual from few milligrams of muscle tissue using a protocol based on Chelex resin (Walsh et al., 1991) with some modiWcations (Ostellari et al., 1996). Two microliters of Chelex extracted DNA, were used in 20
l of PCR mix containing 0.5 U of Taq polymerase (Promega), 1£ reaction buVer, 2.5 mM MgCl2, 50
M of each dNTPs and 250 nM of each primer. PCR primers were designed based on previously obtained sequence information (i.e., 400–500 bp of 5⬘ end of the mitochondrial D-loop region, for eight sparid species), as described in Ostellari et al. (1996). Primers were speciesspeciWc (accessible through Accession No. AF373417– 373527 and AF373528–373555) and allowed amplifying a fragment of approximately 220 bp. Individual PCR products were then subjected to single strand conformation polymorphism (SSCP) analysis as previously described (Ostellari et al., 1996). After the Wrst electrophoretic run, PCR products showing similar mobility patterns were run side by side and compared to already identiWed mobility classes. On average, each sample was run two-three times under identical conditions. A few individuals from each mobility group (1–10, depending on SSCP group frequency) were randomly chosen, and their nucleotide sequence was determined as follows: a large fragment (400 bp) of the control region was PCR ampliWed for each selected individual, using primers speciWcally designed for sparid species (Ostellari et al., 1996). An aliquot (5
l) of each PCR was then run on a 1.8% agarose gel to verify the quality and quantity of the PCR product obtained. A variable amount (5–20 ng) of the ampliWed DNA was directly sequenced, without further puriWcation, using a cycle-sequencing commercial

526

L. Bargelloni et al. / Molecular Phylogenetics and Evolution 36 (2005) 523–535

kit (Amersham Premixed Cycle-sequencing kit). A Xuorescent (Texas Red 5⬘ labelling) internal oligo was used as sequencing primer. Sequencing products were analysed on an automatic DNA sequencer (Amersham Vistra 725). The use of an internal primer and the fact that the sequenced region was larger (about 300 bp) than the SSCP fragment, allowed to determine without ambiguities the complete sequence of the region analysed by means of SSCP. Computer-produced DNA data were aligned using ClustalX (Thompson et al., 1997) with default settings, and individual sequences were re-examined manually by visual inspection of raw Xuorogramdata, with special attention being paid to those sites interested by nucleotide substitutions. MtDNA data for D. puntazzo and D. sargus are presented in Appendix B and C, respectively. 2.4. Data analysis Diversity indexes for allozymes and mtDNA were calculated using the program Arlequin (ver. 2.000, Schneider et al., 2000), and Genepop (ver. 3.1d) (Raymond and Rousset, 1995a). The latter software package was used to perform an exact test of Hardy–Weinberg proportions for multiple alleles (Guo and Thompson, 1992). A similar approach was used to assess independence of allozyme loci. DiVerentiation between population samples was tested by means of an exact test of population diVerentiation based on allelic frequencies (Raymond and Rousset, 1995b). Similarly, Fst values (single locus and multilocus) were estimated both for population pairs and for all populations, using a “weighted” analysis of variance (Weir and Cockerham, 1984). In all tests, signiWcantly deviations from the null hypothesis of genetic homogeneity were assessed by means of a non-parametric approach with 100,000 permutations (reshuZing individuals among populations). For mtDNA, Fst values were estimated either taking into account haplotype frequencies only or consid-

ering also genetic distances between haplotypes. Molecular distances were estimated as uncorrected pairwise nucleotide diVerences. Rate heterogeneity among sites was also evaluated, as described in Bargelloni et al. (2000). Phylogenetic relationships among haplotypes were reconstructed using diVerent approaches. A reduced median network (MN, Bandelt et al., 1995) of mtDNA haplotypes was constructed using the program Network (ver. 2.0b http://www.Xuxus-engineering.com). Insertion– deletion events were recoded as transversions. MtDNA sequence data were also tested for departures from mutation-drift equilibrium, using Tajima’s (1989) D statistics as implemented in Arlequin 2000. A mismatch analysis (Rogers and Harpending, 1992) was also applied. Nucleotide diVerences are counted in each possible pairwise comparison between individual sequences. Frequencies for each mismatch class (0, 1, 2, 3,ƒ diVerences) are then plotted on a graph. Demographic expansions are expected to generate “waves” in the mismatch distribution, while a stable population size should show a less bell-shaped distribution (Rogers and Harpending, 1992). Mismatch distribution for each population was calculated using Taijma and Nei model to estimate diVerences between haplotypes, parameters of the expansion
0,
1,  were estimated using a non linear least square method (Schneider and ExcoYer, 1999) as implemented in the program Arlequin 2000.

3. Results 3.1. Genetic diversity Estimates of genetic variability for both species are reported in Table 1. As for D. puntazzo, results on genetic diversity and population diVerentiation were similar when analysing the complete data set and the reduced one (see Section 2), therefore only the former set of data will be presented hereafter.

