RESEARCH ARTICLE

Sarah Samadi Æ Lionel Bottan Æ Enrique Macpherson

Bertrand Richer De Forges Æ Marie-Catherine Boisselier

Seamount endemism questioned by the geographic distributionand population genetic structure of marine invertebrates

Received: 19 July 2005 / Accepted: 1 March 2006� Springer-Verlag 2006

Abstract Previous studies have suggested that the highdiversity associated with the Norfolk seamounts(Southwest Pacific) could reflect endemism resultingfrom limited dispersal due to hydrological phenomena.Crustaceans of the family Galatheidae are thoroughlystudied in the New Caledonia economic zone permittingthe analysis of species distribution pattern between theNew Caledonia slope and Norfolk ridge seamounts.This analysis has shown that, qualitatively, the samespecies are sampled on seamounts and on the NewCaledonia slope. Local endemism was never detected.However, on each seamount, and therefore on a smallsurface, a very high number of species are usually sam-pled, suggesting that seamounts are biodiversity hotspots. Then, to evaluate whether the seamounts consti-tute patches of isolated habitat, we explore the patternof genetic diversity within several species of crustaceansand gastropods. Analysis of the intra-specific geneticstructure using the mitochondrial marker COI revealsthat populations of two Galatheidae species (Munidathoe and Munida zebra), polymorphic for this marker,are genetically not structured, both among seamounts

and between the seamounts and the island slope. Thegenetic structure over a similar sampling scheme of twoEumunida species (Chirostylidae, the sister family ofGalatheidae) and a planktotrophic gastropod (Sassiaremensa) reveals a similar pattern. Population structureis observed only in Nassaria problematica, a non-planktotrophic gastropod with limited larvae dispersal.Thus, the limitation of gene flow between seamountsappears to be observed only for species with limiteddispersal abilities. Our results suggest that the Norfolkseamounts rather than functioning as areas of ende-mism, instead, may be highly productive zones that cansupport numerous species in small areas.

Introduction

The endemism and species richness, which have beenabundantly documented in the terrestrial biota of oce-anic islands, are often explained by the acceleration ofevolutionary processes due to the fragmentation ofspecies into small isolated populations (Barton 1998).Oceanic water masses are, in this context, seen asphysical barriers between small patches of land eachoccupied by a patch of terrestrial organisms isolatedfrom other patches. An analogy has often been proposedbetween oceanic islands and similar topographical fea-tures such as mountains on land, and underwater sea-mounts in the marine realm. The background hypothesissupporting these analogies is that environmental dis-continuity between prominent topographic features is abarrier to dispersal and thus causes fragmentation intosmall isolated sub-populations. In the case of mountainsummits, the analogy is often supported by the highendemism rates observed. Mountain summits are phys-ically separated by low areas that differ by environ-mental parameters. Organisms are not supposed to besimultaneously adapted to these contrasted environ-mental conditions. Thus, this environmental disconti-nuity is supposed to reduce dispersal from one summit

Communicated by S.A. Poulet, Roscoff

S. Samadi (&) Æ L. Bottan Æ B. R. De Forges Æ M.-C. Boisselier‘‘Systematique, Adaptation et Evolution’’,UR IRD 148 /UMR 7138 UPMC; IRD; MNHN; CNRS,Service de systematique moleculaire (CNRS, IFR101),Departement Systematique et Evolution,Museum National d’Histoire Naturelle, CP 26,57 Rue Cuvier, 75231 Paris Cedex 05, FranceE-mail: sarah@mnhn.frTel.: +33-1-40793759Fax: +33-1-40793844

E. Macpherson Æ B. R. De ForgesCentro de Estudios Avanzados de Blanes (CSIC),C. acc. Cala Sant Francesc 14, 17300 Blanes, Girona, Spain

B. R. De Forges‘‘Systematique, Adaptation et Evolution’’,UMR 7138 UPMC-IRD-MNHN-CNRS (UR IRD 148),Institut de Recherche pour le Developpement, B.P. A5,98848 Noumea Cedex, Nouvelle-Caledonie, France

Marine Biology (2006)DOI 10.1007/s00227-006-0306-4

to another. Similarly, to explain the diversity andapparent endemism of the benthic seamount fauna ofthe southwest Pacific, it has been hypothesized thatseamounts can induce similar isolation between smallpopulations under the sea (Richer de Forges et al. 2000).In the study of Richer de Forges et al. (2000), 36% offish and macro-invertebrate species were new to scienceand thus never sampled from the open seafloor andclosest continental slopes. Moreover, very low speciesoverlap was observed between samples from seamountsseparated by only a few kilometres. These observationswere interpreted as an indication of high endemism rateson these seamounts, thus supporting the analogy be-tween seamounts and terrestrial islands. More precisely,to explain the apparent endemism and low speciesoverlap between seamounts, Richer de Forges et al.(2000) hypothesized that larval dispersal was limited byhydrological phenomena such as Taylor columns(Roden 1987), which result from the interaction betweenwater circulation and topography, promoting larvalretention and aggregation (Boehlert and Mundy 1993;Mullineaux and Mills 1996; Rogers 1994). This phe-nomenon could limit the efficiency of larval dispersionfor organisms inhabiting seamounts, thus inducing iso-lation and permitting subsequent speciation.

However, the few genetic studies available for speciesassociated with seamounts do not support populationisolation at a local scale. For example, Aboim et al.(2005), using two mitochondrial genes, demonstratedthat the populations of the benthopelagic fish Helicole-nus dactylopterus, sampled on seamounts separated by afew hundreds of kilometres, were not differentiated. Inthis study, a genetic differentiation between populationswas found only at larger oceanic scales (transatlantic).However, using micro-satellite loci, Aboim (2005) re-vealed that these seamounts populations were differen-tiated from a population sampled on the continentalslope, about 2,000 km apart. Although most of theavailable genetic studies concerned fishes (reviewed inCreasey and Rogers 1999), some studies on benthicorganisms over seamounts have suggested the sametrend (see for example Smith et al. 2004).

In the marine realm high faunal similarities have,however, been observed between sites 3,000 and4,000 km apart in other isolated deep-sea environmentssuch as hydrothermal vent or cold seeps. Among ventorganisms, recent population genetic studies of bivalvesfrom the genus Bathymodiolus indicate dispersal betweenvery distant sites (Won et al. 2003), and even betweenhydrothermal and cold-seep sites (Miyazaki et al. 2004).These results suggest that although vent and cold-seepenvironments are markedly fragmented, species associ-ated with these environments can be highly dispersive.Thus, such environments are more like oases, i.e. placeswhere a high trophic input allows abundance of speciesand high population density, than like islands, i.e. iso-lated patches of habitat that can accommodate onlysmall populations of a few species without exchangewith other such populations.

