Genetic Connectivity in Corals on the Flower Garden Banks and Surrounding Oil/Gas Platforms, Gulf of Mexico

Journal of Experimental Marine Biology and Ecology 365 (2008) 1–12

Contents lists available at ScienceDirect

Journal of Experimental Marine Biology and Ecology

j ourna l homepage: www.e lsev ie r.com/ locate / jembe

Genetic Connectivity in Corals on the Flower Garden Banks and Surrounding Oil/GasPlatforms, Gulf of Mexico

Amy D. Atchison a,b, Paul W. Sammarco a,b,⁎, Daniel A. Brazeau c

a Department of Oceanography and Coastal Sciences, Louisiana State University A&M College, Baton Rouge, LA 70803, USAb Louisiana Universities Marine Consortium (LUMCON), 8124 Hwy. 56, Chauvin, LA 70344, USAc Pharmaceutical Genetics Laboratory, Department of Pharmaceutical Sciences, 517 Cook Hall, University of Buffalo, Buffalo, NY 14260-1200, USA

⁎ Corresponding author. Louisiana Universities MarinHwy. 56, Chauvin, LA 70344, USA.

E-mail addresses: [email protected] (A.D. Atchi(P.W. Sammarco).

0022-0981/$ – see front matter © 2008 Elsevier B.V. Aldoi:10.1016/j.jembe.2008.07.002

a b s t r a c t
a r t i c l e i n f o
Article history:
The northern Gulf of Mexic Received 31 August 2007Received in revised form 27 March 2008Accepted 1 July 2008
Keywords:AFLPBroadcasterBrooderCoral, genetic affinityConnectivityDiploria strigosaFlower Garden BanksMadracis decactisMeso-scaleMolecular geneticsMontastraea cavernosa

o (GOM) currently possesses ∼3,600 offshore oil and gas production platforms.These platforms serve as artificial reefs on the continental shelf, where, until their introduction, shallow hardsubstrata were rare. This newly available substrate has helped to expand scleractinian coral populations inthe GOM. Here, we conduct molecular genetic analyses on adult scleractinian corals on the Flower GardenBanks (FGB) coral reefs (∼180 km SE of Galveston, TX) and on surrounding oil and gas platforms. We haveattempted to determine the degree of genetic affinity among the natural populations and those on thesurrounding platforms. The three most abundant hermatypic scleractinian species were sampled: Madracisdecactis, Diploria strigosa, and Montastraea cavernosa. Tissue samples were collected from the East (E-) andWest (W-) FGB, and seven platforms within a 65 km radius of the FGB, at a depth range of 5-37 m. Geneticvariation was assessed using Amplified Fragment Length Polymorphisms (AFLPs). The large number ofpolymorphic markers generated by AFLPs allowed for the use of standard genetic analysis tools (AMOVA) aswell as population allocation techniques (AFLPOP). AMOVA analyses indicated that the E- and W-FGB weregenetically homogeneous for populations of Madracis decactis and Diploria strigosa; Montastraea cavernosapopulations, however, were significantly different there. In all species, genetic distance (FST) increasedsignificantly with geographic distance between populations. In the brooding species Madracis decactis, thispattern was even stronger when one considered distance between the platforms and the perimeters of theFGB, particularly the nearest FGB, suggesting that the FGB may be a source of larvae for platform populations.AFLPOP analyses showed that the degree of self-allocation to home sites also increased with inter-sitedistance. Cross allocations between sites dropped significantly and exponentially in all species within onlyone to several kms of the FGB. Madracis decactis, a brooder with extended larval release periods and near-immediate settlement competence, showed greater affinity to the FGB with distance than Diploria strigosa, abroadcaster. This brooder appears to be more effective at colonizing small, nearby target sites and expandingits geographic range at the meso-scale. The low degree of genetic affinity exhibited by all species on theplatforms may be attributed to genetic drift/founder effect or relatively small samples sizes, although totalpopulations were sampled. In general, genetic affinity decreased with inter-site distance. Young coralpopulations are highly differentiated at the meso-scale during early stages of community succession,implying that much time and repeated colonization of patchy habitats around larger potential larval sourceswill be required before genetic equilibrium or homogeneity is reached.

© 2008 Elsevier B.V. All rights reserved.

1. Introduction

Offshore oil and gas platforms in the northern Gulf of Mexico(GOM), currently 3,600 in number (Dauterive, 2000; Sammarco et al.,2004; G. Boland, pers. comm.), have served as substrate for thecolonization of numerous marine organisms since the 1940s. This

e Consortium (LUMCON), 8124

son), [email protected]

l rights reserved.

region is generally characterized by terrigenous, sandy muds withlittle habitat diversity (Rezak et al., 1983; Scarborough-Bull, 1989). Theplatforms extend up through the euphotic zone, providing hardsubstrate in open water (Shinn, 1974) that would otherwise beunavailable to such marine organisms. A 200 ft. platform jacket canprovide several acres of hard substrate which in turn can supportalgae, barnacles, mussels, and other sessile, epibenthic invertebrates(Driessen, 1989; Scarborough-Bull, 1989). Scleractinian corals, includ-ing Diploria spp., Porites astreoides, Madracis decactis, M. asperula, andMillepora alcicornis, have been found on various platforms in the GOM(Bright et al., 1991; Sammarco et al., 2004).

mailto:[email protected]

mailto:[email protected]

http://dx.doi.org/10.1016/j.jembe.2008.07.002

http://www.sciencedirect.com/science/journal/00220981

2 A.D. Atchison et al. / Journal of Experimental Marine Biology and Ecology 365 (2008) 1–12

The Flower Garden Banks (FGB; NOAA Flower Garden BanksNational Marine Sanctuary; Dokken et al., 2002) are located amongthese platforms in the northern GOM (∼180 km SE of Galveston, TX)and are comprised of two banks: the East Bank (27°54′32″ N, 93°36′W) and West Bank (27°52′27″ N, 93°48′47″ W; Bright et al., 1984;Rezak et al., 1983; see Fig. 1). Coral reefs have developed on their caps(Bright et al., 1984; Dokken et al., 1999), rising towithin 18m of the seasurface (Lugo-Fernandez et al., 2001). They are productive (Rezaket al., 1985), healthy reefs, characterized by 24 species of hermatypiccorals (Bright et al., 1984; Dokken et al., 2002), including Madracisdecactis, Diploria strigosa, Montastraea cavernosa, and other species(Gittings et al., 1992; Lugo-Fernandez et al., 2001). These are the onlytwo major coral reefs in the northern Gulf of Mexico. The FGB areseparated by ≥640 km of open ocean from other well-developed coralreefs (Hagman et al., 1998a; Sammarco et al., 2004), with the closestreef system being Lobos-Tuxpan, located 13 km off Cabo Rojo, Mexico(Dokken et al., 2002; Fig. 1). Other banks occur on the northwesternGOM shelf, including Stetson, Sonnier, 32 Fathom, etc., but they are nottrue coral reefs; that is, the benthic substratum is not biogenic(composed of calcium carbonate accreted by corals). Corals arepresent there, but cover is low (Rezak et al., 1985; Lugo-Fernandezet al., 2001; Schmahl, 2003; Sammarco et al., 2004; G. Boland pers.obs.). It is possible that coral populations on these banks could be asource of larvae for platform colonization, but the densities of corals

Fig. 1.Map of the Gulf of Mexico depicting the location of the Flower Garden Banks and its nInset: Map depicting the study sites, including the Flower Garden Banks and thirteen offshopopulations of the target species: (1) HI-A-568A; (2) HI-A-571A; (3) HI-A-376A; (4) HI-A-37

are low there and thus the level of their impact on recruitment is likelyalso low.

We studied three hermatypic scleractinian coral species on oil andgas platforms near the FGB, and on the FGB themselves, to assess thedegree of genetic connectedness among the natural and platformpopulations. Three species were chosen, based upon their abundanceand different reproductive strategies (see Atchison et al., 2006). Thesewere Madracis decactis (Lyman 1859; Pocilloporidae) – a brooder,and two broadcasters - Diploria strigosa (Dana 1846; Faviidae) andMontastraea cavernosa (Linnaeus 1767; Faviidae).Madracis decactis is asimultaneous hermaphrodite that releases planulae from March toDecember, with maximum release occurring from Sept. to Nov.(Vermeij et al., 2003).

There is an abundant literature regarding brooding vs. broad-casting in numerous marine species (e.g., Hellberg et al., 2002; Taylorand Hellberg, 2003; Jones et al., 2005), but wewill limit our commentshere to corals. The fully developed larvae of brooding corals arecompetent to settle in ≥4 hrs (Harrison and Wallace, 1990). Thesespecies typically become reproductive at ages earlier (1-2 yrs.) thanbroadcast spawners (≥4 yrs). Brooders can also planulate monthly,releasing larvae 8-10 times per year (McGuire, 1998), for a period ofdays before and after their peak release date. Broadcast coralspawning is usually seasonal, occurring over only one to severaldays during the summer, being synchronized with a lunar phase and

earest neighboring reefs (e.g., the Lobos-Tuxpan reef system and Campeche Bank reefs).re platforms sampled in this study. Only the FGB and seven platforms possessed coral0A; (5) WC-630A; (6) WC-618; and (7) WC-643A.