Table 1 Genetic diversity at allozyme and mtDNA loci Sample

Allozymes

mtDNA

He

Ho

n alleles/locus

h



Tajima’s D

D. puntazzo FAR ALI OTR MES

0.0287 (0.0682) 0.0501 (0.1275) 0.0482 (0.1228) 0.0449 (0.1288)

0.0286 (0.0659) 0.0451 (0.1260) 0.0564 (0.1498) 0.0436 (0.1288)

1.2857 1.4762 1.2857 1.2381

0.978 0.994 0.995 0.986

0.05 0.040 0.037 0.035

0.79 ¡0.14 ¡0.32 ¡0.27

D. sargus FAR ALI OTR IRK

0.0775 (0.1362) 0.0874 (0.1616) 0.0888 (0.1492) 0.0788 (0.1282)

0.0662 (0.1114) 0.0875 (0.1626) 0.0694 (0.1156) 0.0638 (0.1047)

1.3684 1.4211 1.4737 1.4211

0.810 0.619 0.574 0.797

0.011 0.006 0.007 0.009

¡1.54* ¡1.29 ¡2.08** ¡1.31

He and Ho are respectively expected and observed heterozygosity (associated standard errors are shown in brackets). Average number of allele per locus over all allozyme loci is also reported. For mtDNA haplotype diversity (h) and nucleotide diversity () are presented. Probability values associated to Tajima’s D statistics are *P < 0.05 and **P < 0.01.

L. Bargelloni et al. / Molecular Phylogenetics and Evolution 36 (2005) 523–535

The same allozyme loci (GPI-1*, GPI-2*, MDH-2*, MDH-3*, MEP-1*, MEP-2*, PGM*, PGDH*) are found to be variable in both species, whereas, when a 95% frequency criterion is assumed, less polymorphic loci are found in D. puntazzo (two loci: GPI-1* and MDH-3*) as compared to D. sargus, which shows Wve polymorphic loci (GPI-2*, MDH-3*, MEP-2*, PGM*, PGDH*). On the other hand, mean number of alleles per locus is similar in both species (1.6). Mean expected heterozygosity (He) calculated for each population ranges from 0.078 to 0.089 for D. sargus; in D. puntazzo He is slightly lower (0.029–0.050). When observed genotype frequencies are compared to expectations under HW equilibrium, only one signiWcant exception is observed for a single locus in a single population [D. puntazzo (ALI), MDH-2*], and samples are considered to be at HW equilibrium for the analysed loci. In any case, subsequent analyses on genetic diVerentiation show no diVerences when performed either including or excluding locus MDH-2. Likewise, no signiWcant exception, after sequential Bonferroni correction, is observed when independence of loci is tested. While genetic diversity at allozyme loci is comparable, analysis of mtDNA variation reveals a clear diVerence between the two species. The SSCP approach applied to

527

D. sargus yielded 28 SSCP mobility classes. Sequence analysis on a number of individuals larger than that of mobility groups (a total of 50 randomly selected individuals were sequenced, nine of which were taken out of 44 from the most frequent class) conWrmed the sensitivity of the method, in agreement with a previous work where the method was more thoroughly validated on another sparid species (Ostellari et al., 1996). When the SSCP method was used to evaluate variation at the mtDNA control region in D. puntazzo, almost each individual appeared to belong to a diVerent mobility class. For this reason it was decided to directly sequence the ampliWed fragment for all D. puntazzo individuals. This procedure revealed 111 diVerent haplotypes. The two species share similar features in terms of sequence evolution (compositional and transition–transversion bias, rate heterogeneity across sites; data not shown). However, D. puntazzo shows a much higher mtDNA polymorphism than D. sargus (Table 1). In the latter species both haplotype (h) and nucleotide diversity () are similar to those observed in other marine teleosts, whereas in D. puntazzo h is always close to one (0.98– 0.99), and  (0.035–0.05) is in the high range of values reported for teleost species (Grant and Bowen, 1998; Viñas et al., 2004).

Fig. 2. Median network based on mtDNA sequences of D. sargus. Circles represent haplotypes, with size proportional to relative frequencies. For each haplotype present in more than one population sample, sectors of diVerent colours (black, Portugal; dark grey Spain; light grey Italy; and white, Greece) refer to absolute number of haplotype counts in each population. Haplotype deWnitions are as in Appendix C.

528

L. Bargelloni et al. / Molecular Phylogenetics and Evolution 36 (2005) 523–535

Fig. 3. Median network based on mtDNA sequences of D. puntazzo. Branches represent single substitutional events, if not otherwise indicated (black bars refer to multiple events). Circles represent haplotypes, with size proportional to relative frequencies. For each haplotype present in more than one population sample, sectors of diVerent colours (black, Portugal; dark grey, Spain; light grey, Italy; and white, Greece) refer to absolute number of haplotype counts in each population. Haplotype deWnitions are as in Appendix B. Only haplotypes present more than once are indicated. Haplogroups were squared in dashed lines and numbered with roman numbers (see text).

3.2. Historical and demographic factors Diplodus sargus values of D statistic in the Tajima’s test are substantially negative for all populations, with signiWcant deviations from neutrality for OTR and FAR samples (Table 1). These values may also be underestimated if we consider that strong mutation rate heterogeneity as the one observed in this study (data not shown) tends to shift D toward more positive values (Aris-Brosou and ExcoYer, 1996). Negative D values are often associated with past changes in population size, namely bottlenecks followed by population expansion. D. sargus populations therefore might have experienced large demographic Xuctuations in the past. On the opposite, no signiWcant D value is obtained for D. puntazzo. To investigate the presence of past demographic events we carried out the analysis of mismatch distribu-

tion in the two species. Sudden demographic expansions are expected to produce unimodal (“waves”) distributions of pairwise nucleotide diVerences (mismatch) between alleles (Rogers and Harpending, 1992), while such distribution shape is unlikely under population stationarity. In the case of the two species under study, in consideration of the deviations observed in the neutrality tests, the method of Schneider and ExcoYer (1999) was applied to test whether the data Wtted a sudden expansion model. For both species, the sum of square deviations (SSD, Schneider and ExcoYer, 1999) for 10,000 simulated data sets on the basis of estimated parameters is larger than SSD for observed data (Figs. 4, 5), thereby the model is not rejected. As observed by the authors (Schneider and ExcoYer, 1999), however, this method is quite conservative, rarely rejecting the expansion model. Associated P values, in any case, are much

