Fragmentation of sea bass populations in the western ... - Europe PMC

Fragmentation of sea bass populations in the western and eastern Mediterranean as revealed by microsatellite polymorphism Lilia Bahri-Sfar1, Christophe Lemaire2 , Oum Kalthoum Ben Hassine1 and Franc°ois Bonhomme2* 1Laboratoire

de Biologie et Parasitologie Marines, Faculte¨ des Sciences de Tunis, Campus Universitaire,Tunis,Tunisie Ge¨ nome, Populations, Interactions, CNRS, UMR 5000, Universite¨ de Montpellier 2, Station Me¨diterrane¨ enne de l’Environnement Littoral, 1 Quai de la Daurade, 34200 Se©te, France 2 Laboratoire

We studied the genetic structure at six microsatellite loci of the Mediterranean sea bass (Dicentrarchus labrax) on 19 samples collected from di¡erent localities in the western and eastern Mediterranean basins. Signi¢cant divergence was found between the two basins. The distance tree showed two separate clusters of populations which matched well with geography, with the noticeable exception of one Egyptian sample which grouped within the western clade, a fact attributable to the introduction of aquaculture broodstock. No heterogeneity was observed within the western basin (³^ ˆ 0:0014 and n.s.). However, a signi¢cant level of di¡erentiation was found among samples of the eastern Mediterranean (³^ ˆ 0:026 and p5 0.001). These results match with water currents but probably not with the dispersal abilities of this ¢sh species. We thus hypothesize that selective forces are at play which limit long-range dispersal, a fact to be taken into account in the debate about speciation processes in the marine environment. Keywords: Dicentrarchus labrax; microsatellites; population fragmentation; genetic structure ; Mediterranean Sea The existence of an Atlantic^Mediterranean transition zone has recently been demonstrated in the case of the sea bass (Dicentrarchus labrax (Linne¨ 1758)) (Naciri et al. 1999). This was surprising for this euryhaline and eurythermic demersal species, since adult migratory behaviour has been reported to reach several hundred kilometres (Pickett & Pawson 1994). It poses the question of hypothetical behavioural mechanisms reinforcing genetic isolation. Here, we explore further the relationships between geography and genetic structure in the case of the transition between the western and eastern Mediterranean basins, where the hydrological gradients are weaker. There are few published data which report on this east^west di¡erentiation in the sea bass. Using samples from the Golfe-du-Lion (western basin) and that of Gabe©s (south of Tunisia in the eastern basin), Benharrat et al. (1983) found a small genetic distance at 34 allozyme loci (pairwise DNei ˆ 0.0011^0.0016, equivalent to a global GST (Nei & Chesser 1983) of 0.004). On the basis of mitochondrial cytochrome b sequences, Patarnello et al. (1993) found a non-homogeneous (but statistically non-signi¢cant) distribution of the two most frequent haplotypes (A and B) between the eastern and western basins. More recently, using 28 allozyme loci, Allegrucci et al. (1997) reported an important FST between eastern and western Mediterranean samples. This di¡erence was con¢rmed using the same samples through the study of mitochondrial DNA (Cesaroni et al. 1997) and randomampli¢ed polymorphic DNAs (RAPDs) (Caccone et al. 1997). Therefore, the existence of an east^west di¡erentiation is a good working hypothesis for Mediterranean sea bass, despite a very small representation of eastern populations in the above-mentioned studies. In the

1. INTRODUCTION

Delineating independent breeding units is one of the challenges geneticists have to face with marine organisms if they are to understand the forces which promote speciation in this environment (Strathmann 1990; Palumbi 1994, 1996). In such open habitats apparently without physical barrier to dispersal and where species are abundant and fecund, one could expect a high gene £ow which will lead to genetic homogeneity over large distances (Palumbi 1992; Avise 1994). These long-range gene £ows are likely to counteract local adaptations (Slatkin 1987; Garcia Ramos & Kirkpatrick 1997) which will impede speciation processes in the sea. This should be particularly true in the case of ¢shes which can disperse both as larvae and adults. There are many examples of no or small genetic variation at di¡erent geographical scales (Smith 1991; Blanquer et al. 1992; Elliott et al. 1994; Broughton & Gold 1997; Garc|¨ a de Leo¨n et al. 1997; Chikhi et al. 1998). However, a growing body of literature suggests the existence of higher than expected genetic variation over short geographical distances at the infraspeci¢c level (Waples 1987; Bowen & Avise 1990; Palumbi 1994; Planes et al. 1995; Bembo et al. 1996; Bentzen et al. 1996; Magoulas et al. 1996; Turan et al. 1998). What then are the reasons for this contrasting picture ? The case of transition zones between seas or oceans is particularly interesting in addressing this question because one expects to ¢nd an exacerbation of the various forces likely to have promoted a reduction of gene £ow in these areas (e.g. historical, physical or selective processes) (Borsa et al. 1997). *

Author for correspondence ([email protected]).

