Ecology and Evolution. 2018;8:9503–9515.  | 9503www.ecolevol.org

Received:13December2017  |  Revised:14April2018  |  Accepted:17July2018DOI:10.1002/ece3.4456

O R I G I N A L R E S E A R C H

Integrating complementary methods to improve diet analysis in fishery- targeted species

Jordan K. Matley1,2  | Gregory E. Maes3,4,5,6  | Floriaan Devloo-Delva7  |  Roger Huerlimann5,6  | Gladys Chua5,6 | Andrew J. Tobin5 | Aaron T. Fisk2 |  Colin A. Simpfendorfer5,6  | Michelle R. Heupel5,8

ThisisanopenaccessarticleunderthetermsoftheCreativeCommonsAttributionLicense,whichpermitsuse,distributionandreproductioninanymedium,providedtheoriginalworkisproperlycited.©2018TheAuthors.Ecology and EvolutionpublishedbyJohnWiley&SonsLtd.

1CenterforMarineandEnvironmentalStudies,UniversityoftheVirginIslands,St.Thomas,VirginIslands2GreatLakesInstituteforEnvironmentalResearch,UniversityofWindsor,Windsor,Ontario,Canada3LaboratoryofBiodiversityandEvolutionaryGenomics,KULeuven,Leuven,Belgium4CenterforHumanGenetics,UZLeuven-GenomicsCore,KULeuven,Leuven,Belgium5CentreforSustainableTropicalFisheriesandAquaculture,CollegeofScienceandEngineering,JamesCookUniversity,Townsville,Queensland,Australia6ComparativeGenomicsCentre,CollegeofScienceandEngineering,JamesCookUniversity,Townsville,Queensland,Australia7OceansandAtmosphere,CSIRO,Hobart,Tasmania,Australia8AustralianInstituteofMarineScience,Townsville,Queensland,Australia

CorrespondenceJordanK.Matley,GreatLakesInstituteforEnvironmentalResearch,UniversityofWindsor,Windsor,Ontario,Canada.Email:jordanmatley@gmail.com

Funding informationAustralianGovernment’sNationalEnvironmentalResearchProgram;AustralianResearchCouncilFutureFellowship,Grant/AwardNumber:#FT100101004;CanadianResearchChairsProgram;JamesCookUniversity’sCollegeofMarineandEnvironmentalSciencesandGraduateResearchSchool;AustralianCoralReefSociety;NaturalSciencesandEngineeringResearchCouncil;QueenslandDepartmentofPrimaryIndustriesandFisheries,Grant/AwardNumber:144482

AbstractDevelopingefficient,reliable,cost-effectivewaystoidentifydietisrequiredtoun-derstandtrophicecologyincomplexecosystemsandimprovefoodwebmodels.Acombinationoftechniques,eachvaryingintheirabilitytoproviderobust,spatiallyandtemporallyexplicitinformationcanbeappliedtoclarifydietdataforecologicalresearch. This study applied an integrative analysis of a fishery-targeted speciesgroup—Plectropomusspp.inthecentralGreatBarrierReef,Australia,bycomparingthreediet-identificationapproaches.Visualstomachcontentanalysisprovidedpooridentificationwith~14%ofstomachssampledresultinginidentificationtofamilyorlower.Amolecularapproachwassuccessfulwithpreyfrom~80%ofstomachsidenti-fiedtogenusorspecies,oftenwithseveraluniquepreyinastomach.Stableisotopemixingmodels utilizing experimentally derived assimilation data, identified similarpreyasthemoleculartechniquebutatbroadertemporalscales,particularlywhenprior diet informationwas incorporated.Overall, Caesionidae and Pomacentridaewerethemostabundantpreyfamilies(>50%preycontribution)forallPlectropomus spp.,highlightingtheimportanceofplanktivorousprey.Lessabundantpreycatego-riesdifferedamongspecies/colorphasesindicatingpossiblenichesegregation.Thisstudyisoneofthefirsttodemonstratetheextentoftaxonomicresolutionprovidedbymoleculartechniques,and,likeotherstudies,illustratesthattemporalinvestiga-tionsofdietarypatternsaremoreaccessibleincombinationwithstableisotopes.Theconsumptionofmainlyplanktivorouspreywithinthisspeciesgrouphas importantimplicationswithincoralreeffoodwebsandprovidescautionaryinformationregard-ingtheeffectsthatchangingresourcescouldhaveinreefecosystems.

K E Y W O R D S

coralreef,coraltrout,diet,fisheries,metabarcoding,next-generationsequencing,Plectropomus,stableisotopes,stomachcontents

9504  |     MATLEY ET AL.

1  | INTRODUC TION

Preyacquisitionisafundamentalbiologicalprocessthatdrivesde-velopment and behavior (e.g., growth, reproduction, foraging) ofindividuals,andcontributestopopulation-levelcharacteristics(e.g.,migration, trophic position, habitat selection). Prey selection andavailability can alsohaveongoing andmultiplicative ecological ef-fectswithinanecosystem(e.g.,trophiccascades;Estesetal.,2011)becauseconsumersareoftenresource-limitedorhaveoverlappingdietary preferences (Ross, 1986; Sale, 1977). Empirical diet datahelpquantifytherelativeimportanceofpreyitemsandcharacterizeecologicalinteractions(e.g.,resourcepartitioning,trophodynamics,competition) thatoccurwithin andamong species (Connell, 1980;Schoener,1974).

Forfishes,thereareseveralwaystoidentifyorquantifydiet.Therearealsoseveralconsiderationsinselectingmethodstochar-acterizediet.Thesevaryonacase-by-casebasisandthegoalsoftheresearch,butareconstrainedbythecostofapproach, lethalvs nonlethal sampling, number of samples/individuals required,necessity of repeat sampling, and/or resolution provided by ap-proach (e.g., temporal or identification resolution). One of themost directmethods is a visual examination of identifiable preyfrom stomach contents, and while this provides a snapshot offeeding(e.g.,hours-days),digestionlimitsidentification,stomachsare often empty, and large sample sizes and lethal sampling aregenerallyrequired(StJohn,1999;Vinson&Budy,2011).However,advancesinmolecularapproachesprovideapotentialalternativeto visual stomach content analysis (Carreon-Martinez, Johnson,Ludsin,&Heath,2011;Leray,Meyer,&Mills,2015).Theabilitytosequenceprey itemsfromdegradedstomachcontentsenhancesdiet data and has the capacity to reduce inefficiencies causedby unidentifiable samples.Nevertheless, thismetabarcoding ap-proachisstilllimitedbythecompletenessofreferencesequencedatabasesandthechoiceofgeneticmarkers(Devloo-Delvaetal.,2018);consequently,priorvalidationisneededfornewlystudiedspecies/systems. Another method to characterize diet is stableisotopeanalysis (e.g.,δ15Nandδ13C),abiogeochemical indicatorof prey assimilation in the tissues of consumers (seeNewsome,Clementz,&Koch,2010 for review).Due todifferentmetabolicprocessing within tissues, the timeline (or turnover) represent-ingpreyassimilationvariesdependingonthetissuesampled.Forexample,Matley, Fisk, Tobin, Heupel, and Simpfendorfer (2016)foundthat50%incorporationtimes(or50%turnover)ofδ15Ninplasma,redbloodcells(RBC),andmuscletissuesofthepredatorycoralreeffishPlectropomus leopardus,were66,88,and126days,respectively.Asδ15Nandδ13Cvalues change fromprey to con-sumer by conserved amounts, the identity (e.g., species, family,habitat) and relative importance of different prey sources canbeestimated (e.g.,mixingmodels;Chiaradia,Forero,McInnes,&Ramírez,2014).Thisapproachrequiresmethodicalsamplingofpo-tentialpreyitems,andstandardizationofassimilationparameters(e.g.,diet-tissuediscriminationfactors)thatmaynotexistforthatspecies; thus, it often requires additional sampling/testing over

othermethods.Stable isotopesalso reflectassimilationpatternsofoftenconfoundingdietarysourcesoverrelativelylongperiodsoftimeandthereforeisarepresentationofbroad-scalepatterns(i.e., doesnotnecessarily identifyexactprey)over the temporalscalepertinenttothetissuesampled.Eachmethodforanalyzingdietincludeslimitations;acombinationofapproacheshasthepo-tentialtoprovidegreaterresolutionandclarityatmultiplespatialandtemporalscales.

