MARINE ECOLOGY PROGRESS SERIESMar Ecol Prog Ser

Vol. 532: 112, 2015doi: 10.3354/meps11362

Published July 21

INTRODUCTION

The life history of many marine organisms includesa planktonic dispersal stage that theoretically allowseven species whose adults are sessile or sedentary tomaintain high levels of connectivity over vast dis-tances (Siegel et al. 2003). Understanding the effectsof ocean currents and other environmental factors onconnectivity among populations of such organismshas important implications for improving our knowl-edge about marine ecology, including the manage-ment of exploited species and the design of marinereserves (Cowen et al. 2000). Numerous genetic

Inter-Research 2015 www.int-res.com*Corresponding author: luciano.beheregaray@flinders.edu.au

FEATURE ARTICLE

On-shelf larval retention limits population connectivity in a coastal broadcast spawner

Peter R. Teske1,2, Jonathan Sandoval-Castillo1, Erik van Sebille3,4, Jonathan Waters5,Luciano B. Beheregaray1,*

1Molecular Ecology Lab, School of Biological Sciences, Flinders University, Adelaide, South Australia 5001, Australia2Molecular Zoology Lab, Department of Zoology, University of Johannesburg, Auckland Park 2006, South Africa

3Grantham Institute & Department of Physics, Imperial College London, London SW7 2AZ, UK4ARC Centre of Excellence for Climate System Science, University of New South Wales, Sydney, NSW 2052, Australia

5Allan Wilson Centre for Molecular Ecology and Evolution, Department of Zoology, University of Otago, Dunedin 9054, New Zealand

ABSTRACT: Broadcast-spawning marine organismswith long pelagic larval duration are often expectedto be genetically homogeneous throughout theirranges. When genetic structure is found in such taxa,it may be in the form of chaotic genetic patchiness:i.e. patterns that might seem independent of any un-derlying environmental variation. The joint analysisof population genetic data and marine environmentaldata can elucidate factors driving such spatial geneticdiversity patterns. Using meso-scale sampling (at ascale of 10s to 100s of km), microsatellite data and advection connectivity simulations, we studied the effect of temperate southern Australian ocean circu -lation on the genetic structure of the snail Nerita atramentosa. This species has a long pelagic larvalduration and is represented as a single metapopula-tion throughout its ~3000 km range, but even so, wefound that its dispersal potential is lower than ex-pected. Connectivity simulations indicate that this is aresult of the larvae that remain on the continental shelf(where currents are erratic and often shoreward) re-turning to the coast in much larger numbers than lar-vae that become entrained in the regions shelf-edgeboundary currents. Our study contributes to thegrowing evidence that departures from the expecta-tions of panmixia along continuous and environ -mentally homogeneous coastlines are not limited tolow-dispersal species, and it identifies on-shelf larvalretention as an important factor limiting dispersal.

KEY WORDS: Isolation by distance IBD Marineecology Marine protected areas MPAs Planktoniclarval duration Population genetic structure Seascape genetics

Resale or republication not permitted without written consent of the publisher

A section of the coast of Australia where a seascape geneticanalysis of the snail Nerita atramentosa revealed parti -cularly strong on-shelf larval retention (red: boundary cur-rents; black: on-shelf currents).

Image: L. Beheregaray, E. van Sebille, P. Teske

FREEREE ACCESSCCESS

Mar Ecol Prog Ser 532: 112, 20152

studies have confirmed that in the absence of disper-sal barriers, populations of planktonic dispersersoften show low levels of genetic differentiation com-pared to species with lower dispersal potential, suchas direct developers (e.g. Teske et al. 2007). How-ever, high-dispersal species are often neither pan-mictic (i.e. freely interbreeding) throughout theirranges, nor is the amount of genetic structure in -versely correlated with the time that larvae spendin the plankton (Palumbi 2004, Weersing & Toonen2009). Instead, genetic struc ture often shows a pat-tern of seemingly chaotic genetic patchiness (John-son & Black 1984). This was traditionally attributed tovariability in recruitment success (Hedgecock 1994),but a number of recent studies have shown that suchcomplex genetic patterns could be explained by thestructuring effects of environmental features such ascoastal topography (Banks et al. 2007, Nicastro et al.2008), ocean cur rents (Banks et al. 2007, Piggott etal. 2008, White et al. 2010) and the size of suitablehabitats (Selkoe et al. 2010).

To understand population connectivity in broadcastspawners, one needs to take into consideration thatthe interactions between populations may be subjectto a complex interplay of environmental and biologicalfactors. Detailed assessments of the mechanisms in-volved in limiting larval dispersal have only recentlybecome possible by jointly analysing population ge-netic and oceanographic data, an approach known asseascape genetics. In the case of high-dispersal spe-cies with inherently high rates of migration, the ge-netic data on their own tend to lack sufficient geneticsignal to calculate dispersal rates and directions, butthey can be used to determine which of a number ofoceanographic dispersal scenarios is most appropriatefor the species under investigation (Selkoe et al. 2010).

Here, we present an in-depth seascape geneticanalysis of the temperate Australian coastal snailNerita atramentosa. This species planktonic larvaesettle after several months, which should theoreti-cally allow it to maintain high levels of connectivitythroughout its range. Using a combination of meso-scale sampling, population genetic data and advec-tion connectivity simulations, we tested whethershallow genetic structure among sites could beexplained by the regions oceanography. The findingthat dispersal is limited because it is facilitated pri-marily by weak on-shelf currents rather than offshoreboundary currents highlights the value of the sea-scape genetic ap proach in uncovering biologicallymeaningful patterns in species in which geneticstructure, if present at all, was traditionally believedto be the result of random dispersal processes.

