, 20140556, published 8 May 2014281 2014 Proc. R. Soc. B C. Schunter, M. Pascual, J. C. Garza, N. Raventos and E. Macpherson coastlineKinship analyses identify fish dispersal events on a temperate

Supplementary data

tml http://rspb.royalsocietypublishing.org/content/suppl/2014/05/08/rspb.2014.0556.DC1.h

"Data Supplement"

Referenceshttp://rspb.royalsocietypublishing.org/content/281/1785/20140556.full.html#ref-list-1

This article cites 47 articles, 7 of which can be accessed free

This article is free to access

Subject collections

(147 articles)molecular biology (147 articles)genetics (1782 articles)ecology

Articles on similar topics can be found in the following collections

Email alerting service hereright-hand corner of the article or click Receive free email alerts when new articles cite this article - sign up in the box at the top

http://rspb.royalsocietypublishing.org/subscriptions go to: Proc. R. Soc. BTo subscribe to

on October 15, 2014rspb.royalsocietypublishing.orgDownloaded from on October 15, 2014rspb.royalsocietypublishing.orgDownloaded from

http://rspb.royalsocietypublishing.org/content/suppl/2014/05/08/rspb.2014.0556.DC1.html http://rspb.royalsocietypublishing.org/content/281/1785/20140556.full.html#ref-list-1http://rspb.royalsocietypublishing.org/cgi/collection/ecologyhttp://rspb.royalsocietypublishing.org/cgi/collection/geneticshttp://rspb.royalsocietypublishing.org/cgi/collection/molecular_biologyhttp://rspb.royalsocietypublishing.org/cgi/alerts/ctalert?alertType=citedby&addAlert=cited_by&saveAlert=no&cited_by_criteria_resid=royprsb;281/1785/20140556&return_type=article&return_url=http://rspb.royalsocietypublishing.org/content/281/1785/20140556.full.pdfhttp://rspb.royalsocietypublishing.org/subscriptionshttp://rspb.royalsocietypublishing.org/http://rspb.royalsocietypublishing.org/

on October 15, 2014rspb.royalsocietypublishing.orgDownloaded from rspb.royalsocietypublishing.orgResearchCite this article: Schunter C, Pascual M,Garza JC, Raventos N, Macpherson E. 2014

Kinship analyses identify fish dispersal events

on a temperate coastline. Proc. R. Soc. B 281:20140556.

http://dx.doi.org/10.1098/rspb.2014.0556Received: 6 March 2014

Accepted: 4 April 2014Subject Areas:ecology, molecular biology, genetics

Keywords:parentage analysis, connectivity, dispersal,

self-recruitment, sibship, Tripterygion delaisiAuthor for correspondence:C. Schunter

e-mail: celiaschunter@gmail.comElectronic supplementary material is available

at http://dx.doi.org/10.1098/rspb.2014.0556 or

via http://rspb.royalsocietypublishing.org.& 2014 The Authors. Published by the Royal Society under the terms of the Creative Commons AttributionLicense http://creativecommons.org/licenses/by/3.0/, which permits unrestricted use, provided the originalauthor and source are credited.Kinship analyses identify fish dispersalevents on a temperate coastline

C. Schunter1,2, M. Pascual2, J. C. Garza3, N. Raventos1 and E. Macpherson1

1Centre dEstudis Avancats de Blanes (CEAB-CSIC), Car. Acc. Cala St. Francesc 14, Blanes, Girona 17300, Spain2Department of Genetics and IRBio, Universitat de Barcelona, Diagonal 643, Barcelona 08028, Spain3Southwest Fisheries Science Center, National Marine Fisheries Service and University of California,110 Shaffer Road, Santa Cruz 95060, USA

Connectivity is crucial for the persistence and resilience of marine species,the establishment of networks of marine protected areas and the delineationof fishery management units. In the marine environment, understandingconnectivity is still a major challenge, due to the technical difficulties oftracking larvae. Recently, parentage analysis has provided a means toaddress this question effectively. To be effective, this method requires lim-ited adult movement and extensive sampling of parents, which is oftennot possible for marine species. An alternative approach that is less sensitiveto constraints in parental movement and sampling could be the reconstruc-tion of sibships. Here, we directly measure connectivity and larval dispersalin a temperate marine ecosystem through both analytical approaches. Weuse data from 178 single nucleotide polymorphism markers to performparentage and sibship reconstruction of the black-faced blenny (Tripterygiondelaisi) from an open coastline in the Mediterranean Sea. Parentage analysisrevealed a decrease in dispersal success in the focal area over 1 km distanceand approximately 6.5% of the juveniles were identified as self-recruits.Sibship reconstruction analysis found that, in general, full siblings did notrecruit together to the same location, and that the largest distance betweenrecruitment locations was much higher (11.5 km) than found for parentoffspring pairs (1.2 km). Direct measurements of dispersal are essential tounderstanding connectivity patterns in different marine habitats, and showthe degree of self-replenishment and sustainability of populations of marineorganisms. We demonstrate that sibship reconstruction allows direct measure-ments of dispersal and family structure in marine species while being moreeasily applied in those species for which the collection of the parentalpopulation is difficult or unfeasible.1. IntroductionLarval dispersal determines the connectivity patterns of many species of marinefish. Connectivity counteracts population structuring, and is crucial for the per-sistence and resilience of many species. This has been emphasized inconservation policies, including the design of networks of marine protectedareas [1,2]. However, owing to the technical difficulties in tracking fish throughthe pelagic larval phase, direct measures of connectivity are scarce, and under-standing the distribution of dispersal distances and their direction is still a greatchallenge in marine ecology [3].