The first objective of this study was to compare three di-etary sampling approaches (i.e., visual, genetic, stable isotopeanalysis) to identify the advantages and weaknesses of eachtechnique in isolation and combined. A congeneric group ofcoraltrout(Plectropomusspp.),wereselectedbecausetheyarewidespread mesopredators found throughout the Indo-Pacificwithsignificantfisheryvalue(SadovydeMitchesonetal.,2013).Multiple past studies using visual stomach content analysishave shown that thediet of adultP. leopardus, themost abun-dant Plectropomus species in the Great Barrier Reef MarinePark(GBRMP) inAustralia,consistsof>25preyfamilies,but ismainly comprised of Clupeidae, Pomacentridae, and Labridae(Kingsford,1992;StJohn,1999).DietarycomparisonsbetweensympatricPlectropomusareofinterestbecausetheycanreflectcompetitive interactions or niche partitioning, which can helpelucidate small-scale distributional patterns and capacity forhybridization(e.g.,Harrisonetal.,2017).However,dietarycom-parisons between sympatric Plectropomus are scarce; isotopic(δ15Nandδ13C)nichedifferedbetweenP. laevisandP. leopardus (Matley,Tobin,Simpfendorfer,Fisk,&Heupel,2017),andP. mac-ulatus andP. leopardus (Frisch, Ireland,&Baker, 2014) at reefsoff Townsville and Northwest Island, respectively. However,isotopic niche betweenP. maculatus andP. leoparduswas simi-laratOrpheusIslandReef(Matley,Heupel,Fisk,Simpfendorfer,&Tobin, 2016). Examinationof stomach content has yet tobecompletedforPlectropomusspeciesinsympatry.Therefore,thesecondobjectiveofthisstudywastoidentifyandquantifythecompositionofpreyconsumedbyPlectropomusspp.toexplorenichesegregationandfurtherinformonpreyconsumptionpat-ternsofaniconicspeciesgroup.

2  | MATERIAL S AND METHODS

2.1 | Sample collection

Three species of Plectropomus were collected within the GBRMPbetween August 2013 and May 2014 for visual, molecular, andstable isotope diet analysis (Table1). Plectropomus leopardus (n=90; mean±SE: 455±6mm; range: 276–577mm) and P. laevis (n=36; mean±SE: 522±24mm; range: 299–910mm) were col-lectedatmidshelf reefsoffTownsville,Australia (TSV:HelixReef,YankeeReef,CoilReef;Figure1),andP. leopardus(n=9;mean±SE: 475±16mm; range: 377–610mm) and P. maculatus (n = 10; mean±SE: 358±20mm; range: 280–515mm) were sampled at

     |  9505MATLEY ET AL.

OrpheusIsland(OI)Reef—aninshorereefonwestsideofOrpheusIsland (Figure1). Individuals were taken by speargunwhile divingwithSCUBA(<20mdeep).

2.2 | Visual stomach content identification

Stomachs were removed upon collection and frozen (−20°C).Stomachswerethawed,dissected,andprey itemsclassifiedbasedon thedigestion level (1–4=low–highdigestion:1—littleornodi-gestionexceptsuperficially,forexample,skinandfins;2—moderatedigestionwithheadandtailmostlydigestedandpossibilityofpartsbroken off and oval fleshy remains; 3—major digestionwith small

fleshyremainsandabundanceofbrokenparts;4—completediges-tionwith very small fragments of prey remaining or empty stom-achand clean lining). Prey (digestion level1 and2)wereweighed(0.001g) and identified to the lowest taxonomical level possibleusingRandall,Allen,andSteene(1997)andFroeseandPauly(2016).Anadditional81stomachswerecollected forvisual stomachcon-tent identification: Lodestone Reef (16—P. leopardus inNov 2013;and2—P. laevisfootballer,11—P. leopardusinFeb2014),KeeperReef(1—P. laevisfootballer,17—P. leopardusinAug2013),CentipedeReef(16—P. leopardus inAug2013), andWheelerReef (17—P. leopardus inNov2013).Thesesampleswerenotusedformetabarcodingandstableisotopes,butarepresentedheretofurtherevaluatethesuc-cessofpreyidentificationviathismethod.

2.3 | Diet metabarcoding

StomachcontentsfromeachindividualwerehomogenizedandDNAextracted following the CTAB protocol from Tamari and Hinkley(2016).Devloo-Delvaetal. (2018)establishedthemethod’sabilityforpreydiversityrecoveryinPlectropomusspp.,usingcytochromeoxidase I primers (mlCOIintF/jgHCO2198; Leray etal., 2013). Amplicon polymerase chain reactions (PCR)were completedwiththisprimersetina20μlreactionvolumewith1μloftemplateDNA,1XMyTaqreactionbuffer (Bioline,UK),0.4μMtailedforwardandreverseprimerwith10%untailedprimers(toinitiateamplification),and0.05u/μlMyTaqDNApolymerase (Bioline).PCRamplificationwas performedon aC1000ThermoCycler (BIO-RAD,USA). PCRconditionsweresettoinitialdenaturationof60sat95°C,then40cyclesof30sdenaturationat95°C,annealingat56°Cfor30s,andan extension at 72°C for 30s.Next, PCR productswere indexedusing theNextera IndexKitA (Illumina,USA). Ina finalvolumeof50 μl, we used 5μl of amplicon PCR product, 1XMyTaq reactionbuffer (Bioline), 5μl of each indexing primer and 0.05u/μlMyTaqDNA polymerase (Bioline). After each PCR step, products were

TABLE  1 SummaryofPlectropomusspp.samplecollectionforvisualandDNAstomachcontentanalysis,andstableisotopeanalysis(SIA)

Species Reef DateVisual stomach contents (n)

DNA stomach contents (n) SIA plasma (n) SIA RBC (n) SIA muscle (n)

P. leopardus Helix August2013 11(2) 10(8) 0 0 11

P. laevis(footballer) Coil,Helix,Yankee

November2013 8(1) 10(10) 6 7 8

P. laevis(bluespot) Coil,Helix,Yankee

November2013 28(4) 23(22) 28 27 28

P. leopardus Coil,Helix,Yankee

November2013 58(8) 35(21) 39 49 58

P. leopardus Helix February2014 21(4) 3(2) 19 20 21

P. leopardus Orpheus May2014 10(3) 10(9) 0 0 9

P. maculatus Orpheus May2014 10(4) 10(9) 0 0 10

Note.Bracketsrepresentthenumberofsampleswherepreywereidentified

F IGURE  1 LocationsofcoraltroutcollectionforstableisotopesandDNAgutcontents.Plectropomus leoparduswassampledatalllocations,P. maculatuswassampledatOrpheusIsland(OI)Reef,andP. laeviswassampledatCoilReef,YankeeReef,andHelixReefwithintheTownsville(TSV)sectoroftheGreatBarrierReefMarinePark.ThecityofTownsvilleisidentifiedforreference

9506  |     MATLEY ET AL.

cleanedusingtheserapurebeadsprotocol(Faircloth&Glenn,2014;Rohland&Reich,2012)onaZephyr®G3CompactLiquidHandlingWorkstation (Caliper Life sciences, USA). Finally, the library wasquantified using Qubit dsDNA HS kit (Thermo Fisher Scientific,USA),normalizedandpair-endsequencedonanIlluminaMiSeqplat-formwiththev3ReagentKit(Illumina),anddemultiplexedinMiSeqReporter(v2.5).

Raw sequenceswere filtered using a custom pipeline imple-mented inGeneiousV8.1.8 (Biomatters,NewZealand). In short,primersequenceswereremovedandbaseswerequality-trimmed(basequality<20);subsequentlyreadswerepaired,merged(mini-mumof10bpoverlap),denovoassembled(1%mismatchallowed)tocontigsandblastedagainsttheGenBankCOIdatabaseforfishand invertebrates. Blast results were quality-filtered on a lownumberofreadspersample(<0.1%),lowpairwiseidentity(<98%),andafragmentlengthoutsidea10%rangeoftheexpectedlength(313bp).

2.4 | Stable isotope analysis

Stable isotope sampling procedures and quantification followedMatleyetal. (2017).Briefly, three tissues (plasma, redbloodcells,andmuscle)werecollectedfromPlectropomus individualsandfro-zen (−20°C) until processing.Muscle tissue (no skin)was sampledfromthedorsalmusculatureusing sterile forcepsandscalpel, andbloodcomponentsweresampledfromthe2ndor3rdgillarchwithasterileneedle/syringe.Frozensampleswerefreeze-driedfor48hrandgroundintoapowder,thensampleswerelipid-extractedusinga 2:1 chloroform:methanol solvent. Stable isotope values (δ13Cand δ15N) were calculated using a continuous flow isotope ratiomass spectrometer (Finnigan MAT Deltaplus, Thermo—Finnigan)equipped with a Costech Elemental Analyzer (Costech AnalyticalTechnologies).Stableisotopeanalysisexceededacceptedprecisionandaccuracystandards(Matleyetal.,2017).