MATERIALS AND METHODS

Southern Australian ocean currents

The boundary currents of temperate southern Aus-tralia are unusual in 2 respects. First, the southernwest coast is dominated by the Leeuwin Current (LC),which, unlike many cold eastern boundary currentselsewhere, is of tropical origin (Godfrey & Ridgway1985) and flows in a poleward rather than equator-ward direction (Fig. 1). Second, southern Australiahas perhaps the longest zonal coastal boundary in theworld (Ridgway & Condie 2004). As the LC passesCape Leeuwin (CL) in south-western Australia, itturns eastwards towards the Great Australian Bight(GAB) (Cresswell & Golding 1980). From the easternGAB, zonal current flow continues in what Ridgway& Condie (2004) consider to be a separate current, theSouth Australian Current (SAC), until the currentflow becomes poleward again as the Zeehan Current(ZC) flows along the western Tasmanian coast. Duringthe winter months, there is continuous warm west-to-east current flow (Ridgway & Condie 2004) that po-tentially connects the fauna of the entire region.In summer, there is an overall weakening of currentsand a reversal of boundary flow along the easternsouth coast (Vaux & Olsen 1961). The GAB has a shal-low shelf region up to 100 km wide. The SAC flowsalong the continental slope, and at places is thus faraway from the coastline.

Study species

The intertidal snail Nerita atramentosa is a par -ticularly suitable study organism for in vestigatingthe effects of southern Australias oceanography oncoastal biota because it is represented on rockyshores throughout the temperate southern Aus tralianregion (Spencer et al. 2007). In south-eastern Aus-tralia, N. atramentosa occurs only sporadically be yondWilsons Promontory (WP in Fig. 1) and on the Tas-manian east coast, where it is replaced by its sisterspecies N. melanotragus, whose range is stronglylinked to the region dominated by the East AustralianCurrent (EAC) (Waters et al. 2005, 2014). Australianneritid snails have long spawning seasons of up to9 mo, with peak spawning occurring throughout aus-tral summer (Przeslawski 2008), and long larval dura-tions of around 4 mo (Underwood 1975). Large geo -graphic distances between suitable habitats, a featurethat can affect genetic structure in low-dispersal spe-cies in the absence of any other explanations (e.g.

Teske et al.: Connectivity in a broadcast spawning snail

Hellberg 1996), are thus un likely to impact on N. atra -mentosa. Also, unlike many other species that arerepresented in this region by highly divergent evolu-tionary lineages (e.g. Waters et al. 2004, York et al.2008, Li et al. 2013), reflecting the effects of histori-cal barriers on contemporary genetic structure, thereis no evidence for any division other than the rela-tively recent evolutionary divergence be tween N.atramentosa and its sister species N. mela no tragus(Waters et al. 2005). This suggests that any fine-scalegenetic structure identified in this species is morelikely to be the result of relatively recent eventsrather than historical oceanographic conditions.

Genetic data generation

During 2011 and 2012, tissue samples were excisedfrom the foot of 870 snails from 21 localities (Fig. 1,Table 1) and preserved in 99% ethanol. GenomicDNA was extracted using a salting out protocol (Sun-nucks & Hales 1996). Thirteen microsatellite loci(Neat01, Neat02, Neat03, Neat04, Neat05, Neat07,Neat09, Neat10, Neat12, Neat14, Neat16, Neat18,Neat19) were genotyped for all samples as describedin Sandoval-Castillo et al. (2012), but 3 of these lociwere excluded from the analyses because of high pro-

3

25S

30

35

100

90

80

70

60

50

40

30

20

10

0

40

45115E 120 125 130 135 140 145 150 155

1 23

4

56

71011

12 131415

17 16

202119

18

8 9

LCLC

GABGAB

SACSAC

ZCZC

WPWP EACEAC

LC

GAB

SAC

ZC

WP EAC

SG

GSVCL

Fig. 1. Sampling localities (numbered 121, see Table 1) of Nerita atramentosa along the southern coast of Australia. Arrowsand colours indicate the direction and magnitude (cm s1), respectively, of surface velocity in the Ocean General CirculationModel For the Earth Simulator (OFES) from 1 September to 31 January averaged over the years 1980 to 2010. The black circlebetween Sites 3 and 4 represents an inaccessible site that was not sampled, but for which oceanographic connectivity was simulated (see also Fig. 3 and Fig. S4 in Supplement 3 at www.int-res. com/ articles/ suppl/ m532 p001_ supp. pdf). CL: CapeLeeuwin; EAC: East Australian Current; GAB: Great Australian Bight; GSV: Gulf St. Vincent; LC: Leeuwin Current; SAC:

South Australian Current; SG: Spencer Gulf; WP: Wilsons Promontory; ZC: Zeehan Current

Site Site name N NA PA Ho Heno.

1 Walpole 35 10.9 1 0.77 0.772 Albany 30 10.5 0 0.79 0.803 Esperance 50 11.4 0 0.78 0.784 Penong 35 10.3 0 0.80 0.775 Point Drummond 37 11.7 3 0.81 0.796 Fishery Bay 41 10.5 0 0.81 0.777 Peak Bay 35 10.6 1 0.78 0.788 Point Souttar 41 10.7 1 0.79 0.799 Edithburgh 45 11.7 4 0.82 0.7810 Glenelg 44 11.6 0 0.84 0.7911 Victor Harbor 48 11.3 0 0.80 0.7812 Port Fairy 47 11.3 0 0.77 0.7713 Marengo 48 11.1 0 0.77 0.7714 Walkerville 48 11.9 0 0.74 0.7815 Port Albert 47 11.8 0 0.74 0.7716 Bridport 41 10.9 2 0.78 0.7917 Penguin 40 11.2 2 0.83 0.7918 Couta Rocks 38 11.2 1 0.74 0.8019 Trial Harbour 40 11.0 0 0.80 0.7920 Pirates Bay 42 12.2 3 0.79 0.8121 Swansea 40 10.8 2 0.79 0.79

Table 1. Genetic diversity parameters of Nerita atramentosafrom 21 sampling localities along the southern coast of Aus-tralia. N = sample size; NA = average number of alleles perlocus; PA = private alleles; Ho = observed heterozygosity;

He = expected heterozygosity

http://www.int-res.com/articles/suppl/m532p001_supp.pdf

Mar Ecol Prog Ser 532: 112, 20154

portions of non-amplification (Neat02) or departuresfrom Hardy-Weinberg equilibrium (Neat16 andNeat18, see Table S1 in Supplement 2 at www.int-res.com/ articles/ suppl/ m532 p001_ supp. pdf). Data weregenerated for between 30 and 50 snails per locality(average of 41.5 across all localities).