One solution to overcome this problem is the use of parentage analysisto study dispersal in species with a relatively stationary adult phase [4]. Suchanalysis permits the estimation of connectivity, as the detection of parentoffspring pairs provides direct evidence of offspring dispersal, if adultmovement patterns are known [5]. Since the first application of this methodin the marine environment, a number of studies have been conducted thatdiffer in scale and in study species, and they have revealed wide variation inself-recruitment (from approx. 7.5 to 64% [6,7]). While some studies have

http://crossmark.crossref.org/dialog/?doi=10.1098/rspb.2014.0556&domain=pdf&date_stamp=2014-05-07mailto:celiaschunter@gmail.comhttp://dx.doi.org/10.1098/rspb.2014.0556http://dx.doi.org/10.1098/rspb.2014.0556http://rspb.royalsocietypublishing.orghttp://rspb.royalsocietypublishing.orghttp://rspb.royalsocietypublishing.org/

Arenys (0)

(b)

(a) (c)

Pilona (1)

0

Spain

5 10 km

Palomera(62) Blanes Clotilde (35)

Mar 1 (50)Lloret de

Lloret de Mar 2 (23)

Tossa de Mar (74)

N

946

382

Figure 1. (a) Territorial male of T. delaisi. (b) Map of the sampling area with location names and number of juveniles sampled in brackets. Note that from Palomeratowards the southwest, there are no rocky reefs apart from Pilona and Arenys. The squares represent the number of individuals analysed from the Blanes intensivesampling area, with adult sample size in light shade (online: green) and juvenile sample size in dark shade (online: red). (c) The Blanes intensive sampling area,where all dominant males encountered were sampled in their nests, indicated with light/green dots. (Online version in colour.)

rspb.royalsocietypublishing.orgProc.R.Soc.B

281:20140556

2

on October 15, 2014rspb.royalsocietypublishing.orgDownloaded from shown that larvae not recruiting back to the natal habitatemigrate, with dispersal success rapidly declining overshort distances [8], others have found larval export from amarine reserve in the Great Barrier Reef to reach locationsapproximately 30 km away [9], and in Papua New Guineaup to 35 km in distance [4]. Even so, this could still beconsidered a small scale, especially for species with largegeographical ranges or very discontinuous habitat.

The successful application of parentage analysis to thestudy of larval dispersal requires sampling of a sufficientlylarge proportion of the parental population for the success-ful encounter of parentoffspring pairs [10]. Owing to this,studies applying parentage analysis have been restricted tospecies with confined parental habitats [11,12] or that formreproductive aggregations [12]. However, few common, com-mercially harvested and endangered marine species,especially in temperate and cold water ecosystems, exhibitsuch characteristics. Other genetic approaches, such as recon-struction of sibling groups (sibships), might be useful toidentify dispersal and connectivity patterns. Sibship recon-struction analysis is a common pedigree reconstructionmethod and has been used to study a variety of questionsrelated to family structure and reproductive output in naturalpopulations [13]. In the marine environment, where thehigh fecundity of most species increases the challenge ofencountering kin, large full-sib families can be successfullyidentified and estimates of effective number of breedersderived [14]. In the terrestrial environment, studies using sib-ship analysis have focused on explaining whether siblingsdisperse together [15], but it is also possible to infer therange of the dispersal distances of a population using sibshipreconstruction [16].

In the marine environment, the collection of samplesfrom a large fraction of the population can be extremelydifficult and time-consuming. Performing sibship analysishas the advantage that only one generation needs to besampled, whereas parentage analysis requires sampling ofboth the adult and offspring generations. When extensiveadult collection is not feasible, for example in species withhigh adult motility, sibship reconstruction can be an infor-mative and suitable alternative approach for the estimationof dispersal.In this study, we compare the usefulness of sibship andparentage analyses in deciphering patterns of dispersaland connectivity in the marine environment using a smallrocky shore fish, Tripterygion delaisi (the black-faced blenny)[17], as a model system. It is one of the few temperate speciesthat exhibits the necessary characteristics for parentageanalysis and therefore allows direct comparison of bothmethodologies. We use data from 192 single nucleotidepolymorphisms (SNPs) to perform parentage and sibshipreconstruction analyses in the black-faced blenny on anopen coastline of the northwestern Mediterranean Sea. Thisspecies is a good candidate for parentage analysis, as territorialmales are confined to a very small area during the reproductiveperiod. Recruiting juveniles are readily collected on scuba,which facilitates sibship analysis, and allows us to compareand demonstrate how these two relationship inferencemethods can be used to elucidate patterns of larval dispersal.We use the results of these analyses to provide the first directestimates of the distribution of larval dispersal distances anddirectionality for a fish species in an open-coast temperatemarine ecosystem.2. Material and methods(a) Study species, study area and samplingThe black-faced blenny (T. delaisi) is distributed throughout theMediterranean Sea and the northeastern Atlantic Ocean [17] innear-shore rocky reef habitat, where they live camouflagedwithin the rocks or algae for most of the year. When the reproduc-tive period starts, in spring, some males, known as territorial ordominant males, undergo coloration changes that result in ablack head and bright yellow colouring for the rest of the body, dis-tinguishing them from other members of the species (figure 1a).These males protect a small territory, which is referred to as theirnest, against predators and uncoloured sneaker males [18]. Theblack-faced blenny also displays a high level of homing behaviour[19], and adults do not disperse even small distances in open waterand sandy bottom habitat. Consequently, dispersal, as well as geneflow, is confined to the pelagic larval stage.