2.5 | Data analysis

Unlessindicatedotherwise,samplesfromeachspecieswerepooledbetween reefsanddatesdue to the limitednumberof individualssampled.PreviousresearchindicateddifferentcolorphasesofP. lae-vis(bluespotandfootballer)havedifferentfeedingecology(Matleyetal.,2017);therefore,colorphaseswereanalyzedseparately.Preyitemsweregroupedby familywhenvisually identifieddue to lownumbers. The family Labridae was subdivided into Scarinae and“allothers”becauseofthedifferentfeedingmodesexhibited(e.g.,parrotfishes are typically herbivores/detritivores, other Labridaearemostlypredatory).Dietaryindicesusedtosummarizethefind-ings included: percent prey contribution (Ni), frequency of occur-rence (Oi), percent weight (Wi), and index of relative importance(IRIi)(followingStJohn,2001).PreyfamilycompositionwasplottedafterPlectropomusweredividedinto3sizeclasses(<450mm,450–550mm,>550mm)toinvestigatepreyconsumptionassociatedwithontogeny/growth.

To investigate whether DNA-identified stomach contents in-cluded a sufficient number of samples to formally analyze, thecumulative number of new prey families within each consecutivestomachsampled(randomlyordered)wasplottedforeachspeciesusingthespecaccumfunctionwithinthe“vegan”package(Oksanenetal.,2016)intheRenvironment(RDevelopmentCoreTeam2014).Sampleswereconsideredadequatetocharacterizethedietifcurvesapproachedanasymptote(Ferry&Cailliet,1996).

ComparisonofDNA-identifiedstomachcontentsamongspeciesandcolorphaseswasfacilitatedbynonmetricmultidimensionalscal-ing(nMDS)basedonthepresence/absenceofpreyfamiliesusingtheBray–Curtisdissimilarityindexwithinthe“vegan”package(Oksanenetal.,2016).Ananalysisofsimilarity(ANOSIM)testedforsignificantdifferencesamongspeciesandcolorphases(reefsandsamplingperi-odspooledseparatelyforTSVandOIreefs);aglobalR-statisticvaluebetween−1 and+1wasproducedwith an associated significancelevel (α=0.05).MorepositiveR-statistic values indicatebetween-group differences, whereas values close to zero indicate randomgrouping(i.e.,within-andbetween-groupdissimilaritiesareindistin-guishable).ThedegreeofDNA-baseddietaryoverlapbetweenspe-cieswastestedusingthesimplifiedMorisitaindexandPlectropomus species combinationswith values above 0.60were considered tohave significantly overlapping diets (Langton, 1982). DifferencesinDNAstomachcontentsbetweenTSVreefs(allspeciesandsam-plingperiodscombined)andsamplingperiods (allspeciesandTSVreefscombined)werealsotestedbyANOSIMasdescribedabove.Inaddition,preyfamilycompositionwasplottedafterPlectropomus weredividedinto3sizeclasses(FBT/BST:<450mm,450–650mm,>650mm;CCT:<450mm,450–550mm,>550mm; ICT:<300mm,300–400mm, >400mm) to investigate prey consumption associ-atedwithontogeny/growth.

Although prey abundance/density at each reefwas not deter-mined simultaneouslywithPlectropomus sampling, resource selec-tion was estimated using abundance data from previous surveysat four TSV reefs (Helix Reef, Rib Reef, Chicken Reef, and KnifeReef) during March 2014 as part of the Australian Institute ofMarineScience(AIMS)LongTermMonitoringProgram(Bierwagenetal.—in press).Briefly,thesesurveysincorporatedfishcountsfrom5m belt transects (1m for pomacentrids) along five 50m tran-sects at three sites (i.e., 15 transects). Jacobs’ Electivity Index (D; Jacobs,1974)wascalculatedusingDNA-basedstomachcontentsofPlectropomusatTSVreefstodetermineifpreyfamilieswerespecif-ically selected for independent of their relative abundancewithintheenvironment.Jacobs’Dwascalculatedusingtheequation:D = r − p/(r + p)−(2rp),whererrepresentstheproportionofagivenpreyfamily in thedietandp in theenvironment.ThevalueofDvariesfrom1(maximumavoidance)to+1(maximumpreference).Indexval-uesof0 indicate thatpreyspeciesareconsumed inproportion totheirabundance.Indexvalueswerecalculatedforeachsurveyreeftoprovide95%confidenceintervals;asaconservativeapproach,ifconfidenceintervalsfellbetween−0.25and+0.25,thatpreyfamilywasconsideredtobeconsumedinproportiontoitsabundance(i.e.,neutralselection).

     |  9507MATLEY ET AL.

Preycontribution(family-level)wasestimated(at75%credibilityintervals) foreach species (andcolorphase)usingBayesian stableisotopemixingmodels (adjusted forplasma,RBC, andmuscledis-crimination factors, respectively—Matley, Fisk etal., 2016) withinthe “siar” package (Parnell & Jackson, 2013) in R. The diagnosticcorrelationmatrixplotwasusedtoidentifypreysourcesthatweresimilar;when thisoccurred,model iterationscouldnotdistinguishbetween prey sources resulting in an unknown or biased contri-bution between sources in the posterior model. To address this,confounding sourceswere removed or interpreted as a combinedsource.Priors,basedonDNA-identifiedstomachcontentsforeachspecies(describedabove),wereappliedwithaconservativestandarderrorestimateof0.078(roughlyequivalenttoa20%confidencein-terval)toimprovepreycontributionoutput(Jackson,Inger,Parnell,&Bearhop,2011).Preycontributionestimatesweresimilarlycalcu-latedusingmixingmodelswithoutpriorinformation.Preycomposi-tionoverlapcomparisonsweremadebetweenbothmodeloutputs(i.e.,withandwithoutpriors)foreachspeciesandcolorphase,andtissuetype(basedonthemidvalueof75%credibilityintervals)usingthesimplifiedMorisitaindex.Indexvaluesabove0.60indicatedthatdietcompositionfrommixingmodeloutputswassimilar.

Thecapacityforstableisotopemixingmodelstoaccuratelychar-acterizepreycompositionwasassessedbycomparingthecontribu-tionofpreyinmixingmodelswithtemporallyandspatiallyrelevantcontributionsfromDNA-identifiedstomachcontents.Forexample,anisotopicsamplingofP. leopardusconductedinFebruary2014atHelixReef,roughlycorrespondedtostomachcontentsofindividualssampledinNovember2013(~90days).Also,muscleisotopictrends(50% turnover is ~126days) from February 2014 should incorpo-ratedietfromAugust2013sampling(~180days).Itisimportanttonotethatacombinationofpreysignaturesgraduallybecomeincor-porated into consumer tissues over time, and thus, 50% turnoverperiodsusedhereareanapproximatetemporalestimateofisotopeincorporation.

Thetrophic level (TL)ofeachDNAstomachcontentprey itemwas determined using estimated values from www.fishbase.org(Froese&Pauly,2016).Totestwhetherdifferentfactors(e.g.,spe-cies, colorphase, size,δ15Nvalues, and reef) influenced theTLofconsumedprey,agenerallinearmodel(GLM)wasusedforeachtis-suesampled.ParameterswereestimatedwithrestrictedmaximumlikelihoodandaGaussiandistribution(link:identity).Here,dataweresubsettotheNovember2013(forTSVreefs)andMay2014(forOIReef) samplingperiods to reduceseasonal isotopicbiasand to in-cludeallPlectropomusspeciessampled.Modelassumptions(e.g.,ho-mogeneityofvarianceandnormality)wereverifiedusingdiagnosticplotsandtestswereconsideredsignificantlydifferentifp ≤ 0.05.

3  | RESULTS

Ofthe226stomachsvisuallyexamined100(44%)containedprey,31 (23—P. leopardus; 5—P. laevis; 3—P. maculatus) had identifiableprey items (39different items),whichwere identified to familyor

lower(11oftheseidentifiedtospecies).Caesionidae,Labridae,andPomacentridaewerethemainpreyfamiliesandcomprised~80%ofidentified prey (Table2; Supporting information Figure S1) and atleastoneofthesefamilieswasfoundin~71%ofindividualstomachswithidentifiableprey.