Proc. R. Soc. Lond. B (2000) 267, 929^935 Received 23 November 1999 Accepted 27 January 2000

929

© 2000 The Royal Society

930

L. Bahri-Sfar and others

Genetic structure of Mediterranean sea bass

Table 1. List of samples studied from wild broodstock collected in natural sites geographical origin Western Mediterranean Basin Golfe-du-Lion Gulf of Valencia Gulf of Annaba Siculo-Tunisian Strait Siculo-Tunisian Strait Siculo-Tunisian Strait Siculo-Tunisian Strait Siculo-Tunisian Strait Gulf of Tunis Gulf of Tunis Eastern Mediterranean Basin Lybico-Tunisian Gulf Lybico-Tunisian Gulf Lybico-Tunisian Gulf Lybico-Tunisian Gulf Adriatic Sea Ionian Sea Aegean Sea Aegean Sea Aegean Sea a b

locality

code

n

date of sampling habitat

provider

Se©te Valencia Annaba Marsalaa Ghar el Melh Ghar el Melh Lagoon Ichkeul lake

FSET EGLV AGLA ISCL TGRM TLGR

26 37 34 26 16 7

October 1994 July 1994 January 1994 ö February 1997 May 1997

sea sea sea sea sea coastal lagoon

F. Garc|ä de Leo¨n F. Garc|ä de Leo¨n H. Kara V. Sbordoni L. Bahri-Sfar L. Bahri-Sfar

TISK

40

January 1997

L. Bahri-Sfar

Bizerte lagoon La Goulette Northern Tunis Lagoon

TBIZ TGOU TLNT

54 32 32

February 1997 April 1997 Spring 1997

continental lagoon coastal lagoon sea coastal lagoon

Mahdia Kerkennah Islands Sfax El Biban Lagoon Venezia Lagoon Messolongib Thessaloniki Crete Bardawil Lagoon

TMAH TKER TSFX TELB ILDV GMSL GTSK GCRT YEGL

29 30 11 63 5 33 30 6 30

December1996 February 1997 October 1996 March 1996 ö ö June 1997 ö ö

sea sea sea coastal lagoon coastal lagoon sea sea sea coastal lagoon

L. Bahri-Sfar L. Bahri-Sfar L. Bahri-Sfar L. Bahri-Sfar T. Patarnello T. Patarnello F. Bonhomme T. Patarnello V. Sbordoni

L. Bahri-Sfar L. Bahri-Sfar L. Bahri-Sfar

First generationbred in captivity (G1). Second generationbred in captivity (G2).

present work, we report on the polymorphism of six microsatellite loci in 19 samples of Mediterranean sea bass collected from ten western and nine eastern sites. The sampling e¡ort was concentrated around the SiculoTunisian Strait, which is known to be a biogeographical boundary between the two Mediterranean basins (Quignard 1978). In this manner, we tried to localize the zone of transition between eastern and western populations if, indeed, any exists with greater accuracy. Finally, we compared the levels of homogeneity within each basin and discuss the reasons for the di¡erences revealed.

(c) PCR and genotyping

2. MATERIAL AND METHODS

The mean number of alleles and the observed and unbiased expected heterozygosity (Nei 1978) were computed for each locus. Wright’s ¢xation index (FIS ), which measures the heterozygote de¢ciency within each subpopulation, was estimated using Weir & Cockerham’s (1984) f-estimator. The distribution under the null hypothesis of Hardy^Weinberg equilibrium (HWE) ( f ˆ 0) was obtained after 10 000 allelic permutations for each locus.

(a) Sampling Five hundred and forty-one individuals from wild Mediterranean sea bass were collected, mostly from ¢shermen. A piece of muscle or pectoral ¢n was collected post-mortem and preserved in 80% ethanol. Table 1 reports the details of the populations studied and ¢gure 1 shows their geographical locations. The data for the western basin samples (FSET, AGLA and ISCL) come from the study of Naciri et al. (1999) while the sample (EGLV) has been previously published in Garc|ä de Leo¨n et al. (1997).