3A.D. Atchison et al. / Journal of Experimental Marine Biology and Ecology 365 (2008) 1–12

occurring during a single month (Harrison and Wallace, 1990). Theresultant embryos require 4-7 days for development. Mass spawningin corals occurs on the FGB (Gittings et al., 1992; Hagman et al., 1998a,b) as it does elsewhere in the Atlantic (e.g., de Graaf et al., 1999) andthe Pacific. Spawning in Diploria strigosa and Montastraea cavernosaon the FGB (Bright et al., 1991;Gittings et al., 1992; Hagman et al.,1998a,b) is synchronous with other Caribbean reefs (Lugo-Fernandezet al., 2001).

Some coral studies suggest that brooders are effective at short-distance dispersal, while broadcast spawners are adapted for longer-distance dispersal on and between reefs (see Baums et al., 2005). Onceplanulae are fully developed, however, the potential for long-rangedispersal between brooders and broadcasters are most likelycomparable (Sammarco and Andrews, 1988, 1989; Sammarco, 1994).

Here we examine genetic affinities among populations of thescleractinian corals Madracis decactis, Diploria strigosa, and Montas-traea cavernosa, respectively – one brooder and two broadcasters – onthe Flower Garden Banks and on a set of offshore platformssurrounding them in the first phase of an ongoing study. Theseplatforms represent the proximal platforms around the FGB. We willexamine the degree of connectedness between coral populations onthe natural reefs and the surrounding platforms, the degree of self-allocation, and the degree of connectivity between neighboringplatforms themselves. We will also examine effectiveness of coloniza-tion of the platforms by the brooding vs. broadcasting species, as itpertains to an island-hopping strategy of colonization (see MacArthurand Wilson, 1967; Maltagliati et al., 2002; Atchison, 2005; Baumset al., 2005). Data regarding Madracis decactis and Diploria strigosawere preliminarily analyzed by Atchison et al. (2006). Here, we

Fig. 2. Map of the Gulf of Mexico, depicting examples of general currents known to exist. NoGarden Banks and the northerly jet extending from Alacran to the Flower Garden Banks, in

expand those analyses to include Montastraea cavernosa. Moreimportantly, we apply a more conservative statistical analyticalapproach to examine the data, based on the results of extensivesimulations, and examine the data from a more detailed geographicperspective.

2. Materials and methods

2.1. Study Site

Thirteen oil and gas platforms off the Louisiana and Texas coastwere surveyed for hermatypic scleractinian corals. The target species –Madracis decactis, Diploria strigosa, and Montastraea cavernosa - werefound on seven of these, whichwill from this point on be referred to asPlatforms #1 to #7, west to east (see Fig. 1 legend for details). Allplatforms considered here had been deployed for 15-26 years – aplatform age range known to be associated with the development ofsubstantial adult coral populations (Sammarco et al., 2004).

Tissue was collected from all coral colonies present on theplatforms - i.e., total populations between 5 and 37 m depth. At theEast and West FGB, samples of the larger populations were collectedhaphazardly within a 100 m diameter of our mooring buoy by teamsof divers. The platforms fell within an elongated ellipse, inclusive ofthe FGB - extending from 10 to 15 kmwest of theW-FGB to 50 km eastof E-FGB and 10-15 km north. The sampling regime was skewed to theeast because of the known prevailing westerly currents in the region(Sturges, 1993; Oey, 1995; Lugo-Fernandez, 1998; Lugo-Fernandezet al., 2001; see Fig. 2). Due to problems associated with transfer ofsamples over international borders, it was not possible to sample

te the general westerly current across the continental shelf in the vicinity of the Flowerdicating potential larval dispersal vectors.

Table 1Sequences of the adapters and primers used in the AFLP protocol

Name Sequence

Adapters EcoRI EcoF 5′-CTCGTAGACTGCGTACCEcoR 5′-AATTGGTACGCAGTCTAC

Adapters MseI MseF 5′-GACGATGAGTCCTGAGMseR 5′-TACTCAGGACTCAT

Pre-selective primer EcoRI A 5′-GACTGCGTACC AATTC APre-selective primer MseI C 5′-GATGAGTCCTGAG TAA CSelective primers(Set 1)

EcoRI 5′-GACTGCGTACCAATTC ACTMseI 5′-GATGAGTCCTGAGTAA CAG

Selective primers(Set 2)

EcoRI 5′-GACTGCGTACCAATTC ACCMseI 5′-GATGAGTCCTGAGTAA CTT

Pre-selective and selective nucleotides are indicated in bold.


corals from other potential source populations (Tampico, Bay ofCampeche, Alacran region, etc.).

Teams of SCUBA divers collected tissue samples, each two cm2 insize, from the growing edge of adult colonies of the three targetspecies using small hammers and chisels. Total population sizes on theplatforms were small. For this reason, sample sizes (provided intables) represent the total populations of these species on eachplatform within this depth range. Once on shipboard, the sampleswere sealed in small plastic freezer bags containing SED buffer(saturated NaCl, 250 mM EDTA, pH 7.5, 20% DMSO) to preserve theDNA. This allowed the samples to potentially be stored at roomtemperature, eliminating the need for storage in liquid nitrogen. Bagswere then placed into ice chests containing additional SED buffer andreturned to the laboratory.

2.2. Amplified Fragment Length Polymorphism (AFLP)

AFLPs were used to conduct molecular genetic analyses. It is aDNA-fingerprinting technique (Sunnucks, 2000) that detects poly-morphisms based upon the selective PCR amplification of a subset ofthe massive population of restriction fragments generated using twodifferent restriction enzymes (Vos et al., 1995; Mueller and Wolfen-barger, 1999). AFLPs tend to be highly polymorphic but are not co-dominantly expressed. Bensch and Åkesson (2005) have noted thatAFLPs are commonly used in studies of crop species and economicallyimportant species but have had only a modest impact in animalstudies. This is surprising, since many studies are limited by theavailability of polymorphic markers, something AFLPs can provide.AFLPs have been used successfully to determine migration rates (Heet al., 2004), species boundaries (Lopez et al., 1999; Fukami et al.,2004), and parental contributions to populations (VanToai et al., 1997).Though AFLPs are less than ideal for many population geneticapplications (Sunnucks, 2000), they are well suited for populationassignment or allocation studies (Blears et al., 1998; Mueller andWolfenbarger,1999; Sammarco and Brazeau, 2001) where the numberof polymorphic loci is more important than allelic diversity(Bernatchez and Duchesne, 2000).

One weakness of AFLPs is that some variation detected may not bederived from the target organism (Sunnucks, 2000). In corals, themajor concern is with symbiotic zooxanthellae. Here we usezooxanthella-specific PCR primers to confirm for each sample thatzooxanthellar DNA contamination are at levels far below thosenecessary for AFLP (5-10 pg of zooxanthellar DNA in a backgroundof coral DNA; Brazeau et al., 2005). In an earlier work, we assessed theconsequences of zooxanthellar DNA contamination on coral AFLPs(Brazeau et al., 2005). Using zooxanthella DNA spiked into coral DNApreparations the lower limit of zooxanthellar DNA detected by PCRprimers specific for zooxanthellae was five to six orders of magnitude(4.8 pg) less than the amount of coral DNA present in each sample(minimally 50-100 ng). Attempts to generate AFLP bands using purezooxanthella DNA at amounts ranging from 0.48 to 480 pg wereunsuccessful. Thus, coral samples that show no detectable zoox-anthellae PCR product will have too little zooxanthella DNA tocontribute any substantial numbers of AFLP bands.

2.3. Preparation of Coral Tissue Lysates

DNA was isolated by macerating samples lightly in SED buffer andspinning at 16Xg for 5 min to pelletize the zooxanthellae from thehomogenate. The DNA was then purified using the Wizard® SVGenomic DNA Purification System (Promega Corporation, Madison,WI), following manufacturer's instructions for animal tissue. Allsamples were checked for zooxanthellae DNA contamination usingthe PCR techniques described in Brazeau et al. (2005).

AFLPs, like other similar molecular genetic techniques, generate asubset of markers from a large population of markers. Of the subset

obtained from a given AFLP experiment, a portion is often sensitive tospecific reaction conditions. Thus, extra caution is required inprocessing samples through all procedural steps to maximizerepeatability of results. Here, we processed samples in large lotscontaining members from all populations to uniformly distribute anyerror potentially introduced by reaction conditions between popula-tions in an unbiased fashion. Also, all PCR reactions were done usingone machine and the same thermal cycle profiles.

2.3.1. Genomic DNA digestion and adapter ligationA restriction-ligation “master mix” was prepared using the

following reagents (measures are per sample): 1.1 μl T4 DNA ligase10X buffer (30 mM Tris-HCl, pH 7.8/10 mM MgCl2/10 mM dithio-threitol (DTT)/1 mM ATP), 1.1 μl of 0.5 M NaCl, 0.5 μl bovine serumalbumin (BSA; 1 mg/ml), 1.0 μl Mse I adapters (50 μM), 1.0 μl EcoRIadapter (5 μM), 0.25 μl Mse I (4U/μl; New England BioLabs, Beverly,MA), 0.25 μl of EcoRI (20U/μL; New England BioLabs), and 0.33 μl of T4ligase (3 U/μl; 10 mM Tris-HCl, pH 7.0/50 mM KCl/1 mM DTT/0.1 mMEDTA/50% glycerol). Sequences for the Mse I and EcoRI adapters, andPCR primers, are listed in Table 1. To each new 1.7 ml tube, 5.5 μl of therestriction-ligation mixture plus 5.5 μl (500 ng genomic DNA) of thepurified genomic was added, centrifuged for 15 s, and incubated atroom temperature overnight. At the end of the restriction-ligationreaction, 189 μl of TE buffer (10 mM Tris-HCl, pH 8.0/0.1 mM EDTA)was added (10-fold dilution), serving as the template for the next-step, pre-selective amplification.