The study area is on an open Mediterranean coastline of north-eastern Spain, near the port of Blanes (figure 1b). It is centred on arocky shore of approximately 2 km in length, surrounded by sandy

http://rspb.royalsocietypublishing.org/

rspb.royalsocietypublishing.orgProc.R.Soc.B

281:201

3

on October 15, 2014rspb.royalsocietypublishing.orgDownloaded from bottom habitat that extends for many kilometres in both directions,but especially towards the southwest (figure 1c), and could act as adispersal barrier [20].

A total of 946 adult fish were sampled on scuba with fish netsbetween April and July 2010. Exhaustive searching for completecoverage of all nests in the 2 km section of coastline (Blanes area;figure 1c) was performed and 796 territorial males were sampledfrom their nesting sites. Additionally, 150 camouflaged adults ofunknown sex were sampled. Fish were caught with small nets,body length measured and a small tissue sample taken fromthe dorsal fin before release.

A total of 627 juveniles were collected on scuba between themonths of July and September 2010. Juveniles were sampled inthe Blanes area, as well as at seven external locations, spanning adistance of 48.2 km, between Arenys and Tossa de Mar (figure1b). Four of these locations are found to the northeast ofBlanes, where, after a large sandy bottom stretch, the rockyshore continues providing habitat for T. delaisi. Southwest ofBlanes, juveniles were sampled at all the available sites withrocky habitat (figure 1b).40556(b) Genotyping and genetic analysesAll 1573 samples were genotyped at 192 SNP markers (electronicsupplementary material, table S1), developed for T. delaisi, using96.96 Dynamic SNP Genotyping Array on an EP1 GenotypingSystem (Fluidigm Corporation) [21]. Deviations from HardyWeinberg (HW) and linkage disequilibria were evaluated withGENEPOP v. 4.2 [22] and loci not in HW equilibrium, as well asthe less variable locus of pairs in linkage disequilibrium, wereeliminated. Genotypes from the remaining 178 SNP loci wereused for further analysis. Details of SNP development and of allgenotyping assays are described by Schunter et al. [21].

Two kinship analyses were performed: parentage and sibship.Parentoffspring matches were established using the softwareCERVUS [23], which accommodates a large number of loci andhas been shown to accurately identify parentoffspring pairs inempirical data [24]. The genotyping error rate was set to 1%, anda range of values between 25 and 95% for the parameter adult pro-portion sampled was evaluated. A maximum of three mismatcheswas allowed. The critical LOD value was set at 95% confidence, butthe final LOD value cut-off for identifying parentoffspring pairswas always more than or equal to 2. This is a conservative approachthat accounts for some not-sampled putative parents in the popu-lation. Estimates of the probability of excluding a parent evenwhen the parent was sampled (type I error) and of the probabilityof assigning a false parent whether the true parent was sampled ornot (type II error) were evaluated with CERVUS. The programCOLONY [25] was used to verify parentoffspring pairs identifiedwith CERVUS.

COLONY [25] was also used to identify siblings in the juven-ile samples. The predefined parameters were used, except thatthe mating system was set to polygamy. Full sibship was onlyaccepted for posterior probability values of more than 0.75. Thedataset was also permuted five times with sgm-perm [26] toobtain randomized datasets. The analyses were repeated fivetimes with these permuted datasets to estimate the probabilityof assigning sibship pairs by chance alone. The programML-RELATE [27] was also used to verify sibling assignments.

In order to identify any potential genetic structure within thesamples, a Bayesian clustering analysis was performed withSTRUCTURE v. 2.3.4 [28]. The analyses were carried out with all1573 samples included, as well as with juveniles and adults separ-ately. For each possible number of genetic units (K 13), thevalues over 10 runs were averaged with 100 000 iterations dis-carded as burn-in and 200 000 iterations retained. Geneticstructuring between juveniles from Blanes and the other samplinglocations was evaluated with FST calculated using GENEPOP v. 4.2[22]. Geographical distances between individuals were estimatedfrom the sampling coordinates (electronic supplementary material,tables S2 and S3) with the R package fields [29].

(c) Otolith analysis and wind variablesThe body length of all juveniles was measured, and the lapilliotoliths were extracted and mounted on microscope slides. Theage of individuals was determined in order to establish the cor-responding dates of hatch and settlement, as well as pelagiclarval duration (PLD), following Raventos & Macpherson [30].

The main documented superficial currents in this region ofthe Mediterranean Sea do not reach the coastal area, and there-fore do not affect near shore species [31]. However, windcharacteristics have a strong influence on the inshore circulationpattern in the study area [32,33]. As Tripterygion larvae arealways distributed along inshore waters (less than 2.5 km fromshore) [34], we compared the larval dispersal distance and direc-tion with wind speed and direction to evaluate possiblecorrelations with dispersal patterns. Wind data were obtainedfrom an automated meteorological station belonging to theXMET service (National Weather Service of Catalonia). Winddata were recorded hourly with an anemometer 10 m above theground. The overall wind regime in the study area revealed aclear diurnal pattern with stronger and more variable winds dueto solar heating during the day, and mostly weaker and less vari-able winds blowing along the coast at night [33]. The wind datawere therefore divided into day and night datasets by averagingthe 12.00 and 00.00 measurements, respectively. The settlementdate of each juvenile (as derived from otolith readings) and thecorresponding PLD value were used to determine the periodover which the wind variables were to be averaged. Day- andnight-time daily averages were calculated, and wind speedand direction combined into a single coarse wind variable (elec-tronic supplementary material, table S4) and compared withdispersal patterns using a Spearmans rank correlation test.3. ResultsA total of 25 parentoffspring pairs were identified, all in theBlanes area (figure 2). The same 25 pairs were found withboth parentage methods, CERVUS and COLONY. Usingthe 95% critical LOD value given by CERVUS, the numberof parentoffspring pairs varied from 25 (25% of adultssampled) to 119 (95% of adults sampled). By using a conser-vative threshold of LOD more than or equal to 2, which wasthe 95% critical LOD for 25% of adults sampled, the same 25pairs were detected independent of CERVUS input par-ameters. Type I error estimates were all 0 and the type IIerror varied from 0 to 0.0041 (25% and 95% adults sampled,respectively). We found that 6.5% of the juveniles from theBlanes area were self-recruits (figure 2), with one of theseoffspring found directly adjacent to the fathers nest. Self-recruits did not disperse further than 1.2 km within the2 km intensive sampling area, with 80% settling less than1 km from their natal nest (figure 3).