L. Bargelloni et al. / Molecular Phylogenetics and Evolution 36 (2005) 523–535

529

Fig. 4. Mismatch distribution for D. sargus mtDNA data. On the x-axis are classes of pairwise diVerences (0, 1, 2,ƒ), on the y-axis the absolute number of comparisons yielding a certain number of diVerences. Estimated values for the expansion model are reported for each population with the exception of the Iraklion population for which the program was unable to provide a mismatch distribution analysis.

lower for D. puntazzo (P D 0.1–0.2, Fig. 4) compared to D. sargus (P D 0.6–0.8, Fig. 5), suggesting that model Wtting is rather poor for the former species. Similar evidence is obtained from estimation of raggedness index (Harpending, 1994) for each sample (data not shown), as well as from visual inspection of mismatch distributions for D. puntazzo, which always show a multimodal proWle (Fig. 4). Estimates of expansion parameter  for D. puntazzo samples are larger  D 4.2–19.4, Fig. 4 than those calculated for D. sargus  D 1.4–3.2, Fig. 5. The parameter  is equal to 2t* where t is the time of the expansion and  is the mutation rate. If we assume the same mutation rate for the two species, then the timing of the possible expansion is much more recent for D. sargus than for D. puntazzo. Regarding to the magnitude of expansion, simulation studies (Schneider and ExcoYer, 1999) have showed that while the time of expansion is adequately recovered with valid conWdence intervals, estimate of initial population size (
0) is relatively accurate, but with an overly conservative conWdence interval due to a too large upper limit of the interval, and Wnal population size (
1) is generally biased upward, with an overly large upper bound. Although less reliable than estimates of , values of
0 indicate that even assuming sudden population expansion, the starting population size was large for D. puntazzo (at least for some populations), whereas D. sargus appears to have undergone a drastic bottle-

neck before the expansion, as estimates of the
0 result extremely low (
0 D 0–0.5) in all samples. 3.3. MtDNA networks Evolutionary relationships among haplotype sequences are represented in the form of reduced median networks (MNs), which take into account haplotype frequencies, and allow for parallel substitutions. For D. sargus, the network shows a clear star-like shape (Fig. 2). A single haplotype (A) at high frequency in all population samples is present, from which the majority of remaining sequences stems out, being removed by one or few mutational steps. Evidence of moderate homoplasy is also observed in the form of few network reticulations. In D. puntazzo, network reconstruction reveals a more complex pattern, where four groups of haplotypes (haplogroups) are observed, separated by 8–14 substitution steps (Fig. 3), therefore suggesting a “deep” evolutionary history. The number of haplotypes and individuals, however, is not equal in each haplogroup, with a vast majority of sequences clustering into group II. With regard to the geographical origin of individual haplotypes, three haplogroups (I, II, III in Fig. 3) contain sequences from all four geographic samples, while the fourth one (IV in Fig. 3) is composed by haplotypes found only in the Portuguese sample, except for a single individual collected in Alicante (Spain). The entire

530

L. Bargelloni et al. / Molecular Phylogenetics and Evolution 36 (2005) 523–535

Fig. 5. Mismatch distribution for D. puntazzo mtDNA data. On the x-axis are classes of pairwise diVerences (0, 1, 2,ƒ), on the y-axis the absolute number of comparisons yielding a certain number of diVerences. Estimated values for the expansion model are reported for each population.

network is also characterised by several reticulated instead of linear relationships connecting haplotypes. This evidence seems to suggest a considerable degree of homoplasy. 3.4. Population structure Results from both allozyme and mtDNA data indicate that partitioning of genetic diversity among geographic samples is profoundly diVerent between D. puntazzo and D. sargus. The latter species shows no evidence of genetic structure, whereas a remarkable genetic divergence is present between Atlantic and Mediterranean samples of D. puntazzo. An exact test of population diVerentiation across allozyme loci in D. puntazzo reveals that gene frequencies are signiWcantly (P < 0.0001) diVerent among population samples. At the level of single locus, two loci (GPI-1* and MDH-3*) show signiWcant heterogeneity in both allele and genotype distribution (P < 0.0001 after correction for multiple tests). These two loci are also the only polymorphic ones if a 95% criterion is assumed. Pairwise tests of population diVerentiation (data not shown) reveal that in all comparisons involving the Atlantic sample (FAR) the distribution of allele frequencies signiWcantly (P < 0.0001) deviates from that expected under the hypothesis of genetic homogeneity. A similar albeit less signiWcant diVerentiation is observed between samples from East and West Mediterranean, as OTR and MSL samples are signiWcantly diVerent from ALI sample (P < 0.01). Estimates of Fst values suggest a similar pattern of genetic divergence among populations in D. puntazzo. Global Fst, calculated across loci and popula-