2.3.2. Pre-selective (PS) AmplificationA second “master mix” was made for pre-selection (PS) amplifica-

tion, using the following reagents (per sample measure given): 8.1 μlof nuclease-freewater, 2.0 μl of 10X PCR buffer (15mMMg++ in buffer),0.8 μl of 5 mMdNTP's, 2.0 μl of EcoRI PS primer (2.75 μM), 2.0 μl ofMseIPS primer (2.75 μM), and 0.1 μl of Thermostable (Taq) DNA polymerase(5U/μl), for a total volume of 15.0 μl. To each 0.5 ml tube, 15 μl of thepre-selective amplification master mix was added, and 5 μl of each ofthe diluted restriction ligation reactions samples were vortexed andcentrifuged for 15 s. Amplificationwas performed using a 2-min initialincubation at 72 °C, followed by 20 cycles of 20 s denaturation at 94 °C,30 s annealing at 56 °C, and 2 min extension at 72 °C. Last steps were2 min final extension at 72 °C, and 30 min final incubation at 60 °C.After the cycling was completed, 180 μl of TE buffer was added to eachtube, which consisted of the templates for the final step, selectiveamplification.

2.3.3. Selective AmplificationIn the final step, a selective amplification “master mix” was made,

containing the following components: 8.1 μl of nuclease-free water,2.0 μl of 10X PCR buffer (with Mg++ at 15 mM), 0.8 μl of 5 mM dNTP's,2.0 μl of EcoRI selective primer (0.46 μM), 2.0 μl of MseI selectiveprimer (2.75 μM), and 0.1 μl of Taq DNA polymerase (5U/μl) for a total


volume of 15.0 μl. To each 0.5 ml micro-centrifuge tube, 5 μl of thediluted pre-selection PCR reaction were added to each correspondingtube, mixed, and centrifuged for 15 s. Samples were placed in thethermocycler, and the cycle profile was performed as indicated: 2 mininitial denaturation at 94 °C, followed by 1 cycle of 20 s denaturationat 94 °C, 30 s annealing at 66 °C, and 2 min extension at 72 °C. Next,there were 9 cycles: 20 s at 94 °C, initial 30 s at 66 °C (reduced 1 °C/cycle), and 2 min at 72 °C. The final step consisted of 20 cycles: 20 s at94 °C, 30 s at 56 °C, and 2 min at 72 °C, followed by a 30 min finalincubation at 60 °C.

The products of the selective PCR were separated on a 5%polyacrylamide (sequencing) gel. Banding patterns were analyzedusing Kodak Digital Image Analysis software (Eastman Kodak Co.Scientific Imaging Systems; Bonin et al., 2004). Bandswere assigned tobins based upon 20 bp size intervals. The Selective PCR reactions wererepeated three times for each sample. These “repeat” reactions wererun on different days with populations mixed in each run. Bands wereconsidered present if they appeared in two of the three runs;conversely, bands were scored as absent if two out of the threereactions yielded no band. Of the bands included in the study, N90%yielded the same result in all three PCR runs. These inclusion criteriahelped to exclude bands that were overly sensitive to reactionconditions.

2.4. Statistical Analyses

Two statistical analyses were used to assess population differ-entiation: Analysis of Molecular Variance (AMOVA; Excoffier et al.,1992), and AFLPOP population allocation analysis (v. 1.0; Duchesneand Bernatchez, 2002). AMOVA estimated population differentiationbased upon presence-absence data and generated ΦST values for allpossible pair-wise comparisons between populations by species. Abootstrap of 1000 iterationswas performed to estimate p values -ΦST -for each species for population statistics. Significance levels (p-values)were corrected using a Bonferroni correction for multiple tests(Bonferroni, 1936). No clones were encountered during the study;thus, adjustments for clonality were not necessary.

AFLPOP uses AFLP presence/absence data to calculate log-like-lihood values for any individual's membership in a referencepopulation, based upon their banding patterns. Each individual isallocated to the population showing the highest likelihood for thatgenotype (Duchesne and Bernatchez, 2002; He et al., 2004). Popula-tion assignment tests for individuals based on genetic differentiationamong populations have provided the most promising statisticalmethods used to estimate contemporary long-distance dispersal (Heet al., 2004). When the individual is assigned to a population differentfrom the site fromwhich it was collected, it is interpreted as evidenceof dispersal. Onemajor advantage of using assignmentmethods is thatpopulations do not have to be sampled exhaustively (He et al., 2004).In an AFLPOP simulation, an individual was chosen randomly from theentire population, population marker frequencies were then calcu-lated without that individual, and then the individual was assigned tothe new data set. For each of simulation run, this was repeated 500times. Average assignments to a given site were subsequentlycalculated as a percent value based on 10 repeats of these 500iterations.

The AFLPOP program allows the user to set a log-likelihoodthreshold for each assignment. A log-likelihood threshold set to 0.0will result in the assignment of a colony to the population with thehighest likelihood value. Atchison (2005) found that this may yieldpotentially misleading results, since there may be more than onepopulation with nearly equal likelihood values. Here, we performedsimulations using 1.0 as the comparative log-likelihood threshold.This approach is more conservative and removes potentiallyspurious groupings. With the threshold set to 1.0, assignment of acolony to a population was not made unless the probability of the

given assignment was 10 times more likely than the next mostprobable assignment. If this threshold was not met, the individualwas not assigned to any population; in that case, it was categorizedas “Criteria for Allocation Not Met” (CANM). This does not imply thatthe sample could not be assigned to any population with highprobability. It only means that there may have been at least twopopulations with nearly equivalent probabilities of assignment. Itcould also mean that the individual fits none of the populationswell, in which case it could have been derived from an outsidepopulation (see Atchison et al., 2006 for a preliminary report). Inthis study, we will focus primarily on those cases where clearassignments have been made – including self-allocations to sites oforigin, and cross-population allocations.

The genetic variables derived above were further analyzed viaadditional parametric and non-parametric statistical techniques. Thesoftware used for suchwas BIOMStat V. 3.2 and SigmaPlot V.10.0. Datawere transformed prior to analysis for normalization purposes.

3. Results

A total of 291 tissue samples were examined. Two AFLP primersets were used in the analysis. For Montastraea cavernosa, tissuefrom 60 colonies yielded a total of 40 polymorphic markers. In Di-ploria strigosa, 64 colonies produced a total of 31 polymorphicmarkers and in Madracis decactis, 167 colonies produced 57polymorphic markers. In some cases, only a small number ofcolonies per species were found on individual platforms. We haveincluded all of the populations in the analytical results, however,because low “sample sizes” on some of the platforms are directlyrepresentative of low population sizes at these sites, not decreasedsampling effort; we sampled the entire population at the depthsconcerned. Nonetheless, we have noted instances where smallpopulation sizes may constrain interpretation of the results.

3.1. Madracis decactis (brooder)

Madracis decactis was found on all seven study platforms plus theE- andW-FGB. Since this species was the most abundant encountered,it provided the most information regarding genetic affinities. ΦST orgenetic differentiation values generated by the AMOVA analyses forMadracis decactis revealed that the two FGB populations werehomogeneous, yielding the lowest ΦSTs among all populations inthis comparison (Table 2a). Several populations on the platformsexhibited consistent significant genetic differentiation, despite thesmall population sizes involved. These were the western-mostplatform - #1 - and a far eastern platform - #6. In fact, these platformsplus #7, i.e. those most distant from the FGB, exhibited the highestlevels of significant and most consistent genetic differentiation fromthe other sites. Among the populations with larger sample sizes, thehighest ΦST values were observed between two of the mostgeographically separated populations - #2 and #7.

In Madracis decactis, ΦST increased with inter-site distance, beingdescribed by a significant 2° polynomial regression (Fig. 3). Mostpopulations being considered in a pair-wise manner were signifi-cantly different from each other by AMOVA. It was also clear, however,that all those populations which were not significantly different hadΦST values of ≤0.02 and were clustered at distances of ≤55 km fromeach other (Fig. 3; Table 2a).

We also examined genetic differentiation patterns using the FGB asa reference point forMadracis decactis.ΦST was significantly positivelyassociated with distance from the W-FGB (Fig. 4), and even moreclearly with distance from the E-FGB, as well as with distance from thenearest FGB perimeter. When comparing non-significantly differentpopulations, all populations with homogenous ΦSTs occurred gen-erally at ≤30 km from either FGB perimeter – but up to 55 km in thecase of the W-FGB (Fig. 4).

