The population genetic structure of Littorina littorea (Mollusca: Gastropoda) along a pollution gradient in the Scheldt estuary (The Netherlands) using RAPD analysis

Science of the Total Environment 325(2004) 59–69

0048-9697/04/$ - see front matter� 2003 Elsevier B.V. All rights reserved.doi:10.1016/j.scitoten.2003.11.004

The population genetic structure ofLittorina littorea (Mollusca:Gastropoda) along a pollution gradient in the Scheldt estuary(The

Netherlands) using RAPD analysis

Hans De Wolf *, Ronny Blust , Thierry Backeljaua, a a,b

Departments of Biology, Ecofysiology, Biochemistry and Toxicology, University of Antwerp – RUCA, Groenenborgerlaan 171,a

B-2020 Antwerp, BelgiumRoyal Belgian Institute of Natural Sciences, B-1000 Brussels, Belgiumb

Received 24 June 2003; received in revised form 23 October 2003; accepted 16 November 2003

Abstract

The population genetic structure of the periwinkleLittorina littorea was analysed using random amplifiedpolymorphic DNA (RAPD). Three primers, coding for six putative polymorphic loci were surveyed to infer thegenetic structure of seven populations located along the heavily polluted Western(i.e. in order of decreasing pollutionload W1, W2, W3 and R1) and the relatively clean Eastern Scheldt(E1, E2 and E3) estuary(The Netherlands). Agenetic distance based UPGMA(Unweighted pair group method with arithmetic mean) dendrogram revealed anestuary-related structuring, as Eastern and Western Scheldt sites formed two separate clusters. The Western Scheldtcluster was, however, much more heterogeneous, with three RAPD loci revealing a significant genetic heterogeneitycompared to none when the Eastern Scheldt sites were compared. Overall mean heterozygosity levels were high, butdid not reveal a difference between the estuaries. The current data(1) confirm the patterns of variation previouslyobserved with electrophoretic analyses of esterases and(2) strongly support that these patterns of variation have agenetic basis, in the presence of intense gene flow. In addition, it is suggested that selection, rather than bottleneckeffects, induced by the less favourable living conditions at W1, W2 and W3 are responsible for the genetic patterning.� 2003 Elsevier B.V. All rights reserved.

Keywords: Estuary; Heavy metals;Littorina littorea; Random amplified polymorphic DNA(RAPD); Salinity

1. Introduction

Environmental toxicants are known to affect thegenetic structure of natural populations(e.g. Mor-ton, 1993; Belfiore and Anderson, 2001), eitherthrough the direct action of the toxicant at the

*Corresponding author. Tel.:q32-3-218-04-78; fax:q32-3-218-04-97.

E-mail address: [email protected](H. De Wolf).

DNA level (direct mutagenic effect) or via toxi-cant-mediated mortality andyor curtailment ofreproduction (population genetic effects) (e.g.Becerril et al., 2001; Belfiore and Anderson, 2001;Theodorakis et al., 2001). The latter may indirectlyresult from(1) mutagenic effects(e.g. Hebert andLuiker, 1996; Bickham et al., 2000), but may alsobe achieved when(2) the toxicant interferes withthe physiology of an organism(physiological

60 H. De Wolf et al. / Science of the Total Environment 325 (2004) 59–69

effects) (e.g. Farag et al., 1994; Depledge andBillinghurst, 1999; Troncoso et al., 2000), or when(3) it alters the environment in which the organismhas to live(ecological effects) (e.g. Lande, 1998;Beasley and Kneale, 2002; Edwards, 2002).

A mutagenic toxicant may affect both germ andsomatic cells. When mutagenic effects occur insomatic cells, they may lead to cell death or maytransform the cell into malignancy(Becerril et al.,1999; Bickham et al., 2000). These effects areoften delayed until later in life and result in areduced viability or health of the affected individ-ual (Hebert and Luiker, 1996). In contrast, muta-tions in germ cells can be passed on to theoffspring, introducing new genetic variation intothe population, thus directly affecting its geneticstructure(Staton et al., 2001). Germ cell mutationswith large effects, however, will often lower theviability of gametes, embryos and neonates, lead-ing to a fitness reduction(e.g. Hebert and Luiker,1996; Bickham et al., 2000). Likewise, toxicantsthat interfere with the animals’ physiology, affect-ing the integrity of membranes, homeostasis andother physiological processes will result in a fitnessdecrease(Nony and Schnellmann, 2001), whiletoxicants that affect ecological resources such asfood availability, presence of other species, etc.,may indirectly affect survival andyor reproductionof animals in contaminated areas as well(e.g.Beasley and Kneale, 2002; Edwards, 2002).