In contrast to the hypothesized isolation of seamountpopulations by Taylor columns, several authors havesuggested that the interaction between prominent topo-graphic features and water masses increases turbulenceand mixing, and enhances local biomass production bymoving up nutrients in the euphotic zone (Worm et al.2003; Genin 2004). Thus, by analogy with deep-sea oa-ses, an alternative hypothesis suggests that seamountsare highly productive oases that can accommodate densepopulations of many species in small areas withoutisolation between population patches. Following thishypothesis, the low species overlap observed betweenseamounts and the surrounding open sea or continentalslopes—i.e. the apparent high rate of endemism—maybe explained by insufficient sampling due to differencesin population density and local species abundances be-tween these environments.

The first step to determine which of the insularisolation versus the oases models is the best to explainseamount biodiversity is to test whether the endemismdescribed on seamounts is not an artefactual observa-tion due to (a) a more intensive sampling on sea-mounts than on surrounding open sea floors andnearby continental slopes and/or (b) the concentrationof numerous species on small seamounts areas thatquantitatively increases sampling success for a giveneffort on seamounts as compared to these others areas,and/or (c) a poor taxonomic background that com-promises the assessment of species distribution areas.We thus sought taxa for which taxonomical data areavailable and where the analyses of genetic structureamong populations were possible in order to evaluatedispersal, and thus to find the best fitting hypothesis.Among the organisms found on the seamounts of theNorfolk ridge, the squat lobsters of the families Gal-atheidae and Chirostylidae are highly diversified. Thesefamilies, which have been thoroughly studied in theNew Caledonia economic zone (de Saint Laurent andMacpherson 1990; Macpherson 1993, 1994; Baba andde Saint Laurent 1996; Machordom and Macpherson2004; Macpherson and Machordom 2005), were hereused to evaluate the pattern of specific diversity overthese seamounts.

Using five squat lobster species, we have first testedwhether the association between topography andhydrological phenomena in the seamounts of Norfolkridge leads to population fragmentation. This was doneto test whether population fragmentation and sub-sequent genetic differentiation can induce high rates ofspeciation, which could account for the high speciesdiversity observed. Towards this aim, the genetic vari-ability of three Galatheidae species belonging to thegenus Munida was analysed using mtDNA sequencesfrom individuals collected on several seamounts on theNorfolk ridge. The results were compared to the popu-lation structure observed for two species of the genusEumunida [family Chirostylidae, which is considered asthe sister group of the family Galatheidae (Morrisonet al. 2002)].

Although the taxonomic knowledge was importanton this group, very poor data about the larval life his-tory traits of squat lobsters and their dispersal capabil-ities were available. In order to evaluate the potentialrole of larval dispersal on genetic isolation betweenseamounts, two gastropod species with contrasting lar-val development were also surveyed. In Gastropods,larval development, which is easily determined byexamining the protoconch, is highly correlated to dis-persal abilities of larvae and thus, at least at local geo-graphic scale, to population structure (Collins 2001;Kyle and Boulding 2000; Boisselier-Dubayle and Gofas1999; Todd et al. 1998). If the association of topographyand hydrological phenomena in Norfolk ridge sea-mounts leads to a physical fragmentation of seamounthabitat, we would expect all the analysed species to bemore or less genetically structured. Conversely, if thereis no larval retention, we would expect only species withpoorly dispersive larvae to be structured. Finally, bycomparing results from such ecologically and phyloge-netically distant taxa as gastropods and squat lobsters,we will discuss which of the two models (insular isola-tion vs oasis model), better explains species richness onSouth Pacific seamounts.

Seamounts are also important fishery resources, andKoslow et al. (2001) showed that on Tasmanian sea-mounts intensive fishery activities developed to catchorange roughy (Hoplostethus atlanticus) had drasticallyreduced the invertebrate biomass and thus possibly thebiodiversity. Indeed, the biomass was 83% lower be-tween heavily trawled seamounts (>1,000 trawls) andthose of lightly fished (10–100 trawls). Although the highlevel of apparent endemism associated with seamountshas to be tempered against limited collections andpoorly known systematics (Richer de Forges et al. 2000),seamounts have been suggested as a target for marinebiodiversity protection (Richer de Forges and Chauvin2005). In a biological conservation perspective, it is thusquite important to know whether their richness resultsfrom endemism or from the concentration of popula-tions belonging to numerous widespread species. In bothcases, the protection of limited areas will preserve thehabitat of many species, but in the first instance (insularisolation model), rare endemic species would be pro-tected, while in the second case (oasis model) protectionwould be provided to widespread species in a minimalsurface.

Materials and methods

The fauna of Norfolk ridge seamounts was sampled bydredging and trawling during two cruises of R/V Alis.During the first cruise (NORFOLK1, June 2001), 9seamounts and the ‘‘Isle des Pins’’ slope were explored,and during the second cruise (NORFOLK2, October2003) 13 seamounts along with all sites explored duringthe previous cruise were sampled (Fig. 1). All specimensbelonging to the Galatheidae and specimens of the genus

Eumunida (Chirostylidae) were identified to species leveland stored in 70� ethanol. All the materials of thepresent study are conserved in the Museum Nationald’Histoire Naturelle, Paris.

For Galatheidae, the species accumulation curve overthe 229 stations sampled during the two NORFOLKcruises was calculated using EstimateS 7.5 (Colwell2005). We estimated the species richness using theMichaelis–Menten asymptotic function (Colwell andCoddington 1994) on species accumulation curves ob-tained through 50 random drawing of the 229 stationsusing the Bootstrap estimator and the first- and second-order Jackknife estimators.

In order to evaluate whether seamounts induce pop-ulation fragmentation, we selected three species ofMunida (M. acantha, M. thoe and M. zebra) and twospecies of Eumunida (E. annulosa and E. sternomacula-ta). Finally, to evaluate the potential role of larval dis-persal on population structure, two gastropod species,sampled on most of the seamounts, but which differ intheir larval development, were chosen. These were Sas-sia remensa (Ranellidae), which has a planktotrophicprotoconch, suggesting high larvae dispersal abilities,and Nassaria problematica (Buccinidae) whose proto-conch has non-planktotrophic characteristics suggestingan almost direct larvae development and thus limiteddispersal ability.