tions shows a highly signiWcant value (Table 2), and in pairwise comparisons the highest Fst values are found when the Atlantic sample is compared to Mediterranean ones, especially to the easternmost ones (OTR and MSL). As for the exact test, signiWcant Fst values are also observed comparing the ALI sample with OTR and MSL. For D. sargus, analysis of allozyme variation reveals no evidence for diVerentiation among or between population samples (data not shown). Global Fst across loci and populations is one order of magnitude lower than the one observed in D. puntazzo (Table 2). Similarly, all pairwise Fst values are much lower and never signiWcant. MtDNA data delineate a comparable pattern. In the case of D. puntazzo a strong diVerentiation is observed among population samples (Fst D 0.08 P < 0.0001). Estimates of Fst take into account also molecular distances between haplotypes, because in case of large values of
(eVective size £ mutation rate), it is known that statistical tests of population diVerentiation based on haplotype frequencies are less powerful than sequence-based statistics (Hudson et al., 1992). The highly signiWcant global Fst value appears to be inXuenced by the strong genetic divergence of the Atlantic sample compared to the Mediterranean ones. When populations are examined in pairwise comparisons, the FAR sample results to be signiWcantly diVerent from all three Mediterranean ones (Table 2), with increasing Fst values along a WestEast gradient suggesting a possible mechanism of isolation by distance as revealed by the statistically signiWcant correlation (P < 0.005) between pairwise Fst and geographic distances (Mantel test with 1000 permutations). SigniWcant genetic diVerentiation between

L. Bargelloni et al. / Molecular Phylogenetics and Evolution 36 (2005) 523–535

531

Table 2 Fixation indexes (Fst) across samples and between samples overall allozyme loci and mtDNA Species

Allozymes Global Fst

D. puntazzo

0.160

D. sargus

0.009

***

mtDNA Pairwise Fst

Global Fst **

***

FAR-ALI (0.130) FAR-OTR (0.300)**** FAR-MSL (0.320)**** ALI-OTR (0.06)* ALI-MSL (0.07)* OTR-MSL (0.0)

0.085

FAR-ALI (0.07) FAR-OTR (0.02) FAR-IRK (0.0) ALI-OTR (0.0) ALI-IRK (0.01) OTR-IRK (0.0)

0.007

Pairwise Fst FAR-ALI (0.120)*** FAR-OTR (0.162)*** FAR-MSL (0.202)**** ALI-OTR (¡0.004) ALI-MSL (0.040)* OTR-MSL (0.003) FAR-ALI (0.0) FAR-OTR (0.016) FAR-IRK (¡0.004) ALI-OTR (¡0.012) ALI-IRK (0.034) OTR-IRK (0.048)*

For pairwise Fst estimated values are in brackets. Abbreviations of sampling locations are as in Fig. 1. Probability values after sequential Bonferroni correction: *P < 0.05; **P < 0.01; ***P < 0.001;****P < 0.0001.

Mediterranean samples is observed only when comparing the two most distant locations: ALI and MSL. Likewise, the distribution of haplotypes is signiWcantly heterogeneous (P < 0.00001) when sampling locations and the four haplogroups (I, II, III, IV in Fig. 3) are used as categorical variables in an exact test. This result indicates that individuals collected at diVerent locations are not randomly assorted among the four major clades. In D. sargus the obtained results are rather diVerent. The null hypothesis of panmixia is never rejected when estimating Fst values for population pairs (Table 2) or across all samples (Fst D 0.007, P D 0.2), while the exact test of overall population diVerentiation reveals just a marginally signiWcant heterogeneity (P < 0.05) among samples that could be partly ascribed to divergent haplotype distribution in the IRK sample (data not shown).

4. Discussion As mentioned above, evidence of genetic discontinuity between Atlantic and Mediterranean populations has been reported for several marine species. The transition between the two basins, however, seems to be perceived by diVerent organisms in a diVerent way. Strong to moderate genetic divergence is observed in some species whereas others show no diVerentiation at all. The degree of genetic diVerentiation cannot be easily related to the species biology, as diVerent patterns are found across organisms with similar ecological features, and vice versa. For instance, in the case of the blueWn tuna and the swordWsh, which are both large pelagic Wsh, only in the latter species Atlantic samples are genetically distinct from Mediterranean ones (Chow and Takeyama, 2000; Ely et al., 2002). Divergent patterns are reported also among bivalves or benthic Wsh species (Borsa et al., 1997). Likewise, in the present study D. puntazzo and D. sargus display a remarkable discordance. DiVerences are

observed in the relative levels of genetic diversity, at least for mtDNA, and particularly in the degree of population diVerentiation. These results provide us with relevant information on population structure separately for the two species, and at the same time, allow us a comparative approach to understand the more general importance of the Atlantic–Mediterranean transition on biogeography of marine organisms. Starting from the perspective of a single species, for D. puntazzo geographic distance between sampling locations appears to be correlated with degree of genetic diVerentiation within the Mediterranean as well as when comparing Mediterranean samples with the Atlantic one. Despite sampling sites are rather sparse, this might be considered as preliminary evidence that reduction of gene Xow is associated with increasing geographic distances as suggested by the signiWcant correlation between pairwise Fst and geographic distance between sampling sites. Potential for dispersal is supposed to be relatively limited in D. puntazzo. After a pelagic larval phase, which is relatively short (1 month), juveniles settle into very shallow benthic habitats, which are abandoned several months later to join the adult population in deeper waters (Macpherson, 1998). Adults are considered to be sedentary (Bauchot and Hureau, 1986). Therefore, a model of isolation by distance appears in agreement with the species biology. However, the abrupt change in genetic composition between Atlantic and Mediterranean samples, cannot be explained solely as the eVect of geographic distance. A marked reduction of genetic exchanges appears to be associated with the transition between the Atlantic and the Mediterranean. Moreover, mtDNA analysis shows the presence of four haplotype groups, phylogenetically distinct and strongly associated with sampling locations (especially clade IV, which contains Atlantic haplotypes but one that is from Spain). Although such large phylogenetic distances are generally associated with phylogeographic discontinuities,