If toxicant-mediated mutagenic, physiological orecological effects do result in large-scale mortalityor if they reduce the population recruitment, agenetic bottleneck may occur, decreasing thegenetic variation as low frequency, and rare allelesmay be lost during this process(e.g. Gillespie andGuttman, 1998; Belfiore and Anderson, 2001;Theodorakis et al., 2001). However, if certaingenotypes are more susceptible to mutagenicityandyor toxicant-related physiological or ecologicaleffects than others, the toxicant may act as aselective agent upon loci that are crucial for thesurvival of the animal(e.g. Gillespie and Guttman,1998; Belfiore and Anderson, 2001). Allele fre-quencies at these loci may be shifted as well inrelatively short periods of time, depending on thestrength of the selective force. Besides causingshifts in allele frequencies, both selection and

bottlenecks have the potential to lower the geneticvariability (heterozygosity and polymorphism) inpopulations, thus reducing their future adaptivepotential(e.g. Hebert and Luiker, 1996; Bickhamet al., 2000; Staton et al., 2001).

Hence, toxicant-related mutagenic, physiologicalandyor ecological effects may affect the geneticpopulation structure indirectly via population levelmediated processes, while mutagenic substancesmay affect the genetic structure in a direct manneras well (e.g. Theodorakis et al., 1998, 2001;Bickham et al., 2000; Belfiore and Anderson,2001).

Obviously, effects of environmental toxicantsare difficult to discern in the field from otherfactors that might equally well affect the geneticvariability. Indeed, the genetic make-up of popu-lations has always been affected by natural factors,leading to extinctions in the past and resulting inthe evolution of new species(Bickham et al.,2000). However, due to the accelerated speed inwhich man currently changes the environment,including the production of environmental toxi-cants, it becomes increasingly more difficult forpopulations to adapt(Guttman, 1994). As a con-sequence, the present extinction rate is estimatedto be 10–100 times the historical background(Pimm et al., 1995), thus seriously threatening thebiodiversity(Depledge, 1996; Lande, 1998).

Against this background a population geneticsurvey, using isoelectric focussing of esterase(EST) loci was performed, using populations ofthe periwinkle Littorina littorea, collected alongthe Scheldt estuary(The Netherlands) (De Wolfet al., 2001a). This estuary consists of two adja-cent, but separated arms(i.e. Eastern and WesternScheldt) of which the western one is highly pol-luted, while the eastern one is relatively clean(Gerringa et al., 1996). A few years ago, thewestern arm was even placed among the mostheavily metal polluted estuaries of the world(Bay-ens, 1998), as it is a major drain for industrial anddomestic wastes. Pollution decreases from themainland towards the North Sea(i.e. downstream),creating complex pollution gradients, which arealso reflected in the soft tissues ofL. littorea (DeWolf et al., 2000). In addition, and in contrast tothe eastern part, there is a downstream increasing

61H. De Wolf et al. / Science of the Total Environment 325 (2004) 59–69

salinity gradient, which sets the distribution limitsfor L. littorea along the Western Scheldt(De Wolfet al., 2001b). Despite the periwinkle’s gene flowpotential (via planktonic development) and thusability to continuously introduce lost andyor newgenetic variation into populations, a non-isolationby distance-related genetic structuring was appar-ent (De Wolf et al., 2001a). Indeed, a multi-dimensional scaling procedure, performed on anelectrophoretic EST profile based Dice similaritydistance matrix, revealed two clusters, separatingthe Eastern and Western Scheldt sampling sites(De Wolf et al., 2001a). In addition, geneticvariability, expressed as mean number of isoelec-tric focussed EST bands per site, differed signifi-cantly, as the least polluted and most marine-likesites revealed the highest variable EST profiles(De Wolf et al., 2001a). Apparently, pollutionandyor natural factors like low salinity(1) affectedthe genetic population structure and(2) seemed todecrease genetic variability, thus decreasing theadaptive potential ofL. littorea along the estuary.The mechanisms responsible(i.e. selection andyorbottleneck effects) for this structuring remained,however, to be determined. In addition to selectionandyor bottleneck effects, the expression of proteinmarkers such as EST loci may be affected as wellthrough gene regulation, secondary modificationsof the EST enzyme, etc., which could equally wellexplain the observed molecular patterning. There-fore, additional genetic markers and informationare needed to confirm the EST structuring. Hence,we used RAPD analysis to describe the populationgenetic structure ofL. littorea at different sitesalong the Scheldt estuary(eastqwest), comparingthe current results with the previously obtainedEST results, enabling us to discriminate betweenecophysiological and genetic effects. In addition,the current analysis will help us to formulatehypotheses regarding the underlying mechanism(s)that are responsible for the population geneticstructure ofL. littorea along the polluted Westernand relatively clean Eastern Scheldt estuary.