Genetic diversity and differentiation

Observed and expected heterozygosity, and totalnumber of alleles were calculated per microsatellitelocus and location using ARLEQUIN 3.5 (Ex coffier &Lischer 2010). We also cal culated the number of pri-vate alleles using GENALEX 6.5 (Peakall & Smouse2012). Pairwise values of FST (Wright 1965) and JostsDest (Jost 2008) were estimated between localities,and their significance at = 0.05 was as sessed byrunning 999 permutations in GENALEX. To accountfor multiple comparisons, the B-Y false discovery ratemethod (Benjamini & Yekutieli 2001) was applied.Unlike Bonferroni correction (Rice 1989), this ap -proach not only reduces Type I error but also Type IIerror, and is considered to be more suitable to providecritical values in biological systems (Narum 2006).To determine whether Nerita atramentosa com prisesmultiple regional populations, we used the Bayesiangenetic clustering algorithm implemented in STRUC -TURE 2.3 (Prit chard et al. 2000). We assumed admix-ture and that allele frequencies are correlated be -tween populations, and used sampling locations aspriors, as this can considerably improve the likeli-hood that genetic structure is identified when levelsof genetic divergence are low (Hubisz et al. 2009).For each value of K (number of clusters, ranging from1 to 21) we performed 10 independent runs, eachwith an initial burn-in of 100 000 steps followed by1000000 Markov chain Monte Carlo iterations. Theresults of the 10 replicates were clustered usingCLUMPP 1.1.2 (Jakobsson & Rosenberg 2007), andthe most likely value of K was identified on the basisof having the highest probability (Pritchard et al.2000).

Spatial autocorrelation

Under dispersal models in which individuals sepa-rated by short geographic distances are more closelyrelated to each other than they are to individuals atgreater geographic distance, positive spatial autocor-relation should be evident at shorter distances(Peakall et al. 2003). We performed spatial autocor -

relation analyses in GENA LEX using genetic andgeographic distance matrices. The detection of a pos-itive spatial autocorrelation coefficient r (Smouse &Peakall 1999) that is significantly greater than ex -pected under conditions of panmixia at the smallestdistance classes is indicative of restricted dispersal(see Supplement 1 at www.int-res. com/ articles/ suppl/m532 p001_ supp. pdf for details).

Advection connectivity simulations

To estimate pairwise advection connectivity matri-ces between sites, we used the Connectivity Model-ling System 1.1 (Paris et al. 2013) to integrate virtualLagrangian particles within the Ocean General Cir-culation Model For the Earth Simulator (OFES;Masumoto et al. 2004) (see Supplement 1 for details).To visualise how the geographic position of larvaeaffected their arrival at the coast on a monthly basis,we generated an animation showing the movementof propagules throughout their larval periods. Wegenerated 3 matrices: the first depicted pairwise con-nectivity among sites; the second depicted the per-centage of propagules that remained on the conti-nental shelf throughout their larval phases (i.e. thatnever reached water with depths of >100 m); and thethird showed how many larvae released from aparticu lar site arrived at the coast after completinglarval development. Correlations between the latter2 data sets were compared on a monthly basis bymeans of Spearman rank correlation tests (Spearman1904) performed in SigmaStat 1.0 (Systat Software).To understand the role of ocean circulation in theinaccessible GAB, the simulations also included asite from which no samples could be collected (blackcircle in Fig. 1). The South Australian gulfs wereeach treated as a single location by merging sam-pling sites (SG: Sites 7 and 8; GSV: Sites 9 and 10).

Seascape genetics: genetic vs. environmental data

We determined the relative importance of 3 types ofenvironmental parameters on genetic differen tiationamong sites: geographic distance, thermal gradientsin sea surface temperature and ocean circulation.Mantel tests (Mantel 1967) and Multiple Regressionon Distance Matrices (MRDM; Manly 1986, Legendreet al. 1994) in FSTAT 2.9.3.2 (Goudet 1995) were usedto test for correlations between the environmentaldata and the genetic data. Mantel tests explorewhether there is a correlation between a genetic dis-

http://www.int-res.com/articles/suppl/m532p001_supp.pdfhttp://www.int-res.com/articles/suppl/m532p001_supp.pdfhttp://www.int-res.com/articles/suppl/m532p001_supp.pdfhttp://www.int-res.com/articles/suppl/m532p001_supp.pdf

Teske et al.: Connectivity in a broadcast spawning snail 5

tance matrix (in this case using pairwise FST values asa measure of genetic differentiation) and pairwisematrices of other parameters that may potentially ex-plain genetic patterns. MRDM is a multivariate statis-tical extension of partial Mantel analyses (Smouse etal. 1986) that uses multiple regression to test for thecorrelation between a response variable (in this caseFST) and 2 or more explanatory variables. This makesit possible to simultaneously determine statistical sig-nificance and to make inferences about the relativeimportance of each explanatory variable (Lichstein2007). Oceanographic connectivity data simulatedfrom 1 September to 30 January (austral summer; seeSupplement 1) using data from the years 1980 to 2010were transformed by taking the negative of the natu-ral logarithm of the sum of migrants (i.e. immigrantsand emigrants) for each pair of sites. As we generatedmultiple data sets for temperature and oceanographicconnectivity (see previous paragraph), only thosedata sets were in cluded in the MRDM analyses thatexplained most of the genetic variation in the Manteltests. We also analysed correlations between FST andthe explanatory variables for individual months, andperformed Mantel tests between matrices of ex -planatory variables. Significanceof all tests was based on 10 000permutations.