532

L. Bargelloni et al. / Molecular Phylogenetics and Evolution 36 (2005) 523–535

reciprocal monophyly of Atlantic and Mediterranean haplotypes is not present in the mtDNA network of D. puntazzo. To explain this observation two hypotheses can be put forward: either gene Xow between the Atlantic and Mediterranean regions has been constantly very low or absent, but time since separation of populations has been insuYcient to reach reciprocal monophyly, or recent episodes of genetic exchanges between the two regions have mixed previously separated “regional” clades. It has been shown (Neigel and Avise, 1986) that gene trees are likely to be concordant with population subdivisions after 2 £ Ne generations where Ne is the eVective population size (female eVective size in the case of mtDNA). Assume for the sake of discussion that the two regions have been isolated since the separation of the branch leading to clade 4 from the lineage ancestral to the rest of haplotypes. Uncorrected mean sequence divergence between these two lineages is 0.068, which translates, under a conventional evolutionary rate for Wsh control region (McMillan and Palumbi, 1997) of 0.1–0.12 uncorrected sequence divergence per million year (myr), in a divergence time of 0.56–0.68 myr, or 340,000–187,000 generations, (assuming a 2–3 years generation time for D. puntazzo). This means that Ne of D. puntazzo should have been larger than 2 £ (340,000– 187,000) to prevent complete sorting of haplotypes into reciprocal monophyletic clades. Based on FAO catch data, just in the year 1997, 200 metric tons of D. puntazzo were landed from Wshing vessels only in the North Adriatic, meaning 400,000–600,000 individuals for an average size of 300–500 g. Although admittedly crude, this evidence suggests a census size for D. puntazzo at least in the order of 106–107. A female eVective size of several hundreds thousands individuals therefore is not incompatible with present-day census size of D. puntazzo, and the hypothesis of incomplete monophyly as the consequence of large Ne could not be deWnitively rejected. It should be noticed, however, that in a large number of taxa, especially marine ones, Ne is several orders of magnitude lower than census size (Frankham, 1995). Moreover, some individuals of the FAR sample have haplotypes closely related to sequences found within the Mediterranean region, and included in all the remaining three clades (I, II, and III). For these reasons, the hypothesis that reciprocal monophyly might have been reached once and then lost because of episodes of genetic exchange between the two regions seems more likely. The latter scenario seems also more congruent with the hypothesis that levels of exchanges between the Atlantic and the Mediterranean are variable at diVerent time scales, with short-term, annual changes as well as historical Xuctuations associated with glacial and interglacial periods. In the latter case, geological data (Nilsson, 1982) suggest that historical events of habitat fragmentation have occurred, followed by episodes of

increased circulation between the Atlantic and the Mediterranean. For instance, during the Mindel Glaciation (400,000 years ago) the sea level dropped 115–120 m below the present level, reducing both width and depth of the Gibraltar Passage. Similar drops are suggested for the Riss Glaciation (130,000–170,000 years ago), and other glacial maxima, until the most recent one, occurred 22,000 years ago. During interglacial periods, on the other hand, sea level drops were followed by sudden rises, with movement of large water masses. On a much shorter time scale, oceanographic surveys have demonstrated that water Xow at Gibraltar is asymmetrical, with inXow consisting of surface waters from the Atlantic, while outXow composed by deeper Mediterranean waters. The Xow is also signiWcantly smaller than previously estimated, and variable during the year (Millot, 1999). Under this regard, it should be observed that in D. puntazzo several “Atlantic” individuals are included in clades I, II, and III, while only a single individual from the ALI sample falls into clade IV. However, additional data are needed to make clear whether this result is due to asymmetrical gene Xow favouring immigration from the Atlantic into the Mediterranean (therefore in countercurrent with respect to surface water Xow) or it is just the consequence of unequal sample and clade size. More importantly, the inXowing Atlantic water describes in the Alboran Sea a quasi permanent anticyclonic gyre in the West, a more variable one in the East. The particular water circulation in the Alboran Sea generates an oceanographic front located from Oran (Morocco) to Almeira (Spain), called the Oran-Almeria Front (OAF) (Fig. 1). To date, it is unclear whether in the past, especially during the mentioned climatic changes, the hydrological regime was similar to the present one. Simulation studies suggest that the volume Xow rate into the strait inXuences the growth rate of the gyre(s), but not its general structure (Gleizon et al., 1996). Thereby changes in the section of the Gibraltar Strait associated with climate modiWcations might have led to changes in levels of water Xow, but not to reversal of hydrological conditions. The relevance of the OAF, in the present as well as in the past, is suggested also by genetic evidence. Population genetics studies on sea bass (Naciri et al., 1999) and the mussel Mytilus galloprovincialis (Quesada et al., 1995 and reference therein) where samples from the Alboran Sea are included, demonstrate that a clear shift in gene frequencies (at microsatellite loci in the sea bass, or mtDNA and allozymes in the mussel) is observed to be associated to the OAF. Also the genetic pattern found for D. puntazzo is congruent temporally (large divergence among haplotypes, incomplete monophyly) and spatially (sharp change in genetic composition between Atlantic and Mediterranean samples) with the hypothesis that the mentioned historical and contemporary factors might have reduced, although not constantly, gene Xow between the two marine basins.