2. Materials and methods

2.1. Sample collection and DNA extraction

On 8th August, 1998,L. littorea was collectedat seven sites(20 individuals per site) along the

Western (W) and Eastern(E) Scheldt. In theWestern Scheldt these sites were, in order ofdecreasing pollution and increasing salinity: Han-sweert (W1), Ellewoutsdijk (W2) and Borssele(W3). Westkapelle(R1), located at the very endof the Western Scheldt, facing the North Sea, waschosen as a marine reference site, while Krabben-dijke (E1), Yerseke (E2) and Kattendijke(E3)were sampled in the Eastern Scheldt(Fig. 1).After collection, the shells were removed and theanimals were sexed based on the presenceyabsenceof the vesicula seminalis. Because intersex or themaculinisation condition in which the paliale ovi-duct of females is gradually transformed into aprostate gland and sperm groove and penis maydevelop, has been known to occur along theScheldt estuary(De Wolf et al., 2001c), femaleswere further grouped into ‘normal’ and ‘intersex’individuals. Genomic DNA was extracted fromindividual columellar muscles following the CTAB(Cetyltrimethylammonium bromide) protocol ofWinnepenninckx et al.(1996), modified so thattissues were not ground under liquid nitrogen, butinstead, directly incubated in CTAB buffer.

2.2. RAPD amplification

PCR reactions were performed in a final volumeof 25 ml, containing 25 pmol RAPD primer, 50–100 ng template DNA and a standard quantity‘Ready To Go’ RAPD analysis mixture(Amer-sham Pharmacia Biotech Inc.). PCR conditionswere: 5 min at 958C followed by 45 cycles, of30 s at 958C, 60 s at 408C, and 120 s at 728C.After the last cycle, a final step of 5 min at 728Cwas added to allow complete extension of allamplified fragments. The amplified products(10ml) were separated for 90 min at 100 V on 0.8%agarose, 0.025% ethidium bromide TAE gels. Gelswere photographed under UV light.

Forty random decamer primers were screened(Kits H and G of Operon Technologies), three ofwhich OPG08 (59-TCACGTCCAC-39), OPG20(59-TCTCCCTCAG-39) and OPH10 (59-CCTACGTCAG-39) were retained for further anal-ysis. RAPD bands were scored as presentyabsentand only well-resolved bands were considered.The reproducibility of the RAPD profiles was


Fig. 1. Study area and sampling sites(i.e. Hansweert(W1), Ellewoutsdijk(W2), Borssele(W3), Westkapelle(R1), Krabbendijke(E1), Yerseke(E2) and Kattendijke(E3)).

tested for each primer, by reanalysing five individ-uals from each site.

2.3. Data analysis

Under the assumption that each RAPD bandrepresents a single diallelic locus in Hardy–Wein-berg equilibrium, a condition which seems war-ranted forL. littorea, given previous allozyme data(e.g. Fevolden and Garner, 1987; Foltz et al.,1993), the presentyabsent(1y0) data set can beconverted into allele frequencies(e.g. Apostol etal., 1996; De Wolf et al., 1998). However, becauseRAPD bands behave as dominant markers(i.e.1sAA or Aa and 0saa), allele frequency esti-mates are less accurate than with codominantmarkers(Lynch and Milligan, 1994). Therefore,

unbiased estimates of the recessive allele frequen-cies were obtained using the method of Lynch andMilligan (1994), as implemented by the softwarepackage RAPDFST v.4.0.1(Black, 1997). Thispackage was used to estimate genetic populationdifferentiation, by means ofF andx contingen-2

ST

cy table analysis(x s2NF ; degrees of freedom2ST

(d.f.)snumber of subpopulationsy1; Nsnumberof individuals). F values were calculated withinST

both Scheldt arms and were used to estimate geneflow levels (Nm) according to Wright(1965).

Neutrality was tested at each locus, using thePopgene v1.32 package(Yeh and Boyle, 1996),implementing the Ewens–Watterson test for neu-trality. Upper and lower boundaries of the 95%confidence region were calculated after 1000 per-mutations. Pairwise linkage disequilibrium among


Table 1Allele frequencies(0srecessive; 1sdominant allele;Nsnumber of individuals analysed) at all analysed loci from all samplingsites;(for abbreviations of sampling sites see Section 2)

W1 W2 W3 R1 E1 E2 E3

OPG08-1 N 20 20 20 20 20 20 200 0.8367 0.9220 0.6882 0.8944 0.7071 0.8367 0.89441 0.1633 0.0780 0.3118 0.1056 0.2929 0.1633 0.1056

OPH10-1 N 20 20 20 20 20 20 200 0.8660 0.9747 0.8885 0.7746 0.8367 0.9220 0.89441 0.1340 0.0253 0.1115 0.2254 0.1633 0.0780 0.1056

OPH10-2 N 20 20 20 20 20 20 200 0.9747 1.0000 1.0000 0.8944 0.8944 0.8367 0.89441 0.0253 0.0000 0.0000 0.1056 0.1056 0.1633 0.1056

OPG20-1 N 20 20 20 20 20 20 200 0.8660 0.8660 0.9733 0.7746 0.8062 0.7454 0.70711 0.1340 0.1340 0.0267 0.2254 0.1938 0.2546 0.2929

OPG20-2 N 20 20 20 20 20 20 200 0.6325 0.7071 0.6489 0.8367 0.9220 0.9129 0.92201 0.3675 0.2929 0.3511 0.1633 0.0780 0.0871 0.0780

OPG20-3 N 20 20 20 20 20 20 200 0.9487 0.9487 0.9733 0.7746 0.9487 0.8819 0.94871 0.0513 0.0513 0.0267 0.2254 0.0513 0.1181 0.0513

alleles at all RAPD loci were calculated with theRAPDLD v2.0 software(Black, 1996), using ax goodness-of-fit test.2

Heterozygosity levels were computed at eachlocus for each population, while mean heterozy-gosity levels and standard errors were calculatedfor each sampling site as well, using the Popgenev1.32 package(Yeh and Boyle, 1996). Possibledifferences in heterozygosity levels, between bothestuaries were evaluated for each locus by meansof the non-parametric Mann–WhitneyU-test,using the software package Statistica(Statsoft,1998).