RESULTS

Genetic diversity and differentiation

All microsatellites were poly-morphic at all sampled localities.The number of alleles per locuswas similar between sites, rang-ing from 10.3 at Site 4 to 12.2 atSite 20. Observed and ex pectedheterozygosity were high at allsites, with an average of 0.786and 0.774, respectively. While thenumber of private alleles (allelespresent at a single sampling site)was low, such alleles were foundat most of the sites (Table 1).

Overall genetic differentiationwas low but significant across thespecies range (FST = 0.006, p 100 m).The black circle represents an inaccessible site from which

no genetic samples were available (see Fig. 1)

Mar Ecol Prog Ser 532: 112, 2015

likely affects a greater proportion of planktonicpropagules than do the boundary currents.

The low velocity of the southern Australian boundarycurrents compared to the EAC was interpreted asbeing the cause for lower population connectivity inthis region, promoting a pattern of IBD as comparedto a pattern of chaotic genetic patchiness on the eastcoast (Coleman et al. 2011, 2013). However, in addi-tion to being weaker than the EAC, southern Aus-tralias boundary currents flow at a considerablygreater distance from the coast because in most ofthis region, the continental shelf is considerablywider than the shelf on the east coast (Fig. S5 in Sup-plement 3) (Porter-Smith et al. 2004). This suggeststhat a comparatively smaller proportion of propag-ules will ever reach the boundary currents, and thatthe few that do and are dispersed over greater dis-tances may not reach the coast in time to completelarval development. We conclude that the larvae thatremained on the continental shelf contributed a con-siderably greater number of gametes to the next gen-eration, which strongly affected genetic estimates ofconnectivity.

Comparisons of genetic and environmental data

Seascape genetic approaches represent a way todetermine whether low but significant structureamong sites along a species range may be driven byenvironmental conditions, rather than merely beingthe result of stochastic processes. For Nerita atra-mentosa, we found that advection connectivity andgeographic distance were both strongly correlatedwith genetic structure (and with each other) whenanalysing the complete data set, with the MRDManalysis identifying geographic distance as the mostimportant explanatory variable. Monthly data seemcontradictory in that they often identified advectionconnectivity as being more important than geneticdistance. It is possible that our models describedgene flow poorly when it was mostly facilitated byweak nearshore circulation (particularly during No -vember, the likely peak spawning month), resultingin less support for advection connectivity during thistime and, by extension, the whole data set. We useda global scale model that accurately represents theocean circulation and hence movement of propag-ules in the open ocean and on the continental slope,but performs more poorly when the propagules staywithin the 20 km closest to the coast. To simulate lar-val movement in this near-shore region and therebysomewhat mitigate the lack of resolution there, we

8

Test Corre- p Variance lation explained

Explanatory variable (%)

SeptemberJanuaryMantel test

Coastal Distance 0.24 0.001 5.9Advection connectivity (Model 1) 0.12 0.133 1.3Advection connectivity (Model 2) 0.22 0.003 4.7Summer temperature 0.09 0.245 0.8Delta temperature 0.12 0.106 1.5

MRDM 6.6Coastal distance 0.24 0.160Advection connectivity (Model 2) 0.02 0.722Delta temperature 0.08 0.360

SeptemberMantel test

Advection connectivity (Model 2) 0.23 0.002 5.4MRDM 6.9

Coastal distance 0.11 0.140Advection connectivity (Model 2) 0.23 0.004Delta temperature 0.06 0.395

OctoberMantel test

Advection connectivity (Model 2) 0.19 0.017 3.5MRDM 4.8

Coastal distance 0.09 0.173Advection connectivity (Model 2) 0.19 0.029Delta temperature 0.06 0.430

NovemberMantel test

Advection connectivity (Model 2) 0.16 0.068 3.9MRDM 3.9

Coastal distance 0.11 0.146 Advection connectivity (Model 2) 0.16 0.068Delta temperature 0.05 0.479

DecemberMantel test

Advection connectivity (Model 2) 0.22 0.070 4.8MRDM 5.9

Coastal distance 0.09 0.206Advection connectivity (Model 2) 0.22 0.012Delta temperature 0.06 0.476

JanuaryMantel test

Advection connectivity (Model 2) 0.26 0.0005 6.6MRDM 8.1

Coastal distance 0.01 0.155Advection connectivity (Model 2) 0.26 0.002Delta temperature 0.07 0.366

Table 2. Mantel tests and Multiple Regression on Distance Matri-ces (MRDM) performed on data matrices from Nerita atramen-tosa. For the complete data set, Mantel tests are reported for thecorrelation between genetic distance (FST) and 1 of 5 environ-mental variables, including geographic distance among sites(Coastal distance), the negative ln of advection connectivity(Models 1 and 2), and thermal distance (Summer temperatureand Delta temperature). For monthly data sets, only Advectionconnectivity (Model 2) are shown. MRDM analyses were per-formed using only the connectivity and thermal data sets thatexplained most of the genetic variation on the basis of the Manteltest for combined data. See Supplement 1 for details on model

parameters. Bold: p < 0.05

Teske et al.: Connectivity in a broadcast spawning snail 9

have used an additional sub-grid scale random-walkdiffusion for the particles (e.g. Wood et al. 2013).Given that accurate dynamics close to the shorelineare particularly relevant when studying the settle-ment of particles at the same site at which they werespawned (which was not done in the present study),the effect of limited resolution in this area is likelyminimal.