Table 2Madracis decactis: (a) Assessment of genetic differentiation among populations of the coral Madracis decactis using Analysis of Molecular Variance (AMOVA)

Reference Population

Platform/Site 1 2 W-FGB 3 E-FGB 4 5 6 7

n 8 15 25 20 18 19 28 7 27

(a) AMOVACompared to↓1 Pb0.001 Pb0.001 Pb0.001 Pb0.001 Pb0.001 Pb0.001 Pb0.001 Pb0.0012 0.1285 Pb0.05 Pb0.001 Pb0.05 0.364 Pb0.001 Pb0.001 Pb0.001W-FGB 0.0763 0.0421 Pb0.05 0.556 Pb0.05 0.144 Pb0.05 Pb0.053 0.0740 0.0730 0.0129 0.244 Pb0.001 0.107 Pb0.001 Pb0.05E-FGB 0.0828 0.0250 −0.0057 0.0116 0.303 0.436 Pb0.001 Pb0.0014 0.1133 0.0035 0.0189 0.0348 0.0059 0.117 Pb0.001 Pb0.0015 0.0674 0.0392 0.0176 0.0198 0.0012 0.0178 Pb0.001 Pb0.056 0.0650 0.1140 0.0771 0.0967 0.1060 0.0951 0.1097 Pb0.0017 0.0987 0.1266 0.0476 0.0148 0.0485 0.0858 0.0300 0.1385Overall ΦST=0.073

(b) AFLPOPAllocated to↓1 94.3% 0.0% 0.0% 0.2% 0.0% 0.0% 0.0% 0.0% 0.1%2 0.0% 62.4% 0.2% 0.1% 0.2% 1.2% 0.3% 0.0% 0.0%W-FGB 0.0% 0.3% 32.2% 0.3% 0.3% 0.4% 0.6% 0.0% 0.3%3 0.0% 0.1% 0.4% 38.5% 0.3% 0.4% 0.4% 0.0% 0.4%E-FGB 0.0% 0.2% 0.8% 0.3% 26.2% 0.5% 0.5% 0.0% 0.1%4 0.0% 0.5% 0.6% 0.4% 0.4% 37.8% 0.6% 0.0% 0.0%5 0.0% 0.2% 0.3% 0.2% 0.4% 0.4% 30.9% 0.0% 0.2%6 0.0% 0.0% 0.2% 0.1% 0.2% 0.0% 0.0% 97.1% 0.0%7 0.1% 0.0% 0.3% 1.2% 0.4% 0.2% 1.0% 0.0% 50.4%CANM 5.6% 36.3% 65.0% 58.7% 71.8% 59.1% 65.7% 2.9% 48.5%Totals 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% 100.0%

OverallΦST value for the 9 populations was 0.073. Upper diagonal gives significance levels for genetic differentiation based upon analysis of 1000 iterations. Significant differences inpair-wise comparisons shown in bold, as determined using the Bonferroni correction.ΦST values shown on the lower diagonal. n = sample size. (b) AFLPOP assignment of samples totheir reference home population vs. other sites. Data analyzed via AFLPOP of band frequencies, using a minimum log-likelihood difference of 1.0. Assignments based upon averages of10 simulations of 500 randomly chosen individuals. Levels of allocation back to reference populations (populations of origin) shown on diagonal in bold. CANM = Criteria forAllocation Not Met.


Using the requirement of a minimum log-1 (10-fold difference)threshold for population allocation, most of the AFLPOP assignmentsin Madracis decactis were allocated back to their population of origin

Fig. 3. ΦST or genetic differentiation between populations of the coral Madracis decactisfound on all sites considered here – both offshore oil/gas platforms and the East andWest Flower Garden Banks. Presented as a function of distance between the sites. Dataanalyzed by AMOVA. ● = significantly different populations; ○ = not significantlydifferent. pb0.001, 2° polynomial regression, Y=3.409+0.332X−0.002X2. Y-datatransformed by arcsine for normalization purposes.

(self-assignment; Table 2b). Populations from most platforms and theFGBs showed cross-assignments to many other populations, but theproportions of cross-assignments involving platforms were generallytwo orders of magnitude smaller than the self-assignments. Thewestern-most platform (#1) and one of the eastern-most platforms(#6), yielded high levels of self-assignment (94% and 97%, respec-tively); population sizes here were small (Table 2b). These resultswere consistent with the above AMOVA results.

AFLPOP analytical results supported the concept that the geneticuniqueness of the Madracis decactis populations became increasinglydistinct with increased remoteness from the FGBs (Fig. 5; Table 2b).These results paralleled those yielded by AMOVA and patterns derivedin ΦST. Levels of self-allocation of colonies at a given site increasedsignificantly with distance from the E-FGB (Fig. 5a), and the strengthof this relationship was stronger when considering distance to thenearest FGB perimeter (Fig. 5b). No relationship emerged betweenself-allocation of colonies to sites and distance from the W-FGB whenused as a reference point.

The lowest percentages of self-allocation (26% and 32%) occurredin Madracis decactis populations at the E- and W-FGB, respectively.In the center of the study region, cross-assignments were leastcommon between the FGB and some of those platforms at thefurthest sites (Table 2b). Allocation from the E- to the W-FGB washigher than to any platform. In most cases, the majority of coloniesfrom the FGB were identified as “CANM” (Criteria for Allocation NotMet). There was no correlation between platform age and level ofself-allocation in Madracis decactis (Pearson's product-momentcorrelation analysis, r=−0.654, pN0.05).

AFLPOP analyses also yielded information on cross-allocationbetween Madracis decactis populations at all sites. When all possiblepair-wise comparisons were made between sites, including self- and

Fig. 4.ΦST or genetic differentiation between populations of the coralMadracis decactis found on sites considered in this study, using the perimeter of the Flower Garden Banks (FGB)as a geographic reference point. Presented as a function of distance between the site and the perimeter of (a) E-FGB, (b) W-FGB, and (c) nearest FGB. Data transformed by arcsine fornormalization purposes and analyzed by AMOVA. Number of multiple/over-plotted points shown in parentheses. ● = significantly different populations; ○ = not significantlydifferent. (a) pb0.05, Kendall's rank correlation analysis, T=0.479; and pb0.05, Spearman's rank correlation coefficient, r=0.661. (b) pb0.05, Model-II linear regression analysis,Y=0.457+0.266 X. (c) pb0.001, ANOVA; pb0.05, Model-II linear regression analysis, Y=2.432+0.272 X; pb0.05, correlation analysis, r=0.692.


cross-allocations, a clear pattern of high self-allocation emerged butwith a very low level of cross-site allocation - particularly on theplatforms around the FGB, irrespective of distance from them (Fig. 6a).As distance between sites increased, the level of cross-allocation to asite dropped significantly and highly dramatically, even at distances aslow as 1 km. The pattern of this decrease followed a sharp hyperbolic

Fig. 5. Self-allocation in Madracis decactis colonies; proportion assigned back to their origthreshold of 1.0. Proportions plotted against distance between the site and perimeter of tnormalization purposes. Self-allocation increases significantly with distance from the E-FGB apb0.05, Sum of squared difference in ranks. (b) pb0.05, Kendall's Rank Correlation Analyssquared difference in ranks.

decay curve and was mimicked in the other two species examined(Fig. 6b,c; see below). When the self-allocation data were excludedfrom the analysis, the degree of cross-allocation between populationsdecreased significantly in a linear fashion with distance between sites(Fig. 7), again indicating an increasing amount of genetic isolationbetween populations with distance.

inal sites when analyzed by AFLPOP, using an assignment level with a log-likelihoodhe (a) East Flower Garden Bank and (b) nearest FGB. Data transformed by arcsine fornd nearest FGB perimeter. (a) pb0.05, Spearman's rank correlation coefficient, r=0.683;is, T=0.535; pb0.05, Spearman's Rank Correlation Analysis, r=0.686; pb0.05, Sum of

Fig. 6. (a) Allocation ofMadracis decactis colonies between populations at all study sites, as determined by AFLPOP. A log-linear threshold of 1.0 (highly conservative) used for analysis.Allocation of colonies to populations at other sites vs. inter-site distance. All populations included, included self-assignments to sites of origin. Significant hyperbolic decay; pb0.001,2-parameter hyperbolic regression analysis, Y=(43.417⁎0.382) / (0.382+X). (b) Same, but in Diploria strigosa. Significant hyperbolic decay; pb0.001, 3-parameter hyperbolic decayregression analysis, Y=0.946+(52.507⁎0.103) / (0.103+X). (c) Same, but in Montastraea cavernosa. Significant hyperbolic decay; pb0.01, 2-parameter hyperbolic decay regressionanalysis, Y=(47.383⁎2.740) / (2.740+X).


3.2. Diploria strigosa (broadcaster)

Diploria strigosa, a broadcaster, occurred on the E- andW-FGB, andPlatforms #3, #5, and #7. AMOVA analyses identified variabledifferentiation among populations, with ΦST ranging from 0.012 to0.198. The two FGB populations were not significantly different fromeach other (Table 3a). The E-FGB population was homogeneous with

Fig. 7. Allocation of Madracis decactis colonies between populations at study sites, asdetermined by AFLPOP. A log-linear threshold of 1.0 (highly conservative) used in thisanalysis. Allocation of colonies to populations at other sites vs. inter-site distance. Self-allocation data have been excluded. Significant negative correlation, pb0.05, correlationanalysis, r=0.266; pb0.05, Model-II linear regression analysis, Y=2.768−0.018 X.

all other small platform populations. The small population of Platform#3 was significantly different from the W-FGB.

In comparison to Madracis decactis, Diploria strigosa exhibitedquite a different relationship between ΦST and inter-site distancebetween platforms. ΦST was low at very small inter-site distances(Table 3a). As geographic distance increased, however, ΦST in-creased, peaking at an inter-site distance of 25-35 km. Beyondthat, it decreased significantly out to a distance of ∼65 km in asignificant curvilinear manner (pb0.01, 2° polynomial regression,Y=3.144+0.869 X−0.013 X2). ΦST in Diploria strigosa exhibited nosignificant trends when considered with respect to distance fromthe E-FGB.

For platform populations, AFLPOP analyses assigned ≥83% of theDiploria strigosa colonies back to their original reference sites, withallocations reaching almost 100% on the two platforms furthest east(Table 3b). The E- and W-FGB had the highest number of colonies notreadily assignable to a single site. Similar to Madracis decactis, thecentral populations (E-FGB, W-FGB, and Platform #3) had the largestnumber of assignments to neighboring populations. There was alsono correlation between platform age and level of self-allocation inDiploria strigosa (Pearson's product-moment correlation analysis,r=0.909, pN0.05).