In each selected species we analysed about fourspecimens per seamount (Table 3). The sampling schemewas supplemented whenever possible by specimens col-lected during previous cruises on the same seamountsand stored at the MNHN in Paris. A previously col-lected specimen of M. thoe from Wallis and Futuna(North East Fiji Island) was also analysed.

In both galatheids and gastropods, DNA was isolatedfrom muscle tissue and purified using the DNeasy� 96Tissue Kit (Qiagen) following the manufacturer’sinstructions and then stored at �20�C. The cytochromeoxidase I (COI) mitochondrial gene was amplified, usinguniversal primers LCO1490 (5¢-GGTCAACAAATCATAAAGATATTGG-3¢) and HCO2198 (5¢-TAAACTTCAGG GTGACCAAAAAATCA-3¢) developed byFolmer et al. (1994). PCR reactions were performed in30 ll final volume, containing approximately 3 ng tem-plate DNA, 2.5 mM MgCl2, 0.26 mM of each nucleo-tide, 0.3 lM of each primer, 5% DMSO and 1.5 unit ofTaq Polymerase (Qbiogene). Amplification productswere generated by an initial denaturation step of 4 minat 94�C followed by 40 cycles at 94�C for 30 s, 50�C for30 s and 40 s at 72�C and by a final extension at 72�Cfor 5 min. PCR products were purified using Mon-tageTM PCR Centrifugal Filter Devices (Millipore) andsequenced (Sanger et al. 1977) on a Ceq2000TM auto-mated sequencer (Beckman), in both directions to con-firm accuracy of each haplotype sequence.

Before analysing the pattern of genetic diversity, weevaluated whether observed sequence identity could beinterpreted as identity by descent. Indeed, if the analysedsequence is saturated, an identical but homoplastic

Fig. 1 Bathymetric map of the Caledonian part of the Norfolk ridge and the name given to the explored seamounts. Northern areaAntigonia, Brachiopode, Crypthelia, Jumeau est, Jumeau ouest, Munida, Refractaire, Stylaster. Southern area Aramis, Athos, Eponge,Introuvable, Kaimon Maru, Zorro

sequence could be found in two or more isolated pop-ulation that in fact do not exchange genes. We thusperformed a saturation analysis. Patristic and observedsubstitutions matrices were computed as described byHassanin et al. (1998), using PAUP version 3.1.1(Swofford 1993).

Within each species, analyses of molecular data wereconducted using ARLEQUIN 2.0 software (Schneideret al. 2000). We first evaluated intra-specific geneticdiversity by computing the number of haplotypes, thenumber of polymorphic sites, haplotype diversity(h)—i.e. the probability that two randomly selectedhaplotypes are different (Nei 1987)—and nucleotide se-quence diversity (p)—i.e. the probability that two ran-domly selected homologous nucleotides differ (Tajima1983; Nei 1987).

Second, within each species a genetic distance matrixbetween haplotypes was calculated using the Tamura–Nei method (Tamura and Nei 1993), which was used todraw the corresponding minimum spanning networks(MSNs) that allowed the visualization of the genealog-ical relations between haplotypes and their geographicdistribution on the seamounts.

Third, we tested the correlation between the matrix ofgenetic distances between individuals, using Tamura–Nei genetic distance, and the geographical distancesusing the Mantel test (Mantel 1967), as implemented inARLEQUIN.

Fourth, intra-species population genetic structurewas tested by analysing molecular variance (AMOVA)at two hierarchical levels, i.e. between and within sea-mounts. Only seamounts for which sequence data wereavailable for two or more individuals of the same specieswere included. With haploid data (haplotypes) and

individuals grouped in local populations (i.e. by the se-amount on which they were sampled), the implementedform of the algorithm led to a fixation index FST that isidentical to the weighted average F statistics over locidefined by Weir and Cockerham (1984). The null dis-tribution of FST, which permits inference of the P valueof the estimated FST, was obtained by 10,000 permuta-tions of the haplotypes among seamounts (Excoffieret al. 1992).

Finally, within each species we also investigated usingthe analysis of mismatch distributions of pairwise dif-ferences between all individuals, implemented in Arle-quin version 2.000 software package (Schneider et al.2000) the demographic history. This analysis allows thetesting of whether a population has undergone a rapidpopulation expansion or has remained stable over time(Aboim et al. 2005). We also used Arlequin to test withineach species for departures from mutation-drift equi-librium with Tajima’s D test (Tajima 1989). Because ofsmall sampling size for each species, all populationsfrom the same sampling date (same cruise) were pooledin a single group.

Results

Taxonomic diversity of South Pacific Galatheidae

Sixty-two of the 160 known species of Galatheidae(genera Galathea and Munidopsis excluded) from theSW Pacific Ocean were identified over the 229 stationsexplored during the two NORFOLK cruises. Oursampling included 42% of the Galatheidae diversityknown in the South Pacific (Tables 1, 2) from Norfolk

Table 1 Geographical distribution of species richness (number of species sampled) of Galatheidae genera and Eumunida (Chirostylidae)on Norfolk ridge seamounts

Depth (m) Numberof stationssampled

Galatheidae Eumunidaspp.

Min Max Mean Agononidaspp.

Munidaspp.

Paramunidaspp.

Other genera(pooled)

‘‘Isle des Pins’’ slope 100 1009 423 26 4 5 0 0 1Northern seamounts Antigonia 180 577 344 18 6 15 5 4 3

Munida 86 590 357 11 4 10 5 3 3Jumeau Ouest 234 810 357 18 4 15 4 4 3Crypthelia 185 1190 394 20 2 15 1 1 3Brachiopode 276 762 401 14 2 12 2 4 4Stylaster 420 923 502 17 2 8 1 2 4Jumeau est 377 1434 538 25 4 9 1 3 4Refractaire 640 820 707 9 0 2 0 1 1

Southern seamounts Kaimon Maru 227 896 335 19 2 18 2 5 2Eponge 500 1144 612 18 6 13 3 2 4Introuvable 555 1040 639 17 4 9 3 1 5Zorro 609 1100 762 8 2 5 0 0 1Trois mousquetaires 609 1150 804 9 3 3 0 0 2

Total number of species sampled on the two Norfolk cruises 11 36 8 9 5Total number of species known in the Southwest Pacific 24 79 21 27 14

Minimum, maximum and mean depth and number of sampled stations are indicated for each seamount during the NORFOLK1 andNORFOLK2 campaigns. The first line includes the results from the stations sampled on the ‘‘Isle des Pins’’ slope. The eight following linescorrespond to the northern seamounts ordered by mean depth and the next five lines correspond to the southern seamounts and are alsoordered by mean depth. For each genus the total number of species sampled over the ‘‘Isle des Pins’’ slope and each seamount is indicated