(b) DNA extraction

A rapid DNA extraction protocol using Chelex 1001 (Biorad, Inc., Hercules, CA, USA) was employed. A ca. 1mm 3 piece of tissue was allowed to dry for 1h at room temperature. Each piece was then incubated for 4 h at 56 8 C in 10 m ml Tris EDTA, 300 m ml of 5% Chelex 1001 and 15 m ml of 20 mg ml¡1 proteinase K. The supernatant was used as a polymerase chain reaction (PCR) template. Proc. R. Soc. Lond. B (2000)

The details of the PCR ampli¢cation, technical procedures and genotyping of the six microsatellite loci of the sea bass (Garc|ä de Leo¨n et al. 1995) (Labrax-3, Labrax- 6, Labrax- 8, Labrax-13, Labrax-17 and Labrax-29) are described in Naciri et al. (1999). We used six individuals of known genotypes on each gel to standardize the allelic readings. Autoradiographs were read twice independently to minimize errors.

(d) Data treatment (i) Genetic parameters Genetic variability and departure from Hardy^Weinberg equilibrium

Di¡erentiation between populations The same procedure as above was applied to Weir & Cockerham’s (1984) estimator of the global ¢xation index (FST), ³, but with permutations performed on individual genotypes. FST was preferred over other indexes making explicit use of the size di¡erence between alleles such as RST because the mutation process behind the microsatellites used here probably does not ¢t the stepwise model (Garc|ä de Leo¨n et al. 1997), a case where the variance of RST is expected to be large (Slatkin 1995). We used a jackknife resampling scheme to estimate the variance of ³ according to the contribution of each locus.


t la n At


VENEZIA

n cea ic O

SETE

Ad ri

ati

cS

Black Sea

ea THESSALONIKI MESSOLONGI

VALENCIA ean rran e t i ed rn M e t s We ANNABA

MARSALA

TUNIS

Ionian Sea

Siculo-Tunisian Strait

BIZERTE GHAR el MELH GHAR el MELH lagoon Cape Bon

ISHKEUL

931

Aegean Sea CRETE

Eastern Mediterranean

BARDAWIL

GOULETTE

MAHDIA SFAX

KERKENNAH

EL BIBAN Figure 1. Geographical location of the 19 samples.

Pairwise ³ -values were also computed and the sequential Bonferroni procedure was used to control for multiple tests at the 5% signi¢cance level (Rice 1989). Since all pairwise values are not actually independent from one another, this correction is conservative. In keeping with previous studies on sea bass genetic di¡erentiation, the co-ancestry coe¤cient of Reynolds et al. (1983), calculated using data from the six loci, was used as genetic distance (DReynolds ˆ 7 ln(17 FST)). This choice is appropriate given that the model we test is that of migration^drift equilibrium under which FST (and, hence, DReynolds) is a predictable quantity which decreases with the absolute number of migrants received by each deme. All statistical analyses were performed using the Genetix v. 3.3 software package (Belkhir et al. 1998).

(ii) Phylogenetic algorithms Trees were obtained using the neighbour-joining algorithm (Saitou & Nei 1987) implemented in the phylogenetic package PHYLIP 3.57 (Felsenstein 1993). The robustness of each node was evaluated by bootstrapping 1000 times over loci using the SEQBOOTand CONSENSE programs in the same package. 3. RESULTS

The data concerning gene diversity and the HWE are presented in electronic Appendix A available on The Royal Society Web site. As to the di¡erentiation between the eastern and western Mediterranean sea bass populations, a signi¢cant global FST (³^ ˆ 0:0136 and p5 0.001) was obtained when all samples were considered. Proc. R. Soc. Lond. B (2000)

Jackkni¢ng over loci showed that the loci contributed homogeneously to this di¡erentiation (standard deviation 0.005). Table 2 gives the pairwise ³ -estimates. These ranged from 0.086 between the samples from Sfax (TSFX) and Crete (GCRT) (two eastern samples) to ¡0.0003 between the samples from Se©te (FSET) and the Gulf of Valencia (EGLV) (two western samples). All the values between the ten western samples were small and none were signi¢cant. This resulted into a non-signi¢cant global FST when computed inside the occidental basin (³^ ˆ 0:0014 and p4 0.180). We thus considered this set of ten samples as belonging to the same genetic unit. In contrast, FST was signi¢cantly di¡erent from zero in the oriental basin (³^ ˆ 0:024 and p5 0.001) which appears to be much more heterogeneous. The tree in ¢gure 2 shows that two groups can be distinguished with a bootstrap value of 88%. With the exception of the Egyptian sample, these groups match quite well with the geography, with an occidental clade (to which the Bardawil sample YEGL is related) and an oriental one clustering all the samples south or east of the Gulf of Tunis. When the same procedure was applied locus by locus (data not shown), this east^west di¡erentiation was also supported by three loci (Labrax- 6, Labrax-13 and Labrax- 29), whereas for the remaining three loci the samples Tunis Lagoon (TLNT), Bizerte Lagoon (TBIZ) and Thessaloniki (GTSK) moved from one group to the other. The Egyptian sample always clustered with the western group. Identical conclusions could be obtained on a tree derived from Nei’s (1978) genetic distance (not shown), this last index showing a 0.97 correlation