Using AFLPOP, we made all possible pair-wise comparisonsregarding the allocation of Diploria strigosa colonies to sites vs.inter-site distance. The resulting pattern was very similar to thatderived for Madracis decactis (Fig. 6a). That is, populations had highlyvariable self-allocation levels, but, in all cases, levels of cross-allocation fell off significantly and exponentially, even at distancesof one to several kms (Fig. 6b), suggesting that the populations arehighly dissimilar at even short distances. One of the differencesbetween D. strigosa and M. decactis is that cross-allocation levels atshort inter-site distances were higher in Madracis than in Diploria.

Table 4Montastraea cavernosa: (a) Assessment of genetic differentiation among populations ofthe coral Montastraea cavernosa using Analysis of Molecular Variance (AMOVA)


W-FGB E-FGB Platf.7

ni 27 25 8

(a) AMOVACompared to↓W-FGB Pb0.001 0.995E-FGB 0.0881 Pb0.001Platf. 7 0.0101 0.1154Overall ΦST=0.080

(b) AFLPOPAllocated to↓W-FGB 68.2% 0.2% 0.9%E-FGB 1.5% 85.2% 0.2%Platf. 7 1.2% 0.1% 74.2%CANM 29.1% 14.5% 24.7%Totals 100.0% 100.0% 100.0%

Overall ΦST value for the 3 populations was 0.080. Upper diagonal gives significancelevels for genetic differentiation. Significant differences shown in bold, as determinedusing the Bonferroni correction. ΦST values shown on the lower diagonal. n = samplesize. (b) Results of AFLPOP analyses on same using a minimum log-likelihood differencevalue of 1.0. Levels of allocation back to reference populations (populations of origin)shown on diagonal in bold. CANM = Criteria for Allocation NotMet. See legend of Table 2for additional details.


3.3. Montastraea cavernosa (broadcaster)

Montastraea cavernosa was the least abundant of the three targetspecies on the platforms, yielding low sample numbers. M. cavernosaoccurred only on the E- and W-FGB and Platform #7. Nonetheless,patterns in genetic affinities were still discernible. Unlike the othertwo species, AMOVA identified the two FGB populations as beingsignificantly different from each other (Table 4a). Like Madracisdecactis, however, the small population on Platform #7 was found tobe genetically homogeneous with theW-FGB (ΦST=0.010), but not theE-FGB (ΦST=0.115).

Although the ΦST values were generally low in Montastraeacavernosa, as inter-site distance increased, there was an increase ingenetic differentiation between coral populations (Table 4a; pb0.05,Kendall's rank correlation analysis, T=0.674; pb0.05, Sum ofSquared Difference in Ranks). The variance was high, but theincrease was nonetheless significant. The pattern is similar to thatof Madracis decactis. (The sample sizes encountered here for Mon-tastraea cavernosa prohibit meaningful direct comparisons withresults derived from Diploria strigosa.) AFLPOP analyses allocated≥68% of the Montastraea cavernosa colonies on both FGBs and thesample platform back to their reference populations (Table 4b).Unlike Madracis decactis and D. strigosa, however, Montastraeacavernosa populations on both banks exhibited high levels of self-allocation. The number of platform populations was not sufficient totest for platform-age effects.

As in Madracis decactis and Diploria strigosa, the Montastraeacavernosa populations appear to be highly variablewith respect to self-allocation, although it did reach high levels in some cases (Table 4b).Cross-allocations decrease rapidly, however, becoming significantlydifferent as distances betweenpopulations increase (Fig. 6c; pb0.01, 2-parameter hyperbolic decay regression analysis, Y=[47.383⁎2.740]/

Table 3Diploria strigosa: (a) Assessment of genetic differentiation among populations of thecoral Diploria strigosa using Analysis of Molecular Variance (AMOVA)


W-FGB Platf. E-FGB Platf. Platf.3 5 7

N 29 5 23 4 3

(a) AMOVACompared to↓W-FGB Pb0.001 0.1119 0.3407 0.460Platf. 3 0.0629 0.1249 Pb0.001 0.3926E-FGB 0.0127 0.0508 0.3846 0.3956Platf. 5 0.0222 0.1657 0.0183 Pb0.001Platf. 7 0.0686 0.0121 0.0380 0.1975Overall ΦST=0.073

(b) AFLPOPAllocated to↓W-FGB 34.1% 0.1% 1.2% 0.0% 0.0%Platf. 3 0.4% 83.3% 0.9% 0.0% 0.0%E-FGB 1.4% 0.1% 32.7% 0.0% 0.0%Platf. 5 0.2% 0.0% 0.1% 100.0% 0.0%Platf. 7 0.1% 0.0% 0.2% 0.0% 99.8%CANM 63.8% 16.5% 64.9% 0.0% 0.2%Totals 100.0% 100.0% 100.0% 100.0% 100.0%

Overall ΦST value for the 5 populations was 0.045. Upper diagonal gives significancelevels for genetic differentiation. Significant differences shown in bold, as determinedusing the Bonferroni correction. ΦST values shown on the lower diagonal. n = samplesize. (b) Results of AFLPOP analyses on same using a minimum log-likelihood differencevalue of 1.0. Levels of allocation back to reference populations (populations of origin)shown on diagonal in bold. CANM = Criteria for Allocation NotMet. See legend of Table 2for additional details.

[2.740+X]. The exponential decay is less dramatic than in the other twospecies, but this is most likely an artifact of sample size.

4. Discussion

Homogeneity in Madracis decactis populations between the twoFGBs is not surprising, since these reefs have co-existed in closeproximity (b12 km) for 16,000-18,000 yrs (Rezak et al., 1985). Thisgenetic connectivity also extended to theMadracis populations within10 km to the west (Platform #2) and 25 km to the east (Platform #4).There was less self-assignment in populations of the more centralsites, near and including the FGB's, than at the outermost sites.Assignment to the “CANM” (Criteria for Allocation Not Met) categorydoes not, of course, necessarily imply lack of any assignment for acolony; rather, since we used a threshold of log 1.0, it implies that thelikelihood of a colony “belonging to” one population and only onepopulation at a probability 10-fold higher than the next most probablepopulationwas not met. Generally, this meant that the population hada nearly equal probability (less than a 10-fold difference) of beingassigned to more than one population.

If one considers the genetic divergence between Madracis decactispopulations vs. distance to the FGB, the patterns in ΦST suggest thatpopulations on the FGBs are indeed seeding the surroundingplatforms – or at least are more frequently seeding those in theimmediate vicinity. Also, the increasing dissimilarity with distancecould be a sign of founder effect; that is, fewer and fewer larvae aresuccessfully recruiting on the platforms with distance from the FGB,carrying with them a smaller and smaller sample of the gene pool oforigin and presenting an increasingly skewed representation of thesource population. Thus, the genetic signature appears to be more andmore different with distance.

The fact that genetic differentiation increased in Madracis decactisbetween all sites as geographic distance between them increased,suggests that gene flow is restricted by inter-site distance. Montas-traea cavernosa populations exhibited a similar pattern of a general,low-level increase in genetic differentiation with geographic distance.


In Madracis decactis, the distance beyond which all populationcomparisons were significantly different was 55 km, indicating thatgene flow is probably severely restricted beyond this distance. Thepattern of increasing genetic differentiation with geographic distanceinMontastraea cavernosawas also significant, but it is difficult to makein-depth interpretations of the data with the limited sample sizesencountered here.

This concept is reinforced in Madracis decactis by the AFLPOP self-allocation data. Increasing levels of self-allocation with distance fromthe FGB indicates that the populations become more distinct withdistance from these reefs. It also implies that the FGB may well be asource of larvae for colonization of these structures over the past30 yrs, but againwith strong founder effects occurring due to very lowlevels of successful recruitment on an annual basis. Data that we havecollected in a sister study demonstrate that annual coral settlementrates on the platforms are extremely low – and that M. decactis is oneof the only two species that settled there (Sammarco and Atchison,2002; Sammarco et al., 2005).

Patterns in inter-site allocations in Madracis decactis along withself-allocations yielded by AFLPOP analyses reinforce this conclusion.Clearly, differences in the genetic structure of the platform popula-tions emerge immediately with any distance between sites.The observed pattern in Madracis decactis was mimicked directly byDiploria strigosa and, to a lesser degree, byMontastraea cavernosa. Thehigher variance in this last case is surely due to small sample size.

The linear decreasing pattern in inter-site recognition levels withdistance (excluding self-allocation) in Madracis decactis reinforces theconclusion that populations are becoming increasingly distinct asdistance between them increases. This is evident even when overallcross-recognition levels are very low.

Both Madracis decactis and Diploria strigosa populations on thetwo FGBs were genetically homogeneous. Even though platformpopulation sizes were small, the data suggest gene flow between theFGB's and Platform #3, located between the two banks. As inMadracis,the two eastern-most Diploria populations - #5 and #7 – showed noaffinity to the banks or platform #3. Platforms #3, #5, and #7 all hadhigher self-allocation proportions, ranging from 83-100%, perhaps dueto founder effects. Platform age appears to have had little or no effecton the patterns observed; all platforms were exposed to possiblecolonization over a period of 15-26 yrs.

Patterns of cross-site recognition via AFLPOP in Madracis decactisvs. Diploria strigosa contrasted strikingly. Firstly, this response couldhave been produced via simple genetic drift (founder effects). Samplesizes are indeed different between our two key species, andrecruitment to these sites may be sporadic; though it is surprisingthat, given the possible age of these sites (≤27 yrs), the populationstructure didn't show some integration of multiple recruitmentevents. Secondly, small sample sizes could have produced thephenomenon. The observed pattern is significant, however – i.e.non-random, despite sample size. Once again, our samples representthe entire populations at these sites and thus should be a very goodestimate of population differentiation, despite the low probability ofobserving the phenomenon.