Prevosti genetic distances were calculatedamong all sampling sites and were used to con-struct a UPGMA topology, with the BIOSYSpackage(Swofford and Selander, 1989). A signif-icance level of 5% was used throughout, while aBonferroni correction was applied to correct forfalse assignments of significance by chance alone(multiple test problem).

3. Results

The three primers retained for further analysisyielded six polymorphic and reproducible RAPDbands, representing six putative loci: one with

OPG08; two with OPH10 and three with OPG20(Table 1). None of these RAPD bands was diag-nostic for sexual status, estuary or specific sam-pling sites, although OPH10-2 was only expressedin one individual from the upstream-located West-ern Scheldt sites, whereas it frequently occurredin the downstream Western Scheldt part(i.e. West-kapelle) and in the Eastern Scheldt(Table 1).

The Ewens–Watterson test revealed acceptablefits for neutral expectations, except for OPH10-2in Hansweert (W1), OPH10-1 in Ellewoutsijk(W2), OPG20-1 and OPG20-3 in Borssele(W3)(Table 2). However, given the large number oftests performed(42), these four departures fromneutrality were not considered significant. Linkagewas observed in 16 out of 100 pairwise locicomparisons(Table 3). After Bonferroni correctionfor multiple testing, only the combination OPG20-2 OPG20-3 remained significant and this at allthree Eastern Scheldt sampling sites(Table 3).

F values indicated a high degree of populationST

differentiation in the Western Scheldt, with threeout of six RAPD loci revealing a significantgenetic heterogeneity(Table 4). In contrast, nosignificant population differentiation could bedetected among the sampling sites within the


Table 2Results of the Ewens–Watterson test for neutrality(F ssum of squared observed allele frequencies, L95%slower boundary ofobs

the 95% confidence region, U95%supper boundary of the 95% confidence region.(for abbreviations of sampling sites see Section2)

Locus W1 W2 W3 R1

Fobs L95% U95% Fobs L95% U95% Fobs L95% U95% Fobs L95% U95%

OPG08-1 0.7267 0.5000 0.9050 0.8561 0.5000 0.9050 0.5709 0.5014 0.9003 0.8111 0.5050 0.9050OPH10-1 0.7679 0.5000 0.9050 0.9506 0.5000 0.9050 0.8019 0.5014 0.9003 0.6508 0.5050 0.9050OPH10-2 0.9506 0.5000 0.9050 1.0000 – – 1.0000 – – 0.8111 0.5050 0.9050OPG20-1 0.7679 0.5000 0.9050 0.7679 0.5000 0.9050 0.9481 0.5014 0.9003 0.6508 0.5050 0.9050OPG20-2 0.5351 0.5000 0.9050 0.5858 0.5000 0.9050 0.5443 0.5014 0.9003 0.7267 0.5050 0.9050OPG20-3 0.9026 0.5000 0.9050 0.9026 0.5000 0.9050 0.9481 0.5014 0.9003 0.6508 0.5050 0.9050

Locus E1 E2 E3

Fobs L95% U95% Fobs L95% U95% Fobs L95% U95%

OPG08-1 0.5858 0.5050 0.9050 0.7267 0.5000 0.9050 0.8111 0.5050 0.9050OPH10-1 0.7267 0.5050 0.9050 0.8561 0.5000 0.9050 0.8111 0.5050 0.9050OPH10-2 0.8111 0.5050 0.9050 0.7267 0.5000 0.9050 0.8111 0.5050 0.9050OPG20-1 0.6875 0.5050 0.9050 0.6204 0.5000 0.9050 0.5858 0.5050 0.9050OPG20-2 0.8561 0.5050 0.9050 0.8409 0.5000 0.9050 0.8561 0.5050 0.9050OPG20-3 0.9026 0.5050 0.9050 0.7917 0.5000 0.9050 0.9026 0.5050 0.9050

Table 3Results of the linkage disequilibrium tests performed at each sampling site for all possible pairwise loci combinations(only signif-icant results are tabulated; for abbreviations of sampling sites see Section 2)

Locus W1 W2 W3 R1 E1 E2 E3

OPG08-1 OPG20-2 0.017OPH10-1 OPH10-2 0.0062OPH10-1 OPG20-1 0.0311OPH10-1 OPG20-2 0.0103OPH10-1 OPG20-3 0.0227 0.0388OPH10-2 OPG20-1 0.0186OPH10-2 OPG20-3 0.0426OPG20-1 OPG20-2 0.0100 0.0105 0.0393OPG20-1 OPG20-3 0.0421 0.0144OPG20-2 OPG20-3 0.0004* 0.0007* 0.0004*

*Significant after Bonferroni correction.