Proc. R. Soc. Lond. B (2000)

AGLA

ISCL

¡0.0003 0.0015 0.0004 ö ¡0.0017 0.0060 ö ö 0.0064 ö ö ö ö ö ö ö ö ö ö ö ö ö ö ö ö ö ö ö ö ö ö ö ö ö ö ö ö ö ö ö ö ö ö ö ö ö ö ö ö ö ö ö ö ö

EGLV

¡0.0048 ¡0.0011 ¡0.0006 ¡0.0022 ö ö ö ö ö ö ö ö ö ö ö ö ö ö

TGRM

TBIZ

TISK

TGOU

TLNT

TSFX

TMAH

TKRK

¡0.0152 0.0002 ¡0.0018 ¡0.0015 0.0089 0.0100 0.0072 0.0096 ¡0.0121 0.0045 0.0018 0.0014 0.0043 0.0198 0.0122 0.0163 ¡0.0188 ¡0:0025 0.0028 ¡0.0013 0.0044 0.0061 0.0121 0.0115 ¡0.0036 0.0064 ¡0.0024 0.0054 0.0126 0.0195 0.0218 0.0244 ¡0:0170 ¡0.0011 ¡0.0005 ¡0:0101 0.0081 0.0139 0.0074 0.0097 ö ¡0.0152 ¡0.0112 ¡0.0155 ¡0.0141 ¡0.0143 ¡0.0024 0.0003 ö ö 0.0025 0.0014 0.0042 0.0024 0.0079 0.0095 ö ö ö 0.0032 0.0014 0.0109 0.0101 0.0142 ö ö ö ö 0.0112 0.0162 0.0116 0.0148 ö ö ö ö ö 0.0201 0.0095 0.0156 ö ö ö ö ö ö 0.0059 ¡0.0005 ö ö ö ö ö ö ö ¡0.0029 ö ö ö ö ö ö ö ö ö ö ö ö ö ö ö ö ö ö ö ö ö ö ö ö ö ö ö ö ö ö ö ö ö ö ö ö ö ö ö ö ö ö ö ö ö ö ö ö

TLGR 0.0143 0.0152 0.0105 0.0216 0.0062 0.0011 0.0057 0.0155 0.0133 0.0133 0.0044 0.0002 0.0019 ö ö ö ö ö

TELB 0.0411 0.0389 0.0339 0.0363 0.0184 0.0165 0.0266 0.0276 0.0216 0.0175 0.0292 0.0035 0.0080 0.0056 ö ö ö ö

ILDV 0.0310 0.0299 0.0298 0.0315 0.0352 0.0254 0.0295 0.0222 0.0297 0.0176 0.0552 0.0194 0.0346 0.0292 0.0123 ö ö ö

GMSL

0.0499 0.0436 0.0477 0.0649 0.0230 0.0381 0.0462 0.0455 0.0403 0.0270 0.0864 0.0353 0.0454 0.0458 0.0395 0.0315 ö ö

GCRT

0.0203 0.0226 0.0144 0.0260 0.0069 0.0014 0.0139 0.0127 0.0111 0.0105 0.0256 0.0138 0.0210 0.0163 0.0225 0.0264 0.0311 ö

GTSK

0.0160 0.0201 0.0173 0.0205 0.0280 0.0043 0.0186 0.0208 0.0316 0.0181 0.0313 0.0410 0.0375 0.0385 0.0781 0.0534 0.0771 0.0497

YEGL


FSET EGLV AGLA ISCL TGRM TLGR TBIZ TISK TGOU TLNT TSFX TMAH TKRK TELB ILDV GMSL GCRT GTSK

³

(Signi¢cant values after the sequential Bonferroni procedure are underlined. Samples from the eastern Mediterranean are indicated in italics.)