A third explanatory hypothesis is that the more dissimilar patternswith seemingly greater rates of differentiation at intermediatedistances between sites may have resulted from Diploria's broad-casting mode of reproduction. Larvae settling at the longer distancesmay actually have been more closely related to the source populationthan those at intermediate distances. Upon spawning, buoyant coraleggs are fertilized externally and require 48-96 hrs before they arecompetent to settle (Harrison and Wallace, 1990). During this time,they have little or no swimming capabilities (primarily used for depthcontrol; Sammarco, 1994; see Stake and Sammarco, 2003) and aresubject to currents for advective dispersal. Assuming a patch of larvaeexperiences an average current velocity in the region is ∼15 cm/sec(see Lugo-Fernandez, 1998, 2006; Lugo-Fernandez et al., 2001) during

the spawning period, it would require a min. of 2.3 days to travel30 km. The observed pattern implies that, under these conditions,these larvae may remain in the water column for 4.7 days prior tosettling in larger numbers. Current velocities vary widely in the Gulf ofMexico, of course, and are highly dependent upon numerous large-scale oceanographic features.

These results plus self-allocation patterns with respect to distanceto the FGB suggest that Diploria strigosa is much more easily isolatedthan its brooding counterpart, Madracis decactis. In this case,differentiation of populations appear to be rapid and complete withindistances of ≤25 km vs. 60 km inMadracis decactis.We believe this is afunction of extreme founder effect due to negligible levels ofrecruitment over ecological time (15-27 yrs).

Unlike Madracis and Diploria, Montastraea populations on the E-and W-FGB were genetically distinct. The small population onPlatform #7 (easternmost) was homogeneous with the W-FGB butgenetically distinct from the E-FGB. This suggests that the coloniesmaking up this population may be derived from the W-FGB (or agenetically similar population).

In an earlier study of Agaricia agaricities (a brooder), we found thatpopulations on Crocker and Conch Reefs, Florida Keys, which occurwithin 20 km of each other, although significantly different, exhibitsome genetic affinity (Brazeau et al., 2005). Similar patterns have beenobserved in Seriatopora hystrix in the Red Sea (Maier et al., 2005),Lophelia pertusa in the deep sea of the NE Atlantic (Le Goff-Vitry etal., 2004), and Balanophyllia europaea in the Mediterranean (Goffredoet al., 2004). Regarding the E-FGB, any connectivity observed here(non-significantly different populations) did not extend to the farwestern (Platform #1; 15 km) or far eastern (#6 and #7; 50-65 km)platforms. With respect to the W-FGB, some connectivity wasobserved to the east, but not to the west. There appears to be somecolonization between platforms, although cross-seeding is probablylow. Potential mechanisms of cross-seeding in the region have beenidentified. That is, it is known that larvae may be carried from the FGBinshore to the north, then to a nearshore westerly current, and then toa southerly coastal jet, carrying them back to the W-FGB region again(Lugo-Fernandez, 1998; Lugo-Fernandez et al., 2001).

Small coral population sizes on the platforms are probably due tothe relative size of the platforms, their structure (comprised of anetwork of large cylinders), their distances from the nearest potentiallarval source, and community age. Small sample sizes do decrease thePower of the Test (Sokal and Rohlf, 1994), but they also make it moredifficult to demonstrate significant differences between populations.Here, any significant differences found between small populationsimplies that these differences are relatively robust. Small sample sizesshould thus be less of a problem for interpreting the data here becausewe are dealing with total population size on the platforms; they maystill pose a problem, however, for extrapolating results to other studies.

It is possible that the generally low levels of genetic connectivityobserved in all species here could be the result of colonization bylarvae frommore distant sources (see Ahlroth et al., 2003). The nearestsource would be the Lobos-Tuxpan Reef System (13 km off Cabo Rojo,640 km SW of the FGB), or the Campeche Bank Reefs (181 km off theYucatan Peninsula), although Lugo-Fernandez (2006) has found thatAlacran may also be considered a potential source. Similarities in coralcommunity structure exist between the FGB and Campeche reefs, andthe FGB are believed to be an extension of communities that occur inthe southern GOM (Bright et al., 1984). Such would most likely be arare event. Unfortunately, we have no genetic data from thesesoutherly regions with which to compare our results. Madracisdecactis and other spp. occur on other platforms throughout thenorthern Gulf of Mexico (Sammarco et al., 2005), and relatedmolecular genetic analyses are being performed to determinepotential connectivity there (Sammarco et al., 2005). Anotherexplanation for the generally low levels of genetic affinity is that asource coral could actually have been present on the FGB but that


recruits onto the platforms represent “sampling” error or bias,resulting in founder effect. Such bias might be expected to decreaseover time with successive recruitment events, if self-seeding did notoccur.

Much debate has surrounded the dispersal capabilities of broodingvs. broadcasting corals (see Sammarco and Andrews, 1989; Sammarco,1994;Ayre andHughes, 2000;Miller andMundy, 2003). It is nowknownthat dispersal among brooding species may be more variable thanbroadcasters (Harii and Kayanne, 2003; Magalon et al., 2005; also seeSammarco et al., 2004). The dispersal potential of larvae is influenced by,amongother things, larval competency period and currents carrying thelarvae (Harii and Kayanne, 2003; also see Olson, 1985). The larvalcompetency period ofMadracis decactismay be greater than that of theother two study species, since its planulae are relatively large (Vermeijet al., 2003) and have a high lipid content, making themwell-suited forlong-distance dispersal (Richmond, 1981).

In this study, the brooderMadracis decactiswas the most abundantandwidely dispersed species. It has beenmore successful at recruitingto nearby sites than either of the broadcasting species - Diploria orMontastraea – which release several orders of magnitude morereproductive propagules than Madracis. The major advantage thatMadracis has over the two broadcasters is its repeated release ofplanulae for a period of 8-10 days per month encompassing majorchanges in tidal amplitude. It does this over 8-10 months of the year(see McGuire, 1998), subjecting the larvae to an array of circulationpatterns, traveling in a multitude of directions throughout the year.These conditions would allow Madracis has taken advantage of theplatforms to successfully extend its geographic range. The newlyestablished populations on the platforms can produce larvae that canin turn settle on neighboring platforms (e.g., Barber et al., 2002), actingas a stabilizing force in the ecosystem in the eventof a disturbance. Thissuggests that the species with the greatest potential for widestdispersal will not necessarily exhibit the highest population densitiesandwidest geographic distribution. Some specieswith crawling larvaeand apparently highly limited dispersal capabilities exhibit a verybroad geographic range (Sammarco, 1994), such as some gastropods(Hoskin, 1997), teredinid bivalves (Calloway and Turner, 1983), andcorals (e.g., Balanophyllia elegans; Gerrodette, 1981). Pocilloporadamicornis, a brooding coral, has among the widest geographic rangeof any known coral, extending from the Red Sea to the eastern Pacific.

Another possible reason why Madracis decactis has been moresuccessful at colonization and exhibited the highest level of geneticaffinity between sites is the nearly-immediate readiness of its larvaeto settle (e.g., 4 hrs after release). This is suggested by the distributionpattern of the broadcasters on the two eastern-most sites and thepattern of ΦST in Diploria strigosa. The probability of encounter oftarget site is dependent upon advection and diffusion, and decreasesas a function of the square of the distance between the source and thetarget (Okubo, 1994).

Diploria andMontastraea are annual mass spawners and have onlya single opportunity per year to disperse their larvae. Dispersal isentirely dependent upon current patterns at that time (see Willis andOliver, 1990; Baums et al., 2006). The genetic heterogeneity of thesesmall islands indicates that these broadcasting species are lesseffective colonizers of new habitats patchily-distributed at themeso-scale. One reason that may explain why these two specieshave different population distributions on the platforms and differentpatterns of genetic differentiation on the two FGBs is that thepopulation sizes for Diploria strigosa on the FGB are 1.2-4.5 timesgreater than those of Montastraea cavernosa (using percent-cover;Dokken et al., 1999).

In summary, the genetic data suggest that the coral populations onthe oil and gas platforms in the northern GOM exhibit genetic affinityto the FGB, which decreases w. distance from them – and from eachother. The brooder, Madracis decactis, with its extended larval releaseperiods and its lack of extended delay for competence to settle, shows

greater affinity to the FGB with distance than Diploria strigosa, abroadcaster. They appear to be more effective at colonizing small,nearby target sites and expanding their geographic range at the meso-scale. By contrast, the broadcaster Diploria strigosa is less effective inexpanding its range, colonizing these relatively short-distance targetsin a more haphazard fashion. It may be more effective at colonizinghabitats ≥60 km from their source. All species examinedhere exhibitedsubstantial founder effect on the platforms, generally increasing withinter-site distance. These young coral populations appear to be muchmore highly differentiated at the meso-scale during these early stagesof community succession (in ecological time) than their older, moreestablished counterparts (see Ayre and Hughes, 2004). This impliesthat a great deal of time may be required, encompassing repeatedlarval colonization of patchy habitats around larger establishedpotential sources, before genetic equilibrium or homogeneity isreached between them (in evolutionary time; Hellberg, 2007).