Of the stomachs (n=101) sampled for genetic metabarcod-ing of prey (Table1), 187 prey items (digestion level 1: n=41, 2:n=33, 3: n=68, 4: n=45) from 81 individuals (40—P. leopardus; 32—P. laevis;9—P. maculatus)wereidentifiedwhichincluded50spe-ciesfrom20families (Supporting informationTableS1;Supportinginformation Figure S2). Cumulative prey curves for P. leopardus and P. laevis approached asymptotes at ~20–25 samples, suggest-ing sufficient samples to characterizediet (Supporting informationFigure S3). The footballer phase of P. laevis had <20samples butwas treated separately from the bluespot phase due to previousinvestigations indicatingdistinct feedingecology. Likewise, samplesizesforP. maculatusandP. leopardusatOIReefwerenotadequate,butthemainoutputwasincludedforexploratorypurposes.ForallspeciesandcolorphasesatTSVreefsandOIReef,Pomacentridaeand Caesionidae comprised >50% of identified prey (Figure2a,b;Supporting information Figure S4). Thesemainly included the fol-lowingspecies:Pterocaesio digramma(Caesionidae),Neopomacentrus azysron, Acanthochromis polyacanthus, and Pomacentrus trichrourus (Pomacentridae)(SupportinginformationTableS1;Supportinginfor-mationFigureS2).RemainingpreyfamiliesvariedbetweenspeciesofPlectropomus.Planktivoreswerethemostcommonpreyforallspe-cies(~50%–70%;Figure2c,d;SupportinginformationFigureS5),andherbivorescomprised~10%–15%ofpreyexceptinP. laevis(bluespot),whereitaccountedfor~30%.BasedonANOSIMatTSVreefs,preyfamily differenceswere not found betweenPlectropomus species/color phases (R-statistic=0.050, p = 0.137; Figure3), reefs (R-statistic=0.012,p = 0.292),orsamplingperiods(R-statistic = 0.153,p = 0.067).ANOSIMresultsweresimilarwhenpreyspecies(asop-posed to families)were comparedwithPlectropomus species/colorphases (Supporting informationFigureS6)butcaution interpretingthisoutputissuggestedduetothelargenumberofpreyspeciesinrelationtoPlectropomussamplesize.SimplifiedMorisitaindicesforallTSVspeciesandcolorphasecombinationswere>0.80indicatinghighdietaryoverlap.

Pomacentridae was the most abundant family surveyedduring 2014, followed by Labridae (including Scarinae) andAcanthuridae (Bierwagen etal.—in press). Prey selection pat-terns,asdeterminedbyJacobs’ElectivityIndexshowedselectionfor Labridae (not including Scarinae) for allPlectropomus at TSVreefs (Figure4). Also, no strong selection or avoidance patternswere readily apparent for Pomacentridae despite its high abun-dance. Otherwise, the bluespot P. laevis selected for Siganidae,Serranidae,andLutjanidae,whereasP. leopardusdemonstratedanaffinitytoLethrinidae(Figure4);however,thesefamiliescontrib-utedonlyasmallportionwithinthedietofPlectropomus(Figure2).Caesionidae and a few other families found in the stomachs ofPlectropomuswerenot included in theseabundancesurveysandwerenotincludedinthisanalysis.

9508  |     MATLEY ET AL.

Preycontributionbasedonstable isotopemixingmodelsusingDNA-identified stomach content as prior informationmainly con-sistedofCaesionidaeandPomacentridaeforallspeciesandtissues(Figures5 and 6). At TSV reefs, due to similar isotopic values be-tween these two prey families itwas difficult to distinguish theircontribution values. Nevertheless, both comprised >60% of dietinP. leopardus andP. laevis for all tissues sampled (Figure5). Preycompositionoverlapbetweenmixingmodelswithandwithoutpriorinformationwassignificant(>0.60index)foralltissuesofP. leopar-dus (atTSV reefs andOIReef),P. maculatus, andP. laevis (footbal-ler); however, prey compositiondiffered forP. laevis (bluespot) (alltissues).ThemaindifferencebetweenmixingmodeloutputsatTSVreefswasthatPomacentridaeandCaesionidaecontributed lesstothedietwhenpriorinformationwasnotincludedinmixingmodels,particularlyforbluespotswhichgenerallyshowedagreaterinputofbenthic consumers such as Scarinae, Acanthuridae, and Siganidae(SupportinginformationFigureS7).AtOIReef,Serranidaecontrib-utedalargerportionofthedietwhenpriorinformationwasnotcon-sidered(SupportinginformationFigureS8).

Preycompositionestimatedfromspatiallyandtemporallyequiv-alentDNA-identifiedstomachcontentsandstableisotopeswassim-ilar (Figure7). DNA-identified stomach contents fromAugust andNovember 2013 atHelix Reef consistedmainly of PomacentridaeandCaesionidae(August:40%ofprey;November:65%),aswellasLabridae and Lethrinidae (August: 24% of prey; November: 20%).Corresponding(i.e.,February2014—HelixReef)mixingmodelout-puts(withpriors)alsoindicatedlargecontributionofPomacentridaeand Caesionidae (75% credibility intervals of muscle: 58%–86%;RBC: 69%–100%; plasma: 56%–92%), and Labridae (Lethrinidaewerenotsampled)werealsothethirdmostconsumedprey(muscle:7%–20%;RBC:3%–12%;plasma:5%–19%).

The GLM testing whether factors such as species, colorphase, size,δ15Nvalues, and reef affectedTLofDNA-basedpreyshowed that plasma (F1,43 = 13.4, p < 0.001; δ15N parameter esti-mate±SE = 0.51±0.19)andRBC(F1,42 = 7.5,p = 0.010; δ15Nparam-eterestimate±SE = 0.78±0.21)δ15NvaluesofPlectropomusatTSVreefsweresignificant(positiverelationshipwithpreyTL).Nootherfactors(includingspecies*sizeinteractions)weresignificantatTSVreefsorOIReef(i.e.,p > 0.15).

4  | DISCUSSION

This study demonstrated the utility of multiple sampling tech-niques to characterize thedietofpredatory reef fish. Specifically,DNA stomach analysis provided high prey resolution even whenitems were degraded. Stable isotopes were useful at interpretinglonger-termdietarypatterns,particularlywhencombinedwithDNAstomachanalysis,demonstratingthatwhenrepetitive,long-term,orlethalsampling is impracticalornotpossible,stable isotopeanaly-sis isapowerfulalternative,albeitwith less taxonomic resolution.BothmethodsproducedsimilarpatternsamongPlectropomus,withCaesionidaeandPomacentridaebeingthemainprey.TA

BLE 2 Visualstomachcontentsincludingpercentpreycontribution(%N),frequencyofoccurrence(%O),percentweight(%W),andindexofrelativeimportance(%

IRI)basedon31

individuals(all

Plec

trop

omusspp.combined)

Cae

sion

idae

Labr

idae

Pom

acen

trid

aeLa

brid

ae (S

carin

ae)

Aca

nthu

ridae

Serr

anid

aeSi

gani

dae

Crus

tace

aG

obiid

aeA

pogo

nida

e

%N

26.3

23.7

23.7

5.3

2.6

5.3

2.6

5.3

2.6

2.6

%O

32.3

25.8

19.4

6.5

3.2

6.5

3.2

6.5

3.2

3.2

%W

45.1

9.0

8.9

17.4

9.1

5.2

7.7

0.6

0.9

0.5

%IR

I34.9

16.0

15.9

11.1

5.7

5.1

5.1

2.9

1.7

1.5

Not

e.Stomachsof226individuals(–171—

P. le

opar

dus,–43—

P. la

evis,–12—

P. m

acul

atus)wereexaminedbutonly31hadstomachcontentsidentifiabletofamily.Preyweightwasnotadjustedforpartialdiges-

tion,andconsequently,%

Wand%

IRIarelikelyunderestimatedforsomefamilies.

     |  9509MATLEY ET AL.