L. Bargelloni et al. / Molecular Phylogenetics and Evolution 36 (2005) 523–535

Rather diVerent are the results for D. sargus. Only weak genetic diVerentiation is found, never associated with the transition between Atlantic and Mediterranean.This observation partly contradict the proposed separation in distinct sub-species of the Mediterranean and Atlantic D. sargus populations, classiWed as D. s. sargus and D. s. cadenati, respectively (De la Paz et al., 1973; Bauchot and Hureau, 1986). The present results in fact do not support this view as no appreciable genetic diVerences were found between Atlantic and Mediterranean samples. Similar Wndings were also reported in another molecular investigation which grouped together in a phylogenetic tree D. sargus samples collected in the Mediterranean Sea (Calvi, Corsica, France) and oV the Atlantic African coast (Agadir, Morocco) (Summer et al., 2001). Previous allozyme studies showed diVerences between population collected within the Mediterranean basin (Lenfant and Planes, 1996; González-Wangüemert et al., 2004). In the present study no indication for D. sargus population structure at allozyme loci was evidenced neither between Mediterranean and Atlantic samples nor within Mediterranean populations. It is however worth to note that the average observed heterozygosity at the scored allozyme loci is much lower in the samples analysed in this work as compared to allozyme data which recently appeared in the literature on this species (González-Wangüemert et al., 2004). It diYcult to explain such a large diVerence, polymorphism reduction was reported in this D. sargus as the consequence of selective processes during recruitment (Planes and Romans, 2004). However, drop in the observed heterozygosity could be alternatively due low eVective population size associated to genetic drift. The simplest explanation for (near) absence of genetic structure in D. sargus is that gene Xow is suYciently high to homogenise gene frequencies across geographic populations. Alternatively, even if migration between contemporary populations is eVectively low or null, the eVect of past migrations might still be evident because migrationdrift equilibrium has not been reached yet. Life-history of D. sargus is very similar to that described for D. puntazzo, with a relatively short (1 month) plancktonic larval stage, settlement of juveniles in shallow benthic habitats, and a sedentary adult stage. Capacity for dispersal is thereby supposed to be comparable between the two species (Vigliola et al., 1998). Moreover, experimental conditions (sampling scheme, used markers) are almost identical for the two species. For these reasons, absence of genetic structure in D. sargus is even more striking, if compared to the genetic pattern observed for D. puntazzo. How can be explained this sharp diVerence between the two species? Incongruent phylogeographic patterns might be caused by ecological, genetic or historical factors (Zink, 1996). As already mentioned, the two Diplo-

533

dus species under study have similar dispersal capacity, distributional range and ecological features. DiVerences in ecology therefore appear unlikely to be responsible for divergent genetic patterns. Some relevance for diVerential dispersal might have the observation that D. sargus is more abundant than D. puntazzo both as adults (Macpherson, 1998), and juveniles (Vigliola et al., 1998). If larger census size translates into greater Ne for D. sargus, then higher levels of gene Xow are expected for this species even for dispersal capacity equivalent to D. puntazzo. While this hypothesis might explain the generally lower level of genetic population diVerentiation observed for D. sargus, it seems insuYcient to account for the peculiar diVerence between the two species regarding the Atlantic–Mediterranean transition. Some diVerentiation is indeed detected for D. sargus, but only within the Mediterranean region (signiWcant only in the comparison between Italian and Greek samples for the mtDNA). Moreover, levels of genetic diversity appear to indicate a comparable (allozymes) or much higher (mtDNA) long-term Ne for D. puntazzo compared to D. sargus, thereby not supporting the above hypothesis (D. sargus Ne > D. puntazzo Ne). EVective population size, however, could have experienced important historical variations. For instance, population size might be normally large (in the case of D. sargus larger than D. puntazzo), but bottlenecks may occasionally occur, strongly reducing long-term Ne. Results of Tajima’s neutrality test and mismatch analysis on D. sargus mtDNA data indeed suggests that this species underwent a bottleneck and a subsequent expansion, as all samples show highly negative D values, two of which signiWcantly. If strong population bottlenecks have characterised the evolutionary history of D. sargus, what is the possible inXuence on genetic divergence among populations? The eVect of variation in population size on estimates of population diVerentiation has been explored, suggesting that the classical results for constant population size holds true also for varying population size, with the obvious substitution of long-term eVective size as the harmonic mean of single-generation population sizes (Hudson, 1998). Therefore, lack of genetic diVerentiation in D. sargus cannot be simply ascribed to small long-term eVective size due to past population bottlenecks. However, reductions in population size and subsequent expansions might have been historically associated with occasional, but signiWcant migration events. For instance, if D. sargus had in the past a more limited distribution, not including the Mediterranean region, which might have been colonised only recently, then lack of a phylogeographic pattern could be due to insuYcient time to accumulate genetic diVerences between the two regions. This seems in keeping with the suggestion that the diversiWcation within D. sargus clade has its origin in the centraleastern Atlantic, (probably near Cape Verde Islands) (Summer et al., 2001) which imply that Mediterranean

534

L. Bargelloni et al. / Molecular Phylogenetics and Evolution 36 (2005) 523–535

colonization was a secondary event. Alternatively, events of extinction and recolonization could have produced a similar eVect. Both scenarios are in agreement with evidence of past demographic reduction and sudden expansion of population size of D. sargus, and with the “troubled” paleoclimatic history of the Mediterranean. In conclusion, available evidence seems to suggest that discordant genetic patterns might be associated to accidental factors rather than to ecological diVerences across species. While this suggestion is in agreement with results from other species, studies based on faster evolving loci (e.g. microsatellites) will help to better evaluate the importance of recent and/or present-day phenomena, thereby deWnitively supporting or rejecting the hypothesis that the observed discordant patterns are the legacy of historical events.