Table 2 List of species of Galatheidae and Eumunida (Chirostylidae) identified in the samples from the NORFOLK1 and NORFOLK2campaigns and their occurrences in samples from other campaigns of the Tropical Deep-Sea Benthos program (http://www.mnhn.fr/musorstom) represented in the MNHN collections

Depth range of the stations where each species was sampled is indicated in the last column

seamounts. Except Agononida alisae, all Galatheidaespecies sampled from Norfolk seamounts during thetwo cruises were also known from the New Caledoniaisland slope (Table 2). However, only nine species wereidentified among specimens collected over the 26 sta-tions explored on the slope Isle des Pins, although thesestations covered a range of depth from 100 to 1,009 m.On average, we identified 17.8 species at 15.6 stationsper seamount. When excluding the island slope, a sig-nificant correlation was found between the total num-ber of species and the number of stations sampled perseamount (r=0.618; P=0.024). Thus, the number ofspecies sampled per seamount is a reliable measure ofthe sampling effort. The diversity recovered for Chi-rostylidae (genus Eumunida) was comparable with thatof Galatheidae, since we found 36% of the speciesknown in the South Pacific over all the sampled sea-mounts (Tables 1, 2).

The total richness of the studied area was extrapo-lated to 70 species by Bootstrap estimator and 75 and 77species, by the first- and second-order Jackknife esti-mators, respectively (Fig. 2). Thus, our sampling effortled to a sample of between 83 and 91% of the totalGalatheidae estimated species diversity over this sam-pling area.

Genetic diversity

COI haplotypes were deposited at Genbank (AccessionNos. AY 800009-800046, AY 800048, AY 800050, AY800051, AY 800055-800065, DQ 011181-011220). Gen-bank’s entry for each haplotype included the geographiclocation of the specimens displaying this haplotype, thecruise during which each specimen was sampled, and theM.N.H.N. collection access number when available.

Except M. acantha, all species studied were poly-morphic for the sequenced portion of the COI gene(Table 3). The variability was restricted to the first andthird codon positions, where all substitutions were syn-onymous. No saturation was detected among any of

the species studied at the first and the second codonposition. A very weak saturation for the third positionwas detected only for the two Eumunida species: theslope of the linear regression (S) for the third codonposition was, respectively, S=0.71 for E. annulosaand S=0.73 for E. sternomaculata. No saturation wasobserved for the Munida species and the gastropods.

Within Galatheidae, the two polymorphic Munidaspecies differed by their levels and patterns of geneticdiversity. The higher proportion of polymorphic sites,haplotype number, haplotypic and nucleotidic diversitywas obtained in M. thoe (Table 3). The 29 haplotypesidentified over the 35 M. thoe specimens were geneticallyclose, as seen in the MSN (Fig. 3a). Specimens from thesame seamounts were not obviously closer geneticallythan those from different seamounts.

Although six haplotypes were identified within the 36M. zebra specimens analysed, 31 specimens collected on10 different seamounts shared the same haplotype,leading to low haplotype diversity in this species. Thefive other haplotypes derived from this dominant hap-lotype (Fig. 3b), and only by one substitution. No geo-graphic structure was observed.

Conversely, to those observed in Munida species, theslight differences in genetic diversity between the twoEumunida species could be attributed to sampling bias.Indeed, 23 haplotypes were detected for 34 specimensanalysed in E. annulosa when 10 haplotypes were de-tected among the 20 specimens analysed for E. sterno-maculata (Table 3). The MSN (Figs. 3c, 2d) shows thatthe sequenced gene was diversified within both species,but that there was not any obvious geographic structure.

The haplotype diversity was similar in the two gas-tropod species (Table 3). However, the nucleotidediversity of N. problematica was three times higher thanthat of S. remensa. The MSN showed two groups ofdistant haplotypes in the non-planktotrophic species(N. problematica, Fig. 3f), whereas all the haplotypeswere closely related in the planktotrophic S. remensa(Fig. 3E). Moreover, the pattern was geographicallystructured in N. problematica, as all the individuals

Fig. 2 Species accumulationcurves based on EstimateS 7.5(Colwell 2005); Jackknife (Jack1, Jack 2) and Bootstraprichness estimators

sampled on the island slope were genetically distantfrom those sampled on the seamounts. Furthermore,individuals sampled on the seamounts of the northernpart of the ridge exhibited similar haplotypes, and thosefrom the southern part of the ridge were also similar toone another. However, the genetic distances betweenhaplotypes sampled on the seamounts, whether from thenorth or from the south, remained genetically closer toone another than the haplotypes from the island slope,which were genetically more distant from each other.

The trends observed in the MSN patterns were con-firmed by AMOVA analyses (Table 4), which revealedno significant among-population contribution to thetotal sequence divergences in the galatheid species and inthe planktotrophic gastropod S. remensa. Among-pop-ulation variance components were not significantly dif-ferent from zero, and we therefore were unable to rejectthe null hypothesis of absence of population structure.Conversely, for the non-planktotrophic gastropodN. problematica, the among-population variance com-ponent was highly significant (P<10�3). The nullhypothesis could thus be rejected: the populations ofN. problematica were genetically structured and the FST

estimated to 0.814.Likewise, the Mantel test (Table 5) failed to detect

significant correlation between geographical and geneticdistances in the galatheid species and in the plankto-trophic gastropod S. remensa. Conversely, the correlationcoefficient was positive and highly significant (P<10�3)in the non-planktotrophic gastropod N. problematica,where genetic distance was thus roughly in proportion togeographic distance.

Demographic history

The goodness-of-fit test to the model sudden expansionand the results of Tajima’s D test performed on eachspecies at the two sampling dates are given in Table 6. Forall genetically not structured species, all histograms (datanot shown) presented curves characteristic of populationswith constant size over time. All species presented mod-erate to highly negative Tajima’s D test values, althoughonly one,M. thoe from the second cruise, was significant.The E. sternomaculata sample from the first cruise couldnot be fitted to an expansion model. ForN. problematica,although the populations were markedly structured,due to small sampling size from each cruise, analyseswere also conducted on sub-samples pooled by cruise(Table 6). Tajima’s D test values were positive, althoughnot significant. There was also no indication of popula-tion bottlenecks followed by expansion.