4.1 | Methodological implications

Visualstomachcontentanalysisistypicallyanaffordableapproachtoidentifypreybutrelativelylabor-intensiveandlimitedbybiasesasso-ciatedwithdigestionrates,regurgitationofprey,andemptystomachs

(Arrington,Winemiller,Loftus,&Akin,2002;Vinson&Budy,2011).Thesebiasescanbeproblematicwheninterpretingdietforlargepis-civoresbecauseawidevarietyofpreyisoftenconsumedheteroge-neouslyinspaceandtime(Armstrong&Schindler,2011).Here,preycould only be visually identified in ~14%of individuals because of

F IGURE  2 Preyfamily(a,b)andpreyfunctionalmode(c,d)contribution(%N)calculatedfromDNAstomachanalysisfrom31P. leopardus (CCT),22P. laevis(bluespot;BST),10P. laevis(footballer;FBT)capturedatTownsville(TSV)reefs(Helix,Coil,andDipReefscombined;a,c)and9CCT,9P. maculatus(ICT)capturedatOrpheusIsland(OI)Reef(b,d).Preyfamiliesthatconsistedof<5%oftotalpreyforeachconsumergroupwerecombinedas“Others”

F IGURE  3 Nonmetricmultidimensionalscaling(nMDS)plotcharacterizingDNAstomachanalysisrelativetopreyfamilyofPlectropomus spp.atTSVreefs(Helix,Yankee,andCoilReefs).Atwo-dimensionalBray–Curtisdissimilarityindexwasusedresultinginastresslevelof0.08,ANOSIMR-statisticof0.05,andp-valueof0.14

9510  |     MATLEY ET AL.

emptystomachs(56%)anddigestedstomachcontents(30%).OtherstudieshavehadsimilarlimitationsforP. leopardus(Kingsford,1992;StJohn,1999).Unlesssamplingcanbeconductedonmanyindividu-als (e.g.,>20–25individualswithidentifiablepreypersamplingcat-egory),visualstomachcontentanalysisalonemaybeimpracticalforfisheswithconservationconcernssuchasPlectropomus.

Theuseofmolecularapproaches,especiallynext-generationse-quencing(NGS)barcoding,toidentifypreyoffishesisrelativelynew.However, thesemethods are increasingly utilized to identify preyandexploreecological implicationsofdiet (e.g., Lerayetal.,2013,2015).Herethemolecularapproachidentifiedpreyin~80%ofin-dividuals,includingstomachsthatwerequalifiedasemptybyvisualanalysis. Likewise, Barnett, Redd, Frusher, Stevens, and Semmens(2010)doubled thenumberof identifiablepreycompared tomor-phological analysis in broadnose sevengill sharks (Notorynchus ce-pedianus).Further,morepreyitemsweredetectedineachstomachcomparedtovisualmethods.Forexample,~26%ofstomachswithidentifiablepreycontainedtwoormoreitemsbasedonvisualstom-achcontents(thisstudy;StJohn,1999),but~69%ofstomachshadtwoormoreitemsbasedonDNAidentification.ThemaindrawbackassociatedwiththeDNAapproachiscost(e.g.,~50AUDperindivid-ual)andtheneedforpriorvalidation;however,onceoptimized,moresamplescanbeanalyzedsimultaneouslyatrelativelylowercosts.Inaddition, region-specific genetic markers are needed to broadenavailabledatabasesandavoidtaxonomicuncertainty,especiallyforrarespecies.Nevertheless,theabilitytosuccessfully identifypreyaftermanyhoursofdigestion(<16hr;Carreon-Martinezetal.,2011)providesgreaterscopetocharacterizeshort-termdietarypatternscomparedtoconventionalmethods.

Stable isotopes are now readily used as an alternative or sup-plementtostomachcontentanalysis.Thespecificadvantageisthatbroad-scalefeedingpatternsreflectinghabitatandpreysourcescanbe inferred atmultiple temporal scales (Newsomeetal., 2010). Inaddition, lethal sampling is not necessary and there is no bias as-sociatedwithdigestionratesoremptystomachs(Colborne,Clapp,Longstaffe,&Neff,2015).Amajor limitationwith isotopeanalysisistoobtainacomprehensiveviewofdiet,isotopicallydistinctpreyspeciestypicallyneedtobesampled,whichcanbedifficultandin-flatecosts(~15-30AUDpersample).Heremixingmodelsweredif-ficulttostatisticallycomparewithstomachcontentresultsbecauseofthegreatertemporalscaleassociatedwithtissue-specificisotopicassimilation.Thus,thisstudyisunabletospecificallycomparepreycompositionamongthedifferentapproachesbecausedietmanipu-lationandstandardizationwerenotcompleted. Inaddition,notallprey typesdetected in thestomachsweresampled forstable iso-tope values. Nevertheless, based on the corresponding temporalproxies of diet betweenFebruary2014muscle tissue andAugust2013 DNA-identified stomach contents, and between February2014bloodcomponentsandNovember2013DNA-identifiedstom-achcontents,thedietaryoutputfrommixingmodelswastypicallywithinestimatedmarginsforDNA-identifiedstomachcontentsforP. leopardus atHelix Reef. Admittedly, stable isotopemixingmod-elsincorporatedstomachcontentdata,butconservativemarginsoferror(i.e.,20%confidenceinterval)wereusedtonotguidethemod-els too strongly andmixingmodelswithout prior information stillidentifiedthemainpreygroups.

Resultsofthisanalysishighlightthevalueofusingmultiplecom-plementary approaches. The visual analysis provided a baseline

F IGURE  4 SummaryofresourceselectionasindicatedbyJacobs’ElectivityIndex(D)forPlectropomusatTSVreefs.TheproportionofpreyconsumedwascalculatedusingDNAstomachcontentsandtheproportionofpreyavailablewasestimatedfromstandardizedabundancesurveysatfourTSVreefs(HelixReef,RibReef,ChickenReef,andKnifeReef)during2014(Bierwagenetal.—in press).ThevalueofDvariesfrom1(maximumavoidance)to+1(maximumpreference).Indexvaluesof0indicatethatpreyspeciesareconsumedinproportiontotheirabundance.Confidenceintervalsthatfellbetween−0.25and+0.25weredeemedtobeconsumedinproportiontoitsabundance(i.e.,neutralselection)

     |  9511MATLEY ET AL.

understandingofdiet,butlackedthedetailandresolutionprovidedbythemolecularapproach.Theseresults,incombinationwithlong-terminformationsuppliedbyisotopeanalysis,provideamorecom-prehensiveunderstandingoffeedingpatterns.Theinclusionofpriorknowledge(i.e.,stomachcontentdata)intoBayesianmixingmodelshastheability to improveprecision inestimatingdietcompositionat monthly temporal scales (Chiaradia etal., 2014; Franco-Trecuetal.,2013).ItwasparticularlyusefulidentifyingpreycompositioninP. laevis (bluespot),whichappearedtounderestimatethecontri-bution of Pomacentridae andCaesionidaewhenprior information

wasnotincorporated.Ifpossible,futurestudiesshouldvalidatebothtechniques at multiple andmutually relevant time-scales to trackseasonaldietarychangesthatmayoccuramongspecies(notappar-entinthisstudy).

4.2 | Ecological implications

This study demonstrated that planktivorous Pomacentridae andCaesionidae are important components of Plectropomus diet atshort- and long-term temporal scales. This has been previouslydemonstrated for P. leopardus based on stomach content results(Kingsford,1992;St John,1999).Plectropomus leopardus areoftenconsideredopportunisticgeneralistsconsumingpreyrelativetotheirabundancewithincoralreefsystems;however,thishasnotspecifi-callybeentested.Amorecomprehensivesamplingregimeisneededto fully understand resource selection patterns for Plectropomus; however,thepreliminaryinvestigationofpreyselectioninthisstudysupports this conclusion for P. leopardus and P. laevis, particularlyconsideringthatPomacentridaeweredominantinthediet.AlthoughCaesionidaewerenotsampledintheabundancesurveysutilizedinthisstudy,video-recordedsurveyswithsimilarsamplingdesignhaveshownthatPomacentridaeandCaesionidaeareoverwhelminglythemost abundant families on the TSV reefs (Stacy Bierwagen, pers.comm.).Therefore,Plectropomusappeartofollowgeneralistandop-portunistic prey selection, at least for a large component of theirdiet. Notwithstanding relative abundance of Pomacentridae andCaesionidae, planktivorous species in these familiesmay bemorevulnerabletopredationwhenforagingabovereefstructurebecausePlectropomusareambushpredators(StJohn,2001).