Acknowledgment This work was part of a project funded by the European Union, contract No. AIR3-CT94-1926.

Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/ j.ympev.2005.04.017

References Aris-Brosou, S., ExcoYer, L., 1996. The impact of population expansion and mutation rate heterogeneity on DNA sequence polymorphism. Mol. Biol. Evol. 13, 494–504. Bandelt, H.J., Forster, P., Sykes, B.C., Richards, M.B., 1995. Mitochondrial portraits of human populations using median networks. Genetics 141, 743–753. Bargelloni, L., Alarcon, J.A., Alvarez, M.C., Penzo, E., Magoulas, A., Reis, C., Patarnello, T., 2003. Discord in the family Sparidae (Teleostei): divergent phylogeographical patterns across the Atlantic– Mediterranean divide. J. Evol. Biol. 16, 1149–1158. Bargelloni, L., Marcato, S., Zane, L., Patarnello, T., 2000. Mitochondrial phylogeny of notothenioids: a molecular approach to Antarctic Wsh evolution and biogeography. Syst. Biol. 49, 114–129. Bauchot, M.-L., Hureau, J.C., 1986. Sparidae. In: Whitehead, P.J.P., Bauchot, M.L., Hureau, J.C., Nielsen, J., Tortonese, E. (Eds.), Fishes of the North-Eastern Atlantic and the Mediterranean. Unesco, Paris, pp. 883–907. Borsa, P., Naciri, M., Bahri, L., et al., 1997. Zoogeographie infra-speciWque de la Mer Mediterranee. Vie et Milieu 47, 295–305. Chow, S., Takeyama, H., 2000. Nuclear and mitochondrial DNA analyses reveal four genetically separated breeding units of the swordWsh (Xiphias gladius). J. Fish Biol. 565, 1087–1098. De la Paz, R.M., Bauchot, M.L., Daget, J., 1973. Les Diplodus (Perciformes, Sparidae) du groupe Sargus: systématique et phylogénie. Ichthyologia 5, 113–128. Ely, B., Stoner, D.S., Bremer, A.J., Dean, J.M., Addis, P., Cau, A., Thelen, E.J., Jones, W.J., Black, D.E., Smith, L., Scott, K., Naseri, I.,

Quattro, J.M., 2002. Analyses of nuclear ldhA gene and mtDNA control region sequences of Atlantic Northern BlueWn Tuna populations. Mar. Biotechnol. 4, 583–588. Frankham, R., 1995. EVective population size/adult population size in wildlife: a review. Genet. Res. 66, 95–107. Gleizon, P., Chabert D’Hieres, G., Renouard, D., 1996. Experimental study of the Alboran Sea gyres. Oceanol. Acta 19, 499–511. González-Wangüemert, M., Perez-Ruzafa, A., Marcos, C., GarciaCharton, J.A., 2004. Genetic diVerentiation of Diplodus sargus (Pisces: Sparidae) populations in the south-west Mediterranean. Biol. J. Linn. Soc. 82, 249–261. Grant, W.S., Bowen, B.W., 1998. Shallow population histories in deep evolutionary lineages of marine Wshes: insights from sardines and anchovies and lessons for conservation. J. Hered. 89, 415–426. Guo, S.W., Thompson, E.A., 1992. Performing the exact test of Hardy– Weinberg proportions for multiple alleles. Biometrics 48, 361–372. Hanel, R., Sturmbauer, C., 2000. Multiple recurrent evolution of trophic types in Northeastern Atlantic and Mediterranean seabreams (Sparidae, Percoidei). J. Mol. Evol. 50, 276–283. Harpending, H.C., 1994. Signature of ancient population growth in a low-resolution mitochondrial DNA mismatch distribution. Hum. Biol. 66, 591–600. Hudson, R.R., Slatkin, M., Maddison, W.P., 1992. Estimation of levels of gene Xow from DNA sequence data. Genetics 132, 583–589. Hudson, R.R., 1998. Island models and the coalescent process. Mol. Ecol. 7, 413–418. Lenfant, P., Planes, S., 1996. Genetic diVerentiation of white sea bream within the Lion’s Gulf and the Ligurian Sea (Mediterranean Sea). J. Fish Biol. 49, 613–621. Macpherson, E., 1998. Ontogenetic shifts in habitat use and aggregation in juvenile sparid Wshes. J. Exp. Mar. Biol. Ecol. 220, 127–150. McMillan, W.O., Palumbi, S.R., 1997. Rapid rate of control-region evolution in PaciWc butterXyWshes (Chaetodontidae). J. Mol. Evol. 45, 473–484. Millot, C., 1999. Circulation in the Western Mediterranean sea. J. Mar. Syst. 20, 423–442. Naciri, M., Lemaire, C., Borsa, P., Bonhomme, F., 1999. Genetic study of the Atlantic/Mediterranean transition in sea bass (Dicentrarchus labrax). J. Hered. 90, 591–596. Neigel, J.E., Avise, J.C., 1986. Phylogenetic relationships of mitochondrial DNA under various demographic models of spetiation. In: Nevo, E., Karlin, S. (Eds.), Evolutionary Processes and Theory. Academic Press, New York, pp. 59–68. Nilsson, T., 1982. The Pleistocene: Geology and Life in the Quaternary Age. Ridel D Publishing Company, Dordrecht, Holland. Orrell, T.M., Carpenter, K.E., 2004. A phylogeny of the Wsh family Sparidae (porgies) inferred from mitochondrial DNA data. Mol. Phylogenet. Evol. 32, 425–434. Ostellari, L., Bargelloni, L., Penzo, E., Patarnello, P., Patarnello, T., 1996. Optimization of single-strand conformation polymorphism and sequence analysis of the mitochondrial control region in Pagellus bogaraveo (Sparidae, Teleostei): rationalized tools in Wsh population biology. Anim. Genet. 27, 423–427. Pannacciulli, F.G., Bishop, J.D.D., Hawkins, S.J., 1997. Genetic structure of populations of two species of Chthamalus (Crustacea: Cirripedia) in the north-east Atlantic and Mediterranean. Mar. Biol. 128, 73–82. Perez-Losada, M., Guerra, A., Sanjuan, A., 1999. Allozyme diVerentiation in the cuttleWsh Sepia oYcinalis (Mollusca: Cephalopoda) from the NE Atlantic and Mediterranean. Heredity 83, 280–289. Planes, S., Romans, P., 2004. Evidence of genetic selection for growth in new recruits of a marine Wsh. Mol. Ecol. 17, 2049–2060. Quesada, H., Beynon, C.M., Skibinski, D.O., 1995. A mitochondrial DNA discontinuity in the mussel Mytilus galloprovincialis Lmk: pleistocene vicariance biogeography and secondary intergradation. Mol. Biol. Evol. 12, 521–524. Quignard, J.P., 1978. La Mediterranee, creuset ichthyologique. Bollettino di Zoologia 45 (Suppl.), 23–36.