Sibship analysis with COLONY identified 21 pairs ofjuveniles as full siblings. ML-RELATE identified six of these21 pairs as full siblings, whereas the other 15 pairs had simi-lar likelihoods of being full or half siblings. No parents wereidentified for any of the 42 juveniles in these pairs. By per-muting the juvenile dataset, we found that the probabilityof assignments due to chance is low (0.0028%). As T. delaisiadults do not disperse and males territorially guard onenest, even half siblings will come from either the same nest

http://rspb.royalsocietypublishing.org/

0

N

0.25 0.5 km

southwest

5

4

3

2

1

01.21.4 1.01.2 0.81.0 0.60.8 0.40.6 0.40.6 0.60.8 0.81.0 1.01.2 1.21.40.20.4 0.20.4

rspb.royalsocietypublishing.orgProc.R.Soc.B

281:20140556

5

on October 15, 2014rspb.royalsocietypublishing.orgDownloaded from temperate ecosystem. Self-recruitment, which is the percentageof larvae that settle back to the natal location, in marine fisheshas long been debated [35], but empirical studies over the last15 years have begun to elucidate patterns [35,36]. The degreeto which a population receives self-recruits has ecological impli-cations, as it influences the level of self-replenishment andpopulation sustainability [37]. Studies using parentage analysisto study dispersal in several species of coral reef fishesinhabiting embayments in Papua New Guinea revealed highself-recruitment [4,7], with 4064% of juveniles recruiting totheir natal site. Studies using a similar approach with coralreef fishes but on open coastlines found the proportion of self-recruiting juveniles to average 4.67.5% per site [11,38]. Thepercentage of self-recruitment for T. delaisi on a temperateopen coastline found here (6.5%) was similar. Saenz-Agudeloet al. [11] argue that coastal geography may be crucial inexplaining the relatively low rates of self-recruitment. It hasbeen suggested that habitat patchiness determines the opennessof populations in marine species with different dispersal abil-ities [39]. Temperate open coastlines typically lack a networkof suitable habitat, as is often found in a tropical reef environ-ment, and recruitment is only possible on a narrow strip ofrocky shoreline habitat. This suggests that the proportion ofself-recruiting individuals is reflected by the openness orpatchiness of the habitat geography, or lack of available habitat.

Pelagic larval duration has been suggested to have alarge influence on the rate of self-recruitment [5,29]. How-ever, this does not seem to be the case for T. delaisi, as wedetected a low proportion of self-recruits, with an averagePLD of 18 days, and a similar proportion was estimated fora coral reef species with a PLD of just several days [6]. Never-theless, the percentage of self-recruitment found in T. delaisi isa minimum estimate, because sneaker males may also con-tribute to the progeny. We sampled all the territorial malesobserved in the study area and tried to capture sneakermales, but were largely unsuccessful, as they are camou-flaged and quite difficult to see. However, sneaker malesare rarely observed to participate in mating, suggestinggenerally low reproductive output for this behaviouralstrategy [40] and a minor contribution to the proportion ofself-recruits in this population.

Large-scale investigations that integrate population gen-etics and oceanographic modelling have shown thatconnectivity can be influenced, or even dominated, by physicalprocesses [4144]. However, near-shore oceanographic fea-tures are generally hard to model, because they are mostlyinfluenced by wind and coastal morphology [32]. For example,dispersal patterns of the coral reef fish Amphyprion polymnuswere found not to be influenced by physical processes, suchas currents [11]. For T. delaisi, we found no significant corre-lation between the dispersal of self-recruits and thecomponent wind variable, probably due to the lack of a patternof directional dispersal. None of the juveniles collected in thesampling areas northeast of Blanes were offspring of theBlanes parents, which could be due to the large populationsize. For those recruits that dispersed southwestward, thecoastline is formed by sandy beaches with no available rockyreef habitat for approximately 100 km, suggesting little suc-cessful dispersal occurring beyond this area. It waspreviously reported that the population of Blanes is geneticallydistinct from that of the next analysed locality approximately250 km to the southwest, with a low number of juvenilesassigned to distant locations [17]. However, these populationsare not completely disconnected, as isolation by distance wasfound along the Spanish coast [45], suggesting that a smallnumber of larval recruits must successfully disperse long dis-tances to the southwest. As such, the majority of larvae fromthis open-coastline habitat settle in suitable habitat adjacentto their natal location, but small numbers of larvae dispersefurther and allow large-scale connectivity.

Parentage analysis in T. delaisi revealed limited successfuldispersal and only short-distance dispersal events in the focalstudy area. Identified dispersal events declined precipitouslymore than 1 km from the natal site, as previously found inreef fish species [8,38]. The low level of self-recruitment,though, would indicate high influx of recruits from otherareas, which in turn suggests a larger dispersal range.Indeed, the T. delaisi population in the study area was foundto be one genetic unit connected by high gene flow, consistentwith previous studies [17,45].