Eastern Scheldt(Table 4). Average gene flowestimates(Nm) are nonetheless high, ranging from4.5 within the Western Scheldt to 17.8 within theEastern Scheldt(Table 4).

Locus-specific heterozygosity levels, mean het-erozygosity levels and standard errors are graphi-cally represented in Fig. 2. Although meanheterozygosity levels are comparable between thedifferent sites, locus-specific heterozygosity levelsdiffer significantly. Indeed, heterozygosity at locusOPH10-2 is significantly lower in the Western

Scheldt (Ps0.026), whereas, heterozygosity atlocus OPG20-2 is significantly lower in the EasternScheldt(Ps0.032). Although not significant(Ps0.074), heterozygosity at locus OPG20-1 is moreelevated in the Eastern Scheldt. The heterozygositylevels at the remaining loci are comparablebetween both Scheldt arms.

The Prevosti genetic distance based UPGMAtopology is shown in Fig. 3. The topology reflectsthe geographical positioning of the different sam-pling sites, yielding two clusters, one joining the


Table 4Summary of gene flow estimates(Nm), F -statistics andST

results of ax test for possible genetic heterogeneity within2

the Eastern and Western Scheldt

Western Scheldt Eastern Scheldt

Locus FST P Nm FST P Nm

OPG08-1 0.059 0.0261 4.0 0.040 0.0891 6.0OPH10-1 0.047 0.0588 5.0 0.012 0.4766 20.6OPH10-2 0.059 0.0256 4.0 0.007 0.6655 35.5OPG20-1 0.043 0.0801 5.6 0.009 0.5845 27.5OPG20-2 0.031 0.1763 7.8 0.000 0.9863OPG20-3 0.078 0.0063 3.0 0.014 0.4370 17.6Mean 0.053 0.0001 4.5 0.014 0.6255 17.6

Fig. 3. Prevosti’s genetic distance UPGMA topology, based onall six RAPD loci.

Fig. 2. Mean(A) and locus-specific heterozygosity levels(i.e. A: OPG08-1; B: OPH10-1, 2; C: OPG20-1, 2, 3) and(D) genotypicdistribution at locus OPH10-2 observed in the Eastern and Western Scheldt sampling.

Western Scheldt sites, except for Westkapelle, andone joining the Eastern Scheldt sites andWestkapelle.

4. Discussion

L. littorea has been the subject of many popu-lation genetic surveys, using allozyme data. Most


of these studies failed, however, to detect geneticstructuring both at micro- and macro-geographicalscales(e.g. Berger, 1973; Wilkins and O’Reagan1980; Janson, 1987; Johannesson, 1992). This hasbeen explained by the planktonic development andthus presumed gene flow potential of the species(Johannesson, 1992). Indeed,L. littorea is a plank-tonic developer, producing long-lived(4–7 weeks)free-floating larvae which are presumed to disperseover long distances. Consequently, new recruitsmay arise from a variety of parental genotypes,thus counteracting possible local genetic differen-tiation by genetic drift. This seems to hold for thepopulations ofL. littorea in the Eastern Scheldt,but does not seem to apply to the populations fromthe Western Scheldt, where three loci revealedsignificant intra estuary differences. These resultsare in line with previous EST data from the samesampling sites(De Wolf et al., 2001a), revealing(1) an Eastern–Western Scheldt related geneticdifferentiation;(2) a genetic structuring within theWestern Scheldt which seems to correlate with asalinityypollution gradient; and(3) an EasternScheldt positioning of the Western Scheldt ‘refer-ence’ sampling site Westkapelle. Hence, the ESTpatterns ofL. littorea along the Scheldt estuaryseem to have a genetic basis. In order to ensurethat the current patterns of genetic variation aretemporarily stable, the RAPD analysis will haveto be performed again in the future, on individualscollected at the same sampling sites, using thesame RAPD loci. Nevertheless, the fact that strik-ingly similar patterns of genetic variation emergeat different points in time using EST and RAPDanalysis, suggests temporal stability.

The question that arises now is how to explainthese patterns of variation among and within theEastern and Western Scheldt. Heavy metals suchas Cd, Cr, Cu, Fe and Ni, of which the ambientand soft tissue concentrations differ between bothestuaries and display upstream increasing soft tis-sue concentration gradients along the WesternScheldt(De Wolf et al., 2000), are known to bemutagenic(Forni, 1994; Reid et al., 1994). Theymay also affect the physiology(e.g. Farag et al.,1994; Nicholson, 2001) andyor the resources ofthe animals(e.g. Clements et al., 2000; Beasleyand Kneale, 2002). Likewise, salinity which also

differs between both estuaries and which displaysa downstream increasing gradient along the West-ern Scheldt, is known to affect the physiologyandyor ecological resources of estuarine organ-isms, while it may interact with the heavy metaluptake as well(e.g. Blust et al., 1992; Bjerregaardand Depledge, 1994; Jackson et al., 2000). At highsalinities, Cd will, for instance, be complexed withchlorides making it less available for uptake(Blustet al., 1992; Jackson et al., 2000). In addition, theuptake of the remaining uncomplexed bioavailableCd fraction will be hampered at higher salinities,2q

due to a more intense competition with Ca for2q

uptake via the Ca channels(Bjerregaard and2q

Depledge, 1994). Given the higher ambient heavymetal levels and(for many trace metals) increaseduptake in the lower saline Western Scheldt sam-pling points, we expect more metal mutagenicityand adverse physiological andyor ecologicaleffects in the Western Scheldt, especially atupstream located sites.