ThesimilarityofpreyselectionamongPlectropomusspecieshasecologicalimplicationswithincoralreefecosystems.Sharedresource-useamongPlectropomuscouldleadtocompetitiveinteractionsoral-teredcommunitycomposition(Boström-Einarsson,Bonin,Munday,&Jones,2014;Papastamatiou,Wetherbee,Lowe,&Crow,2006).This

F IGURE  6 Preycontributionestimates(75%credibilityintervals)forPlectropomus leopardus(CCT)andP. maculatus(ICT)atOrpheusIslandReefbasedonBayesianstableisotopemixingmodels(adjustedformusclediscriminationfactors(Matley,Fisketal.,2016))usingpriorsconsistingofDNAstomachcontentanalysisforeachspecies

F IGURE  5 Preycontributionestimates(75%credibilityintervals)forPlectropomusspp.atTSVreefs(Helix,Yankee,andCoilReefscombined)basedonBayesianstableisotopemixingmodels(adjustedforplasma,RBC,andmusclediscriminationfactors,respectively(Matley,Fisketal.,2016))usingpriorsconsistingofDNAstomachcontentanalysisforeachspecies.stomach

9512  |     MATLEY ET AL.

issuewilllikelybeamplifiedunderpredictedclimatechangescenar-iosifmetabolicdemandsoflargepredatorsarenotmetduetopreyavailability (Johansenetal., 2015;Pörtner&Peck, 2010) andhabi-tat degradation alters community composition (Jones,McCormick,Srinivasan,&Eagle,2004;Wen,Bonin,Harrison,Williamson,&Jones,2016).AlthoughPlectropomusarelikelycapableofadaptingtochang-ingresourcepools(Grahametal.,2007;Johansenetal.,2015),plank-tonicfoodsourcesareimportant.Indeed,thefourmainpreyspecies(P. digramma, N. azysron,A. polyacanthus, and P. trichrourus) are pre-dominantlyplanktivorous(Froese&Pauly,2016).Changesinprimaryproductionandplankton-basedtrophodynamics(e.g.,Doney,Fabry,Feely,&Kleypas,2009)willlikelyhaveastrongeffectonhowmeso-predatorssuchasPlectropomus,selectandpartitionprey(Audzijonyte,Kuparinen,Gorton,&Fulton,2013;Hempsonetal.,2017).

Despitemajorpreyitemsbeingsimilaramongspeciesandcolorphases,theircontribution,andthatof lesserpreydiffered.Forex-ample, more planktonic prey (Clupeidae, Caesionidae) were de-tected in P. leopardus compared to P. maculatus, which consumedmore benthic/midwater consumers (Gobiidae, Lethrinidae), at OIReef.Thisdifferencemaybearesultofverticalsegregation(Matley,Heupeletal.,2016),buta larger samplesize isneeded toconfirmthese results.BluespotP. laevis appeared to selectpredatorycon-sumers such as Serranidae, Lutjanidae; however, low abundancesofthesefamiliesinsurveysmayhaveoverinflatedthefewfoundinstomachs.Benthicherbivores(Acanthuridae,Blenniidae,Siganidae)

were~15%–20%moreabundantinbluespotDNAstomachcontentscomparedtootherspeciesandcolorphases.Differencesinbenthiccarbon sources between P. leopardus and P. laevis (bluespot) werealsofoundbasedon isotopicnichebreadth,whichshowed limitedoverlap(0%–21%)forplasma,RBC,andmuscletissue(Matleyetal.,2017).Within-reefisotopiccompositionofpreylikelyvariesbasedonphysicalandbiologicalprocessesassociatedwithhabitator lo-cationon the reef (e.g.,depthorproximity toocean floor;Wyatt,Waite,&Humphries,2012).Therefore,thesamepreyspeciesmayhave different isotope values depending on foraging habitat. Thismay help explain why the stomach contents of P. leopardus andP. laeviswerenotmarkedlydifferent,asopposedto isotopicnichebreadth,asbothspeciesexhibiteddifferenthomerangesandmove-mentpatterns(i.e.,differentforagingmodes;Matley,Tobin,Ledee,Heupel, & Simpfendorfer, 2016). Different feeding modes withinprey familiesmayalsodrive isotopicdifferences.Furtherexplora-tionsofwithin-reefisotopicvariationareneededtoassesstheex-tenttowhichhabitatandprimaryproductioninfluencevaluesatalocalscale.

Temporalchangesinfeedingpatternswithinandbetweenspe-cies and color phaseswere identified. Inmixingmodels, benthicpreygroupscontributedmoretoP. laevis(bluespot)muscletissuecomparedtoplasmaandRBCs,indicatingthatoveralongertimes-cale,benthic,andmidwaterpreywereconsumedbybluespots.Incontrast, tissue-specific differences in prey composition (based

F IGURE  7  Interpretiverepresentationofstableisotopeturnover(i.e.,half-life,T0.5)andpreycompositionestimatesfrommixingmodels(using75%credibilityintervals)indifferenttissuesofPlectropomus leopardussampledinFebruary2014relativetopreycompositionbasedonDNAstomachcontentsinAugustandNovember2013

     |  9513MATLEY ET AL.

onmixingmodels)werenotevidentwithinP. laevis(footballer)orP. leopardus. Prey abundance surveys incorporated in this studywerelimitedtooneperiod(March2014)outsideofDNAsampling,and therefore, seasonal fluctuations in recruitment could affectpreyselectivity;however,pastresearchhasfoundthatthedietofP. leopardusdoesnotchangeseasonally (StJohn,2001).Findingsin this study align with this for P. leopardus, as well as P. laevis (footballer),whichishypothesizedtohaveanintermediatefeedingecologybetweenP. laevis(bluespot)andP. leopardus(Matleyetal.,2017).

Consumers typically select prey that optimize energetic gainssuch as larger prey (offset by foraging costs; Pyke etal. 1977).However, the sizeof consumedprey isoften limitedbyconsumersize due to limitations such as gape size (Mittelbach and Persson1998).Inthisstudy,therewasnostrongevidenceofPlectropomus sizeaffectingpreysizeasdemonstratedbytheGLMwithpreyTLastheresponsevariable;however,within-speciesdifferencesinsizecouldnotbetested.Likewise,whenPlectropomusweredividedintodifferentsizeclasses,therewasnoevidentseparationinmajorpreygroups,butmoreequitablesamplesizeswouldaidinterpretingthisoutput.Thesefindingsarenotsurprisingbecauseontogeneticshiftsinpreyselectionmainlyoccurpriortomaturity(StJohn,1999;Wen,Almany,Williamson,Pratchett,&Jones,2012).

5  | CONCLUSION

ThisstudyshowedDNAstomachanalysiswasamorecomprehen-sivetooltocharacterizedietcomparedtovisualstomachanalysisbyincreasingresolutionofpreyidentificationanddetectinggreatertaxonomicaldiversity.Theuseofstableisotopesprovideddietaryestimates over longer periods of time and quantification of preyvia Bayesian mixing models matched well with temporally con-gruent samples after incorporating stomach content information.Plectropomus δ15Nvalues inplasmaandRBCs reflected theTLofprey;however,muscleδ15Nvaluesdidnot,highlightingthelimita-tionsofstomachcontentstocharacterizedietoverlongerperiods.Thus, interpretation of muscle-derived mixing models should betreatedcautiouslybecausegreateruncertaintyinpreyitemsexists.Still,similaritiesbetweentemporallyrelevantdietaryoutputwereidentified, suggesting prey isotope values, discrimination factors,andpriorstomachcontentinformationweresuitablyapplied.Thus,unless repeat samplingof sufficient sample size ispossible toac-quirepreyviastomachcontents,multi-tissuestableisotopeinvesti-gationsprovideavaluablealternative,particularlyifcombinedwithwell-informed(andtemporallyrelevant)stomachcontentanalysis.Smallsamplesizes,particularlyatOIReefandforfootballerP. lae-vis,hinderedthereliabilityofcomparisonsamongspeciesandex-tentofconclusionsthatcouldbemadeinthisstudy.However,likeinmoststudies, samplesizeswereconstrainedbyseveral factorssuchasprocessingcostsandlegalcatch-limits.Broad-scalepreyse-lectionpatternsamongPlectropomuswereevidentdespitesamplesizes,withCaesionidaeandPomacentridaeasthemaincontributors

forallspecies.TheimplicationofashareddietamongPlectropomus isrelevanttoresourcemanagersbecausechangingenvironmentalconditionswill likelyhavea strongeffectonpreyavailabilityandresourcepartitioningamongpredators.Furthermore,basedonpreycompositionofPlectropomus,plankton-basedresourcesplayakeyroleinstructuringenergeticpathwaysofpredatoryreeffish.

ACKNOWLEDG MENTS

Funding was provided by the Australian Government’s NationalEnvironmentalResearchProgram(TropicalEcosystemsHubProject6.1).M.R.H. andA.T.F.were supportedby anAustralianResearchCouncilFutureFellowship(#FT100101004),andCanadianResearchChairsProgram,respectively.FundingwasprovidedtoJ.K.M.fromJames Cook University’s College of Marine and EnvironmentalSciencesandGraduateResearchSchool,theAustralianCoralReefSocietyandNaturalSciencesandEngineeringResearchCouncil.AllresearchwasconductedunderJCUAnimalEthicsPermitA1933andresearchpermitsfromtheGreatBarrierReefMarineParkAuthority(G12/35236.1 and G14/36624.1) and Queensland Department ofPrimaryIndustriesandFisheries(144482).LaboratoryanalysiswasassistedbyA.HusseyandK.Johnson.TheauthorsthankcrewfromtheR/V“JamesKirby”andOrpheusIslandResearchStation,andthefollowingcolleagues:F.deFaria,S.Moore,P.Yates,M.Espinoza,E.Lédée,J.Smart,M.Green,S.Bierwagen,andA.Schlaff.