The assumption that N. atramentosa spawnsthrough out the austral summer is a likely simplifica-tion, as spawning in marine organisms typicallypeaks during some months and is much lower duringthe remainder of the spawning season (Zastrow et al.1991). No such data are available for N. atramentosa.However, the fact that recruitment success remainsfairly constant throughout the spawning seasonwhen comparing different sites, as well as the factthat spawning patterns may differ be tween years(Coombs et al. 2006), suggests that our assumptionthat spawning occurs throughout summer is reason-able. Our advection connectivity simulations furtherassumed that the larvae of N. atramentosa disperselike passive particles, which at large geographicalscales is often an acceptable assumption (McQuaid &Phillips 2000); nonetheless it is possible that incorpo-rating larval behaviour (such as diel vertical migra-tions, Barile et al. 1994) into the model would resultin even stronger correlations between oceanographicand genetic data. However, since so little is knownabout larval be haviour for Nerita spp., it is impossibleto incorporate any meaningful behaviour. Further-more, the most relevant behaviour (for connectivity)happens on small scales near shores (see e.g. Staater-man et al. 2012), while in this model we are dealingwith offshore, large-scale dispersion of larvae.

Water temperature is considered to be very impor-tant in shaping marine biogeography (Murawski1993) and, by extension, genetic structure (Briggs &Bowen 2012), but we found no clear evidence for ther-mal selection playing a role in driving genetic struc-ture along the range inhabited by N. atramentosa.Thermal gradients along the temperate southern Aus-tralian coast are minimal when compared to condi-tions in adjacent regions (Wernberg et al. 2013), andthe areas dominated by the LC, SAC and ZC are suffi-ciently connected and homogeneous to be considereda single biogeographic province (Waters et al. 2010).

Implications for MPA design

Marine protected areas are an important tool forlimiting the negative effects of anthropogenic

activities on the ecological functioning of coastalbiotas (Lubchenco et al. 2003, Edgar et al. 2014).Networks of marine reserves rather than large,single reserves are considered to be particularlyimportant for ensuring the long-term persistence ofmarine communities (Lubchenco et al. 2003). Eventhough IBD was found in N. atramentosa, this spe-cies is likely able to maintain a single metapopula-tion by means of meso-scale dispersal over severalspawning cycles. However, the finding that theregions boundary currents contribute compara-tively little to dispersing propagules, particularlyin South Australia, suggests that population per-sistence in species with smaller population sizesand lower dispersal potential could be jeo pardisedwhen reserves are spaced too far apart. The gov-ernment of South Australia is currently establishinga network of 19 marine parks, but only 6% ofcoastal South Australian waters have been pro-posed as sanctuaries (Government of South Aus-tralia 2012). Based on a recent meta-analysis ofkey features to be considered during marinereserve design (Edgar et al. 2014), it appears thatmany South Australian sanctuary zones are toosmall to achieve their desired conservation value.Nonetheless, given that dispersal distances ofcoastal species in this region can be ex pected tobe comparatively small, the closely spaced pro-posed sanctuary zones probably represent a rea-sonable starting point to maintain population con-nectivity.

CONCLUSIONS

Studies of genetic population structure in coastalorganisms often attribute genetic connectivity to dis-persal driven by boundary currents (Hoskin 2000,Neethling et al. 2008), with stronger boundary cur-rents having a greater potential of homogenisinggenetic structure than weaker currents (Coleman etal. 2011). Here, we show that genetic connectivity ina coastal broadcast disperser in much of southernAustralia is primarily influenced by on-shelf currentflow. A large proportion of planktonic propagules donot reach the regions boundary currents during theirspawning season, and many of those that becomeentrained in the boundary currents do not return tothe coast to settle. This demonstrates that a detailedunderstanding of current flows in the hydrodynami-cally complex coastal and shelf regions is needed toexplain the dispersal of planktonic propagules alongcontinuous coastlines.

Mar Ecol Prog Ser 532: 112, 2015

Data accessibility. Microsatellite data: Dryad, doi: 10. 5061/dryad. n9v91

Acknowledgements. This study was funded by the Aus-tralian Research Council (DP110101275 to L.B.B, J.W. andLuciana Mller and DE130101336 to E.v.S). The present arti-cle is publication no. 53 of the Molecular Ecology Group for Marine Research (MEGMAR).

LITERATURE CITED

Aiken CM, Navarrete SA, Castillo MI, Castilla JC (2007)Along-shore larval dispersal kernels in a numericalocean model of the central Chilean coast. Mar Ecol ProgSer 339: 1324

Banks SC, Piggott MP, Williamson JE, Bov U, Holbrook NJ,Beheregaray LB (2007) Oceanic variability and coastaltopography shape genetic structure in a long-dispersingsea urchin. Ecology 88: 30553064

Barile PJ, Stoner AW, Young CM (1994) Phototaxis and ver-tical migration of the queen conch (Strombus gigasLinne) veliger larvae. J Exp Mar Biol Ecol 183: 147162

Benjamini Y, Yekutieli D (2001) The control of false discov-ery rate under dependency. Ann Stat 29: 11651188

Briggs JC, Bowen BW (2012) A realignment of marine bio-geographic provinces with particular reference to fishdistributions. J Biogeogr 39: 1230

Cockroft AC, Wooldridge T (1987) Reproduction and larvaldistribution of the penaeid prawn Macropetasma afri -canus (Balss) in Algoa Bay. S Afr J Zool 22: 228234

Coleman MA, Roughan M, Macdonald HS, Connell SD,Gillanders BM, Kelaher BP, Steinberg PD (2011) Varia-tion in the strength of continental boundary currentsdetermines continent-wide connectivity in kelp. J Ecol99: 10261032

Coleman MA, Feng M, Roughan M, Cetina-Heredia P, Con-nell SD (2013) Temperate shelf water dispersal by Aus-tralian boundary currents: implications for populationconnectivity. Limnol Oceanogr 3: 295309

Coombs SH, Smyth TJ, Conway DVP, Halliday NC, BernalM, Stratoudakis Y, Alvarez P (2006) Spawning seasonand temperature relationships for sardine (Sardinapilchardus) in the eastern North Atlantic. J Mar BiolAssoc UK 86: 12451252

Cowen RK, Lwiza KMM, Sponaugle S, Paris CB, Olson DB(2000) Connectivity of marine populations: open orclosed? Science 287: 857859

Cresswell GR, Golding TJ (1980) Observations of a south-flowing current in the southeastern Indian Ocean. Deep-Sea Res A 27: 449466