Acknowledgements

We thank J. Benzie for suggesting the use of AFLPs on coral tissuefor the molecular genetics and L. Rouse and M. Hellberg for helpfuladvice and comments that have greatly strengthened the study andmanuscript. Many thanks to all of the divers who assisted in samplecollection: A. Barker, G. Boland, G. Bunch, C. Chauffe, S. Childs, J.Collins, J. Conklin, J. Culbertson, L. Dauterive, B. DeFelice, T. Dempre, D.Dowdy, T. Empey, C. Gentry, S. Kolian, H. Leedy, L. Logan, D. Marcel, K.de Mutsert, G. Rainey, A. Sebastian, J. Sinclair, G. Speyrer, J. Van derHam, K. Wheelock, and D. Woodall. We also thank the crew of theM/VFling (Gulf Diving, Inc.) and all oil and gas companies who provided usaccess to their offshore platforms: Anadarko Petroleum, Apache Oil,Coastal Oil and Gas, Devon Oil, Dominion Exploration and Production,El Paso Energy, Forcenergy/Forest Oil, Kerr-McGee Oil and Gas, MeritEnergy, Newfield Exploration, Noble Energy, UNOCAL, Samedan Oil,Chevron-Texaco Exploration and Production, and W&T Offshore. ADAwould like to thank her parents for their never-ending support andadvice, without which her achievements here would not have beenpossible. This project was funded by the U.S. Department of theInterior - Minerals Management Service (Cooperative Agreement#32806/TO#36188 to PWS), the Louisiana State University MMS-Coastal Marine Institute, the Louisiana Universities Marine Consor-tium, and the J. Bennett Johnston Science Foundation (to ADA). [SS]

References

Ahlroth, P., Alatalo, R.V., Holopainen, A., Kumpulainen, T., Suhonen, J., 2003. Founderpopulation size and number of source populations enhance colonization success inwaterstriders. Oecologia 137, 617–620.

Atchison, A.D., 2005. Offshore oil and gas platforms as stepping stones for expansion ofcoral communities: A molecular genetic analysis. M.Sc. thesis, Oceanography &Coastal Sciences, Louisiana State University, Baton Rouge, LA. 86 pp.

Atchison, A.D., Sammarco, P.W., Brazeau, D.A., 2006. Genetic affinities of coralpopulations between the Flower Garden Banks and oil and gas platforms in thenorthern Gulf of Mexico: Preliminary data. Proc. 10th Intl. Coral Reef Symp.,Okinawa, Japan, 2004.

Ayre, D.J., Hughes, T.P., 2000. Genotypic diversity and gene flow in brooding andspawning corals along the Great Barrier Reef, Australia. Evolution 54, 1590–1605.

Ayre, D.J., Hughes, T.P., 2004. Climate change, genotypic diversity, and gene flow in reef-building corals. Ecol. Lett. 7, 273–278.

Barber, P.H., Moosa, M.K., Palumbi, S.R., 2002. Rapid recovery of genetic diversity ofstomatopod populations on Krakatau: Temporal and spatial scales of marine larvaldispersal. Proc. R. Soc. Lond. B 269, 1591–1597.

Baums, I.B., Miller, M.W., Hellberg, M.E., 2005. Regionally isolated populations of animperiled Caribbean coral, Acropora palmata. Mol. Ecol. 14, 137–1390.

Baums, I.B., Paris, C.B., Cherubim, L.M., 2006. A bio-oceanographic filter to larvaldispersal in a reef-building coral. Limnol. Oceanogr. 51, 1969–1981.

Bensch, S., Åkesson, M., 2005. Ten years of AFLP in ecology and evolution: why so fewanimals? Mol. Ecol. 14, 2899–2914.

Bernatchez, L., Duchesne, P., 2000. Individual-based genotype analysis in studies ofparentage and population assignment: How many loci, how many alleles? Can. J.Fish. Aquat. Sci. 57, 1–12.

Blears, M.J., de Grandis, S.A., Lee, H., Trevors, J.T., 1998. Amplified fragment lengthpolymorphism (AFLP): A review of the procedure and its applications. J. Ind.Microbiol. Biotechnol. 21, 99–114.


Bonferroni, C.E., 1936. Teoria statistica delle classi e calcolo delle probabilità.Pubblicazioni del R Istituto Superiore di Scienze Economiche e Commerciali diFirenze 8, 3–62.

Bonin, A., Bellemain, E., Bronken Eidesen, P., Pomponon, F., Brochmann, C., Taberlet, P.,2004. How to track and assess genotyping errors in population genetics studies.Molec. Ecol. 13, 3261–3273.

Brazeau, D., Sammarco, P.W., Gleason, D.F., 2005. A multi-locus genetic assignmenttechnique to assess local recruitment of Agaricia agaricites on coral reefs. Mar. Biol.147, 1141–1148.

Bright, T.J., Kraemer, G.P., Minnery, G.A., Viada, S.T., 1984. Hermatypes of the FlowerGarden Banks, northwestern Gulf of Mexico: A comparison to other westernAtlantic reefs. Bull. Mar. Sci. 34, 461–476.

Bright, T.J., Gittings, S.R., Zingula, R., 1991. Occurrence of Atlantic reef corals on offshoreplatforms in the northwestern Gulf of Mexico. Northeast Gulf Sci. 12, 55–60.

Calloway, C.B., Turner, R.D., 1983. Documentation and implications of rapid successivegametogenic cycles and broods in the shipworm Lyrodus floridanus (Bartsch)(Bivalvia, Teredinidae). J. Shellfish Res. 3, 65–69.

Dauterive, L., 2000. Rigs-to-Reefs policy, progress, and perspective. NewOrleans, LA: U.S.Dept. of Interior, Minerals Management Service, Gulf of Mexico OCS Region; OCSStudy MMS 2000-073. 8 pp.

de Graaf, M., Geertjes, G.J., Videler, J.J., 1999. Observations on spawning of scleractiniancorals and other invertebrates on the reefs of Bonaire (Netherlands Antilles,Caribbean). Bull. Mar. Sci. 64, 189–194.

Dokken, Q.R., MacDonald, I.R., Tunnell Jr., J.W., Beaver, C.R., Boland, G.S., Hagman, D.K.,1999. Long-term monitoring at the East and West Flower Garden Banks NationalMarine Sanctuary, 1996-1997. U.S. Dept. of the Interior, Minerals ManagementService, Gulf of Mexico OCS Region, New Orleans, LA. 101 pp.

Dokken, Q.R., MacDonald, I.R., Tunnell Jr., J.W., Wade, T., Beaver, C.R., Childs, S.A.,Withers, K., Bates, T.W., 2002. Long-term monitoring at the East and West FlowerGarden Banks National Marine Sanctuary, 1998-1999. U.S. Dept. of the Interior,Minerals Management Service, Gulf of Mexico OCS Region, NewOrleans, LA. 119 pp.

Driessen, P.K., 1989. Offshore oil platforms: Mini-ecosystems. In: Reggio, V.C. (Ed.),Petroleum structures as artificial reefs: A compendium. Fourth Int. Conf. onArtificial Habitats for Fisheries, Rigs-to-Reefs Special Session, pp. 3–6. Miami, FL. USDept. Interior, Minerals Mgt. Service, New Orleans, OCS Study/MMS 89-0021.

Duchesne, P., Bernatchez, L., 2002. AFLPOP: A computer program for simulated and realpopulation allocation using AFLP markers. Mol. Ecol. Notes 2, 380–383.

Excoffier, L., Smouse, P.E., Quattro, J.M., 1992. Analysis of molecular variation inferredfrom metric distances among DNA haplotypes: Application to human mitochon-drial DNA restriction data. Genetics 131, 479–491.

Fukami, H., Budd, A.F., Levitan, D.R., Jara, J., Kersanach, R., Knowlton, N., 2004. Geo-graphic differences in species boundaries among members of the Montastraeaannularis complex based on molecular and morphological markers. Evol. 58,324–337.

Gerrodette, T., 1981. Dispersal of the solitary coral Balanophyllia elegans by demersalplanular larvae. Ecology 62, 611–619.

Gittings, S.R., Boland, G.B., Deslarzes, K.J.P., Combs, C.L., Holland, B.S., Bright, T.J., 1992.Mass spawning and reproductive viability of reef corals at the East Flower GardenBank, northwest Gulf of Mexico. Bull. Mar. Sci. 51, 420–428.

Goffredo, S., Mezzomonaco, L., Zaccanti, F., 2004. Genetic differentiation amongpopulations of the Mediterranean hermaphroditic brooding coral Balanophylliaeuropaea (Scleractinia: Dendrophylliidae). Mar. Biol. 145, 1075–1083.

Hagman, D.K., Gittings, S.R., Deslarzes, K.J.P., 1998a. Timing, species participation, andenvironmental factors influencing annual mass spawning at the Flower GardenBanks (Northwestern Gulf of Mexico). Gulf Mex. Sci. 2, 170–179.

Hagman, D.K., Gittings, S.R., Vize, P.D., 1998b. Fertilization of broadcast-spawning coralsof the Flower Garden Banks National Marine Sanctuary. Gulf Mex. Sci. 16, 180–187.

Harii, S., Kayanne, H., 2003. Larval dispersal, recruitment, and adult distribution of thebrooding stony octocoral Heliopora coerulea on Ishigaki Island, southwest Japan.Coral Reefs 22, 188–196.

Harrison, P.L., Wallace, C.C., 1990. Reproduction, dispersal, and recruitment ofscleractinian corals. In: Dubinsky, Z. (Ed.), Ecosystems of the World 25: Coralreefs. Elsevier Science Publishing Company Inc., New York, NY, pp. 133–207.