The high connectivity found in population structure ana-lyses was reflected in the sibling pairs identified. Sibshipanalysis has previously been used to successfully elucidatefamily structure and mating strategies in several other fishspecies [4648]. Here, we use it to provide information onthe dispersal potential of a species in the marine environ-ment. Whereas paternity analysis can provide informationon both dispersal distance and directionality, identification ofsiblings recruiting to different locations provides estimatesof a possible range of dispersal distances, because full sib-lings begin dispersal from the same natal nest. The meandistance between T. delaisi sibling pairs was much largerthan the maximum distance between parentoffspring pairs(figure 3). Nonetheless, sibship analysis could still underesti-mate dispersal distance, because the natal location is notknown. If siblings disperse in opposite directions, then thedistance between them will be twice their mean dispersal dis-tance, but if they do not, then the observed distance will beless than their combined dispersal distance.

Combining kin relationship estimation with otolith analysis,it is also possible to retrace whether siblings recruited together.Indeed, a recent study on a coral reef fish provided evidencethat siblings can travel together throughout the entire pelagiclarval phase [49]. Nonetheless, there was only one T. delaisisibling pair encountered at the same recruitment location. Oto-lith analysis found that both siblings hatched within 2 days ofeach other and spent 19 days in the pelagic larval phase, indicat-ing that they probably left the nest at the same time andrecruited together. Even though many groups of juvenileswere collected simultaneously from the same locations, onlythis one pair of siblings was identified as having recruitedtogether. This indicates that the majority of full siblings do notrecruit to the same location, suggesting a high degree of larvalmixing and recruitment heterogeneity [49].

In this study, by selecting to work with a species for whichparentage analysis is possible due to the sedentary, territorialbehaviour of adults, we can compare the results of the two kin-ship methods. With parentage analysis, the direction ofdispersal events is known, which cannot be accomplishedwith sibship reconstruction as the location of origin of the sib-lings is generally not known. Nonetheless, in our study therewas no clear pattern of dispersal direction, even though parent-age analysis was performed. With sibship reconstruction,which does not require sedentary life history or sampling ofparents, the dispersal range of the species is estimated, andinference regarding family structure and dispersal behaviour

http://rspb.royalsocietypublishing.org/

rspb.royalsocietypublishing.orgProc.R.Soc.B

281:20140

6

on October 15, 2014rspb.royalsocietypublishing.orgDownloaded from can be derived. For T. delaisi, the parentoffspring comparisonsindicated that larvae tend to move short distances alongthe coast. By contrast, the sibling pairs identified showedthat recruitment can be at locations quite distant from theirnatal site, and from each other, reflecting the high degree ofconnectivity inferred from population structure analyses.

The information provided by these two methods can becomplementary, and provides a more complete picture of dis-persal dynamics and population connectivity than either onealone. The direct measurement of dispersal distance anddirection by applying parentage analysis is dramaticallyincreasing our knowledge of larval dispersal in marineorganisms. However, for many species, including those ofconservation concern, sampling a large proportion of theparental population is difficult or unfeasible, or the life his-tory may make it difficult to infer the origin of dispersalevents. In such cases, sibship reconstruction can be moreeasily applied, as only one generation needs be sampled, toelucidate dispersal patterns and structure in marine species.Collection and field procedures followed Spanish Law (Royal Execu-tive Order, 53/2013) for Animal Experimentation, in accordance with

European Union directive 2010/63/UE.

Acknowledgement. We are grateful to the Southwest Fisheries Science CenterMolecular Ecology and Genetic Analysis Team, especially E. Anderson,A. Abada-Cardoso, V. Apkenas, C. Columbus, E. Gilbert-Horvath andD. Pearse, for support in the laboratory and helpful comments. Thanksto C. Arenas and F. Mestres for their statistical input. We also thankE. Ballesteros for providing a photo of the black-faced blenny.Data accessibility. Genotypes of T. delaisi adults and recruits at the 178SNP markers can be found in the electronic supplementary material,table S1. Coordinates for each collected individual sample are in theelectronic supplementary material, tables S2 and S3, and wind vari-ables including wind direction and strength can be found in theelectronic supplementary material, table S4.Funding statement. This work was partially funded by the SpanishMinistry of Science and Innovation through the BENTHOMICS(CTM2010-22218-C02-01-02) projects and the FBBVA project(BIOCON 08-187?09). The authors C.S., M.P. and E.M. are part ofthe research group 2009SGR-636 and 2009SGR-665 of the Generalitatde Catalunya, and C.S. was funded by a JAE-Predoctoral Fellowship.556References1. Botsford LW, White JW, Coffroth M-A, Paris CB,Planes S, Shearer TL, Thorrold SR, Jones GP. 2009Connectivity and resilience of coral reefmetapopulations in marine protected areas: matchingempirical efforts to predictive needs. Coral Reefs 28,327337. (doi:10.1007/s00338-009-0466-z)

2. Palumbi SR. 2003 Population genetics, demographicconnectivity, and the design of marine reserves.Ecol. Appl. 13, 146158. (doi:10.1890/1051-0761(2003)013[0146:PGDCAT]2.0.CO;2)

3. Selkoe K, Toonen RJ. 2011 Marine connectivity: anew look at pelagic larval duration and geneticmetrics of dispersal. Mar. Ecol. Prog. Ser. 436,291305. (doi:10.3354/meps09238)

4. Planes S, Jones GP, Thorrold SR. 2009 Larvaldispersal connects fish populations in a network ofmarine protected areas. Proc. Natl Acad. Sci. USA106, 56935697. (doi:10.1073/pnas.0808007106)