CONFLIC T OF INTERE S T

Nonedeclared.

AUTHORS’ CONTRIBUTIONS

JKMatleyandAJTobincollectedmostsamples;JKMatleyandATFiskprocessedandanalyzedisotopesamples;JKMatley,GEMaes,FDevloo-Delva,RHuerlimann,andGChuaprocessedandanalyzedgenetic samples;andall authorswere involved in theconceptanddesign of the project, data analysis, and contributed to writing/revisions.

DATA ACCE SSIBILIT Y S TATEMENT

Supporting data have been uploaded to Dryad (doi: 10.5061/dryad.4v80379).

ORCID

Jordan K. Matley http://orcid.org/0000-0003-4286-597X

Gregory E. Maes http://orcid.org/0000-0002-1531-7321

Floriaan Devloo-Delva http://orcid.org/0000-0002-6757-2949

Roger Huerlimann http://orcid.org/0000-0002-6020-334X

Colin A. Simpfendorfer http://orcid.org/0000-0002-0295-2238

Michelle R. Heupel http://orcid.org/0000-0002-8245-7332

9514  |     MATLEY ET AL.

R E FE R E N C E SArmstrong, J. B., & Schindler, D. E. (2011). Excess digestive capacity

in predators reflect a life of feast and famine.Nature,476, 84–87.https://doi.org/10.1038/nature10240

Arrington,D.A.,Winemiller,K.O.,Loftus,W.F.,&Akin,S.(2002).Howoftendofishes‘runonempty’?Ecology,83,2145–2151.

Audzijonyte, A., Kuparinen, A., Gorton, R., & Fulton, E. A. (2013).Ecologicalconsequencesofbodysizedeclineinharvestedfishspe-cies:Positivefeedbackloopsintrophicinteractionsamplifyhumanimpact. Biology Letters, 9, 20121103. https://doi.org/10.1098/rsbl.2012.1103

Barnett,A.,Redd,K.S.,Frusher,S.D.,Stevens,J.D.,&Semmens,J.M.(2010).Non-lethalmethodtoobtainstomachsamplesfroma largemarinepredatorandtheuseofDNAanalysistoimprovedietaryin-formation. Journal of Experimental Marine Biology and Ecology,393,188–192.https://doi.org/10.1016/j.jembe.2010.07.022

Bierwagen, S.L., Emslie, M. J., Chin, A., & Simpfendorfer, C. A. (In-Press). Reef-scale variability in fish and coral assemblages on thecentralGreatBarrierReef.Marine Biology https://doi.org/10.1007/s00227-018-3400-5

Boström-Einarsson,L.,Bonin,M.C.,Munday,P.L.,&Jones,G.P.(2014).Habitatdegradationmodifiesthestrengthofinterspecificcompeti-tionincoraldwellingdamselfishes.Ecology,95,3056–3067.https://doi.org/10.1890/13-1345.1

Carreon-Martinez,L.,Johnson,T.B.,Ludsin,S.A.,&Heath,D.D.(2011).Utilizationof stomach contentDNA to determine diet diversity inpiscivorousfishes.Journal of Fish Biology,78,1170–1182.https://doi.org/10.1111/j.1095-8649.2011.02925.x

Chiaradia,A.,Forero,M.G.,McInnes,J.C.,&Ramírez,F.(2014).Searchingforthetruedietofmarinepredators:IncorporatingBayesianpriorsintostableisotopemixingmodels.PLoS ONE,9,e92665.https://doi.org/10.1371/journal.pone.0092665

Colborne,S.F.,Clapp,A.D.M.,Longstaffe,F. J.,&Neff,B.D. (2015).Foragingecologyofnativepumpkinseed (Lepomis gibbosus) follow-ing the invasionof zebramussels (Dreissena polymorpha).Canadian Journal of Fisheries and Aquatic Sciences, 72, 983–990. https://doi.org/10.1139/cjfas-2014-0372

Connell, J. H. (1980). Diversity and the coevolution of competitors,or the ghost of competition past.Oikos, 35, 131–138. https://doi.org/10.2307/3544421

Devloo-Delva, F.,Huerlimann,R.,Chua,G.,Matley, J.K.,Heupel,M.R.,Simpfendorfer,C.A.,&Maes,G.E. (2018).Howdoesmarkerchoiceaffect your diet analysis: Comparing genetic markers and digestionlevels for dietmetabarcoding of tropical-reef piscivores.Marine and Freshwater Research.https://doi.org/10.1071/MF17209

Doney,S.C.,Fabry,V.J.,Feely,R.A.,&Kleypas,J.A.(2009).Oceanacid-ification:TheotherCO2problem.Annual Review of Marine Science,1,169–192.https://doi.org/10.1146/annurev.marine.010908.163834

Estes,J.A.,Terborgh,J.,Brashares,J.S.,Power,M.E.,Berger,J.,Bond,W.J.,…Wardle,D.A.(2011).TrophicdowngradingofplanetEarth.Science,333,301–306.https://doi.org/10.1126/science.1205106

Faircloth,B.,&Glenn,T. (2014).Protocol: Preparation of an AMPure XP substitute(a.k.a. Serapure).https://doi.org/10.6079/J9MW2F26

Ferry, L., & Cailliet, G. M. (1996). Sample size and data analysis: Arewe characterizing and comparing diet properly? In K. Shearer, &D.MacKinlay (Eds.),Feeding ecology and nutrition in fish, symposium proceedings, international congress on the biology of fishes, physiology section (pp. 71–80). San Francisco, California: American FisheriesSociety.

Franco-Trecu,V.,Drago,M., Riet-Sapriza, F.G., Parnell,A., Frau,R.,&Inchausti,P. (2013).Bias indietdetermination: Incorporatingtradi-tionalmethodsinBayesianmixingmodels.PLoS ONE,8,1–8.

Frisch,A.J.,Ireland,M.,&Baker,R.(2014).Trophicecologyoflargepred-atoryreeffishes:Energypathways,trophiclevel,andimplicationsfor

fisheries inachangingclimate.Marine Biology,161,61–73.https://doi.org/10.1007/s00227-013-2315-4

Froese,R.,&Pauly,D.(2016).FishBase.WorldWideWebelectronicpub-lication.Retrievedfromwww.fishbase.org(accessed12/20/2016).

Graham, N. A. J., Wilson, S. K., Jennings, S., Polunin, N. V. C.,Robinson, J., Bijoux, J. P., &Daw, T.M. (2007). Lag effects in theimpacts of mass coral bleaching on coral reef fish, fisheries, andecosystems. Conservation Biology, 21, 1291–1300. https://doi.org/10.1111/j.1523-1739.2007.00754.x

Harrison,H.B.,Berumen,M.L.,Saenz-Agudelo,P.,Salas,E.,Williamson,D.H.,&Jones,G.P. (2017).Widespreadhybridizationandbidirec-tionalintrogressioninsympatricspeciesofcoralreeffish.Molecular Ecology,26,5692–5704.https://doi.org/10.1111/mec.14279

Hempson,T.N.,Graham,N.A.J.,MacNeil,M.A.,Williamson,D.H.,Jones,G.P.,&Almany,G.R.(2017).Coralreefmesopredatorsswitchprey,shorteningfoodchains,inresponsetohabitatdegradation.Ecology and Evolution,7,2626–2635.https://doi.org/10.1002/ece3.2805

Jackson,A.L.,Inger,R.,Parnell,A.C.,&Bearhop,S.(2011).Comparingisotopicnichewidthsamongandwithincommunities:SIBER-StableIsotopeBayesianEllipsesinR.The Journal of Animal Ecology,80,595–602.https://doi.org/10.1111/j.1365-2656.2011.01806.x

Jacobs,J.(1974).Quantitativemeasurementoffoodselection:Amodi-ficationoftheforageratioandIvlevselectivityindex.Oecologia,14,413–417.https://doi.org/10.1007/BF00384581

Johansen,J.L.,Pratchett,M.S.,Messmer,V.,Coker,D.J.,Tobin,A.J.,&Hoey,A.S.(2015).Largepredatorycoraltroutspeciesunlikelytomeet increasing energetic demands in a warming ocean. Scientific Reports,5,13830.https://doi.org/10.1038/srep13830

Jones,G.P.,McCormick,M.I.,Srinivasan,M.,&Eagle,J.V.(2004).Coraldeclinethreatensfishbiodiversityinmarinereserves.Proceedings of the National Academy of Sciences of the United States of America,101,8251–8253.https://doi.org/10.1073/pnas.0401277101

Kingsford,M. J. (1992).Spatial and temporal variation inpredationonreeffishesbycoraltrout(Plectropomus leopardus,Serranidae).Coral Reefs,11,193–198.https://doi.org/10.1007/BF00301993

Langton,R.W.(1982).DietoverlapbetweenAtlanticcod,Gadus morhua,silver hake, Mer- luccius bilinearis, and fifteen other northwestAtlanticfinfish.Fishery Bulletin,80,745–759.