He, T., Krauss, S.L., Lamont, B.B.,Miller, B.P., Enright, N.J., 2004. Long-distance seed dispersalin a metapopulation of Banksia hookeriana inferred from a population allocationanalysis of amplified fragment length polymorphism data. Mol. Ecol. 13, 1099–1109.

Hellberg, M.E., 2007. Footprints on water: The genetic wake of dispersal among reefs.Coral Reefs 26, 463–473.

Hellberg, M.E., Burton, R.S., Neigel, J.E., Palumbi, S.R., 2002. Genetic assessment ofconnectivity among marine populations. Bull. Mar. Sci. 70, 273–290.

Hoskin, M.G., 1997. Effects of contrasting modes of larval development on the geneticstructure of populations of three species of prosobranch gastropods. Mar. Biol. 127,647–656.

Jones, G.P., Planes, S., Thorrold, S.R., 2005. Coral reef fish larvae settle close to home.Curr. Biol. 15, 1314–1318.

Le Goff-Vitry, M.C., Pybus, O.G., Rogers, A.D., 2004. Genetic structure of the deep-seacoral Lophelia pertusa in the northeast Atlantic revealed by microsatellites andinternal transcribed spacer sequences. Mol. Ecol. 13, 537–549.

Lopez, J.V., Kersanach, R., Rehner, S.A., Knowlton, N., 1999. Molecular determination ofspecies boundaries in corals: Genetic analysis of theMontastraea annularis complexusing Amplified Fragment Length Polymorphisms and a microsatellite marker. Biol.Bull. 196, 80–93.

Lugo-Fernandez, A., 1998. Ecological Implications of hydrography and circulation to theFlower Garden Banks, northwest Gulf of Mexico. Gulf Mex. Sci. 1998, 144–160.

Lugo-Fernandez, A., 2006. Travel times of passive drifters from thewestern Caribbean tothe Gulf of Mexico and Florida Bahamas. Gulf Mex. Sci. 2006, 61–67.

Lugo-Fernandez, A., Deslarzes, K.J.P., Price, J.M., Boland, G.S., Morin, M.V., 2001. Inferringprobable dispersal of Flower Garden Banks coral larvae (Gulf of Mexico) usingobserved and simulated drifter trajectories. Cont. Shelf Res. 21, 47–67.

MacArthur, R.H., Wilson, E.O., 1967. The theory of island biogeography. PrincetonUniversity Press, NJ. 203 pp.

Maier, E., Tollrian, R., Rinkevich, B., Nuernberger, B., 2005. Isolation by distance in thescleractinian coral Seriatopora hystrix in the Red Sea. Mar. Biol. 147, 1109–1120.

Maltagliati, F., Belcari, P., Casu, D., Casu, M., Sartor, P., Varqiu, G., Castelli, A., 2002.Allozyme genetic variability and gene flow in Octopus vulgaris (Cephalopoda,Octopodidae) from the Mediterranean Sea. Bull. Mar. Sci. 71, 473–486.

Magalon, H., Adjeroud, M., Veuille, M., 2005. Patterns of genetic variation do notcorrelate with geographical distribution of the reef-building coral Pocilloporameandrina in the South Pacific. Mol. Ecol. 14, 1861–1868.

McGuire, M.P., 1998. Timing of larval release by Porites astreoides in the northern FloridaKeys. Coral Reefs 17, 369–375.

Miller, K., Mundy, C., 2003. Rapid settlement in broadcast spawning corals: Implicationsfor larval dispersal. Coral Reefs 22, 99–106.

Mueller, U.G., Wolfenbarger, L.L., 1999. AFLP genotyping and fingerprinting. Trends Ecol.Evol. 14, 389–394.

Oey, L., 1995. Eddy- and wind-forced shelf circulation. J. Geophys. Res. 100, 8621–8637.Okubo, A., 1994. The role of diffusion and related physical processes in dispersal and

recruitment of marine populations. In: Sammarco, P.W., Heron, M.L. (Eds.), The Bio-Physics of Marine Larval Dispersal, Am. Geophys. Union, Washington, DC, pp. 5–34.

Olson, R.R., 1985. The consequences of short-distance larval dispersal in a sessile marineinvertebrate. Ecol. 66, 30–39.

Rezak, R., Bright, T.J., McGrail, D.W., 1983. Reefs and banks of the northwestern Gulf ofMexico: Their geological, biological, and physical dynamics: Final report. U.S. Dept.of the Interior, Minerals Management Service, Outer Continental Shelf Office, NewOrleans, LA. 502 pp.

Rezak, R., Bright, T.J., McGrail, D.W., 1985. Reefs and banks of the northwestern Gulf ofMexico: Their geological, biological, and physical dynamics. John Wiley and Sons,Inc., New York, NY. 259 pp.

Richmond, R., 1981. Energetic considerations in the dispersal of Pocillopora damicornia(Linnaeus) planulae. Proc. Fourth Int. Coral Reef Symposium. Marine SciencesCenter, University of the Philippines, Quezon City, Philippines, pp. 153–156.

Sammarco, P.W., 1994. Larval dispersal and recruitment processes in Great Barrier Reefcorals: Analysis and synthesis. In: Sammarco, P.W., Heron, M.L. (Eds.), The Bio-Physics of Marine Larval Dispersal. American Geophysical Union, Washington, D.C.,pp. 35–72.

Sammarco, P.W., Andrews, J.C., 1988. Localized dispersal and recruitment in GreatBarrier Reef corals: The Helix experiment. Science 239, 1422–1424.

Sammarco, P.W., Andrews, J.C.,1989. The Helix experiment: Differential localized dispersaland recruitment patterns in Great Barrier Reef corals. Limnol. Oceanogr. 34, 896–912.

Sammarco, P.W., Atchison, A., 2002. Coral communities and recruitment on offshoredrillingplatforms in thenorthernGulf ofMexico: Summary. USDept. InteriorMineralsManagement Service Information Transfer Meeting, New Orleans, LA, Jan. 2002.

Sammarco, P.W., Atchison, A.D., Boland, G.S., 2004. Offshore oil and gas platforms andexpansion of coral communities within the Northern Gulf of Mexico. Mar. Ecol.Prog. Ser. 280, 129–143.

Sammarco, P.W., Atchison, A.D., Boland, G.S., 2005. Geographic expansion and limits ofcorals in the NW Gulf of Mexico: Colonization of Offshore Oil and Gas Platforms.Proc. 10th Intl. Coral Reef Symp., Okinawa, Japan, 2004.

Sammarco, P.W., Brazeau, D.A., 2001. Genetic affinity between corals, including spat, atthree tropical W. Atlantic sites: Where do the larvae go? Proc. 30th Sci. Meeting,Assn. Mar. Labs. Caribb. (AMLC), La Parguera, Puerto Rico, p. 31.

Scarborough-Bull, A., 1989. Some comparisons between communities beneath thepetroleum platforms off California and in the Gulf of Mexico. In: Reggio, V.C. (Ed.),Petroleum structures as artificial reefs: A compendium. Fourth Int. Conf. onArtificial Habitats for Fisheries, Rigs-to-Reefs Special Session, pp. 47–50. Miami, FL.US Dept. Interior, Minerals Mgt. Service, New Orleans, OCS Study MMS 89-0021.

Schmahl, G.P., 2003. Biodiversity associated with topographic features in the north-western Gulf of Mexico. Proc. U.S. Dept. of the Interior, Minerals ManagementService Information Transfer Meeting, Gulf of Mexico OCS Region, Kenner, LA.

Shinn, E.A.,1974. Oil structures as artificial reefs. In: Colunga, L., Stone, R. (Eds.), Proc. Int.Conf. on Artificial Reefs, March 1974. Houston, TX, pp. 91–96. TAMU-SG-74-103.

Sokal, R.R., Rohlf, F.J., 1994. Biometry, 3rd ed. W.H. Freeman and Co., San Francisco.Stake, J.L., Sammarco, P.W., 2003. Effects of pressure on swimming behavior in planula

larvae of the coral Porites astreoides (Coelenterata, Scleractinia). J. Exp. Mar. Biol.Ecol. 288, 181–201.

Sturges, W., 1993. The annual cycle of the Western Boundary Current in the Gulf ofMexico. J. Geophys. Res. 98, 18053–18068.

Sunnucks, P., 2000. Efficient genetic markers for population biology. Trends Ecol. Evol.15, 199–203.

Taylor, M.S., Hellberg, M.E., 2003. Genetic evidence for local retention of pelagic larvaein a Caribbean reef fish. Science 299, 107–109.

VanToai, T.T., Peng, J., St Martin, S.K., 1997. Using AFLP markers to determine thegenomic contribution of parents to populations. Crop Sci. 37, 1370–1374.

Vermeij, M.J.A., Sampayo, E., Broker, K., Bak, R.P.M., 2003. Variation in planulae releaseof closely related coral species. Mar. Ecol. Prog. Ser. 247, 75–84.

Vos, P., Hogers, R., Bleeker, M., Reijans, M., van de Lee, T., Hornes, M., Frijters, A., Pot, J.,Peleman, J., Kuiper, M., Zabeau, M., 1995. AFLP: a new technique for DNAfingerprinting. Nucleic Acids Res. 23, 4407–4414.

Willis, B.L., Oliver, J.K., 1990. Direct tracking of coral larvae: Implications for dispersalstudies of planktonic larvae in topographically complex environments. Ophelia 32,145–162.

Documents

Genetic Connectivity in Corals on the Flower Garden Banks and Surrounding Oil/Gas Platforms, Gulf of Mexico