Increased mutagenic activity could explain thenear absence of OPH10-2 amplification in theupstream parts of the Western Scheldt, if thisactivity is directed towards the OPH10-2 primingsites andyor interpriming site distance. However,chromosomal andyor DNA damage research isneeded to investigate this hypothesis in the future.Alternatively, selection and or bottlenecks, eitherrelated to the mutagenic, physiological andyorecological effects might be implicated.

A bottleneck is expected to affect most loci,reducing their heterozygosity and polymorphismlevels(e.g. Gillespie and Guttman, 1998; Belfioreand Anderson, 2001; Theodorakis et al., 2001),while selection is expected to affect particular locionly (e.g. Gillespie and Guttman, 1998; Belfioreand Anderson, 2001), not necessarily reducingheterozygosity(i.e. heterosis). Examples of bothselection and bottleneck-related effects can befound in RAPD-related toxicological literature(e.g. Theodorakis and Shugart, 1997; Ma et al.,2000). Based on our results it is, however, difficultto differentiate between both mechanisms. Whenwe compare the overall mean heterozygosity levelsfrom the Western and Eastern Scheldt(respectively0.21 and 0.23) with previously published allozymebased heterozygosity levels forL. littorea (0.07)


(Foltz et al., 1993), (0.04) (Fevolden and Garner,1987; Janson, 1987), it is clear that they areelevated and of the same magnitude at both estu-aries. Given these high heterozygosity values andconsidering the gene flow levels(Nm)1), it israther unlikely that the genetic patterning resultedfrom a genetic bottleneck. Nevertheless, the currenthigh heterozygosity levels cannot completelyrefute the bottleneck hypothesis, as heterozygositylevels obtained via allozyme analysis have, insome cases, been reported to underestimate thegenetic variability, when compared to DNA mark-ers(Apostol et al., 1996).

Evidence for selection-related effects might, onthe other hand, be found when considering theallelicygenotypic distribution at locus OPH10-2(Fig. 2). Indeed, OPH10-2 was expressed in 28%of the individuals from the Eastern Scheldt, where-as it could only be detected in one individual fromthe Western Scheldt when Westkapelle was omittedfrom the analysis. Consequently, all but one indi-vidual from the three upstream-located WesternScheldt sampling sites shared the same homozy-gotic recessive genotype at OPH10-2. The factthat the recessive allele in OPH10-2 is almostcompletely fixed in the upstream-located WesternScheldt sampling sites, while both the dominantas well as the recessive allele are frequently foundin the downstream part of the Western and in theEastern Scheldt, suggests a selection-related effectthrough hitchhiking(Yan et al., 1998; Belfiore andAnderson, 2001).

5. Conclusion

In conclusion, one can say that the geneticpatterning obtained with the protein EST markeris confirmed in the current study, using DNAmarkers. Hence, the EST patterns did not resultfrom an ecophysiological-induced phenotypicexpression, but truly reflected the periwinkles’genome. These genetic patterns may result fromthe direct mutagenic activity at metal pollutedsites, but may also result from population-mediatedprocesses, induced by salinityyheavy metal pollu-tion-related mutagenic, physiological andyor eco-logical adverse effects. The fact that(1) geneticpatterns persist in time, despite intense gene flow;

(2) heterozygosity levels are elevated and compa-rable between polluted and pristine sites; and(3)alleles and genotypes at OPH10-2 are not random-ly distributed along the estuary, suggest that selec-tion, rather than a bottleneck, is responsible forthe genetic patterning.

In order to test this hypothesis, laboratory exper-iments will have to be set up, determining geno-types, DNA damage, chromosomal damage,physiological and fitness parameters in periwinklesthat are challenged to a wide salinity andyor heavymetal range. In addition, all sampling sites willhave to be ecologically characterised in order tolook for possible environmentally affected ecolog-ical differences among the sites, which could affectthe life history of the periwinkles’ as well.

Acknowledgments

This research was supported by the Fifth Frame-work Programme of the European Commissionunder contract number EVK3-CT-2001-00048, byan F.W.O. ‘Krediet aan navorsers’ and RAFO1DEWOH KP02 grant. HDW is a PostdoctoralFellow of the Fund for Scientific Research-Flan-ders(Belgium) (F.W.O.).