Edgar GJ, Stuart-Smith RD, Willis TJ, Kininmonth S and others (2014) Global conservation outcomes depend onmarine protected areas with five key features. Nature506: 216220

Epperson BK (1995) Spatial distributions of genotypes underisolation by distance. Genetics 140: 14311440

Excoffier L, Lischer HE (2010) Arlequin suite ver 3.5: a newseries of programs to perform population genetics analy-ses under Linux and Windows. Mol Ecol Res 10:564567

Godfrey JS, Ridgway KR (1985) The large-scale environ-ment of the poleward-flowing Leeuwin Current, WesternAustralia: longshore steric height patterns, wind stressesand geostrophic flow. J Phys Oceanogr 15: 481495

Goudet J (1995) FSTAT (version 1.2): a computer program to

calculate F-statistics. J Hered 86:485486Government of South Australia (2012) South Australias

marine parks network, explanatory document. Depart-ment of Environment, Water and Natural Resources,Government of South Australia, Adelaide

Hedgecock D (1994) Temporal and spatial genetic structureof marine animal populations in the California Current.Calif Coop Ocean Fish Investig Rep 35: 7381

Hellberg ME (1996). Dependence of gene flow on geo-graphic distance in two solitary corals with different lar-val dispersal capabilities. Evolution 50: 11671175

Hohenlohe PA (2004) Limits to gene flow in marine ani-mals with planktonic larvae: models of Littorina speciesaround Point Conception, California. Biol J Linn Soc 82: 169187

Hoskin MG (2000) Effects of the East Australian Current onthe genetic structure of a direct developing muricid snail(Bedeva hanleyi, Angas): variability within and amonglocal populations. Biol J Linn Soc 69: 245262

Hubisz MJ, Falush D, Stephens M, Pritchard JK (2009) Infer-ring weak population structure with the assistance ofsample group information. Mol Ecol Res 9:13221332

Hutchings L, Beckley LE, Griffiths MH, Roberts MJ, SundbyS, van der Lingen C (2002) Spawning on the edge: spawning grounds and nursery areas around the south-ern African coastline. Mar Freshw Res 53: 307318

Jackson JM, Rainville L, Roberts MJ, McQuaid CD, Lutje-harms JRE (2012) Mesoscale bio-physical interactionsbetween the Agulhas Current and the Agulhas Bank,South Africa. Cont Shelf Res 49: 1024

Jakobsson M, Rosenberg NA (2007) CLUMPP: a clustermatching and permutation program for dealing withlabel switching and multimodality in analysis of popula-tion structure. Bioinformatics 23:18011806

Johnson MS, Black R (1984) Pattern beneath the chaos: theeffect of recruitment on genetic patchiness in an inter-tidal limpet. Evolution 38: 13711383

Jost L (2008) GST and its relatives do not measure differenti-ation. Mol Ecol 17: 40154026

Legendre P, Fortin MJ (2010) Comparison of the Mantel testand alternative approaches for detecting complex multi-variate relationships in the spatial analysis of geneticdata. Mol Ecol Resour 10: 831844

Legendre P, Lapointe FJ, Casgrain P (1994) Modeling brainevolution from behaviour: a permutational regressionapproach. Evolution 48: 14871499

Li J, Foighil D, Park JK (2013) Tritons trident: cryptic Neo-gene divergences in a marine clam (Lasaea australis)correspond to Australias 3 temperate biogeographicprovinces. Mol Ecol 22: 19331946

Lichstein JW (2007) Multiple regression on distance ma trices: a multivariate spatial analysis tool. Plant Ecol 188: 117131

Lubchenco J, Palumbi SR, Gaines SD, Andelman S (2003)Plugging a hole in the ocean: the emerging science ofmarine reserves. Ecol Appl 13: 37

Manly BF (1986) Randomization and regression methodsfor testing for associations with geographical, environ-mental and biological distances between populations.Res Pop Ecol 28: 201218

Mantel N (1967) The detection of disease clustering and ageneralized regression approach. Cancer Res 27: 209220

Masumoto Y, Sasaki H, Kagimoto T, Komori N and others(2004) A fifty-year eddy-resolving simulation of theworld oceanpreliminary outcomes of OFES (OGCMfor the Earth Simulator). J Earth Simul 1: 3556

10

http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&list_uids=6018555&dopt=Abstracthttp://dx.doi.org/10.1007/BF02515450http://dx.doi.org/10.1890/1051-0761(2003)013[0003%3APAHITO]2.0.CO%3B2http://dx.doi.org/10.1007/s11258-006-9126-3http://dx.doi.org/10.1111/mec.12220http://dx.doi.org/10.2307/2410243http://dx.doi.org/10.1111/j.1755-0998.2010.02866.xhttp://dx.doi.org/10.1111/j.1365-294X.2008.03887.xhttp://dx.doi.org/10.2307/2408642http://dx.doi.org/10.1016/j.csr.2012.09.005http://dx.doi.org/10.1071/MF01147http://dx.doi.org/10.1111/j.1095-8312.2000.tb01201.xhttp://dx.doi.org/10.1111/j.1095-8312.2004.00318.xhttp://dx.doi.org/10.1175/1520-0485(1985)015%3C0481%3ATLSEOT%3E2.0.CO%3B2http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&list_uids=7498782&dopt=Abstracthttp://dx.doi.org/10.1038/nature13022http://dx.doi.org/10.1016/0198-0149(80)90055-2http://dx.doi.org/10.1126/science.287.5454.857http://dx.doi.org/10.1017/S0025315406014251http://dx.doi.org/10.1215/21573689-2409306http://dx.doi.org/10.1111/j.1365-2745.2011.01822.xhttp://dx.doi.org/10.1111/j.1365-2699.2011.02613.xhttp://dx.doi.org/10.1214/aos/1013699998http://dx.doi.org/10.1016/0022-0981(94)90084-1http://dx.doi.org/10.1890/07-0091.1http://dx.doi.org/10.3354/meps339013