5. Hedgecock D, Barber P, Edmands S. 2007 Geneticapproaches to measuring connectivity. Oceanography20, 7079. (doi:10.5670/oceanog.2007.30)

6. Saenz-Agudelo P, Jones GP, Thorrold SR, Planes S.2012 Patterns and persistence of larval retentionand connectivity in a marine fish metapopulation.Mol. Ecol. 21, 46954705. (doi:10.1111/j.1365-294X.2012.05726.x)

7. Berumen ML, Almany GR, Planes S, Jones GP,Saenz-Agudelo P, Thorrold SR. 2012 Persistence ofself-recruitment and patterns of larval connectivityin a marine protected area network. Ecol. Evol. 2,444452. (doi:10.1002/ece3.208)

8. Buston PM, Jones GP, Planes S, Thorrold SR. 2012Probability of successful larval dispersal declinesfivefold over 1 km in a coral reef fish. Proc. R. Soc. B279, 18831888. (doi:10.1098/rspb.2011.2041)

9. Harrison HB et al. 2012 Larval export from marinereserves and the recruitment benefit for fish andfisheries. Curr. Biol. 22, 10231028. (doi:10.1016/j.cub.2012.04.008)10. Jones AG, Ardren WR. 2003 Methods of parentageanalysis in natural populations. Mol. Ecol. 12,25112523. (doi:10.1046/j.1365-294X.2003.01928.x)

11. Saenz-Agudelo P, Jones GP, Thorrold SR, Planes S.2011 Connectivity dominates larval replenishmentin a coastal reef fish metapopulation. Proc. R. Soc. B278, 29542961. (doi:10.1098/rspb.2010.2780)

12. Almany GR et al. 2013 Dispersal of grouper larvaedrives local resource sharing in a coral reef fishery.Curr. Biol. 23, 626630. (doi:10.1016/j.cub.2013.03.006)

13. Blouin MS. 2003 DNA-based methods for pedigreereconstruction and kinship analysis in naturalpopulations. Trends Ecol. Evol. 18, 503511.(doi:10.1016/S0169-5347(03)00225-8)

14. Liu J-X, Ely B. 2009 Sibship reconstructiondemonstrates the extremely low effective populationsize of striped bass Morone saxatilis in the Santee-Cooper system, South Carolina, USA. Mol. Ecol.18, 41124120. (doi:10.1111/j.1365-294X.2009.04343.x)

15. Sharp SP, Simeoni M, Hatchwell BJ. 2008 Dispersalof sibling coalitions promotes helping amongimmigrants in a cooperatively breeding bird.Proc. R. Soc. B 275, 21252130. (doi:10.1098/rspb.2008.0398)

16. Lepais O, Darvill B, OConnor S, Osborne JL,Sanderson RA, Cussans J, Goffe L, Goulson D. 2010Estimation of bumblebee queen dispersal distancesusing sibship reconstruction method. Mol. Ecol. 19,819831. (doi:10.1111/j.1365-294X.2009.04500.x)

17. Carreras-Carbonell J, Macpherson E, Pascual M. 2005Rapid radiation and cryptic speciation inmediterranean triplefin blennies (Pisces:Tripterygiidae) combining multiple genes. Mol.Phylogenet. Evol. 37, 751761. (doi:10.1016/j.ympev.2005.04.021)

18. Wirtz P. 1978 The behaviour of the MediterraneanTripterygion species (Pisces, Blennioidei).Z. Tierpsychol. 48, 142174. (doi:10.1111/j.1439-0310.1978.tb00253.x)

19. Heymer A. 1977 Experiences subaquatiques sur lesperformances dorientation et de retour au gite chezTripterygion tripteronotus et Tripterygionxanthosoma (Blennioidei, Tripterygiidae). Vie Milieu27, 425435.

20. Carreras-Carbonell J, Macpherson E, Pascual M. 2007High self-recruitment levels in a Mediterraneanlittoral fish population revealed by microsatellitemarkers. Mar. Biol. 151, 719727. (doi:10.1007/s00227-006-0513-z)

21. Schunter C, Garza JC, Macpherson E, Pascual M.2014 SNP development from RNA-seq data in anonmodel fish: how many individuals areneeded for accurate allele frequency prediction?Mol. Ecol. Resour. 14, 157165. (doi:10.1111/1755-0998.12155)

22. Rousset F. 2008 genepop007: a complete re-implementation of the Genepop software forWindows and Linux. Mol. Ecol. Resour. 8, 103106.(doi:10.1111/j.1471-8286.2007.01931.x)

23. Kalinowski ST, Taper ML, Marshall TC. 2007 Revisinghow the computer program CERVUS accommodatesgenotyping error increases success in paternityassignment. Mol. Ecol. 16, 10991106. (doi:10.1111/j.1365-294X.2007.03089.x)

24. Slate J, Marshall T, Pemberton J. 2000 A retrospectiveassessment of the accuracy of the paternity inferenceprogram CERVUS. Mol. Ecol. 9, 801808. (doi:10.1046/j.1365-294x.2000.00930.x)

25. Jones OR, Wang J. 2010 COLONY: a program forparentage and sibship inference from multilocusgenotype data. Mol. Ecol. Resour. 10, 551555.(doi:10.1111/j.1755-0998.2009.02787.x)

26. Anderson EC. 2012 sgm_permpermute a simplegenetic matrix. See http://users.soe.ucsc.edu/~eriq/dokuwiki/doku.php?id=sisg:sibship_inference_practical2012.