Table 2. Pairwise ³^ matrix calculated according to Weir & Cockerham (1984)

932 Genetic structure of Mediterranean sea bass

Genetic structure of Mediterranean sea bass Bardawil Lagoon (YEGL) SÙ ette (FSET) Sicilia (ISCL) 52% Ishkeul Lagoon (TISK) Valencia (EGLV) Goulette (TGOU) 59% Ghar El Melh (TGRM) Annaba (AGLA) Ghar El Melh Lagoon (TLGR)


933

with a su¤ciently high amount of gene £ow between them so that genetic di¡erentiation cannot accumulate. The case of the Egyptian sample from Bardawil Lagoon is interesting because it does not cluster with the samples belonging to the same geographical origin, but rather with western samples (although with a long branch) (¢gure 2). A similar result was obtained for the same sample using allozyme data (Allegrucci et al. 1997). This suggests a recent introduction of animals of western origin. This is not surprising, since eggs or ¢ngerlings originating from the western basin were used to seed many hatcheries around the Mediterranean when sea bass aquaculture began.

Bizerte (TBIZ)

(b) East^west di¡erentiation

Tunis North Lagoon (TLNT) Messolongi (GMSL) Crete (GCRT) Venezia Lagoon (ILDV) Mahdia (TMAH) El Biban Lagoon (TELB) 88%

Sfax (TSFX) Kerkennah (TKER) Thessaloniki (GTSK)

0.01 Figure 2. Neighbour-joining tree of D. labrax populations in the Mediterranean Sea. The bootstrap values are from a consensus tree. Only values above 50% are indicated.

coe¤cient on our data set with Reynold et al.’s (1983) coancestry coe¤cient. 4. DISCUSSION AND CONCLUSION

According to our results, Mediterranean sea bass populations are divided into two major groups corresponding to populations living in the eastern and western basins. (a) Western basin

If we omit locus Labrax- 6 for the reasons detailed in the electronic Appendix A, the samples show either no or very slight departures from the HWE. This is consistent with the non-signi¢cant value (F^ IS ˆ 7 0.004) found with a slightly di¡erent sample of western Mediterranean sea bass analysed at the same loci by Naciri et al. (1999). Therefore, we may provisionally consider that every sample is the product of a primarily panmictic reproduction event. If we extrapolate this to the interpopulation scale, the pairwise and global F ST provided for the ten western samples are either extremely small or non-signi¢cant. Similarly, low F ST for western Mediterranean populations has also been reported in Garc|¨ a de Leo¨n et al. (1997) and Naciri et al. (1999), which together bring the total number of closely related (if not identical) samples studied to date in this part of the Mediterranean to 14. Since there is no single breeding ground for western Mediterranean sea bass, this means that the occidental basin probably functions as a series of local populations Proc. R. Soc. Lond. B (2000)

In the present study, the between-basin FST (computed as the mean of the pairwise values) averages ³^ ˆ 0:0198 (excluding Bardawil Lagoon), compared with the average of ³^ ˆ 0:023 found between western Mediterranean and Alboran SeaÂtlantic samples (Naciri et al. 1999) (this ¢gure would amount to ³^ ˆ 0:046 if one computes the Atlanticêastern Mediterranean with the samples from both studies (not shown)). The sea bass populations from the Oriental Mediterranean Basin are thus clearly di¡erent from the western populations, as has already been suggested by studies using other genetic markers (see } 1). Until now, studies which have shown such a western ^ eastern Mediterranean subdivision for ¢shes have often been inconclusive because of limited sampling. (c) Eastern basin

In contrast with western populations, sea bass from the Eastern Mediterranean Basin, with a global FST of 0.017 ( p5 0.001), appear to come from di¡erentiated units. However, there was heterogeneity in the levels of di¡erentiation between the di¡erent samples. For example, the south-Tunisian samples formed a homogeneous group (F^ ST ˆ 0.0011 and p4 0.05) and di¡ered signi¢cantly from all other eastern basin samples, but the Greek and Adriatic samples were signi¢cantly di¡erent (F^ ST ˆ 0.026 and p 5 0.001). Indeed, the genetic structure of eastern sea bass populations is consistent with the subdivision of the eastern Mediterranean into several basins (e.g. the Adriatic, Ionian and Aegean Seas and the LybicoTunisian Gulf ). The genetic distinctiveness of each subset probably means that each sub-basin is quite independent from the others in terms of spawning grounds and larval circulation. Can this be explained by the history and hydrography of the di¡erent sub-basins ? There are many forces in the marine environment which can contribute to the structure of inter- and intraspeci¢c biodiversity, such as depth, light, temperature and nutrient gradients. Hydrodynamics is another force which is often put forward to explain why organisms with planktonic larvae can or cannot complete a life cycle in a given area. Larvae have to get to a place where they can grow and become reproductive adults and these in turn have to emit larvae to a place where they can subsequently recruit (Roughgarden et al. 1988; Sinclair & Iles 1989). As far as genetic di¡erentiation is concerned, a logical consequence of the preceding constraints is to allow predictions as to the existence of distinct units whenever more or less closed hydrological features are observed.