Discussion and conclusions

The two NORFOLK cruises allowed a comprehensivesampling of the species richness of the sampled areasince 83–91% of the estimated total species richness ofT

able

3Sample

size

andsummary

statisticsofthegeneCOIforeach

analysedspeciescollectedonWallisandFutuna,‘‘Isle

des

Pins’’slopeandNorfolk

seamounts

Species

Wallisand

Futuna

Isle

des

Pins

Northernarea

Southernarea

Sample

size

Polymorphic

sites

Haplotype

number

Haplotype

diversity

(h)

Nucleotide

diversity

(p)

AN

BR

CR

JEJO

ST

MU

EP

INKM

Munidathoe

1(1)

14(1)

23

44

32

44

335

5%

29

0.985

0.007

Munidazebra

44

54

44(1)

42

41(1)

36

1%

60.308

0.001

Munidaacantha

8(8)

33

33

1(1)

21

0%

10

0Eumunidaannulosa

14

24

44

42

44

134

5%

23

0.963

0.009

Eumunidasternomaculata

41

45(1)

33

20

2%

10

0.863

0.003

Sassia

remensa

34

44

14

424

5%

21

0.986

0.009

Nassariaproblematica

44

44

44

24

8%

14

0.946

0.021

Thenumber

ofindividualsanalysedfrom

previouscampaignsare

given

inparentheses

AN

Antigonia,BR

Brachiopode,

CR

Crypthelia,JEJumeauest,JO

Jumeauouest,ST

Stylaster,EPEponge,

INIntrouvable,KM

Kaim

on-M

aru

4 3

4

3

3

2

2

Jumeau ouest

2

2

4

2

22

0.02

E. annulosa

E. sternomaculata

A

D

E

B C

M. thoe M. zebra

N. problematica

S. remensa

Jumeau est

Crypthelia

Antigonia Introuvable

Éponge

Brachiopode

Stylaster Kaimon-Maru

île des Pins

Wallis-et-Futuna

Nor

ther

nar

ea

Sou

ther

nar

eaO

ut o

frid

ge

Key

2

2

5

3

2

2

2

Munida

2

MT12

MT21MT20

MT15MT07

MT11

MT01 MT29

MT27

MT23

MT05MT04

MT16

MT24

MT28

MT22MT19

MT09

MT02

MT13MT17

MT10MT26

MT08 MT18

MT14MT25

MT03MT06

MZ06

MZ04MZ03

MZ01

MZ05 MZ02

EA04

EA09

EA08

EA10 EA20

EA01

EA12

EA06

EA19

EA02

EA03

EA07

EA22

EA13

EA11

EA15

EA14EA16

EA17

EA21

EA18

SR01

SR02

SR20

SR13SR10

SR07

SR03SR04

SR05

SR06

SR21

SR16

SR15

SR14

SR12

SR11

SR09

SR08

SR19

SR17SR18

NP01

NP02

NP08

NP10

NP14

NP05NP03

NP04

NP09

NP07NP06

NP11 NP12

NP13

ES02

ES07

ES04

ES03

ES08

ES05ES06

ES01

ES09

ES10

EA05

EA23

2

F

Fig. 3 Minimum spanning networks constructed using Tamura–Nei distances between mitochondrial COI haplotypes (represented bycircles); areas proportional to the number of individuals sharing a given haplotype. The symbols inside the circle represent the localitieswhere the haplotype was found (see key)

Galatheidae (excluding genera Galathea and Munidop-sis) was sampled. However, each individual seamountappeared to be richer than the restricted explored areaon the New Caledonia slope, although a similar area tothat of one seamount was sampled. Indeed, over eachseamount an average of more than one species was ad-ded per station explored and a positive correlation be-tween the number of stations explored and the numberof species sampled indicates that not all the species livingon each seamount were sampled despite the high sam-pling effort. In the two NORFOLK cruises, the numberof species added per sampled station from the NewCaledonian slope was about three times lower eventhough sampling effort was similar to that made on eachseamount explored. It is therefore likely that, for anygiven restricted area, species richness is lower along the‘‘continental’’ slope than in each explored seamountalong the Norfolk ridge (i.e. species density is lower inthe former).

The observed Galatheoidea group species richness isremarkably high on the Norfolk ridge seamounts andthe Isle des Pins slope since it represents 42% of theknown diversity of the South Pacific for Galatheideaand 36% for the genus Eumunida. The species of thefamilies Galatheidae and Chirostylidae are among thebest-inventoried marine taxa in New Caledonia. AllMuseum collections obtained from more than 28 cruises,

covering more than 1,500 stations during the last20 years, have been thoroughly studied (de Saint Lau-rent and Macpherson 1990; Macpherson 1993, 1994;Baba and de Saint Laurent 1996; Machordom andMacpherson 2004; Macpherson and Machordom 2005;Baba 2004, 2005 and K. Baba personal communication)and have allowed the description of more than 150 newspecies for the Galatheoidea group. This remarkableinventory reveals that species diversity is particularlyhigh in these families. Nevertheless, despite the highdiversity found on the Norfolk seamounts, none of thespecies sampled was found to be endemic either to oneseamount or even to the entire ridge. Indeed, all thesampled species were known from the island slope ofNew Caledonia, indicating that the Norfolk seamountsare a diversity hotspot for the family Galatheidae, butnot an area of endemism. Other studies have also sug-gested that seamounts, like other prominent topographicfeatures such as reef islands or shelf breaks, are biodi-versity hotspots (see for example Worm et al. (2003) forvertebrate predators and Heinz et al. (2004) for fora-minifera). However, our results suggest that while sea-mounts are not an area of high endemism, they arehighly productive zones where many species co-occur inlarge populations. This finding is in accordance with thesuggestion that high productivity is a prominent eco-logical feature of seamounts (Fock et al. 2002; Genin2004; Rogers 1994). The high productivity observed onseamounts could either result from enhanced primaryproduction or be due to the fact that seamounts offer alocation where the deep scattering layers can intersecthard substrate.