Leray,M.,Meyer,C.P.,&Mills,S.C.(2015).Metabarcodingdietaryanal-ysisofcoraldwellingpredatoryfishdemonstratestheminorcontri-butionofcoralmutualiststotheirhighlypartitioned,generalistdiet.PeerJ,3,e1047.https://doi.org/10.7717/peerj.1047

Leray,M., Yang, J. Y.,Meyer, C. P., Mills, S. C., Agudelo, N., Ranwez,V., … Machida, R. J. (2013). A new versatile primer set targetinga short fragment of the mitochondrial COI region for metabar-coding metazoan diversity: Application for characterizing coralreef fish gut contents. Frontiers in Zoology, 10, 34. https://doi.org/10.1186/1742-9994-10-34

Matley,J.K.,Fisk,A.T.,Tobin,A.J.,Heupel,M.R.,&Simpfendorfer,C.A.(2016).Diet-tissuediscriminationfactorsandturnoverofcarbonandnitrogen stable isotopes in tissuesof anadultpredatory coralreef fish, Plectropomus leopardus. Rapid Communications in Mass Spectrometry,30,29–44.https://doi.org/10.1002/rcm.7406

Matley,J.K.,Heupel,M.R.,Fisk,A.T.,Simpfendorfer,C.A.,&Tobin,A.J.(2016).Measuringnicheoverlapbetweenco-occurringPlectropomus spp. using acoustic telemetry and stable isotopes. Marine and Freshwater Research,68,1468.https://doi.org/10.1071/MF16120

Matley,J.K.,Tobin,A.J.,Ledee,E.J.I.,Heupel,M.R.,&Simpfendorfer,C.A. (2016).Contrastingpatternsof vertical andhorizontal spaceuseof twoexploited and sympatric coral reef fish.Marine Biology,163,253.https://doi.org/10.1007/s00227-016-3023-7

Matley,J.K.,Tobin,A.J.,Simpfendorfer,C.A.,Fisk,A.T.,&Heupel,M.R. (2017). Trophic niche and spatio-temporal changes in the feed-ing ecology of two sympatric species of coral trout (Plectropomus

     |  9515MATLEY ET AL.

leopardusandP. laevis).Marine Ecology Progress Series,563,197–210.https://doi.org/10.3354/meps11971

Mittelbach,G.G.,&Persson,L. (1998).Theontogenyofpiscivoryanditsecologicalconsequences.Canadian Journal of Fisheries and Aquatic Sciences,55,1454–1465.https://doi.org/10.1139/f98-041

Newsome,S.D.,Clementz,M.T.,&Koch,P.L.(2010).Usingstableiso-tope biogeochemistry to study marine mammal ecology. Marine Mammal Science,26,509–572.

Oksanen,J.,Blanchet,F.G.,Friendly,M.,Kindt,R.,Legendre,P.,McGlinn,D.,…Wagner,H.(2016).Vegan: Community ecology package.Rpack-age version 2.4-1. Retrieved from https://CRAN.R-project.org/package=vegan

Papastamatiou,Y.P.,Wetherbee,B.M.,Lowe,C.G.,&Crow,G.L.(2006).Distribution and diet of four species of carcharhinid shark in theHawaiianIslands:Evidenceforresourcepartitioningandcompetitiveexclusion.Marine Ecology Progress Series,320,239–251.https://doi.org/10.3354/meps320239

Parnell, A., & Jackson, A. (2013). Siar: Stable isotope analysis in R. R package version 4.2. Retrieved from https://CRAN.R-project.org/package=siar

Pörtner,H.O.,&Peck,M.A.(2010).Climatechangeeffectsonfishesandfish-eries:Towardsacause-and-effectunderstanding.Journal of Fish Biology,77,1745–1779.https://doi.org/10.1111/j.1095-8649.2010.02783.x

Pyke,G.H.,Pulliam,H.R.,&Charnov,E.L.(1977).Optimalforaging:Aselectivereviewoftheoryandtests.The Quarterly Review of Biology,52,137–154.http://www.jstor.org/stable/2824020

RDevelopmentCoreTeam(2014).R: A language and environment for statisti-cal computing.Vienna,Austria:RFoundationforStatisticalComputing.www.r-project.orghttps://doi.org/10.1007/978-1-4939-3185-9

Randall,J.E.,Allen,G.R.,&Steene,R.C.(1997).Fishes of the great barrier reef and Coral Sea.Honolulu,HI:UniversityofHawaiiPress.

Rohland,N.,&Reich,D.(2012).Cost-effective,high-throughputDNAse-quencinglibrariesformultiplexedtargetcapture.Genome Research,22,939–946.https://doi.org/10.1101/gr.128124.111

Ross, S. T. (1986). Resource partitioning in fish assemblages: A re-view of field studies. Copeia, 1986, 352–388. https://doi.org/10.2307/1444996

SadovydeMitcheson,Y.,Craig,M.T.,Bertoncini,A.A.,Carpenter,K.E.,Cheung,W.W.L.,Choat, J.H.,…Sanciangco,J. (2013).Fishinggrouperstowardsextinction:Aglobalassessmentofthreatsandex-tinctionrisksinabilliondollarfishery.Fish and Fisheries,14,119–136.https://doi.org/10.1111/j.1467-2979.2011.00455.x

Sale,P.F.(1977).Maintenanceofhighdiversityincoralreeffishcommunities.The American Naturalist,111,337–359.https://doi.org/10.1086/283164

Schoener,T.W.(1974).Resourcepartitioninginecologicalcommunities.Science,185,27–39.https://doi.org/10.1126/science.185.4145.27

StJohn,J.(1999).Ontogeneticchangesinthedietofthecoralreefgrou-perPlectropomus leopardus (Serranidae): Patterns in taxa, size andhabitatofprey.Marine Ecology Progress Series,180,233–246.https://doi.org/10.3354/meps180233

StJohn,J. (2001).Temporalvariation inthedietofacoral reefpisciv-ore(Pisces:Serranidae)wasnotseasonal.Coral Reefs,20,163–170.https://doi.org/10.1007/s003380100152

Tamari,F.,&Hinkley,C.S.(2016).ExtractionofDNAfromplanttissue:Review and protocols. In M. Mićić (Ed.), Sample preparation tech-niques for soil, plant, and animal samples(pp.245–263).NewYork,NY:HumanaPress.

Vinson, M. R., & Budy, P. (2011). Sources of variability and compa-rability between salmonid stomach contents and isotopic anal-yses: Study design lessons and recommendations. Canadian Journal of Fisheries and Aquatic Sciences,68, 137–151.https://doi.org/10.1139/F10-117

Wen,C.,Almany,G.,Williamson,D.,Pratchett,M.,&Jones,G. (2012).Evaluating the effects of marine reserves on diet, prey availabil-ity andprey selectionby juvenilepredatory fishes.Marine Ecology Progress Series,469,133–144.https://doi.org/10.3354/meps09949

Wen,C.K.C.,Bonin,M.C.,Harrison,H.B.,Williamson,D.H.,&Jones,G. P. (2016). Dietary shift in juvenile coral trout (Plectropomus maculatus) following coral reef degradation from a flood plumedisturbance. Coral Reefs, 35, 451–455. https://doi.org/10.1007/s00338-016-1398-z

Wyatt, A. S. J.,Waite, A.M., & Humphries, S. (2012). Stable isotopeanalysis reveals community-level variation in fish trophodynamicsacrossafringingcoralreef.Coral Reefs,31,1029–1044.https://doi.org/10.1007/s00338-012-0923-y

SUPPORTING INFORMATIONAdditional supporting information may be found online in theSupportingInformationsectionattheendofthearticle.

How to cite this article:MatleyJK,MaesGE,Devloo-DelvaF,etal.Integratingcomplementarymethodstoimprovedietanalysisinfishery-targetedspecies.Ecol Evol. 2018;8:9503–9515. https://doi.org/10.1002/ece3.4456