934



Looking at past and present hydrographic regimes in the Mediterranean (Ovchinnikov 1966), one can easily ¢nd such structures. During the last glaciation, the lower sea level modi¢ed coast lines, splitting apart the eastern and western basins (Bonatti 1966; Thiede 1978). Since then and probably even before, the two basins have had di¡erent hydrographic regimes, the western one being much more uniform than the eastern one because of their respective geographies. The water circulation in the Siculo-Tunisian Strait is characterized by a unidirectional east^south-east £ow of pelagic currents (coming from the Atlantic via Gibraltar) which go round Cape Bon and leave the coast of Tunisia at the latitude of Kelibia (368 50’N), while eastern Mediterranean waters stay in the Lybico-Tunisian Gulf. The circulation is not very active in the rest of the eastern Mediterranean, with the Adriatic and Aegean Seas, which are under the in£uence of cool and low salinity waters, showing cyclonic circulation causing isolation of their northern parts (Pinardi et al. 1997). These facts match rather well with our ¢nding of an east ^west split of sea bass populations, as well as the separation of the Adriatic, Ionian, Aegean and south Tunisian samples. These hydrographic patterns may be rather ancient, which could have allowed progressive genetic di¡erentiation. However, as was noted in an earlier study concerning the transition across the AlmeriaÔran front (Naciri et al. 1999), all these purely mechanistic and historical explanations may be insu¤cient in the case of euryhaline and eurythermic ¢shes such as sea bass. If populations are now at migration^drift equilibrium, and we have reasons to believe they are because a few hundred generations are generally su¤cient to reach this equilibrium (Slatkin 1980), the pairwise F^ ST-values we ¢nd between subpopulations would amount to an e¡ective number of migrants, Nm (N m ˆ 17 F^ ST/4ST), ranging between 4 and 41. As rough as these ¢gures are, they are not high and if F^ ST-values were underestimated for some reason such as allelic homoplasy, the mutation rate being much higher than the migration rate or an equivalent phenomenon, they would be overestimates. Are these values biologically realistic ? Larvae may be entrapped in hydrological structures which keep them apart from other pools of larvae, but juveniles are potentially active swimmers and adults are able to undertake long migrations (Pickett & Pawson 1994). Therefore, we once again suggest, as proposed by Naciri et al. (1999), that the hypothesis according to which the number of successful migrant ¢shes is actually smaller than might be expected based on their apparent dispersal abilities and forces other than purely passive hydrological mechanisms are implied, such as active homing or other behavioural traits. In other words, while hydrological features appear to be a good primary template for a ¢sh population structure to establish as the present data set seems to show, the migration capabilities of adults and juveniles greatly exceed those which would be necessary to prevent any genetic di¡erentiation. Since this di¡erentiation nevertheless occurs, one could thus hypothesize that selective processes are at play which limit longdistance dispersal, an already recognized phenomenon in the marine world. This opens up the debate as to the real mechanisms preventing individuals from migrating Proc. R. Soc. Lond. B (2000)

from one stock to the other less than they should be able to. The authors wish to thank F. Garc|ä de Leo¨n for providing unpublished data, M. Naciri for the genotyping of the Sicilian and Egyptian samples, V. Sbordoni, H. Kara and T. Patarnello for providing some of the samples and two anonymous reviewers for their helpful comments. They also extend their thanks to P. Borsa and J. C. Garza for helpful discussions. This work was partly ¢nanced by contract IFREMER URM no. 16 and CNRS-Diversitas `GENDIV’ to F.B.