The analysis of population genetic structure in speciesof Galatheidae and Chirostylidae suggested that theirpopulations are genetically connected between the ex-plored seamounts. Except for M. acantha, the within-species diversities of the studied fragment of COI genewere important but no genetic structure among popu-lations was observed in most species. However, with thesame gene and with a lower level of haplotypic andnucleotide diversity, Fratini and Vannini (2002) revealedthat geographically close populations of the swimming

Table 4 Hierarchical partitioning of molecular variation within and among seamounts (AMOVA) based on COI haplotypes

Species Source of variation Degree of freedom Variance component FST P value

Eumunida annulosa Among populations 8 �0.091 NS 0.666Within populations 23 2.737

Eumunida sternomaculata Among populations 4 0.024 NS 0.304Within populations 14 0.857

Munida thoe Among populations 9 0.038 NS 0.309Within populations 23 1.917

Munida zebra Among populations 8 0.017 NS 0.056Within populations 25 0.131

Sassia remensa Among populations 5 �0.084 NS 0.745Within populations 17 2.521

Nassaria problematica Among populations 5 5.770 0.811 <10�3

Within populations 18 1.345

P value: Probability of obtaining by chance equal or more extreme random variance component and FST than the observed values

Table 5 Mantel correlation test between genetic distances (Tam-ura–Nei) matrix based on COI haplotypes and geographical dis-tances matrix between individuals

Species Correlationcoefficient

P value

Eumunida annulosa 0.000 0.89Eumunida sternomaculata �0.007 0.49Munida thoe 0.040 0.41Munida zebra 0.155 0.12Sassia remensa 0.249 0.05Nassaria problematica 0.701 <10�3

P value: Probability of obtaining by chance equal or more extremerandom correlation coefficient than the observed value

crab Scylla serrata, which has an extended planktoniclarval phase, were differentiated. Thus, the absence ofgenetic structure appears not to be a consequence of alack of variability for the marker used. The saturationanalysis confirmed that genetic homogeneity amongseamounts was also not due to homoplasy. Moreover,the variability of the marker is likely to be neutral sinceall the mutations identified were found to be synony-mous. Similarity between haplotypes is thus due tocommunity of descent and, since no selective pressuremay be inferred from the observed variability, the ab-sence of genetic structure can be interpreted to resultfrom gene flow among populations. However, this pat-tern of genetic structure may also be found in recentlyisolated population that have not experienced completelineage sorting. Moreover, if isolation was not associ-ated with a population bottleneck (e.g. an importantreduction of effective size) the time before completelineage sorting may be long. However, the high vari-ability of reproductive success in many marine organ-isms should lead to reduced effective population size(Flowers et al. 2002; Hedgecock 1994; Turner et al.1999) and thus rapid lineage sorting when two popula-tions are isolated. Moreover, seamounts are notephemeral structures (typically millions to tens of mil-lions of years in age). Thus, if hydrological phenomenaefficiently prevent gene flow among close seamountsover time, the lineage sorting should be observed in mostpopulations. Lastly, for each species, at sampling scaleof each cruise, Tajima D test revealed no departure frommutation-drift equilibrium and the mismatch analysis ofpairwise differences were characteristic of populationswith constant size over time.

In M. thoe, the haplotype of the specimen collectedfrom Wallis and Futuna, ca. 2,200 km distant, wasgenetically very close to those of specimens sampledfrom the Norfolk seamounts. This observation suggeststhat the dispersal abilities of this species allow large-scale gene flow among populations, or recent connectionbetween the populations. Our results parallel those ob-tained by Smith et al. (2004) from deep-sea bamboocorals, among which taxonomists had traditionallysuggested a high rate of endemism on seamounts andlow gene flow among distant populations. Their genetic

survey confirmed that specific diversity is high, butshowed that species are not endemic to seamounts andthat distant populations are genetically interconnected.This pattern may therefore be a general phenomenon.

Two gastropod species were used to differentiate be-tween the effects of their life cycles and those of physicalfragmentation of populations resulting from hydrologi-cal phenomena. In gastropod the morphology of theprotoconch allows inference of the type of larval devel-opment, which is itself related to dispersal ability. Thisrelation has been confirmed by many studies of popula-tion genetic structure that, comparing close species con-trasting in the duration of pelagic larval development,have shown that species with poorly dispersive larvae arehighly structured, whereas those with highly dispersivelarvae are poorly structured (Boisselier-Dubayle andGofas 1999; Collin 2001; Kyle and Boulding 2000; Toddet al. 1998). Our data confirm this trend by showing thatthe non-planktotrophic species in our species sample, N.problematica, is highly structured. By contrast, like thegalatheid species, in the planktotrophic S. remensa nogenetic structure was revealed by our study. An alter-native hypothesis is that the two species differ by theirpopulation effective size. Following this hypothesis, theabsence of genetic differentiation within S. remensa is notsubsequent to gene flow among seamounts but may bedue to a very high effective population size that has notyet permitted lineage sorting. Last, given the very highFST value estimated among populations of N. problem-atica, it might be possible that the observed populationstructure is not due to a high limitation of gene flowwithin species but to the presence of cryptic species. TheMSN revealed that the specimens from the New Cale-donia slope are genetically highly distant from the spec-imens found on the seamounts. The MSN also revealedthat, although the seamount populations are differenti-ated, the genetic distances between haplotypes are weak.It is thus possible that the species present on the NewCaledonia slope differs from that sampled on seamounts.

Other studies on Atlantic seamounts revealed thesame trends. Indeed, in the North Atlantic, Gofas(2000), examining Fasciolariidae species, and Dijkstraand Gofas (2004), studying Pectinoidea species, foundno seamount-to-seamount endemism, even between

Table 6 Neutrality test and test of goodness of fit to the model sudden expansion, for each species and each sampling date

Eumunida an-nulosa

Eunmunidasternomaculata

Munida zebra Munida thoe Sassia remensa Nassariaproblematica

N1 N2 N1 N2 N1 N2 N1 N2 N1 N2 N1 N2

Neutrality testsTajima’s D �0.62 �0.97 �0.54 �0.97 �1.29 �1.11 �1.16 �1.65 �1.06 �1.10 0.14 1.21P value 0.29 0.19 0.32 0.19 0.10 0.15 0.13 0.04 0.16 0.16 -0.44 �0.14Test of goodness-of-fitSum of squared deviation 0.03 0.01 No fit 0.90 0.00 0.00 0.01 0.00 0.01 0.05 0.04 0.12P (Sim. Ssd >= Obs. Ssd) 0.05 0.64 0.17 0.29 0.33 0.58 0.41 0.23 0.26 0.24 0.05

The two sampling dates correspond to the two cruises NORFOLK1 (N1) and NORFOLK2 (N2)In bold, significant P values

seamounts separated as much as 100 km. These resultsobtained both for galatheid species and for plankto-trophic and non-planktotrophic gastropod species sug-gest that seamounts are not particularly isolated patchesof habitat. Moreover, when comparing molluscan faunafrom North Atlantic and Azores seamounts to themainland, planktotrophic development appears over-represented (Gofas and Beu 2002). The trends observedon the distribution of molluscs on North Atlantic sea-mounts are similar to those we observed for squat-lob-ster in Norfolk ridge seamounts; there is no limitation ofdispersal and seamounts do not represent isolated pat-ches of habitat.