http://dx.doi.org/10.1007/s00338-009-0466-zhttp://dx.doi.org/10.1890/1051-0761(2003)013[0146:PGDCAT]2.0.CO;2http://dx.doi.org/10.1890/1051-0761(2003)013[0146:PGDCAT]2.0.CO;2http://dx.doi.org/10.3354/meps09238http://dx.doi.org/10.1073/pnas.0808007106http://dx.doi.org/10.5670/oceanog.2007.30http://dx.doi.org/10.1111/j.1365-294X.2012.05726.xhttp://dx.doi.org/10.1111/j.1365-294X.2012.05726.xhttp://dx.doi.org/10.1002/ece3.208http://dx.doi.org/10.1098/rspb.2011.2041http://dx.doi.org/10.1016/j.cub.2012.04.008http://dx.doi.org/10.1016/j.cub.2012.04.008http://dx.doi.org/10.1046/j.1365-294X.2003.01928.xhttp://dx.doi.org/10.1098/rspb.2010.2780http://dx.doi.org/10.1016/j.cub.2013.03.006http://dx.doi.org/10.1016/j.cub.2013.03.006http://dx.doi.org/10.1016/S0169-5347(03)00225-8http://dx.doi.org/10.1111/j.1365-294X.2009.04343.xhttp://dx.doi.org/10.1111/j.1365-294X.2009.04343.xhttp://dx.doi.org/10.1098/rspb.2008.0398http://dx.doi.org/10.1098/rspb.2008.0398http://dx.doi.org/10.1111/j.1365-294X.2009.04500.xhttp://dx.doi.org/10.1016/j.ympev.2005.04.021http://dx.doi.org/10.1016/j.ympev.2005.04.021http://dx.doi.org/10.1111/j.1439-0310.1978.tb00253.xhttp://dx.doi.org/10.1111/j.1439-0310.1978.tb00253.xhttp://dx.doi.org/10.1007/s00227-006-0513-zhttp://dx.doi.org/10.1007/s00227-006-0513-zhttp://dx.doi.org/10.1111/1755-0998.12155http://dx.doi.org/10.1111/1755-0998.12155http://dx.doi.org/10.1111/j.1471-8286.2007.01931.xhttp://dx.doi.org/10.1111/j.1365-294X.2007.03089.xhttp://dx.doi.org/10.1111/j.1365-294X.2007.03089.xhttp://dx.doi.org/10.1046/j.1365-294x.2000.00930.xhttp://dx.doi.org/10.1046/j.1365-294x.2000.00930.xhttp://dx.doi.org/10.1111/j.1755-0998.2009.02787.xhttp://users.soe.ucsc.edu/~eriq/dokuwiki/doku.php?id=sisg:sibship_inference_practical2012http://users.soe.ucsc.edu/~eriq/dokuwiki/doku.php?id=sisg:sibship_inference_practical2012http://users.soe.ucsc.edu/~eriq/dokuwiki/doku.php?id=sisg:sibship_inference_practical2012http://users.soe.ucsc.edu/~eriq/dokuwiki/doku.php?id=sisg:sibship_inference_practical2012http://rspb.royalsocietypublishing.org/

rspb.royalsocietypublishing.orgProc.R.Soc.B

281:20140556

7

on October 15, 2014rspb.royalsocietypublishing.orgDownloaded from 27. Kalinowski ST, Wagner AP, Taper ML. 2006ML-RELATE: a computer program for maximumlikelihood estimation of relatedness andrelationship. Mol. Ecol. Notes 6, 576579. (doi:10.1111/j.1471-8286.2006.01256.x)

28. Pritchard JK, Stephens M, Donnelly P. 2000Inference of population structure using multilocusgenotype data. Genetics 155, 945959.

29. Fields Development Team. 2006 Fields: toolsfor spatial data. See http://www.cgd.ucar.edu/Software/Fields.

30. Raventos N, Macpherson E. 2001 Planktonic larvalduration and settlement marks on the otolithsof Mediterranean littoral fishes. Mar. Biol. 138,11151120. (doi:10.1007/s002270000535)

31. Font J, Garcialadona E, Gorriz E. 1995 Theseasonality of mesoscale motion in the northerncurrent of the western Mediterranean: several yearsof evidence. Oceanol. Acta 18, 207219.

32. Grifoll M, Aretxabaleta AL, Espino M, Warner JC.2012 Along-shelf current variability on theCatalan inner-shelf (NW Mediterranean).J. Geophys. Res. 117, C09027. (doi:10.1029/2012JC008182)

33. Raventos N, Macpherson E. 2005 Environmentalinfluences on temporal patterns of settlement intwo littoral labrid fishes in the Mediterranean Sea.Estuar. Coast. Shelf Sci. 63, 479487. (doi:10.1016/j.ecss.2004.11.018)

34. Sabates A, Zabala M, Garca-Rubies A. 2003 Larval fishcommunities in the Medes Islands Marine Reserve(north-west Mediterranean). J. Plankton Res. 25,10351046. (doi:10.1093/plankt/25.9.1035)

35. Jones GP, Milicich MJ, Emslie MJ, Lunow C. 1999Self-recruitment in a coral reef fish population.Nature 402, 802804. (doi:10.1038/45538)36. Jones GP, Planes S, Thorrold SR. 2005 Coral reeffish larvae settle close to home. Curr. Biol. 15,13141318. (doi:10.1016/j.cub.2005.06.061)

37. Sponaugle S et al. 2002 Predicting self-recruitmentin marine populations: biophysical correlates andmechanisms. Bull. Mar. Sci. 70, 341375.