REFERENCES Allegrucci, G., Fortunato, C. & Sbordoni, V. 1997 Genetic structure and allozyme variation of sea bass (Dicentrarchus labrax and D. punctatus) in the Mediterranean Sea. Mar. Biol. 128, 347^358. Avise, J. C. 1994 Molecular markers, natural history and evolution. London: Chapman & Hall. Belkhir, K., Borsa, P., Goudet, J., Chikhi, L. & Bonhomme, F. 1998 Genetix v. 3.3. Logiciel sous WindowsTM pour la ge¨ ne¨ tique des populations. Montpellier, France: Laboratoire Ge¨nome et Populations, Universite¨ de Montpellier 2. Bembo, D. G., Carvalho, G. R., Cingolani, N. & Pitcher, T. J. 1996 Electrophoretic analysis of stock structure in northern Mediterranean anchovies, Engraulis encrasicolus. ICES J. Mar. Sci. 53, 115^128. Benharrat, K., Pasteur, N., Siau, Y. & Bouain, A. 1983 Polymorphisme biochimique de loups (Dicentrarchus labrax) originaires de quatre populations naturelles et d’un e¨levage. In Recherches biologiques et aquaculture, vol. 1, pp. 1^17. Brest, France: CNEXO^BNDO. Bentzen, P., Taggart, C. T., Ruzzante, D. E. & Cook, D. 1996 Microsatellite polymorphism and the population structure of Atlantic cod (Gadus morhua) in the nothwest Atlantic. Can. J. Fish Aquat. Sci. 53, 2706^2721. Blanquer, A., Alayse, J. P., Berrada-Rkhami, O. & Berrebi, P. 1992 Allozyme variation in turbot (Psetta maxima) and brill (Scophtalmus rhombus) (Osteichthyes, Pleuronectiformes, Scophtalmidae) throughout their range in Europe. J. Fish Biol. 41, 725^736. Bonatti, E. 1966 North Mediterranean climate during the last Wurm glaciation. Nature 205, 984^985. Borsa, P., Naciri, M., Bahri, L., Chikhi, L., Garcia de Leon, F. J., Kotoulas, G. & Bonhomme, F. 1997 Zoogeögraphie infraspe¨ci¢que de la mer Me¨ diterraneë: analyse des donneës ge¨ne¨tiques populationnelles sur seize espe©ces AtlantoMe¨ diterraneënnes (poissons et inverte¨bre¨s). Vie et Milieu 47, 295^305. Bowen, B. W. & Avise, J. C. 1990 Genetic structure of Atlantic and Gulf of Mexico populations of sea bass, menhaden, and sturgeon: in£uence of zoogeographic factors and life-history patterns. Mar. Biol. 107, 371^381. Broughton, R. E. & Gold, J. R. 1997 Microsatellite development and survey of variation in northern blue¢n tuna (Thunnus thynnus). Mol. Mar. Biol. Biotechnol. 6, 308^314. Caccone, A., Allegrucci, G., Fortunato, C. & Sbordoni, V. 1997 Genetic di¡erentiation within the European sea bass (D. labrax) as revealed by RAPD^ PCR assays. J. Hered. 88, 316^324. Cesaroni, D., Venanzetti, F., Allegrucci, G. & Sbordoni, V. 1997 Mitochondrial DNA length variation and heteroplasmy in natural populations of the European sea bass, Dicentrarchus labrax. Mol. Biol. Evol. 14, 560^568. Chikhi, L., Agne©se, J. F. & Bonhomme, F. 1998 Low genetic variability in a widely distributed and abundant clupeid species, Sardinella aurita. New empirical results and interpretations. J. Fish Biol. 52, 861^878.