Overall, these results strongly suggest that thehydrological phenomena that may be associated withseamounts are not strong physical barriers for the spe-cies inhabiting them. The endemism associated withoceanic islands is traditionally explained by the accel-eration of evolutionary processes in isolated populations(Barton 1998) in which small population size and lack ofgenetic exchange with larger gene pools induce rapidgenetic drift. Moreover, local adaptation, resulting fromlocal selection, is thought to be facilitated by geneticisolation as migration does not bring misadapted geno-types. Our data suggest that, contrary to this insularisolation model, populations living on seamounts appearto be genetically connected, and the high diversity ob-served on seamounts therefore may not result from highlocal speciation rate. Both species distribution patternswithin Galatheidae and the genetic structure withinseveral species sampled on nine seamounts of the Nor-folk ridge indicate that, rather than representing areas ofendemism, these areas should be regarded as highlyproductive zones that may accommodate many speciesin densities far higher than in surrounding less-produc-tive areas. Thus, our data fit best with the oases model.

Seamounts are of great interest for biodiversity con-servation because widespread species concentrate in largenumbers and in dense populations within these smallareas. Contrary to previously held belief, however, theseintriguing areas are not characterized by distinctivefaunas that do not occur elsewhere. Protecting seamountarea would thus preserve particularly rich communitiesand ecosystems (Roberts 2002). Moreover, their preser-vation would protect number of species in a minimal areathat are easier to control. Ecosystems, at least as much asindividual species, if not more, justify protection becausethey constitute the level at which survival of individualspecies is sustainable. Apparent lack of endemism (or atleast its low level) in seamounts does not diminish theirpotential as biodiversity reserves, but rather changes theperspective in which such reserves should be created.However, as exemplified byN. problematica in our study,some species of invertebrates inhabiting seamounts havea low dispersal capacity (Parker and Tunniclife 1994); inorder to develop a better global understanding of se-amount habitats, a special effort will have to be done todescribe the reproductive strategy of additional deep-seainvertebrate groups.

Acknowledgements We are grateful to the crew of R/V Alis and thetechnical support of the IRD at Noumea; the staff of the ‘‘Servicede Systematique Moleculaire’’ at the MNHN for technical facili-ties; Philippe Bouchet, Pierre Lozouet, and Alan Beu for theidentification of gastropod specimen; Alain Crosnier for providingaccess to the MNHN crustacean collection; Simon Tillier forconstructive comments and discussion on the manuscript; Porter P.Lowry for English improvement of the manuscript; Robin Wapplesfor comments on a previous version of this manuscript. Theexperiments complied with the current laws of France.

References

Aboim MA (2005) Population genetics and evolutionary history ofsome deep-sea demersal fishes from the Azores - North Atlan-tic. University of Southampton, Faculty of Enginering Scienceand Mathematics, School of Ocean and Earth Science, PhDThesis, 167 pp

Aboim MA, Menezes GM, Schlitt T, Rogers AD (2005) Geneticstructure and history of populations of the deep-sea fish Heli-colenus dactylopterus (Delaroche, 1809) inferred from mtDNAsequence analysis. Mol Ecol 14:1343–1354

Baba K (2004) Uroptychodes, new genus of Chirostylidae (Crus-tacea: Decapoda: Anomura), with description of three newspecies. Sci Mar 68:97–116

Baba K (2005) Deep-sea chirostylid and galatheid crustaceans(Decapoda: Anomura) from the Indo-Pacific, with a list ofspecies. Galathea Rep 20:1–317

Baba K, Saint-Laurent M de (1996) Crustacea Decapoda: rev on ofthe genus Bathymunida Balss, 1914, and description of six newrelated genera (Galatheidae). Mem Mus Nat Hist Nat 168:433–502

Barton NH (1998) Natural selection and random genetic drift ascauses of evolution on islands. In: Grant PR (ed) Evolution onislands. Oxford University Press, Oxford, pp 102–123

Boehlert GW, Mundy BC (1993) Ichtyoplancton assemblages atseamounts and oceanic islands. Bull Mar Sci 53:336–361

Boisselier-Dubayle MC, Gofas S (1999) Genetic relationships be-tween marine and marginal-marine populations of Cerithiumspecies from the Mediterranean Sea. Mar Biol 135:671–682

Collin R (2001) The effects of mode of development on phyloge-ography and population structure of North Atlantic Crepidula(Gastropoda: Calyptraeidae). Mol Ecol 10:2249–2262

Colwell RK (2005) EstimateS: Statistical estimation of speciesrichness and shared species from samples. Version 7.5. User’sguide and application published at: http://www.purl.oclc.org/estimates

Colwell RK, Coddington JA (1994) Estimating terrestrial biodi-versity through extrapolation. Phil Trans Roy Soc (B) 345:101–118

Creasey S, Rogers AD (1999) Population genetics of bathyal andabyssal organisms. Adv Mar Biol 35:3–151

Dijkstra HH, Gofas S (2004) Pectinoidea (Bivalvia: Propeamuss-iidae and Pectinidae) from some northeastern Atlantic sea-mounts. Sarsia 89(1):33–78

Excoffier L, Smouse PE, Quattro JM (1992) Analysis of molecularvariance inferred from metric distances among DNA haplo-types: application to human mitochondrial DNA restrictiondata. Genetics 131:479–491

Flowers JM, Schroeter SC, Burton RS (2002) The recruitmentsweepstakes has many winners: genetic evidence from the seaurchin Strongylocentrotus purpuratus. Evolution 56:1445–1453

Fock H, Uiblein F, Koester F, von Westernhagen H (2002) Bio-diversity and species-environment relationships of the demersalfish assemblage at the Great Meteor Seamount (subtropical NEAtlantic), sampled by different trawls. Mar Biol 141:185–199

Folmer O, Black M, Hoen W, Lutz R, Vrijenhoek R (1994) DNAprimers for amplification of mitochondrial cytochrome c oxi-dase subunit I from diverse metazoan invertebrates. Mol MarBiol Biotechnol 3:294–299

Fratini S, Vannini M (2002) Genetic differentiation in the mud crabScylla serrata (Decapoda: Portunidae) within the Indian Ocean.J Exp Mar Biol Ecol 272:103–116

Genin A (2004) Bio-physical coupling in the formation of zoo-plankton and fish aggregations over abrupt topographies.J Mar Syst 50:3–20