References

Apostol BL, Black WC, Reiter P, Miller BR. Populationgenetics with RAPD-PCR markers: the breeding structureof Aedes aegypti in Puerto Rico. Heredity 1996;76:325–334.

Bayens W. Evolution of trace metal concentrations in theScheldt estuary(1978–1995). A comparison with estuarineand ocean levels. Hydrobiologia 1998;366:157–167.

Beasley G, Kneale P. Reviewing the impact of metals andPAH’s on macro invertebrates in urban watercourses. ProgPhys Geogr 2002;26(2):236–270.

Becerril C, Ferrero M, Sanz F, Castano A. Detection of˜mitomycin C-induced genetic damage in fish cells by useof RAPD. Mutagenesis 1999;14:449–456.

Becerril C, Acevedo H, Ferrero M, Sanz F, Castano A. DNA˜fingerprint comparison of rainbow trout and RTG-2 cell lineusing random amplified polymorphic DNA. Ecotoxicology2001;10:115–124.

Belfiore NM, Anderson SL. Effects of contaminants on geneticpatterns in aquatic organisms: a review. Mutat Res2001;489:97–122.

Berger EM. Gene–enzyme variation in three sympatric speciesof Littorina. II The Roscoff population, with a note on the


origin of North AmericanL. littorea. Biol Bull Mar BiolLab Woods Hole 1973;153:255–264.

Bickham JW, Sandhu S, Hebert PDN, Chikhi L, Athwal R.Effects of chemical contaminants on genetic diversity innatural populations: implications for biomonitoring andecotoxicology. Mutat Res 2000;463:33–51.

Bjerregaard P, Depledge MH. Cadmium accumulation inLit-torina littorea, Mytilus edulis and Cracinus maenas: theinfluence of salinity and calcium ion concentrations. MarBiol 1994;119:385–395.

Black B. RAPDLD 2.0—A FORTRAN program to computepair-wise linkage disequilibrium coefficients among allelesat RAPD-PCR loci. Colorado State University, 1996.

Black B. RAPDFST 4.0.1—A FORTRAN program to estimateF(ST) and effective migration rates among subpopulationsusing RAPD-PCR files. Colorado State University, 1997.

Blust R, Kockelbergh E, Baillieul M. Effect of salinity on theuptake of cadmium by the brine shrimpArtemia franciscana.Mar Ecol Prog Ser 1992;84:245–254.

Clements WH, Carlisle DM, Lazochak JM, Johnson PC. Heavymetals structure benthic communities in Colorado mountainstreams. Ecol Appl 2000;10(2):626–638.

Depledge MH. Genetic ecotoxicology: an overview. J ExpMar Biol Ecol 1996;200:57–66.

Depledge MH, Billinghurst Z. Ecological significance of endo-crine disruption in marine invertebrates. Mar Pollut Bull1999;39:32–38.

De Wolf H, Backeljau T, Verhagen R. Congruence betweenallozyme and RAPD data in assessing macrogeographicalgenetic variation in the periwinkleLittorina striata (Mol-lusca, Gastropoda). Heredity 1998;81:486–492.

De Wolf H, Backeljau T, Blust R. Heavy metal accumulationin the periwinkleLittorina littorea, along a pollution gradi-ent in the Scheldt estuary. Sci Total Environ 2000;262:111–121.

De Wolf H, Jordaens K, Reusens E, Blust R, Backeljau T.Esterase variation in the periwinkleLittorina littorea, alongthe Western and Eastern Scheldt estuarium. Mar EnvironRes 2001a;52:373–382.

De Wolf H, Blust R, Backeljau T. Shell size variation inLittorina littorea in the Western Scheldt estuary. J ShellfishRes 2001b;20(1):427–430.

De Wolf H, De Coen W, Backeljau T, Blust R. Intersex andsterility in the periwinkleLittorina littorea (Mollusca: Gas-tropoda) along the Western Scheldt estuary, The Nether-lands. Mar Environ Res 2001c;52:249–255.

Edwards CA. Assessing the effects of environmental pollutantson soil organisms, communities, processes and ecosystems.Eur J Soil Biol 2002;38:225–231.

Farag AM, Boese CJ, Woodward DF, Bergman HL. Physiolog-ical-changes and tissue metal accumulation in rainbow-troutexposed to foodborne and waterborne metals. Environ Tox-icol Chem 1994;13(12):2021–2029.

Fevolden SE, Garner SP. Environmental stress and allozymevariation in Littorina littorea (Prosobranchia). Mar EcolProg Ser 1987;39:129–136.

Foltz DW, Shumway SE, Crisp D. Genetic structure andheterozygosity-related fitness effects in the marine snailLittorina littorea. Am Malacol Bull 1993;10(1):55–60.

Forni A. Comparison of chromosome aberrations and micro-nuclei in testing genotoxicity in humans. Toxicol Lett1994;72:185–190.

Gerringa LJA, Poortvliet TCW, Hummel H. Comparison ofchemical speciation of copper in the Oosterschelde andWesterschelde estuaries, The Netherlands. Estuar CoastShelf Sci 1996;42:629–643.