Genetic structure of Mediterranean sea bass Elliot, N. G., Smolenski, A. J. & Ward, R. D. 1994 Allozyme and mitochondrial DNA variation in orange roughy, Hoplostethus atlanticus (Teleostei: Trachichtyidae): little di¡erentiation between Australian and North Atlantic populations. Mar. Biol. 119, 621^627. Felsenstein, J. 1993 Phylip 3.57, phylogeny inference package. Seattle, WA: University of Washington. Garc|ä de Leo¨n, F. J., Dallas, J., Chatain, B., Canone, M., Versini, J. & Bonhomme, F. 1995 Development and use of microsatellite markers in sea bass, Dicentrarchus labrax (Perciformes Serranidae). Mol. Mar. Biol. Biotech. 4, 62^68. Garc|ä de Leo¨n, F. J., Chikhi, L. & Bonhomme, F. 1997 Microsatellite polymorphism and population subdivision in natural populations of European sea bass Dicentrarchus labrax L. 1758. Mol. Ecol. 6, 51^62. Garcia Ramos, G. & Kirkpatrick, M. 1997 Genetic models of adaptation and gene £ow in peripheral populations. Evolution 51, 21^28. Magoulas, A., Tsimenides, N. & Zouros, E. 1996 Mitochondrial DNA phylogeny and the reconstruction of population history of a species: the case of the European anchovy (Engraulus encrasicolus). Mol. Biol. Evol. 13, 178^190. 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. Nei, M. 1978 Estimation of average heterozygosity and genetic distance from a small number of individuals. Genetics 89, 583^590. Nei, M. & Chesser, R. K. 1983 Estimation of ¢xation indices and gene diversities. A. Hum. Genet. 47, 253^259. Ovchinnikov, I. M. 1966 Circulation in the surface and intermediate layers of the Mediterranean. Oceanology 6, 48^59. Palumbi, S. R. 1992 Marine speciation on a small planet. Trends Ecol. Evol. 7, 114^118. Palumbi, S. R. 1994 Genetic divergence, reproductive isolation, and marine speciation. A. Rev. Ecol. Syst. 25, 547^572. Palumbi, S. R.1996 Macrospatial genetic structure and speciation in marine taxa with high dispersal abilities. In Molecular zoology. Advances, strategies and protocols (ed. J. D. Ferraris & S. R. Palumbi), pp. 101^118. NewYork:Wiley-Liss. Patarnello, T., Bargelloni, L., Caldera, F. & Colombo, L. 1993 Mitochondrial DNA sequence variation in the European sea bass Dicentrarchuslabrax L. (Serranidae): evidence of di¡erential haplotype distribution in natural and farmed population. Mol. Mar. Biol. Biotechnol. 2, 333^337. Pickett, G. D. & Pawson, M. G. 1994 Sea bass, biology, exploitation and conservation. London: Chapman & Hall.

Proc. R. Soc. Lond. B (2000)


935

Pinardi, N., Korres, G., Lascarotos, A., Roussenov, V. & Stanev, E. 1997 Numerical simulation of the interannual variability of the Mediterranean sea upper ocean circulation. Geophys. Res. Lett. 24, 425^428. Planes, S., Galzin, R. & Bonhomme, F. 1995 A genetic metapopulation model for reef ¢shes in oceanic islands: the case of the surgeon¢sh, Acanthurus triostetus. J. Evol. Biol. 8, 103^117. Quignard, J. P. 1978 La Me¨diterraneë: creuset ichtyologique. Boll. Zool. 45, 23^26. Reynolds, J., Weir, B. & Cockerham, C. 1983 Estimation of the coancestry coe¤cient, basis for a short term genetic distance. Genetics 105, 767^779. Rice, W. 1989 Analysis tables of statistical tests. Evolution 43, 223^225. Roughgarden, J., Gaines, S. & Possingham, H. 1988 Recruitment dynamics in complex life cycles. Science 241, 1460^1466. Saitou, N. & Nei, M. 1987 The neighbor-joining method: a new method for reconstructing phylogenetic trees. Mol. Biol. Evol. 4, 406^425. Sinclair, M. & Iles, T. D. 1989 Population regulation and speciation in the oceans. J. Conserv. Int. d’Explorat. Mer 45, 165^175. Slatkin, M. 1980 Estimating levels of gene £ow in natural populations. Genetics 99, 323^335. Slatkin, M. 1987 Gene £ow and the geographic structure of natural populations. Science 236, 787^792. Slatkin, M. 1995 A measure of population subdivision based on microsatellite allele frequencies. Genetics 139, 457^462. Smith, P. J. 1986 Genetic similarity between samples of the orange roughly Hoplostetus atlanticus from the Tasman Sea, south-west Paci¢c Ocean and north-east Atlantic Ocean. Mar. Biol. 91, 173^180. Strathmann, R. R. 1990 Why life histories evolve di¡erently in the sea? Am. Zool. 30, 197^207. Thiede, J. 1978 A glacial Mediterranean. Nature 276, 680^683. Turan, C., Carvalho, G. R. & Mork, J. 1998 Molecular genetic analysis of Atlanto-Scandian herring (Clupea harengus) populations using allozymes and mitochondrial DNA markers. J. Mar. Biol. Assoc. UK 78, 1^15. Waples, R. S. 1987 A multispecies approach to the analysis of gene £ow in marine shore species. Evolution 4, 385^400. Weir, B. & Cockerham, C. 1984 Estimating F-statistics for the analysis of population structure. Evolution 38, 1358^1370. As this paper exceeds the maximum length normally permitted, the authors have agreed to contribute to production costs.