Gofas S (2000) Four species of the family Fasciolariidae (Gastro-poda) from the North Atlantic seamounts. J Conchol 37:7–16

Gofas S, BeuA (2002) Tonnoidean gastropods of theNorthAtlanticseamounts and the Azores. Amer Malacol Bull 17:91–108

Hassanin A, Lecointre G, Tillier S (1998) The ‘evolutionary signal’ ofhomoplasy in protein-coding gene sequences and its consequencesfor a priori weighting in phylogeny. Evolution 321:611–620

Hedgecock D (1994) Does variance in reproductive success limiteffective population sizes of marine organisms? In: BeaumontAR (ed) Genetics and evolution of aquatic organisms. Chap-man and Hall, London, pp 122–134

Heinz P, Ruepp D, Hemleben C (2004) Benthic foraminiferaassemblages at Great Meteor Seamount. Mar Biol 144:985–998

Koslow JA, Gowlett-Holmes K, Lowry J, O’Hara T, Poore G,Williams A (2001) The seamount benthic macrofauna offsouthern Tasmania: community structure and impacts oftrawling. Mar Ecol Progr Ser 213:111–125

Kyle CJ, Boulding EG (2000) Comparative population geneticstructure of marine gastropods (Littorina spp.) with and with-out pelagic larval dispersal. Mar Biol 137:835–845

Machordom A, Macpherson E (2004) Rapid radiation and crypticspeciation in galatheid crabs of the genus Munida and relatedgenera in the South West Pacific: molecular and morphologicalevidence. Mol Phyl Evol 33:259–279

Macpherson E (1993) Crustacea Decapoda: Species of the genusParamunida Baba, 1988 (Galatheidae) from the Philippines,Indonesia and New Caledonia. Mem Mus Nat Hist Nat156:443–473

Macpherson E (1994) Crustacea Decapoda: Studies on the genusMunida Leach, 1820 (Galatheidae) in New Caledonian andadjacent waters with descriptions of 56 new species. Mem MusNat Hist Nat 161:421–569

Macpherson E, Machordom A (2005) Description of three siblingnew species of the genus Munida Leach, 1820 (Decapoda,Galatheidae) from New Caledonia using morphological andmolecular data. J Nat Hist 39:819–834

Mantel N (1967) The detection of disease clustering and a gener-alized regression approach. Cancer Res 27:209–220

Miyazaki JI, Shintaku M, Kyuno A, Fujiwara Y, Hashinoto J,Iwasaki H (2004) Phylogenetic relationships of deep-sea musselsof the genus Bathymodiolus (Bivalvia: Mythilidae). Mar Biol144:527–535

Morrison CL, Harvey AW, Lavery S, Tieu K, Huang Y,Cunningham CW (2002) Mitochondrial gene rearrangementconfirm the parallel evolution of the crab-like form. Proc R SocLondon B Biol Sci 269:345–350

Mullineaux LS, Mills SW (1996) A test of the larval retentionhypothesis in semount-generated flows. Deep Sea Res I 44:745–770

Nei M (1987) Molecular evolutionary genetics. Columbia Univer-sity Press, New York

Parker T, Tunnicliffe V (1994) Dispersal strategies of the biota onan oceanic seamount: implications for ecology and biogeogra-phy. Biol Bull 187:336–345

Richer de Forges B, Koslow JA, Poore GC (2000) Diversity andendemism of the benthic seamount fauna in the southwest Pa-cific. Nature 405:944–947

Richer de Forges B, Chauvin C (2005) Indo-Pacific deep-sea fauna:species richness and vulnerability of seamount fauna. Assisesde la Recherche Francaise dans le Pacifique 24–27 Aout 2004,37–38 (Abstract)

Roberts CM (2002) Deep impact: the rising toll of fishing in thedeep sea. TREE 17:242–245

Roden GI (1987) Effects of seamounts and seamount chains onoceanic circulation and thermocline structure. In: Keating BHet al. (eds) Seamounts, islands and atolls, Geophysical Mono-graphs Ser 43. AGU, Washington DC pp 335–354

Rogers AD (1994) The biology of seamounts. Adv Mar Biol30:305–350

Saint Laurent M de, Macpherson E (1990) Crustacea Decapoda:Le genre Eumunida (Chirostylidae) dans les eaux neo-caledon-iennes. Mem Mus Nat Hist Nat 145:227–288

Sanger F, Nicklen S, Coulson AR (1977) DNA sequencing withchain-terminating inhibitors. Proc Natl Acad Sci USA 74:5463–5467

Schneider S, Dueffer JM, Roessli D, Excoffier L (2000) Arlequinver 2.0: A software for population genetics data analysis.Genetics and Biometry Laboratory, University of Geneva,Switzerland. URL: http://www.anthropologie.unige.ch/arle-quin

Smith PJ, McVeagh SM, Mingoia JT, France SC (2004) Mito-chondrial DNA sequence variation in deep-sea bamboo coral(Keratoisidinae) species in the southwest and northwest PacificOcean. Mar Biol 144:253–261

Swofford DL (1993) PAUP: phylogenetic analysis using parsimony.Illinois Natural History Survey, Champaign

Tajima F (1983) Evolutionary relationship of DNA sequences infinite populations. Genetics 105:437–460

Tajima F (1989) Statistical method for testing the neutral mutationhypothesis by DNA polymorphism. Genetics 123:585–595

Tamura K, Nei M (1993) Estimation of the number of nucleotidesubstitutions in the control region of mitochondrial DNA inhumans and chimpanzees. Mol Biol Evol 10:512–526

Todd CD, Lambert WJ, Thorpe JP (1998) The genetic structure ofintertidal populations of two species of nudibranch molluscswith planktotrophic and pelagic lecithotrophic larval stages: arepelagic larvae ‘‘for’’ dispersal? J Exp Mar Biol Ecol 228:1–28

Turner TF, Richardson LR, Gold JR (1999) Temporal geneticvariation of mtDNA and effective female population size of reddrum in the northern Gulf of Mexico. Mol Ecol 8:1223–1230

Weir BS, Cockerham CC (1984) Estimating F-statistics for theanalysis of population structure. Evolution 38:1358–1370

Won Y, Hallan SJ, O’Mullan GD, Vrijenhoek RC (2003) Dispersalbarriers and isolation among deep-sea mussel populations(Mytilidae: Bathymodiolus) from eastern Pacific hydrothermalvents. Mol Ecol 12:3185–3190

Worm B, Lotze HK, Myers RA (2003) Predators diversity hotspotsin the blue ocean. Proc Natl Acad Sci USA 100:9884–9888