38. DAloia CC, Bogdanowicz SM, Majoris JE,Harrison RG, Buston PM. 2013 Self-recruitmentin a Caribbean reef fish: a method forapproximating dispersal kernels accountingfor seascape. Mol. Ecol. 22, 25632572.(doi:10.1111/mec.12274)

39. Pinsky ML, Palumbi SR, Andrefouet S, Purkis SJ.2012 Open and closed seascapes: where doeshabitat patchiness create populations with highfractions of self-recruitment? Ecol. Appl. 22,12571267. (doi:10.1890/11-1240.1)

40. De Jonge J, Videler JJ. 1989 Differences betweenthe reproductive biologies of Tripterygiontripteronotus and T. delaisi (Pisces, Perciformes,Tripterygiidae): the adaptive significance of analternative mating strategy and a red instead of ayellow nuptial colour. Mar. Biol. 100, 431437.(doi:10.1007/BF00394818)

41. White C, Selkoe K, Watson J, Siegel DA, Zacherl DC,Toonen RJ. 2010 Ocean currents help explainpopulation genetic structure. Proc. R. Soc. B 277,16851694. (doi:10.1098/rspb.2009.2214)

42. Schunter C, Carreras-Carbonell J, Macpherson E,Tintore J, Vidal-Vijande E, Pascual A, Guidetti P,Pascual M. 2011 Matching genetics withoceanography: directional gene flow in aMediterranean fish species. Mol. Ecol. 20,51675181. (doi:10.1111/j.1365-294X.2011.05355.x)

43. Selkoe KA, Gaines SD, Caselle JE, Warner RR. 2006Current shifts and kin aggregation explain geneticpatchiness in fish recruits. Ecology 87, 30823094.(doi:10.1890/0012-9658(2006)87[3082:CSAKAE]2.0.CO;2)

44. Pujolar JM, Schiavina M, Di Franco A, Melia P,Guidetti P, Gatto M, De Leo GA, Zane L. 2013Understanding the effectiveness of marine protectedareas using genetic connectivity patterns andLagrangian simulations. Divers. Distrib. 19,15311542. (doi:10.1111/ddi.12114)

45. Carreras-Carbonell J, Macpherson E, Pascual M. 2006Population structure within and between subspeciesof the Mediterranean triplefin fish Tripterygiondelaisi revealed by highly polymorphic microsatelliteloci. Mol. Ecol. 15, 35273539. (doi:10.1111/j.1365-294X.2006.03003.x)

46. De Mestral LG, Herbinger CM, OReilly PT, Taylor EB.2012 Mating structure of an endangered populationof wild Atlantic salmon (Salmo salar) as determinedusing sibship reconstruction and a novel methodof sex inference. Can. J. Fish. Aquat. Sci. 69,13521361. (doi:10.1139/f2012-065)

47. Dibattista JD, Feldheim KA, Garant D, Gruber SH,Hendry AP. 2009 Evolutionary potential of a largemarine vertebrate: quantitative genetic parametersin a wild population. Evolution 63, 10511067.(doi:10.1111/j.1558-5646.2008.00605.x)

48. Kanno Y, Vokoun JC, Letcher BH. 2010 Sibshipreconstruction for inferring mating systems,dispersal and effective population size in headwaterbrook trout (Salvelinus fontinalis) populations.Conserv. Genet. 12, 619628. (doi:10.1007/s10592-010-0166-9)

49. Bernardi G, Beldade R, Holbrook SJ, Schmitt RJ.2012 Full-sibs in cohorts of newly settled coral reeffishes. PLoS ONE 7, e44953. (doi:10.1371/journal.pone.0044953)

http://dx.doi.org/10.1111/j.1471-8286.2006.01256.xhttp://dx.doi.org/10.1111/j.1471-8286.2006.01256.xhttp://www.cgd.ucar.edu/Software/Fieldshttp://www.cgd.ucar.edu/Software/Fieldshttp://www.cgd.ucar.edu/Software/Fieldshttp://dx.doi.org/10.1007/s002270000535http://dx.doi.org/10.1029/2012JC008182http://dx.doi.org/10.1029/2012JC008182http://dx.doi.org/10.1016/j.ecss.2004.11.018http://dx.doi.org/10.1016/j.ecss.2004.11.018http://dx.doi.org/10.1093/plankt/25.9.1035http://dx.doi.org/10.1038/45538http://dx.doi.org/10.1016/j.cub.2005.06.061http://dx.doi.org/10.1111/mec.12274http://dx.doi.org/10.1890/11-1240.1http://dx.doi.org/10.1007/BF00394818http://dx.doi.org/10.1098/rspb.2009.2214http://dx.doi.org/10.1111/j.1365-294X.2011.05355.xhttp://dx.doi.org/10.1890/0012-9658(2006)87[3082:CSAKAE]2.0.CO;2http://dx.doi.org/10.1890/0012-9658(2006)87[3082:CSAKAE]2.0.CO;2http://dx.doi.org/10.1111/ddi.12114http://dx.doi.org/10.1111/j.1365-294X.2006.03003.xhttp://dx.doi.org/10.1111/j.1365-294X.2006.03003.xhttp://dx.doi.org/10.1139/f2012-065http://dx.doi.org/10.1111/j.1558-5646.2008.00605.xhttp://dx.doi.org/10.1007/s10592-010-0166-9http://dx.doi.org/10.1007/s10592-010-0166-9http://dx.doi.org/10.1371/journal.pone.0044953http://dx.doi.org/10.1371/journal.pone.0044953http://rspb.royalsocietypublishing.org/

Kinship analyses identify fish dispersal events on a temperate coastlineIntroductionMaterial and methodsStudy species, study area and samplingGenotyping and genetic analysesOtolith analysis and wind variables

ResultsDiscussionAcknowledgementData accessibilityFunding statementReferences