L. Bargelloni et al. / Molecular Phylogenetics and Evolution 36 (2005) 523–535 Raymond, M., Rousset, F., 1995a. GENEPOP (version 1.2): population genetics software for exact tests and ecumenicism. J. Hered. 86, 248–249. Raymond, M., Rousset, F., 1995b. An exact tes for population diVerentiation. Evolution 49, 1280–1283. Reina, J., Martinez, G., Amores, A., Alvarez, M.C., 1994. InterspeciWc genetic diVerentiation in Western Mediterranean sparid Wsh. Aquaculture 125, 47–57. Rogers, A.R., Harpending, H.C., 1992. Population growth makes waves in the distribution of pairwise genetic diVerences. Mol. Biol. Evol. 9, 552–569. Schneider, S., ExcoYer, L., 1999. Estimation of past demographic parameters from the distribution of pairwise diVerences when the mutation rates vary among sites: application to human mitochondrial DNA. Genetics 152, 1079–1089. Schneider, S., Roessli, D., ExcoYer, L., 2000. Arlequin Ver. 2.000: A Software for Population Genetic Data Analysis. Genetics and Biometry Laboratory, University of Geneva, Geneva, Switzerland. Summer, M., Hanel, R., Sturmbauer, C., 2001. Mitochondrial phylogeny and biogeographic aYnities of sea bream of the genus Diplodus (Sparidae). J. Fish Biol. 59, 1639–1652. Tajima, F., 1989. Statistical method for testing the neutral mutation hypothesis by DNA polymorphism. Genetics 123, 585–595.

535

Thompson, J.D., Gibson, T.J., Plewniak, F., Jeanmougin, F., Higgins, D.G., 1997. The CLUSTAL_X windows interface: Xexible strategies for multiple sequence alignment aided by quality analysis tools. Nucleic Acids Res. 25, 4876–4882. Vigliola, L., Harmelin Vivien, M.L., Biagi, F., et al., 1998. Spatial and temporal patterns of settlement among sparid Wshes of the genus Diplodus in the northwestern Mediterranean. Mar. Ecol. Prog. Ser. 168, 45–56. Viñas, J., Alvarado Bremer, J., Pla, C., 2004. Phylogeography of the Atlantic bonito (Sarda sarda) in the northern Mediterranean: the combined eVects of historical vicariance, population expansion, secondary invasion, and isolation by distance. Mol. Phylogenet. Evol. 33, 32–42. Walsh, P.S., Metzger, D.A., Higuchi, R., 1991. Chelex 100 as a medium for simple extraction of DNA for PCR based typing from forensic material. Biotechniques 10, 506–513. Weir, B.S., Cockerham, C.C., 1984. Estimating F-statistics for the analysis of population structure. Evolution 38, 1358–1370. Zane, L., Ostellari, L., Maccatrozzo, L., Bargelloni, L., Cuzin-Roudy, J., Buchholz, F., Patarnello, T., 2000. Genetic diVerentiation in a pelagic crustacean (Meganyctiphanes norvegica: Euphausiacea) from the North East and the Mediterranean Sea. Mar. Biol. 136, 191–199. Zink, R.M., 1996. Comparative phylogeography in North American birds. Evolution 50, 308–317.