Gillespie RB, Guttman SI. Chemical-induced changes in thegenetic structure of populations: effects on allozymes. In:Forbes VE, editor. Genetics and ecotoxicology. Philadelphia:Taylor and Francis, 1998. p. 55–77.

Guttman SI. Population genetic structure and ecotoxicology.Environ Health Perspect 1994;102(12):97–100.

Hebert PDN, Luiker MM. Genetic effects of contaminantexposure—towards an assessment of impacts on animalpopulations. Sci Total Environ 1996;191:23–58.

Jackson BP, Lasier PJ, Miller WP, Winger PW. Effects ofcalcium, magnesium and sodium on ellevating cadmiumtoxicity to Hyalella azteca. Bull Environ Contam Toxicol2000;64:279–286.

Janson K. Allozyme and shell variation in two marine snails(Littorina, Prosobranchia) with different dispersal abilities.Biol J Linn Soc 1987;30:245–256.

Johannesson K. Genetic variability and large scale differenti-ation in two species of Littorinid gastropods with plankto-trophic development,Littorina littorea (L.) andMelarhaphe(Littorina) neritoides (L.) (Prosobranchia: Littorinacea),with notes on a mass occurrence ofM. neritoides in Sweden.Biol J Linn Soc 1992;47:285–299.

Lande R. Anthropogenic, ecological and genetic factors inextinction and conservation. Res Popul Ecol1998;40(3):259–269.

Lynch M, Milligan BG. Analysis of population genetic struc-ture with RAPD markers. Mol Ecol 1994;3:91–99.

Ma XL, Cowles DL, Carter RL. Effect of pollution on geneticdiversity in the bay musselMytilus galloprovincialis andthe acorn barnacleBalanus glandula. Mar Environ Res2000;50:559–563.

Morton RA. Evolution of Drosophila insecticide resistance.Genome 1993;36:1–7.

Nicholson S. Ecocytological and toxicological responses tocopper inPerna viridis (L.) (Bivalvia: Mytilidae) haemo-cyte lysosomal membranes. Chemosphere 2001;45:399–407.

Nony PA, Schnellmann RG. Interactions between collagen IVand collagen-binding integrins in renal cell repair aftersublethal injury. Mol Pharmacol 2001;60(6):1226–1234.

Pimm SL, Russel GL, Gittleman JL, Brooks TM. The futureof biodiversity. Science 1995;269:347–350.

Reid TM, Dunn JF, Loeb LA. Mutagenesis by metal-inducedoxygen radicals. Environ Health Perspect 1994;102:57–61.

Staton JL, Schizas NV, Chandler GT, Coull BC, Quattro JM.Ecotoxicology and population genetics: the emergence of


‘phylogeographic and evolutionary ecotoxicology’. Ecotox-icology 2001;10:217–222.

Statsoft. STATISTICA for Windows(computer manual). Tul-sa, Oklahoma: Statsoft Inc, 1998.

Swofford DL, Selander RB. BIOSYS. A computer programfor the analysis of allelic variation in population geneticsand biochemical systematics. Release 1.7. Urbana: Univer-sity of Illinois, 1989.

Theodorakis CW, Shugart LR. Genetic ecotoxicology: II.Population genetic structure in mosquitofish exposed in situto radionuclides. Ecotoxicology 1997;6:335–354.

Theodorakis CW, Bickham JW, Elbl T, Shugart LR, ChesserRK. Genetics of radionuclide-contaminated mosquitofishpopulations and homology betweenGambusia affinis andG. holbrooki. Environ Toxicol Chem 1998;17(10):1992–1998.

Theodorakis CW, Bickham JW, Lamb T. Integration of geno-toxicity and population genetic analyses in kangaroo rats(Dipodomys merriami) exposed to radionuclide contamina-

tion at the Nevada test site, USA. Environ Toxicol Chem2001;20(2):317–326.

Troncoso L, Galleguillos R, Larrain A. Effects of copper onthe fitness of the Chilean scallopArgopecten purpratus(Mollusca: Bivalia). Hydrobiologia 2000;420:185–189.

Wilkins NP, O’Reagan D. Genetic variation in sympatricspecies ofLittorina. Veliger 1980;22:355–359.

Winnepenninckx B, Backeljau T, De Wachter R. Investigationof molluscan phylogeny on the basis of 18S rRNA sequenc-es. Mol Biol Evol 1996;13:1306–1313.

Wright S. The interpretation of population structure byF-statistics with special regard to systems of mating. Evolution1965;19:395–420.

Yan GY, Chadee DD, Severson DW. Evidence for genetichitchhiking effect associated with insecticide resistance inAedes aegypti. Genetics 1998;148(2):793–800.

Yeh FC, Boyle T. Popgene v1.32 Microsoft Windows basedsoftware for population genetic analysis. Edmonton: Uni-versity of Alberta, 1996.

Documents

The population genetic structure of Littorina littorea (Mollusca: Gastropoda) along a pollution gradient in the Scheldt estuary (The Netherlands) using RAPD analysis