Teske et al.: Connectivity in a broadcast spawning snail

McQuaid CD, Phillips TE (2000) Limited wind-driven dis-persal of intertidal mussel larvae: in situ evidence fromthe plankton and the spread of the invasive speciesMytilus galloprovincialis in South Africa. Mar Ecol ProgSer 201: 211220

Mitarai S, Siegel DA, Watson JR, Dong C, McWilliams JC(2009) Quantifying connectivity in the coastal ocean withapplication to the Southern California Bight. J GeophysRes 114: C10026, doi: 10.1029/2008JC005166

Murawski SA (1993) Climate change and marine fish distri-butions: forecasting from historical analogy. Trans AmFish Soc 122:647658

Narum SR (2006) Beyond Bonferroni: less conservative analy-ses for conservation genetics. Conserv Genet 7: 783787

Neethling M, Matthee CA, Bowie RCK, von der Heyden S(2008) Evidence for panmixia despite barriers to geneflow in the southern African endemic, Caffrogobius caffer (Teleostei: Gobiidae). BMC Evol Biol 8: 325

Nicastro KR, Zardi GI, McQuaid CD, Teske PR, Barker NP(2008) Coastal topography drives genetic structure inmarine mussels. Mar Ecol Prog Ser 368: 189195

Palumbi SR (2004) Marine reserves and ocean neighbor-hoods: the spatial scale of marine populations and theirmanagement. Annu Rev Environ Resour 29: 3168

Paris CB, Helgers J, van Sebille E, Srinivasan A (2013) Con-nectivity modelling system: a probabilistic modelling toolfor the multi-scale tracking of biotic and abiotic variabil-ity in the ocean. Environ Model Softw 42: 4754

Peakall R, Smouse PE (2012) GenAlEx 6.5: genetic analysisin Excel. Population genetic software for teaching andresearch an update. Bioinformatics 28:25372539

Peakall R, Ruibal M, Lindenmayer DB (2003) Spatial auto-correlation analysis offers new insights into gene flow inthe Australian bush rat, Rattus fuscipes. Evolution 57:11821195

Piggott M, Banks S, Tung P, Beheregaray LB (2008) Geneticevidence for different scales of connectivity in a marinemollusc. Mar Ecol Prog Ser 365: 127136

Pogson GH, Taggart CT, Mesa KA, Boutilier RG (2001) Iso-lation by distance in the Atlantic cod, Gadus morhua, atlarge and small geographic scales. Evolution 55: 131146

Polato NR, Concepcion GT, Toonen RJ, Baums IB (2010) Iso-lation by distance across the Hawaiian Archipelago inthe reef-building coral Porites lobata. Mol Ecol 19:46614677

Porri F, Jackson JM, von der Meden CEO, Weidberg N,McQuaid CD (2014) The effect of mesoscale oceano-graphic features on the distribution of mussel larvaealong the south coast of South Africa. J Mar Syst 132: 162173

Porter-Smith R, Harris PT, Andersen OB, Coleman R,Greenslade D, Jenkins CJ (2004) Classification of theAustralian continental shelf based on predicted sedimentthreshold exceedance from tidal currents and swellwaves. Mar Geol 211: 120

Pritchard JK, Stephens M, Donnelly P (2000) Inference ofpopulation structure using multilocus genotype data.Genetics 155:945959

Przeslawski R (2008) Temporal patterns of gastropod eggmass deposition on southeastern Australian shores. MarFreshw Res 59: 457466

Rice WR (1989) Analyzing tables of statistical tests. Evolu-tion 43: 223225

Ridgway KR, Condie SA (2004) The 5500-km-long boundaryflow off western and southern Australia. J Geophys Res

109: C04017, doi:10.1029/2003JC001921Rothlisberg PC, Church JA, Fandry CB (1995) A mechanism

for near-shore concentration and estuarine recruitmentof post-larval Penaeus plebejus Hess (Decapoda, Penaei-dae). Estuar Coast Shelf Sci 40: 115138

Sandoval-Castillo J, Gardner MG, Beheregaray LB (2012)Isolation and characterization of microsatellite markersfor the marine black nerite Nerita atramentosa: tools forassessment and design of marine protected areas. Con-serv Genet Resour 4:625627

Selkoe KA, Watson JR, White C, Horin TB, Iacchei M,Mitarai S, Siegel DA (2010) Taking the chaos out ofgenetic patchiness: seascape genetics reveals ecologicaland oceanographic drivers of genetic patterns in threetemperate reef species. Mol Ecol 19: 37083726

Siegel DA, Kinlan BP, Gaylord B, Gaines SD (2003) Lag -rangian descriptions of marine larval dispersion. MarEcol Prog Ser 260: 8396

Smouse PE, Peakall R (1999) Spatial autocorrelation analysisof individual multiallele and multilocus genetic struc-ture. Heredity 82: 561573

Smouse PE, Long JC, Sokal RR (1986) Multiple regressionand correlation extensions of the Mantel test of matrixcorrespondence. Syst Zool 35: 627632

Spearman C (1904) The proof and measurement of associa-tion between two things. Am J Psychol 15: 72101

Spencer HG, Waters JM, Eichhorst TE (2007) Taxonomy andnomenclature of black nerites (Gastropoda: Neritimor-pha: Nerita) from the South Pacific. Invertebr Syst 21: 229237

Staaterman E, Paris CB, Helgers J (2012) Orientation behav-ior in fish larvae: a missing piece to Hjorts critical periodhypothesis. J Theor Biol 304: 188196

Sunnucks P, Hales DF (1996) Numerous transposedsequences of mitochondrial cytochrome oxidase I-II inaphids of the genus Sitobion (Hemiptera: Aphididae).Mol Biol Evol 13:510524

Teske PR, Papadopoulos I, Zardi GI, McQuaid CD, EdkinsMT, Griffiths CL, Barker NP (2007) Implications of lifehistory for genetic structure and migration rates of south-ern African coastal invertebrates: planktonic, abbrevi-ated and direct developers. Mar Biol 152: 697711

Teske PR, Zardi GI, McQuaid CD, Nicastro K (2013) Twosides of the same coin: extinctions and originationsacross the Atlantic/Indian Ocean boundary as conse-quences of the same climate oscillation. Front Biogeogr5: 4859

Underwood AJ (1975) Comparative studies on the biology ofNerita atramentosa (Reeve), Bembicium nanum (Lamarck)and Cellana tramoserica (Sowerby) (Gastropoda: Proso-branchia) in SE Australia. J Exp Mar Biol Ecol 18: 153172

Vaux D, Olsen AM (1961) Use of drift bottles in fisheriesresearch. Aust Fish News 20: 1720

Waters JM, OLoughlin PM, Roy MS (2004) Cladogenesis ina starfish species complex from southern Australia: evi-dence for vicariant speciation? Mol Phylogenet Evol 32:236245

Waters JM, King TM, OLoughlin PM, Spencer HG (2005)Phylogeographical disjunction in abundant high-disper-sal littoral gastropods. Mol Ecol 14: 27892802

Waters JM, Wernberg T, Connell SD, Thomsen MS and oth-ers (2010) Australias marine biogeography revisited: back to the future? Austral Ecol 35: 988992

Waters JM, Condie SA, Beheregaray LB (2014) Does coastaltopography constrain marine biogeography at an oceano -

11

http://dx.doi.org/10.1111/j.1442-9993.2010.02114.xhttp://dx.doi.org/10.1111/j.1365-294X.2005.02635.xhttp://dx.doi.org/10.1016/0022-0981(75)90071-4http://dx.doi.org/10.1007/s00227-007-0724-yhttp://dx.doi.org/10.1016/j.jtbi.2012.03.016http://dx.doi.org/10.1071/IS06038http://dx.doi.org/10.2307/1412159http://dx.doi.org/10.2307/2413122http://dx.doi.org/10.1038/sj.hdy.6885180http://dx.doi.org/10.3354/meps260083http://dx.doi.org/10.1111/j.1365-294X.2010.04658.xhttp://dx.doi.org/10.1016/S0272-7714(05)80001-0http://dx.doi.org/10.1029/2003JC001921http://dx.doi.org/10.2307/2409177http://dx.doi.org/10.1071/MF07229http://dx.doi.org/10.1016/j.margeo.2004.05.031http://dx.doi.org/10.1016/j.jmarsys.2014.02.001http://dx.doi.org/10.1111/j.0014-3820.2001.tb01279.xhttp://dx.doi.org/10.3354/meps07478http://dx.doi.org/10.1016/j.envsoft.2012.12.006http://dx.doi.org/10.1146/annurev.energy.29.062403.102254http://dx.doi.org/10.3354/meps07607http://dx.doi.org/10.1186/1471-2148-8-325http://dx.doi.org/10.1007/s10592-005-9056-yhttp://dx.doi.org/10.1029/2008JC005166http://dx.doi.org/10.1111/j.1755-0998.2010.02927.xhttp://dx.doi.org/10.3354/meps201211

Mar Ecol Prog Ser 532: 112, 2015

graphic interface? Mar Freshw Res 65: 969977Weersing K, Toonen RJ (2009) Population genetics, larval

dispersal, and connectivity in marine systems. Mar EcolProg Ser 393:112

Wernberg T, Smale DA, Tuya F, Thomsen MS and others(2013) An extreme climatic event alters marine ecosys-tem structure in a global biodiversity hotspot. Nat ClimChange 3: 7882

White C, Selkoe KA, Watson J, Siegel DA, Zacherl DC, Too-nen RJ (2010) Ocean currents help explain populationgenetic structure. Proc R Soc B 277: 16851694

Wood S, Paris CB, Ridgwell A, Hendy EJ (2013) Modellingdispersal and connectivity of broadcast spawning coralsat the global scale. Global Ecol Biogeogr 5: 111

Wright S (1965) The interpretation of population structure

by F-statistics with special regard to systems of mating.Evolution 19: 395420

York KL, Blacket MJ, Appleton BR (2008) The Bassian Isth-mus and the major ocean currents of southeast Australiainfluence the phylogeography and population structureof a southern Australian intertidal barnacle Catomeruspolymerus (Darwin). Mol Ecol 17: 19481961

Zardi GI, Nicastro KR, McQuaid CD, Hancke L, Helmuth B(2011) The combination of selection and dispersal helpsexplain genetic structure in intertidal mussels. Oecologia165: 947958

Zastrow CE, Houde ED, Morin LG (1991) Spawning, fecun-dity, hatch-date frequency and young-of-the year growthof bay anchovy Anchoa mitchilli in mid-Chesapeake Bay.Mar Ecol Prog Ser 73: 161171

12

Editorial responsibility: Christine Paetzold, Oldendorf/Luhe, Germany

Submitted: January 8, 2015; Accepted: May 20, 2015Proofs received from author(s): June 18, 2015

http://dx.doi.org/10.3354/meps073161http://dx.doi.org/10.1007/s00442-010-1788-9http://dx.doi.org/10.1111/j.1365-294X.2008.03735.xhttp://dx.doi.org/10.2307/2406450http://dx.doi.org/10.1098/rspb.2009.2214http://dx.doi.org/10.1038/nclimate1627

cite10: cite12: cite21: cite16: cite30: cite18: cite32: cite27: cite41: cite4: cite43: cite50: cite8: cite38: cite52: cite34: cite54: cite47: cite61: cite49: cite63: cite29: cite65: cite58: cite72: cite67: cite74: cite69: cite76: cite5: cite9: cite13: cite20: cite22: cite15: cite24: cite17: cite2: cite26: cite40: cite19: cite42: cite6: cite33: cite37: cite51: cite31: cite46: cite60: cite44: cite55: cite48: cite57: cite64: cite71: cite68: cite77: cite7: