Transcript
Page 1: Gigapixel big data movies provide cost-effective seascape

Ecology and Evolution 20181ndash12 emsp|emsp1wwwecolevolorg

Received16November2017emsp |emsp Revised14May2018emsp |emsp Accepted17May2018DOI101002ece34301

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

Gigapixel big data movies provide cost- effective seascape scale direct measurements of open- access coastal human use such as recreational fisheries

David J H Flynn12emsp|emspTim P Lynch1 emsp|emspNeville S Barrett2emsp|emspLincoln S C Wong12emsp|emsp Carlie Devine1emsp|emspDavid Hughes1

ThisisanopenaccessarticleunderthetermsoftheCreativeCommonsAttributionLicensewhichpermitsusedistributionandreproductioninanymediumprovidedtheoriginalworkisproperlycitedcopy2018TheAuthorsEcology and EvolutionpublishedbyJohnWileyampSonsLtd

TweetBigdataseascapescalemoviesallowforcost-effectivehigh-frequencymonitoringofcoastalrecreationaluseCSIROCRAGSTuna

1TheCommonwealthScientificandIndustrialResearchOrganisation(CSIRO)OceansampAtmosphereHobartTASAustralia2TheInstituteforMarineandAntarcticStudies(IMAS)UniversityofTasmaniaHobartTASAustralia

CorrespondenceTimPLynchTheCommonwealthScientificandIndustrialResearchOrganisation(CSIRO)OceansampAtmosphereCastrayEsplanadeHobartTAS7000AustraliaEmailtimlynchcsiroau

Funding informationUniversityofTasmaniaPartoftheIMASHonoursProjectTheCoastsProgramCSIROCAPEXGigapanruggerdizationMkII201415

AbstractCollectingdataonunlicensedopen-accesscoastalactivitiessuchassometypesofrecreationalfishinghasoftenreliedontelephoneinterviewsselectedfromlandlinedirectoriesHoweverthisapproachisbecomingobsoleteduetochangesincommuni-cationtechnologysuchasaswitchtounlistedmobilephonesOthermethodssuchasboatrampinterviewsareoftenimpracticalduetohighlaborcostWetrialedanau-tonomousultra-high-resolutionphotosamplingmethodasacosteffectsolutionfordirect measurements of a recreational fishery Our sequential photosampling wasbatchedprocessedusinganovelsoftwareapplicationtoproduceldquobigdatardquotimese-riesmoviesfromaspatialsubsetofthefisheryandwevalidatedthiswitharegionalbus-routesurveyandinterviewswithparticipantsataccesspointsWealsocomparedlaborcostsbetweenthesetwomethodsMost trailerboatuserswererecreationalfishers targeting tuna sppOur camera systemcloselymatched trends in temporalvariationfromthelargerscaleregionalsurveybutasthecameradatawereatmuchhigher frequency we could additionally describe strong daily variability in effortPeakswerenormallyassociatedwithweekendsbutconsecutiveweekendtunafish-ingcompetitionsledtoananomalyofhigheffortacrossthenormalweekdaylullsByreducingfieldtimeandbatchprocessingimagerymonthlylaborcostsforthecamerasamplingwereaquarterofthebus-routesurveyandindividualcamerasamplescost25of bus route samples toobtainGigapixel panoramic cameraobservationsoffishingwererepresentativeofthetemporalvariabilityofregionalfishingeffortandcould be used to develop a cost-efficient indexHigh-frequency sampling had theaddedbenefitofbeingmorelikelytodetectabnormalpatternsofuseCombinationsofremotesensingandon-siteinterviewsmayprovideasolutiontodescribinghighlyvariableeffortinrecreationalfisherieswhilealsovalidatingactivityandcatch

K E Y W O R D S

bigdatacoastalCSIRORuggerdisedAutonomousGigapixelSystemGigaPanopen-accessfisheriesphotomosaicremotesensingsouthernbluefintunatimeintervalcounttime-lapse

2emsp |emsp emspensp FLYNN et aL

1emsp |emspINTRODUC TION

Worldwidecoastalzonesaresomeofthemostheavilyimpactedpartsoftheoceanbypeoplebothintermsofactivitytypesandparticipationrates(Halpernetal2008)Aparticularfocusofre-search on coastal use has been recreational fishing as this pop-ular activity can significantly affect fish and invertebrate stocks(Arlinghaus 2006 Cooke amp Cowx 2004 Lewin Arlinghaus ampMehner2006McPheeLeadbitterampSkilleter2002)asformanyspecies harvest can exceed the take of the commercial fishery(GiriampHall2015LyleStarkampTracey2014ZischkeGriffithsampTibbetts2012)

Unlikecommercialfisherieswhicharecommonlydocumentedbyoperatorsloggingtheircatchandeffortassessmentsofopen-accessnonreporting activities suchas recreational fisheries require sam-plingThiscanbebothdifficultandexpensiveduetotheiroftenlargespatialextenthightemporalvariationandforthefisheriesagenciesundertaking the work which often requires weekend and holidaysurveyinghighlaborcosts(Maetal2018PollockJonesampBrown1994 Rocklin Levrel Drogou Herfaut amp Veron 2014 VenturelliHyderampSkov2017)Due to these issues assessmentshave typi-callyrelieduponindirectdatacollectionfromoff-sitetelephonesur-veysbasedondataframesdevelopedaroundregionallsquowhitepagesrsquoor landline telephonedirectorieswhichtraditionallyhavehadhighresponseratesandeasilyscalablesamplesizes(Mooreetal2015Pollocketal1994)Suchapproachesmaynowriskundersamplingactivefishersbecauseofdemographicallybiasandoveralldecreasedlandlinephoneuseunlistedandnonregionalcodedmobilephonesandlowresponsesuccessfromvoicecallstomobilephones(Badcocketal 2016 Blumberg amp Luke 2009 Teixeira Zischke ampWebley2016)

Other formsof recreational fisheryassessmentsemploydirecton-sitesurveymethodssuchasbus-routesurveyswhichuseclerkstraveling around a fisherymdashfollowing randomly selected predeter-minedschedulesmdashinterviewingandobservingfisherstotrackfish-ingeffortcatchandrelease(McGlennonampKinloch1997Pollocketal 1994) On-sitemethods that cover broad geographic areashowever are seldom performed due to high labor costs (JonesampPollock2012)

Given these limitations remote sensing tools and technol-ogies such as autonomous photography are an emerging fieldofferingdirectmonitoringasanalternativeorsupplementtoon-site interviews (HartillPayneRushampBian2016KellerSteffeLowryMurphyampSuthers2016ParnellDaytonFisherLoarieampDarrow2010PowersampAnson2016vanPoortenCarruthersWardampVarkey 2015Wood LynchDevineKelleramp Figueira2016) One such system is the CSIRO Ruggedised AutonomousGigapixel System (CRAGS) which is a programmable weather-proofcameratrapthatprovidesultra-highresolutiontime-lapseandhigh-frequencypanoramic imagesTheCRAGSutilizesmod-ified commercially available hardware and software known as aGigaPanreg (httpgigapancom)tocapturephotographicsampleswith such high pixel density (gigapans) that wide fields of view

andassociatedobjectsofinterestscanbeexaminedcloselywith-out the loss of broader environmental context at landscape orseascape scales (LynchAldermanampHobday 2015) (SupportingInformationAppendixS1)

Thisroboticcamerasystemallowsformuchbroaderscaledphoto-graphicassessmentsthansimplecameratrapsbyautonomouslytak-ingandthenstitchingtogethermultipletelephotomegapixelimagesinto high-resolution gigapixel panorama The resulting tiled imageallowsfullyzoomableviewingforeithermultiplesmalltargetssuchasAlbatrossnestsacrossacolony(Lynchetal2015)ormorewidelyspacedlargertargetssuchasidentifyingcommercialvsrecreationalvessels around an offshore artificial reef (Wood etal 2016) Thebenefitofthesystemcomparedtosimplercameratrapsisbeingabletoobserve inhighdetail fromtheone imagestreammanyobjectssimultaneously across landscapeor seascape scalesWith somuchinformationdatahandlingcanbecomeoverwhelmingsoabatchpro-cessingmethod calledGigapanTimemachinehasbeendevelopedOriginallyusedforviewingcosmologicalsimulations(Yuetal2011)TimeMachineproducesavideostreamthatallowsviewerstofluidlyexploregigatotetrapixel-scaledvideosacrossbothspaceandtime

InAustraliaannualrecreationalfishingparticipationhasbeenes-timatedat195(HenryampLyle2003)whichiswellabovetheglobalaverageofaround10(ArlinghausTillnerampBork2015)AroundtheAustralian island state of Tasmania the annual participation rate is293whichwellexceedsthenationalaverageTheTasmanPeninsulainTasmaniarsquossoutheast (Figure1) isapopularfishingregionneartothelargestcityandstatecapitalofHobartwhereasteepbathymetricprofileandmigratorygame-fishpathwayoverlapprovidingcoastalac-cesstonormallyoffshorefishingopportunitiestorecreationalfisherintrailerboatsPelagicgamefishsuchassouthernbluefintuna(Thunnus maccoyii)aretargetedinlateAustralSummerandAutumn(MortonampLyle2004)butasthefisheryisunlicensedandepisodictrendsinef-fortandcatcharepoorlyunderstood(LowryampMurphy2003Mooreetal2015)Thisparticularrecreationalfisheryisalsoofmoregeneralinterestasthetargetsarecommerciallyimportantspecieswhichin-cludethosewithinternationallynegotiatedquotas(Ponsetal2017Zischke etal 2012) Recreational fishers are also often associatedwithotherusersofthemarineenvironment(FarrStoecklampSutton2014Kearney2002)orinthiscasetouristsenjoyingthelocalnationalparksHencehighusesitesmaybeimportanttomonitornotonlyforfisheriesandecologicalreasonsbutformanagingsocialvaluescon-flict resolutionandplanning (AlessaKliskeyampBrown2008Lynchetal2004)

WetrailedourCRAGSmethodtoseewhetherwecoulddevelopa high-frequency time series of temporal trends of observed trailerboat fishingeffortatanoffshoreconcentrationpoint for the recre-ationalgamefisheryWealsotestedboththesuitabilityandlaborcost-effectivenessofthehardwareandanovelapplicationofTimeMachinebatchprocessingsoftwaretoautomaticallyproducealdquobigdatardquointer-activemoviesmadefromourgigapixelpanoramasAsthepeninsulais isolatedandservicedbyonlya limitednumberofboatrampsweaimed tovalidate the representativenessof the temporal trendsweobservedwithCRAGSbycorrelatingtomatchsamplesobtainedfrom

lyn088
Cross-Out
lyn088
Cross-Out
lyn088
Inserted Text
M

emspensp emsp | emsp3FLYNN et aL

asimultaneousregionallyscaledbus-routetrailerboatsurveyatthethreerampsonthepeninsularsuitableforlaunchinglargetrailerboatsWe also undertook on-ground interviewswith trailer boat users togroundtruthactivitytype(fishingornonfishing)andtargetspeciesByundertakingbothcameraandboatrampsurveysourover-overarchingaimwastoinvestigatehowremoteimagerymayreplacecomplimentoroptimizeconventionalon-siteapproachesforrecreationalfishing

2emsp |emspMATERIAL S AND METHODS

WedeployedtheCRAGSbetween1May2015and30June2015toassesstrailerboatsoffshorefromwateradjacenttotheTasman

Peninsula (Figure1) This period was chosen as it correspondswith the main tuna fishing season For large trailer boats (gt5mlength) localaccess to the fishery is limitedto threeboat ramps(Figure1)althoughtheareamayalsobevisitedbylargervesselssteaming frommarinas and anchorages further north (MortonampLyle2004)butforthepurposesofourstudyweexcludedtheserelativelyrare largervesselsWefocusedourCRAGSsurveyataknown game fishing concentration the Hippolyte Rock MarineConservationArea (HRMCA)which are a group of small islandsabout6-kmoffshore (Figures1and2)aroundwhichrecreationalfishingispermittedWhiletheCRAGSdeploymentwasmarginallycloser to the FortescueBayRampwe commonly observe boatsapproachingtheHRMCAfromdirectionsthatcorrespondedtothe

F IGURE 1emspTheTasmanPeninsulainSETasmaniaBus-routesurveysamplingoccurredatramps locatedatPiratesFortescueandStuartsBayTheCRAGSlocationisshownasacameraiconandtheareaobservedbythecamerawhichincludestheHippolyteRocksMarineConservationArea(HRMCA)isdescribedasatriangle

Text

Pirates Bay

Fortesque Bay

StewartsBay

0 3 6 9 1215Kilometers

0 30 60 90 12015Kilometers

Tasman Is

Australia

Tasmania

Hobart

TasmanPeninsular

HippolyteRocksMCA

43deg 01 30

147deg 52 30

TAS

4emsp |emsp emspensp FLYNN et aL

locationof all three rampsWe fixed theCRAGS to a clifftop atWaterfall Baywhich overlooks the site fromvery high sea cliffs(~250m)withthecamerafocusingonthewesternsideofHRMCA(Figure1)ForeachsampleCRAGScapturedaseriesof68pho-tographsthroughatelephotolenswhichwhenstitchedtogetherprovide a high-resolution panoramic image with a wide field ofview (~140deg) (Supporting Information Appendix S1) The size ofareasurveyedbyCRAGScomparedtotheregionalareaaccessibleby trailer boatswas estimated using google earth polygons andtheUniversityofNewHampshireKMLextensiontoolmdashhttpsex-tensionunhedukmlToolsindexcfm

Thetimeof the initialCRAGSpanoramasamplewasrandomlyselected Samples were then collected sequentially through-out daylight hours (0600ndash1800) As less recreational fishing fortuna occurred at night no samples were taken with the CRAGSprogrammed to ldquoblankrdquo these hours from the sampling sched-ule (Supporting InformationAppendix S1) Photographic sessionswereprogrammedtobeseparatedbyarecurring93-mingapthus

producing a forward shift in sample capture time andminimizingpotentialautocorrelationsbetweenboatingactivitiesandsamplingtimesPhotographswerebatchprocessedtobothformpanoramicimagesforeachsampleandforcompilingintoaninteractivetime-lapsemovieusingCREATELabrsquosTimeMachinesoftware (CREATELabregPittsburgPAUSA)

TimeMachinemovies are fully interactive and are able to bezoomedintopausedandplayedatvariousspeedsandarealsotimeanddate-stamped(Figure3)UsingthisinteractivemoviesoftwareeachGigaPanmoviewassystematicallypausedateachsamplefullyzoomedandscannedlefttorighttoptobottomforobjectsintheseascapebytheleadauthor(Figure3)Vesselcountswererecordedforeachsamplewithobjectslargerthan~10minlength(nottrailerboats)or requiring longer than10seconds todistinguish (eg lowresolutionandimageartifact)beingdiscountedSampleswerecat-egorized into two strataweekdays andweekendspublic holidaysusing the time and date stamp provided on each GigaPan scene(Figure3)

F IGURE 2emspAnunmodifiedGigaPanEpicPROunit(left)andaCRAGSdeployedatWaterfallBaySETasmania(right)

F IGURE 3emspScreencaptureoutputsfromtheTimeMachineinteractivedataviewershowinganozoomandpanoramicextentoftheHippolyteRockMarineConservationArea(HRMCA)b=13zoommainHippolyterock~6kmfromcamerac=12zoomofalargetrailerboatd=fullzoom

lyn088
Cross-Out
lyn088
Inserted Text
(a)
lyn088
Cross-Out
lyn088
Inserted Text
(b)
lyn088
Cross-Out
lyn088
Inserted Text
(c)
lyn088
Cross-Out
lyn088
Inserted Text
(d)
lyn088
Inserted Text
s

emspensp emsp | emsp5FLYNN et aL

Thecountofboatsforeachsamplewasthenextrapolatedintoanestimateofdailyaverageboatingeffort(hoursplusmnSE)usingthein-stantaneouscountmethod(Equation1)

where ei is theextrapolatedboatinghours for the ithday Ii is theaveragenumberoftrailerboatsobservedacrosstheinstantaneoussamplesfordayiandTisthedailysamplingperiod(daylighthours)Asbadweathercanbeamajorfactoraffectingthedecisiontoboatorfish(ForbesTraceyampLyle2009)boatingeffortwasassumedtobe0whenadverseweathermadeobservationsofboatsimpossible

Next total effort for each day-type strata (Ejk) within sam-pling months (k) was calculated to estimate the monthly effort(Equation2)AsCRAGSallowedforcontinuoussamplingthroughoutthesurveyperiodthesamplingprobability(πk)wassetto1

Standarderrorandeffortvarianceformonthlyestimateswerecalculated(Equation3)accordingtoPollocketal(1994)forstrata

where nj is thenumberofsampleddaysofstrata j Njk is thetotalnumberofdaysofstratajinmonthk

Asvariationforeachstratawascalculatedseparatelyweusedasumofsquaredstandarderrorsapproach(Equation4)toestimatethetotalmonthlyvariation

Assampleswereobtainedinaserialandcontinuousfashionatimeseriesmodelusingrollingaverageswasalsoappliedtothedata(Diggle1995)(Equation5)

where smt is the smoothed average of extrapolated effort e persampletwhichisweightedacrossthreesampleseachseparatedby93minThisprovidedasmoothedaveragewithindays

Wealsoconductedabus-routestylesurveytoestimatefishingeffortfromtrailerboatsfortheregion(Pollocketal1994Robsonamp Jones1989)WhileCRAGSsampleswerecollected throughoutMayandJunethecomparablebus-routesurveywasonlyconductedinMaydue to logistical constraintsWesurveyed the threemajorboatrampsintheregionatPiratesBayFortescueBayandStewartsBaybetween3May2015and30May2015(Figure1)

Atotalof20bus-routesampleswerescheduledWeundertookrandomstratifiedsamplingwithdaytypedividedintoweekdaysorweekendpublic holidays To reflect expected higher recreationaleffortduringweekendsandpublicholidays(McCluskeyampLewison2008)thesamplingprobability(πk)wasweightedtowardthisstrata

over weekdays at 12 to eight samples (Supporting InformationAppendix S2) We stratified the sample time into morning (am0600ndash1159)andafternoon(pm1200ndash1800)withequalproba-bilitiesofsamplingTominimizetemporalautocorrelationthestart-ing location and travel direction of the routewere also randomlyselectedwithoutreplacementEqualsamplingweightwasprovidedtoeachrampduetolackofpriorknowledgeofuserates

At each boat ramp surveys were conducted by first countingparkedboattrailersthenbymonitoringvesselentryandexitsduringa1-hrsamplingperiod(KinlochMcGlennonNicollampPike1997Pollocketal1994)The1-hrperiodwaschosendueto travel timeanddis-tancebetweensitesallowingforabus-routesampletobecompletedforallrampswithinthetemporalstrata(AMorPM)ofthedailysam-plingframeInterviewswereconductedwithallpartieslaunchingandretrievingvesselsduringtheclerkrsquoswaittimedeterminingpartysizetargetspeciesgroupandactivity(iefishingornot)forestimatesoffishing effort This well-established procedure produced a monthlyextrapolatedestimationofboateffort instandardizedunitsofefforthoursplusmnSEwhichcouldthenbeadjustedtofishingeffortviathepro-portionofintervieweesactivities(Pollocketal1994RobsonampJones1989)(SupportingInformationAppendixS3)PartysizebetweenrampswastestedforanydifferenceusingaonefactorANOVA

Thebus-route surveyhenceprovidesanestimationofall fish-ing effort from trailer boats around this isolated peninsular TheCRAGSdataarethusahigh-frequencyspatialsubsetoftheentiretrailer boat fishing effort for the period thatwas extrapolated bytheregionalsurveyThisallowedustobothtesttherelativeuseoftheHRMCAcomparedtowiderpeninsularuseandtheabilityofourremoteobservationswithourCRAGsunitofanactualsubsetofthefisherytotracklargertemporalpatternsofpeninsularwideeffort

TovalidatetherepresentativenessoftheCRAGShigh-frequencytrackingofeffortmetricsover timeweundertookaSpearmanrankcorrelationbetweentime-matchedeffortestimationsbyCRAGSandthe regional bus-route survey For graphing bus-routeextrapolationestimates were overlaid onto extrapolations from daily averages ofboatfishingeffortfromCRAGSWealsocarefullyloggedalltimespentontheprojecttocomparetotallaborcostsbetweenCRAGSandthebusrouteforbothdatacollectionandprocessingtodetermineeffortacrossthecomparativemonthaswellasforcostpersample

3emsp |emspRESULTS

The CRAGS operated successfully and continuously throughoutthestudycapturing545plusmn198SE(May)and483plusmn088SE(June)panoramasperdaymakinga totalof314 samplesWemade895observationsofboatsbut51samplesacross21daysreturnedzerocountsofboats(17oftotalsamples)Thesewereconcentratedonweekdaysmdashwith 44 zero-count samples across 18daysmdashwhile onweekendsandpublicholidaytherewereonlysevensamplesacross3dayswithnoactivityInclementweatherrendered27samplesun-usableTheimagesandbatch-processedTimeMachinemoviescon-sumed4274GBofmemory

(1)ei=IitimesT

(2)Ejk=

nsum

i=1

(

ei

120587k

)

(3)SEjk=

1

njminus1

sumn

i=1(eiminus ei)

2

njtimes (Njk)

2

(4)SEk=

radic

SE21k+SE2

2khellip+SE2

jk

(5)smt=etminus1+ et+ et+1

3

6emsp |emsp emspensp FLYNN et aL

Based on boat ramp interviews the average trailer boat partysize (P) was 287 people per vessel throughout May (plusmn0133 SEn=84)andpartysizebetweensitesshowednosignificantdiffer-ence (df=2F = 302p = 068) The averagemonthly boat fishingeffort(hours)extrapolatedforourCRAGSobservationsofHRMCAwas5629hr(plusmn711SE)with4096hr(plusmn607SE)inMayand1534hr(plusmn370 SE) in June (Figure4)mdashthis could be extrapolated to fisherhoursbymultiplyingbytheaveragepartysizebutasnointerviewswereconductedinJuneweleftboththeCRAGSandbusrouteex-trapolationsasboatfishingeffortonly

The total subset area observed by CRAGSwas 96km2 which comparedto318km2thatwaseasilyavailabletotrailerboatsaroundtheTasmanPeninsularwhichcorrespondstoaround3ofthetotalareaComparedtothetotalfishingeffortforMayextrapolatedfromtheregionalbus-routesurvey therelativelysmallareaaroundtheHRM(~3)receivedalargeamountofthefishingeffort(~23)

ThehighfrequencyofCRAGSsamplingpermittedtheexamina-tionoffine-scaleinter-dailytemporalvariationwhichdisplayedarag-gedsinusoidoscillation(Figure5)Effortwasgenerallylowwithinthe

weekdaystrataandthen increasedrapidly to largepeaksonweek-ends (Figure5) One exception to this was between the 18th and22ndMaywhichhadmoreeffortonweekdays thanotherperiods(Figure5)Bracketingthisperiodweretwoweekendsoffishingcom-petitions theldquoTunaClubofTasmaniarsquosrsquoFarSouthClassicrdquo (16ndash17thMay)andRally4Northerninvitationalweekendrally(23rdMay)

A total of 16 bus-route samples were conducted logistic fail-uresledtotheabandonmentoffour(20)oftheplannedsamplesRegionalTrailerboateffortfromaroundtheTasmanPeninsularwasestimatedtobe19650hr (plusmn2858SE) inMay (Figure4)Notrailersat individual rampswere recorded four timeswith all of theseoc-curringonweekdaysThetotalnumberoftrailerssightedperramprangedfrom0to16duringweekdaysand1ndash54forweekendswiththehighestnumberrecordedon16May2015atPiratesBayDuringhigh-intensitydays(particularlyweekendsholidaysmornings)therewereoftenlargeoverflowsfromrampcarparkswithparkedcarsandboattrailersextendingasmuchasonekilometerfromtheboatramp

During thebus-route survey 84 interviewswere conductedThemajorityofmariners(905n=96)indicatedfishingwastheir

F IGURE 4emspExtrapolatedmonthlytrailerboateffort(hoursplusmnSE)fortheentireTasmanPeninsulaassessedbyabus-routesurvey(May)andforthesubsetoftheareaaroundtheHRMCAsampledwithCRAGS(MayandJune)

F IGURE 5emspHigh-frequencyinterdailyextrapolatedCRAGSestimationsofboatefforthoursaroundthe(HRMCA)

0

100

200

300

400

500

600

Rolli

ng m

ean

daily

effo

rt in

hou

rs

DayWeekdays Weekends + Public Holidays

lyn088
Inserted Text
CA
lyn088
Cross-Out
lyn088
Cross-Out
lyn088
Inserted Text
t

emspensp emsp | emsp7FLYNN et aL

primaryactivity (Table1)Overall643 (n=54) indicated theirprimary target were tuna with 179 (n=15) specifically tar-geting southern bluefin tuna (Thunnus maccoyii)Other commonrecreationaltargetsincludedflathead(Platycephalusspp595)Southernrocklobster(Jasus edwardsii107)andsquidcalamari(Sepioteuthis australisandNototodarus gouldi476)(Table1)

Basedon the responsesofactivity type from interviews totalboatfishingeffortinMaywasestimatedtobe17783hr(plusmn2586SE)Asmostnonfishingactivitieswerenear shoreweassumed trailerboatingactivityatHMRCAobservedbyCRAGSwasallfishing

Therewere16directcomparisonsamplesbetweenthebusrouteandCRAGSmethodsbetweenthe2ndandthe29thofMaySamplesconsisted of 7 weekdays and 9 weekendspublic holiday samples

(Figure6)Infourcaseswheretwobus-routesamplesweretakenonthe same day (Supporting Information Appendix S2) they are com-bined toadailyestimation for tabulation (Table2)Theproportionaldifferencebetweenestimatedeffortsusingthetwomethodsrangedfrom078-to418-folddifferencehoweverranksestimatesbetweenCRAGS and bus-route samples were strongly associated (n=16ρ(12)=087p=lt001)Interdailysamplingstrata(ampm)werealsoexaminedseparatelyforcorrelationwhichremainedstrongforbotham(n=9ρ(7)=080p=lt001)andpm(n=7ρ(5)=089p=lt001)

ForthecomparativemonthCRAGStook169samplescomparedtothebusroutesrsquo16(Table3)WehadtwofieldtripdaysforCRAGSwith four people on the setup and three on the breakdown daywhichresultedinsevenfull-timeequivalent(FTE)daysofworkThiscomparedto16fielddaysforthebus-routemethodwhichincludedtraveltothefieldsiteforbetweentwoandthreestaffwhichaccu-mulatedto40FTEdaysofworkSettingupourCRAGSautomatedstitching took one FTE day (though this is faster for experiencedofficers)whiledataextractionperimagewasestimatedtotakeon

TABLE 1emsp Intervieweesreportedprimarytargettaxaoractivity

Primary target taxaspeciesactivity Interview responses

Tuna(unspecified) 37 440

SouthernBluefinTuna 15 179

AlbacoreTuna 2 238

AllTuna 54 643

Flathead 5 595

Southernrocklobster 9 107

Squidcalamari 4 476

Abalone 1 119

SmallPelagics 1 119

SnottyTrevalla 2 238

Allnearshorespecies 22 262

Surfing 1 119

SCUBA 3 357

DemonCavevisitors 3 357

PleasureCruise 1 119

Allnonextractive 8 952

Note ldquoTunardquo responsesdenote anglers targeting all specieswithin thefamilyThunnini

F IGURE 6emspDailyboatingeffort(hours)extrapolatedusingbus-routeaccesspointsurvey(bar)andCRAGS(line)Effortestimatefrombus-routesurveywascalculatedfrommorning(amdarkgray)andafternoon(pmlightgray)surveyCRAGSsurveywasconductedbetweenMayandJunewhilethebus-routesurveywasonlyconductedinMay

TABLE 2emspRankingofeffortforpairedsamplesbetweenCRAGSandthebusroute

DateCRAGS hours (rank)

Bus- route hours (rank) Δ Rank

2052015 330(9) 1208(10) minus1

3052015 303(8) 1471(11) minus3

12052015 0(1) 48(2) minus1

14052015 0(1) 24(1) 0

16052015 387(11) 1794(12) minus1

19052015 93(4) 89(4) 0

20052015 57(3) 117(5) minus2

22052015 228(7) 283(6) 1

23052015 513(12) 414(8) 4

24052015 345(10) 1155(9) 1

25052015 122(6) 323(7) minus1

29052015 102(5) 79(3) 1

8emsp |emsp emspensp FLYNN et aL

average75mincomparedto1minforenteringthebus-routedatasheetWhiledataextraction forCRAGS took longer compared tothe bus routemdashat 381 FTE daysmdashtotal monthly labor for CRAGStookonlyaquarterofthetimeandonaper-samplecomparisonwasonly25ofthebus-routelaborcost

4emsp |emspDISCUSSION

Over our autumn samplingmost trailer boat owners accessing thewaters around theTasmanPeninsula via the regional boating infra-structurewererecreationallyfishingwithfisherspredominatelytar-getingtunaTrailerboatrecreationalfishersconstitutealargetrancheof coastal users in Australia (McPhee etal 2002) and our resultssuggestedthatthisregionremainsonewhererecreationalfishing isconcentrated(MortonampLyle2004)EffortobservedbyCRAGSwasa spatial subsetof thegeneral areaeasily accessedby trailer boatsaroundtheTasmanPeninsularwhichweregionallyassessedwithourbus-routerovingsurveyAsCRAGSonlyobserved~3of thetotalarea but corresponded to ~23 of themonthly bus-route boat ef-fort theHippolyteRocksMarineConservationArea (HRMCA) is afurtherconcentrationpointwithintheregionforrecreationalfishingAsthisoffshorefishingeffortaroundtheHRMCAwasmostprobablyfocusedontotunamdashwhichaccountedfor634ofallinterviewedfish-ers activitiesmdashthe actual proportionofpeninsularwide tuna fishingthatwasobservedbyCRAGSwaspotentiallymuchhigherthan23

For developing effort metrics finding a representative site toobservemaybe important forprecise trackingof trendsThetightrankcorrelationbetweenpairedregionalbus-routeandCRAGScam-era samples suggest that effort atHRMCAwas representative ofregionaltemporaltrendsOurresultsarealsosimilartootherstud-ieswheresamplingmethodsforrecreationalfisheriesoverbroaderscaleswhencomparedtoindextrendsofeffortcollectedfrompointsourcesshowstrongcorrelations(Hartilletal2016)

CRAGSwithitsseascapescaleofdatacaptureobservesactualeffortonfishinggroundsratherthan indirectmetricsfrommove-ment past access points (Hartill etal 2016 Smallwood PollockWise Hall amp Gaughan 2012 van Poorten amp Brydle 2018) This

makesthechoiceofaccesspointstomonitorimmaterialfortrendcol-lectionandmaybeofparticularinterestforpelagicfisheriesWhilenonintuitiverecreationalfishingforwide-rangingmigratorypelagicspeciesisoftenspatiallyconcentratedintosmallareas(Lynch2006)duetotightoverlapsbetweeneaseofaccessbyfishersandfishbe-haviorrelatedtobathymetrymigrationhabitatandpreyavailability(Lynchetal2004PattersonEvansCarterampGunn2008)

Unlikespatialpatternstheintensityofcoastalrecreationaltemporaleffortcanbehighlyvariableovertime(Lynch20082014WiseTelferLaiHallampJackson2012)thoughitisgenerallythoughttofollowpre-dictablepatternsrelativetoholidayandnonholidayperiods(JonesampPollock2012)Ourresultsdemonstratedtheselargefluctuationsrang-ingfromzerouptogt400hroftrailerboateffortforadjacentdaysIncombinationwithholidayperiodsothersourcesofvariabilitymayalsooccurduetoindividualorcombinationsoffactorssuchasfishingcluborcompetitionactivitiesandweatherconditionsHigh-frequencysam-plingcanovercomethesecommontemporaldisjunctionsinobservedfishingeffortwherebychanceduetolownumbersofsampleslogis-ticallypermissiblewithtraditionalrovingoraccesspointsurveyslargeerrorsintheestimationofeffortcanbeintroduced(Hartilletal2016vanPoortenampBrydle2018vanPoortenetal2015)

With multiple daily observations abnormal episodes are morelikelytobeobservedForinstanceaweek-longperiodofintenseef-fortwasrecordedbetween15May2015and27May2015whichwasbracketedbytwoconsecutiveweekendswherefishingcompetitionswereheldontheTasmanPeninsularThenormalsinusoidpatternoflowweekdayandhighweekendeffortwasmuchlesspronouncedastheweekdaylullsineffortwerepartiallybridgedbyanglerscontinuingtofishAsbus-routedesignsusedaytypeasstratawithlowerprob-abilitiesforsamplingweekdaysandmuchfewersamplesoverallthistypeofeffectcouldeasilybemissedWhileourbus-routesamplesdidoverlapwiththetunafishingcompetitionsthiswaspurelybychanceSucheventswhichresultinlargespikesofeffortcanhavemarkedim-pactsonlocalecosystemsandharvestestimates(McPheeetal2002MonzPickeringampHadwen2013)andwithoutpriorknowledgearemorelikelytobecapturedwithmonitoringathigherfrequencies

Methodsforgatheringdataoncoastalusesuchasrecreationalfishing moved away from direct approaches and toward off-sitemethodssuchastelephoneinterviews(McCluskeyampLewison2008Pollock etal 1994) due to unsustainable costs from staffing andoperationsparticularlyduringperiodsof intenseusesuchasearlymornings weekends and public holidays which require overtimepaymentsHoweversamplingissueswithoff-sitemethodsareofpar-ticularconcernforopen-accessactivitiessuchasunlicensedfisheriesasthereisnolicensedatabaseofcontactphonenumbersonwhichtobaseasampleframeThisisparticularlysofornicherecreationalfisheriessuchaspelagicgamefishingasthefishersbecomeincreas-inglydilutewithinthegeneralpopulationandhencearehardtoac-cessparticularlyviaoff-sitemethodssuchastelephone interviews(Griffithsetal2010)

Remoteandautonomousdeploymentofsensorsmaythereforeprovideanalternativeandcost-effectivemethodforlong-termmon-itoring particularly for effort across a range of applicationsOur

TABLE 3emspFulltimeequivalent(FTE)labordaysforCRAGSandbusroutesurveystoestimatepatternsofeffort

CRAGS Bus route

Numberofsamples 169 16

Fielddays 2 16

Averagepeopleperfieldday 35 25

Field FTE days subtotals 7 40

StitchingsetupFTE 1 0

DataextractionFTE(persample) 0017 0002

Subtotal data extraction FTE 381 004

FTEdayspersample 006 250

Total FTE days 108 400

emspensp emsp | emsp9FLYNN et aL

comparisonsofmatchedpairsofdatafromtwomethodssuggestedthat theCRAGSoutputswerewell correlatedwith regionalefforttrendsOuranalysisoflabordemonstratesthattryingtounderstandthistemporalvariabilitywithsamplingviaboatrampsurveyswouldbeprohibitivelyexpensiveOurCRAGSallowsforatemporaleffortmetric derived fromactual observationsof fishing to be cheaplydeveloped with sampling frequency at subdaily scales This high-frequencysamplingapproachwasvalidatedagainstour largerbutmuchmorecostlybus-routesurveyOuranalysisoflaborrequiredtobothcollectandprocessdatashowedthatCRAGSrequiredconsid-erablylessresourcestocollectapproximatelyninetimesmorefre-quenttemporaldatathanthebus-routemethodMajorlaborsavingswereduetotheruggedizedandautonomouscapabilityofCRAGSasonlytwofieldtripswererequired(deploymentandretrievaltook7FTEdays)comparedtothebus-routersquos16fieldtrips inamonthwhichconsumed40FTEdaysandincludedanadditional780kmofinter-ramproadtravel(SupportingInformationAppendixS3)AsitisweatherproofautonomousandabletobeprogrammedwithatimedelaysetupCRAGSldquofireandforgetrdquonaturebothremovestheneedforobservertobephysicallylocatedattheregiontocountactivity(EdwardsampSchindler2017)andworkduringovertimeperiodsTheldquoblankingperiodrdquowhichstopsandstartssamplingmdashinthiscaseover-nightmdashalso reduces battery and memory use In combined thesetechnological advances have allowed long-term CRAGS deploy-mentsacross4yearsof4ndash6monthrotationsontooffshoreislandsand for several seasons in Antarctica to monitor seabirds (Lynchetal2017)Thelaborestimateforthebusroutewasconservativeasitdidnotincludeovertimeloadingforthe6weekenddaysofsam-plingorearlymorningcommencementofamsamples

Theadditional381FTEprocessingtimerequiredbythecameraapproach did not impact greatly on the overall labor budgetwhenconsideredagainstfieldworkTheuseoftheTimeMachinebatchpro-cessingsoftwaretoautomaticallyproducetheldquobigdatardquointeractivemoviewasalsoatechnologicalinnovationthatsignificantlyreducedthelaborofdatahandlingcomparedtothemanualopeningofindivid-ualGigaPansusedinpreviousstudies(Lynchetal2015Woodetal2016)IncomparativetermsattheindividualsampleleveltotracktheeffortmetriceachCRAGSsamplerequired24ofthelaborofthebus-routesurvey

Thedevelopmentandapplicationofimagerecognitionsoftware(machine learning) to detect classify and count boats automati-cally (Bousetouane amp Morris 2015) would further reduce post-processingcostsTherequiredfrequencyofsamplingshouldalsobeconsideredonacase-by-casebasisdependentontheresearchquestion as image collection can be somewhat open-ended andresult inunwieldydata sets that require subsampling tobe cost-effective(Parnelletal2010)Inourcasethesettingof~5imagesperdayappearedtoadequatelydescribetheinterdailyvariation

Whileouroneautonomouscameraprovidedastrongcorrelationwithtemporaltrendsineffortcomparedtotheregionalsurveyitdidnotprovide relativespatialdistributionsofeffortacross the fisheryThesedistributionshowevercanbepredictableandhighlyconcen-trated(Lynch20062014Parnelletal2010)anobservationthatwas

alsoshownbyourresultsandmuchsmallerlevelsofsamplingeffortwouldprobablyberequiredtoresolvevariationinspatialdistributionscomparedtowhatisrequiredtoresolvethehightemporalvariationsineffortUnderstandingfine-scalespatialdistributionsofrecreationalfishingeffortiscommonlyachievedusingaerialcounton-watersur-veysorvantagepointsurveys(Lynch2006Smallwoodetal2012)thoughotherremotemethodssuchashigh-resolutionsatelliteimag-ery(FretwellScofieldampPhillips2017)mayprovideacheapersolu-tionThesetypesofwidersurveyshavealsobeensuggestedashighlycompatiblewithpointsourcemetricssuchasremotecamerasystemstoallowforcalibrationorldquoup-ratingrdquoforeffortexpansionstoestimateoveralleffortwithinthefishery(Hartilletal2016)

Thelongdistance(~6km)betweentheHMRCAandthedeploy-ment location stretched the resolution limit of the current CRAGSsetup aswewere unable to distinguish party size or activity typepurelyusing thecamerasystemThis isunlikepreviousCRAGSde-ploymentwhichhadahighsuccessrateofidentifyingpartysizeac-tivity typeata shorter rangeofapproximately19km (Woodetal2016) Thus ground-truthing information about the fisherywas re-quiredforthesystemtoworkeffectivelyAutonomouscamerasystemalsodonotcollectcreel(catch)databutbyallowingforcost-effectivecollectionofthehighlyvariabletemporaleffortdatatheymayallowforhumanresourcestobespentelsewhereduringsurveyperiods(egtoundertakeboatrampinterviews)(vanPoortenampBrydle2018)

Our studywas a small-scale trial of new technology andwhilepreliminary results are encouraging validated seasonal effort esti-mationsobservedviaCRAGSwereoutsidethescopeofthisprojectOur comparisons would hence have been improved by extendingCRAGSdeploymentsandwithmorecomparisonstoregionalsurveysSimilarly usingmultiple CRAGS simultaneously would have bettertested for differences in subregional patterns of temporal trendsorconcentrationof infishingeffortfor instanceboatsaroundthenearbyTasmanIsland

Insummarytheprogrammablesamplinganddataextractionof-feredbyhigh-resolutionautonomouscamerasystemscanallowforfine-scalelong-termmonitoringoffishingefforttrendsorpotentiallymanyothertypesofactivitiesatseascapescalesMorewidespreadadoptionofhigh-frequencysamplingapproachesforeffortestima-tionwouldalsohelpavoidcollectionofmisrepresentativedatabyidentifyingeffortanomalies(McCluskeyampLewison2008)suchastheweekdaycontinuationeffectofraisedeffortbetweenadjacentweekendfishingcompetitionByimprovingtheinformationdensityovertraditionalsurveymethodsautonomouscamerasystemsandotherremotesystemsmayprovideanewwaveofoptimizationfordirectmeasurementofcoastalhumanusesuchasrecreationalfish-eriesalthoughcomplimentarymethodsarestillrequiredtodeter-minespatialdistributionstargetspeciesandharvest

ACKNOWLEDG MENTS

We thankMsGeorginaWoodMs SaraAdkinsMsBelleMonkMrRSherringtonandMsPaigeKellyforconstructiveinsighten-thusiasmandassistanceduringfieldworkWealsothankMrPhillip

10emsp |emsp emspensp FLYNN et aL

CulversonMrStuartDudgeonandMrMattStapenellKarenEvensprovidedhelpfulcommentsonaninternaldraftGreatestthankstothemany recreational anglerswhowillingly shared information tosurveyclerkswithouttheirinvolvementthisprojectwouldnothavecometofruitionThreeanonymousreviewersprovidedcommentsthatimprovedthemanuscript

CONFLIC T OF INTERE S TS

Theauthorsdeclarenoconflictofinterests

AUTHOR CONTRIBUTIONS

MrFlynnhelpeddesignthestudyledthefieldworkundertookmostoftheanalysisandhelpwritethemanuscriptDrLynchhelpdesignthestudyassisted inthefieldworkprovidedadviceontheanaly-sisanddidsomeanalysisconstructedfiguresandafterMrFlynncontributedthemosttothedraftingofthemanuscriptDrBarretthelpeddesign the study assisted in some fieldwork provided ad-viceonanalysisandeditedthemanuscriptMrWonghelpedexten-sivelywithfieldworkassistedwithanalysisconstructedfiguresandhelpedwritethemanuscriptMsDevineassistedwithfieldworkandconfigurationoftheequipmentincludingtheapplicationofthebigdatamoviesoftwareandhandlingofphotographsandalsohelpedwritethemanuscriptMrHughesdevelopedtheelectronicsfortheequipmentoversaw thebuildofgearwrote the software topro-gramtherobotandreviewedandeditedthemanuscript

DATA ACCE SSIBILIT Y

Survey sampling schedules data collection sheets extrapolationproceduresandtechnicalCRAGSspecificationsarealluploadedasonlinesupportinginformationappendixRawdatafromsurveysandCRAGSDOIhttpdoiorg105061dryadv10d218

RE SE ARCH E THIC S COMPLIANCE AND FUNDERS

ResearchabidesbytheAustraliancodeofhumanresearchandex-perimentation(NationalStatementonEthicalConductinHumanResearch of 2007) Permission to conduct research on humanparticipants was granted by the University of Tasmania SocialScience under reference H0014772mdashldquoRemote investigation ofcoastal area user group quantification and variationrdquo aswell asfromtheCSIROSocialScienceHumanResearchEthicsCommitteeunder reference 08115mdashldquoHawkesbury Region Project HumanUsePatternsrdquoFundingwasbyUTASasPartoftheIMASHonoursProject funding and through internal support from the CoastsProgram

ORCID

Tim P Lynch httporcidorg0000-0002-5346-8454

R E FE R E N C E S

Alessa L Kliskey A amp Brown G (2008) Socialndashecological hotspotsmapping A spatial approach for identifying coupled socialndasheco-logicalspaceLandscape and Urban Planning8527ndash39httpsdoiorg101016jlandurbplan200709007

ArlinghausR(2006)Overcominghumanobstaclestoconservationofrecreational fishery resources with emphasis on central EuropeEnvironmental Conservation 33 46ndash59 httpsdoiorg101017S0376892906002700

Arlinghaus R Tillner R amp Bork M (2015) Explaining participationratesinrecreationalfishingacrossindustrialisedcountriesFisheries Management and Ecology 22 45ndash55 httpsdoiorg101111fme12075

BadcockPBPatrickKSmithAMASimpsonJMPennayDRisselCEhellipRichtersJ(2016)Differencesbetweenlandlineandmobilephoneusers in sexualbehavior researchArchives of Sexual Behavior46(6)1711ndash1721

Blumberg S J amp Luke J V (2009) Reevaluating the need for con-cern regarding noncoverage bias in landline surveys American Journal of Public Health 99 1806ndash1810 httpsdoiorg102105AJPH2008152835

Bousetouane F ampMorris B (2015) Off-the-shelf CNN features forfine-grainedclassificationofvesselsinamaritimeenvironmentInGBebisRBoyleBParvinDKoracinIPavlidisRFerisTMcGrawMElendtRKopperERaganZYeampGWeber(Eds)Advances in visual computing 11th International symposium ISVC 2015 Las Vegas NV USA December 14ndash16 2015 proceedings Part II (pp379ndash388)Cham Switzerland Springer International Publishing httpsdoiorg101007978-3-319-27863-6

Cooke S J amp Cowx I G (2004) The role of recreational fish-ing in global fish crises BioScience 54 857ndash859 httpsdoiorg1016410006-3568(2004)054[0857TRORFI]20CO2

DiggleP J (1995)Time-series a biostatistical introductionOxfordUKClarendonPress

EdwardsJampSchindlerE (2017)Avideomonitoringsystemtoeval-uate ocean recreational fishing effort in Astoria OregonMarine recreational information program (pp 31) Newport OR OregonDepartmentofFishandWildlife

FarrMStoecklNampSuttonS(2014)RecreationalfishingandboatingArethedeterminantsthesameMarine Policy47126ndash137httpsdoiorg101016jmarpol201402014

ForbesETraceySampLyle JM (2009)Assessment of the 2008 rec-reational gamefish fishery of Southeast Tasmania with particular refer-ence to Southern Bluefin TunaHobartTASTasmanianAquacultureandFisheriesInstituteUniversityofTasmania

FretwellPTScofieldPampPhillipsRA(2017)Usingsuper-highres-olutionsatelliteimagerytocensusthreatenedalbatrossesIbis159481ndash490httpsdoiorg101111ibi12482

GiriKampHallK (2015)SouthAustralian recreational fishingsurveyFisheries Victoria internal report series (pp 63) Queenscliff VicFisheriesVictoria

Griffiths S P Pollock K H Lyle J M Pepperell J G TonksM L amp Sawynok W (2010) Following the chain to elu-sive anglers Fish and Fisheries 11 220ndash228 httpsdoiorg101111j1467-2979201000354x

Halpern B SWalbridge S Selkoe K A Kappel C V Micheli FDrsquoAgrosaChellipWatsonR(2008)AglobalmapofhumanimpactonmarineecosystemsScience319948ndash952httpsdoiorg101126science1149345

Hartill BW Payne GW Rush N amp Bian R (2016) Bridging thetemporal gap Continuous and cost-effective monitoring of dy-namic recreational fisheries by web cameras and creel surveysFisheries Research 183 488ndash497 httpsdoiorg101016jfishres201606002

emspensp emsp | emsp11FLYNN et aL

HenryGWampLyleJM(2003)NationalrecreationalandindigenousfishingsurveyFinal report to the fisheries research amp development cor-poration and the fisheries action program (pp 188) Canberra ACTFRDC

JonesCMampPollockKH(2012)Recreationalanglersurveymeth-odsEstimationofeffortharvestandreleasedcatchInAZaleDLParrishampTMSutton(Eds)Fisheries techniquesBethesdaMDAmericanFisheriesSociety

Kearney R E (2002) Co-management The resolution of conflict be-tween commercial and recreational fishers in Victoria AustraliaOcean amp Coastal Management45201ndash214httpsdoiorg101016S0964-5691(02)00055-8

KellerKSteffeASLowryMMurphyJJampSuthersIM(2016)Monitoringboat-basedrecreationalfishingeffortatanearshorearti-ficialreefwithashore-basedcameraFisheries Research18184ndash92httpsdoiorg101016jfishres201603025

KinlochMAMcGlennonDNicollGampPikePG(1997)Evaluationof the bus-route creel surveymethod in a largeAustralianmarinerecreationalfisheryISurveydesignFisheries Research33101ndash121httpsdoiorg101016S0165-7836(97)00068-4

LewinWCArlinghausRampMehnerT(2006)Documentedandpo-tential biological impacts of recreational fishing Insights forman-agementandconservationReviews in Fisheries Science14305ndash367httpsdoiorg10108010641260600886455

LowryMampMurphy J (2003)Monitoring the recreational gamefishfisheryoffsouth-easternAustraliaMarine and Freshwater Research54425ndash434httpsdoiorg101071MF01269

LyleJMStarkKEampTraceySR(2014)2012ndash2013 Survey of rec-reational fishing in TasmaniaHobart TAS Institute forMarine andAntarcticStudies

Lynch T P (2006) Incorporation of recreational fishing ef-fort into design of marine protected areas Incorporacioacuten delEsfuerzo de Pesca Recreativa en el Disentildeo de Aacutereas MarinasProtegidas Conservation Biology 20 1466ndash1476 httpsdoiorg101111j1523-1739200600509x

LynchTP(2008)Thedifferencebetweenspatialandtemporalvaria-tioninrecreationalfisheriesforplanningofmarineprotectedareasResponse toSteffeConservation Biology22486ndash491httpsdoiorg101111j1523-1739200800888x

LynchTP (2014)Adecadal time-seriesof recreational fishingeffortcollectedduringandafterimplementationofamultipleusemarinepark shows high inter-annual but low spatial variability Fisheries Research15185ndash90httpsdoiorg101016jfishres201309014

LynchTPAldermanRampHobdayAJ(2015)Ahigh-resolutionpan-orama camera system formonitoring colony-wide seabird nestingbehaviourMethods in Ecology and Evolution6491ndash499httpsdoiorg1011112041-210X12339

LynchTFlynnDWoodGKimJ-HParkJampHobdayAJ(2017)Metrics parameters and results from fisheries and wildlife studies with the CSIRO Ruggedized Autonomous Gigapixel Systems (CRAGS)DarwinNTAustralianMarineScienceAssociation

LynchTPWilkinsonEMellingLHamiltonRMacreadyAampFearyS(2004)ConflictandimpactsofdiversandanglersinamarineparkEnvironmental Management 33 196ndash211 httpsdoiorg101007s00267-003-3014-6

MaHOgawaTKSminkeyTRBreidtFJLesserVMOpsomerJDhellipVanVoorheesDA(2018)Pilotsurveystoimprovemonitoringofmarine recreational fisheries inHawaiʻiFisheries Research204197ndash208httpsdoiorg101016jfishres201802010

McCluskey S M amp Lewison R L (2008) Quantifying fish-ing effort A synthesis of current methods and their ap-plications Fish and Fisheries 9 188ndash200 httpsdoiorg101111j1467-2979200800283x

McGlennonDampKinlochMA(1997)Evaluationofthebus-routecreelsurveymethod in a largeAustralianmarine recreational fishery II

PilotsurveysandoptimalsamplingallocationFisheries Research3389ndash99httpsdoiorg101016S0165-7836(97)00067-2

McPheeDLeadbitterDampSkilleterGA(2002)SwallowingthebaitIs recreational fishing in Australia ecologically sustainable Pacific Conservation Biology840httpsdoiorg101071PC020040

Monz C A Pickering C M amp HadwenW L (2013) Recent ad-vances in recreation ecology and the implications of differentrelationships between recreation use and ecological impactsFrontiers in Ecology and the Environment11441ndash446httpsdoiorg101890120358

MooreAHallKGiriKTraceySPenroseLHansenShellipBrownP(2015)Developingrobustandcost-effectivemethodsforestimatingthenationalrecreationalcatchofSouthernBluefinTunainAustraliaFRDC Projectpp123

Morton A J amp Lyle J M (2004) Preliminary assessment of the rec-reational gamefish fishery in Tasmania with particular reference to Southern Bluefin Tuna Taroona TAS Tasmanian Aquaculture andFisheriesInstitute

ParnellPEDaytonPKFisherRALoarieCCampDarrowRD(2010)SpatialpatternsoffishingeffortoffSanDiegoImplicationsfor zonal management and ecosystem function Ecological Applications202203ndash2222httpsdoiorg10189009-15431

PattersonTAEvansKCarterTLampGunnJS(2008)Movementand behaviour of large southern bluefin tuna (Thunnus maccoyii)in the Australian region determined using pop-up satellite ar-chival tags Fisheries Oceanography 17 352ndash367 httpsdoiorg101111j1365-2419200800483x

PollockKH JonesCMampBrownTL (1994)Angler survey meth-ods and their applications in fisheries management Bethesda MDAmericanFisheriesSociety

PonsM Branch T AMelnychukM C JensenO P Brodziak JFromentinJMhellipHilbornR(2017)EffectsofbiologicaleconomicandmanagementfactorsontunaandbillfishstockstatusFish and Fisheries181ndash21httpsdoiorg101111faf12163

PowersSPampAnsonK (2016)Estimating recreationaleffort in theGulf of Mexico Red Snapper Fishery using boat ramp camerasReduction in federalseason lengthdoesnotproportionally reducecatchNorth American Journal of Fisheries Management 36 1156ndash1166httpsdoiorg1010800275594720161198284

Robson D amp Jones C M (1989) The theoretical basis of an ac-cess site angler survey design Biometrics 45 83ndash98 httpsdoiorg1023072532036

Rocklin D Levrel H Drogou M Herfaut J amp Veron G (2014)Combining telephone surveys and fishing catches self-reportTheFrenchseabass recreational fisheryassessmentPLoS ONE9e87271httpsdoiorg101371journalpone0087271

Smallwood C B Pollock K HWise B S Hall N G amp GaughanD J (2012) Expanding aerial-roving surveys to include counts ofshore-based recreational fishers from remotely operated camerasBenefitslimitationsandcosteffectivenessNorth American Journal of Fisheries Management321265ndash1276httpsdoiorg101080027559472012728181

Teixeira D ZischkeM T ampWebley J A C (2016) Investigatingbias in recreational fishing surveys Fishers listed in public tele-phone directories fish similarly to their unlisted counterpartsFisheries Research 181 127ndash136 httpsdoiorg101016jfishres201604012

van Poorten B T amp Brydle S (2018) Estimating fishing effortfrom remote traffic counters Opportunities and challengesFisheries Research 204 231ndash238 httpsdoiorg101016jfishres201802024

van Poorten B T Carruthers T RWardHGM ampVarkeyD A(2015)Imputingrecreationalanglingeffortfromtime-lapsecamerasusinganhierarchicalBayesianmodelFisheries Research172265ndash273httpsdoiorg101016jfishres201507032

12emsp |emsp emspensp FLYNN et aL

VenturelliPAHyderKampSkovC(2017)AnglerappsasasourceofrecreationalfisheriesdataOpportunitieschallengesandproposedstandardsFish and Fisheries18578ndash595httpsdoiorg101111faf12189

WiseBSTelferCFLaiEKMHallNGampJacksonG(2012)Long-termmonitoringofboat-basedrecreational fishing inSharkBayWesternAustraliaProvidingscientificadviceforsustainablemanagement in a World Heritage Area Marine and Freshwater Research631129ndash1141httpsdoiorg101071MF12054

WoodGLynchTPDevineCKellerKampFigueiraW(2016)High-resolutionphoto-mosaictime-seriesimageryformonitoringhumanuseofanartificialreefEcology and Evolution66963ndash6968httpsdoiorg101002ece32342

YuFRupertACCDiTizianaMNishikantaKRandySIllahNhellipJeffreyG(2011)TerapixelimagingofcosmologicalsimulationsThe Astrophysical Journal Supplement Series19718

Zischke M T Griffiths S P amp Tibbetts I R (2012) Catch and ef-fort from a specialised recreational pelagic sport fishery off

eastern Australia Fisheries Research 127ndash128 61ndash72 httpsdoiorg101016jfishres201204011

SUPPORTING INFORMATION

Additional supporting information may be found online in theSupportingInformationsectionattheendofthearticle

How to cite this articleFlynnDJHLynchTPBarrettNSWongLSCDevineCHughesDGigapixelbigdatamoviesprovidecost-effectiveseascapescaledirectmeasurementsofopen-accesscoastalhumanusesuchasrecreationalfisheriesEcol Evol 2018001ndash12 httpsdoiorg101002ece34301

Graphical AbstractThecontentsofthispagewillbeusedaspartofthegraphicalabstractofhtmlonly

Itwillnotbepublishedaspartofmainarticle

Wetrailedaremoteautonomousultra-high-resolutionphotosamplingmethodasacosteffectsolutionfordirectseascape-scaledmeasure-mentsOursequentialphotosamplingwasbatchedprocessesusingnovelTimeMachinedatahandlingsoftwaretoproduceldquobigdatardquotimeseriesmoviesfromaspatialsubsetofanoffshorefisherywhichwecomparedtoaregionalbus-routesurveyofboattrailersandinterviewswithparticipantsOurcamerasamplescloselytrackedlargerscaleregionaltrendsbutwithlaborcostsaquarterofthebus-routesurveycostsforanindividualsample25ofthebusroutes

Page 2: Gigapixel big data movies provide cost-effective seascape

2emsp |emsp emspensp FLYNN et aL

1emsp |emspINTRODUC TION

Worldwidecoastalzonesaresomeofthemostheavilyimpactedpartsoftheoceanbypeoplebothintermsofactivitytypesandparticipationrates(Halpernetal2008)Aparticularfocusofre-search on coastal use has been recreational fishing as this pop-ular activity can significantly affect fish and invertebrate stocks(Arlinghaus 2006 Cooke amp Cowx 2004 Lewin Arlinghaus ampMehner2006McPheeLeadbitterampSkilleter2002)asformanyspecies harvest can exceed the take of the commercial fishery(GiriampHall2015LyleStarkampTracey2014ZischkeGriffithsampTibbetts2012)

Unlikecommercialfisherieswhicharecommonlydocumentedbyoperatorsloggingtheircatchandeffortassessmentsofopen-accessnonreporting activities suchas recreational fisheries require sam-plingThiscanbebothdifficultandexpensiveduetotheiroftenlargespatialextenthightemporalvariationandforthefisheriesagenciesundertaking the work which often requires weekend and holidaysurveyinghighlaborcosts(Maetal2018PollockJonesampBrown1994 Rocklin Levrel Drogou Herfaut amp Veron 2014 VenturelliHyderampSkov2017)Due to these issues assessmentshave typi-callyrelieduponindirectdatacollectionfromoff-sitetelephonesur-veysbasedondataframesdevelopedaroundregionallsquowhitepagesrsquoor landline telephonedirectorieswhichtraditionallyhavehadhighresponseratesandeasilyscalablesamplesizes(Mooreetal2015Pollocketal1994)Suchapproachesmaynowriskundersamplingactivefishersbecauseofdemographicallybiasandoveralldecreasedlandlinephoneuseunlistedandnonregionalcodedmobilephonesandlowresponsesuccessfromvoicecallstomobilephones(Badcocketal 2016 Blumberg amp Luke 2009 Teixeira Zischke ampWebley2016)

Other formsof recreational fisheryassessmentsemploydirecton-sitesurveymethodssuchasbus-routesurveyswhichuseclerkstraveling around a fisherymdashfollowing randomly selected predeter-minedschedulesmdashinterviewingandobservingfisherstotrackfish-ingeffortcatchandrelease(McGlennonampKinloch1997Pollocketal 1994) On-sitemethods that cover broad geographic areashowever are seldom performed due to high labor costs (JonesampPollock2012)

Given these limitations remote sensing tools and technol-ogies such as autonomous photography are an emerging fieldofferingdirectmonitoringasanalternativeorsupplementtoon-site interviews (HartillPayneRushampBian2016KellerSteffeLowryMurphyampSuthers2016ParnellDaytonFisherLoarieampDarrow2010PowersampAnson2016vanPoortenCarruthersWardampVarkey 2015Wood LynchDevineKelleramp Figueira2016) One such system is the CSIRO Ruggedised AutonomousGigapixel System (CRAGS) which is a programmable weather-proofcameratrapthatprovidesultra-highresolutiontime-lapseandhigh-frequencypanoramic imagesTheCRAGSutilizesmod-ified commercially available hardware and software known as aGigaPanreg (httpgigapancom)tocapturephotographicsampleswith such high pixel density (gigapans) that wide fields of view

andassociatedobjectsofinterestscanbeexaminedcloselywith-out the loss of broader environmental context at landscape orseascape scales (LynchAldermanampHobday 2015) (SupportingInformationAppendixS1)

Thisroboticcamerasystemallowsformuchbroaderscaledphoto-graphicassessmentsthansimplecameratrapsbyautonomouslytak-ingandthenstitchingtogethermultipletelephotomegapixelimagesinto high-resolution gigapixel panorama The resulting tiled imageallowsfullyzoomableviewingforeithermultiplesmalltargetssuchasAlbatrossnestsacrossacolony(Lynchetal2015)ormorewidelyspacedlargertargetssuchasidentifyingcommercialvsrecreationalvessels around an offshore artificial reef (Wood etal 2016) Thebenefitofthesystemcomparedtosimplercameratrapsisbeingabletoobserve inhighdetail fromtheone imagestreammanyobjectssimultaneously across landscapeor seascape scalesWith somuchinformationdatahandlingcanbecomeoverwhelmingsoabatchpro-cessingmethod calledGigapanTimemachinehasbeendevelopedOriginallyusedforviewingcosmologicalsimulations(Yuetal2011)TimeMachineproducesavideostreamthatallowsviewerstofluidlyexploregigatotetrapixel-scaledvideosacrossbothspaceandtime

InAustraliaannualrecreationalfishingparticipationhasbeenes-timatedat195(HenryampLyle2003)whichiswellabovetheglobalaverageofaround10(ArlinghausTillnerampBork2015)AroundtheAustralian island state of Tasmania the annual participation rate is293whichwellexceedsthenationalaverageTheTasmanPeninsulainTasmaniarsquossoutheast (Figure1) isapopularfishingregionneartothelargestcityandstatecapitalofHobartwhereasteepbathymetricprofileandmigratorygame-fishpathwayoverlapprovidingcoastalac-cesstonormallyoffshorefishingopportunitiestorecreationalfisherintrailerboatsPelagicgamefishsuchassouthernbluefintuna(Thunnus maccoyii)aretargetedinlateAustralSummerandAutumn(MortonampLyle2004)butasthefisheryisunlicensedandepisodictrendsinef-fortandcatcharepoorlyunderstood(LowryampMurphy2003Mooreetal2015)Thisparticularrecreationalfisheryisalsoofmoregeneralinterestasthetargetsarecommerciallyimportantspecieswhichin-cludethosewithinternationallynegotiatedquotas(Ponsetal2017Zischke etal 2012) Recreational fishers are also often associatedwithotherusersofthemarineenvironment(FarrStoecklampSutton2014Kearney2002)orinthiscasetouristsenjoyingthelocalnationalparksHencehighusesitesmaybeimportanttomonitornotonlyforfisheriesandecologicalreasonsbutformanagingsocialvaluescon-flict resolutionandplanning (AlessaKliskeyampBrown2008Lynchetal2004)

WetrailedourCRAGSmethodtoseewhetherwecoulddevelopa high-frequency time series of temporal trends of observed trailerboat fishingeffortatanoffshoreconcentrationpoint for the recre-ationalgamefisheryWealsotestedboththesuitabilityandlaborcost-effectivenessofthehardwareandanovelapplicationofTimeMachinebatchprocessingsoftwaretoautomaticallyproducealdquobigdatardquointer-activemoviesmadefromourgigapixelpanoramasAsthepeninsulais isolatedandservicedbyonlya limitednumberofboatrampsweaimed tovalidate the representativenessof the temporal trendsweobservedwithCRAGSbycorrelatingtomatchsamplesobtainedfrom

lyn088
Cross-Out
lyn088
Cross-Out
lyn088
Inserted Text
M

emspensp emsp | emsp3FLYNN et aL

asimultaneousregionallyscaledbus-routetrailerboatsurveyatthethreerampsonthepeninsularsuitableforlaunchinglargetrailerboatsWe also undertook on-ground interviewswith trailer boat users togroundtruthactivitytype(fishingornonfishing)andtargetspeciesByundertakingbothcameraandboatrampsurveysourover-overarchingaimwastoinvestigatehowremoteimagerymayreplacecomplimentoroptimizeconventionalon-siteapproachesforrecreationalfishing

2emsp |emspMATERIAL S AND METHODS

WedeployedtheCRAGSbetween1May2015and30June2015toassesstrailerboatsoffshorefromwateradjacenttotheTasman

Peninsula (Figure1) This period was chosen as it correspondswith the main tuna fishing season For large trailer boats (gt5mlength) localaccess to the fishery is limitedto threeboat ramps(Figure1)althoughtheareamayalsobevisitedbylargervesselssteaming frommarinas and anchorages further north (MortonampLyle2004)butforthepurposesofourstudyweexcludedtheserelativelyrare largervesselsWefocusedourCRAGSsurveyataknown game fishing concentration the Hippolyte Rock MarineConservationArea (HRMCA)which are a group of small islandsabout6-kmoffshore (Figures1and2)aroundwhichrecreationalfishingispermittedWhiletheCRAGSdeploymentwasmarginallycloser to the FortescueBayRampwe commonly observe boatsapproachingtheHRMCAfromdirectionsthatcorrespondedtothe

F IGURE 1emspTheTasmanPeninsulainSETasmaniaBus-routesurveysamplingoccurredatramps locatedatPiratesFortescueandStuartsBayTheCRAGSlocationisshownasacameraiconandtheareaobservedbythecamerawhichincludestheHippolyteRocksMarineConservationArea(HRMCA)isdescribedasatriangle

Text

Pirates Bay

Fortesque Bay

StewartsBay

0 3 6 9 1215Kilometers

0 30 60 90 12015Kilometers

Tasman Is

Australia

Tasmania

Hobart

TasmanPeninsular

HippolyteRocksMCA

43deg 01 30

147deg 52 30

TAS

4emsp |emsp emspensp FLYNN et aL

locationof all three rampsWe fixed theCRAGS to a clifftop atWaterfall Baywhich overlooks the site fromvery high sea cliffs(~250m)withthecamerafocusingonthewesternsideofHRMCA(Figure1)ForeachsampleCRAGScapturedaseriesof68pho-tographsthroughatelephotolenswhichwhenstitchedtogetherprovide a high-resolution panoramic image with a wide field ofview (~140deg) (Supporting Information Appendix S1) The size ofareasurveyedbyCRAGScomparedtotheregionalareaaccessibleby trailer boatswas estimated using google earth polygons andtheUniversityofNewHampshireKMLextensiontoolmdashhttpsex-tensionunhedukmlToolsindexcfm

Thetimeof the initialCRAGSpanoramasamplewasrandomlyselected Samples were then collected sequentially through-out daylight hours (0600ndash1800) As less recreational fishing fortuna occurred at night no samples were taken with the CRAGSprogrammed to ldquoblankrdquo these hours from the sampling sched-ule (Supporting InformationAppendix S1) Photographic sessionswereprogrammedtobeseparatedbyarecurring93-mingapthus

producing a forward shift in sample capture time andminimizingpotentialautocorrelationsbetweenboatingactivitiesandsamplingtimesPhotographswerebatchprocessedtobothformpanoramicimagesforeachsampleandforcompilingintoaninteractivetime-lapsemovieusingCREATELabrsquosTimeMachinesoftware (CREATELabregPittsburgPAUSA)

TimeMachinemovies are fully interactive and are able to bezoomedintopausedandplayedatvariousspeedsandarealsotimeanddate-stamped(Figure3)UsingthisinteractivemoviesoftwareeachGigaPanmoviewassystematicallypausedateachsamplefullyzoomedandscannedlefttorighttoptobottomforobjectsintheseascapebytheleadauthor(Figure3)Vesselcountswererecordedforeachsamplewithobjectslargerthan~10minlength(nottrailerboats)or requiring longer than10seconds todistinguish (eg lowresolutionandimageartifact)beingdiscountedSampleswerecat-egorized into two strataweekdays andweekendspublic holidaysusing the time and date stamp provided on each GigaPan scene(Figure3)

F IGURE 2emspAnunmodifiedGigaPanEpicPROunit(left)andaCRAGSdeployedatWaterfallBaySETasmania(right)

F IGURE 3emspScreencaptureoutputsfromtheTimeMachineinteractivedataviewershowinganozoomandpanoramicextentoftheHippolyteRockMarineConservationArea(HRMCA)b=13zoommainHippolyterock~6kmfromcamerac=12zoomofalargetrailerboatd=fullzoom

lyn088
Cross-Out
lyn088
Inserted Text
(a)
lyn088
Cross-Out
lyn088
Inserted Text
(b)
lyn088
Cross-Out
lyn088
Inserted Text
(c)
lyn088
Cross-Out
lyn088
Inserted Text
(d)
lyn088
Inserted Text
s

emspensp emsp | emsp5FLYNN et aL

Thecountofboatsforeachsamplewasthenextrapolatedintoanestimateofdailyaverageboatingeffort(hoursplusmnSE)usingthein-stantaneouscountmethod(Equation1)

where ei is theextrapolatedboatinghours for the ithday Ii is theaveragenumberoftrailerboatsobservedacrosstheinstantaneoussamplesfordayiandTisthedailysamplingperiod(daylighthours)Asbadweathercanbeamajorfactoraffectingthedecisiontoboatorfish(ForbesTraceyampLyle2009)boatingeffortwasassumedtobe0whenadverseweathermadeobservationsofboatsimpossible

Next total effort for each day-type strata (Ejk) within sam-pling months (k) was calculated to estimate the monthly effort(Equation2)AsCRAGSallowedforcontinuoussamplingthroughoutthesurveyperiodthesamplingprobability(πk)wassetto1

Standarderrorandeffortvarianceformonthlyestimateswerecalculated(Equation3)accordingtoPollocketal(1994)forstrata

where nj is thenumberofsampleddaysofstrata j Njk is thetotalnumberofdaysofstratajinmonthk

Asvariationforeachstratawascalculatedseparatelyweusedasumofsquaredstandarderrorsapproach(Equation4)toestimatethetotalmonthlyvariation

Assampleswereobtainedinaserialandcontinuousfashionatimeseriesmodelusingrollingaverageswasalsoappliedtothedata(Diggle1995)(Equation5)

where smt is the smoothed average of extrapolated effort e persampletwhichisweightedacrossthreesampleseachseparatedby93minThisprovidedasmoothedaveragewithindays

Wealsoconductedabus-routestylesurveytoestimatefishingeffortfromtrailerboatsfortheregion(Pollocketal1994Robsonamp Jones1989)WhileCRAGSsampleswerecollected throughoutMayandJunethecomparablebus-routesurveywasonlyconductedinMaydue to logistical constraintsWesurveyed the threemajorboatrampsintheregionatPiratesBayFortescueBayandStewartsBaybetween3May2015and30May2015(Figure1)

Atotalof20bus-routesampleswerescheduledWeundertookrandomstratifiedsamplingwithdaytypedividedintoweekdaysorweekendpublic holidays To reflect expected higher recreationaleffortduringweekendsandpublicholidays(McCluskeyampLewison2008)thesamplingprobability(πk)wasweightedtowardthisstrata

over weekdays at 12 to eight samples (Supporting InformationAppendix S2) We stratified the sample time into morning (am0600ndash1159)andafternoon(pm1200ndash1800)withequalproba-bilitiesofsamplingTominimizetemporalautocorrelationthestart-ing location and travel direction of the routewere also randomlyselectedwithoutreplacementEqualsamplingweightwasprovidedtoeachrampduetolackofpriorknowledgeofuserates

At each boat ramp surveys were conducted by first countingparkedboattrailersthenbymonitoringvesselentryandexitsduringa1-hrsamplingperiod(KinlochMcGlennonNicollampPike1997Pollocketal1994)The1-hrperiodwaschosendueto travel timeanddis-tancebetweensitesallowingforabus-routesampletobecompletedforallrampswithinthetemporalstrata(AMorPM)ofthedailysam-plingframeInterviewswereconductedwithallpartieslaunchingandretrievingvesselsduringtheclerkrsquoswaittimedeterminingpartysizetargetspeciesgroupandactivity(iefishingornot)forestimatesoffishing effort This well-established procedure produced a monthlyextrapolatedestimationofboateffort instandardizedunitsofefforthoursplusmnSEwhichcouldthenbeadjustedtofishingeffortviathepro-portionofintervieweesactivities(Pollocketal1994RobsonampJones1989)(SupportingInformationAppendixS3)PartysizebetweenrampswastestedforanydifferenceusingaonefactorANOVA

Thebus-route surveyhenceprovidesanestimationofall fish-ing effort from trailer boats around this isolated peninsular TheCRAGSdataarethusahigh-frequencyspatialsubsetoftheentiretrailer boat fishing effort for the period thatwas extrapolated bytheregionalsurveyThisallowedustobothtesttherelativeuseoftheHRMCAcomparedtowiderpeninsularuseandtheabilityofourremoteobservationswithourCRAGsunitofanactualsubsetofthefisherytotracklargertemporalpatternsofpeninsularwideeffort

TovalidatetherepresentativenessoftheCRAGShigh-frequencytrackingofeffortmetricsover timeweundertookaSpearmanrankcorrelationbetweentime-matchedeffortestimationsbyCRAGSandthe regional bus-route survey For graphing bus-routeextrapolationestimates were overlaid onto extrapolations from daily averages ofboatfishingeffortfromCRAGSWealsocarefullyloggedalltimespentontheprojecttocomparetotallaborcostsbetweenCRAGSandthebusrouteforbothdatacollectionandprocessingtodetermineeffortacrossthecomparativemonthaswellasforcostpersample

3emsp |emspRESULTS

The CRAGS operated successfully and continuously throughoutthestudycapturing545plusmn198SE(May)and483plusmn088SE(June)panoramasperdaymakinga totalof314 samplesWemade895observationsofboatsbut51samplesacross21daysreturnedzerocountsofboats(17oftotalsamples)Thesewereconcentratedonweekdaysmdashwith 44 zero-count samples across 18daysmdashwhile onweekendsandpublicholidaytherewereonlysevensamplesacross3dayswithnoactivityInclementweatherrendered27samplesun-usableTheimagesandbatch-processedTimeMachinemoviescon-sumed4274GBofmemory

(1)ei=IitimesT

(2)Ejk=

nsum

i=1

(

ei

120587k

)

(3)SEjk=

1

njminus1

sumn

i=1(eiminus ei)

2

njtimes (Njk)

2

(4)SEk=

radic

SE21k+SE2

2khellip+SE2

jk

(5)smt=etminus1+ et+ et+1

3

6emsp |emsp emspensp FLYNN et aL

Based on boat ramp interviews the average trailer boat partysize (P) was 287 people per vessel throughout May (plusmn0133 SEn=84)andpartysizebetweensitesshowednosignificantdiffer-ence (df=2F = 302p = 068) The averagemonthly boat fishingeffort(hours)extrapolatedforourCRAGSobservationsofHRMCAwas5629hr(plusmn711SE)with4096hr(plusmn607SE)inMayand1534hr(plusmn370 SE) in June (Figure4)mdashthis could be extrapolated to fisherhoursbymultiplyingbytheaveragepartysizebutasnointerviewswereconductedinJuneweleftboththeCRAGSandbusrouteex-trapolationsasboatfishingeffortonly

The total subset area observed by CRAGSwas 96km2 which comparedto318km2thatwaseasilyavailabletotrailerboatsaroundtheTasmanPeninsularwhichcorrespondstoaround3ofthetotalareaComparedtothetotalfishingeffortforMayextrapolatedfromtheregionalbus-routesurvey therelativelysmallareaaroundtheHRM(~3)receivedalargeamountofthefishingeffort(~23)

ThehighfrequencyofCRAGSsamplingpermittedtheexamina-tionoffine-scaleinter-dailytemporalvariationwhichdisplayedarag-gedsinusoidoscillation(Figure5)Effortwasgenerallylowwithinthe

weekdaystrataandthen increasedrapidly to largepeaksonweek-ends (Figure5) One exception to this was between the 18th and22ndMaywhichhadmoreeffortonweekdays thanotherperiods(Figure5)Bracketingthisperiodweretwoweekendsoffishingcom-petitions theldquoTunaClubofTasmaniarsquosrsquoFarSouthClassicrdquo (16ndash17thMay)andRally4Northerninvitationalweekendrally(23rdMay)

A total of 16 bus-route samples were conducted logistic fail-uresledtotheabandonmentoffour(20)oftheplannedsamplesRegionalTrailerboateffortfromaroundtheTasmanPeninsularwasestimatedtobe19650hr (plusmn2858SE) inMay (Figure4)Notrailersat individual rampswere recorded four timeswith all of theseoc-curringonweekdaysThetotalnumberoftrailerssightedperramprangedfrom0to16duringweekdaysand1ndash54forweekendswiththehighestnumberrecordedon16May2015atPiratesBayDuringhigh-intensitydays(particularlyweekendsholidaysmornings)therewereoftenlargeoverflowsfromrampcarparkswithparkedcarsandboattrailersextendingasmuchasonekilometerfromtheboatramp

During thebus-route survey 84 interviewswere conductedThemajorityofmariners(905n=96)indicatedfishingwastheir

F IGURE 4emspExtrapolatedmonthlytrailerboateffort(hoursplusmnSE)fortheentireTasmanPeninsulaassessedbyabus-routesurvey(May)andforthesubsetoftheareaaroundtheHRMCAsampledwithCRAGS(MayandJune)

F IGURE 5emspHigh-frequencyinterdailyextrapolatedCRAGSestimationsofboatefforthoursaroundthe(HRMCA)

0

100

200

300

400

500

600

Rolli

ng m

ean

daily

effo

rt in

hou

rs

DayWeekdays Weekends + Public Holidays

lyn088
Inserted Text
CA
lyn088
Cross-Out
lyn088
Cross-Out
lyn088
Inserted Text
t

emspensp emsp | emsp7FLYNN et aL

primaryactivity (Table1)Overall643 (n=54) indicated theirprimary target were tuna with 179 (n=15) specifically tar-geting southern bluefin tuna (Thunnus maccoyii)Other commonrecreationaltargetsincludedflathead(Platycephalusspp595)Southernrocklobster(Jasus edwardsii107)andsquidcalamari(Sepioteuthis australisandNototodarus gouldi476)(Table1)

Basedon the responsesofactivity type from interviews totalboatfishingeffortinMaywasestimatedtobe17783hr(plusmn2586SE)Asmostnonfishingactivitieswerenear shoreweassumed trailerboatingactivityatHMRCAobservedbyCRAGSwasallfishing

Therewere16directcomparisonsamplesbetweenthebusrouteandCRAGSmethodsbetweenthe2ndandthe29thofMaySamplesconsisted of 7 weekdays and 9 weekendspublic holiday samples

(Figure6)Infourcaseswheretwobus-routesamplesweretakenonthe same day (Supporting Information Appendix S2) they are com-bined toadailyestimation for tabulation (Table2)Theproportionaldifferencebetweenestimatedeffortsusingthetwomethodsrangedfrom078-to418-folddifferencehoweverranksestimatesbetweenCRAGS and bus-route samples were strongly associated (n=16ρ(12)=087p=lt001)Interdailysamplingstrata(ampm)werealsoexaminedseparatelyforcorrelationwhichremainedstrongforbotham(n=9ρ(7)=080p=lt001)andpm(n=7ρ(5)=089p=lt001)

ForthecomparativemonthCRAGStook169samplescomparedtothebusroutesrsquo16(Table3)WehadtwofieldtripdaysforCRAGSwith four people on the setup and three on the breakdown daywhichresultedinsevenfull-timeequivalent(FTE)daysofworkThiscomparedto16fielddaysforthebus-routemethodwhichincludedtraveltothefieldsiteforbetweentwoandthreestaffwhichaccu-mulatedto40FTEdaysofworkSettingupourCRAGSautomatedstitching took one FTE day (though this is faster for experiencedofficers)whiledataextractionperimagewasestimatedtotakeon

TABLE 1emsp Intervieweesreportedprimarytargettaxaoractivity

Primary target taxaspeciesactivity Interview responses

Tuna(unspecified) 37 440

SouthernBluefinTuna 15 179

AlbacoreTuna 2 238

AllTuna 54 643

Flathead 5 595

Southernrocklobster 9 107

Squidcalamari 4 476

Abalone 1 119

SmallPelagics 1 119

SnottyTrevalla 2 238

Allnearshorespecies 22 262

Surfing 1 119

SCUBA 3 357

DemonCavevisitors 3 357

PleasureCruise 1 119

Allnonextractive 8 952

Note ldquoTunardquo responsesdenote anglers targeting all specieswithin thefamilyThunnini

F IGURE 6emspDailyboatingeffort(hours)extrapolatedusingbus-routeaccesspointsurvey(bar)andCRAGS(line)Effortestimatefrombus-routesurveywascalculatedfrommorning(amdarkgray)andafternoon(pmlightgray)surveyCRAGSsurveywasconductedbetweenMayandJunewhilethebus-routesurveywasonlyconductedinMay

TABLE 2emspRankingofeffortforpairedsamplesbetweenCRAGSandthebusroute

DateCRAGS hours (rank)

Bus- route hours (rank) Δ Rank

2052015 330(9) 1208(10) minus1

3052015 303(8) 1471(11) minus3

12052015 0(1) 48(2) minus1

14052015 0(1) 24(1) 0

16052015 387(11) 1794(12) minus1

19052015 93(4) 89(4) 0

20052015 57(3) 117(5) minus2

22052015 228(7) 283(6) 1

23052015 513(12) 414(8) 4

24052015 345(10) 1155(9) 1

25052015 122(6) 323(7) minus1

29052015 102(5) 79(3) 1

8emsp |emsp emspensp FLYNN et aL

average75mincomparedto1minforenteringthebus-routedatasheetWhiledataextraction forCRAGS took longer compared tothe bus routemdashat 381 FTE daysmdashtotal monthly labor for CRAGStookonlyaquarterofthetimeandonaper-samplecomparisonwasonly25ofthebus-routelaborcost

4emsp |emspDISCUSSION

Over our autumn samplingmost trailer boat owners accessing thewaters around theTasmanPeninsula via the regional boating infra-structurewererecreationallyfishingwithfisherspredominatelytar-getingtunaTrailerboatrecreationalfishersconstitutealargetrancheof coastal users in Australia (McPhee etal 2002) and our resultssuggestedthatthisregionremainsonewhererecreationalfishing isconcentrated(MortonampLyle2004)EffortobservedbyCRAGSwasa spatial subsetof thegeneral areaeasily accessedby trailer boatsaroundtheTasmanPeninsularwhichweregionallyassessedwithourbus-routerovingsurveyAsCRAGSonlyobserved~3of thetotalarea but corresponded to ~23 of themonthly bus-route boat ef-fort theHippolyteRocksMarineConservationArea (HRMCA) is afurtherconcentrationpointwithintheregionforrecreationalfishingAsthisoffshorefishingeffortaroundtheHRMCAwasmostprobablyfocusedontotunamdashwhichaccountedfor634ofallinterviewedfish-ers activitiesmdashthe actual proportionofpeninsularwide tuna fishingthatwasobservedbyCRAGSwaspotentiallymuchhigherthan23

For developing effort metrics finding a representative site toobservemaybe important forprecise trackingof trendsThetightrankcorrelationbetweenpairedregionalbus-routeandCRAGScam-era samples suggest that effort atHRMCAwas representative ofregionaltemporaltrendsOurresultsarealsosimilartootherstud-ieswheresamplingmethodsforrecreationalfisheriesoverbroaderscaleswhencomparedtoindextrendsofeffortcollectedfrompointsourcesshowstrongcorrelations(Hartilletal2016)

CRAGSwithitsseascapescaleofdatacaptureobservesactualeffortonfishinggroundsratherthan indirectmetricsfrommove-ment past access points (Hartill etal 2016 Smallwood PollockWise Hall amp Gaughan 2012 van Poorten amp Brydle 2018) This

makesthechoiceofaccesspointstomonitorimmaterialfortrendcol-lectionandmaybeofparticularinterestforpelagicfisheriesWhilenonintuitiverecreationalfishingforwide-rangingmigratorypelagicspeciesisoftenspatiallyconcentratedintosmallareas(Lynch2006)duetotightoverlapsbetweeneaseofaccessbyfishersandfishbe-haviorrelatedtobathymetrymigrationhabitatandpreyavailability(Lynchetal2004PattersonEvansCarterampGunn2008)

Unlikespatialpatternstheintensityofcoastalrecreationaltemporaleffortcanbehighlyvariableovertime(Lynch20082014WiseTelferLaiHallampJackson2012)thoughitisgenerallythoughttofollowpre-dictablepatternsrelativetoholidayandnonholidayperiods(JonesampPollock2012)Ourresultsdemonstratedtheselargefluctuationsrang-ingfromzerouptogt400hroftrailerboateffortforadjacentdaysIncombinationwithholidayperiodsothersourcesofvariabilitymayalsooccurduetoindividualorcombinationsoffactorssuchasfishingcluborcompetitionactivitiesandweatherconditionsHigh-frequencysam-plingcanovercomethesecommontemporaldisjunctionsinobservedfishingeffortwherebychanceduetolownumbersofsampleslogis-ticallypermissiblewithtraditionalrovingoraccesspointsurveyslargeerrorsintheestimationofeffortcanbeintroduced(Hartilletal2016vanPoortenampBrydle2018vanPoortenetal2015)

With multiple daily observations abnormal episodes are morelikelytobeobservedForinstanceaweek-longperiodofintenseef-fortwasrecordedbetween15May2015and27May2015whichwasbracketedbytwoconsecutiveweekendswherefishingcompetitionswereheldontheTasmanPeninsularThenormalsinusoidpatternoflowweekdayandhighweekendeffortwasmuchlesspronouncedastheweekdaylullsineffortwerepartiallybridgedbyanglerscontinuingtofishAsbus-routedesignsusedaytypeasstratawithlowerprob-abilitiesforsamplingweekdaysandmuchfewersamplesoverallthistypeofeffectcouldeasilybemissedWhileourbus-routesamplesdidoverlapwiththetunafishingcompetitionsthiswaspurelybychanceSucheventswhichresultinlargespikesofeffortcanhavemarkedim-pactsonlocalecosystemsandharvestestimates(McPheeetal2002MonzPickeringampHadwen2013)andwithoutpriorknowledgearemorelikelytobecapturedwithmonitoringathigherfrequencies

Methodsforgatheringdataoncoastalusesuchasrecreationalfishing moved away from direct approaches and toward off-sitemethodssuchastelephoneinterviews(McCluskeyampLewison2008Pollock etal 1994) due to unsustainable costs from staffing andoperationsparticularlyduringperiodsof intenseusesuchasearlymornings weekends and public holidays which require overtimepaymentsHoweversamplingissueswithoff-sitemethodsareofpar-ticularconcernforopen-accessactivitiessuchasunlicensedfisheriesasthereisnolicensedatabaseofcontactphonenumbersonwhichtobaseasampleframeThisisparticularlysofornicherecreationalfisheriessuchaspelagicgamefishingasthefishersbecomeincreas-inglydilutewithinthegeneralpopulationandhencearehardtoac-cessparticularlyviaoff-sitemethodssuchastelephone interviews(Griffithsetal2010)

Remoteandautonomousdeploymentofsensorsmaythereforeprovideanalternativeandcost-effectivemethodforlong-termmon-itoring particularly for effort across a range of applicationsOur

TABLE 3emspFulltimeequivalent(FTE)labordaysforCRAGSandbusroutesurveystoestimatepatternsofeffort

CRAGS Bus route

Numberofsamples 169 16

Fielddays 2 16

Averagepeopleperfieldday 35 25

Field FTE days subtotals 7 40

StitchingsetupFTE 1 0

DataextractionFTE(persample) 0017 0002

Subtotal data extraction FTE 381 004

FTEdayspersample 006 250

Total FTE days 108 400

emspensp emsp | emsp9FLYNN et aL

comparisonsofmatchedpairsofdatafromtwomethodssuggestedthat theCRAGSoutputswerewell correlatedwith regionalefforttrendsOuranalysisoflabordemonstratesthattryingtounderstandthistemporalvariabilitywithsamplingviaboatrampsurveyswouldbeprohibitivelyexpensiveOurCRAGSallowsforatemporaleffortmetric derived fromactual observationsof fishing to be cheaplydeveloped with sampling frequency at subdaily scales This high-frequencysamplingapproachwasvalidatedagainstour largerbutmuchmorecostlybus-routesurveyOuranalysisoflaborrequiredtobothcollectandprocessdatashowedthatCRAGSrequiredconsid-erablylessresourcestocollectapproximatelyninetimesmorefre-quenttemporaldatathanthebus-routemethodMajorlaborsavingswereduetotheruggedizedandautonomouscapabilityofCRAGSasonlytwofieldtripswererequired(deploymentandretrievaltook7FTEdays)comparedtothebus-routersquos16fieldtrips inamonthwhichconsumed40FTEdaysandincludedanadditional780kmofinter-ramproadtravel(SupportingInformationAppendixS3)AsitisweatherproofautonomousandabletobeprogrammedwithatimedelaysetupCRAGSldquofireandforgetrdquonaturebothremovestheneedforobservertobephysicallylocatedattheregiontocountactivity(EdwardsampSchindler2017)andworkduringovertimeperiodsTheldquoblankingperiodrdquowhichstopsandstartssamplingmdashinthiscaseover-nightmdashalso reduces battery and memory use In combined thesetechnological advances have allowed long-term CRAGS deploy-mentsacross4yearsof4ndash6monthrotationsontooffshoreislandsand for several seasons in Antarctica to monitor seabirds (Lynchetal2017)Thelaborestimateforthebusroutewasconservativeasitdidnotincludeovertimeloadingforthe6weekenddaysofsam-plingorearlymorningcommencementofamsamples

Theadditional381FTEprocessingtimerequiredbythecameraapproach did not impact greatly on the overall labor budgetwhenconsideredagainstfieldworkTheuseoftheTimeMachinebatchpro-cessingsoftwaretoautomaticallyproducetheldquobigdatardquointeractivemoviewasalsoatechnologicalinnovationthatsignificantlyreducedthelaborofdatahandlingcomparedtothemanualopeningofindivid-ualGigaPansusedinpreviousstudies(Lynchetal2015Woodetal2016)IncomparativetermsattheindividualsampleleveltotracktheeffortmetriceachCRAGSsamplerequired24ofthelaborofthebus-routesurvey

Thedevelopmentandapplicationofimagerecognitionsoftware(machine learning) to detect classify and count boats automati-cally (Bousetouane amp Morris 2015) would further reduce post-processingcostsTherequiredfrequencyofsamplingshouldalsobeconsideredonacase-by-casebasisdependentontheresearchquestion as image collection can be somewhat open-ended andresult inunwieldydata sets that require subsampling tobe cost-effective(Parnelletal2010)Inourcasethesettingof~5imagesperdayappearedtoadequatelydescribetheinterdailyvariation

Whileouroneautonomouscameraprovidedastrongcorrelationwithtemporaltrendsineffortcomparedtotheregionalsurveyitdidnotprovide relativespatialdistributionsofeffortacross the fisheryThesedistributionshowevercanbepredictableandhighlyconcen-trated(Lynch20062014Parnelletal2010)anobservationthatwas

alsoshownbyourresultsandmuchsmallerlevelsofsamplingeffortwouldprobablyberequiredtoresolvevariationinspatialdistributionscomparedtowhatisrequiredtoresolvethehightemporalvariationsineffortUnderstandingfine-scalespatialdistributionsofrecreationalfishingeffortiscommonlyachievedusingaerialcounton-watersur-veysorvantagepointsurveys(Lynch2006Smallwoodetal2012)thoughotherremotemethodssuchashigh-resolutionsatelliteimag-ery(FretwellScofieldampPhillips2017)mayprovideacheapersolu-tionThesetypesofwidersurveyshavealsobeensuggestedashighlycompatiblewithpointsourcemetricssuchasremotecamerasystemstoallowforcalibrationorldquoup-ratingrdquoforeffortexpansionstoestimateoveralleffortwithinthefishery(Hartilletal2016)

Thelongdistance(~6km)betweentheHMRCAandthedeploy-ment location stretched the resolution limit of the current CRAGSsetup aswewere unable to distinguish party size or activity typepurelyusing thecamerasystemThis isunlikepreviousCRAGSde-ploymentwhichhadahighsuccessrateofidentifyingpartysizeac-tivity typeata shorter rangeofapproximately19km (Woodetal2016) Thus ground-truthing information about the fisherywas re-quiredforthesystemtoworkeffectivelyAutonomouscamerasystemalsodonotcollectcreel(catch)databutbyallowingforcost-effectivecollectionofthehighlyvariabletemporaleffortdatatheymayallowforhumanresourcestobespentelsewhereduringsurveyperiods(egtoundertakeboatrampinterviews)(vanPoortenampBrydle2018)

Our studywas a small-scale trial of new technology andwhilepreliminary results are encouraging validated seasonal effort esti-mationsobservedviaCRAGSwereoutsidethescopeofthisprojectOur comparisons would hence have been improved by extendingCRAGSdeploymentsandwithmorecomparisonstoregionalsurveysSimilarly usingmultiple CRAGS simultaneously would have bettertested for differences in subregional patterns of temporal trendsorconcentrationof infishingeffortfor instanceboatsaroundthenearbyTasmanIsland

Insummarytheprogrammablesamplinganddataextractionof-feredbyhigh-resolutionautonomouscamerasystemscanallowforfine-scalelong-termmonitoringoffishingefforttrendsorpotentiallymanyothertypesofactivitiesatseascapescalesMorewidespreadadoptionofhigh-frequencysamplingapproachesforeffortestima-tionwouldalsohelpavoidcollectionofmisrepresentativedatabyidentifyingeffortanomalies(McCluskeyampLewison2008)suchastheweekdaycontinuationeffectofraisedeffortbetweenadjacentweekendfishingcompetitionByimprovingtheinformationdensityovertraditionalsurveymethodsautonomouscamerasystemsandotherremotesystemsmayprovideanewwaveofoptimizationfordirectmeasurementofcoastalhumanusesuchasrecreationalfish-eriesalthoughcomplimentarymethodsarestillrequiredtodeter-minespatialdistributionstargetspeciesandharvest

ACKNOWLEDG MENTS

We thankMsGeorginaWoodMs SaraAdkinsMsBelleMonkMrRSherringtonandMsPaigeKellyforconstructiveinsighten-thusiasmandassistanceduringfieldworkWealsothankMrPhillip

10emsp |emsp emspensp FLYNN et aL

CulversonMrStuartDudgeonandMrMattStapenellKarenEvensprovidedhelpfulcommentsonaninternaldraftGreatestthankstothemany recreational anglerswhowillingly shared information tosurveyclerkswithouttheirinvolvementthisprojectwouldnothavecometofruitionThreeanonymousreviewersprovidedcommentsthatimprovedthemanuscript

CONFLIC T OF INTERE S TS

Theauthorsdeclarenoconflictofinterests

AUTHOR CONTRIBUTIONS

MrFlynnhelpeddesignthestudyledthefieldworkundertookmostoftheanalysisandhelpwritethemanuscriptDrLynchhelpdesignthestudyassisted inthefieldworkprovidedadviceontheanaly-sisanddidsomeanalysisconstructedfiguresandafterMrFlynncontributedthemosttothedraftingofthemanuscriptDrBarretthelpeddesign the study assisted in some fieldwork provided ad-viceonanalysisandeditedthemanuscriptMrWonghelpedexten-sivelywithfieldworkassistedwithanalysisconstructedfiguresandhelpedwritethemanuscriptMsDevineassistedwithfieldworkandconfigurationoftheequipmentincludingtheapplicationofthebigdatamoviesoftwareandhandlingofphotographsandalsohelpedwritethemanuscriptMrHughesdevelopedtheelectronicsfortheequipmentoversaw thebuildofgearwrote the software topro-gramtherobotandreviewedandeditedthemanuscript

DATA ACCE SSIBILIT Y

Survey sampling schedules data collection sheets extrapolationproceduresandtechnicalCRAGSspecificationsarealluploadedasonlinesupportinginformationappendixRawdatafromsurveysandCRAGSDOIhttpdoiorg105061dryadv10d218

RE SE ARCH E THIC S COMPLIANCE AND FUNDERS

ResearchabidesbytheAustraliancodeofhumanresearchandex-perimentation(NationalStatementonEthicalConductinHumanResearch of 2007) Permission to conduct research on humanparticipants was granted by the University of Tasmania SocialScience under reference H0014772mdashldquoRemote investigation ofcoastal area user group quantification and variationrdquo aswell asfromtheCSIROSocialScienceHumanResearchEthicsCommitteeunder reference 08115mdashldquoHawkesbury Region Project HumanUsePatternsrdquoFundingwasbyUTASasPartoftheIMASHonoursProject funding and through internal support from the CoastsProgram

ORCID

Tim P Lynch httporcidorg0000-0002-5346-8454

R E FE R E N C E S

Alessa L Kliskey A amp Brown G (2008) Socialndashecological hotspotsmapping A spatial approach for identifying coupled socialndasheco-logicalspaceLandscape and Urban Planning8527ndash39httpsdoiorg101016jlandurbplan200709007

ArlinghausR(2006)Overcominghumanobstaclestoconservationofrecreational fishery resources with emphasis on central EuropeEnvironmental Conservation 33 46ndash59 httpsdoiorg101017S0376892906002700

Arlinghaus R Tillner R amp Bork M (2015) Explaining participationratesinrecreationalfishingacrossindustrialisedcountriesFisheries Management and Ecology 22 45ndash55 httpsdoiorg101111fme12075

BadcockPBPatrickKSmithAMASimpsonJMPennayDRisselCEhellipRichtersJ(2016)Differencesbetweenlandlineandmobilephoneusers in sexualbehavior researchArchives of Sexual Behavior46(6)1711ndash1721

Blumberg S J amp Luke J V (2009) Reevaluating the need for con-cern regarding noncoverage bias in landline surveys American Journal of Public Health 99 1806ndash1810 httpsdoiorg102105AJPH2008152835

Bousetouane F ampMorris B (2015) Off-the-shelf CNN features forfine-grainedclassificationofvesselsinamaritimeenvironmentInGBebisRBoyleBParvinDKoracinIPavlidisRFerisTMcGrawMElendtRKopperERaganZYeampGWeber(Eds)Advances in visual computing 11th International symposium ISVC 2015 Las Vegas NV USA December 14ndash16 2015 proceedings Part II (pp379ndash388)Cham Switzerland Springer International Publishing httpsdoiorg101007978-3-319-27863-6

Cooke S J amp Cowx I G (2004) The role of recreational fish-ing in global fish crises BioScience 54 857ndash859 httpsdoiorg1016410006-3568(2004)054[0857TRORFI]20CO2

DiggleP J (1995)Time-series a biostatistical introductionOxfordUKClarendonPress

EdwardsJampSchindlerE (2017)Avideomonitoringsystemtoeval-uate ocean recreational fishing effort in Astoria OregonMarine recreational information program (pp 31) Newport OR OregonDepartmentofFishandWildlife

FarrMStoecklNampSuttonS(2014)RecreationalfishingandboatingArethedeterminantsthesameMarine Policy47126ndash137httpsdoiorg101016jmarpol201402014

ForbesETraceySampLyle JM (2009)Assessment of the 2008 rec-reational gamefish fishery of Southeast Tasmania with particular refer-ence to Southern Bluefin TunaHobartTASTasmanianAquacultureandFisheriesInstituteUniversityofTasmania

FretwellPTScofieldPampPhillipsRA(2017)Usingsuper-highres-olutionsatelliteimagerytocensusthreatenedalbatrossesIbis159481ndash490httpsdoiorg101111ibi12482

GiriKampHallK (2015)SouthAustralian recreational fishingsurveyFisheries Victoria internal report series (pp 63) Queenscliff VicFisheriesVictoria

Griffiths S P Pollock K H Lyle J M Pepperell J G TonksM L amp Sawynok W (2010) Following the chain to elu-sive anglers Fish and Fisheries 11 220ndash228 httpsdoiorg101111j1467-2979201000354x

Halpern B SWalbridge S Selkoe K A Kappel C V Micheli FDrsquoAgrosaChellipWatsonR(2008)AglobalmapofhumanimpactonmarineecosystemsScience319948ndash952httpsdoiorg101126science1149345

Hartill BW Payne GW Rush N amp Bian R (2016) Bridging thetemporal gap Continuous and cost-effective monitoring of dy-namic recreational fisheries by web cameras and creel surveysFisheries Research 183 488ndash497 httpsdoiorg101016jfishres201606002

emspensp emsp | emsp11FLYNN et aL

HenryGWampLyleJM(2003)NationalrecreationalandindigenousfishingsurveyFinal report to the fisheries research amp development cor-poration and the fisheries action program (pp 188) Canberra ACTFRDC

JonesCMampPollockKH(2012)Recreationalanglersurveymeth-odsEstimationofeffortharvestandreleasedcatchInAZaleDLParrishampTMSutton(Eds)Fisheries techniquesBethesdaMDAmericanFisheriesSociety

Kearney R E (2002) Co-management The resolution of conflict be-tween commercial and recreational fishers in Victoria AustraliaOcean amp Coastal Management45201ndash214httpsdoiorg101016S0964-5691(02)00055-8

KellerKSteffeASLowryMMurphyJJampSuthersIM(2016)Monitoringboat-basedrecreationalfishingeffortatanearshorearti-ficialreefwithashore-basedcameraFisheries Research18184ndash92httpsdoiorg101016jfishres201603025

KinlochMAMcGlennonDNicollGampPikePG(1997)Evaluationof the bus-route creel surveymethod in a largeAustralianmarinerecreationalfisheryISurveydesignFisheries Research33101ndash121httpsdoiorg101016S0165-7836(97)00068-4

LewinWCArlinghausRampMehnerT(2006)Documentedandpo-tential biological impacts of recreational fishing Insights forman-agementandconservationReviews in Fisheries Science14305ndash367httpsdoiorg10108010641260600886455

LowryMampMurphy J (2003)Monitoring the recreational gamefishfisheryoffsouth-easternAustraliaMarine and Freshwater Research54425ndash434httpsdoiorg101071MF01269

LyleJMStarkKEampTraceySR(2014)2012ndash2013 Survey of rec-reational fishing in TasmaniaHobart TAS Institute forMarine andAntarcticStudies

Lynch T P (2006) Incorporation of recreational fishing ef-fort into design of marine protected areas Incorporacioacuten delEsfuerzo de Pesca Recreativa en el Disentildeo de Aacutereas MarinasProtegidas Conservation Biology 20 1466ndash1476 httpsdoiorg101111j1523-1739200600509x

LynchTP(2008)Thedifferencebetweenspatialandtemporalvaria-tioninrecreationalfisheriesforplanningofmarineprotectedareasResponse toSteffeConservation Biology22486ndash491httpsdoiorg101111j1523-1739200800888x

LynchTP (2014)Adecadal time-seriesof recreational fishingeffortcollectedduringandafterimplementationofamultipleusemarinepark shows high inter-annual but low spatial variability Fisheries Research15185ndash90httpsdoiorg101016jfishres201309014

LynchTPAldermanRampHobdayAJ(2015)Ahigh-resolutionpan-orama camera system formonitoring colony-wide seabird nestingbehaviourMethods in Ecology and Evolution6491ndash499httpsdoiorg1011112041-210X12339

LynchTFlynnDWoodGKimJ-HParkJampHobdayAJ(2017)Metrics parameters and results from fisheries and wildlife studies with the CSIRO Ruggedized Autonomous Gigapixel Systems (CRAGS)DarwinNTAustralianMarineScienceAssociation

LynchTPWilkinsonEMellingLHamiltonRMacreadyAampFearyS(2004)ConflictandimpactsofdiversandanglersinamarineparkEnvironmental Management 33 196ndash211 httpsdoiorg101007s00267-003-3014-6

MaHOgawaTKSminkeyTRBreidtFJLesserVMOpsomerJDhellipVanVoorheesDA(2018)Pilotsurveystoimprovemonitoringofmarine recreational fisheries inHawaiʻiFisheries Research204197ndash208httpsdoiorg101016jfishres201802010

McCluskey S M amp Lewison R L (2008) Quantifying fish-ing effort A synthesis of current methods and their ap-plications Fish and Fisheries 9 188ndash200 httpsdoiorg101111j1467-2979200800283x

McGlennonDampKinlochMA(1997)Evaluationofthebus-routecreelsurveymethod in a largeAustralianmarine recreational fishery II

PilotsurveysandoptimalsamplingallocationFisheries Research3389ndash99httpsdoiorg101016S0165-7836(97)00067-2

McPheeDLeadbitterDampSkilleterGA(2002)SwallowingthebaitIs recreational fishing in Australia ecologically sustainable Pacific Conservation Biology840httpsdoiorg101071PC020040

Monz C A Pickering C M amp HadwenW L (2013) Recent ad-vances in recreation ecology and the implications of differentrelationships between recreation use and ecological impactsFrontiers in Ecology and the Environment11441ndash446httpsdoiorg101890120358

MooreAHallKGiriKTraceySPenroseLHansenShellipBrownP(2015)Developingrobustandcost-effectivemethodsforestimatingthenationalrecreationalcatchofSouthernBluefinTunainAustraliaFRDC Projectpp123

Morton A J amp Lyle J M (2004) Preliminary assessment of the rec-reational gamefish fishery in Tasmania with particular reference to Southern Bluefin Tuna Taroona TAS Tasmanian Aquaculture andFisheriesInstitute

ParnellPEDaytonPKFisherRALoarieCCampDarrowRD(2010)SpatialpatternsoffishingeffortoffSanDiegoImplicationsfor zonal management and ecosystem function Ecological Applications202203ndash2222httpsdoiorg10189009-15431

PattersonTAEvansKCarterTLampGunnJS(2008)Movementand behaviour of large southern bluefin tuna (Thunnus maccoyii)in the Australian region determined using pop-up satellite ar-chival tags Fisheries Oceanography 17 352ndash367 httpsdoiorg101111j1365-2419200800483x

PollockKH JonesCMampBrownTL (1994)Angler survey meth-ods and their applications in fisheries management Bethesda MDAmericanFisheriesSociety

PonsM Branch T AMelnychukM C JensenO P Brodziak JFromentinJMhellipHilbornR(2017)EffectsofbiologicaleconomicandmanagementfactorsontunaandbillfishstockstatusFish and Fisheries181ndash21httpsdoiorg101111faf12163

PowersSPampAnsonK (2016)Estimating recreationaleffort in theGulf of Mexico Red Snapper Fishery using boat ramp camerasReduction in federalseason lengthdoesnotproportionally reducecatchNorth American Journal of Fisheries Management 36 1156ndash1166httpsdoiorg1010800275594720161198284

Robson D amp Jones C M (1989) The theoretical basis of an ac-cess site angler survey design Biometrics 45 83ndash98 httpsdoiorg1023072532036

Rocklin D Levrel H Drogou M Herfaut J amp Veron G (2014)Combining telephone surveys and fishing catches self-reportTheFrenchseabass recreational fisheryassessmentPLoS ONE9e87271httpsdoiorg101371journalpone0087271

Smallwood C B Pollock K HWise B S Hall N G amp GaughanD J (2012) Expanding aerial-roving surveys to include counts ofshore-based recreational fishers from remotely operated camerasBenefitslimitationsandcosteffectivenessNorth American Journal of Fisheries Management321265ndash1276httpsdoiorg101080027559472012728181

Teixeira D ZischkeM T ampWebley J A C (2016) Investigatingbias in recreational fishing surveys Fishers listed in public tele-phone directories fish similarly to their unlisted counterpartsFisheries Research 181 127ndash136 httpsdoiorg101016jfishres201604012

van Poorten B T amp Brydle S (2018) Estimating fishing effortfrom remote traffic counters Opportunities and challengesFisheries Research 204 231ndash238 httpsdoiorg101016jfishres201802024

van Poorten B T Carruthers T RWardHGM ampVarkeyD A(2015)Imputingrecreationalanglingeffortfromtime-lapsecamerasusinganhierarchicalBayesianmodelFisheries Research172265ndash273httpsdoiorg101016jfishres201507032

12emsp |emsp emspensp FLYNN et aL

VenturelliPAHyderKampSkovC(2017)AnglerappsasasourceofrecreationalfisheriesdataOpportunitieschallengesandproposedstandardsFish and Fisheries18578ndash595httpsdoiorg101111faf12189

WiseBSTelferCFLaiEKMHallNGampJacksonG(2012)Long-termmonitoringofboat-basedrecreational fishing inSharkBayWesternAustraliaProvidingscientificadviceforsustainablemanagement in a World Heritage Area Marine and Freshwater Research631129ndash1141httpsdoiorg101071MF12054

WoodGLynchTPDevineCKellerKampFigueiraW(2016)High-resolutionphoto-mosaictime-seriesimageryformonitoringhumanuseofanartificialreefEcology and Evolution66963ndash6968httpsdoiorg101002ece32342

YuFRupertACCDiTizianaMNishikantaKRandySIllahNhellipJeffreyG(2011)TerapixelimagingofcosmologicalsimulationsThe Astrophysical Journal Supplement Series19718

Zischke M T Griffiths S P amp Tibbetts I R (2012) Catch and ef-fort from a specialised recreational pelagic sport fishery off

eastern Australia Fisheries Research 127ndash128 61ndash72 httpsdoiorg101016jfishres201204011

SUPPORTING INFORMATION

Additional supporting information may be found online in theSupportingInformationsectionattheendofthearticle

How to cite this articleFlynnDJHLynchTPBarrettNSWongLSCDevineCHughesDGigapixelbigdatamoviesprovidecost-effectiveseascapescaledirectmeasurementsofopen-accesscoastalhumanusesuchasrecreationalfisheriesEcol Evol 2018001ndash12 httpsdoiorg101002ece34301

Graphical AbstractThecontentsofthispagewillbeusedaspartofthegraphicalabstractofhtmlonly

Itwillnotbepublishedaspartofmainarticle

Wetrailedaremoteautonomousultra-high-resolutionphotosamplingmethodasacosteffectsolutionfordirectseascape-scaledmeasure-mentsOursequentialphotosamplingwasbatchedprocessesusingnovelTimeMachinedatahandlingsoftwaretoproduceldquobigdatardquotimeseriesmoviesfromaspatialsubsetofanoffshorefisherywhichwecomparedtoaregionalbus-routesurveyofboattrailersandinterviewswithparticipantsOurcamerasamplescloselytrackedlargerscaleregionaltrendsbutwithlaborcostsaquarterofthebus-routesurveycostsforanindividualsample25ofthebusroutes

Page 3: Gigapixel big data movies provide cost-effective seascape

emspensp emsp | emsp3FLYNN et aL

asimultaneousregionallyscaledbus-routetrailerboatsurveyatthethreerampsonthepeninsularsuitableforlaunchinglargetrailerboatsWe also undertook on-ground interviewswith trailer boat users togroundtruthactivitytype(fishingornonfishing)andtargetspeciesByundertakingbothcameraandboatrampsurveysourover-overarchingaimwastoinvestigatehowremoteimagerymayreplacecomplimentoroptimizeconventionalon-siteapproachesforrecreationalfishing

2emsp |emspMATERIAL S AND METHODS

WedeployedtheCRAGSbetween1May2015and30June2015toassesstrailerboatsoffshorefromwateradjacenttotheTasman

Peninsula (Figure1) This period was chosen as it correspondswith the main tuna fishing season For large trailer boats (gt5mlength) localaccess to the fishery is limitedto threeboat ramps(Figure1)althoughtheareamayalsobevisitedbylargervesselssteaming frommarinas and anchorages further north (MortonampLyle2004)butforthepurposesofourstudyweexcludedtheserelativelyrare largervesselsWefocusedourCRAGSsurveyataknown game fishing concentration the Hippolyte Rock MarineConservationArea (HRMCA)which are a group of small islandsabout6-kmoffshore (Figures1and2)aroundwhichrecreationalfishingispermittedWhiletheCRAGSdeploymentwasmarginallycloser to the FortescueBayRampwe commonly observe boatsapproachingtheHRMCAfromdirectionsthatcorrespondedtothe

F IGURE 1emspTheTasmanPeninsulainSETasmaniaBus-routesurveysamplingoccurredatramps locatedatPiratesFortescueandStuartsBayTheCRAGSlocationisshownasacameraiconandtheareaobservedbythecamerawhichincludestheHippolyteRocksMarineConservationArea(HRMCA)isdescribedasatriangle

Text

Pirates Bay

Fortesque Bay

StewartsBay

0 3 6 9 1215Kilometers

0 30 60 90 12015Kilometers

Tasman Is

Australia

Tasmania

Hobart

TasmanPeninsular

HippolyteRocksMCA

43deg 01 30

147deg 52 30

TAS

4emsp |emsp emspensp FLYNN et aL

locationof all three rampsWe fixed theCRAGS to a clifftop atWaterfall Baywhich overlooks the site fromvery high sea cliffs(~250m)withthecamerafocusingonthewesternsideofHRMCA(Figure1)ForeachsampleCRAGScapturedaseriesof68pho-tographsthroughatelephotolenswhichwhenstitchedtogetherprovide a high-resolution panoramic image with a wide field ofview (~140deg) (Supporting Information Appendix S1) The size ofareasurveyedbyCRAGScomparedtotheregionalareaaccessibleby trailer boatswas estimated using google earth polygons andtheUniversityofNewHampshireKMLextensiontoolmdashhttpsex-tensionunhedukmlToolsindexcfm

Thetimeof the initialCRAGSpanoramasamplewasrandomlyselected Samples were then collected sequentially through-out daylight hours (0600ndash1800) As less recreational fishing fortuna occurred at night no samples were taken with the CRAGSprogrammed to ldquoblankrdquo these hours from the sampling sched-ule (Supporting InformationAppendix S1) Photographic sessionswereprogrammedtobeseparatedbyarecurring93-mingapthus

producing a forward shift in sample capture time andminimizingpotentialautocorrelationsbetweenboatingactivitiesandsamplingtimesPhotographswerebatchprocessedtobothformpanoramicimagesforeachsampleandforcompilingintoaninteractivetime-lapsemovieusingCREATELabrsquosTimeMachinesoftware (CREATELabregPittsburgPAUSA)

TimeMachinemovies are fully interactive and are able to bezoomedintopausedandplayedatvariousspeedsandarealsotimeanddate-stamped(Figure3)UsingthisinteractivemoviesoftwareeachGigaPanmoviewassystematicallypausedateachsamplefullyzoomedandscannedlefttorighttoptobottomforobjectsintheseascapebytheleadauthor(Figure3)Vesselcountswererecordedforeachsamplewithobjectslargerthan~10minlength(nottrailerboats)or requiring longer than10seconds todistinguish (eg lowresolutionandimageartifact)beingdiscountedSampleswerecat-egorized into two strataweekdays andweekendspublic holidaysusing the time and date stamp provided on each GigaPan scene(Figure3)

F IGURE 2emspAnunmodifiedGigaPanEpicPROunit(left)andaCRAGSdeployedatWaterfallBaySETasmania(right)

F IGURE 3emspScreencaptureoutputsfromtheTimeMachineinteractivedataviewershowinganozoomandpanoramicextentoftheHippolyteRockMarineConservationArea(HRMCA)b=13zoommainHippolyterock~6kmfromcamerac=12zoomofalargetrailerboatd=fullzoom

lyn088
Cross-Out
lyn088
Inserted Text
(a)
lyn088
Cross-Out
lyn088
Inserted Text
(b)
lyn088
Cross-Out
lyn088
Inserted Text
(c)
lyn088
Cross-Out
lyn088
Inserted Text
(d)
lyn088
Inserted Text
s

emspensp emsp | emsp5FLYNN et aL

Thecountofboatsforeachsamplewasthenextrapolatedintoanestimateofdailyaverageboatingeffort(hoursplusmnSE)usingthein-stantaneouscountmethod(Equation1)

where ei is theextrapolatedboatinghours for the ithday Ii is theaveragenumberoftrailerboatsobservedacrosstheinstantaneoussamplesfordayiandTisthedailysamplingperiod(daylighthours)Asbadweathercanbeamajorfactoraffectingthedecisiontoboatorfish(ForbesTraceyampLyle2009)boatingeffortwasassumedtobe0whenadverseweathermadeobservationsofboatsimpossible

Next total effort for each day-type strata (Ejk) within sam-pling months (k) was calculated to estimate the monthly effort(Equation2)AsCRAGSallowedforcontinuoussamplingthroughoutthesurveyperiodthesamplingprobability(πk)wassetto1

Standarderrorandeffortvarianceformonthlyestimateswerecalculated(Equation3)accordingtoPollocketal(1994)forstrata

where nj is thenumberofsampleddaysofstrata j Njk is thetotalnumberofdaysofstratajinmonthk

Asvariationforeachstratawascalculatedseparatelyweusedasumofsquaredstandarderrorsapproach(Equation4)toestimatethetotalmonthlyvariation

Assampleswereobtainedinaserialandcontinuousfashionatimeseriesmodelusingrollingaverageswasalsoappliedtothedata(Diggle1995)(Equation5)

where smt is the smoothed average of extrapolated effort e persampletwhichisweightedacrossthreesampleseachseparatedby93minThisprovidedasmoothedaveragewithindays

Wealsoconductedabus-routestylesurveytoestimatefishingeffortfromtrailerboatsfortheregion(Pollocketal1994Robsonamp Jones1989)WhileCRAGSsampleswerecollected throughoutMayandJunethecomparablebus-routesurveywasonlyconductedinMaydue to logistical constraintsWesurveyed the threemajorboatrampsintheregionatPiratesBayFortescueBayandStewartsBaybetween3May2015and30May2015(Figure1)

Atotalof20bus-routesampleswerescheduledWeundertookrandomstratifiedsamplingwithdaytypedividedintoweekdaysorweekendpublic holidays To reflect expected higher recreationaleffortduringweekendsandpublicholidays(McCluskeyampLewison2008)thesamplingprobability(πk)wasweightedtowardthisstrata

over weekdays at 12 to eight samples (Supporting InformationAppendix S2) We stratified the sample time into morning (am0600ndash1159)andafternoon(pm1200ndash1800)withequalproba-bilitiesofsamplingTominimizetemporalautocorrelationthestart-ing location and travel direction of the routewere also randomlyselectedwithoutreplacementEqualsamplingweightwasprovidedtoeachrampduetolackofpriorknowledgeofuserates

At each boat ramp surveys were conducted by first countingparkedboattrailersthenbymonitoringvesselentryandexitsduringa1-hrsamplingperiod(KinlochMcGlennonNicollampPike1997Pollocketal1994)The1-hrperiodwaschosendueto travel timeanddis-tancebetweensitesallowingforabus-routesampletobecompletedforallrampswithinthetemporalstrata(AMorPM)ofthedailysam-plingframeInterviewswereconductedwithallpartieslaunchingandretrievingvesselsduringtheclerkrsquoswaittimedeterminingpartysizetargetspeciesgroupandactivity(iefishingornot)forestimatesoffishing effort This well-established procedure produced a monthlyextrapolatedestimationofboateffort instandardizedunitsofefforthoursplusmnSEwhichcouldthenbeadjustedtofishingeffortviathepro-portionofintervieweesactivities(Pollocketal1994RobsonampJones1989)(SupportingInformationAppendixS3)PartysizebetweenrampswastestedforanydifferenceusingaonefactorANOVA

Thebus-route surveyhenceprovidesanestimationofall fish-ing effort from trailer boats around this isolated peninsular TheCRAGSdataarethusahigh-frequencyspatialsubsetoftheentiretrailer boat fishing effort for the period thatwas extrapolated bytheregionalsurveyThisallowedustobothtesttherelativeuseoftheHRMCAcomparedtowiderpeninsularuseandtheabilityofourremoteobservationswithourCRAGsunitofanactualsubsetofthefisherytotracklargertemporalpatternsofpeninsularwideeffort

TovalidatetherepresentativenessoftheCRAGShigh-frequencytrackingofeffortmetricsover timeweundertookaSpearmanrankcorrelationbetweentime-matchedeffortestimationsbyCRAGSandthe regional bus-route survey For graphing bus-routeextrapolationestimates were overlaid onto extrapolations from daily averages ofboatfishingeffortfromCRAGSWealsocarefullyloggedalltimespentontheprojecttocomparetotallaborcostsbetweenCRAGSandthebusrouteforbothdatacollectionandprocessingtodetermineeffortacrossthecomparativemonthaswellasforcostpersample

3emsp |emspRESULTS

The CRAGS operated successfully and continuously throughoutthestudycapturing545plusmn198SE(May)and483plusmn088SE(June)panoramasperdaymakinga totalof314 samplesWemade895observationsofboatsbut51samplesacross21daysreturnedzerocountsofboats(17oftotalsamples)Thesewereconcentratedonweekdaysmdashwith 44 zero-count samples across 18daysmdashwhile onweekendsandpublicholidaytherewereonlysevensamplesacross3dayswithnoactivityInclementweatherrendered27samplesun-usableTheimagesandbatch-processedTimeMachinemoviescon-sumed4274GBofmemory

(1)ei=IitimesT

(2)Ejk=

nsum

i=1

(

ei

120587k

)

(3)SEjk=

1

njminus1

sumn

i=1(eiminus ei)

2

njtimes (Njk)

2

(4)SEk=

radic

SE21k+SE2

2khellip+SE2

jk

(5)smt=etminus1+ et+ et+1

3

6emsp |emsp emspensp FLYNN et aL

Based on boat ramp interviews the average trailer boat partysize (P) was 287 people per vessel throughout May (plusmn0133 SEn=84)andpartysizebetweensitesshowednosignificantdiffer-ence (df=2F = 302p = 068) The averagemonthly boat fishingeffort(hours)extrapolatedforourCRAGSobservationsofHRMCAwas5629hr(plusmn711SE)with4096hr(plusmn607SE)inMayand1534hr(plusmn370 SE) in June (Figure4)mdashthis could be extrapolated to fisherhoursbymultiplyingbytheaveragepartysizebutasnointerviewswereconductedinJuneweleftboththeCRAGSandbusrouteex-trapolationsasboatfishingeffortonly

The total subset area observed by CRAGSwas 96km2 which comparedto318km2thatwaseasilyavailabletotrailerboatsaroundtheTasmanPeninsularwhichcorrespondstoaround3ofthetotalareaComparedtothetotalfishingeffortforMayextrapolatedfromtheregionalbus-routesurvey therelativelysmallareaaroundtheHRM(~3)receivedalargeamountofthefishingeffort(~23)

ThehighfrequencyofCRAGSsamplingpermittedtheexamina-tionoffine-scaleinter-dailytemporalvariationwhichdisplayedarag-gedsinusoidoscillation(Figure5)Effortwasgenerallylowwithinthe

weekdaystrataandthen increasedrapidly to largepeaksonweek-ends (Figure5) One exception to this was between the 18th and22ndMaywhichhadmoreeffortonweekdays thanotherperiods(Figure5)Bracketingthisperiodweretwoweekendsoffishingcom-petitions theldquoTunaClubofTasmaniarsquosrsquoFarSouthClassicrdquo (16ndash17thMay)andRally4Northerninvitationalweekendrally(23rdMay)

A total of 16 bus-route samples were conducted logistic fail-uresledtotheabandonmentoffour(20)oftheplannedsamplesRegionalTrailerboateffortfromaroundtheTasmanPeninsularwasestimatedtobe19650hr (plusmn2858SE) inMay (Figure4)Notrailersat individual rampswere recorded four timeswith all of theseoc-curringonweekdaysThetotalnumberoftrailerssightedperramprangedfrom0to16duringweekdaysand1ndash54forweekendswiththehighestnumberrecordedon16May2015atPiratesBayDuringhigh-intensitydays(particularlyweekendsholidaysmornings)therewereoftenlargeoverflowsfromrampcarparkswithparkedcarsandboattrailersextendingasmuchasonekilometerfromtheboatramp

During thebus-route survey 84 interviewswere conductedThemajorityofmariners(905n=96)indicatedfishingwastheir

F IGURE 4emspExtrapolatedmonthlytrailerboateffort(hoursplusmnSE)fortheentireTasmanPeninsulaassessedbyabus-routesurvey(May)andforthesubsetoftheareaaroundtheHRMCAsampledwithCRAGS(MayandJune)

F IGURE 5emspHigh-frequencyinterdailyextrapolatedCRAGSestimationsofboatefforthoursaroundthe(HRMCA)

0

100

200

300

400

500

600

Rolli

ng m

ean

daily

effo

rt in

hou

rs

DayWeekdays Weekends + Public Holidays

lyn088
Inserted Text
CA
lyn088
Cross-Out
lyn088
Cross-Out
lyn088
Inserted Text
t

emspensp emsp | emsp7FLYNN et aL

primaryactivity (Table1)Overall643 (n=54) indicated theirprimary target were tuna with 179 (n=15) specifically tar-geting southern bluefin tuna (Thunnus maccoyii)Other commonrecreationaltargetsincludedflathead(Platycephalusspp595)Southernrocklobster(Jasus edwardsii107)andsquidcalamari(Sepioteuthis australisandNototodarus gouldi476)(Table1)

Basedon the responsesofactivity type from interviews totalboatfishingeffortinMaywasestimatedtobe17783hr(plusmn2586SE)Asmostnonfishingactivitieswerenear shoreweassumed trailerboatingactivityatHMRCAobservedbyCRAGSwasallfishing

Therewere16directcomparisonsamplesbetweenthebusrouteandCRAGSmethodsbetweenthe2ndandthe29thofMaySamplesconsisted of 7 weekdays and 9 weekendspublic holiday samples

(Figure6)Infourcaseswheretwobus-routesamplesweretakenonthe same day (Supporting Information Appendix S2) they are com-bined toadailyestimation for tabulation (Table2)Theproportionaldifferencebetweenestimatedeffortsusingthetwomethodsrangedfrom078-to418-folddifferencehoweverranksestimatesbetweenCRAGS and bus-route samples were strongly associated (n=16ρ(12)=087p=lt001)Interdailysamplingstrata(ampm)werealsoexaminedseparatelyforcorrelationwhichremainedstrongforbotham(n=9ρ(7)=080p=lt001)andpm(n=7ρ(5)=089p=lt001)

ForthecomparativemonthCRAGStook169samplescomparedtothebusroutesrsquo16(Table3)WehadtwofieldtripdaysforCRAGSwith four people on the setup and three on the breakdown daywhichresultedinsevenfull-timeequivalent(FTE)daysofworkThiscomparedto16fielddaysforthebus-routemethodwhichincludedtraveltothefieldsiteforbetweentwoandthreestaffwhichaccu-mulatedto40FTEdaysofworkSettingupourCRAGSautomatedstitching took one FTE day (though this is faster for experiencedofficers)whiledataextractionperimagewasestimatedtotakeon

TABLE 1emsp Intervieweesreportedprimarytargettaxaoractivity

Primary target taxaspeciesactivity Interview responses

Tuna(unspecified) 37 440

SouthernBluefinTuna 15 179

AlbacoreTuna 2 238

AllTuna 54 643

Flathead 5 595

Southernrocklobster 9 107

Squidcalamari 4 476

Abalone 1 119

SmallPelagics 1 119

SnottyTrevalla 2 238

Allnearshorespecies 22 262

Surfing 1 119

SCUBA 3 357

DemonCavevisitors 3 357

PleasureCruise 1 119

Allnonextractive 8 952

Note ldquoTunardquo responsesdenote anglers targeting all specieswithin thefamilyThunnini

F IGURE 6emspDailyboatingeffort(hours)extrapolatedusingbus-routeaccesspointsurvey(bar)andCRAGS(line)Effortestimatefrombus-routesurveywascalculatedfrommorning(amdarkgray)andafternoon(pmlightgray)surveyCRAGSsurveywasconductedbetweenMayandJunewhilethebus-routesurveywasonlyconductedinMay

TABLE 2emspRankingofeffortforpairedsamplesbetweenCRAGSandthebusroute

DateCRAGS hours (rank)

Bus- route hours (rank) Δ Rank

2052015 330(9) 1208(10) minus1

3052015 303(8) 1471(11) minus3

12052015 0(1) 48(2) minus1

14052015 0(1) 24(1) 0

16052015 387(11) 1794(12) minus1

19052015 93(4) 89(4) 0

20052015 57(3) 117(5) minus2

22052015 228(7) 283(6) 1

23052015 513(12) 414(8) 4

24052015 345(10) 1155(9) 1

25052015 122(6) 323(7) minus1

29052015 102(5) 79(3) 1

8emsp |emsp emspensp FLYNN et aL

average75mincomparedto1minforenteringthebus-routedatasheetWhiledataextraction forCRAGS took longer compared tothe bus routemdashat 381 FTE daysmdashtotal monthly labor for CRAGStookonlyaquarterofthetimeandonaper-samplecomparisonwasonly25ofthebus-routelaborcost

4emsp |emspDISCUSSION

Over our autumn samplingmost trailer boat owners accessing thewaters around theTasmanPeninsula via the regional boating infra-structurewererecreationallyfishingwithfisherspredominatelytar-getingtunaTrailerboatrecreationalfishersconstitutealargetrancheof coastal users in Australia (McPhee etal 2002) and our resultssuggestedthatthisregionremainsonewhererecreationalfishing isconcentrated(MortonampLyle2004)EffortobservedbyCRAGSwasa spatial subsetof thegeneral areaeasily accessedby trailer boatsaroundtheTasmanPeninsularwhichweregionallyassessedwithourbus-routerovingsurveyAsCRAGSonlyobserved~3of thetotalarea but corresponded to ~23 of themonthly bus-route boat ef-fort theHippolyteRocksMarineConservationArea (HRMCA) is afurtherconcentrationpointwithintheregionforrecreationalfishingAsthisoffshorefishingeffortaroundtheHRMCAwasmostprobablyfocusedontotunamdashwhichaccountedfor634ofallinterviewedfish-ers activitiesmdashthe actual proportionofpeninsularwide tuna fishingthatwasobservedbyCRAGSwaspotentiallymuchhigherthan23

For developing effort metrics finding a representative site toobservemaybe important forprecise trackingof trendsThetightrankcorrelationbetweenpairedregionalbus-routeandCRAGScam-era samples suggest that effort atHRMCAwas representative ofregionaltemporaltrendsOurresultsarealsosimilartootherstud-ieswheresamplingmethodsforrecreationalfisheriesoverbroaderscaleswhencomparedtoindextrendsofeffortcollectedfrompointsourcesshowstrongcorrelations(Hartilletal2016)

CRAGSwithitsseascapescaleofdatacaptureobservesactualeffortonfishinggroundsratherthan indirectmetricsfrommove-ment past access points (Hartill etal 2016 Smallwood PollockWise Hall amp Gaughan 2012 van Poorten amp Brydle 2018) This

makesthechoiceofaccesspointstomonitorimmaterialfortrendcol-lectionandmaybeofparticularinterestforpelagicfisheriesWhilenonintuitiverecreationalfishingforwide-rangingmigratorypelagicspeciesisoftenspatiallyconcentratedintosmallareas(Lynch2006)duetotightoverlapsbetweeneaseofaccessbyfishersandfishbe-haviorrelatedtobathymetrymigrationhabitatandpreyavailability(Lynchetal2004PattersonEvansCarterampGunn2008)

Unlikespatialpatternstheintensityofcoastalrecreationaltemporaleffortcanbehighlyvariableovertime(Lynch20082014WiseTelferLaiHallampJackson2012)thoughitisgenerallythoughttofollowpre-dictablepatternsrelativetoholidayandnonholidayperiods(JonesampPollock2012)Ourresultsdemonstratedtheselargefluctuationsrang-ingfromzerouptogt400hroftrailerboateffortforadjacentdaysIncombinationwithholidayperiodsothersourcesofvariabilitymayalsooccurduetoindividualorcombinationsoffactorssuchasfishingcluborcompetitionactivitiesandweatherconditionsHigh-frequencysam-plingcanovercomethesecommontemporaldisjunctionsinobservedfishingeffortwherebychanceduetolownumbersofsampleslogis-ticallypermissiblewithtraditionalrovingoraccesspointsurveyslargeerrorsintheestimationofeffortcanbeintroduced(Hartilletal2016vanPoortenampBrydle2018vanPoortenetal2015)

With multiple daily observations abnormal episodes are morelikelytobeobservedForinstanceaweek-longperiodofintenseef-fortwasrecordedbetween15May2015and27May2015whichwasbracketedbytwoconsecutiveweekendswherefishingcompetitionswereheldontheTasmanPeninsularThenormalsinusoidpatternoflowweekdayandhighweekendeffortwasmuchlesspronouncedastheweekdaylullsineffortwerepartiallybridgedbyanglerscontinuingtofishAsbus-routedesignsusedaytypeasstratawithlowerprob-abilitiesforsamplingweekdaysandmuchfewersamplesoverallthistypeofeffectcouldeasilybemissedWhileourbus-routesamplesdidoverlapwiththetunafishingcompetitionsthiswaspurelybychanceSucheventswhichresultinlargespikesofeffortcanhavemarkedim-pactsonlocalecosystemsandharvestestimates(McPheeetal2002MonzPickeringampHadwen2013)andwithoutpriorknowledgearemorelikelytobecapturedwithmonitoringathigherfrequencies

Methodsforgatheringdataoncoastalusesuchasrecreationalfishing moved away from direct approaches and toward off-sitemethodssuchastelephoneinterviews(McCluskeyampLewison2008Pollock etal 1994) due to unsustainable costs from staffing andoperationsparticularlyduringperiodsof intenseusesuchasearlymornings weekends and public holidays which require overtimepaymentsHoweversamplingissueswithoff-sitemethodsareofpar-ticularconcernforopen-accessactivitiessuchasunlicensedfisheriesasthereisnolicensedatabaseofcontactphonenumbersonwhichtobaseasampleframeThisisparticularlysofornicherecreationalfisheriessuchaspelagicgamefishingasthefishersbecomeincreas-inglydilutewithinthegeneralpopulationandhencearehardtoac-cessparticularlyviaoff-sitemethodssuchastelephone interviews(Griffithsetal2010)

Remoteandautonomousdeploymentofsensorsmaythereforeprovideanalternativeandcost-effectivemethodforlong-termmon-itoring particularly for effort across a range of applicationsOur

TABLE 3emspFulltimeequivalent(FTE)labordaysforCRAGSandbusroutesurveystoestimatepatternsofeffort

CRAGS Bus route

Numberofsamples 169 16

Fielddays 2 16

Averagepeopleperfieldday 35 25

Field FTE days subtotals 7 40

StitchingsetupFTE 1 0

DataextractionFTE(persample) 0017 0002

Subtotal data extraction FTE 381 004

FTEdayspersample 006 250

Total FTE days 108 400

emspensp emsp | emsp9FLYNN et aL

comparisonsofmatchedpairsofdatafromtwomethodssuggestedthat theCRAGSoutputswerewell correlatedwith regionalefforttrendsOuranalysisoflabordemonstratesthattryingtounderstandthistemporalvariabilitywithsamplingviaboatrampsurveyswouldbeprohibitivelyexpensiveOurCRAGSallowsforatemporaleffortmetric derived fromactual observationsof fishing to be cheaplydeveloped with sampling frequency at subdaily scales This high-frequencysamplingapproachwasvalidatedagainstour largerbutmuchmorecostlybus-routesurveyOuranalysisoflaborrequiredtobothcollectandprocessdatashowedthatCRAGSrequiredconsid-erablylessresourcestocollectapproximatelyninetimesmorefre-quenttemporaldatathanthebus-routemethodMajorlaborsavingswereduetotheruggedizedandautonomouscapabilityofCRAGSasonlytwofieldtripswererequired(deploymentandretrievaltook7FTEdays)comparedtothebus-routersquos16fieldtrips inamonthwhichconsumed40FTEdaysandincludedanadditional780kmofinter-ramproadtravel(SupportingInformationAppendixS3)AsitisweatherproofautonomousandabletobeprogrammedwithatimedelaysetupCRAGSldquofireandforgetrdquonaturebothremovestheneedforobservertobephysicallylocatedattheregiontocountactivity(EdwardsampSchindler2017)andworkduringovertimeperiodsTheldquoblankingperiodrdquowhichstopsandstartssamplingmdashinthiscaseover-nightmdashalso reduces battery and memory use In combined thesetechnological advances have allowed long-term CRAGS deploy-mentsacross4yearsof4ndash6monthrotationsontooffshoreislandsand for several seasons in Antarctica to monitor seabirds (Lynchetal2017)Thelaborestimateforthebusroutewasconservativeasitdidnotincludeovertimeloadingforthe6weekenddaysofsam-plingorearlymorningcommencementofamsamples

Theadditional381FTEprocessingtimerequiredbythecameraapproach did not impact greatly on the overall labor budgetwhenconsideredagainstfieldworkTheuseoftheTimeMachinebatchpro-cessingsoftwaretoautomaticallyproducetheldquobigdatardquointeractivemoviewasalsoatechnologicalinnovationthatsignificantlyreducedthelaborofdatahandlingcomparedtothemanualopeningofindivid-ualGigaPansusedinpreviousstudies(Lynchetal2015Woodetal2016)IncomparativetermsattheindividualsampleleveltotracktheeffortmetriceachCRAGSsamplerequired24ofthelaborofthebus-routesurvey

Thedevelopmentandapplicationofimagerecognitionsoftware(machine learning) to detect classify and count boats automati-cally (Bousetouane amp Morris 2015) would further reduce post-processingcostsTherequiredfrequencyofsamplingshouldalsobeconsideredonacase-by-casebasisdependentontheresearchquestion as image collection can be somewhat open-ended andresult inunwieldydata sets that require subsampling tobe cost-effective(Parnelletal2010)Inourcasethesettingof~5imagesperdayappearedtoadequatelydescribetheinterdailyvariation

Whileouroneautonomouscameraprovidedastrongcorrelationwithtemporaltrendsineffortcomparedtotheregionalsurveyitdidnotprovide relativespatialdistributionsofeffortacross the fisheryThesedistributionshowevercanbepredictableandhighlyconcen-trated(Lynch20062014Parnelletal2010)anobservationthatwas

alsoshownbyourresultsandmuchsmallerlevelsofsamplingeffortwouldprobablyberequiredtoresolvevariationinspatialdistributionscomparedtowhatisrequiredtoresolvethehightemporalvariationsineffortUnderstandingfine-scalespatialdistributionsofrecreationalfishingeffortiscommonlyachievedusingaerialcounton-watersur-veysorvantagepointsurveys(Lynch2006Smallwoodetal2012)thoughotherremotemethodssuchashigh-resolutionsatelliteimag-ery(FretwellScofieldampPhillips2017)mayprovideacheapersolu-tionThesetypesofwidersurveyshavealsobeensuggestedashighlycompatiblewithpointsourcemetricssuchasremotecamerasystemstoallowforcalibrationorldquoup-ratingrdquoforeffortexpansionstoestimateoveralleffortwithinthefishery(Hartilletal2016)

Thelongdistance(~6km)betweentheHMRCAandthedeploy-ment location stretched the resolution limit of the current CRAGSsetup aswewere unable to distinguish party size or activity typepurelyusing thecamerasystemThis isunlikepreviousCRAGSde-ploymentwhichhadahighsuccessrateofidentifyingpartysizeac-tivity typeata shorter rangeofapproximately19km (Woodetal2016) Thus ground-truthing information about the fisherywas re-quiredforthesystemtoworkeffectivelyAutonomouscamerasystemalsodonotcollectcreel(catch)databutbyallowingforcost-effectivecollectionofthehighlyvariabletemporaleffortdatatheymayallowforhumanresourcestobespentelsewhereduringsurveyperiods(egtoundertakeboatrampinterviews)(vanPoortenampBrydle2018)

Our studywas a small-scale trial of new technology andwhilepreliminary results are encouraging validated seasonal effort esti-mationsobservedviaCRAGSwereoutsidethescopeofthisprojectOur comparisons would hence have been improved by extendingCRAGSdeploymentsandwithmorecomparisonstoregionalsurveysSimilarly usingmultiple CRAGS simultaneously would have bettertested for differences in subregional patterns of temporal trendsorconcentrationof infishingeffortfor instanceboatsaroundthenearbyTasmanIsland

Insummarytheprogrammablesamplinganddataextractionof-feredbyhigh-resolutionautonomouscamerasystemscanallowforfine-scalelong-termmonitoringoffishingefforttrendsorpotentiallymanyothertypesofactivitiesatseascapescalesMorewidespreadadoptionofhigh-frequencysamplingapproachesforeffortestima-tionwouldalsohelpavoidcollectionofmisrepresentativedatabyidentifyingeffortanomalies(McCluskeyampLewison2008)suchastheweekdaycontinuationeffectofraisedeffortbetweenadjacentweekendfishingcompetitionByimprovingtheinformationdensityovertraditionalsurveymethodsautonomouscamerasystemsandotherremotesystemsmayprovideanewwaveofoptimizationfordirectmeasurementofcoastalhumanusesuchasrecreationalfish-eriesalthoughcomplimentarymethodsarestillrequiredtodeter-minespatialdistributionstargetspeciesandharvest

ACKNOWLEDG MENTS

We thankMsGeorginaWoodMs SaraAdkinsMsBelleMonkMrRSherringtonandMsPaigeKellyforconstructiveinsighten-thusiasmandassistanceduringfieldworkWealsothankMrPhillip

10emsp |emsp emspensp FLYNN et aL

CulversonMrStuartDudgeonandMrMattStapenellKarenEvensprovidedhelpfulcommentsonaninternaldraftGreatestthankstothemany recreational anglerswhowillingly shared information tosurveyclerkswithouttheirinvolvementthisprojectwouldnothavecometofruitionThreeanonymousreviewersprovidedcommentsthatimprovedthemanuscript

CONFLIC T OF INTERE S TS

Theauthorsdeclarenoconflictofinterests

AUTHOR CONTRIBUTIONS

MrFlynnhelpeddesignthestudyledthefieldworkundertookmostoftheanalysisandhelpwritethemanuscriptDrLynchhelpdesignthestudyassisted inthefieldworkprovidedadviceontheanaly-sisanddidsomeanalysisconstructedfiguresandafterMrFlynncontributedthemosttothedraftingofthemanuscriptDrBarretthelpeddesign the study assisted in some fieldwork provided ad-viceonanalysisandeditedthemanuscriptMrWonghelpedexten-sivelywithfieldworkassistedwithanalysisconstructedfiguresandhelpedwritethemanuscriptMsDevineassistedwithfieldworkandconfigurationoftheequipmentincludingtheapplicationofthebigdatamoviesoftwareandhandlingofphotographsandalsohelpedwritethemanuscriptMrHughesdevelopedtheelectronicsfortheequipmentoversaw thebuildofgearwrote the software topro-gramtherobotandreviewedandeditedthemanuscript

DATA ACCE SSIBILIT Y

Survey sampling schedules data collection sheets extrapolationproceduresandtechnicalCRAGSspecificationsarealluploadedasonlinesupportinginformationappendixRawdatafromsurveysandCRAGSDOIhttpdoiorg105061dryadv10d218

RE SE ARCH E THIC S COMPLIANCE AND FUNDERS

ResearchabidesbytheAustraliancodeofhumanresearchandex-perimentation(NationalStatementonEthicalConductinHumanResearch of 2007) Permission to conduct research on humanparticipants was granted by the University of Tasmania SocialScience under reference H0014772mdashldquoRemote investigation ofcoastal area user group quantification and variationrdquo aswell asfromtheCSIROSocialScienceHumanResearchEthicsCommitteeunder reference 08115mdashldquoHawkesbury Region Project HumanUsePatternsrdquoFundingwasbyUTASasPartoftheIMASHonoursProject funding and through internal support from the CoastsProgram

ORCID

Tim P Lynch httporcidorg0000-0002-5346-8454

R E FE R E N C E S

Alessa L Kliskey A amp Brown G (2008) Socialndashecological hotspotsmapping A spatial approach for identifying coupled socialndasheco-logicalspaceLandscape and Urban Planning8527ndash39httpsdoiorg101016jlandurbplan200709007

ArlinghausR(2006)Overcominghumanobstaclestoconservationofrecreational fishery resources with emphasis on central EuropeEnvironmental Conservation 33 46ndash59 httpsdoiorg101017S0376892906002700

Arlinghaus R Tillner R amp Bork M (2015) Explaining participationratesinrecreationalfishingacrossindustrialisedcountriesFisheries Management and Ecology 22 45ndash55 httpsdoiorg101111fme12075

BadcockPBPatrickKSmithAMASimpsonJMPennayDRisselCEhellipRichtersJ(2016)Differencesbetweenlandlineandmobilephoneusers in sexualbehavior researchArchives of Sexual Behavior46(6)1711ndash1721

Blumberg S J amp Luke J V (2009) Reevaluating the need for con-cern regarding noncoverage bias in landline surveys American Journal of Public Health 99 1806ndash1810 httpsdoiorg102105AJPH2008152835

Bousetouane F ampMorris B (2015) Off-the-shelf CNN features forfine-grainedclassificationofvesselsinamaritimeenvironmentInGBebisRBoyleBParvinDKoracinIPavlidisRFerisTMcGrawMElendtRKopperERaganZYeampGWeber(Eds)Advances in visual computing 11th International symposium ISVC 2015 Las Vegas NV USA December 14ndash16 2015 proceedings Part II (pp379ndash388)Cham Switzerland Springer International Publishing httpsdoiorg101007978-3-319-27863-6

Cooke S J amp Cowx I G (2004) The role of recreational fish-ing in global fish crises BioScience 54 857ndash859 httpsdoiorg1016410006-3568(2004)054[0857TRORFI]20CO2

DiggleP J (1995)Time-series a biostatistical introductionOxfordUKClarendonPress

EdwardsJampSchindlerE (2017)Avideomonitoringsystemtoeval-uate ocean recreational fishing effort in Astoria OregonMarine recreational information program (pp 31) Newport OR OregonDepartmentofFishandWildlife

FarrMStoecklNampSuttonS(2014)RecreationalfishingandboatingArethedeterminantsthesameMarine Policy47126ndash137httpsdoiorg101016jmarpol201402014

ForbesETraceySampLyle JM (2009)Assessment of the 2008 rec-reational gamefish fishery of Southeast Tasmania with particular refer-ence to Southern Bluefin TunaHobartTASTasmanianAquacultureandFisheriesInstituteUniversityofTasmania

FretwellPTScofieldPampPhillipsRA(2017)Usingsuper-highres-olutionsatelliteimagerytocensusthreatenedalbatrossesIbis159481ndash490httpsdoiorg101111ibi12482

GiriKampHallK (2015)SouthAustralian recreational fishingsurveyFisheries Victoria internal report series (pp 63) Queenscliff VicFisheriesVictoria

Griffiths S P Pollock K H Lyle J M Pepperell J G TonksM L amp Sawynok W (2010) Following the chain to elu-sive anglers Fish and Fisheries 11 220ndash228 httpsdoiorg101111j1467-2979201000354x

Halpern B SWalbridge S Selkoe K A Kappel C V Micheli FDrsquoAgrosaChellipWatsonR(2008)AglobalmapofhumanimpactonmarineecosystemsScience319948ndash952httpsdoiorg101126science1149345

Hartill BW Payne GW Rush N amp Bian R (2016) Bridging thetemporal gap Continuous and cost-effective monitoring of dy-namic recreational fisheries by web cameras and creel surveysFisheries Research 183 488ndash497 httpsdoiorg101016jfishres201606002

emspensp emsp | emsp11FLYNN et aL

HenryGWampLyleJM(2003)NationalrecreationalandindigenousfishingsurveyFinal report to the fisheries research amp development cor-poration and the fisheries action program (pp 188) Canberra ACTFRDC

JonesCMampPollockKH(2012)Recreationalanglersurveymeth-odsEstimationofeffortharvestandreleasedcatchInAZaleDLParrishampTMSutton(Eds)Fisheries techniquesBethesdaMDAmericanFisheriesSociety

Kearney R E (2002) Co-management The resolution of conflict be-tween commercial and recreational fishers in Victoria AustraliaOcean amp Coastal Management45201ndash214httpsdoiorg101016S0964-5691(02)00055-8

KellerKSteffeASLowryMMurphyJJampSuthersIM(2016)Monitoringboat-basedrecreationalfishingeffortatanearshorearti-ficialreefwithashore-basedcameraFisheries Research18184ndash92httpsdoiorg101016jfishres201603025

KinlochMAMcGlennonDNicollGampPikePG(1997)Evaluationof the bus-route creel surveymethod in a largeAustralianmarinerecreationalfisheryISurveydesignFisheries Research33101ndash121httpsdoiorg101016S0165-7836(97)00068-4

LewinWCArlinghausRampMehnerT(2006)Documentedandpo-tential biological impacts of recreational fishing Insights forman-agementandconservationReviews in Fisheries Science14305ndash367httpsdoiorg10108010641260600886455

LowryMampMurphy J (2003)Monitoring the recreational gamefishfisheryoffsouth-easternAustraliaMarine and Freshwater Research54425ndash434httpsdoiorg101071MF01269

LyleJMStarkKEampTraceySR(2014)2012ndash2013 Survey of rec-reational fishing in TasmaniaHobart TAS Institute forMarine andAntarcticStudies

Lynch T P (2006) Incorporation of recreational fishing ef-fort into design of marine protected areas Incorporacioacuten delEsfuerzo de Pesca Recreativa en el Disentildeo de Aacutereas MarinasProtegidas Conservation Biology 20 1466ndash1476 httpsdoiorg101111j1523-1739200600509x

LynchTP(2008)Thedifferencebetweenspatialandtemporalvaria-tioninrecreationalfisheriesforplanningofmarineprotectedareasResponse toSteffeConservation Biology22486ndash491httpsdoiorg101111j1523-1739200800888x

LynchTP (2014)Adecadal time-seriesof recreational fishingeffortcollectedduringandafterimplementationofamultipleusemarinepark shows high inter-annual but low spatial variability Fisheries Research15185ndash90httpsdoiorg101016jfishres201309014

LynchTPAldermanRampHobdayAJ(2015)Ahigh-resolutionpan-orama camera system formonitoring colony-wide seabird nestingbehaviourMethods in Ecology and Evolution6491ndash499httpsdoiorg1011112041-210X12339

LynchTFlynnDWoodGKimJ-HParkJampHobdayAJ(2017)Metrics parameters and results from fisheries and wildlife studies with the CSIRO Ruggedized Autonomous Gigapixel Systems (CRAGS)DarwinNTAustralianMarineScienceAssociation

LynchTPWilkinsonEMellingLHamiltonRMacreadyAampFearyS(2004)ConflictandimpactsofdiversandanglersinamarineparkEnvironmental Management 33 196ndash211 httpsdoiorg101007s00267-003-3014-6

MaHOgawaTKSminkeyTRBreidtFJLesserVMOpsomerJDhellipVanVoorheesDA(2018)Pilotsurveystoimprovemonitoringofmarine recreational fisheries inHawaiʻiFisheries Research204197ndash208httpsdoiorg101016jfishres201802010

McCluskey S M amp Lewison R L (2008) Quantifying fish-ing effort A synthesis of current methods and their ap-plications Fish and Fisheries 9 188ndash200 httpsdoiorg101111j1467-2979200800283x

McGlennonDampKinlochMA(1997)Evaluationofthebus-routecreelsurveymethod in a largeAustralianmarine recreational fishery II

PilotsurveysandoptimalsamplingallocationFisheries Research3389ndash99httpsdoiorg101016S0165-7836(97)00067-2

McPheeDLeadbitterDampSkilleterGA(2002)SwallowingthebaitIs recreational fishing in Australia ecologically sustainable Pacific Conservation Biology840httpsdoiorg101071PC020040

Monz C A Pickering C M amp HadwenW L (2013) Recent ad-vances in recreation ecology and the implications of differentrelationships between recreation use and ecological impactsFrontiers in Ecology and the Environment11441ndash446httpsdoiorg101890120358

MooreAHallKGiriKTraceySPenroseLHansenShellipBrownP(2015)Developingrobustandcost-effectivemethodsforestimatingthenationalrecreationalcatchofSouthernBluefinTunainAustraliaFRDC Projectpp123

Morton A J amp Lyle J M (2004) Preliminary assessment of the rec-reational gamefish fishery in Tasmania with particular reference to Southern Bluefin Tuna Taroona TAS Tasmanian Aquaculture andFisheriesInstitute

ParnellPEDaytonPKFisherRALoarieCCampDarrowRD(2010)SpatialpatternsoffishingeffortoffSanDiegoImplicationsfor zonal management and ecosystem function Ecological Applications202203ndash2222httpsdoiorg10189009-15431

PattersonTAEvansKCarterTLampGunnJS(2008)Movementand behaviour of large southern bluefin tuna (Thunnus maccoyii)in the Australian region determined using pop-up satellite ar-chival tags Fisheries Oceanography 17 352ndash367 httpsdoiorg101111j1365-2419200800483x

PollockKH JonesCMampBrownTL (1994)Angler survey meth-ods and their applications in fisheries management Bethesda MDAmericanFisheriesSociety

PonsM Branch T AMelnychukM C JensenO P Brodziak JFromentinJMhellipHilbornR(2017)EffectsofbiologicaleconomicandmanagementfactorsontunaandbillfishstockstatusFish and Fisheries181ndash21httpsdoiorg101111faf12163

PowersSPampAnsonK (2016)Estimating recreationaleffort in theGulf of Mexico Red Snapper Fishery using boat ramp camerasReduction in federalseason lengthdoesnotproportionally reducecatchNorth American Journal of Fisheries Management 36 1156ndash1166httpsdoiorg1010800275594720161198284

Robson D amp Jones C M (1989) The theoretical basis of an ac-cess site angler survey design Biometrics 45 83ndash98 httpsdoiorg1023072532036

Rocklin D Levrel H Drogou M Herfaut J amp Veron G (2014)Combining telephone surveys and fishing catches self-reportTheFrenchseabass recreational fisheryassessmentPLoS ONE9e87271httpsdoiorg101371journalpone0087271

Smallwood C B Pollock K HWise B S Hall N G amp GaughanD J (2012) Expanding aerial-roving surveys to include counts ofshore-based recreational fishers from remotely operated camerasBenefitslimitationsandcosteffectivenessNorth American Journal of Fisheries Management321265ndash1276httpsdoiorg101080027559472012728181

Teixeira D ZischkeM T ampWebley J A C (2016) Investigatingbias in recreational fishing surveys Fishers listed in public tele-phone directories fish similarly to their unlisted counterpartsFisheries Research 181 127ndash136 httpsdoiorg101016jfishres201604012

van Poorten B T amp Brydle S (2018) Estimating fishing effortfrom remote traffic counters Opportunities and challengesFisheries Research 204 231ndash238 httpsdoiorg101016jfishres201802024

van Poorten B T Carruthers T RWardHGM ampVarkeyD A(2015)Imputingrecreationalanglingeffortfromtime-lapsecamerasusinganhierarchicalBayesianmodelFisheries Research172265ndash273httpsdoiorg101016jfishres201507032

12emsp |emsp emspensp FLYNN et aL

VenturelliPAHyderKampSkovC(2017)AnglerappsasasourceofrecreationalfisheriesdataOpportunitieschallengesandproposedstandardsFish and Fisheries18578ndash595httpsdoiorg101111faf12189

WiseBSTelferCFLaiEKMHallNGampJacksonG(2012)Long-termmonitoringofboat-basedrecreational fishing inSharkBayWesternAustraliaProvidingscientificadviceforsustainablemanagement in a World Heritage Area Marine and Freshwater Research631129ndash1141httpsdoiorg101071MF12054

WoodGLynchTPDevineCKellerKampFigueiraW(2016)High-resolutionphoto-mosaictime-seriesimageryformonitoringhumanuseofanartificialreefEcology and Evolution66963ndash6968httpsdoiorg101002ece32342

YuFRupertACCDiTizianaMNishikantaKRandySIllahNhellipJeffreyG(2011)TerapixelimagingofcosmologicalsimulationsThe Astrophysical Journal Supplement Series19718

Zischke M T Griffiths S P amp Tibbetts I R (2012) Catch and ef-fort from a specialised recreational pelagic sport fishery off

eastern Australia Fisheries Research 127ndash128 61ndash72 httpsdoiorg101016jfishres201204011

SUPPORTING INFORMATION

Additional supporting information may be found online in theSupportingInformationsectionattheendofthearticle

How to cite this articleFlynnDJHLynchTPBarrettNSWongLSCDevineCHughesDGigapixelbigdatamoviesprovidecost-effectiveseascapescaledirectmeasurementsofopen-accesscoastalhumanusesuchasrecreationalfisheriesEcol Evol 2018001ndash12 httpsdoiorg101002ece34301

Graphical AbstractThecontentsofthispagewillbeusedaspartofthegraphicalabstractofhtmlonly

Itwillnotbepublishedaspartofmainarticle

Wetrailedaremoteautonomousultra-high-resolutionphotosamplingmethodasacosteffectsolutionfordirectseascape-scaledmeasure-mentsOursequentialphotosamplingwasbatchedprocessesusingnovelTimeMachinedatahandlingsoftwaretoproduceldquobigdatardquotimeseriesmoviesfromaspatialsubsetofanoffshorefisherywhichwecomparedtoaregionalbus-routesurveyofboattrailersandinterviewswithparticipantsOurcamerasamplescloselytrackedlargerscaleregionaltrendsbutwithlaborcostsaquarterofthebus-routesurveycostsforanindividualsample25ofthebusroutes

Page 4: Gigapixel big data movies provide cost-effective seascape

4emsp |emsp emspensp FLYNN et aL

locationof all three rampsWe fixed theCRAGS to a clifftop atWaterfall Baywhich overlooks the site fromvery high sea cliffs(~250m)withthecamerafocusingonthewesternsideofHRMCA(Figure1)ForeachsampleCRAGScapturedaseriesof68pho-tographsthroughatelephotolenswhichwhenstitchedtogetherprovide a high-resolution panoramic image with a wide field ofview (~140deg) (Supporting Information Appendix S1) The size ofareasurveyedbyCRAGScomparedtotheregionalareaaccessibleby trailer boatswas estimated using google earth polygons andtheUniversityofNewHampshireKMLextensiontoolmdashhttpsex-tensionunhedukmlToolsindexcfm

Thetimeof the initialCRAGSpanoramasamplewasrandomlyselected Samples were then collected sequentially through-out daylight hours (0600ndash1800) As less recreational fishing fortuna occurred at night no samples were taken with the CRAGSprogrammed to ldquoblankrdquo these hours from the sampling sched-ule (Supporting InformationAppendix S1) Photographic sessionswereprogrammedtobeseparatedbyarecurring93-mingapthus

producing a forward shift in sample capture time andminimizingpotentialautocorrelationsbetweenboatingactivitiesandsamplingtimesPhotographswerebatchprocessedtobothformpanoramicimagesforeachsampleandforcompilingintoaninteractivetime-lapsemovieusingCREATELabrsquosTimeMachinesoftware (CREATELabregPittsburgPAUSA)

TimeMachinemovies are fully interactive and are able to bezoomedintopausedandplayedatvariousspeedsandarealsotimeanddate-stamped(Figure3)UsingthisinteractivemoviesoftwareeachGigaPanmoviewassystematicallypausedateachsamplefullyzoomedandscannedlefttorighttoptobottomforobjectsintheseascapebytheleadauthor(Figure3)Vesselcountswererecordedforeachsamplewithobjectslargerthan~10minlength(nottrailerboats)or requiring longer than10seconds todistinguish (eg lowresolutionandimageartifact)beingdiscountedSampleswerecat-egorized into two strataweekdays andweekendspublic holidaysusing the time and date stamp provided on each GigaPan scene(Figure3)

F IGURE 2emspAnunmodifiedGigaPanEpicPROunit(left)andaCRAGSdeployedatWaterfallBaySETasmania(right)

F IGURE 3emspScreencaptureoutputsfromtheTimeMachineinteractivedataviewershowinganozoomandpanoramicextentoftheHippolyteRockMarineConservationArea(HRMCA)b=13zoommainHippolyterock~6kmfromcamerac=12zoomofalargetrailerboatd=fullzoom

lyn088
Cross-Out
lyn088
Inserted Text
(a)
lyn088
Cross-Out
lyn088
Inserted Text
(b)
lyn088
Cross-Out
lyn088
Inserted Text
(c)
lyn088
Cross-Out
lyn088
Inserted Text
(d)
lyn088
Inserted Text
s

emspensp emsp | emsp5FLYNN et aL

Thecountofboatsforeachsamplewasthenextrapolatedintoanestimateofdailyaverageboatingeffort(hoursplusmnSE)usingthein-stantaneouscountmethod(Equation1)

where ei is theextrapolatedboatinghours for the ithday Ii is theaveragenumberoftrailerboatsobservedacrosstheinstantaneoussamplesfordayiandTisthedailysamplingperiod(daylighthours)Asbadweathercanbeamajorfactoraffectingthedecisiontoboatorfish(ForbesTraceyampLyle2009)boatingeffortwasassumedtobe0whenadverseweathermadeobservationsofboatsimpossible

Next total effort for each day-type strata (Ejk) within sam-pling months (k) was calculated to estimate the monthly effort(Equation2)AsCRAGSallowedforcontinuoussamplingthroughoutthesurveyperiodthesamplingprobability(πk)wassetto1

Standarderrorandeffortvarianceformonthlyestimateswerecalculated(Equation3)accordingtoPollocketal(1994)forstrata

where nj is thenumberofsampleddaysofstrata j Njk is thetotalnumberofdaysofstratajinmonthk

Asvariationforeachstratawascalculatedseparatelyweusedasumofsquaredstandarderrorsapproach(Equation4)toestimatethetotalmonthlyvariation

Assampleswereobtainedinaserialandcontinuousfashionatimeseriesmodelusingrollingaverageswasalsoappliedtothedata(Diggle1995)(Equation5)

where smt is the smoothed average of extrapolated effort e persampletwhichisweightedacrossthreesampleseachseparatedby93minThisprovidedasmoothedaveragewithindays

Wealsoconductedabus-routestylesurveytoestimatefishingeffortfromtrailerboatsfortheregion(Pollocketal1994Robsonamp Jones1989)WhileCRAGSsampleswerecollected throughoutMayandJunethecomparablebus-routesurveywasonlyconductedinMaydue to logistical constraintsWesurveyed the threemajorboatrampsintheregionatPiratesBayFortescueBayandStewartsBaybetween3May2015and30May2015(Figure1)

Atotalof20bus-routesampleswerescheduledWeundertookrandomstratifiedsamplingwithdaytypedividedintoweekdaysorweekendpublic holidays To reflect expected higher recreationaleffortduringweekendsandpublicholidays(McCluskeyampLewison2008)thesamplingprobability(πk)wasweightedtowardthisstrata

over weekdays at 12 to eight samples (Supporting InformationAppendix S2) We stratified the sample time into morning (am0600ndash1159)andafternoon(pm1200ndash1800)withequalproba-bilitiesofsamplingTominimizetemporalautocorrelationthestart-ing location and travel direction of the routewere also randomlyselectedwithoutreplacementEqualsamplingweightwasprovidedtoeachrampduetolackofpriorknowledgeofuserates

At each boat ramp surveys were conducted by first countingparkedboattrailersthenbymonitoringvesselentryandexitsduringa1-hrsamplingperiod(KinlochMcGlennonNicollampPike1997Pollocketal1994)The1-hrperiodwaschosendueto travel timeanddis-tancebetweensitesallowingforabus-routesampletobecompletedforallrampswithinthetemporalstrata(AMorPM)ofthedailysam-plingframeInterviewswereconductedwithallpartieslaunchingandretrievingvesselsduringtheclerkrsquoswaittimedeterminingpartysizetargetspeciesgroupandactivity(iefishingornot)forestimatesoffishing effort This well-established procedure produced a monthlyextrapolatedestimationofboateffort instandardizedunitsofefforthoursplusmnSEwhichcouldthenbeadjustedtofishingeffortviathepro-portionofintervieweesactivities(Pollocketal1994RobsonampJones1989)(SupportingInformationAppendixS3)PartysizebetweenrampswastestedforanydifferenceusingaonefactorANOVA

Thebus-route surveyhenceprovidesanestimationofall fish-ing effort from trailer boats around this isolated peninsular TheCRAGSdataarethusahigh-frequencyspatialsubsetoftheentiretrailer boat fishing effort for the period thatwas extrapolated bytheregionalsurveyThisallowedustobothtesttherelativeuseoftheHRMCAcomparedtowiderpeninsularuseandtheabilityofourremoteobservationswithourCRAGsunitofanactualsubsetofthefisherytotracklargertemporalpatternsofpeninsularwideeffort

TovalidatetherepresentativenessoftheCRAGShigh-frequencytrackingofeffortmetricsover timeweundertookaSpearmanrankcorrelationbetweentime-matchedeffortestimationsbyCRAGSandthe regional bus-route survey For graphing bus-routeextrapolationestimates were overlaid onto extrapolations from daily averages ofboatfishingeffortfromCRAGSWealsocarefullyloggedalltimespentontheprojecttocomparetotallaborcostsbetweenCRAGSandthebusrouteforbothdatacollectionandprocessingtodetermineeffortacrossthecomparativemonthaswellasforcostpersample

3emsp |emspRESULTS

The CRAGS operated successfully and continuously throughoutthestudycapturing545plusmn198SE(May)and483plusmn088SE(June)panoramasperdaymakinga totalof314 samplesWemade895observationsofboatsbut51samplesacross21daysreturnedzerocountsofboats(17oftotalsamples)Thesewereconcentratedonweekdaysmdashwith 44 zero-count samples across 18daysmdashwhile onweekendsandpublicholidaytherewereonlysevensamplesacross3dayswithnoactivityInclementweatherrendered27samplesun-usableTheimagesandbatch-processedTimeMachinemoviescon-sumed4274GBofmemory

(1)ei=IitimesT

(2)Ejk=

nsum

i=1

(

ei

120587k

)

(3)SEjk=

1

njminus1

sumn

i=1(eiminus ei)

2

njtimes (Njk)

2

(4)SEk=

radic

SE21k+SE2

2khellip+SE2

jk

(5)smt=etminus1+ et+ et+1

3

6emsp |emsp emspensp FLYNN et aL

Based on boat ramp interviews the average trailer boat partysize (P) was 287 people per vessel throughout May (plusmn0133 SEn=84)andpartysizebetweensitesshowednosignificantdiffer-ence (df=2F = 302p = 068) The averagemonthly boat fishingeffort(hours)extrapolatedforourCRAGSobservationsofHRMCAwas5629hr(plusmn711SE)with4096hr(plusmn607SE)inMayand1534hr(plusmn370 SE) in June (Figure4)mdashthis could be extrapolated to fisherhoursbymultiplyingbytheaveragepartysizebutasnointerviewswereconductedinJuneweleftboththeCRAGSandbusrouteex-trapolationsasboatfishingeffortonly

The total subset area observed by CRAGSwas 96km2 which comparedto318km2thatwaseasilyavailabletotrailerboatsaroundtheTasmanPeninsularwhichcorrespondstoaround3ofthetotalareaComparedtothetotalfishingeffortforMayextrapolatedfromtheregionalbus-routesurvey therelativelysmallareaaroundtheHRM(~3)receivedalargeamountofthefishingeffort(~23)

ThehighfrequencyofCRAGSsamplingpermittedtheexamina-tionoffine-scaleinter-dailytemporalvariationwhichdisplayedarag-gedsinusoidoscillation(Figure5)Effortwasgenerallylowwithinthe

weekdaystrataandthen increasedrapidly to largepeaksonweek-ends (Figure5) One exception to this was between the 18th and22ndMaywhichhadmoreeffortonweekdays thanotherperiods(Figure5)Bracketingthisperiodweretwoweekendsoffishingcom-petitions theldquoTunaClubofTasmaniarsquosrsquoFarSouthClassicrdquo (16ndash17thMay)andRally4Northerninvitationalweekendrally(23rdMay)

A total of 16 bus-route samples were conducted logistic fail-uresledtotheabandonmentoffour(20)oftheplannedsamplesRegionalTrailerboateffortfromaroundtheTasmanPeninsularwasestimatedtobe19650hr (plusmn2858SE) inMay (Figure4)Notrailersat individual rampswere recorded four timeswith all of theseoc-curringonweekdaysThetotalnumberoftrailerssightedperramprangedfrom0to16duringweekdaysand1ndash54forweekendswiththehighestnumberrecordedon16May2015atPiratesBayDuringhigh-intensitydays(particularlyweekendsholidaysmornings)therewereoftenlargeoverflowsfromrampcarparkswithparkedcarsandboattrailersextendingasmuchasonekilometerfromtheboatramp

During thebus-route survey 84 interviewswere conductedThemajorityofmariners(905n=96)indicatedfishingwastheir

F IGURE 4emspExtrapolatedmonthlytrailerboateffort(hoursplusmnSE)fortheentireTasmanPeninsulaassessedbyabus-routesurvey(May)andforthesubsetoftheareaaroundtheHRMCAsampledwithCRAGS(MayandJune)

F IGURE 5emspHigh-frequencyinterdailyextrapolatedCRAGSestimationsofboatefforthoursaroundthe(HRMCA)

0

100

200

300

400

500

600

Rolli

ng m

ean

daily

effo

rt in

hou

rs

DayWeekdays Weekends + Public Holidays

lyn088
Inserted Text
CA
lyn088
Cross-Out
lyn088
Cross-Out
lyn088
Inserted Text
t

emspensp emsp | emsp7FLYNN et aL

primaryactivity (Table1)Overall643 (n=54) indicated theirprimary target were tuna with 179 (n=15) specifically tar-geting southern bluefin tuna (Thunnus maccoyii)Other commonrecreationaltargetsincludedflathead(Platycephalusspp595)Southernrocklobster(Jasus edwardsii107)andsquidcalamari(Sepioteuthis australisandNototodarus gouldi476)(Table1)

Basedon the responsesofactivity type from interviews totalboatfishingeffortinMaywasestimatedtobe17783hr(plusmn2586SE)Asmostnonfishingactivitieswerenear shoreweassumed trailerboatingactivityatHMRCAobservedbyCRAGSwasallfishing

Therewere16directcomparisonsamplesbetweenthebusrouteandCRAGSmethodsbetweenthe2ndandthe29thofMaySamplesconsisted of 7 weekdays and 9 weekendspublic holiday samples

(Figure6)Infourcaseswheretwobus-routesamplesweretakenonthe same day (Supporting Information Appendix S2) they are com-bined toadailyestimation for tabulation (Table2)Theproportionaldifferencebetweenestimatedeffortsusingthetwomethodsrangedfrom078-to418-folddifferencehoweverranksestimatesbetweenCRAGS and bus-route samples were strongly associated (n=16ρ(12)=087p=lt001)Interdailysamplingstrata(ampm)werealsoexaminedseparatelyforcorrelationwhichremainedstrongforbotham(n=9ρ(7)=080p=lt001)andpm(n=7ρ(5)=089p=lt001)

ForthecomparativemonthCRAGStook169samplescomparedtothebusroutesrsquo16(Table3)WehadtwofieldtripdaysforCRAGSwith four people on the setup and three on the breakdown daywhichresultedinsevenfull-timeequivalent(FTE)daysofworkThiscomparedto16fielddaysforthebus-routemethodwhichincludedtraveltothefieldsiteforbetweentwoandthreestaffwhichaccu-mulatedto40FTEdaysofworkSettingupourCRAGSautomatedstitching took one FTE day (though this is faster for experiencedofficers)whiledataextractionperimagewasestimatedtotakeon

TABLE 1emsp Intervieweesreportedprimarytargettaxaoractivity

Primary target taxaspeciesactivity Interview responses

Tuna(unspecified) 37 440

SouthernBluefinTuna 15 179

AlbacoreTuna 2 238

AllTuna 54 643

Flathead 5 595

Southernrocklobster 9 107

Squidcalamari 4 476

Abalone 1 119

SmallPelagics 1 119

SnottyTrevalla 2 238

Allnearshorespecies 22 262

Surfing 1 119

SCUBA 3 357

DemonCavevisitors 3 357

PleasureCruise 1 119

Allnonextractive 8 952

Note ldquoTunardquo responsesdenote anglers targeting all specieswithin thefamilyThunnini

F IGURE 6emspDailyboatingeffort(hours)extrapolatedusingbus-routeaccesspointsurvey(bar)andCRAGS(line)Effortestimatefrombus-routesurveywascalculatedfrommorning(amdarkgray)andafternoon(pmlightgray)surveyCRAGSsurveywasconductedbetweenMayandJunewhilethebus-routesurveywasonlyconductedinMay

TABLE 2emspRankingofeffortforpairedsamplesbetweenCRAGSandthebusroute

DateCRAGS hours (rank)

Bus- route hours (rank) Δ Rank

2052015 330(9) 1208(10) minus1

3052015 303(8) 1471(11) minus3

12052015 0(1) 48(2) minus1

14052015 0(1) 24(1) 0

16052015 387(11) 1794(12) minus1

19052015 93(4) 89(4) 0

20052015 57(3) 117(5) minus2

22052015 228(7) 283(6) 1

23052015 513(12) 414(8) 4

24052015 345(10) 1155(9) 1

25052015 122(6) 323(7) minus1

29052015 102(5) 79(3) 1

8emsp |emsp emspensp FLYNN et aL

average75mincomparedto1minforenteringthebus-routedatasheetWhiledataextraction forCRAGS took longer compared tothe bus routemdashat 381 FTE daysmdashtotal monthly labor for CRAGStookonlyaquarterofthetimeandonaper-samplecomparisonwasonly25ofthebus-routelaborcost

4emsp |emspDISCUSSION

Over our autumn samplingmost trailer boat owners accessing thewaters around theTasmanPeninsula via the regional boating infra-structurewererecreationallyfishingwithfisherspredominatelytar-getingtunaTrailerboatrecreationalfishersconstitutealargetrancheof coastal users in Australia (McPhee etal 2002) and our resultssuggestedthatthisregionremainsonewhererecreationalfishing isconcentrated(MortonampLyle2004)EffortobservedbyCRAGSwasa spatial subsetof thegeneral areaeasily accessedby trailer boatsaroundtheTasmanPeninsularwhichweregionallyassessedwithourbus-routerovingsurveyAsCRAGSonlyobserved~3of thetotalarea but corresponded to ~23 of themonthly bus-route boat ef-fort theHippolyteRocksMarineConservationArea (HRMCA) is afurtherconcentrationpointwithintheregionforrecreationalfishingAsthisoffshorefishingeffortaroundtheHRMCAwasmostprobablyfocusedontotunamdashwhichaccountedfor634ofallinterviewedfish-ers activitiesmdashthe actual proportionofpeninsularwide tuna fishingthatwasobservedbyCRAGSwaspotentiallymuchhigherthan23

For developing effort metrics finding a representative site toobservemaybe important forprecise trackingof trendsThetightrankcorrelationbetweenpairedregionalbus-routeandCRAGScam-era samples suggest that effort atHRMCAwas representative ofregionaltemporaltrendsOurresultsarealsosimilartootherstud-ieswheresamplingmethodsforrecreationalfisheriesoverbroaderscaleswhencomparedtoindextrendsofeffortcollectedfrompointsourcesshowstrongcorrelations(Hartilletal2016)

CRAGSwithitsseascapescaleofdatacaptureobservesactualeffortonfishinggroundsratherthan indirectmetricsfrommove-ment past access points (Hartill etal 2016 Smallwood PollockWise Hall amp Gaughan 2012 van Poorten amp Brydle 2018) This

makesthechoiceofaccesspointstomonitorimmaterialfortrendcol-lectionandmaybeofparticularinterestforpelagicfisheriesWhilenonintuitiverecreationalfishingforwide-rangingmigratorypelagicspeciesisoftenspatiallyconcentratedintosmallareas(Lynch2006)duetotightoverlapsbetweeneaseofaccessbyfishersandfishbe-haviorrelatedtobathymetrymigrationhabitatandpreyavailability(Lynchetal2004PattersonEvansCarterampGunn2008)

Unlikespatialpatternstheintensityofcoastalrecreationaltemporaleffortcanbehighlyvariableovertime(Lynch20082014WiseTelferLaiHallampJackson2012)thoughitisgenerallythoughttofollowpre-dictablepatternsrelativetoholidayandnonholidayperiods(JonesampPollock2012)Ourresultsdemonstratedtheselargefluctuationsrang-ingfromzerouptogt400hroftrailerboateffortforadjacentdaysIncombinationwithholidayperiodsothersourcesofvariabilitymayalsooccurduetoindividualorcombinationsoffactorssuchasfishingcluborcompetitionactivitiesandweatherconditionsHigh-frequencysam-plingcanovercomethesecommontemporaldisjunctionsinobservedfishingeffortwherebychanceduetolownumbersofsampleslogis-ticallypermissiblewithtraditionalrovingoraccesspointsurveyslargeerrorsintheestimationofeffortcanbeintroduced(Hartilletal2016vanPoortenampBrydle2018vanPoortenetal2015)

With multiple daily observations abnormal episodes are morelikelytobeobservedForinstanceaweek-longperiodofintenseef-fortwasrecordedbetween15May2015and27May2015whichwasbracketedbytwoconsecutiveweekendswherefishingcompetitionswereheldontheTasmanPeninsularThenormalsinusoidpatternoflowweekdayandhighweekendeffortwasmuchlesspronouncedastheweekdaylullsineffortwerepartiallybridgedbyanglerscontinuingtofishAsbus-routedesignsusedaytypeasstratawithlowerprob-abilitiesforsamplingweekdaysandmuchfewersamplesoverallthistypeofeffectcouldeasilybemissedWhileourbus-routesamplesdidoverlapwiththetunafishingcompetitionsthiswaspurelybychanceSucheventswhichresultinlargespikesofeffortcanhavemarkedim-pactsonlocalecosystemsandharvestestimates(McPheeetal2002MonzPickeringampHadwen2013)andwithoutpriorknowledgearemorelikelytobecapturedwithmonitoringathigherfrequencies

Methodsforgatheringdataoncoastalusesuchasrecreationalfishing moved away from direct approaches and toward off-sitemethodssuchastelephoneinterviews(McCluskeyampLewison2008Pollock etal 1994) due to unsustainable costs from staffing andoperationsparticularlyduringperiodsof intenseusesuchasearlymornings weekends and public holidays which require overtimepaymentsHoweversamplingissueswithoff-sitemethodsareofpar-ticularconcernforopen-accessactivitiessuchasunlicensedfisheriesasthereisnolicensedatabaseofcontactphonenumbersonwhichtobaseasampleframeThisisparticularlysofornicherecreationalfisheriessuchaspelagicgamefishingasthefishersbecomeincreas-inglydilutewithinthegeneralpopulationandhencearehardtoac-cessparticularlyviaoff-sitemethodssuchastelephone interviews(Griffithsetal2010)

Remoteandautonomousdeploymentofsensorsmaythereforeprovideanalternativeandcost-effectivemethodforlong-termmon-itoring particularly for effort across a range of applicationsOur

TABLE 3emspFulltimeequivalent(FTE)labordaysforCRAGSandbusroutesurveystoestimatepatternsofeffort

CRAGS Bus route

Numberofsamples 169 16

Fielddays 2 16

Averagepeopleperfieldday 35 25

Field FTE days subtotals 7 40

StitchingsetupFTE 1 0

DataextractionFTE(persample) 0017 0002

Subtotal data extraction FTE 381 004

FTEdayspersample 006 250

Total FTE days 108 400

emspensp emsp | emsp9FLYNN et aL

comparisonsofmatchedpairsofdatafromtwomethodssuggestedthat theCRAGSoutputswerewell correlatedwith regionalefforttrendsOuranalysisoflabordemonstratesthattryingtounderstandthistemporalvariabilitywithsamplingviaboatrampsurveyswouldbeprohibitivelyexpensiveOurCRAGSallowsforatemporaleffortmetric derived fromactual observationsof fishing to be cheaplydeveloped with sampling frequency at subdaily scales This high-frequencysamplingapproachwasvalidatedagainstour largerbutmuchmorecostlybus-routesurveyOuranalysisoflaborrequiredtobothcollectandprocessdatashowedthatCRAGSrequiredconsid-erablylessresourcestocollectapproximatelyninetimesmorefre-quenttemporaldatathanthebus-routemethodMajorlaborsavingswereduetotheruggedizedandautonomouscapabilityofCRAGSasonlytwofieldtripswererequired(deploymentandretrievaltook7FTEdays)comparedtothebus-routersquos16fieldtrips inamonthwhichconsumed40FTEdaysandincludedanadditional780kmofinter-ramproadtravel(SupportingInformationAppendixS3)AsitisweatherproofautonomousandabletobeprogrammedwithatimedelaysetupCRAGSldquofireandforgetrdquonaturebothremovestheneedforobservertobephysicallylocatedattheregiontocountactivity(EdwardsampSchindler2017)andworkduringovertimeperiodsTheldquoblankingperiodrdquowhichstopsandstartssamplingmdashinthiscaseover-nightmdashalso reduces battery and memory use In combined thesetechnological advances have allowed long-term CRAGS deploy-mentsacross4yearsof4ndash6monthrotationsontooffshoreislandsand for several seasons in Antarctica to monitor seabirds (Lynchetal2017)Thelaborestimateforthebusroutewasconservativeasitdidnotincludeovertimeloadingforthe6weekenddaysofsam-plingorearlymorningcommencementofamsamples

Theadditional381FTEprocessingtimerequiredbythecameraapproach did not impact greatly on the overall labor budgetwhenconsideredagainstfieldworkTheuseoftheTimeMachinebatchpro-cessingsoftwaretoautomaticallyproducetheldquobigdatardquointeractivemoviewasalsoatechnologicalinnovationthatsignificantlyreducedthelaborofdatahandlingcomparedtothemanualopeningofindivid-ualGigaPansusedinpreviousstudies(Lynchetal2015Woodetal2016)IncomparativetermsattheindividualsampleleveltotracktheeffortmetriceachCRAGSsamplerequired24ofthelaborofthebus-routesurvey

Thedevelopmentandapplicationofimagerecognitionsoftware(machine learning) to detect classify and count boats automati-cally (Bousetouane amp Morris 2015) would further reduce post-processingcostsTherequiredfrequencyofsamplingshouldalsobeconsideredonacase-by-casebasisdependentontheresearchquestion as image collection can be somewhat open-ended andresult inunwieldydata sets that require subsampling tobe cost-effective(Parnelletal2010)Inourcasethesettingof~5imagesperdayappearedtoadequatelydescribetheinterdailyvariation

Whileouroneautonomouscameraprovidedastrongcorrelationwithtemporaltrendsineffortcomparedtotheregionalsurveyitdidnotprovide relativespatialdistributionsofeffortacross the fisheryThesedistributionshowevercanbepredictableandhighlyconcen-trated(Lynch20062014Parnelletal2010)anobservationthatwas

alsoshownbyourresultsandmuchsmallerlevelsofsamplingeffortwouldprobablyberequiredtoresolvevariationinspatialdistributionscomparedtowhatisrequiredtoresolvethehightemporalvariationsineffortUnderstandingfine-scalespatialdistributionsofrecreationalfishingeffortiscommonlyachievedusingaerialcounton-watersur-veysorvantagepointsurveys(Lynch2006Smallwoodetal2012)thoughotherremotemethodssuchashigh-resolutionsatelliteimag-ery(FretwellScofieldampPhillips2017)mayprovideacheapersolu-tionThesetypesofwidersurveyshavealsobeensuggestedashighlycompatiblewithpointsourcemetricssuchasremotecamerasystemstoallowforcalibrationorldquoup-ratingrdquoforeffortexpansionstoestimateoveralleffortwithinthefishery(Hartilletal2016)

Thelongdistance(~6km)betweentheHMRCAandthedeploy-ment location stretched the resolution limit of the current CRAGSsetup aswewere unable to distinguish party size or activity typepurelyusing thecamerasystemThis isunlikepreviousCRAGSde-ploymentwhichhadahighsuccessrateofidentifyingpartysizeac-tivity typeata shorter rangeofapproximately19km (Woodetal2016) Thus ground-truthing information about the fisherywas re-quiredforthesystemtoworkeffectivelyAutonomouscamerasystemalsodonotcollectcreel(catch)databutbyallowingforcost-effectivecollectionofthehighlyvariabletemporaleffortdatatheymayallowforhumanresourcestobespentelsewhereduringsurveyperiods(egtoundertakeboatrampinterviews)(vanPoortenampBrydle2018)

Our studywas a small-scale trial of new technology andwhilepreliminary results are encouraging validated seasonal effort esti-mationsobservedviaCRAGSwereoutsidethescopeofthisprojectOur comparisons would hence have been improved by extendingCRAGSdeploymentsandwithmorecomparisonstoregionalsurveysSimilarly usingmultiple CRAGS simultaneously would have bettertested for differences in subregional patterns of temporal trendsorconcentrationof infishingeffortfor instanceboatsaroundthenearbyTasmanIsland

Insummarytheprogrammablesamplinganddataextractionof-feredbyhigh-resolutionautonomouscamerasystemscanallowforfine-scalelong-termmonitoringoffishingefforttrendsorpotentiallymanyothertypesofactivitiesatseascapescalesMorewidespreadadoptionofhigh-frequencysamplingapproachesforeffortestima-tionwouldalsohelpavoidcollectionofmisrepresentativedatabyidentifyingeffortanomalies(McCluskeyampLewison2008)suchastheweekdaycontinuationeffectofraisedeffortbetweenadjacentweekendfishingcompetitionByimprovingtheinformationdensityovertraditionalsurveymethodsautonomouscamerasystemsandotherremotesystemsmayprovideanewwaveofoptimizationfordirectmeasurementofcoastalhumanusesuchasrecreationalfish-eriesalthoughcomplimentarymethodsarestillrequiredtodeter-minespatialdistributionstargetspeciesandharvest

ACKNOWLEDG MENTS

We thankMsGeorginaWoodMs SaraAdkinsMsBelleMonkMrRSherringtonandMsPaigeKellyforconstructiveinsighten-thusiasmandassistanceduringfieldworkWealsothankMrPhillip

10emsp |emsp emspensp FLYNN et aL

CulversonMrStuartDudgeonandMrMattStapenellKarenEvensprovidedhelpfulcommentsonaninternaldraftGreatestthankstothemany recreational anglerswhowillingly shared information tosurveyclerkswithouttheirinvolvementthisprojectwouldnothavecometofruitionThreeanonymousreviewersprovidedcommentsthatimprovedthemanuscript

CONFLIC T OF INTERE S TS

Theauthorsdeclarenoconflictofinterests

AUTHOR CONTRIBUTIONS

MrFlynnhelpeddesignthestudyledthefieldworkundertookmostoftheanalysisandhelpwritethemanuscriptDrLynchhelpdesignthestudyassisted inthefieldworkprovidedadviceontheanaly-sisanddidsomeanalysisconstructedfiguresandafterMrFlynncontributedthemosttothedraftingofthemanuscriptDrBarretthelpeddesign the study assisted in some fieldwork provided ad-viceonanalysisandeditedthemanuscriptMrWonghelpedexten-sivelywithfieldworkassistedwithanalysisconstructedfiguresandhelpedwritethemanuscriptMsDevineassistedwithfieldworkandconfigurationoftheequipmentincludingtheapplicationofthebigdatamoviesoftwareandhandlingofphotographsandalsohelpedwritethemanuscriptMrHughesdevelopedtheelectronicsfortheequipmentoversaw thebuildofgearwrote the software topro-gramtherobotandreviewedandeditedthemanuscript

DATA ACCE SSIBILIT Y

Survey sampling schedules data collection sheets extrapolationproceduresandtechnicalCRAGSspecificationsarealluploadedasonlinesupportinginformationappendixRawdatafromsurveysandCRAGSDOIhttpdoiorg105061dryadv10d218

RE SE ARCH E THIC S COMPLIANCE AND FUNDERS

ResearchabidesbytheAustraliancodeofhumanresearchandex-perimentation(NationalStatementonEthicalConductinHumanResearch of 2007) Permission to conduct research on humanparticipants was granted by the University of Tasmania SocialScience under reference H0014772mdashldquoRemote investigation ofcoastal area user group quantification and variationrdquo aswell asfromtheCSIROSocialScienceHumanResearchEthicsCommitteeunder reference 08115mdashldquoHawkesbury Region Project HumanUsePatternsrdquoFundingwasbyUTASasPartoftheIMASHonoursProject funding and through internal support from the CoastsProgram

ORCID

Tim P Lynch httporcidorg0000-0002-5346-8454

R E FE R E N C E S

Alessa L Kliskey A amp Brown G (2008) Socialndashecological hotspotsmapping A spatial approach for identifying coupled socialndasheco-logicalspaceLandscape and Urban Planning8527ndash39httpsdoiorg101016jlandurbplan200709007

ArlinghausR(2006)Overcominghumanobstaclestoconservationofrecreational fishery resources with emphasis on central EuropeEnvironmental Conservation 33 46ndash59 httpsdoiorg101017S0376892906002700

Arlinghaus R Tillner R amp Bork M (2015) Explaining participationratesinrecreationalfishingacrossindustrialisedcountriesFisheries Management and Ecology 22 45ndash55 httpsdoiorg101111fme12075

BadcockPBPatrickKSmithAMASimpsonJMPennayDRisselCEhellipRichtersJ(2016)Differencesbetweenlandlineandmobilephoneusers in sexualbehavior researchArchives of Sexual Behavior46(6)1711ndash1721

Blumberg S J amp Luke J V (2009) Reevaluating the need for con-cern regarding noncoverage bias in landline surveys American Journal of Public Health 99 1806ndash1810 httpsdoiorg102105AJPH2008152835

Bousetouane F ampMorris B (2015) Off-the-shelf CNN features forfine-grainedclassificationofvesselsinamaritimeenvironmentInGBebisRBoyleBParvinDKoracinIPavlidisRFerisTMcGrawMElendtRKopperERaganZYeampGWeber(Eds)Advances in visual computing 11th International symposium ISVC 2015 Las Vegas NV USA December 14ndash16 2015 proceedings Part II (pp379ndash388)Cham Switzerland Springer International Publishing httpsdoiorg101007978-3-319-27863-6

Cooke S J amp Cowx I G (2004) The role of recreational fish-ing in global fish crises BioScience 54 857ndash859 httpsdoiorg1016410006-3568(2004)054[0857TRORFI]20CO2

DiggleP J (1995)Time-series a biostatistical introductionOxfordUKClarendonPress

EdwardsJampSchindlerE (2017)Avideomonitoringsystemtoeval-uate ocean recreational fishing effort in Astoria OregonMarine recreational information program (pp 31) Newport OR OregonDepartmentofFishandWildlife

FarrMStoecklNampSuttonS(2014)RecreationalfishingandboatingArethedeterminantsthesameMarine Policy47126ndash137httpsdoiorg101016jmarpol201402014

ForbesETraceySampLyle JM (2009)Assessment of the 2008 rec-reational gamefish fishery of Southeast Tasmania with particular refer-ence to Southern Bluefin TunaHobartTASTasmanianAquacultureandFisheriesInstituteUniversityofTasmania

FretwellPTScofieldPampPhillipsRA(2017)Usingsuper-highres-olutionsatelliteimagerytocensusthreatenedalbatrossesIbis159481ndash490httpsdoiorg101111ibi12482

GiriKampHallK (2015)SouthAustralian recreational fishingsurveyFisheries Victoria internal report series (pp 63) Queenscliff VicFisheriesVictoria

Griffiths S P Pollock K H Lyle J M Pepperell J G TonksM L amp Sawynok W (2010) Following the chain to elu-sive anglers Fish and Fisheries 11 220ndash228 httpsdoiorg101111j1467-2979201000354x

Halpern B SWalbridge S Selkoe K A Kappel C V Micheli FDrsquoAgrosaChellipWatsonR(2008)AglobalmapofhumanimpactonmarineecosystemsScience319948ndash952httpsdoiorg101126science1149345

Hartill BW Payne GW Rush N amp Bian R (2016) Bridging thetemporal gap Continuous and cost-effective monitoring of dy-namic recreational fisheries by web cameras and creel surveysFisheries Research 183 488ndash497 httpsdoiorg101016jfishres201606002

emspensp emsp | emsp11FLYNN et aL

HenryGWampLyleJM(2003)NationalrecreationalandindigenousfishingsurveyFinal report to the fisheries research amp development cor-poration and the fisheries action program (pp 188) Canberra ACTFRDC

JonesCMampPollockKH(2012)Recreationalanglersurveymeth-odsEstimationofeffortharvestandreleasedcatchInAZaleDLParrishampTMSutton(Eds)Fisheries techniquesBethesdaMDAmericanFisheriesSociety

Kearney R E (2002) Co-management The resolution of conflict be-tween commercial and recreational fishers in Victoria AustraliaOcean amp Coastal Management45201ndash214httpsdoiorg101016S0964-5691(02)00055-8

KellerKSteffeASLowryMMurphyJJampSuthersIM(2016)Monitoringboat-basedrecreationalfishingeffortatanearshorearti-ficialreefwithashore-basedcameraFisheries Research18184ndash92httpsdoiorg101016jfishres201603025

KinlochMAMcGlennonDNicollGampPikePG(1997)Evaluationof the bus-route creel surveymethod in a largeAustralianmarinerecreationalfisheryISurveydesignFisheries Research33101ndash121httpsdoiorg101016S0165-7836(97)00068-4

LewinWCArlinghausRampMehnerT(2006)Documentedandpo-tential biological impacts of recreational fishing Insights forman-agementandconservationReviews in Fisheries Science14305ndash367httpsdoiorg10108010641260600886455

LowryMampMurphy J (2003)Monitoring the recreational gamefishfisheryoffsouth-easternAustraliaMarine and Freshwater Research54425ndash434httpsdoiorg101071MF01269

LyleJMStarkKEampTraceySR(2014)2012ndash2013 Survey of rec-reational fishing in TasmaniaHobart TAS Institute forMarine andAntarcticStudies

Lynch T P (2006) Incorporation of recreational fishing ef-fort into design of marine protected areas Incorporacioacuten delEsfuerzo de Pesca Recreativa en el Disentildeo de Aacutereas MarinasProtegidas Conservation Biology 20 1466ndash1476 httpsdoiorg101111j1523-1739200600509x

LynchTP(2008)Thedifferencebetweenspatialandtemporalvaria-tioninrecreationalfisheriesforplanningofmarineprotectedareasResponse toSteffeConservation Biology22486ndash491httpsdoiorg101111j1523-1739200800888x

LynchTP (2014)Adecadal time-seriesof recreational fishingeffortcollectedduringandafterimplementationofamultipleusemarinepark shows high inter-annual but low spatial variability Fisheries Research15185ndash90httpsdoiorg101016jfishres201309014

LynchTPAldermanRampHobdayAJ(2015)Ahigh-resolutionpan-orama camera system formonitoring colony-wide seabird nestingbehaviourMethods in Ecology and Evolution6491ndash499httpsdoiorg1011112041-210X12339

LynchTFlynnDWoodGKimJ-HParkJampHobdayAJ(2017)Metrics parameters and results from fisheries and wildlife studies with the CSIRO Ruggedized Autonomous Gigapixel Systems (CRAGS)DarwinNTAustralianMarineScienceAssociation

LynchTPWilkinsonEMellingLHamiltonRMacreadyAampFearyS(2004)ConflictandimpactsofdiversandanglersinamarineparkEnvironmental Management 33 196ndash211 httpsdoiorg101007s00267-003-3014-6

MaHOgawaTKSminkeyTRBreidtFJLesserVMOpsomerJDhellipVanVoorheesDA(2018)Pilotsurveystoimprovemonitoringofmarine recreational fisheries inHawaiʻiFisheries Research204197ndash208httpsdoiorg101016jfishres201802010

McCluskey S M amp Lewison R L (2008) Quantifying fish-ing effort A synthesis of current methods and their ap-plications Fish and Fisheries 9 188ndash200 httpsdoiorg101111j1467-2979200800283x

McGlennonDampKinlochMA(1997)Evaluationofthebus-routecreelsurveymethod in a largeAustralianmarine recreational fishery II

PilotsurveysandoptimalsamplingallocationFisheries Research3389ndash99httpsdoiorg101016S0165-7836(97)00067-2

McPheeDLeadbitterDampSkilleterGA(2002)SwallowingthebaitIs recreational fishing in Australia ecologically sustainable Pacific Conservation Biology840httpsdoiorg101071PC020040

Monz C A Pickering C M amp HadwenW L (2013) Recent ad-vances in recreation ecology and the implications of differentrelationships between recreation use and ecological impactsFrontiers in Ecology and the Environment11441ndash446httpsdoiorg101890120358

MooreAHallKGiriKTraceySPenroseLHansenShellipBrownP(2015)Developingrobustandcost-effectivemethodsforestimatingthenationalrecreationalcatchofSouthernBluefinTunainAustraliaFRDC Projectpp123

Morton A J amp Lyle J M (2004) Preliminary assessment of the rec-reational gamefish fishery in Tasmania with particular reference to Southern Bluefin Tuna Taroona TAS Tasmanian Aquaculture andFisheriesInstitute

ParnellPEDaytonPKFisherRALoarieCCampDarrowRD(2010)SpatialpatternsoffishingeffortoffSanDiegoImplicationsfor zonal management and ecosystem function Ecological Applications202203ndash2222httpsdoiorg10189009-15431

PattersonTAEvansKCarterTLampGunnJS(2008)Movementand behaviour of large southern bluefin tuna (Thunnus maccoyii)in the Australian region determined using pop-up satellite ar-chival tags Fisheries Oceanography 17 352ndash367 httpsdoiorg101111j1365-2419200800483x

PollockKH JonesCMampBrownTL (1994)Angler survey meth-ods and their applications in fisheries management Bethesda MDAmericanFisheriesSociety

PonsM Branch T AMelnychukM C JensenO P Brodziak JFromentinJMhellipHilbornR(2017)EffectsofbiologicaleconomicandmanagementfactorsontunaandbillfishstockstatusFish and Fisheries181ndash21httpsdoiorg101111faf12163

PowersSPampAnsonK (2016)Estimating recreationaleffort in theGulf of Mexico Red Snapper Fishery using boat ramp camerasReduction in federalseason lengthdoesnotproportionally reducecatchNorth American Journal of Fisheries Management 36 1156ndash1166httpsdoiorg1010800275594720161198284

Robson D amp Jones C M (1989) The theoretical basis of an ac-cess site angler survey design Biometrics 45 83ndash98 httpsdoiorg1023072532036

Rocklin D Levrel H Drogou M Herfaut J amp Veron G (2014)Combining telephone surveys and fishing catches self-reportTheFrenchseabass recreational fisheryassessmentPLoS ONE9e87271httpsdoiorg101371journalpone0087271

Smallwood C B Pollock K HWise B S Hall N G amp GaughanD J (2012) Expanding aerial-roving surveys to include counts ofshore-based recreational fishers from remotely operated camerasBenefitslimitationsandcosteffectivenessNorth American Journal of Fisheries Management321265ndash1276httpsdoiorg101080027559472012728181

Teixeira D ZischkeM T ampWebley J A C (2016) Investigatingbias in recreational fishing surveys Fishers listed in public tele-phone directories fish similarly to their unlisted counterpartsFisheries Research 181 127ndash136 httpsdoiorg101016jfishres201604012

van Poorten B T amp Brydle S (2018) Estimating fishing effortfrom remote traffic counters Opportunities and challengesFisheries Research 204 231ndash238 httpsdoiorg101016jfishres201802024

van Poorten B T Carruthers T RWardHGM ampVarkeyD A(2015)Imputingrecreationalanglingeffortfromtime-lapsecamerasusinganhierarchicalBayesianmodelFisheries Research172265ndash273httpsdoiorg101016jfishres201507032

12emsp |emsp emspensp FLYNN et aL

VenturelliPAHyderKampSkovC(2017)AnglerappsasasourceofrecreationalfisheriesdataOpportunitieschallengesandproposedstandardsFish and Fisheries18578ndash595httpsdoiorg101111faf12189

WiseBSTelferCFLaiEKMHallNGampJacksonG(2012)Long-termmonitoringofboat-basedrecreational fishing inSharkBayWesternAustraliaProvidingscientificadviceforsustainablemanagement in a World Heritage Area Marine and Freshwater Research631129ndash1141httpsdoiorg101071MF12054

WoodGLynchTPDevineCKellerKampFigueiraW(2016)High-resolutionphoto-mosaictime-seriesimageryformonitoringhumanuseofanartificialreefEcology and Evolution66963ndash6968httpsdoiorg101002ece32342

YuFRupertACCDiTizianaMNishikantaKRandySIllahNhellipJeffreyG(2011)TerapixelimagingofcosmologicalsimulationsThe Astrophysical Journal Supplement Series19718

Zischke M T Griffiths S P amp Tibbetts I R (2012) Catch and ef-fort from a specialised recreational pelagic sport fishery off

eastern Australia Fisheries Research 127ndash128 61ndash72 httpsdoiorg101016jfishres201204011

SUPPORTING INFORMATION

Additional supporting information may be found online in theSupportingInformationsectionattheendofthearticle

How to cite this articleFlynnDJHLynchTPBarrettNSWongLSCDevineCHughesDGigapixelbigdatamoviesprovidecost-effectiveseascapescaledirectmeasurementsofopen-accesscoastalhumanusesuchasrecreationalfisheriesEcol Evol 2018001ndash12 httpsdoiorg101002ece34301

Graphical AbstractThecontentsofthispagewillbeusedaspartofthegraphicalabstractofhtmlonly

Itwillnotbepublishedaspartofmainarticle

Wetrailedaremoteautonomousultra-high-resolutionphotosamplingmethodasacosteffectsolutionfordirectseascape-scaledmeasure-mentsOursequentialphotosamplingwasbatchedprocessesusingnovelTimeMachinedatahandlingsoftwaretoproduceldquobigdatardquotimeseriesmoviesfromaspatialsubsetofanoffshorefisherywhichwecomparedtoaregionalbus-routesurveyofboattrailersandinterviewswithparticipantsOurcamerasamplescloselytrackedlargerscaleregionaltrendsbutwithlaborcostsaquarterofthebus-routesurveycostsforanindividualsample25ofthebusroutes

Page 5: Gigapixel big data movies provide cost-effective seascape

emspensp emsp | emsp5FLYNN et aL

Thecountofboatsforeachsamplewasthenextrapolatedintoanestimateofdailyaverageboatingeffort(hoursplusmnSE)usingthein-stantaneouscountmethod(Equation1)

where ei is theextrapolatedboatinghours for the ithday Ii is theaveragenumberoftrailerboatsobservedacrosstheinstantaneoussamplesfordayiandTisthedailysamplingperiod(daylighthours)Asbadweathercanbeamajorfactoraffectingthedecisiontoboatorfish(ForbesTraceyampLyle2009)boatingeffortwasassumedtobe0whenadverseweathermadeobservationsofboatsimpossible

Next total effort for each day-type strata (Ejk) within sam-pling months (k) was calculated to estimate the monthly effort(Equation2)AsCRAGSallowedforcontinuoussamplingthroughoutthesurveyperiodthesamplingprobability(πk)wassetto1

Standarderrorandeffortvarianceformonthlyestimateswerecalculated(Equation3)accordingtoPollocketal(1994)forstrata

where nj is thenumberofsampleddaysofstrata j Njk is thetotalnumberofdaysofstratajinmonthk

Asvariationforeachstratawascalculatedseparatelyweusedasumofsquaredstandarderrorsapproach(Equation4)toestimatethetotalmonthlyvariation

Assampleswereobtainedinaserialandcontinuousfashionatimeseriesmodelusingrollingaverageswasalsoappliedtothedata(Diggle1995)(Equation5)

where smt is the smoothed average of extrapolated effort e persampletwhichisweightedacrossthreesampleseachseparatedby93minThisprovidedasmoothedaveragewithindays

Wealsoconductedabus-routestylesurveytoestimatefishingeffortfromtrailerboatsfortheregion(Pollocketal1994Robsonamp Jones1989)WhileCRAGSsampleswerecollected throughoutMayandJunethecomparablebus-routesurveywasonlyconductedinMaydue to logistical constraintsWesurveyed the threemajorboatrampsintheregionatPiratesBayFortescueBayandStewartsBaybetween3May2015and30May2015(Figure1)

Atotalof20bus-routesampleswerescheduledWeundertookrandomstratifiedsamplingwithdaytypedividedintoweekdaysorweekendpublic holidays To reflect expected higher recreationaleffortduringweekendsandpublicholidays(McCluskeyampLewison2008)thesamplingprobability(πk)wasweightedtowardthisstrata

over weekdays at 12 to eight samples (Supporting InformationAppendix S2) We stratified the sample time into morning (am0600ndash1159)andafternoon(pm1200ndash1800)withequalproba-bilitiesofsamplingTominimizetemporalautocorrelationthestart-ing location and travel direction of the routewere also randomlyselectedwithoutreplacementEqualsamplingweightwasprovidedtoeachrampduetolackofpriorknowledgeofuserates

At each boat ramp surveys were conducted by first countingparkedboattrailersthenbymonitoringvesselentryandexitsduringa1-hrsamplingperiod(KinlochMcGlennonNicollampPike1997Pollocketal1994)The1-hrperiodwaschosendueto travel timeanddis-tancebetweensitesallowingforabus-routesampletobecompletedforallrampswithinthetemporalstrata(AMorPM)ofthedailysam-plingframeInterviewswereconductedwithallpartieslaunchingandretrievingvesselsduringtheclerkrsquoswaittimedeterminingpartysizetargetspeciesgroupandactivity(iefishingornot)forestimatesoffishing effort This well-established procedure produced a monthlyextrapolatedestimationofboateffort instandardizedunitsofefforthoursplusmnSEwhichcouldthenbeadjustedtofishingeffortviathepro-portionofintervieweesactivities(Pollocketal1994RobsonampJones1989)(SupportingInformationAppendixS3)PartysizebetweenrampswastestedforanydifferenceusingaonefactorANOVA

Thebus-route surveyhenceprovidesanestimationofall fish-ing effort from trailer boats around this isolated peninsular TheCRAGSdataarethusahigh-frequencyspatialsubsetoftheentiretrailer boat fishing effort for the period thatwas extrapolated bytheregionalsurveyThisallowedustobothtesttherelativeuseoftheHRMCAcomparedtowiderpeninsularuseandtheabilityofourremoteobservationswithourCRAGsunitofanactualsubsetofthefisherytotracklargertemporalpatternsofpeninsularwideeffort

TovalidatetherepresentativenessoftheCRAGShigh-frequencytrackingofeffortmetricsover timeweundertookaSpearmanrankcorrelationbetweentime-matchedeffortestimationsbyCRAGSandthe regional bus-route survey For graphing bus-routeextrapolationestimates were overlaid onto extrapolations from daily averages ofboatfishingeffortfromCRAGSWealsocarefullyloggedalltimespentontheprojecttocomparetotallaborcostsbetweenCRAGSandthebusrouteforbothdatacollectionandprocessingtodetermineeffortacrossthecomparativemonthaswellasforcostpersample

3emsp |emspRESULTS

The CRAGS operated successfully and continuously throughoutthestudycapturing545plusmn198SE(May)and483plusmn088SE(June)panoramasperdaymakinga totalof314 samplesWemade895observationsofboatsbut51samplesacross21daysreturnedzerocountsofboats(17oftotalsamples)Thesewereconcentratedonweekdaysmdashwith 44 zero-count samples across 18daysmdashwhile onweekendsandpublicholidaytherewereonlysevensamplesacross3dayswithnoactivityInclementweatherrendered27samplesun-usableTheimagesandbatch-processedTimeMachinemoviescon-sumed4274GBofmemory

(1)ei=IitimesT

(2)Ejk=

nsum

i=1

(

ei

120587k

)

(3)SEjk=

1

njminus1

sumn

i=1(eiminus ei)

2

njtimes (Njk)

2

(4)SEk=

radic

SE21k+SE2

2khellip+SE2

jk

(5)smt=etminus1+ et+ et+1

3

6emsp |emsp emspensp FLYNN et aL

Based on boat ramp interviews the average trailer boat partysize (P) was 287 people per vessel throughout May (plusmn0133 SEn=84)andpartysizebetweensitesshowednosignificantdiffer-ence (df=2F = 302p = 068) The averagemonthly boat fishingeffort(hours)extrapolatedforourCRAGSobservationsofHRMCAwas5629hr(plusmn711SE)with4096hr(plusmn607SE)inMayand1534hr(plusmn370 SE) in June (Figure4)mdashthis could be extrapolated to fisherhoursbymultiplyingbytheaveragepartysizebutasnointerviewswereconductedinJuneweleftboththeCRAGSandbusrouteex-trapolationsasboatfishingeffortonly

The total subset area observed by CRAGSwas 96km2 which comparedto318km2thatwaseasilyavailabletotrailerboatsaroundtheTasmanPeninsularwhichcorrespondstoaround3ofthetotalareaComparedtothetotalfishingeffortforMayextrapolatedfromtheregionalbus-routesurvey therelativelysmallareaaroundtheHRM(~3)receivedalargeamountofthefishingeffort(~23)

ThehighfrequencyofCRAGSsamplingpermittedtheexamina-tionoffine-scaleinter-dailytemporalvariationwhichdisplayedarag-gedsinusoidoscillation(Figure5)Effortwasgenerallylowwithinthe

weekdaystrataandthen increasedrapidly to largepeaksonweek-ends (Figure5) One exception to this was between the 18th and22ndMaywhichhadmoreeffortonweekdays thanotherperiods(Figure5)Bracketingthisperiodweretwoweekendsoffishingcom-petitions theldquoTunaClubofTasmaniarsquosrsquoFarSouthClassicrdquo (16ndash17thMay)andRally4Northerninvitationalweekendrally(23rdMay)

A total of 16 bus-route samples were conducted logistic fail-uresledtotheabandonmentoffour(20)oftheplannedsamplesRegionalTrailerboateffortfromaroundtheTasmanPeninsularwasestimatedtobe19650hr (plusmn2858SE) inMay (Figure4)Notrailersat individual rampswere recorded four timeswith all of theseoc-curringonweekdaysThetotalnumberoftrailerssightedperramprangedfrom0to16duringweekdaysand1ndash54forweekendswiththehighestnumberrecordedon16May2015atPiratesBayDuringhigh-intensitydays(particularlyweekendsholidaysmornings)therewereoftenlargeoverflowsfromrampcarparkswithparkedcarsandboattrailersextendingasmuchasonekilometerfromtheboatramp

During thebus-route survey 84 interviewswere conductedThemajorityofmariners(905n=96)indicatedfishingwastheir

F IGURE 4emspExtrapolatedmonthlytrailerboateffort(hoursplusmnSE)fortheentireTasmanPeninsulaassessedbyabus-routesurvey(May)andforthesubsetoftheareaaroundtheHRMCAsampledwithCRAGS(MayandJune)

F IGURE 5emspHigh-frequencyinterdailyextrapolatedCRAGSestimationsofboatefforthoursaroundthe(HRMCA)

0

100

200

300

400

500

600

Rolli

ng m

ean

daily

effo

rt in

hou

rs

DayWeekdays Weekends + Public Holidays

lyn088
Inserted Text
CA
lyn088
Cross-Out
lyn088
Cross-Out
lyn088
Inserted Text
t

emspensp emsp | emsp7FLYNN et aL

primaryactivity (Table1)Overall643 (n=54) indicated theirprimary target were tuna with 179 (n=15) specifically tar-geting southern bluefin tuna (Thunnus maccoyii)Other commonrecreationaltargetsincludedflathead(Platycephalusspp595)Southernrocklobster(Jasus edwardsii107)andsquidcalamari(Sepioteuthis australisandNototodarus gouldi476)(Table1)

Basedon the responsesofactivity type from interviews totalboatfishingeffortinMaywasestimatedtobe17783hr(plusmn2586SE)Asmostnonfishingactivitieswerenear shoreweassumed trailerboatingactivityatHMRCAobservedbyCRAGSwasallfishing

Therewere16directcomparisonsamplesbetweenthebusrouteandCRAGSmethodsbetweenthe2ndandthe29thofMaySamplesconsisted of 7 weekdays and 9 weekendspublic holiday samples

(Figure6)Infourcaseswheretwobus-routesamplesweretakenonthe same day (Supporting Information Appendix S2) they are com-bined toadailyestimation for tabulation (Table2)Theproportionaldifferencebetweenestimatedeffortsusingthetwomethodsrangedfrom078-to418-folddifferencehoweverranksestimatesbetweenCRAGS and bus-route samples were strongly associated (n=16ρ(12)=087p=lt001)Interdailysamplingstrata(ampm)werealsoexaminedseparatelyforcorrelationwhichremainedstrongforbotham(n=9ρ(7)=080p=lt001)andpm(n=7ρ(5)=089p=lt001)

ForthecomparativemonthCRAGStook169samplescomparedtothebusroutesrsquo16(Table3)WehadtwofieldtripdaysforCRAGSwith four people on the setup and three on the breakdown daywhichresultedinsevenfull-timeequivalent(FTE)daysofworkThiscomparedto16fielddaysforthebus-routemethodwhichincludedtraveltothefieldsiteforbetweentwoandthreestaffwhichaccu-mulatedto40FTEdaysofworkSettingupourCRAGSautomatedstitching took one FTE day (though this is faster for experiencedofficers)whiledataextractionperimagewasestimatedtotakeon

TABLE 1emsp Intervieweesreportedprimarytargettaxaoractivity

Primary target taxaspeciesactivity Interview responses

Tuna(unspecified) 37 440

SouthernBluefinTuna 15 179

AlbacoreTuna 2 238

AllTuna 54 643

Flathead 5 595

Southernrocklobster 9 107

Squidcalamari 4 476

Abalone 1 119

SmallPelagics 1 119

SnottyTrevalla 2 238

Allnearshorespecies 22 262

Surfing 1 119

SCUBA 3 357

DemonCavevisitors 3 357

PleasureCruise 1 119

Allnonextractive 8 952

Note ldquoTunardquo responsesdenote anglers targeting all specieswithin thefamilyThunnini

F IGURE 6emspDailyboatingeffort(hours)extrapolatedusingbus-routeaccesspointsurvey(bar)andCRAGS(line)Effortestimatefrombus-routesurveywascalculatedfrommorning(amdarkgray)andafternoon(pmlightgray)surveyCRAGSsurveywasconductedbetweenMayandJunewhilethebus-routesurveywasonlyconductedinMay

TABLE 2emspRankingofeffortforpairedsamplesbetweenCRAGSandthebusroute

DateCRAGS hours (rank)

Bus- route hours (rank) Δ Rank

2052015 330(9) 1208(10) minus1

3052015 303(8) 1471(11) minus3

12052015 0(1) 48(2) minus1

14052015 0(1) 24(1) 0

16052015 387(11) 1794(12) minus1

19052015 93(4) 89(4) 0

20052015 57(3) 117(5) minus2

22052015 228(7) 283(6) 1

23052015 513(12) 414(8) 4

24052015 345(10) 1155(9) 1

25052015 122(6) 323(7) minus1

29052015 102(5) 79(3) 1

8emsp |emsp emspensp FLYNN et aL

average75mincomparedto1minforenteringthebus-routedatasheetWhiledataextraction forCRAGS took longer compared tothe bus routemdashat 381 FTE daysmdashtotal monthly labor for CRAGStookonlyaquarterofthetimeandonaper-samplecomparisonwasonly25ofthebus-routelaborcost

4emsp |emspDISCUSSION

Over our autumn samplingmost trailer boat owners accessing thewaters around theTasmanPeninsula via the regional boating infra-structurewererecreationallyfishingwithfisherspredominatelytar-getingtunaTrailerboatrecreationalfishersconstitutealargetrancheof coastal users in Australia (McPhee etal 2002) and our resultssuggestedthatthisregionremainsonewhererecreationalfishing isconcentrated(MortonampLyle2004)EffortobservedbyCRAGSwasa spatial subsetof thegeneral areaeasily accessedby trailer boatsaroundtheTasmanPeninsularwhichweregionallyassessedwithourbus-routerovingsurveyAsCRAGSonlyobserved~3of thetotalarea but corresponded to ~23 of themonthly bus-route boat ef-fort theHippolyteRocksMarineConservationArea (HRMCA) is afurtherconcentrationpointwithintheregionforrecreationalfishingAsthisoffshorefishingeffortaroundtheHRMCAwasmostprobablyfocusedontotunamdashwhichaccountedfor634ofallinterviewedfish-ers activitiesmdashthe actual proportionofpeninsularwide tuna fishingthatwasobservedbyCRAGSwaspotentiallymuchhigherthan23

For developing effort metrics finding a representative site toobservemaybe important forprecise trackingof trendsThetightrankcorrelationbetweenpairedregionalbus-routeandCRAGScam-era samples suggest that effort atHRMCAwas representative ofregionaltemporaltrendsOurresultsarealsosimilartootherstud-ieswheresamplingmethodsforrecreationalfisheriesoverbroaderscaleswhencomparedtoindextrendsofeffortcollectedfrompointsourcesshowstrongcorrelations(Hartilletal2016)

CRAGSwithitsseascapescaleofdatacaptureobservesactualeffortonfishinggroundsratherthan indirectmetricsfrommove-ment past access points (Hartill etal 2016 Smallwood PollockWise Hall amp Gaughan 2012 van Poorten amp Brydle 2018) This

makesthechoiceofaccesspointstomonitorimmaterialfortrendcol-lectionandmaybeofparticularinterestforpelagicfisheriesWhilenonintuitiverecreationalfishingforwide-rangingmigratorypelagicspeciesisoftenspatiallyconcentratedintosmallareas(Lynch2006)duetotightoverlapsbetweeneaseofaccessbyfishersandfishbe-haviorrelatedtobathymetrymigrationhabitatandpreyavailability(Lynchetal2004PattersonEvansCarterampGunn2008)

Unlikespatialpatternstheintensityofcoastalrecreationaltemporaleffortcanbehighlyvariableovertime(Lynch20082014WiseTelferLaiHallampJackson2012)thoughitisgenerallythoughttofollowpre-dictablepatternsrelativetoholidayandnonholidayperiods(JonesampPollock2012)Ourresultsdemonstratedtheselargefluctuationsrang-ingfromzerouptogt400hroftrailerboateffortforadjacentdaysIncombinationwithholidayperiodsothersourcesofvariabilitymayalsooccurduetoindividualorcombinationsoffactorssuchasfishingcluborcompetitionactivitiesandweatherconditionsHigh-frequencysam-plingcanovercomethesecommontemporaldisjunctionsinobservedfishingeffortwherebychanceduetolownumbersofsampleslogis-ticallypermissiblewithtraditionalrovingoraccesspointsurveyslargeerrorsintheestimationofeffortcanbeintroduced(Hartilletal2016vanPoortenampBrydle2018vanPoortenetal2015)

With multiple daily observations abnormal episodes are morelikelytobeobservedForinstanceaweek-longperiodofintenseef-fortwasrecordedbetween15May2015and27May2015whichwasbracketedbytwoconsecutiveweekendswherefishingcompetitionswereheldontheTasmanPeninsularThenormalsinusoidpatternoflowweekdayandhighweekendeffortwasmuchlesspronouncedastheweekdaylullsineffortwerepartiallybridgedbyanglerscontinuingtofishAsbus-routedesignsusedaytypeasstratawithlowerprob-abilitiesforsamplingweekdaysandmuchfewersamplesoverallthistypeofeffectcouldeasilybemissedWhileourbus-routesamplesdidoverlapwiththetunafishingcompetitionsthiswaspurelybychanceSucheventswhichresultinlargespikesofeffortcanhavemarkedim-pactsonlocalecosystemsandharvestestimates(McPheeetal2002MonzPickeringampHadwen2013)andwithoutpriorknowledgearemorelikelytobecapturedwithmonitoringathigherfrequencies

Methodsforgatheringdataoncoastalusesuchasrecreationalfishing moved away from direct approaches and toward off-sitemethodssuchastelephoneinterviews(McCluskeyampLewison2008Pollock etal 1994) due to unsustainable costs from staffing andoperationsparticularlyduringperiodsof intenseusesuchasearlymornings weekends and public holidays which require overtimepaymentsHoweversamplingissueswithoff-sitemethodsareofpar-ticularconcernforopen-accessactivitiessuchasunlicensedfisheriesasthereisnolicensedatabaseofcontactphonenumbersonwhichtobaseasampleframeThisisparticularlysofornicherecreationalfisheriessuchaspelagicgamefishingasthefishersbecomeincreas-inglydilutewithinthegeneralpopulationandhencearehardtoac-cessparticularlyviaoff-sitemethodssuchastelephone interviews(Griffithsetal2010)

Remoteandautonomousdeploymentofsensorsmaythereforeprovideanalternativeandcost-effectivemethodforlong-termmon-itoring particularly for effort across a range of applicationsOur

TABLE 3emspFulltimeequivalent(FTE)labordaysforCRAGSandbusroutesurveystoestimatepatternsofeffort

CRAGS Bus route

Numberofsamples 169 16

Fielddays 2 16

Averagepeopleperfieldday 35 25

Field FTE days subtotals 7 40

StitchingsetupFTE 1 0

DataextractionFTE(persample) 0017 0002

Subtotal data extraction FTE 381 004

FTEdayspersample 006 250

Total FTE days 108 400

emspensp emsp | emsp9FLYNN et aL

comparisonsofmatchedpairsofdatafromtwomethodssuggestedthat theCRAGSoutputswerewell correlatedwith regionalefforttrendsOuranalysisoflabordemonstratesthattryingtounderstandthistemporalvariabilitywithsamplingviaboatrampsurveyswouldbeprohibitivelyexpensiveOurCRAGSallowsforatemporaleffortmetric derived fromactual observationsof fishing to be cheaplydeveloped with sampling frequency at subdaily scales This high-frequencysamplingapproachwasvalidatedagainstour largerbutmuchmorecostlybus-routesurveyOuranalysisoflaborrequiredtobothcollectandprocessdatashowedthatCRAGSrequiredconsid-erablylessresourcestocollectapproximatelyninetimesmorefre-quenttemporaldatathanthebus-routemethodMajorlaborsavingswereduetotheruggedizedandautonomouscapabilityofCRAGSasonlytwofieldtripswererequired(deploymentandretrievaltook7FTEdays)comparedtothebus-routersquos16fieldtrips inamonthwhichconsumed40FTEdaysandincludedanadditional780kmofinter-ramproadtravel(SupportingInformationAppendixS3)AsitisweatherproofautonomousandabletobeprogrammedwithatimedelaysetupCRAGSldquofireandforgetrdquonaturebothremovestheneedforobservertobephysicallylocatedattheregiontocountactivity(EdwardsampSchindler2017)andworkduringovertimeperiodsTheldquoblankingperiodrdquowhichstopsandstartssamplingmdashinthiscaseover-nightmdashalso reduces battery and memory use In combined thesetechnological advances have allowed long-term CRAGS deploy-mentsacross4yearsof4ndash6monthrotationsontooffshoreislandsand for several seasons in Antarctica to monitor seabirds (Lynchetal2017)Thelaborestimateforthebusroutewasconservativeasitdidnotincludeovertimeloadingforthe6weekenddaysofsam-plingorearlymorningcommencementofamsamples

Theadditional381FTEprocessingtimerequiredbythecameraapproach did not impact greatly on the overall labor budgetwhenconsideredagainstfieldworkTheuseoftheTimeMachinebatchpro-cessingsoftwaretoautomaticallyproducetheldquobigdatardquointeractivemoviewasalsoatechnologicalinnovationthatsignificantlyreducedthelaborofdatahandlingcomparedtothemanualopeningofindivid-ualGigaPansusedinpreviousstudies(Lynchetal2015Woodetal2016)IncomparativetermsattheindividualsampleleveltotracktheeffortmetriceachCRAGSsamplerequired24ofthelaborofthebus-routesurvey

Thedevelopmentandapplicationofimagerecognitionsoftware(machine learning) to detect classify and count boats automati-cally (Bousetouane amp Morris 2015) would further reduce post-processingcostsTherequiredfrequencyofsamplingshouldalsobeconsideredonacase-by-casebasisdependentontheresearchquestion as image collection can be somewhat open-ended andresult inunwieldydata sets that require subsampling tobe cost-effective(Parnelletal2010)Inourcasethesettingof~5imagesperdayappearedtoadequatelydescribetheinterdailyvariation

Whileouroneautonomouscameraprovidedastrongcorrelationwithtemporaltrendsineffortcomparedtotheregionalsurveyitdidnotprovide relativespatialdistributionsofeffortacross the fisheryThesedistributionshowevercanbepredictableandhighlyconcen-trated(Lynch20062014Parnelletal2010)anobservationthatwas

alsoshownbyourresultsandmuchsmallerlevelsofsamplingeffortwouldprobablyberequiredtoresolvevariationinspatialdistributionscomparedtowhatisrequiredtoresolvethehightemporalvariationsineffortUnderstandingfine-scalespatialdistributionsofrecreationalfishingeffortiscommonlyachievedusingaerialcounton-watersur-veysorvantagepointsurveys(Lynch2006Smallwoodetal2012)thoughotherremotemethodssuchashigh-resolutionsatelliteimag-ery(FretwellScofieldampPhillips2017)mayprovideacheapersolu-tionThesetypesofwidersurveyshavealsobeensuggestedashighlycompatiblewithpointsourcemetricssuchasremotecamerasystemstoallowforcalibrationorldquoup-ratingrdquoforeffortexpansionstoestimateoveralleffortwithinthefishery(Hartilletal2016)

Thelongdistance(~6km)betweentheHMRCAandthedeploy-ment location stretched the resolution limit of the current CRAGSsetup aswewere unable to distinguish party size or activity typepurelyusing thecamerasystemThis isunlikepreviousCRAGSde-ploymentwhichhadahighsuccessrateofidentifyingpartysizeac-tivity typeata shorter rangeofapproximately19km (Woodetal2016) Thus ground-truthing information about the fisherywas re-quiredforthesystemtoworkeffectivelyAutonomouscamerasystemalsodonotcollectcreel(catch)databutbyallowingforcost-effectivecollectionofthehighlyvariabletemporaleffortdatatheymayallowforhumanresourcestobespentelsewhereduringsurveyperiods(egtoundertakeboatrampinterviews)(vanPoortenampBrydle2018)

Our studywas a small-scale trial of new technology andwhilepreliminary results are encouraging validated seasonal effort esti-mationsobservedviaCRAGSwereoutsidethescopeofthisprojectOur comparisons would hence have been improved by extendingCRAGSdeploymentsandwithmorecomparisonstoregionalsurveysSimilarly usingmultiple CRAGS simultaneously would have bettertested for differences in subregional patterns of temporal trendsorconcentrationof infishingeffortfor instanceboatsaroundthenearbyTasmanIsland

Insummarytheprogrammablesamplinganddataextractionof-feredbyhigh-resolutionautonomouscamerasystemscanallowforfine-scalelong-termmonitoringoffishingefforttrendsorpotentiallymanyothertypesofactivitiesatseascapescalesMorewidespreadadoptionofhigh-frequencysamplingapproachesforeffortestima-tionwouldalsohelpavoidcollectionofmisrepresentativedatabyidentifyingeffortanomalies(McCluskeyampLewison2008)suchastheweekdaycontinuationeffectofraisedeffortbetweenadjacentweekendfishingcompetitionByimprovingtheinformationdensityovertraditionalsurveymethodsautonomouscamerasystemsandotherremotesystemsmayprovideanewwaveofoptimizationfordirectmeasurementofcoastalhumanusesuchasrecreationalfish-eriesalthoughcomplimentarymethodsarestillrequiredtodeter-minespatialdistributionstargetspeciesandharvest

ACKNOWLEDG MENTS

We thankMsGeorginaWoodMs SaraAdkinsMsBelleMonkMrRSherringtonandMsPaigeKellyforconstructiveinsighten-thusiasmandassistanceduringfieldworkWealsothankMrPhillip

10emsp |emsp emspensp FLYNN et aL

CulversonMrStuartDudgeonandMrMattStapenellKarenEvensprovidedhelpfulcommentsonaninternaldraftGreatestthankstothemany recreational anglerswhowillingly shared information tosurveyclerkswithouttheirinvolvementthisprojectwouldnothavecometofruitionThreeanonymousreviewersprovidedcommentsthatimprovedthemanuscript

CONFLIC T OF INTERE S TS

Theauthorsdeclarenoconflictofinterests

AUTHOR CONTRIBUTIONS

MrFlynnhelpeddesignthestudyledthefieldworkundertookmostoftheanalysisandhelpwritethemanuscriptDrLynchhelpdesignthestudyassisted inthefieldworkprovidedadviceontheanaly-sisanddidsomeanalysisconstructedfiguresandafterMrFlynncontributedthemosttothedraftingofthemanuscriptDrBarretthelpeddesign the study assisted in some fieldwork provided ad-viceonanalysisandeditedthemanuscriptMrWonghelpedexten-sivelywithfieldworkassistedwithanalysisconstructedfiguresandhelpedwritethemanuscriptMsDevineassistedwithfieldworkandconfigurationoftheequipmentincludingtheapplicationofthebigdatamoviesoftwareandhandlingofphotographsandalsohelpedwritethemanuscriptMrHughesdevelopedtheelectronicsfortheequipmentoversaw thebuildofgearwrote the software topro-gramtherobotandreviewedandeditedthemanuscript

DATA ACCE SSIBILIT Y

Survey sampling schedules data collection sheets extrapolationproceduresandtechnicalCRAGSspecificationsarealluploadedasonlinesupportinginformationappendixRawdatafromsurveysandCRAGSDOIhttpdoiorg105061dryadv10d218

RE SE ARCH E THIC S COMPLIANCE AND FUNDERS

ResearchabidesbytheAustraliancodeofhumanresearchandex-perimentation(NationalStatementonEthicalConductinHumanResearch of 2007) Permission to conduct research on humanparticipants was granted by the University of Tasmania SocialScience under reference H0014772mdashldquoRemote investigation ofcoastal area user group quantification and variationrdquo aswell asfromtheCSIROSocialScienceHumanResearchEthicsCommitteeunder reference 08115mdashldquoHawkesbury Region Project HumanUsePatternsrdquoFundingwasbyUTASasPartoftheIMASHonoursProject funding and through internal support from the CoastsProgram

ORCID

Tim P Lynch httporcidorg0000-0002-5346-8454

R E FE R E N C E S

Alessa L Kliskey A amp Brown G (2008) Socialndashecological hotspotsmapping A spatial approach for identifying coupled socialndasheco-logicalspaceLandscape and Urban Planning8527ndash39httpsdoiorg101016jlandurbplan200709007

ArlinghausR(2006)Overcominghumanobstaclestoconservationofrecreational fishery resources with emphasis on central EuropeEnvironmental Conservation 33 46ndash59 httpsdoiorg101017S0376892906002700

Arlinghaus R Tillner R amp Bork M (2015) Explaining participationratesinrecreationalfishingacrossindustrialisedcountriesFisheries Management and Ecology 22 45ndash55 httpsdoiorg101111fme12075

BadcockPBPatrickKSmithAMASimpsonJMPennayDRisselCEhellipRichtersJ(2016)Differencesbetweenlandlineandmobilephoneusers in sexualbehavior researchArchives of Sexual Behavior46(6)1711ndash1721

Blumberg S J amp Luke J V (2009) Reevaluating the need for con-cern regarding noncoverage bias in landline surveys American Journal of Public Health 99 1806ndash1810 httpsdoiorg102105AJPH2008152835

Bousetouane F ampMorris B (2015) Off-the-shelf CNN features forfine-grainedclassificationofvesselsinamaritimeenvironmentInGBebisRBoyleBParvinDKoracinIPavlidisRFerisTMcGrawMElendtRKopperERaganZYeampGWeber(Eds)Advances in visual computing 11th International symposium ISVC 2015 Las Vegas NV USA December 14ndash16 2015 proceedings Part II (pp379ndash388)Cham Switzerland Springer International Publishing httpsdoiorg101007978-3-319-27863-6

Cooke S J amp Cowx I G (2004) The role of recreational fish-ing in global fish crises BioScience 54 857ndash859 httpsdoiorg1016410006-3568(2004)054[0857TRORFI]20CO2

DiggleP J (1995)Time-series a biostatistical introductionOxfordUKClarendonPress

EdwardsJampSchindlerE (2017)Avideomonitoringsystemtoeval-uate ocean recreational fishing effort in Astoria OregonMarine recreational information program (pp 31) Newport OR OregonDepartmentofFishandWildlife

FarrMStoecklNampSuttonS(2014)RecreationalfishingandboatingArethedeterminantsthesameMarine Policy47126ndash137httpsdoiorg101016jmarpol201402014

ForbesETraceySampLyle JM (2009)Assessment of the 2008 rec-reational gamefish fishery of Southeast Tasmania with particular refer-ence to Southern Bluefin TunaHobartTASTasmanianAquacultureandFisheriesInstituteUniversityofTasmania

FretwellPTScofieldPampPhillipsRA(2017)Usingsuper-highres-olutionsatelliteimagerytocensusthreatenedalbatrossesIbis159481ndash490httpsdoiorg101111ibi12482

GiriKampHallK (2015)SouthAustralian recreational fishingsurveyFisheries Victoria internal report series (pp 63) Queenscliff VicFisheriesVictoria

Griffiths S P Pollock K H Lyle J M Pepperell J G TonksM L amp Sawynok W (2010) Following the chain to elu-sive anglers Fish and Fisheries 11 220ndash228 httpsdoiorg101111j1467-2979201000354x

Halpern B SWalbridge S Selkoe K A Kappel C V Micheli FDrsquoAgrosaChellipWatsonR(2008)AglobalmapofhumanimpactonmarineecosystemsScience319948ndash952httpsdoiorg101126science1149345

Hartill BW Payne GW Rush N amp Bian R (2016) Bridging thetemporal gap Continuous and cost-effective monitoring of dy-namic recreational fisheries by web cameras and creel surveysFisheries Research 183 488ndash497 httpsdoiorg101016jfishres201606002

emspensp emsp | emsp11FLYNN et aL

HenryGWampLyleJM(2003)NationalrecreationalandindigenousfishingsurveyFinal report to the fisheries research amp development cor-poration and the fisheries action program (pp 188) Canberra ACTFRDC

JonesCMampPollockKH(2012)Recreationalanglersurveymeth-odsEstimationofeffortharvestandreleasedcatchInAZaleDLParrishampTMSutton(Eds)Fisheries techniquesBethesdaMDAmericanFisheriesSociety

Kearney R E (2002) Co-management The resolution of conflict be-tween commercial and recreational fishers in Victoria AustraliaOcean amp Coastal Management45201ndash214httpsdoiorg101016S0964-5691(02)00055-8

KellerKSteffeASLowryMMurphyJJampSuthersIM(2016)Monitoringboat-basedrecreationalfishingeffortatanearshorearti-ficialreefwithashore-basedcameraFisheries Research18184ndash92httpsdoiorg101016jfishres201603025

KinlochMAMcGlennonDNicollGampPikePG(1997)Evaluationof the bus-route creel surveymethod in a largeAustralianmarinerecreationalfisheryISurveydesignFisheries Research33101ndash121httpsdoiorg101016S0165-7836(97)00068-4

LewinWCArlinghausRampMehnerT(2006)Documentedandpo-tential biological impacts of recreational fishing Insights forman-agementandconservationReviews in Fisheries Science14305ndash367httpsdoiorg10108010641260600886455

LowryMampMurphy J (2003)Monitoring the recreational gamefishfisheryoffsouth-easternAustraliaMarine and Freshwater Research54425ndash434httpsdoiorg101071MF01269

LyleJMStarkKEampTraceySR(2014)2012ndash2013 Survey of rec-reational fishing in TasmaniaHobart TAS Institute forMarine andAntarcticStudies

Lynch T P (2006) Incorporation of recreational fishing ef-fort into design of marine protected areas Incorporacioacuten delEsfuerzo de Pesca Recreativa en el Disentildeo de Aacutereas MarinasProtegidas Conservation Biology 20 1466ndash1476 httpsdoiorg101111j1523-1739200600509x

LynchTP(2008)Thedifferencebetweenspatialandtemporalvaria-tioninrecreationalfisheriesforplanningofmarineprotectedareasResponse toSteffeConservation Biology22486ndash491httpsdoiorg101111j1523-1739200800888x

LynchTP (2014)Adecadal time-seriesof recreational fishingeffortcollectedduringandafterimplementationofamultipleusemarinepark shows high inter-annual but low spatial variability Fisheries Research15185ndash90httpsdoiorg101016jfishres201309014

LynchTPAldermanRampHobdayAJ(2015)Ahigh-resolutionpan-orama camera system formonitoring colony-wide seabird nestingbehaviourMethods in Ecology and Evolution6491ndash499httpsdoiorg1011112041-210X12339

LynchTFlynnDWoodGKimJ-HParkJampHobdayAJ(2017)Metrics parameters and results from fisheries and wildlife studies with the CSIRO Ruggedized Autonomous Gigapixel Systems (CRAGS)DarwinNTAustralianMarineScienceAssociation

LynchTPWilkinsonEMellingLHamiltonRMacreadyAampFearyS(2004)ConflictandimpactsofdiversandanglersinamarineparkEnvironmental Management 33 196ndash211 httpsdoiorg101007s00267-003-3014-6

MaHOgawaTKSminkeyTRBreidtFJLesserVMOpsomerJDhellipVanVoorheesDA(2018)Pilotsurveystoimprovemonitoringofmarine recreational fisheries inHawaiʻiFisheries Research204197ndash208httpsdoiorg101016jfishres201802010

McCluskey S M amp Lewison R L (2008) Quantifying fish-ing effort A synthesis of current methods and their ap-plications Fish and Fisheries 9 188ndash200 httpsdoiorg101111j1467-2979200800283x

McGlennonDampKinlochMA(1997)Evaluationofthebus-routecreelsurveymethod in a largeAustralianmarine recreational fishery II

PilotsurveysandoptimalsamplingallocationFisheries Research3389ndash99httpsdoiorg101016S0165-7836(97)00067-2

McPheeDLeadbitterDampSkilleterGA(2002)SwallowingthebaitIs recreational fishing in Australia ecologically sustainable Pacific Conservation Biology840httpsdoiorg101071PC020040

Monz C A Pickering C M amp HadwenW L (2013) Recent ad-vances in recreation ecology and the implications of differentrelationships between recreation use and ecological impactsFrontiers in Ecology and the Environment11441ndash446httpsdoiorg101890120358

MooreAHallKGiriKTraceySPenroseLHansenShellipBrownP(2015)Developingrobustandcost-effectivemethodsforestimatingthenationalrecreationalcatchofSouthernBluefinTunainAustraliaFRDC Projectpp123

Morton A J amp Lyle J M (2004) Preliminary assessment of the rec-reational gamefish fishery in Tasmania with particular reference to Southern Bluefin Tuna Taroona TAS Tasmanian Aquaculture andFisheriesInstitute

ParnellPEDaytonPKFisherRALoarieCCampDarrowRD(2010)SpatialpatternsoffishingeffortoffSanDiegoImplicationsfor zonal management and ecosystem function Ecological Applications202203ndash2222httpsdoiorg10189009-15431

PattersonTAEvansKCarterTLampGunnJS(2008)Movementand behaviour of large southern bluefin tuna (Thunnus maccoyii)in the Australian region determined using pop-up satellite ar-chival tags Fisheries Oceanography 17 352ndash367 httpsdoiorg101111j1365-2419200800483x

PollockKH JonesCMampBrownTL (1994)Angler survey meth-ods and their applications in fisheries management Bethesda MDAmericanFisheriesSociety

PonsM Branch T AMelnychukM C JensenO P Brodziak JFromentinJMhellipHilbornR(2017)EffectsofbiologicaleconomicandmanagementfactorsontunaandbillfishstockstatusFish and Fisheries181ndash21httpsdoiorg101111faf12163

PowersSPampAnsonK (2016)Estimating recreationaleffort in theGulf of Mexico Red Snapper Fishery using boat ramp camerasReduction in federalseason lengthdoesnotproportionally reducecatchNorth American Journal of Fisheries Management 36 1156ndash1166httpsdoiorg1010800275594720161198284

Robson D amp Jones C M (1989) The theoretical basis of an ac-cess site angler survey design Biometrics 45 83ndash98 httpsdoiorg1023072532036

Rocklin D Levrel H Drogou M Herfaut J amp Veron G (2014)Combining telephone surveys and fishing catches self-reportTheFrenchseabass recreational fisheryassessmentPLoS ONE9e87271httpsdoiorg101371journalpone0087271

Smallwood C B Pollock K HWise B S Hall N G amp GaughanD J (2012) Expanding aerial-roving surveys to include counts ofshore-based recreational fishers from remotely operated camerasBenefitslimitationsandcosteffectivenessNorth American Journal of Fisheries Management321265ndash1276httpsdoiorg101080027559472012728181

Teixeira D ZischkeM T ampWebley J A C (2016) Investigatingbias in recreational fishing surveys Fishers listed in public tele-phone directories fish similarly to their unlisted counterpartsFisheries Research 181 127ndash136 httpsdoiorg101016jfishres201604012

van Poorten B T amp Brydle S (2018) Estimating fishing effortfrom remote traffic counters Opportunities and challengesFisheries Research 204 231ndash238 httpsdoiorg101016jfishres201802024

van Poorten B T Carruthers T RWardHGM ampVarkeyD A(2015)Imputingrecreationalanglingeffortfromtime-lapsecamerasusinganhierarchicalBayesianmodelFisheries Research172265ndash273httpsdoiorg101016jfishres201507032

12emsp |emsp emspensp FLYNN et aL

VenturelliPAHyderKampSkovC(2017)AnglerappsasasourceofrecreationalfisheriesdataOpportunitieschallengesandproposedstandardsFish and Fisheries18578ndash595httpsdoiorg101111faf12189

WiseBSTelferCFLaiEKMHallNGampJacksonG(2012)Long-termmonitoringofboat-basedrecreational fishing inSharkBayWesternAustraliaProvidingscientificadviceforsustainablemanagement in a World Heritage Area Marine and Freshwater Research631129ndash1141httpsdoiorg101071MF12054

WoodGLynchTPDevineCKellerKampFigueiraW(2016)High-resolutionphoto-mosaictime-seriesimageryformonitoringhumanuseofanartificialreefEcology and Evolution66963ndash6968httpsdoiorg101002ece32342

YuFRupertACCDiTizianaMNishikantaKRandySIllahNhellipJeffreyG(2011)TerapixelimagingofcosmologicalsimulationsThe Astrophysical Journal Supplement Series19718

Zischke M T Griffiths S P amp Tibbetts I R (2012) Catch and ef-fort from a specialised recreational pelagic sport fishery off

eastern Australia Fisheries Research 127ndash128 61ndash72 httpsdoiorg101016jfishres201204011

SUPPORTING INFORMATION

Additional supporting information may be found online in theSupportingInformationsectionattheendofthearticle

How to cite this articleFlynnDJHLynchTPBarrettNSWongLSCDevineCHughesDGigapixelbigdatamoviesprovidecost-effectiveseascapescaledirectmeasurementsofopen-accesscoastalhumanusesuchasrecreationalfisheriesEcol Evol 2018001ndash12 httpsdoiorg101002ece34301

Graphical AbstractThecontentsofthispagewillbeusedaspartofthegraphicalabstractofhtmlonly

Itwillnotbepublishedaspartofmainarticle

Wetrailedaremoteautonomousultra-high-resolutionphotosamplingmethodasacosteffectsolutionfordirectseascape-scaledmeasure-mentsOursequentialphotosamplingwasbatchedprocessesusingnovelTimeMachinedatahandlingsoftwaretoproduceldquobigdatardquotimeseriesmoviesfromaspatialsubsetofanoffshorefisherywhichwecomparedtoaregionalbus-routesurveyofboattrailersandinterviewswithparticipantsOurcamerasamplescloselytrackedlargerscaleregionaltrendsbutwithlaborcostsaquarterofthebus-routesurveycostsforanindividualsample25ofthebusroutes

Page 6: Gigapixel big data movies provide cost-effective seascape

6emsp |emsp emspensp FLYNN et aL

Based on boat ramp interviews the average trailer boat partysize (P) was 287 people per vessel throughout May (plusmn0133 SEn=84)andpartysizebetweensitesshowednosignificantdiffer-ence (df=2F = 302p = 068) The averagemonthly boat fishingeffort(hours)extrapolatedforourCRAGSobservationsofHRMCAwas5629hr(plusmn711SE)with4096hr(plusmn607SE)inMayand1534hr(plusmn370 SE) in June (Figure4)mdashthis could be extrapolated to fisherhoursbymultiplyingbytheaveragepartysizebutasnointerviewswereconductedinJuneweleftboththeCRAGSandbusrouteex-trapolationsasboatfishingeffortonly

The total subset area observed by CRAGSwas 96km2 which comparedto318km2thatwaseasilyavailabletotrailerboatsaroundtheTasmanPeninsularwhichcorrespondstoaround3ofthetotalareaComparedtothetotalfishingeffortforMayextrapolatedfromtheregionalbus-routesurvey therelativelysmallareaaroundtheHRM(~3)receivedalargeamountofthefishingeffort(~23)

ThehighfrequencyofCRAGSsamplingpermittedtheexamina-tionoffine-scaleinter-dailytemporalvariationwhichdisplayedarag-gedsinusoidoscillation(Figure5)Effortwasgenerallylowwithinthe

weekdaystrataandthen increasedrapidly to largepeaksonweek-ends (Figure5) One exception to this was between the 18th and22ndMaywhichhadmoreeffortonweekdays thanotherperiods(Figure5)Bracketingthisperiodweretwoweekendsoffishingcom-petitions theldquoTunaClubofTasmaniarsquosrsquoFarSouthClassicrdquo (16ndash17thMay)andRally4Northerninvitationalweekendrally(23rdMay)

A total of 16 bus-route samples were conducted logistic fail-uresledtotheabandonmentoffour(20)oftheplannedsamplesRegionalTrailerboateffortfromaroundtheTasmanPeninsularwasestimatedtobe19650hr (plusmn2858SE) inMay (Figure4)Notrailersat individual rampswere recorded four timeswith all of theseoc-curringonweekdaysThetotalnumberoftrailerssightedperramprangedfrom0to16duringweekdaysand1ndash54forweekendswiththehighestnumberrecordedon16May2015atPiratesBayDuringhigh-intensitydays(particularlyweekendsholidaysmornings)therewereoftenlargeoverflowsfromrampcarparkswithparkedcarsandboattrailersextendingasmuchasonekilometerfromtheboatramp

During thebus-route survey 84 interviewswere conductedThemajorityofmariners(905n=96)indicatedfishingwastheir

F IGURE 4emspExtrapolatedmonthlytrailerboateffort(hoursplusmnSE)fortheentireTasmanPeninsulaassessedbyabus-routesurvey(May)andforthesubsetoftheareaaroundtheHRMCAsampledwithCRAGS(MayandJune)

F IGURE 5emspHigh-frequencyinterdailyextrapolatedCRAGSestimationsofboatefforthoursaroundthe(HRMCA)

0

100

200

300

400

500

600

Rolli

ng m

ean

daily

effo

rt in

hou

rs

DayWeekdays Weekends + Public Holidays

lyn088
Inserted Text
CA
lyn088
Cross-Out
lyn088
Cross-Out
lyn088
Inserted Text
t

emspensp emsp | emsp7FLYNN et aL

primaryactivity (Table1)Overall643 (n=54) indicated theirprimary target were tuna with 179 (n=15) specifically tar-geting southern bluefin tuna (Thunnus maccoyii)Other commonrecreationaltargetsincludedflathead(Platycephalusspp595)Southernrocklobster(Jasus edwardsii107)andsquidcalamari(Sepioteuthis australisandNototodarus gouldi476)(Table1)

Basedon the responsesofactivity type from interviews totalboatfishingeffortinMaywasestimatedtobe17783hr(plusmn2586SE)Asmostnonfishingactivitieswerenear shoreweassumed trailerboatingactivityatHMRCAobservedbyCRAGSwasallfishing

Therewere16directcomparisonsamplesbetweenthebusrouteandCRAGSmethodsbetweenthe2ndandthe29thofMaySamplesconsisted of 7 weekdays and 9 weekendspublic holiday samples

(Figure6)Infourcaseswheretwobus-routesamplesweretakenonthe same day (Supporting Information Appendix S2) they are com-bined toadailyestimation for tabulation (Table2)Theproportionaldifferencebetweenestimatedeffortsusingthetwomethodsrangedfrom078-to418-folddifferencehoweverranksestimatesbetweenCRAGS and bus-route samples were strongly associated (n=16ρ(12)=087p=lt001)Interdailysamplingstrata(ampm)werealsoexaminedseparatelyforcorrelationwhichremainedstrongforbotham(n=9ρ(7)=080p=lt001)andpm(n=7ρ(5)=089p=lt001)

ForthecomparativemonthCRAGStook169samplescomparedtothebusroutesrsquo16(Table3)WehadtwofieldtripdaysforCRAGSwith four people on the setup and three on the breakdown daywhichresultedinsevenfull-timeequivalent(FTE)daysofworkThiscomparedto16fielddaysforthebus-routemethodwhichincludedtraveltothefieldsiteforbetweentwoandthreestaffwhichaccu-mulatedto40FTEdaysofworkSettingupourCRAGSautomatedstitching took one FTE day (though this is faster for experiencedofficers)whiledataextractionperimagewasestimatedtotakeon

TABLE 1emsp Intervieweesreportedprimarytargettaxaoractivity

Primary target taxaspeciesactivity Interview responses

Tuna(unspecified) 37 440

SouthernBluefinTuna 15 179

AlbacoreTuna 2 238

AllTuna 54 643

Flathead 5 595

Southernrocklobster 9 107

Squidcalamari 4 476

Abalone 1 119

SmallPelagics 1 119

SnottyTrevalla 2 238

Allnearshorespecies 22 262

Surfing 1 119

SCUBA 3 357

DemonCavevisitors 3 357

PleasureCruise 1 119

Allnonextractive 8 952

Note ldquoTunardquo responsesdenote anglers targeting all specieswithin thefamilyThunnini

F IGURE 6emspDailyboatingeffort(hours)extrapolatedusingbus-routeaccesspointsurvey(bar)andCRAGS(line)Effortestimatefrombus-routesurveywascalculatedfrommorning(amdarkgray)andafternoon(pmlightgray)surveyCRAGSsurveywasconductedbetweenMayandJunewhilethebus-routesurveywasonlyconductedinMay

TABLE 2emspRankingofeffortforpairedsamplesbetweenCRAGSandthebusroute

DateCRAGS hours (rank)

Bus- route hours (rank) Δ Rank

2052015 330(9) 1208(10) minus1

3052015 303(8) 1471(11) minus3

12052015 0(1) 48(2) minus1

14052015 0(1) 24(1) 0

16052015 387(11) 1794(12) minus1

19052015 93(4) 89(4) 0

20052015 57(3) 117(5) minus2

22052015 228(7) 283(6) 1

23052015 513(12) 414(8) 4

24052015 345(10) 1155(9) 1

25052015 122(6) 323(7) minus1

29052015 102(5) 79(3) 1

8emsp |emsp emspensp FLYNN et aL

average75mincomparedto1minforenteringthebus-routedatasheetWhiledataextraction forCRAGS took longer compared tothe bus routemdashat 381 FTE daysmdashtotal monthly labor for CRAGStookonlyaquarterofthetimeandonaper-samplecomparisonwasonly25ofthebus-routelaborcost

4emsp |emspDISCUSSION

Over our autumn samplingmost trailer boat owners accessing thewaters around theTasmanPeninsula via the regional boating infra-structurewererecreationallyfishingwithfisherspredominatelytar-getingtunaTrailerboatrecreationalfishersconstitutealargetrancheof coastal users in Australia (McPhee etal 2002) and our resultssuggestedthatthisregionremainsonewhererecreationalfishing isconcentrated(MortonampLyle2004)EffortobservedbyCRAGSwasa spatial subsetof thegeneral areaeasily accessedby trailer boatsaroundtheTasmanPeninsularwhichweregionallyassessedwithourbus-routerovingsurveyAsCRAGSonlyobserved~3of thetotalarea but corresponded to ~23 of themonthly bus-route boat ef-fort theHippolyteRocksMarineConservationArea (HRMCA) is afurtherconcentrationpointwithintheregionforrecreationalfishingAsthisoffshorefishingeffortaroundtheHRMCAwasmostprobablyfocusedontotunamdashwhichaccountedfor634ofallinterviewedfish-ers activitiesmdashthe actual proportionofpeninsularwide tuna fishingthatwasobservedbyCRAGSwaspotentiallymuchhigherthan23

For developing effort metrics finding a representative site toobservemaybe important forprecise trackingof trendsThetightrankcorrelationbetweenpairedregionalbus-routeandCRAGScam-era samples suggest that effort atHRMCAwas representative ofregionaltemporaltrendsOurresultsarealsosimilartootherstud-ieswheresamplingmethodsforrecreationalfisheriesoverbroaderscaleswhencomparedtoindextrendsofeffortcollectedfrompointsourcesshowstrongcorrelations(Hartilletal2016)

CRAGSwithitsseascapescaleofdatacaptureobservesactualeffortonfishinggroundsratherthan indirectmetricsfrommove-ment past access points (Hartill etal 2016 Smallwood PollockWise Hall amp Gaughan 2012 van Poorten amp Brydle 2018) This

makesthechoiceofaccesspointstomonitorimmaterialfortrendcol-lectionandmaybeofparticularinterestforpelagicfisheriesWhilenonintuitiverecreationalfishingforwide-rangingmigratorypelagicspeciesisoftenspatiallyconcentratedintosmallareas(Lynch2006)duetotightoverlapsbetweeneaseofaccessbyfishersandfishbe-haviorrelatedtobathymetrymigrationhabitatandpreyavailability(Lynchetal2004PattersonEvansCarterampGunn2008)

Unlikespatialpatternstheintensityofcoastalrecreationaltemporaleffortcanbehighlyvariableovertime(Lynch20082014WiseTelferLaiHallampJackson2012)thoughitisgenerallythoughttofollowpre-dictablepatternsrelativetoholidayandnonholidayperiods(JonesampPollock2012)Ourresultsdemonstratedtheselargefluctuationsrang-ingfromzerouptogt400hroftrailerboateffortforadjacentdaysIncombinationwithholidayperiodsothersourcesofvariabilitymayalsooccurduetoindividualorcombinationsoffactorssuchasfishingcluborcompetitionactivitiesandweatherconditionsHigh-frequencysam-plingcanovercomethesecommontemporaldisjunctionsinobservedfishingeffortwherebychanceduetolownumbersofsampleslogis-ticallypermissiblewithtraditionalrovingoraccesspointsurveyslargeerrorsintheestimationofeffortcanbeintroduced(Hartilletal2016vanPoortenampBrydle2018vanPoortenetal2015)

With multiple daily observations abnormal episodes are morelikelytobeobservedForinstanceaweek-longperiodofintenseef-fortwasrecordedbetween15May2015and27May2015whichwasbracketedbytwoconsecutiveweekendswherefishingcompetitionswereheldontheTasmanPeninsularThenormalsinusoidpatternoflowweekdayandhighweekendeffortwasmuchlesspronouncedastheweekdaylullsineffortwerepartiallybridgedbyanglerscontinuingtofishAsbus-routedesignsusedaytypeasstratawithlowerprob-abilitiesforsamplingweekdaysandmuchfewersamplesoverallthistypeofeffectcouldeasilybemissedWhileourbus-routesamplesdidoverlapwiththetunafishingcompetitionsthiswaspurelybychanceSucheventswhichresultinlargespikesofeffortcanhavemarkedim-pactsonlocalecosystemsandharvestestimates(McPheeetal2002MonzPickeringampHadwen2013)andwithoutpriorknowledgearemorelikelytobecapturedwithmonitoringathigherfrequencies

Methodsforgatheringdataoncoastalusesuchasrecreationalfishing moved away from direct approaches and toward off-sitemethodssuchastelephoneinterviews(McCluskeyampLewison2008Pollock etal 1994) due to unsustainable costs from staffing andoperationsparticularlyduringperiodsof intenseusesuchasearlymornings weekends and public holidays which require overtimepaymentsHoweversamplingissueswithoff-sitemethodsareofpar-ticularconcernforopen-accessactivitiessuchasunlicensedfisheriesasthereisnolicensedatabaseofcontactphonenumbersonwhichtobaseasampleframeThisisparticularlysofornicherecreationalfisheriessuchaspelagicgamefishingasthefishersbecomeincreas-inglydilutewithinthegeneralpopulationandhencearehardtoac-cessparticularlyviaoff-sitemethodssuchastelephone interviews(Griffithsetal2010)

Remoteandautonomousdeploymentofsensorsmaythereforeprovideanalternativeandcost-effectivemethodforlong-termmon-itoring particularly for effort across a range of applicationsOur

TABLE 3emspFulltimeequivalent(FTE)labordaysforCRAGSandbusroutesurveystoestimatepatternsofeffort

CRAGS Bus route

Numberofsamples 169 16

Fielddays 2 16

Averagepeopleperfieldday 35 25

Field FTE days subtotals 7 40

StitchingsetupFTE 1 0

DataextractionFTE(persample) 0017 0002

Subtotal data extraction FTE 381 004

FTEdayspersample 006 250

Total FTE days 108 400

emspensp emsp | emsp9FLYNN et aL

comparisonsofmatchedpairsofdatafromtwomethodssuggestedthat theCRAGSoutputswerewell correlatedwith regionalefforttrendsOuranalysisoflabordemonstratesthattryingtounderstandthistemporalvariabilitywithsamplingviaboatrampsurveyswouldbeprohibitivelyexpensiveOurCRAGSallowsforatemporaleffortmetric derived fromactual observationsof fishing to be cheaplydeveloped with sampling frequency at subdaily scales This high-frequencysamplingapproachwasvalidatedagainstour largerbutmuchmorecostlybus-routesurveyOuranalysisoflaborrequiredtobothcollectandprocessdatashowedthatCRAGSrequiredconsid-erablylessresourcestocollectapproximatelyninetimesmorefre-quenttemporaldatathanthebus-routemethodMajorlaborsavingswereduetotheruggedizedandautonomouscapabilityofCRAGSasonlytwofieldtripswererequired(deploymentandretrievaltook7FTEdays)comparedtothebus-routersquos16fieldtrips inamonthwhichconsumed40FTEdaysandincludedanadditional780kmofinter-ramproadtravel(SupportingInformationAppendixS3)AsitisweatherproofautonomousandabletobeprogrammedwithatimedelaysetupCRAGSldquofireandforgetrdquonaturebothremovestheneedforobservertobephysicallylocatedattheregiontocountactivity(EdwardsampSchindler2017)andworkduringovertimeperiodsTheldquoblankingperiodrdquowhichstopsandstartssamplingmdashinthiscaseover-nightmdashalso reduces battery and memory use In combined thesetechnological advances have allowed long-term CRAGS deploy-mentsacross4yearsof4ndash6monthrotationsontooffshoreislandsand for several seasons in Antarctica to monitor seabirds (Lynchetal2017)Thelaborestimateforthebusroutewasconservativeasitdidnotincludeovertimeloadingforthe6weekenddaysofsam-plingorearlymorningcommencementofamsamples

Theadditional381FTEprocessingtimerequiredbythecameraapproach did not impact greatly on the overall labor budgetwhenconsideredagainstfieldworkTheuseoftheTimeMachinebatchpro-cessingsoftwaretoautomaticallyproducetheldquobigdatardquointeractivemoviewasalsoatechnologicalinnovationthatsignificantlyreducedthelaborofdatahandlingcomparedtothemanualopeningofindivid-ualGigaPansusedinpreviousstudies(Lynchetal2015Woodetal2016)IncomparativetermsattheindividualsampleleveltotracktheeffortmetriceachCRAGSsamplerequired24ofthelaborofthebus-routesurvey

Thedevelopmentandapplicationofimagerecognitionsoftware(machine learning) to detect classify and count boats automati-cally (Bousetouane amp Morris 2015) would further reduce post-processingcostsTherequiredfrequencyofsamplingshouldalsobeconsideredonacase-by-casebasisdependentontheresearchquestion as image collection can be somewhat open-ended andresult inunwieldydata sets that require subsampling tobe cost-effective(Parnelletal2010)Inourcasethesettingof~5imagesperdayappearedtoadequatelydescribetheinterdailyvariation

Whileouroneautonomouscameraprovidedastrongcorrelationwithtemporaltrendsineffortcomparedtotheregionalsurveyitdidnotprovide relativespatialdistributionsofeffortacross the fisheryThesedistributionshowevercanbepredictableandhighlyconcen-trated(Lynch20062014Parnelletal2010)anobservationthatwas

alsoshownbyourresultsandmuchsmallerlevelsofsamplingeffortwouldprobablyberequiredtoresolvevariationinspatialdistributionscomparedtowhatisrequiredtoresolvethehightemporalvariationsineffortUnderstandingfine-scalespatialdistributionsofrecreationalfishingeffortiscommonlyachievedusingaerialcounton-watersur-veysorvantagepointsurveys(Lynch2006Smallwoodetal2012)thoughotherremotemethodssuchashigh-resolutionsatelliteimag-ery(FretwellScofieldampPhillips2017)mayprovideacheapersolu-tionThesetypesofwidersurveyshavealsobeensuggestedashighlycompatiblewithpointsourcemetricssuchasremotecamerasystemstoallowforcalibrationorldquoup-ratingrdquoforeffortexpansionstoestimateoveralleffortwithinthefishery(Hartilletal2016)

Thelongdistance(~6km)betweentheHMRCAandthedeploy-ment location stretched the resolution limit of the current CRAGSsetup aswewere unable to distinguish party size or activity typepurelyusing thecamerasystemThis isunlikepreviousCRAGSde-ploymentwhichhadahighsuccessrateofidentifyingpartysizeac-tivity typeata shorter rangeofapproximately19km (Woodetal2016) Thus ground-truthing information about the fisherywas re-quiredforthesystemtoworkeffectivelyAutonomouscamerasystemalsodonotcollectcreel(catch)databutbyallowingforcost-effectivecollectionofthehighlyvariabletemporaleffortdatatheymayallowforhumanresourcestobespentelsewhereduringsurveyperiods(egtoundertakeboatrampinterviews)(vanPoortenampBrydle2018)

Our studywas a small-scale trial of new technology andwhilepreliminary results are encouraging validated seasonal effort esti-mationsobservedviaCRAGSwereoutsidethescopeofthisprojectOur comparisons would hence have been improved by extendingCRAGSdeploymentsandwithmorecomparisonstoregionalsurveysSimilarly usingmultiple CRAGS simultaneously would have bettertested for differences in subregional patterns of temporal trendsorconcentrationof infishingeffortfor instanceboatsaroundthenearbyTasmanIsland

Insummarytheprogrammablesamplinganddataextractionof-feredbyhigh-resolutionautonomouscamerasystemscanallowforfine-scalelong-termmonitoringoffishingefforttrendsorpotentiallymanyothertypesofactivitiesatseascapescalesMorewidespreadadoptionofhigh-frequencysamplingapproachesforeffortestima-tionwouldalsohelpavoidcollectionofmisrepresentativedatabyidentifyingeffortanomalies(McCluskeyampLewison2008)suchastheweekdaycontinuationeffectofraisedeffortbetweenadjacentweekendfishingcompetitionByimprovingtheinformationdensityovertraditionalsurveymethodsautonomouscamerasystemsandotherremotesystemsmayprovideanewwaveofoptimizationfordirectmeasurementofcoastalhumanusesuchasrecreationalfish-eriesalthoughcomplimentarymethodsarestillrequiredtodeter-minespatialdistributionstargetspeciesandharvest

ACKNOWLEDG MENTS

We thankMsGeorginaWoodMs SaraAdkinsMsBelleMonkMrRSherringtonandMsPaigeKellyforconstructiveinsighten-thusiasmandassistanceduringfieldworkWealsothankMrPhillip

10emsp |emsp emspensp FLYNN et aL

CulversonMrStuartDudgeonandMrMattStapenellKarenEvensprovidedhelpfulcommentsonaninternaldraftGreatestthankstothemany recreational anglerswhowillingly shared information tosurveyclerkswithouttheirinvolvementthisprojectwouldnothavecometofruitionThreeanonymousreviewersprovidedcommentsthatimprovedthemanuscript

CONFLIC T OF INTERE S TS

Theauthorsdeclarenoconflictofinterests

AUTHOR CONTRIBUTIONS

MrFlynnhelpeddesignthestudyledthefieldworkundertookmostoftheanalysisandhelpwritethemanuscriptDrLynchhelpdesignthestudyassisted inthefieldworkprovidedadviceontheanaly-sisanddidsomeanalysisconstructedfiguresandafterMrFlynncontributedthemosttothedraftingofthemanuscriptDrBarretthelpeddesign the study assisted in some fieldwork provided ad-viceonanalysisandeditedthemanuscriptMrWonghelpedexten-sivelywithfieldworkassistedwithanalysisconstructedfiguresandhelpedwritethemanuscriptMsDevineassistedwithfieldworkandconfigurationoftheequipmentincludingtheapplicationofthebigdatamoviesoftwareandhandlingofphotographsandalsohelpedwritethemanuscriptMrHughesdevelopedtheelectronicsfortheequipmentoversaw thebuildofgearwrote the software topro-gramtherobotandreviewedandeditedthemanuscript

DATA ACCE SSIBILIT Y

Survey sampling schedules data collection sheets extrapolationproceduresandtechnicalCRAGSspecificationsarealluploadedasonlinesupportinginformationappendixRawdatafromsurveysandCRAGSDOIhttpdoiorg105061dryadv10d218

RE SE ARCH E THIC S COMPLIANCE AND FUNDERS

ResearchabidesbytheAustraliancodeofhumanresearchandex-perimentation(NationalStatementonEthicalConductinHumanResearch of 2007) Permission to conduct research on humanparticipants was granted by the University of Tasmania SocialScience under reference H0014772mdashldquoRemote investigation ofcoastal area user group quantification and variationrdquo aswell asfromtheCSIROSocialScienceHumanResearchEthicsCommitteeunder reference 08115mdashldquoHawkesbury Region Project HumanUsePatternsrdquoFundingwasbyUTASasPartoftheIMASHonoursProject funding and through internal support from the CoastsProgram

ORCID

Tim P Lynch httporcidorg0000-0002-5346-8454

R E FE R E N C E S

Alessa L Kliskey A amp Brown G (2008) Socialndashecological hotspotsmapping A spatial approach for identifying coupled socialndasheco-logicalspaceLandscape and Urban Planning8527ndash39httpsdoiorg101016jlandurbplan200709007

ArlinghausR(2006)Overcominghumanobstaclestoconservationofrecreational fishery resources with emphasis on central EuropeEnvironmental Conservation 33 46ndash59 httpsdoiorg101017S0376892906002700

Arlinghaus R Tillner R amp Bork M (2015) Explaining participationratesinrecreationalfishingacrossindustrialisedcountriesFisheries Management and Ecology 22 45ndash55 httpsdoiorg101111fme12075

BadcockPBPatrickKSmithAMASimpsonJMPennayDRisselCEhellipRichtersJ(2016)Differencesbetweenlandlineandmobilephoneusers in sexualbehavior researchArchives of Sexual Behavior46(6)1711ndash1721

Blumberg S J amp Luke J V (2009) Reevaluating the need for con-cern regarding noncoverage bias in landline surveys American Journal of Public Health 99 1806ndash1810 httpsdoiorg102105AJPH2008152835

Bousetouane F ampMorris B (2015) Off-the-shelf CNN features forfine-grainedclassificationofvesselsinamaritimeenvironmentInGBebisRBoyleBParvinDKoracinIPavlidisRFerisTMcGrawMElendtRKopperERaganZYeampGWeber(Eds)Advances in visual computing 11th International symposium ISVC 2015 Las Vegas NV USA December 14ndash16 2015 proceedings Part II (pp379ndash388)Cham Switzerland Springer International Publishing httpsdoiorg101007978-3-319-27863-6

Cooke S J amp Cowx I G (2004) The role of recreational fish-ing in global fish crises BioScience 54 857ndash859 httpsdoiorg1016410006-3568(2004)054[0857TRORFI]20CO2

DiggleP J (1995)Time-series a biostatistical introductionOxfordUKClarendonPress

EdwardsJampSchindlerE (2017)Avideomonitoringsystemtoeval-uate ocean recreational fishing effort in Astoria OregonMarine recreational information program (pp 31) Newport OR OregonDepartmentofFishandWildlife

FarrMStoecklNampSuttonS(2014)RecreationalfishingandboatingArethedeterminantsthesameMarine Policy47126ndash137httpsdoiorg101016jmarpol201402014

ForbesETraceySampLyle JM (2009)Assessment of the 2008 rec-reational gamefish fishery of Southeast Tasmania with particular refer-ence to Southern Bluefin TunaHobartTASTasmanianAquacultureandFisheriesInstituteUniversityofTasmania

FretwellPTScofieldPampPhillipsRA(2017)Usingsuper-highres-olutionsatelliteimagerytocensusthreatenedalbatrossesIbis159481ndash490httpsdoiorg101111ibi12482

GiriKampHallK (2015)SouthAustralian recreational fishingsurveyFisheries Victoria internal report series (pp 63) Queenscliff VicFisheriesVictoria

Griffiths S P Pollock K H Lyle J M Pepperell J G TonksM L amp Sawynok W (2010) Following the chain to elu-sive anglers Fish and Fisheries 11 220ndash228 httpsdoiorg101111j1467-2979201000354x

Halpern B SWalbridge S Selkoe K A Kappel C V Micheli FDrsquoAgrosaChellipWatsonR(2008)AglobalmapofhumanimpactonmarineecosystemsScience319948ndash952httpsdoiorg101126science1149345

Hartill BW Payne GW Rush N amp Bian R (2016) Bridging thetemporal gap Continuous and cost-effective monitoring of dy-namic recreational fisheries by web cameras and creel surveysFisheries Research 183 488ndash497 httpsdoiorg101016jfishres201606002

emspensp emsp | emsp11FLYNN et aL

HenryGWampLyleJM(2003)NationalrecreationalandindigenousfishingsurveyFinal report to the fisheries research amp development cor-poration and the fisheries action program (pp 188) Canberra ACTFRDC

JonesCMampPollockKH(2012)Recreationalanglersurveymeth-odsEstimationofeffortharvestandreleasedcatchInAZaleDLParrishampTMSutton(Eds)Fisheries techniquesBethesdaMDAmericanFisheriesSociety

Kearney R E (2002) Co-management The resolution of conflict be-tween commercial and recreational fishers in Victoria AustraliaOcean amp Coastal Management45201ndash214httpsdoiorg101016S0964-5691(02)00055-8

KellerKSteffeASLowryMMurphyJJampSuthersIM(2016)Monitoringboat-basedrecreationalfishingeffortatanearshorearti-ficialreefwithashore-basedcameraFisheries Research18184ndash92httpsdoiorg101016jfishres201603025

KinlochMAMcGlennonDNicollGampPikePG(1997)Evaluationof the bus-route creel surveymethod in a largeAustralianmarinerecreationalfisheryISurveydesignFisheries Research33101ndash121httpsdoiorg101016S0165-7836(97)00068-4

LewinWCArlinghausRampMehnerT(2006)Documentedandpo-tential biological impacts of recreational fishing Insights forman-agementandconservationReviews in Fisheries Science14305ndash367httpsdoiorg10108010641260600886455

LowryMampMurphy J (2003)Monitoring the recreational gamefishfisheryoffsouth-easternAustraliaMarine and Freshwater Research54425ndash434httpsdoiorg101071MF01269

LyleJMStarkKEampTraceySR(2014)2012ndash2013 Survey of rec-reational fishing in TasmaniaHobart TAS Institute forMarine andAntarcticStudies

Lynch T P (2006) Incorporation of recreational fishing ef-fort into design of marine protected areas Incorporacioacuten delEsfuerzo de Pesca Recreativa en el Disentildeo de Aacutereas MarinasProtegidas Conservation Biology 20 1466ndash1476 httpsdoiorg101111j1523-1739200600509x

LynchTP(2008)Thedifferencebetweenspatialandtemporalvaria-tioninrecreationalfisheriesforplanningofmarineprotectedareasResponse toSteffeConservation Biology22486ndash491httpsdoiorg101111j1523-1739200800888x

LynchTP (2014)Adecadal time-seriesof recreational fishingeffortcollectedduringandafterimplementationofamultipleusemarinepark shows high inter-annual but low spatial variability Fisheries Research15185ndash90httpsdoiorg101016jfishres201309014

LynchTPAldermanRampHobdayAJ(2015)Ahigh-resolutionpan-orama camera system formonitoring colony-wide seabird nestingbehaviourMethods in Ecology and Evolution6491ndash499httpsdoiorg1011112041-210X12339

LynchTFlynnDWoodGKimJ-HParkJampHobdayAJ(2017)Metrics parameters and results from fisheries and wildlife studies with the CSIRO Ruggedized Autonomous Gigapixel Systems (CRAGS)DarwinNTAustralianMarineScienceAssociation

LynchTPWilkinsonEMellingLHamiltonRMacreadyAampFearyS(2004)ConflictandimpactsofdiversandanglersinamarineparkEnvironmental Management 33 196ndash211 httpsdoiorg101007s00267-003-3014-6

MaHOgawaTKSminkeyTRBreidtFJLesserVMOpsomerJDhellipVanVoorheesDA(2018)Pilotsurveystoimprovemonitoringofmarine recreational fisheries inHawaiʻiFisheries Research204197ndash208httpsdoiorg101016jfishres201802010

McCluskey S M amp Lewison R L (2008) Quantifying fish-ing effort A synthesis of current methods and their ap-plications Fish and Fisheries 9 188ndash200 httpsdoiorg101111j1467-2979200800283x

McGlennonDampKinlochMA(1997)Evaluationofthebus-routecreelsurveymethod in a largeAustralianmarine recreational fishery II

PilotsurveysandoptimalsamplingallocationFisheries Research3389ndash99httpsdoiorg101016S0165-7836(97)00067-2

McPheeDLeadbitterDampSkilleterGA(2002)SwallowingthebaitIs recreational fishing in Australia ecologically sustainable Pacific Conservation Biology840httpsdoiorg101071PC020040

Monz C A Pickering C M amp HadwenW L (2013) Recent ad-vances in recreation ecology and the implications of differentrelationships between recreation use and ecological impactsFrontiers in Ecology and the Environment11441ndash446httpsdoiorg101890120358

MooreAHallKGiriKTraceySPenroseLHansenShellipBrownP(2015)Developingrobustandcost-effectivemethodsforestimatingthenationalrecreationalcatchofSouthernBluefinTunainAustraliaFRDC Projectpp123

Morton A J amp Lyle J M (2004) Preliminary assessment of the rec-reational gamefish fishery in Tasmania with particular reference to Southern Bluefin Tuna Taroona TAS Tasmanian Aquaculture andFisheriesInstitute

ParnellPEDaytonPKFisherRALoarieCCampDarrowRD(2010)SpatialpatternsoffishingeffortoffSanDiegoImplicationsfor zonal management and ecosystem function Ecological Applications202203ndash2222httpsdoiorg10189009-15431

PattersonTAEvansKCarterTLampGunnJS(2008)Movementand behaviour of large southern bluefin tuna (Thunnus maccoyii)in the Australian region determined using pop-up satellite ar-chival tags Fisheries Oceanography 17 352ndash367 httpsdoiorg101111j1365-2419200800483x

PollockKH JonesCMampBrownTL (1994)Angler survey meth-ods and their applications in fisheries management Bethesda MDAmericanFisheriesSociety

PonsM Branch T AMelnychukM C JensenO P Brodziak JFromentinJMhellipHilbornR(2017)EffectsofbiologicaleconomicandmanagementfactorsontunaandbillfishstockstatusFish and Fisheries181ndash21httpsdoiorg101111faf12163

PowersSPampAnsonK (2016)Estimating recreationaleffort in theGulf of Mexico Red Snapper Fishery using boat ramp camerasReduction in federalseason lengthdoesnotproportionally reducecatchNorth American Journal of Fisheries Management 36 1156ndash1166httpsdoiorg1010800275594720161198284

Robson D amp Jones C M (1989) The theoretical basis of an ac-cess site angler survey design Biometrics 45 83ndash98 httpsdoiorg1023072532036

Rocklin D Levrel H Drogou M Herfaut J amp Veron G (2014)Combining telephone surveys and fishing catches self-reportTheFrenchseabass recreational fisheryassessmentPLoS ONE9e87271httpsdoiorg101371journalpone0087271

Smallwood C B Pollock K HWise B S Hall N G amp GaughanD J (2012) Expanding aerial-roving surveys to include counts ofshore-based recreational fishers from remotely operated camerasBenefitslimitationsandcosteffectivenessNorth American Journal of Fisheries Management321265ndash1276httpsdoiorg101080027559472012728181

Teixeira D ZischkeM T ampWebley J A C (2016) Investigatingbias in recreational fishing surveys Fishers listed in public tele-phone directories fish similarly to their unlisted counterpartsFisheries Research 181 127ndash136 httpsdoiorg101016jfishres201604012

van Poorten B T amp Brydle S (2018) Estimating fishing effortfrom remote traffic counters Opportunities and challengesFisheries Research 204 231ndash238 httpsdoiorg101016jfishres201802024

van Poorten B T Carruthers T RWardHGM ampVarkeyD A(2015)Imputingrecreationalanglingeffortfromtime-lapsecamerasusinganhierarchicalBayesianmodelFisheries Research172265ndash273httpsdoiorg101016jfishres201507032

12emsp |emsp emspensp FLYNN et aL

VenturelliPAHyderKampSkovC(2017)AnglerappsasasourceofrecreationalfisheriesdataOpportunitieschallengesandproposedstandardsFish and Fisheries18578ndash595httpsdoiorg101111faf12189

WiseBSTelferCFLaiEKMHallNGampJacksonG(2012)Long-termmonitoringofboat-basedrecreational fishing inSharkBayWesternAustraliaProvidingscientificadviceforsustainablemanagement in a World Heritage Area Marine and Freshwater Research631129ndash1141httpsdoiorg101071MF12054

WoodGLynchTPDevineCKellerKampFigueiraW(2016)High-resolutionphoto-mosaictime-seriesimageryformonitoringhumanuseofanartificialreefEcology and Evolution66963ndash6968httpsdoiorg101002ece32342

YuFRupertACCDiTizianaMNishikantaKRandySIllahNhellipJeffreyG(2011)TerapixelimagingofcosmologicalsimulationsThe Astrophysical Journal Supplement Series19718

Zischke M T Griffiths S P amp Tibbetts I R (2012) Catch and ef-fort from a specialised recreational pelagic sport fishery off

eastern Australia Fisheries Research 127ndash128 61ndash72 httpsdoiorg101016jfishres201204011

SUPPORTING INFORMATION

Additional supporting information may be found online in theSupportingInformationsectionattheendofthearticle

How to cite this articleFlynnDJHLynchTPBarrettNSWongLSCDevineCHughesDGigapixelbigdatamoviesprovidecost-effectiveseascapescaledirectmeasurementsofopen-accesscoastalhumanusesuchasrecreationalfisheriesEcol Evol 2018001ndash12 httpsdoiorg101002ece34301

Graphical AbstractThecontentsofthispagewillbeusedaspartofthegraphicalabstractofhtmlonly

Itwillnotbepublishedaspartofmainarticle

Wetrailedaremoteautonomousultra-high-resolutionphotosamplingmethodasacosteffectsolutionfordirectseascape-scaledmeasure-mentsOursequentialphotosamplingwasbatchedprocessesusingnovelTimeMachinedatahandlingsoftwaretoproduceldquobigdatardquotimeseriesmoviesfromaspatialsubsetofanoffshorefisherywhichwecomparedtoaregionalbus-routesurveyofboattrailersandinterviewswithparticipantsOurcamerasamplescloselytrackedlargerscaleregionaltrendsbutwithlaborcostsaquarterofthebus-routesurveycostsforanindividualsample25ofthebusroutes

Page 7: Gigapixel big data movies provide cost-effective seascape

emspensp emsp | emsp7FLYNN et aL

primaryactivity (Table1)Overall643 (n=54) indicated theirprimary target were tuna with 179 (n=15) specifically tar-geting southern bluefin tuna (Thunnus maccoyii)Other commonrecreationaltargetsincludedflathead(Platycephalusspp595)Southernrocklobster(Jasus edwardsii107)andsquidcalamari(Sepioteuthis australisandNototodarus gouldi476)(Table1)

Basedon the responsesofactivity type from interviews totalboatfishingeffortinMaywasestimatedtobe17783hr(plusmn2586SE)Asmostnonfishingactivitieswerenear shoreweassumed trailerboatingactivityatHMRCAobservedbyCRAGSwasallfishing

Therewere16directcomparisonsamplesbetweenthebusrouteandCRAGSmethodsbetweenthe2ndandthe29thofMaySamplesconsisted of 7 weekdays and 9 weekendspublic holiday samples

(Figure6)Infourcaseswheretwobus-routesamplesweretakenonthe same day (Supporting Information Appendix S2) they are com-bined toadailyestimation for tabulation (Table2)Theproportionaldifferencebetweenestimatedeffortsusingthetwomethodsrangedfrom078-to418-folddifferencehoweverranksestimatesbetweenCRAGS and bus-route samples were strongly associated (n=16ρ(12)=087p=lt001)Interdailysamplingstrata(ampm)werealsoexaminedseparatelyforcorrelationwhichremainedstrongforbotham(n=9ρ(7)=080p=lt001)andpm(n=7ρ(5)=089p=lt001)

ForthecomparativemonthCRAGStook169samplescomparedtothebusroutesrsquo16(Table3)WehadtwofieldtripdaysforCRAGSwith four people on the setup and three on the breakdown daywhichresultedinsevenfull-timeequivalent(FTE)daysofworkThiscomparedto16fielddaysforthebus-routemethodwhichincludedtraveltothefieldsiteforbetweentwoandthreestaffwhichaccu-mulatedto40FTEdaysofworkSettingupourCRAGSautomatedstitching took one FTE day (though this is faster for experiencedofficers)whiledataextractionperimagewasestimatedtotakeon

TABLE 1emsp Intervieweesreportedprimarytargettaxaoractivity

Primary target taxaspeciesactivity Interview responses

Tuna(unspecified) 37 440

SouthernBluefinTuna 15 179

AlbacoreTuna 2 238

AllTuna 54 643

Flathead 5 595

Southernrocklobster 9 107

Squidcalamari 4 476

Abalone 1 119

SmallPelagics 1 119

SnottyTrevalla 2 238

Allnearshorespecies 22 262

Surfing 1 119

SCUBA 3 357

DemonCavevisitors 3 357

PleasureCruise 1 119

Allnonextractive 8 952

Note ldquoTunardquo responsesdenote anglers targeting all specieswithin thefamilyThunnini

F IGURE 6emspDailyboatingeffort(hours)extrapolatedusingbus-routeaccesspointsurvey(bar)andCRAGS(line)Effortestimatefrombus-routesurveywascalculatedfrommorning(amdarkgray)andafternoon(pmlightgray)surveyCRAGSsurveywasconductedbetweenMayandJunewhilethebus-routesurveywasonlyconductedinMay

TABLE 2emspRankingofeffortforpairedsamplesbetweenCRAGSandthebusroute

DateCRAGS hours (rank)

Bus- route hours (rank) Δ Rank

2052015 330(9) 1208(10) minus1

3052015 303(8) 1471(11) minus3

12052015 0(1) 48(2) minus1

14052015 0(1) 24(1) 0

16052015 387(11) 1794(12) minus1

19052015 93(4) 89(4) 0

20052015 57(3) 117(5) minus2

22052015 228(7) 283(6) 1

23052015 513(12) 414(8) 4

24052015 345(10) 1155(9) 1

25052015 122(6) 323(7) minus1

29052015 102(5) 79(3) 1

8emsp |emsp emspensp FLYNN et aL

average75mincomparedto1minforenteringthebus-routedatasheetWhiledataextraction forCRAGS took longer compared tothe bus routemdashat 381 FTE daysmdashtotal monthly labor for CRAGStookonlyaquarterofthetimeandonaper-samplecomparisonwasonly25ofthebus-routelaborcost

4emsp |emspDISCUSSION

Over our autumn samplingmost trailer boat owners accessing thewaters around theTasmanPeninsula via the regional boating infra-structurewererecreationallyfishingwithfisherspredominatelytar-getingtunaTrailerboatrecreationalfishersconstitutealargetrancheof coastal users in Australia (McPhee etal 2002) and our resultssuggestedthatthisregionremainsonewhererecreationalfishing isconcentrated(MortonampLyle2004)EffortobservedbyCRAGSwasa spatial subsetof thegeneral areaeasily accessedby trailer boatsaroundtheTasmanPeninsularwhichweregionallyassessedwithourbus-routerovingsurveyAsCRAGSonlyobserved~3of thetotalarea but corresponded to ~23 of themonthly bus-route boat ef-fort theHippolyteRocksMarineConservationArea (HRMCA) is afurtherconcentrationpointwithintheregionforrecreationalfishingAsthisoffshorefishingeffortaroundtheHRMCAwasmostprobablyfocusedontotunamdashwhichaccountedfor634ofallinterviewedfish-ers activitiesmdashthe actual proportionofpeninsularwide tuna fishingthatwasobservedbyCRAGSwaspotentiallymuchhigherthan23

For developing effort metrics finding a representative site toobservemaybe important forprecise trackingof trendsThetightrankcorrelationbetweenpairedregionalbus-routeandCRAGScam-era samples suggest that effort atHRMCAwas representative ofregionaltemporaltrendsOurresultsarealsosimilartootherstud-ieswheresamplingmethodsforrecreationalfisheriesoverbroaderscaleswhencomparedtoindextrendsofeffortcollectedfrompointsourcesshowstrongcorrelations(Hartilletal2016)

CRAGSwithitsseascapescaleofdatacaptureobservesactualeffortonfishinggroundsratherthan indirectmetricsfrommove-ment past access points (Hartill etal 2016 Smallwood PollockWise Hall amp Gaughan 2012 van Poorten amp Brydle 2018) This

makesthechoiceofaccesspointstomonitorimmaterialfortrendcol-lectionandmaybeofparticularinterestforpelagicfisheriesWhilenonintuitiverecreationalfishingforwide-rangingmigratorypelagicspeciesisoftenspatiallyconcentratedintosmallareas(Lynch2006)duetotightoverlapsbetweeneaseofaccessbyfishersandfishbe-haviorrelatedtobathymetrymigrationhabitatandpreyavailability(Lynchetal2004PattersonEvansCarterampGunn2008)

Unlikespatialpatternstheintensityofcoastalrecreationaltemporaleffortcanbehighlyvariableovertime(Lynch20082014WiseTelferLaiHallampJackson2012)thoughitisgenerallythoughttofollowpre-dictablepatternsrelativetoholidayandnonholidayperiods(JonesampPollock2012)Ourresultsdemonstratedtheselargefluctuationsrang-ingfromzerouptogt400hroftrailerboateffortforadjacentdaysIncombinationwithholidayperiodsothersourcesofvariabilitymayalsooccurduetoindividualorcombinationsoffactorssuchasfishingcluborcompetitionactivitiesandweatherconditionsHigh-frequencysam-plingcanovercomethesecommontemporaldisjunctionsinobservedfishingeffortwherebychanceduetolownumbersofsampleslogis-ticallypermissiblewithtraditionalrovingoraccesspointsurveyslargeerrorsintheestimationofeffortcanbeintroduced(Hartilletal2016vanPoortenampBrydle2018vanPoortenetal2015)

With multiple daily observations abnormal episodes are morelikelytobeobservedForinstanceaweek-longperiodofintenseef-fortwasrecordedbetween15May2015and27May2015whichwasbracketedbytwoconsecutiveweekendswherefishingcompetitionswereheldontheTasmanPeninsularThenormalsinusoidpatternoflowweekdayandhighweekendeffortwasmuchlesspronouncedastheweekdaylullsineffortwerepartiallybridgedbyanglerscontinuingtofishAsbus-routedesignsusedaytypeasstratawithlowerprob-abilitiesforsamplingweekdaysandmuchfewersamplesoverallthistypeofeffectcouldeasilybemissedWhileourbus-routesamplesdidoverlapwiththetunafishingcompetitionsthiswaspurelybychanceSucheventswhichresultinlargespikesofeffortcanhavemarkedim-pactsonlocalecosystemsandharvestestimates(McPheeetal2002MonzPickeringampHadwen2013)andwithoutpriorknowledgearemorelikelytobecapturedwithmonitoringathigherfrequencies

Methodsforgatheringdataoncoastalusesuchasrecreationalfishing moved away from direct approaches and toward off-sitemethodssuchastelephoneinterviews(McCluskeyampLewison2008Pollock etal 1994) due to unsustainable costs from staffing andoperationsparticularlyduringperiodsof intenseusesuchasearlymornings weekends and public holidays which require overtimepaymentsHoweversamplingissueswithoff-sitemethodsareofpar-ticularconcernforopen-accessactivitiessuchasunlicensedfisheriesasthereisnolicensedatabaseofcontactphonenumbersonwhichtobaseasampleframeThisisparticularlysofornicherecreationalfisheriessuchaspelagicgamefishingasthefishersbecomeincreas-inglydilutewithinthegeneralpopulationandhencearehardtoac-cessparticularlyviaoff-sitemethodssuchastelephone interviews(Griffithsetal2010)

Remoteandautonomousdeploymentofsensorsmaythereforeprovideanalternativeandcost-effectivemethodforlong-termmon-itoring particularly for effort across a range of applicationsOur

TABLE 3emspFulltimeequivalent(FTE)labordaysforCRAGSandbusroutesurveystoestimatepatternsofeffort

CRAGS Bus route

Numberofsamples 169 16

Fielddays 2 16

Averagepeopleperfieldday 35 25

Field FTE days subtotals 7 40

StitchingsetupFTE 1 0

DataextractionFTE(persample) 0017 0002

Subtotal data extraction FTE 381 004

FTEdayspersample 006 250

Total FTE days 108 400

emspensp emsp | emsp9FLYNN et aL

comparisonsofmatchedpairsofdatafromtwomethodssuggestedthat theCRAGSoutputswerewell correlatedwith regionalefforttrendsOuranalysisoflabordemonstratesthattryingtounderstandthistemporalvariabilitywithsamplingviaboatrampsurveyswouldbeprohibitivelyexpensiveOurCRAGSallowsforatemporaleffortmetric derived fromactual observationsof fishing to be cheaplydeveloped with sampling frequency at subdaily scales This high-frequencysamplingapproachwasvalidatedagainstour largerbutmuchmorecostlybus-routesurveyOuranalysisoflaborrequiredtobothcollectandprocessdatashowedthatCRAGSrequiredconsid-erablylessresourcestocollectapproximatelyninetimesmorefre-quenttemporaldatathanthebus-routemethodMajorlaborsavingswereduetotheruggedizedandautonomouscapabilityofCRAGSasonlytwofieldtripswererequired(deploymentandretrievaltook7FTEdays)comparedtothebus-routersquos16fieldtrips inamonthwhichconsumed40FTEdaysandincludedanadditional780kmofinter-ramproadtravel(SupportingInformationAppendixS3)AsitisweatherproofautonomousandabletobeprogrammedwithatimedelaysetupCRAGSldquofireandforgetrdquonaturebothremovestheneedforobservertobephysicallylocatedattheregiontocountactivity(EdwardsampSchindler2017)andworkduringovertimeperiodsTheldquoblankingperiodrdquowhichstopsandstartssamplingmdashinthiscaseover-nightmdashalso reduces battery and memory use In combined thesetechnological advances have allowed long-term CRAGS deploy-mentsacross4yearsof4ndash6monthrotationsontooffshoreislandsand for several seasons in Antarctica to monitor seabirds (Lynchetal2017)Thelaborestimateforthebusroutewasconservativeasitdidnotincludeovertimeloadingforthe6weekenddaysofsam-plingorearlymorningcommencementofamsamples

Theadditional381FTEprocessingtimerequiredbythecameraapproach did not impact greatly on the overall labor budgetwhenconsideredagainstfieldworkTheuseoftheTimeMachinebatchpro-cessingsoftwaretoautomaticallyproducetheldquobigdatardquointeractivemoviewasalsoatechnologicalinnovationthatsignificantlyreducedthelaborofdatahandlingcomparedtothemanualopeningofindivid-ualGigaPansusedinpreviousstudies(Lynchetal2015Woodetal2016)IncomparativetermsattheindividualsampleleveltotracktheeffortmetriceachCRAGSsamplerequired24ofthelaborofthebus-routesurvey

Thedevelopmentandapplicationofimagerecognitionsoftware(machine learning) to detect classify and count boats automati-cally (Bousetouane amp Morris 2015) would further reduce post-processingcostsTherequiredfrequencyofsamplingshouldalsobeconsideredonacase-by-casebasisdependentontheresearchquestion as image collection can be somewhat open-ended andresult inunwieldydata sets that require subsampling tobe cost-effective(Parnelletal2010)Inourcasethesettingof~5imagesperdayappearedtoadequatelydescribetheinterdailyvariation

Whileouroneautonomouscameraprovidedastrongcorrelationwithtemporaltrendsineffortcomparedtotheregionalsurveyitdidnotprovide relativespatialdistributionsofeffortacross the fisheryThesedistributionshowevercanbepredictableandhighlyconcen-trated(Lynch20062014Parnelletal2010)anobservationthatwas

alsoshownbyourresultsandmuchsmallerlevelsofsamplingeffortwouldprobablyberequiredtoresolvevariationinspatialdistributionscomparedtowhatisrequiredtoresolvethehightemporalvariationsineffortUnderstandingfine-scalespatialdistributionsofrecreationalfishingeffortiscommonlyachievedusingaerialcounton-watersur-veysorvantagepointsurveys(Lynch2006Smallwoodetal2012)thoughotherremotemethodssuchashigh-resolutionsatelliteimag-ery(FretwellScofieldampPhillips2017)mayprovideacheapersolu-tionThesetypesofwidersurveyshavealsobeensuggestedashighlycompatiblewithpointsourcemetricssuchasremotecamerasystemstoallowforcalibrationorldquoup-ratingrdquoforeffortexpansionstoestimateoveralleffortwithinthefishery(Hartilletal2016)

Thelongdistance(~6km)betweentheHMRCAandthedeploy-ment location stretched the resolution limit of the current CRAGSsetup aswewere unable to distinguish party size or activity typepurelyusing thecamerasystemThis isunlikepreviousCRAGSde-ploymentwhichhadahighsuccessrateofidentifyingpartysizeac-tivity typeata shorter rangeofapproximately19km (Woodetal2016) Thus ground-truthing information about the fisherywas re-quiredforthesystemtoworkeffectivelyAutonomouscamerasystemalsodonotcollectcreel(catch)databutbyallowingforcost-effectivecollectionofthehighlyvariabletemporaleffortdatatheymayallowforhumanresourcestobespentelsewhereduringsurveyperiods(egtoundertakeboatrampinterviews)(vanPoortenampBrydle2018)

Our studywas a small-scale trial of new technology andwhilepreliminary results are encouraging validated seasonal effort esti-mationsobservedviaCRAGSwereoutsidethescopeofthisprojectOur comparisons would hence have been improved by extendingCRAGSdeploymentsandwithmorecomparisonstoregionalsurveysSimilarly usingmultiple CRAGS simultaneously would have bettertested for differences in subregional patterns of temporal trendsorconcentrationof infishingeffortfor instanceboatsaroundthenearbyTasmanIsland

Insummarytheprogrammablesamplinganddataextractionof-feredbyhigh-resolutionautonomouscamerasystemscanallowforfine-scalelong-termmonitoringoffishingefforttrendsorpotentiallymanyothertypesofactivitiesatseascapescalesMorewidespreadadoptionofhigh-frequencysamplingapproachesforeffortestima-tionwouldalsohelpavoidcollectionofmisrepresentativedatabyidentifyingeffortanomalies(McCluskeyampLewison2008)suchastheweekdaycontinuationeffectofraisedeffortbetweenadjacentweekendfishingcompetitionByimprovingtheinformationdensityovertraditionalsurveymethodsautonomouscamerasystemsandotherremotesystemsmayprovideanewwaveofoptimizationfordirectmeasurementofcoastalhumanusesuchasrecreationalfish-eriesalthoughcomplimentarymethodsarestillrequiredtodeter-minespatialdistributionstargetspeciesandharvest

ACKNOWLEDG MENTS

We thankMsGeorginaWoodMs SaraAdkinsMsBelleMonkMrRSherringtonandMsPaigeKellyforconstructiveinsighten-thusiasmandassistanceduringfieldworkWealsothankMrPhillip

10emsp |emsp emspensp FLYNN et aL

CulversonMrStuartDudgeonandMrMattStapenellKarenEvensprovidedhelpfulcommentsonaninternaldraftGreatestthankstothemany recreational anglerswhowillingly shared information tosurveyclerkswithouttheirinvolvementthisprojectwouldnothavecometofruitionThreeanonymousreviewersprovidedcommentsthatimprovedthemanuscript

CONFLIC T OF INTERE S TS

Theauthorsdeclarenoconflictofinterests

AUTHOR CONTRIBUTIONS

MrFlynnhelpeddesignthestudyledthefieldworkundertookmostoftheanalysisandhelpwritethemanuscriptDrLynchhelpdesignthestudyassisted inthefieldworkprovidedadviceontheanaly-sisanddidsomeanalysisconstructedfiguresandafterMrFlynncontributedthemosttothedraftingofthemanuscriptDrBarretthelpeddesign the study assisted in some fieldwork provided ad-viceonanalysisandeditedthemanuscriptMrWonghelpedexten-sivelywithfieldworkassistedwithanalysisconstructedfiguresandhelpedwritethemanuscriptMsDevineassistedwithfieldworkandconfigurationoftheequipmentincludingtheapplicationofthebigdatamoviesoftwareandhandlingofphotographsandalsohelpedwritethemanuscriptMrHughesdevelopedtheelectronicsfortheequipmentoversaw thebuildofgearwrote the software topro-gramtherobotandreviewedandeditedthemanuscript

DATA ACCE SSIBILIT Y

Survey sampling schedules data collection sheets extrapolationproceduresandtechnicalCRAGSspecificationsarealluploadedasonlinesupportinginformationappendixRawdatafromsurveysandCRAGSDOIhttpdoiorg105061dryadv10d218

RE SE ARCH E THIC S COMPLIANCE AND FUNDERS

ResearchabidesbytheAustraliancodeofhumanresearchandex-perimentation(NationalStatementonEthicalConductinHumanResearch of 2007) Permission to conduct research on humanparticipants was granted by the University of Tasmania SocialScience under reference H0014772mdashldquoRemote investigation ofcoastal area user group quantification and variationrdquo aswell asfromtheCSIROSocialScienceHumanResearchEthicsCommitteeunder reference 08115mdashldquoHawkesbury Region Project HumanUsePatternsrdquoFundingwasbyUTASasPartoftheIMASHonoursProject funding and through internal support from the CoastsProgram

ORCID

Tim P Lynch httporcidorg0000-0002-5346-8454

R E FE R E N C E S

Alessa L Kliskey A amp Brown G (2008) Socialndashecological hotspotsmapping A spatial approach for identifying coupled socialndasheco-logicalspaceLandscape and Urban Planning8527ndash39httpsdoiorg101016jlandurbplan200709007

ArlinghausR(2006)Overcominghumanobstaclestoconservationofrecreational fishery resources with emphasis on central EuropeEnvironmental Conservation 33 46ndash59 httpsdoiorg101017S0376892906002700

Arlinghaus R Tillner R amp Bork M (2015) Explaining participationratesinrecreationalfishingacrossindustrialisedcountriesFisheries Management and Ecology 22 45ndash55 httpsdoiorg101111fme12075

BadcockPBPatrickKSmithAMASimpsonJMPennayDRisselCEhellipRichtersJ(2016)Differencesbetweenlandlineandmobilephoneusers in sexualbehavior researchArchives of Sexual Behavior46(6)1711ndash1721

Blumberg S J amp Luke J V (2009) Reevaluating the need for con-cern regarding noncoverage bias in landline surveys American Journal of Public Health 99 1806ndash1810 httpsdoiorg102105AJPH2008152835

Bousetouane F ampMorris B (2015) Off-the-shelf CNN features forfine-grainedclassificationofvesselsinamaritimeenvironmentInGBebisRBoyleBParvinDKoracinIPavlidisRFerisTMcGrawMElendtRKopperERaganZYeampGWeber(Eds)Advances in visual computing 11th International symposium ISVC 2015 Las Vegas NV USA December 14ndash16 2015 proceedings Part II (pp379ndash388)Cham Switzerland Springer International Publishing httpsdoiorg101007978-3-319-27863-6

Cooke S J amp Cowx I G (2004) The role of recreational fish-ing in global fish crises BioScience 54 857ndash859 httpsdoiorg1016410006-3568(2004)054[0857TRORFI]20CO2

DiggleP J (1995)Time-series a biostatistical introductionOxfordUKClarendonPress

EdwardsJampSchindlerE (2017)Avideomonitoringsystemtoeval-uate ocean recreational fishing effort in Astoria OregonMarine recreational information program (pp 31) Newport OR OregonDepartmentofFishandWildlife

FarrMStoecklNampSuttonS(2014)RecreationalfishingandboatingArethedeterminantsthesameMarine Policy47126ndash137httpsdoiorg101016jmarpol201402014

ForbesETraceySampLyle JM (2009)Assessment of the 2008 rec-reational gamefish fishery of Southeast Tasmania with particular refer-ence to Southern Bluefin TunaHobartTASTasmanianAquacultureandFisheriesInstituteUniversityofTasmania

FretwellPTScofieldPampPhillipsRA(2017)Usingsuper-highres-olutionsatelliteimagerytocensusthreatenedalbatrossesIbis159481ndash490httpsdoiorg101111ibi12482

GiriKampHallK (2015)SouthAustralian recreational fishingsurveyFisheries Victoria internal report series (pp 63) Queenscliff VicFisheriesVictoria

Griffiths S P Pollock K H Lyle J M Pepperell J G TonksM L amp Sawynok W (2010) Following the chain to elu-sive anglers Fish and Fisheries 11 220ndash228 httpsdoiorg101111j1467-2979201000354x

Halpern B SWalbridge S Selkoe K A Kappel C V Micheli FDrsquoAgrosaChellipWatsonR(2008)AglobalmapofhumanimpactonmarineecosystemsScience319948ndash952httpsdoiorg101126science1149345

Hartill BW Payne GW Rush N amp Bian R (2016) Bridging thetemporal gap Continuous and cost-effective monitoring of dy-namic recreational fisheries by web cameras and creel surveysFisheries Research 183 488ndash497 httpsdoiorg101016jfishres201606002

emspensp emsp | emsp11FLYNN et aL

HenryGWampLyleJM(2003)NationalrecreationalandindigenousfishingsurveyFinal report to the fisheries research amp development cor-poration and the fisheries action program (pp 188) Canberra ACTFRDC

JonesCMampPollockKH(2012)Recreationalanglersurveymeth-odsEstimationofeffortharvestandreleasedcatchInAZaleDLParrishampTMSutton(Eds)Fisheries techniquesBethesdaMDAmericanFisheriesSociety

Kearney R E (2002) Co-management The resolution of conflict be-tween commercial and recreational fishers in Victoria AustraliaOcean amp Coastal Management45201ndash214httpsdoiorg101016S0964-5691(02)00055-8

KellerKSteffeASLowryMMurphyJJampSuthersIM(2016)Monitoringboat-basedrecreationalfishingeffortatanearshorearti-ficialreefwithashore-basedcameraFisheries Research18184ndash92httpsdoiorg101016jfishres201603025

KinlochMAMcGlennonDNicollGampPikePG(1997)Evaluationof the bus-route creel surveymethod in a largeAustralianmarinerecreationalfisheryISurveydesignFisheries Research33101ndash121httpsdoiorg101016S0165-7836(97)00068-4

LewinWCArlinghausRampMehnerT(2006)Documentedandpo-tential biological impacts of recreational fishing Insights forman-agementandconservationReviews in Fisheries Science14305ndash367httpsdoiorg10108010641260600886455

LowryMampMurphy J (2003)Monitoring the recreational gamefishfisheryoffsouth-easternAustraliaMarine and Freshwater Research54425ndash434httpsdoiorg101071MF01269

LyleJMStarkKEampTraceySR(2014)2012ndash2013 Survey of rec-reational fishing in TasmaniaHobart TAS Institute forMarine andAntarcticStudies

Lynch T P (2006) Incorporation of recreational fishing ef-fort into design of marine protected areas Incorporacioacuten delEsfuerzo de Pesca Recreativa en el Disentildeo de Aacutereas MarinasProtegidas Conservation Biology 20 1466ndash1476 httpsdoiorg101111j1523-1739200600509x

LynchTP(2008)Thedifferencebetweenspatialandtemporalvaria-tioninrecreationalfisheriesforplanningofmarineprotectedareasResponse toSteffeConservation Biology22486ndash491httpsdoiorg101111j1523-1739200800888x

LynchTP (2014)Adecadal time-seriesof recreational fishingeffortcollectedduringandafterimplementationofamultipleusemarinepark shows high inter-annual but low spatial variability Fisheries Research15185ndash90httpsdoiorg101016jfishres201309014

LynchTPAldermanRampHobdayAJ(2015)Ahigh-resolutionpan-orama camera system formonitoring colony-wide seabird nestingbehaviourMethods in Ecology and Evolution6491ndash499httpsdoiorg1011112041-210X12339

LynchTFlynnDWoodGKimJ-HParkJampHobdayAJ(2017)Metrics parameters and results from fisheries and wildlife studies with the CSIRO Ruggedized Autonomous Gigapixel Systems (CRAGS)DarwinNTAustralianMarineScienceAssociation

LynchTPWilkinsonEMellingLHamiltonRMacreadyAampFearyS(2004)ConflictandimpactsofdiversandanglersinamarineparkEnvironmental Management 33 196ndash211 httpsdoiorg101007s00267-003-3014-6

MaHOgawaTKSminkeyTRBreidtFJLesserVMOpsomerJDhellipVanVoorheesDA(2018)Pilotsurveystoimprovemonitoringofmarine recreational fisheries inHawaiʻiFisheries Research204197ndash208httpsdoiorg101016jfishres201802010

McCluskey S M amp Lewison R L (2008) Quantifying fish-ing effort A synthesis of current methods and their ap-plications Fish and Fisheries 9 188ndash200 httpsdoiorg101111j1467-2979200800283x

McGlennonDampKinlochMA(1997)Evaluationofthebus-routecreelsurveymethod in a largeAustralianmarine recreational fishery II

PilotsurveysandoptimalsamplingallocationFisheries Research3389ndash99httpsdoiorg101016S0165-7836(97)00067-2

McPheeDLeadbitterDampSkilleterGA(2002)SwallowingthebaitIs recreational fishing in Australia ecologically sustainable Pacific Conservation Biology840httpsdoiorg101071PC020040

Monz C A Pickering C M amp HadwenW L (2013) Recent ad-vances in recreation ecology and the implications of differentrelationships between recreation use and ecological impactsFrontiers in Ecology and the Environment11441ndash446httpsdoiorg101890120358

MooreAHallKGiriKTraceySPenroseLHansenShellipBrownP(2015)Developingrobustandcost-effectivemethodsforestimatingthenationalrecreationalcatchofSouthernBluefinTunainAustraliaFRDC Projectpp123

Morton A J amp Lyle J M (2004) Preliminary assessment of the rec-reational gamefish fishery in Tasmania with particular reference to Southern Bluefin Tuna Taroona TAS Tasmanian Aquaculture andFisheriesInstitute

ParnellPEDaytonPKFisherRALoarieCCampDarrowRD(2010)SpatialpatternsoffishingeffortoffSanDiegoImplicationsfor zonal management and ecosystem function Ecological Applications202203ndash2222httpsdoiorg10189009-15431

PattersonTAEvansKCarterTLampGunnJS(2008)Movementand behaviour of large southern bluefin tuna (Thunnus maccoyii)in the Australian region determined using pop-up satellite ar-chival tags Fisheries Oceanography 17 352ndash367 httpsdoiorg101111j1365-2419200800483x

PollockKH JonesCMampBrownTL (1994)Angler survey meth-ods and their applications in fisheries management Bethesda MDAmericanFisheriesSociety

PonsM Branch T AMelnychukM C JensenO P Brodziak JFromentinJMhellipHilbornR(2017)EffectsofbiologicaleconomicandmanagementfactorsontunaandbillfishstockstatusFish and Fisheries181ndash21httpsdoiorg101111faf12163

PowersSPampAnsonK (2016)Estimating recreationaleffort in theGulf of Mexico Red Snapper Fishery using boat ramp camerasReduction in federalseason lengthdoesnotproportionally reducecatchNorth American Journal of Fisheries Management 36 1156ndash1166httpsdoiorg1010800275594720161198284

Robson D amp Jones C M (1989) The theoretical basis of an ac-cess site angler survey design Biometrics 45 83ndash98 httpsdoiorg1023072532036

Rocklin D Levrel H Drogou M Herfaut J amp Veron G (2014)Combining telephone surveys and fishing catches self-reportTheFrenchseabass recreational fisheryassessmentPLoS ONE9e87271httpsdoiorg101371journalpone0087271

Smallwood C B Pollock K HWise B S Hall N G amp GaughanD J (2012) Expanding aerial-roving surveys to include counts ofshore-based recreational fishers from remotely operated camerasBenefitslimitationsandcosteffectivenessNorth American Journal of Fisheries Management321265ndash1276httpsdoiorg101080027559472012728181

Teixeira D ZischkeM T ampWebley J A C (2016) Investigatingbias in recreational fishing surveys Fishers listed in public tele-phone directories fish similarly to their unlisted counterpartsFisheries Research 181 127ndash136 httpsdoiorg101016jfishres201604012

van Poorten B T amp Brydle S (2018) Estimating fishing effortfrom remote traffic counters Opportunities and challengesFisheries Research 204 231ndash238 httpsdoiorg101016jfishres201802024

van Poorten B T Carruthers T RWardHGM ampVarkeyD A(2015)Imputingrecreationalanglingeffortfromtime-lapsecamerasusinganhierarchicalBayesianmodelFisheries Research172265ndash273httpsdoiorg101016jfishres201507032

12emsp |emsp emspensp FLYNN et aL

VenturelliPAHyderKampSkovC(2017)AnglerappsasasourceofrecreationalfisheriesdataOpportunitieschallengesandproposedstandardsFish and Fisheries18578ndash595httpsdoiorg101111faf12189

WiseBSTelferCFLaiEKMHallNGampJacksonG(2012)Long-termmonitoringofboat-basedrecreational fishing inSharkBayWesternAustraliaProvidingscientificadviceforsustainablemanagement in a World Heritage Area Marine and Freshwater Research631129ndash1141httpsdoiorg101071MF12054

WoodGLynchTPDevineCKellerKampFigueiraW(2016)High-resolutionphoto-mosaictime-seriesimageryformonitoringhumanuseofanartificialreefEcology and Evolution66963ndash6968httpsdoiorg101002ece32342

YuFRupertACCDiTizianaMNishikantaKRandySIllahNhellipJeffreyG(2011)TerapixelimagingofcosmologicalsimulationsThe Astrophysical Journal Supplement Series19718

Zischke M T Griffiths S P amp Tibbetts I R (2012) Catch and ef-fort from a specialised recreational pelagic sport fishery off

eastern Australia Fisheries Research 127ndash128 61ndash72 httpsdoiorg101016jfishres201204011

SUPPORTING INFORMATION

Additional supporting information may be found online in theSupportingInformationsectionattheendofthearticle

How to cite this articleFlynnDJHLynchTPBarrettNSWongLSCDevineCHughesDGigapixelbigdatamoviesprovidecost-effectiveseascapescaledirectmeasurementsofopen-accesscoastalhumanusesuchasrecreationalfisheriesEcol Evol 2018001ndash12 httpsdoiorg101002ece34301

Graphical AbstractThecontentsofthispagewillbeusedaspartofthegraphicalabstractofhtmlonly

Itwillnotbepublishedaspartofmainarticle

Wetrailedaremoteautonomousultra-high-resolutionphotosamplingmethodasacosteffectsolutionfordirectseascape-scaledmeasure-mentsOursequentialphotosamplingwasbatchedprocessesusingnovelTimeMachinedatahandlingsoftwaretoproduceldquobigdatardquotimeseriesmoviesfromaspatialsubsetofanoffshorefisherywhichwecomparedtoaregionalbus-routesurveyofboattrailersandinterviewswithparticipantsOurcamerasamplescloselytrackedlargerscaleregionaltrendsbutwithlaborcostsaquarterofthebus-routesurveycostsforanindividualsample25ofthebusroutes

Page 8: Gigapixel big data movies provide cost-effective seascape

8emsp |emsp emspensp FLYNN et aL

average75mincomparedto1minforenteringthebus-routedatasheetWhiledataextraction forCRAGS took longer compared tothe bus routemdashat 381 FTE daysmdashtotal monthly labor for CRAGStookonlyaquarterofthetimeandonaper-samplecomparisonwasonly25ofthebus-routelaborcost

4emsp |emspDISCUSSION

Over our autumn samplingmost trailer boat owners accessing thewaters around theTasmanPeninsula via the regional boating infra-structurewererecreationallyfishingwithfisherspredominatelytar-getingtunaTrailerboatrecreationalfishersconstitutealargetrancheof coastal users in Australia (McPhee etal 2002) and our resultssuggestedthatthisregionremainsonewhererecreationalfishing isconcentrated(MortonampLyle2004)EffortobservedbyCRAGSwasa spatial subsetof thegeneral areaeasily accessedby trailer boatsaroundtheTasmanPeninsularwhichweregionallyassessedwithourbus-routerovingsurveyAsCRAGSonlyobserved~3of thetotalarea but corresponded to ~23 of themonthly bus-route boat ef-fort theHippolyteRocksMarineConservationArea (HRMCA) is afurtherconcentrationpointwithintheregionforrecreationalfishingAsthisoffshorefishingeffortaroundtheHRMCAwasmostprobablyfocusedontotunamdashwhichaccountedfor634ofallinterviewedfish-ers activitiesmdashthe actual proportionofpeninsularwide tuna fishingthatwasobservedbyCRAGSwaspotentiallymuchhigherthan23

For developing effort metrics finding a representative site toobservemaybe important forprecise trackingof trendsThetightrankcorrelationbetweenpairedregionalbus-routeandCRAGScam-era samples suggest that effort atHRMCAwas representative ofregionaltemporaltrendsOurresultsarealsosimilartootherstud-ieswheresamplingmethodsforrecreationalfisheriesoverbroaderscaleswhencomparedtoindextrendsofeffortcollectedfrompointsourcesshowstrongcorrelations(Hartilletal2016)

CRAGSwithitsseascapescaleofdatacaptureobservesactualeffortonfishinggroundsratherthan indirectmetricsfrommove-ment past access points (Hartill etal 2016 Smallwood PollockWise Hall amp Gaughan 2012 van Poorten amp Brydle 2018) This

makesthechoiceofaccesspointstomonitorimmaterialfortrendcol-lectionandmaybeofparticularinterestforpelagicfisheriesWhilenonintuitiverecreationalfishingforwide-rangingmigratorypelagicspeciesisoftenspatiallyconcentratedintosmallareas(Lynch2006)duetotightoverlapsbetweeneaseofaccessbyfishersandfishbe-haviorrelatedtobathymetrymigrationhabitatandpreyavailability(Lynchetal2004PattersonEvansCarterampGunn2008)

Unlikespatialpatternstheintensityofcoastalrecreationaltemporaleffortcanbehighlyvariableovertime(Lynch20082014WiseTelferLaiHallampJackson2012)thoughitisgenerallythoughttofollowpre-dictablepatternsrelativetoholidayandnonholidayperiods(JonesampPollock2012)Ourresultsdemonstratedtheselargefluctuationsrang-ingfromzerouptogt400hroftrailerboateffortforadjacentdaysIncombinationwithholidayperiodsothersourcesofvariabilitymayalsooccurduetoindividualorcombinationsoffactorssuchasfishingcluborcompetitionactivitiesandweatherconditionsHigh-frequencysam-plingcanovercomethesecommontemporaldisjunctionsinobservedfishingeffortwherebychanceduetolownumbersofsampleslogis-ticallypermissiblewithtraditionalrovingoraccesspointsurveyslargeerrorsintheestimationofeffortcanbeintroduced(Hartilletal2016vanPoortenampBrydle2018vanPoortenetal2015)

With multiple daily observations abnormal episodes are morelikelytobeobservedForinstanceaweek-longperiodofintenseef-fortwasrecordedbetween15May2015and27May2015whichwasbracketedbytwoconsecutiveweekendswherefishingcompetitionswereheldontheTasmanPeninsularThenormalsinusoidpatternoflowweekdayandhighweekendeffortwasmuchlesspronouncedastheweekdaylullsineffortwerepartiallybridgedbyanglerscontinuingtofishAsbus-routedesignsusedaytypeasstratawithlowerprob-abilitiesforsamplingweekdaysandmuchfewersamplesoverallthistypeofeffectcouldeasilybemissedWhileourbus-routesamplesdidoverlapwiththetunafishingcompetitionsthiswaspurelybychanceSucheventswhichresultinlargespikesofeffortcanhavemarkedim-pactsonlocalecosystemsandharvestestimates(McPheeetal2002MonzPickeringampHadwen2013)andwithoutpriorknowledgearemorelikelytobecapturedwithmonitoringathigherfrequencies

Methodsforgatheringdataoncoastalusesuchasrecreationalfishing moved away from direct approaches and toward off-sitemethodssuchastelephoneinterviews(McCluskeyampLewison2008Pollock etal 1994) due to unsustainable costs from staffing andoperationsparticularlyduringperiodsof intenseusesuchasearlymornings weekends and public holidays which require overtimepaymentsHoweversamplingissueswithoff-sitemethodsareofpar-ticularconcernforopen-accessactivitiessuchasunlicensedfisheriesasthereisnolicensedatabaseofcontactphonenumbersonwhichtobaseasampleframeThisisparticularlysofornicherecreationalfisheriessuchaspelagicgamefishingasthefishersbecomeincreas-inglydilutewithinthegeneralpopulationandhencearehardtoac-cessparticularlyviaoff-sitemethodssuchastelephone interviews(Griffithsetal2010)

Remoteandautonomousdeploymentofsensorsmaythereforeprovideanalternativeandcost-effectivemethodforlong-termmon-itoring particularly for effort across a range of applicationsOur

TABLE 3emspFulltimeequivalent(FTE)labordaysforCRAGSandbusroutesurveystoestimatepatternsofeffort

CRAGS Bus route

Numberofsamples 169 16

Fielddays 2 16

Averagepeopleperfieldday 35 25

Field FTE days subtotals 7 40

StitchingsetupFTE 1 0

DataextractionFTE(persample) 0017 0002

Subtotal data extraction FTE 381 004

FTEdayspersample 006 250

Total FTE days 108 400

emspensp emsp | emsp9FLYNN et aL

comparisonsofmatchedpairsofdatafromtwomethodssuggestedthat theCRAGSoutputswerewell correlatedwith regionalefforttrendsOuranalysisoflabordemonstratesthattryingtounderstandthistemporalvariabilitywithsamplingviaboatrampsurveyswouldbeprohibitivelyexpensiveOurCRAGSallowsforatemporaleffortmetric derived fromactual observationsof fishing to be cheaplydeveloped with sampling frequency at subdaily scales This high-frequencysamplingapproachwasvalidatedagainstour largerbutmuchmorecostlybus-routesurveyOuranalysisoflaborrequiredtobothcollectandprocessdatashowedthatCRAGSrequiredconsid-erablylessresourcestocollectapproximatelyninetimesmorefre-quenttemporaldatathanthebus-routemethodMajorlaborsavingswereduetotheruggedizedandautonomouscapabilityofCRAGSasonlytwofieldtripswererequired(deploymentandretrievaltook7FTEdays)comparedtothebus-routersquos16fieldtrips inamonthwhichconsumed40FTEdaysandincludedanadditional780kmofinter-ramproadtravel(SupportingInformationAppendixS3)AsitisweatherproofautonomousandabletobeprogrammedwithatimedelaysetupCRAGSldquofireandforgetrdquonaturebothremovestheneedforobservertobephysicallylocatedattheregiontocountactivity(EdwardsampSchindler2017)andworkduringovertimeperiodsTheldquoblankingperiodrdquowhichstopsandstartssamplingmdashinthiscaseover-nightmdashalso reduces battery and memory use In combined thesetechnological advances have allowed long-term CRAGS deploy-mentsacross4yearsof4ndash6monthrotationsontooffshoreislandsand for several seasons in Antarctica to monitor seabirds (Lynchetal2017)Thelaborestimateforthebusroutewasconservativeasitdidnotincludeovertimeloadingforthe6weekenddaysofsam-plingorearlymorningcommencementofamsamples

Theadditional381FTEprocessingtimerequiredbythecameraapproach did not impact greatly on the overall labor budgetwhenconsideredagainstfieldworkTheuseoftheTimeMachinebatchpro-cessingsoftwaretoautomaticallyproducetheldquobigdatardquointeractivemoviewasalsoatechnologicalinnovationthatsignificantlyreducedthelaborofdatahandlingcomparedtothemanualopeningofindivid-ualGigaPansusedinpreviousstudies(Lynchetal2015Woodetal2016)IncomparativetermsattheindividualsampleleveltotracktheeffortmetriceachCRAGSsamplerequired24ofthelaborofthebus-routesurvey

Thedevelopmentandapplicationofimagerecognitionsoftware(machine learning) to detect classify and count boats automati-cally (Bousetouane amp Morris 2015) would further reduce post-processingcostsTherequiredfrequencyofsamplingshouldalsobeconsideredonacase-by-casebasisdependentontheresearchquestion as image collection can be somewhat open-ended andresult inunwieldydata sets that require subsampling tobe cost-effective(Parnelletal2010)Inourcasethesettingof~5imagesperdayappearedtoadequatelydescribetheinterdailyvariation

Whileouroneautonomouscameraprovidedastrongcorrelationwithtemporaltrendsineffortcomparedtotheregionalsurveyitdidnotprovide relativespatialdistributionsofeffortacross the fisheryThesedistributionshowevercanbepredictableandhighlyconcen-trated(Lynch20062014Parnelletal2010)anobservationthatwas

alsoshownbyourresultsandmuchsmallerlevelsofsamplingeffortwouldprobablyberequiredtoresolvevariationinspatialdistributionscomparedtowhatisrequiredtoresolvethehightemporalvariationsineffortUnderstandingfine-scalespatialdistributionsofrecreationalfishingeffortiscommonlyachievedusingaerialcounton-watersur-veysorvantagepointsurveys(Lynch2006Smallwoodetal2012)thoughotherremotemethodssuchashigh-resolutionsatelliteimag-ery(FretwellScofieldampPhillips2017)mayprovideacheapersolu-tionThesetypesofwidersurveyshavealsobeensuggestedashighlycompatiblewithpointsourcemetricssuchasremotecamerasystemstoallowforcalibrationorldquoup-ratingrdquoforeffortexpansionstoestimateoveralleffortwithinthefishery(Hartilletal2016)

Thelongdistance(~6km)betweentheHMRCAandthedeploy-ment location stretched the resolution limit of the current CRAGSsetup aswewere unable to distinguish party size or activity typepurelyusing thecamerasystemThis isunlikepreviousCRAGSde-ploymentwhichhadahighsuccessrateofidentifyingpartysizeac-tivity typeata shorter rangeofapproximately19km (Woodetal2016) Thus ground-truthing information about the fisherywas re-quiredforthesystemtoworkeffectivelyAutonomouscamerasystemalsodonotcollectcreel(catch)databutbyallowingforcost-effectivecollectionofthehighlyvariabletemporaleffortdatatheymayallowforhumanresourcestobespentelsewhereduringsurveyperiods(egtoundertakeboatrampinterviews)(vanPoortenampBrydle2018)

Our studywas a small-scale trial of new technology andwhilepreliminary results are encouraging validated seasonal effort esti-mationsobservedviaCRAGSwereoutsidethescopeofthisprojectOur comparisons would hence have been improved by extendingCRAGSdeploymentsandwithmorecomparisonstoregionalsurveysSimilarly usingmultiple CRAGS simultaneously would have bettertested for differences in subregional patterns of temporal trendsorconcentrationof infishingeffortfor instanceboatsaroundthenearbyTasmanIsland

Insummarytheprogrammablesamplinganddataextractionof-feredbyhigh-resolutionautonomouscamerasystemscanallowforfine-scalelong-termmonitoringoffishingefforttrendsorpotentiallymanyothertypesofactivitiesatseascapescalesMorewidespreadadoptionofhigh-frequencysamplingapproachesforeffortestima-tionwouldalsohelpavoidcollectionofmisrepresentativedatabyidentifyingeffortanomalies(McCluskeyampLewison2008)suchastheweekdaycontinuationeffectofraisedeffortbetweenadjacentweekendfishingcompetitionByimprovingtheinformationdensityovertraditionalsurveymethodsautonomouscamerasystemsandotherremotesystemsmayprovideanewwaveofoptimizationfordirectmeasurementofcoastalhumanusesuchasrecreationalfish-eriesalthoughcomplimentarymethodsarestillrequiredtodeter-minespatialdistributionstargetspeciesandharvest

ACKNOWLEDG MENTS

We thankMsGeorginaWoodMs SaraAdkinsMsBelleMonkMrRSherringtonandMsPaigeKellyforconstructiveinsighten-thusiasmandassistanceduringfieldworkWealsothankMrPhillip

10emsp |emsp emspensp FLYNN et aL

CulversonMrStuartDudgeonandMrMattStapenellKarenEvensprovidedhelpfulcommentsonaninternaldraftGreatestthankstothemany recreational anglerswhowillingly shared information tosurveyclerkswithouttheirinvolvementthisprojectwouldnothavecometofruitionThreeanonymousreviewersprovidedcommentsthatimprovedthemanuscript

CONFLIC T OF INTERE S TS

Theauthorsdeclarenoconflictofinterests

AUTHOR CONTRIBUTIONS

MrFlynnhelpeddesignthestudyledthefieldworkundertookmostoftheanalysisandhelpwritethemanuscriptDrLynchhelpdesignthestudyassisted inthefieldworkprovidedadviceontheanaly-sisanddidsomeanalysisconstructedfiguresandafterMrFlynncontributedthemosttothedraftingofthemanuscriptDrBarretthelpeddesign the study assisted in some fieldwork provided ad-viceonanalysisandeditedthemanuscriptMrWonghelpedexten-sivelywithfieldworkassistedwithanalysisconstructedfiguresandhelpedwritethemanuscriptMsDevineassistedwithfieldworkandconfigurationoftheequipmentincludingtheapplicationofthebigdatamoviesoftwareandhandlingofphotographsandalsohelpedwritethemanuscriptMrHughesdevelopedtheelectronicsfortheequipmentoversaw thebuildofgearwrote the software topro-gramtherobotandreviewedandeditedthemanuscript

DATA ACCE SSIBILIT Y

Survey sampling schedules data collection sheets extrapolationproceduresandtechnicalCRAGSspecificationsarealluploadedasonlinesupportinginformationappendixRawdatafromsurveysandCRAGSDOIhttpdoiorg105061dryadv10d218

RE SE ARCH E THIC S COMPLIANCE AND FUNDERS

ResearchabidesbytheAustraliancodeofhumanresearchandex-perimentation(NationalStatementonEthicalConductinHumanResearch of 2007) Permission to conduct research on humanparticipants was granted by the University of Tasmania SocialScience under reference H0014772mdashldquoRemote investigation ofcoastal area user group quantification and variationrdquo aswell asfromtheCSIROSocialScienceHumanResearchEthicsCommitteeunder reference 08115mdashldquoHawkesbury Region Project HumanUsePatternsrdquoFundingwasbyUTASasPartoftheIMASHonoursProject funding and through internal support from the CoastsProgram

ORCID

Tim P Lynch httporcidorg0000-0002-5346-8454

R E FE R E N C E S

Alessa L Kliskey A amp Brown G (2008) Socialndashecological hotspotsmapping A spatial approach for identifying coupled socialndasheco-logicalspaceLandscape and Urban Planning8527ndash39httpsdoiorg101016jlandurbplan200709007

ArlinghausR(2006)Overcominghumanobstaclestoconservationofrecreational fishery resources with emphasis on central EuropeEnvironmental Conservation 33 46ndash59 httpsdoiorg101017S0376892906002700

Arlinghaus R Tillner R amp Bork M (2015) Explaining participationratesinrecreationalfishingacrossindustrialisedcountriesFisheries Management and Ecology 22 45ndash55 httpsdoiorg101111fme12075

BadcockPBPatrickKSmithAMASimpsonJMPennayDRisselCEhellipRichtersJ(2016)Differencesbetweenlandlineandmobilephoneusers in sexualbehavior researchArchives of Sexual Behavior46(6)1711ndash1721

Blumberg S J amp Luke J V (2009) Reevaluating the need for con-cern regarding noncoverage bias in landline surveys American Journal of Public Health 99 1806ndash1810 httpsdoiorg102105AJPH2008152835

Bousetouane F ampMorris B (2015) Off-the-shelf CNN features forfine-grainedclassificationofvesselsinamaritimeenvironmentInGBebisRBoyleBParvinDKoracinIPavlidisRFerisTMcGrawMElendtRKopperERaganZYeampGWeber(Eds)Advances in visual computing 11th International symposium ISVC 2015 Las Vegas NV USA December 14ndash16 2015 proceedings Part II (pp379ndash388)Cham Switzerland Springer International Publishing httpsdoiorg101007978-3-319-27863-6

Cooke S J amp Cowx I G (2004) The role of recreational fish-ing in global fish crises BioScience 54 857ndash859 httpsdoiorg1016410006-3568(2004)054[0857TRORFI]20CO2

DiggleP J (1995)Time-series a biostatistical introductionOxfordUKClarendonPress

EdwardsJampSchindlerE (2017)Avideomonitoringsystemtoeval-uate ocean recreational fishing effort in Astoria OregonMarine recreational information program (pp 31) Newport OR OregonDepartmentofFishandWildlife

FarrMStoecklNampSuttonS(2014)RecreationalfishingandboatingArethedeterminantsthesameMarine Policy47126ndash137httpsdoiorg101016jmarpol201402014

ForbesETraceySampLyle JM (2009)Assessment of the 2008 rec-reational gamefish fishery of Southeast Tasmania with particular refer-ence to Southern Bluefin TunaHobartTASTasmanianAquacultureandFisheriesInstituteUniversityofTasmania

FretwellPTScofieldPampPhillipsRA(2017)Usingsuper-highres-olutionsatelliteimagerytocensusthreatenedalbatrossesIbis159481ndash490httpsdoiorg101111ibi12482

GiriKampHallK (2015)SouthAustralian recreational fishingsurveyFisheries Victoria internal report series (pp 63) Queenscliff VicFisheriesVictoria

Griffiths S P Pollock K H Lyle J M Pepperell J G TonksM L amp Sawynok W (2010) Following the chain to elu-sive anglers Fish and Fisheries 11 220ndash228 httpsdoiorg101111j1467-2979201000354x

Halpern B SWalbridge S Selkoe K A Kappel C V Micheli FDrsquoAgrosaChellipWatsonR(2008)AglobalmapofhumanimpactonmarineecosystemsScience319948ndash952httpsdoiorg101126science1149345

Hartill BW Payne GW Rush N amp Bian R (2016) Bridging thetemporal gap Continuous and cost-effective monitoring of dy-namic recreational fisheries by web cameras and creel surveysFisheries Research 183 488ndash497 httpsdoiorg101016jfishres201606002

emspensp emsp | emsp11FLYNN et aL

HenryGWampLyleJM(2003)NationalrecreationalandindigenousfishingsurveyFinal report to the fisheries research amp development cor-poration and the fisheries action program (pp 188) Canberra ACTFRDC

JonesCMampPollockKH(2012)Recreationalanglersurveymeth-odsEstimationofeffortharvestandreleasedcatchInAZaleDLParrishampTMSutton(Eds)Fisheries techniquesBethesdaMDAmericanFisheriesSociety

Kearney R E (2002) Co-management The resolution of conflict be-tween commercial and recreational fishers in Victoria AustraliaOcean amp Coastal Management45201ndash214httpsdoiorg101016S0964-5691(02)00055-8

KellerKSteffeASLowryMMurphyJJampSuthersIM(2016)Monitoringboat-basedrecreationalfishingeffortatanearshorearti-ficialreefwithashore-basedcameraFisheries Research18184ndash92httpsdoiorg101016jfishres201603025

KinlochMAMcGlennonDNicollGampPikePG(1997)Evaluationof the bus-route creel surveymethod in a largeAustralianmarinerecreationalfisheryISurveydesignFisheries Research33101ndash121httpsdoiorg101016S0165-7836(97)00068-4

LewinWCArlinghausRampMehnerT(2006)Documentedandpo-tential biological impacts of recreational fishing Insights forman-agementandconservationReviews in Fisheries Science14305ndash367httpsdoiorg10108010641260600886455

LowryMampMurphy J (2003)Monitoring the recreational gamefishfisheryoffsouth-easternAustraliaMarine and Freshwater Research54425ndash434httpsdoiorg101071MF01269

LyleJMStarkKEampTraceySR(2014)2012ndash2013 Survey of rec-reational fishing in TasmaniaHobart TAS Institute forMarine andAntarcticStudies

Lynch T P (2006) Incorporation of recreational fishing ef-fort into design of marine protected areas Incorporacioacuten delEsfuerzo de Pesca Recreativa en el Disentildeo de Aacutereas MarinasProtegidas Conservation Biology 20 1466ndash1476 httpsdoiorg101111j1523-1739200600509x

LynchTP(2008)Thedifferencebetweenspatialandtemporalvaria-tioninrecreationalfisheriesforplanningofmarineprotectedareasResponse toSteffeConservation Biology22486ndash491httpsdoiorg101111j1523-1739200800888x

LynchTP (2014)Adecadal time-seriesof recreational fishingeffortcollectedduringandafterimplementationofamultipleusemarinepark shows high inter-annual but low spatial variability Fisheries Research15185ndash90httpsdoiorg101016jfishres201309014

LynchTPAldermanRampHobdayAJ(2015)Ahigh-resolutionpan-orama camera system formonitoring colony-wide seabird nestingbehaviourMethods in Ecology and Evolution6491ndash499httpsdoiorg1011112041-210X12339

LynchTFlynnDWoodGKimJ-HParkJampHobdayAJ(2017)Metrics parameters and results from fisheries and wildlife studies with the CSIRO Ruggedized Autonomous Gigapixel Systems (CRAGS)DarwinNTAustralianMarineScienceAssociation

LynchTPWilkinsonEMellingLHamiltonRMacreadyAampFearyS(2004)ConflictandimpactsofdiversandanglersinamarineparkEnvironmental Management 33 196ndash211 httpsdoiorg101007s00267-003-3014-6

MaHOgawaTKSminkeyTRBreidtFJLesserVMOpsomerJDhellipVanVoorheesDA(2018)Pilotsurveystoimprovemonitoringofmarine recreational fisheries inHawaiʻiFisheries Research204197ndash208httpsdoiorg101016jfishres201802010

McCluskey S M amp Lewison R L (2008) Quantifying fish-ing effort A synthesis of current methods and their ap-plications Fish and Fisheries 9 188ndash200 httpsdoiorg101111j1467-2979200800283x

McGlennonDampKinlochMA(1997)Evaluationofthebus-routecreelsurveymethod in a largeAustralianmarine recreational fishery II

PilotsurveysandoptimalsamplingallocationFisheries Research3389ndash99httpsdoiorg101016S0165-7836(97)00067-2

McPheeDLeadbitterDampSkilleterGA(2002)SwallowingthebaitIs recreational fishing in Australia ecologically sustainable Pacific Conservation Biology840httpsdoiorg101071PC020040

Monz C A Pickering C M amp HadwenW L (2013) Recent ad-vances in recreation ecology and the implications of differentrelationships between recreation use and ecological impactsFrontiers in Ecology and the Environment11441ndash446httpsdoiorg101890120358

MooreAHallKGiriKTraceySPenroseLHansenShellipBrownP(2015)Developingrobustandcost-effectivemethodsforestimatingthenationalrecreationalcatchofSouthernBluefinTunainAustraliaFRDC Projectpp123

Morton A J amp Lyle J M (2004) Preliminary assessment of the rec-reational gamefish fishery in Tasmania with particular reference to Southern Bluefin Tuna Taroona TAS Tasmanian Aquaculture andFisheriesInstitute

ParnellPEDaytonPKFisherRALoarieCCampDarrowRD(2010)SpatialpatternsoffishingeffortoffSanDiegoImplicationsfor zonal management and ecosystem function Ecological Applications202203ndash2222httpsdoiorg10189009-15431

PattersonTAEvansKCarterTLampGunnJS(2008)Movementand behaviour of large southern bluefin tuna (Thunnus maccoyii)in the Australian region determined using pop-up satellite ar-chival tags Fisheries Oceanography 17 352ndash367 httpsdoiorg101111j1365-2419200800483x

PollockKH JonesCMampBrownTL (1994)Angler survey meth-ods and their applications in fisheries management Bethesda MDAmericanFisheriesSociety

PonsM Branch T AMelnychukM C JensenO P Brodziak JFromentinJMhellipHilbornR(2017)EffectsofbiologicaleconomicandmanagementfactorsontunaandbillfishstockstatusFish and Fisheries181ndash21httpsdoiorg101111faf12163

PowersSPampAnsonK (2016)Estimating recreationaleffort in theGulf of Mexico Red Snapper Fishery using boat ramp camerasReduction in federalseason lengthdoesnotproportionally reducecatchNorth American Journal of Fisheries Management 36 1156ndash1166httpsdoiorg1010800275594720161198284

Robson D amp Jones C M (1989) The theoretical basis of an ac-cess site angler survey design Biometrics 45 83ndash98 httpsdoiorg1023072532036

Rocklin D Levrel H Drogou M Herfaut J amp Veron G (2014)Combining telephone surveys and fishing catches self-reportTheFrenchseabass recreational fisheryassessmentPLoS ONE9e87271httpsdoiorg101371journalpone0087271

Smallwood C B Pollock K HWise B S Hall N G amp GaughanD J (2012) Expanding aerial-roving surveys to include counts ofshore-based recreational fishers from remotely operated camerasBenefitslimitationsandcosteffectivenessNorth American Journal of Fisheries Management321265ndash1276httpsdoiorg101080027559472012728181

Teixeira D ZischkeM T ampWebley J A C (2016) Investigatingbias in recreational fishing surveys Fishers listed in public tele-phone directories fish similarly to their unlisted counterpartsFisheries Research 181 127ndash136 httpsdoiorg101016jfishres201604012

van Poorten B T amp Brydle S (2018) Estimating fishing effortfrom remote traffic counters Opportunities and challengesFisheries Research 204 231ndash238 httpsdoiorg101016jfishres201802024

van Poorten B T Carruthers T RWardHGM ampVarkeyD A(2015)Imputingrecreationalanglingeffortfromtime-lapsecamerasusinganhierarchicalBayesianmodelFisheries Research172265ndash273httpsdoiorg101016jfishres201507032

12emsp |emsp emspensp FLYNN et aL

VenturelliPAHyderKampSkovC(2017)AnglerappsasasourceofrecreationalfisheriesdataOpportunitieschallengesandproposedstandardsFish and Fisheries18578ndash595httpsdoiorg101111faf12189

WiseBSTelferCFLaiEKMHallNGampJacksonG(2012)Long-termmonitoringofboat-basedrecreational fishing inSharkBayWesternAustraliaProvidingscientificadviceforsustainablemanagement in a World Heritage Area Marine and Freshwater Research631129ndash1141httpsdoiorg101071MF12054

WoodGLynchTPDevineCKellerKampFigueiraW(2016)High-resolutionphoto-mosaictime-seriesimageryformonitoringhumanuseofanartificialreefEcology and Evolution66963ndash6968httpsdoiorg101002ece32342

YuFRupertACCDiTizianaMNishikantaKRandySIllahNhellipJeffreyG(2011)TerapixelimagingofcosmologicalsimulationsThe Astrophysical Journal Supplement Series19718

Zischke M T Griffiths S P amp Tibbetts I R (2012) Catch and ef-fort from a specialised recreational pelagic sport fishery off

eastern Australia Fisheries Research 127ndash128 61ndash72 httpsdoiorg101016jfishres201204011

SUPPORTING INFORMATION

Additional supporting information may be found online in theSupportingInformationsectionattheendofthearticle

How to cite this articleFlynnDJHLynchTPBarrettNSWongLSCDevineCHughesDGigapixelbigdatamoviesprovidecost-effectiveseascapescaledirectmeasurementsofopen-accesscoastalhumanusesuchasrecreationalfisheriesEcol Evol 2018001ndash12 httpsdoiorg101002ece34301

Graphical AbstractThecontentsofthispagewillbeusedaspartofthegraphicalabstractofhtmlonly

Itwillnotbepublishedaspartofmainarticle

Wetrailedaremoteautonomousultra-high-resolutionphotosamplingmethodasacosteffectsolutionfordirectseascape-scaledmeasure-mentsOursequentialphotosamplingwasbatchedprocessesusingnovelTimeMachinedatahandlingsoftwaretoproduceldquobigdatardquotimeseriesmoviesfromaspatialsubsetofanoffshorefisherywhichwecomparedtoaregionalbus-routesurveyofboattrailersandinterviewswithparticipantsOurcamerasamplescloselytrackedlargerscaleregionaltrendsbutwithlaborcostsaquarterofthebus-routesurveycostsforanindividualsample25ofthebusroutes

Page 9: Gigapixel big data movies provide cost-effective seascape

emspensp emsp | emsp9FLYNN et aL

comparisonsofmatchedpairsofdatafromtwomethodssuggestedthat theCRAGSoutputswerewell correlatedwith regionalefforttrendsOuranalysisoflabordemonstratesthattryingtounderstandthistemporalvariabilitywithsamplingviaboatrampsurveyswouldbeprohibitivelyexpensiveOurCRAGSallowsforatemporaleffortmetric derived fromactual observationsof fishing to be cheaplydeveloped with sampling frequency at subdaily scales This high-frequencysamplingapproachwasvalidatedagainstour largerbutmuchmorecostlybus-routesurveyOuranalysisoflaborrequiredtobothcollectandprocessdatashowedthatCRAGSrequiredconsid-erablylessresourcestocollectapproximatelyninetimesmorefre-quenttemporaldatathanthebus-routemethodMajorlaborsavingswereduetotheruggedizedandautonomouscapabilityofCRAGSasonlytwofieldtripswererequired(deploymentandretrievaltook7FTEdays)comparedtothebus-routersquos16fieldtrips inamonthwhichconsumed40FTEdaysandincludedanadditional780kmofinter-ramproadtravel(SupportingInformationAppendixS3)AsitisweatherproofautonomousandabletobeprogrammedwithatimedelaysetupCRAGSldquofireandforgetrdquonaturebothremovestheneedforobservertobephysicallylocatedattheregiontocountactivity(EdwardsampSchindler2017)andworkduringovertimeperiodsTheldquoblankingperiodrdquowhichstopsandstartssamplingmdashinthiscaseover-nightmdashalso reduces battery and memory use In combined thesetechnological advances have allowed long-term CRAGS deploy-mentsacross4yearsof4ndash6monthrotationsontooffshoreislandsand for several seasons in Antarctica to monitor seabirds (Lynchetal2017)Thelaborestimateforthebusroutewasconservativeasitdidnotincludeovertimeloadingforthe6weekenddaysofsam-plingorearlymorningcommencementofamsamples

Theadditional381FTEprocessingtimerequiredbythecameraapproach did not impact greatly on the overall labor budgetwhenconsideredagainstfieldworkTheuseoftheTimeMachinebatchpro-cessingsoftwaretoautomaticallyproducetheldquobigdatardquointeractivemoviewasalsoatechnologicalinnovationthatsignificantlyreducedthelaborofdatahandlingcomparedtothemanualopeningofindivid-ualGigaPansusedinpreviousstudies(Lynchetal2015Woodetal2016)IncomparativetermsattheindividualsampleleveltotracktheeffortmetriceachCRAGSsamplerequired24ofthelaborofthebus-routesurvey

Thedevelopmentandapplicationofimagerecognitionsoftware(machine learning) to detect classify and count boats automati-cally (Bousetouane amp Morris 2015) would further reduce post-processingcostsTherequiredfrequencyofsamplingshouldalsobeconsideredonacase-by-casebasisdependentontheresearchquestion as image collection can be somewhat open-ended andresult inunwieldydata sets that require subsampling tobe cost-effective(Parnelletal2010)Inourcasethesettingof~5imagesperdayappearedtoadequatelydescribetheinterdailyvariation

Whileouroneautonomouscameraprovidedastrongcorrelationwithtemporaltrendsineffortcomparedtotheregionalsurveyitdidnotprovide relativespatialdistributionsofeffortacross the fisheryThesedistributionshowevercanbepredictableandhighlyconcen-trated(Lynch20062014Parnelletal2010)anobservationthatwas

alsoshownbyourresultsandmuchsmallerlevelsofsamplingeffortwouldprobablyberequiredtoresolvevariationinspatialdistributionscomparedtowhatisrequiredtoresolvethehightemporalvariationsineffortUnderstandingfine-scalespatialdistributionsofrecreationalfishingeffortiscommonlyachievedusingaerialcounton-watersur-veysorvantagepointsurveys(Lynch2006Smallwoodetal2012)thoughotherremotemethodssuchashigh-resolutionsatelliteimag-ery(FretwellScofieldampPhillips2017)mayprovideacheapersolu-tionThesetypesofwidersurveyshavealsobeensuggestedashighlycompatiblewithpointsourcemetricssuchasremotecamerasystemstoallowforcalibrationorldquoup-ratingrdquoforeffortexpansionstoestimateoveralleffortwithinthefishery(Hartilletal2016)

Thelongdistance(~6km)betweentheHMRCAandthedeploy-ment location stretched the resolution limit of the current CRAGSsetup aswewere unable to distinguish party size or activity typepurelyusing thecamerasystemThis isunlikepreviousCRAGSde-ploymentwhichhadahighsuccessrateofidentifyingpartysizeac-tivity typeata shorter rangeofapproximately19km (Woodetal2016) Thus ground-truthing information about the fisherywas re-quiredforthesystemtoworkeffectivelyAutonomouscamerasystemalsodonotcollectcreel(catch)databutbyallowingforcost-effectivecollectionofthehighlyvariabletemporaleffortdatatheymayallowforhumanresourcestobespentelsewhereduringsurveyperiods(egtoundertakeboatrampinterviews)(vanPoortenampBrydle2018)

Our studywas a small-scale trial of new technology andwhilepreliminary results are encouraging validated seasonal effort esti-mationsobservedviaCRAGSwereoutsidethescopeofthisprojectOur comparisons would hence have been improved by extendingCRAGSdeploymentsandwithmorecomparisonstoregionalsurveysSimilarly usingmultiple CRAGS simultaneously would have bettertested for differences in subregional patterns of temporal trendsorconcentrationof infishingeffortfor instanceboatsaroundthenearbyTasmanIsland

Insummarytheprogrammablesamplinganddataextractionof-feredbyhigh-resolutionautonomouscamerasystemscanallowforfine-scalelong-termmonitoringoffishingefforttrendsorpotentiallymanyothertypesofactivitiesatseascapescalesMorewidespreadadoptionofhigh-frequencysamplingapproachesforeffortestima-tionwouldalsohelpavoidcollectionofmisrepresentativedatabyidentifyingeffortanomalies(McCluskeyampLewison2008)suchastheweekdaycontinuationeffectofraisedeffortbetweenadjacentweekendfishingcompetitionByimprovingtheinformationdensityovertraditionalsurveymethodsautonomouscamerasystemsandotherremotesystemsmayprovideanewwaveofoptimizationfordirectmeasurementofcoastalhumanusesuchasrecreationalfish-eriesalthoughcomplimentarymethodsarestillrequiredtodeter-minespatialdistributionstargetspeciesandharvest

ACKNOWLEDG MENTS

We thankMsGeorginaWoodMs SaraAdkinsMsBelleMonkMrRSherringtonandMsPaigeKellyforconstructiveinsighten-thusiasmandassistanceduringfieldworkWealsothankMrPhillip

10emsp |emsp emspensp FLYNN et aL

CulversonMrStuartDudgeonandMrMattStapenellKarenEvensprovidedhelpfulcommentsonaninternaldraftGreatestthankstothemany recreational anglerswhowillingly shared information tosurveyclerkswithouttheirinvolvementthisprojectwouldnothavecometofruitionThreeanonymousreviewersprovidedcommentsthatimprovedthemanuscript

CONFLIC T OF INTERE S TS

Theauthorsdeclarenoconflictofinterests

AUTHOR CONTRIBUTIONS

MrFlynnhelpeddesignthestudyledthefieldworkundertookmostoftheanalysisandhelpwritethemanuscriptDrLynchhelpdesignthestudyassisted inthefieldworkprovidedadviceontheanaly-sisanddidsomeanalysisconstructedfiguresandafterMrFlynncontributedthemosttothedraftingofthemanuscriptDrBarretthelpeddesign the study assisted in some fieldwork provided ad-viceonanalysisandeditedthemanuscriptMrWonghelpedexten-sivelywithfieldworkassistedwithanalysisconstructedfiguresandhelpedwritethemanuscriptMsDevineassistedwithfieldworkandconfigurationoftheequipmentincludingtheapplicationofthebigdatamoviesoftwareandhandlingofphotographsandalsohelpedwritethemanuscriptMrHughesdevelopedtheelectronicsfortheequipmentoversaw thebuildofgearwrote the software topro-gramtherobotandreviewedandeditedthemanuscript

DATA ACCE SSIBILIT Y

Survey sampling schedules data collection sheets extrapolationproceduresandtechnicalCRAGSspecificationsarealluploadedasonlinesupportinginformationappendixRawdatafromsurveysandCRAGSDOIhttpdoiorg105061dryadv10d218

RE SE ARCH E THIC S COMPLIANCE AND FUNDERS

ResearchabidesbytheAustraliancodeofhumanresearchandex-perimentation(NationalStatementonEthicalConductinHumanResearch of 2007) Permission to conduct research on humanparticipants was granted by the University of Tasmania SocialScience under reference H0014772mdashldquoRemote investigation ofcoastal area user group quantification and variationrdquo aswell asfromtheCSIROSocialScienceHumanResearchEthicsCommitteeunder reference 08115mdashldquoHawkesbury Region Project HumanUsePatternsrdquoFundingwasbyUTASasPartoftheIMASHonoursProject funding and through internal support from the CoastsProgram

ORCID

Tim P Lynch httporcidorg0000-0002-5346-8454

R E FE R E N C E S

Alessa L Kliskey A amp Brown G (2008) Socialndashecological hotspotsmapping A spatial approach for identifying coupled socialndasheco-logicalspaceLandscape and Urban Planning8527ndash39httpsdoiorg101016jlandurbplan200709007

ArlinghausR(2006)Overcominghumanobstaclestoconservationofrecreational fishery resources with emphasis on central EuropeEnvironmental Conservation 33 46ndash59 httpsdoiorg101017S0376892906002700

Arlinghaus R Tillner R amp Bork M (2015) Explaining participationratesinrecreationalfishingacrossindustrialisedcountriesFisheries Management and Ecology 22 45ndash55 httpsdoiorg101111fme12075

BadcockPBPatrickKSmithAMASimpsonJMPennayDRisselCEhellipRichtersJ(2016)Differencesbetweenlandlineandmobilephoneusers in sexualbehavior researchArchives of Sexual Behavior46(6)1711ndash1721

Blumberg S J amp Luke J V (2009) Reevaluating the need for con-cern regarding noncoverage bias in landline surveys American Journal of Public Health 99 1806ndash1810 httpsdoiorg102105AJPH2008152835

Bousetouane F ampMorris B (2015) Off-the-shelf CNN features forfine-grainedclassificationofvesselsinamaritimeenvironmentInGBebisRBoyleBParvinDKoracinIPavlidisRFerisTMcGrawMElendtRKopperERaganZYeampGWeber(Eds)Advances in visual computing 11th International symposium ISVC 2015 Las Vegas NV USA December 14ndash16 2015 proceedings Part II (pp379ndash388)Cham Switzerland Springer International Publishing httpsdoiorg101007978-3-319-27863-6

Cooke S J amp Cowx I G (2004) The role of recreational fish-ing in global fish crises BioScience 54 857ndash859 httpsdoiorg1016410006-3568(2004)054[0857TRORFI]20CO2

DiggleP J (1995)Time-series a biostatistical introductionOxfordUKClarendonPress

EdwardsJampSchindlerE (2017)Avideomonitoringsystemtoeval-uate ocean recreational fishing effort in Astoria OregonMarine recreational information program (pp 31) Newport OR OregonDepartmentofFishandWildlife

FarrMStoecklNampSuttonS(2014)RecreationalfishingandboatingArethedeterminantsthesameMarine Policy47126ndash137httpsdoiorg101016jmarpol201402014

ForbesETraceySampLyle JM (2009)Assessment of the 2008 rec-reational gamefish fishery of Southeast Tasmania with particular refer-ence to Southern Bluefin TunaHobartTASTasmanianAquacultureandFisheriesInstituteUniversityofTasmania

FretwellPTScofieldPampPhillipsRA(2017)Usingsuper-highres-olutionsatelliteimagerytocensusthreatenedalbatrossesIbis159481ndash490httpsdoiorg101111ibi12482

GiriKampHallK (2015)SouthAustralian recreational fishingsurveyFisheries Victoria internal report series (pp 63) Queenscliff VicFisheriesVictoria

Griffiths S P Pollock K H Lyle J M Pepperell J G TonksM L amp Sawynok W (2010) Following the chain to elu-sive anglers Fish and Fisheries 11 220ndash228 httpsdoiorg101111j1467-2979201000354x

Halpern B SWalbridge S Selkoe K A Kappel C V Micheli FDrsquoAgrosaChellipWatsonR(2008)AglobalmapofhumanimpactonmarineecosystemsScience319948ndash952httpsdoiorg101126science1149345

Hartill BW Payne GW Rush N amp Bian R (2016) Bridging thetemporal gap Continuous and cost-effective monitoring of dy-namic recreational fisheries by web cameras and creel surveysFisheries Research 183 488ndash497 httpsdoiorg101016jfishres201606002

emspensp emsp | emsp11FLYNN et aL

HenryGWampLyleJM(2003)NationalrecreationalandindigenousfishingsurveyFinal report to the fisheries research amp development cor-poration and the fisheries action program (pp 188) Canberra ACTFRDC

JonesCMampPollockKH(2012)Recreationalanglersurveymeth-odsEstimationofeffortharvestandreleasedcatchInAZaleDLParrishampTMSutton(Eds)Fisheries techniquesBethesdaMDAmericanFisheriesSociety

Kearney R E (2002) Co-management The resolution of conflict be-tween commercial and recreational fishers in Victoria AustraliaOcean amp Coastal Management45201ndash214httpsdoiorg101016S0964-5691(02)00055-8

KellerKSteffeASLowryMMurphyJJampSuthersIM(2016)Monitoringboat-basedrecreationalfishingeffortatanearshorearti-ficialreefwithashore-basedcameraFisheries Research18184ndash92httpsdoiorg101016jfishres201603025

KinlochMAMcGlennonDNicollGampPikePG(1997)Evaluationof the bus-route creel surveymethod in a largeAustralianmarinerecreationalfisheryISurveydesignFisheries Research33101ndash121httpsdoiorg101016S0165-7836(97)00068-4

LewinWCArlinghausRampMehnerT(2006)Documentedandpo-tential biological impacts of recreational fishing Insights forman-agementandconservationReviews in Fisheries Science14305ndash367httpsdoiorg10108010641260600886455

LowryMampMurphy J (2003)Monitoring the recreational gamefishfisheryoffsouth-easternAustraliaMarine and Freshwater Research54425ndash434httpsdoiorg101071MF01269

LyleJMStarkKEampTraceySR(2014)2012ndash2013 Survey of rec-reational fishing in TasmaniaHobart TAS Institute forMarine andAntarcticStudies

Lynch T P (2006) Incorporation of recreational fishing ef-fort into design of marine protected areas Incorporacioacuten delEsfuerzo de Pesca Recreativa en el Disentildeo de Aacutereas MarinasProtegidas Conservation Biology 20 1466ndash1476 httpsdoiorg101111j1523-1739200600509x

LynchTP(2008)Thedifferencebetweenspatialandtemporalvaria-tioninrecreationalfisheriesforplanningofmarineprotectedareasResponse toSteffeConservation Biology22486ndash491httpsdoiorg101111j1523-1739200800888x

LynchTP (2014)Adecadal time-seriesof recreational fishingeffortcollectedduringandafterimplementationofamultipleusemarinepark shows high inter-annual but low spatial variability Fisheries Research15185ndash90httpsdoiorg101016jfishres201309014

LynchTPAldermanRampHobdayAJ(2015)Ahigh-resolutionpan-orama camera system formonitoring colony-wide seabird nestingbehaviourMethods in Ecology and Evolution6491ndash499httpsdoiorg1011112041-210X12339

LynchTFlynnDWoodGKimJ-HParkJampHobdayAJ(2017)Metrics parameters and results from fisheries and wildlife studies with the CSIRO Ruggedized Autonomous Gigapixel Systems (CRAGS)DarwinNTAustralianMarineScienceAssociation

LynchTPWilkinsonEMellingLHamiltonRMacreadyAampFearyS(2004)ConflictandimpactsofdiversandanglersinamarineparkEnvironmental Management 33 196ndash211 httpsdoiorg101007s00267-003-3014-6

MaHOgawaTKSminkeyTRBreidtFJLesserVMOpsomerJDhellipVanVoorheesDA(2018)Pilotsurveystoimprovemonitoringofmarine recreational fisheries inHawaiʻiFisheries Research204197ndash208httpsdoiorg101016jfishres201802010

McCluskey S M amp Lewison R L (2008) Quantifying fish-ing effort A synthesis of current methods and their ap-plications Fish and Fisheries 9 188ndash200 httpsdoiorg101111j1467-2979200800283x

McGlennonDampKinlochMA(1997)Evaluationofthebus-routecreelsurveymethod in a largeAustralianmarine recreational fishery II

PilotsurveysandoptimalsamplingallocationFisheries Research3389ndash99httpsdoiorg101016S0165-7836(97)00067-2

McPheeDLeadbitterDampSkilleterGA(2002)SwallowingthebaitIs recreational fishing in Australia ecologically sustainable Pacific Conservation Biology840httpsdoiorg101071PC020040

Monz C A Pickering C M amp HadwenW L (2013) Recent ad-vances in recreation ecology and the implications of differentrelationships between recreation use and ecological impactsFrontiers in Ecology and the Environment11441ndash446httpsdoiorg101890120358

MooreAHallKGiriKTraceySPenroseLHansenShellipBrownP(2015)Developingrobustandcost-effectivemethodsforestimatingthenationalrecreationalcatchofSouthernBluefinTunainAustraliaFRDC Projectpp123

Morton A J amp Lyle J M (2004) Preliminary assessment of the rec-reational gamefish fishery in Tasmania with particular reference to Southern Bluefin Tuna Taroona TAS Tasmanian Aquaculture andFisheriesInstitute

ParnellPEDaytonPKFisherRALoarieCCampDarrowRD(2010)SpatialpatternsoffishingeffortoffSanDiegoImplicationsfor zonal management and ecosystem function Ecological Applications202203ndash2222httpsdoiorg10189009-15431

PattersonTAEvansKCarterTLampGunnJS(2008)Movementand behaviour of large southern bluefin tuna (Thunnus maccoyii)in the Australian region determined using pop-up satellite ar-chival tags Fisheries Oceanography 17 352ndash367 httpsdoiorg101111j1365-2419200800483x

PollockKH JonesCMampBrownTL (1994)Angler survey meth-ods and their applications in fisheries management Bethesda MDAmericanFisheriesSociety

PonsM Branch T AMelnychukM C JensenO P Brodziak JFromentinJMhellipHilbornR(2017)EffectsofbiologicaleconomicandmanagementfactorsontunaandbillfishstockstatusFish and Fisheries181ndash21httpsdoiorg101111faf12163

PowersSPampAnsonK (2016)Estimating recreationaleffort in theGulf of Mexico Red Snapper Fishery using boat ramp camerasReduction in federalseason lengthdoesnotproportionally reducecatchNorth American Journal of Fisheries Management 36 1156ndash1166httpsdoiorg1010800275594720161198284

Robson D amp Jones C M (1989) The theoretical basis of an ac-cess site angler survey design Biometrics 45 83ndash98 httpsdoiorg1023072532036

Rocklin D Levrel H Drogou M Herfaut J amp Veron G (2014)Combining telephone surveys and fishing catches self-reportTheFrenchseabass recreational fisheryassessmentPLoS ONE9e87271httpsdoiorg101371journalpone0087271

Smallwood C B Pollock K HWise B S Hall N G amp GaughanD J (2012) Expanding aerial-roving surveys to include counts ofshore-based recreational fishers from remotely operated camerasBenefitslimitationsandcosteffectivenessNorth American Journal of Fisheries Management321265ndash1276httpsdoiorg101080027559472012728181

Teixeira D ZischkeM T ampWebley J A C (2016) Investigatingbias in recreational fishing surveys Fishers listed in public tele-phone directories fish similarly to their unlisted counterpartsFisheries Research 181 127ndash136 httpsdoiorg101016jfishres201604012

van Poorten B T amp Brydle S (2018) Estimating fishing effortfrom remote traffic counters Opportunities and challengesFisheries Research 204 231ndash238 httpsdoiorg101016jfishres201802024

van Poorten B T Carruthers T RWardHGM ampVarkeyD A(2015)Imputingrecreationalanglingeffortfromtime-lapsecamerasusinganhierarchicalBayesianmodelFisheries Research172265ndash273httpsdoiorg101016jfishres201507032

12emsp |emsp emspensp FLYNN et aL

VenturelliPAHyderKampSkovC(2017)AnglerappsasasourceofrecreationalfisheriesdataOpportunitieschallengesandproposedstandardsFish and Fisheries18578ndash595httpsdoiorg101111faf12189

WiseBSTelferCFLaiEKMHallNGampJacksonG(2012)Long-termmonitoringofboat-basedrecreational fishing inSharkBayWesternAustraliaProvidingscientificadviceforsustainablemanagement in a World Heritage Area Marine and Freshwater Research631129ndash1141httpsdoiorg101071MF12054

WoodGLynchTPDevineCKellerKampFigueiraW(2016)High-resolutionphoto-mosaictime-seriesimageryformonitoringhumanuseofanartificialreefEcology and Evolution66963ndash6968httpsdoiorg101002ece32342

YuFRupertACCDiTizianaMNishikantaKRandySIllahNhellipJeffreyG(2011)TerapixelimagingofcosmologicalsimulationsThe Astrophysical Journal Supplement Series19718

Zischke M T Griffiths S P amp Tibbetts I R (2012) Catch and ef-fort from a specialised recreational pelagic sport fishery off

eastern Australia Fisheries Research 127ndash128 61ndash72 httpsdoiorg101016jfishres201204011

SUPPORTING INFORMATION

Additional supporting information may be found online in theSupportingInformationsectionattheendofthearticle

How to cite this articleFlynnDJHLynchTPBarrettNSWongLSCDevineCHughesDGigapixelbigdatamoviesprovidecost-effectiveseascapescaledirectmeasurementsofopen-accesscoastalhumanusesuchasrecreationalfisheriesEcol Evol 2018001ndash12 httpsdoiorg101002ece34301

Graphical AbstractThecontentsofthispagewillbeusedaspartofthegraphicalabstractofhtmlonly

Itwillnotbepublishedaspartofmainarticle

Wetrailedaremoteautonomousultra-high-resolutionphotosamplingmethodasacosteffectsolutionfordirectseascape-scaledmeasure-mentsOursequentialphotosamplingwasbatchedprocessesusingnovelTimeMachinedatahandlingsoftwaretoproduceldquobigdatardquotimeseriesmoviesfromaspatialsubsetofanoffshorefisherywhichwecomparedtoaregionalbus-routesurveyofboattrailersandinterviewswithparticipantsOurcamerasamplescloselytrackedlargerscaleregionaltrendsbutwithlaborcostsaquarterofthebus-routesurveycostsforanindividualsample25ofthebusroutes

Page 10: Gigapixel big data movies provide cost-effective seascape

10emsp |emsp emspensp FLYNN et aL

CulversonMrStuartDudgeonandMrMattStapenellKarenEvensprovidedhelpfulcommentsonaninternaldraftGreatestthankstothemany recreational anglerswhowillingly shared information tosurveyclerkswithouttheirinvolvementthisprojectwouldnothavecometofruitionThreeanonymousreviewersprovidedcommentsthatimprovedthemanuscript

CONFLIC T OF INTERE S TS

Theauthorsdeclarenoconflictofinterests

AUTHOR CONTRIBUTIONS

MrFlynnhelpeddesignthestudyledthefieldworkundertookmostoftheanalysisandhelpwritethemanuscriptDrLynchhelpdesignthestudyassisted inthefieldworkprovidedadviceontheanaly-sisanddidsomeanalysisconstructedfiguresandafterMrFlynncontributedthemosttothedraftingofthemanuscriptDrBarretthelpeddesign the study assisted in some fieldwork provided ad-viceonanalysisandeditedthemanuscriptMrWonghelpedexten-sivelywithfieldworkassistedwithanalysisconstructedfiguresandhelpedwritethemanuscriptMsDevineassistedwithfieldworkandconfigurationoftheequipmentincludingtheapplicationofthebigdatamoviesoftwareandhandlingofphotographsandalsohelpedwritethemanuscriptMrHughesdevelopedtheelectronicsfortheequipmentoversaw thebuildofgearwrote the software topro-gramtherobotandreviewedandeditedthemanuscript

DATA ACCE SSIBILIT Y

Survey sampling schedules data collection sheets extrapolationproceduresandtechnicalCRAGSspecificationsarealluploadedasonlinesupportinginformationappendixRawdatafromsurveysandCRAGSDOIhttpdoiorg105061dryadv10d218

RE SE ARCH E THIC S COMPLIANCE AND FUNDERS

ResearchabidesbytheAustraliancodeofhumanresearchandex-perimentation(NationalStatementonEthicalConductinHumanResearch of 2007) Permission to conduct research on humanparticipants was granted by the University of Tasmania SocialScience under reference H0014772mdashldquoRemote investigation ofcoastal area user group quantification and variationrdquo aswell asfromtheCSIROSocialScienceHumanResearchEthicsCommitteeunder reference 08115mdashldquoHawkesbury Region Project HumanUsePatternsrdquoFundingwasbyUTASasPartoftheIMASHonoursProject funding and through internal support from the CoastsProgram

ORCID

Tim P Lynch httporcidorg0000-0002-5346-8454

R E FE R E N C E S

Alessa L Kliskey A amp Brown G (2008) Socialndashecological hotspotsmapping A spatial approach for identifying coupled socialndasheco-logicalspaceLandscape and Urban Planning8527ndash39httpsdoiorg101016jlandurbplan200709007

ArlinghausR(2006)Overcominghumanobstaclestoconservationofrecreational fishery resources with emphasis on central EuropeEnvironmental Conservation 33 46ndash59 httpsdoiorg101017S0376892906002700

Arlinghaus R Tillner R amp Bork M (2015) Explaining participationratesinrecreationalfishingacrossindustrialisedcountriesFisheries Management and Ecology 22 45ndash55 httpsdoiorg101111fme12075

BadcockPBPatrickKSmithAMASimpsonJMPennayDRisselCEhellipRichtersJ(2016)Differencesbetweenlandlineandmobilephoneusers in sexualbehavior researchArchives of Sexual Behavior46(6)1711ndash1721

Blumberg S J amp Luke J V (2009) Reevaluating the need for con-cern regarding noncoverage bias in landline surveys American Journal of Public Health 99 1806ndash1810 httpsdoiorg102105AJPH2008152835

Bousetouane F ampMorris B (2015) Off-the-shelf CNN features forfine-grainedclassificationofvesselsinamaritimeenvironmentInGBebisRBoyleBParvinDKoracinIPavlidisRFerisTMcGrawMElendtRKopperERaganZYeampGWeber(Eds)Advances in visual computing 11th International symposium ISVC 2015 Las Vegas NV USA December 14ndash16 2015 proceedings Part II (pp379ndash388)Cham Switzerland Springer International Publishing httpsdoiorg101007978-3-319-27863-6

Cooke S J amp Cowx I G (2004) The role of recreational fish-ing in global fish crises BioScience 54 857ndash859 httpsdoiorg1016410006-3568(2004)054[0857TRORFI]20CO2

DiggleP J (1995)Time-series a biostatistical introductionOxfordUKClarendonPress

EdwardsJampSchindlerE (2017)Avideomonitoringsystemtoeval-uate ocean recreational fishing effort in Astoria OregonMarine recreational information program (pp 31) Newport OR OregonDepartmentofFishandWildlife

FarrMStoecklNampSuttonS(2014)RecreationalfishingandboatingArethedeterminantsthesameMarine Policy47126ndash137httpsdoiorg101016jmarpol201402014

ForbesETraceySampLyle JM (2009)Assessment of the 2008 rec-reational gamefish fishery of Southeast Tasmania with particular refer-ence to Southern Bluefin TunaHobartTASTasmanianAquacultureandFisheriesInstituteUniversityofTasmania

FretwellPTScofieldPampPhillipsRA(2017)Usingsuper-highres-olutionsatelliteimagerytocensusthreatenedalbatrossesIbis159481ndash490httpsdoiorg101111ibi12482

GiriKampHallK (2015)SouthAustralian recreational fishingsurveyFisheries Victoria internal report series (pp 63) Queenscliff VicFisheriesVictoria

Griffiths S P Pollock K H Lyle J M Pepperell J G TonksM L amp Sawynok W (2010) Following the chain to elu-sive anglers Fish and Fisheries 11 220ndash228 httpsdoiorg101111j1467-2979201000354x

Halpern B SWalbridge S Selkoe K A Kappel C V Micheli FDrsquoAgrosaChellipWatsonR(2008)AglobalmapofhumanimpactonmarineecosystemsScience319948ndash952httpsdoiorg101126science1149345

Hartill BW Payne GW Rush N amp Bian R (2016) Bridging thetemporal gap Continuous and cost-effective monitoring of dy-namic recreational fisheries by web cameras and creel surveysFisheries Research 183 488ndash497 httpsdoiorg101016jfishres201606002

emspensp emsp | emsp11FLYNN et aL

HenryGWampLyleJM(2003)NationalrecreationalandindigenousfishingsurveyFinal report to the fisheries research amp development cor-poration and the fisheries action program (pp 188) Canberra ACTFRDC

JonesCMampPollockKH(2012)Recreationalanglersurveymeth-odsEstimationofeffortharvestandreleasedcatchInAZaleDLParrishampTMSutton(Eds)Fisheries techniquesBethesdaMDAmericanFisheriesSociety

Kearney R E (2002) Co-management The resolution of conflict be-tween commercial and recreational fishers in Victoria AustraliaOcean amp Coastal Management45201ndash214httpsdoiorg101016S0964-5691(02)00055-8

KellerKSteffeASLowryMMurphyJJampSuthersIM(2016)Monitoringboat-basedrecreationalfishingeffortatanearshorearti-ficialreefwithashore-basedcameraFisheries Research18184ndash92httpsdoiorg101016jfishres201603025

KinlochMAMcGlennonDNicollGampPikePG(1997)Evaluationof the bus-route creel surveymethod in a largeAustralianmarinerecreationalfisheryISurveydesignFisheries Research33101ndash121httpsdoiorg101016S0165-7836(97)00068-4

LewinWCArlinghausRampMehnerT(2006)Documentedandpo-tential biological impacts of recreational fishing Insights forman-agementandconservationReviews in Fisheries Science14305ndash367httpsdoiorg10108010641260600886455

LowryMampMurphy J (2003)Monitoring the recreational gamefishfisheryoffsouth-easternAustraliaMarine and Freshwater Research54425ndash434httpsdoiorg101071MF01269

LyleJMStarkKEampTraceySR(2014)2012ndash2013 Survey of rec-reational fishing in TasmaniaHobart TAS Institute forMarine andAntarcticStudies

Lynch T P (2006) Incorporation of recreational fishing ef-fort into design of marine protected areas Incorporacioacuten delEsfuerzo de Pesca Recreativa en el Disentildeo de Aacutereas MarinasProtegidas Conservation Biology 20 1466ndash1476 httpsdoiorg101111j1523-1739200600509x

LynchTP(2008)Thedifferencebetweenspatialandtemporalvaria-tioninrecreationalfisheriesforplanningofmarineprotectedareasResponse toSteffeConservation Biology22486ndash491httpsdoiorg101111j1523-1739200800888x

LynchTP (2014)Adecadal time-seriesof recreational fishingeffortcollectedduringandafterimplementationofamultipleusemarinepark shows high inter-annual but low spatial variability Fisheries Research15185ndash90httpsdoiorg101016jfishres201309014

LynchTPAldermanRampHobdayAJ(2015)Ahigh-resolutionpan-orama camera system formonitoring colony-wide seabird nestingbehaviourMethods in Ecology and Evolution6491ndash499httpsdoiorg1011112041-210X12339

LynchTFlynnDWoodGKimJ-HParkJampHobdayAJ(2017)Metrics parameters and results from fisheries and wildlife studies with the CSIRO Ruggedized Autonomous Gigapixel Systems (CRAGS)DarwinNTAustralianMarineScienceAssociation

LynchTPWilkinsonEMellingLHamiltonRMacreadyAampFearyS(2004)ConflictandimpactsofdiversandanglersinamarineparkEnvironmental Management 33 196ndash211 httpsdoiorg101007s00267-003-3014-6

MaHOgawaTKSminkeyTRBreidtFJLesserVMOpsomerJDhellipVanVoorheesDA(2018)Pilotsurveystoimprovemonitoringofmarine recreational fisheries inHawaiʻiFisheries Research204197ndash208httpsdoiorg101016jfishres201802010

McCluskey S M amp Lewison R L (2008) Quantifying fish-ing effort A synthesis of current methods and their ap-plications Fish and Fisheries 9 188ndash200 httpsdoiorg101111j1467-2979200800283x

McGlennonDampKinlochMA(1997)Evaluationofthebus-routecreelsurveymethod in a largeAustralianmarine recreational fishery II

PilotsurveysandoptimalsamplingallocationFisheries Research3389ndash99httpsdoiorg101016S0165-7836(97)00067-2

McPheeDLeadbitterDampSkilleterGA(2002)SwallowingthebaitIs recreational fishing in Australia ecologically sustainable Pacific Conservation Biology840httpsdoiorg101071PC020040

Monz C A Pickering C M amp HadwenW L (2013) Recent ad-vances in recreation ecology and the implications of differentrelationships between recreation use and ecological impactsFrontiers in Ecology and the Environment11441ndash446httpsdoiorg101890120358

MooreAHallKGiriKTraceySPenroseLHansenShellipBrownP(2015)Developingrobustandcost-effectivemethodsforestimatingthenationalrecreationalcatchofSouthernBluefinTunainAustraliaFRDC Projectpp123

Morton A J amp Lyle J M (2004) Preliminary assessment of the rec-reational gamefish fishery in Tasmania with particular reference to Southern Bluefin Tuna Taroona TAS Tasmanian Aquaculture andFisheriesInstitute

ParnellPEDaytonPKFisherRALoarieCCampDarrowRD(2010)SpatialpatternsoffishingeffortoffSanDiegoImplicationsfor zonal management and ecosystem function Ecological Applications202203ndash2222httpsdoiorg10189009-15431

PattersonTAEvansKCarterTLampGunnJS(2008)Movementand behaviour of large southern bluefin tuna (Thunnus maccoyii)in the Australian region determined using pop-up satellite ar-chival tags Fisheries Oceanography 17 352ndash367 httpsdoiorg101111j1365-2419200800483x

PollockKH JonesCMampBrownTL (1994)Angler survey meth-ods and their applications in fisheries management Bethesda MDAmericanFisheriesSociety

PonsM Branch T AMelnychukM C JensenO P Brodziak JFromentinJMhellipHilbornR(2017)EffectsofbiologicaleconomicandmanagementfactorsontunaandbillfishstockstatusFish and Fisheries181ndash21httpsdoiorg101111faf12163

PowersSPampAnsonK (2016)Estimating recreationaleffort in theGulf of Mexico Red Snapper Fishery using boat ramp camerasReduction in federalseason lengthdoesnotproportionally reducecatchNorth American Journal of Fisheries Management 36 1156ndash1166httpsdoiorg1010800275594720161198284

Robson D amp Jones C M (1989) The theoretical basis of an ac-cess site angler survey design Biometrics 45 83ndash98 httpsdoiorg1023072532036

Rocklin D Levrel H Drogou M Herfaut J amp Veron G (2014)Combining telephone surveys and fishing catches self-reportTheFrenchseabass recreational fisheryassessmentPLoS ONE9e87271httpsdoiorg101371journalpone0087271

Smallwood C B Pollock K HWise B S Hall N G amp GaughanD J (2012) Expanding aerial-roving surveys to include counts ofshore-based recreational fishers from remotely operated camerasBenefitslimitationsandcosteffectivenessNorth American Journal of Fisheries Management321265ndash1276httpsdoiorg101080027559472012728181

Teixeira D ZischkeM T ampWebley J A C (2016) Investigatingbias in recreational fishing surveys Fishers listed in public tele-phone directories fish similarly to their unlisted counterpartsFisheries Research 181 127ndash136 httpsdoiorg101016jfishres201604012

van Poorten B T amp Brydle S (2018) Estimating fishing effortfrom remote traffic counters Opportunities and challengesFisheries Research 204 231ndash238 httpsdoiorg101016jfishres201802024

van Poorten B T Carruthers T RWardHGM ampVarkeyD A(2015)Imputingrecreationalanglingeffortfromtime-lapsecamerasusinganhierarchicalBayesianmodelFisheries Research172265ndash273httpsdoiorg101016jfishres201507032

12emsp |emsp emspensp FLYNN et aL

VenturelliPAHyderKampSkovC(2017)AnglerappsasasourceofrecreationalfisheriesdataOpportunitieschallengesandproposedstandardsFish and Fisheries18578ndash595httpsdoiorg101111faf12189

WiseBSTelferCFLaiEKMHallNGampJacksonG(2012)Long-termmonitoringofboat-basedrecreational fishing inSharkBayWesternAustraliaProvidingscientificadviceforsustainablemanagement in a World Heritage Area Marine and Freshwater Research631129ndash1141httpsdoiorg101071MF12054

WoodGLynchTPDevineCKellerKampFigueiraW(2016)High-resolutionphoto-mosaictime-seriesimageryformonitoringhumanuseofanartificialreefEcology and Evolution66963ndash6968httpsdoiorg101002ece32342

YuFRupertACCDiTizianaMNishikantaKRandySIllahNhellipJeffreyG(2011)TerapixelimagingofcosmologicalsimulationsThe Astrophysical Journal Supplement Series19718

Zischke M T Griffiths S P amp Tibbetts I R (2012) Catch and ef-fort from a specialised recreational pelagic sport fishery off

eastern Australia Fisheries Research 127ndash128 61ndash72 httpsdoiorg101016jfishres201204011

SUPPORTING INFORMATION

Additional supporting information may be found online in theSupportingInformationsectionattheendofthearticle

How to cite this articleFlynnDJHLynchTPBarrettNSWongLSCDevineCHughesDGigapixelbigdatamoviesprovidecost-effectiveseascapescaledirectmeasurementsofopen-accesscoastalhumanusesuchasrecreationalfisheriesEcol Evol 2018001ndash12 httpsdoiorg101002ece34301

Graphical AbstractThecontentsofthispagewillbeusedaspartofthegraphicalabstractofhtmlonly

Itwillnotbepublishedaspartofmainarticle

Wetrailedaremoteautonomousultra-high-resolutionphotosamplingmethodasacosteffectsolutionfordirectseascape-scaledmeasure-mentsOursequentialphotosamplingwasbatchedprocessesusingnovelTimeMachinedatahandlingsoftwaretoproduceldquobigdatardquotimeseriesmoviesfromaspatialsubsetofanoffshorefisherywhichwecomparedtoaregionalbus-routesurveyofboattrailersandinterviewswithparticipantsOurcamerasamplescloselytrackedlargerscaleregionaltrendsbutwithlaborcostsaquarterofthebus-routesurveycostsforanindividualsample25ofthebusroutes

Page 11: Gigapixel big data movies provide cost-effective seascape

emspensp emsp | emsp11FLYNN et aL

HenryGWampLyleJM(2003)NationalrecreationalandindigenousfishingsurveyFinal report to the fisheries research amp development cor-poration and the fisheries action program (pp 188) Canberra ACTFRDC

JonesCMampPollockKH(2012)Recreationalanglersurveymeth-odsEstimationofeffortharvestandreleasedcatchInAZaleDLParrishampTMSutton(Eds)Fisheries techniquesBethesdaMDAmericanFisheriesSociety

Kearney R E (2002) Co-management The resolution of conflict be-tween commercial and recreational fishers in Victoria AustraliaOcean amp Coastal Management45201ndash214httpsdoiorg101016S0964-5691(02)00055-8

KellerKSteffeASLowryMMurphyJJampSuthersIM(2016)Monitoringboat-basedrecreationalfishingeffortatanearshorearti-ficialreefwithashore-basedcameraFisheries Research18184ndash92httpsdoiorg101016jfishres201603025

KinlochMAMcGlennonDNicollGampPikePG(1997)Evaluationof the bus-route creel surveymethod in a largeAustralianmarinerecreationalfisheryISurveydesignFisheries Research33101ndash121httpsdoiorg101016S0165-7836(97)00068-4

LewinWCArlinghausRampMehnerT(2006)Documentedandpo-tential biological impacts of recreational fishing Insights forman-agementandconservationReviews in Fisheries Science14305ndash367httpsdoiorg10108010641260600886455

LowryMampMurphy J (2003)Monitoring the recreational gamefishfisheryoffsouth-easternAustraliaMarine and Freshwater Research54425ndash434httpsdoiorg101071MF01269

LyleJMStarkKEampTraceySR(2014)2012ndash2013 Survey of rec-reational fishing in TasmaniaHobart TAS Institute forMarine andAntarcticStudies

Lynch T P (2006) Incorporation of recreational fishing ef-fort into design of marine protected areas Incorporacioacuten delEsfuerzo de Pesca Recreativa en el Disentildeo de Aacutereas MarinasProtegidas Conservation Biology 20 1466ndash1476 httpsdoiorg101111j1523-1739200600509x

LynchTP(2008)Thedifferencebetweenspatialandtemporalvaria-tioninrecreationalfisheriesforplanningofmarineprotectedareasResponse toSteffeConservation Biology22486ndash491httpsdoiorg101111j1523-1739200800888x

LynchTP (2014)Adecadal time-seriesof recreational fishingeffortcollectedduringandafterimplementationofamultipleusemarinepark shows high inter-annual but low spatial variability Fisheries Research15185ndash90httpsdoiorg101016jfishres201309014

LynchTPAldermanRampHobdayAJ(2015)Ahigh-resolutionpan-orama camera system formonitoring colony-wide seabird nestingbehaviourMethods in Ecology and Evolution6491ndash499httpsdoiorg1011112041-210X12339

LynchTFlynnDWoodGKimJ-HParkJampHobdayAJ(2017)Metrics parameters and results from fisheries and wildlife studies with the CSIRO Ruggedized Autonomous Gigapixel Systems (CRAGS)DarwinNTAustralianMarineScienceAssociation

LynchTPWilkinsonEMellingLHamiltonRMacreadyAampFearyS(2004)ConflictandimpactsofdiversandanglersinamarineparkEnvironmental Management 33 196ndash211 httpsdoiorg101007s00267-003-3014-6

MaHOgawaTKSminkeyTRBreidtFJLesserVMOpsomerJDhellipVanVoorheesDA(2018)Pilotsurveystoimprovemonitoringofmarine recreational fisheries inHawaiʻiFisheries Research204197ndash208httpsdoiorg101016jfishres201802010

McCluskey S M amp Lewison R L (2008) Quantifying fish-ing effort A synthesis of current methods and their ap-plications Fish and Fisheries 9 188ndash200 httpsdoiorg101111j1467-2979200800283x

McGlennonDampKinlochMA(1997)Evaluationofthebus-routecreelsurveymethod in a largeAustralianmarine recreational fishery II

PilotsurveysandoptimalsamplingallocationFisheries Research3389ndash99httpsdoiorg101016S0165-7836(97)00067-2

McPheeDLeadbitterDampSkilleterGA(2002)SwallowingthebaitIs recreational fishing in Australia ecologically sustainable Pacific Conservation Biology840httpsdoiorg101071PC020040

Monz C A Pickering C M amp HadwenW L (2013) Recent ad-vances in recreation ecology and the implications of differentrelationships between recreation use and ecological impactsFrontiers in Ecology and the Environment11441ndash446httpsdoiorg101890120358

MooreAHallKGiriKTraceySPenroseLHansenShellipBrownP(2015)Developingrobustandcost-effectivemethodsforestimatingthenationalrecreationalcatchofSouthernBluefinTunainAustraliaFRDC Projectpp123

Morton A J amp Lyle J M (2004) Preliminary assessment of the rec-reational gamefish fishery in Tasmania with particular reference to Southern Bluefin Tuna Taroona TAS Tasmanian Aquaculture andFisheriesInstitute

ParnellPEDaytonPKFisherRALoarieCCampDarrowRD(2010)SpatialpatternsoffishingeffortoffSanDiegoImplicationsfor zonal management and ecosystem function Ecological Applications202203ndash2222httpsdoiorg10189009-15431

PattersonTAEvansKCarterTLampGunnJS(2008)Movementand behaviour of large southern bluefin tuna (Thunnus maccoyii)in the Australian region determined using pop-up satellite ar-chival tags Fisheries Oceanography 17 352ndash367 httpsdoiorg101111j1365-2419200800483x

PollockKH JonesCMampBrownTL (1994)Angler survey meth-ods and their applications in fisheries management Bethesda MDAmericanFisheriesSociety

PonsM Branch T AMelnychukM C JensenO P Brodziak JFromentinJMhellipHilbornR(2017)EffectsofbiologicaleconomicandmanagementfactorsontunaandbillfishstockstatusFish and Fisheries181ndash21httpsdoiorg101111faf12163

PowersSPampAnsonK (2016)Estimating recreationaleffort in theGulf of Mexico Red Snapper Fishery using boat ramp camerasReduction in federalseason lengthdoesnotproportionally reducecatchNorth American Journal of Fisheries Management 36 1156ndash1166httpsdoiorg1010800275594720161198284

Robson D amp Jones C M (1989) The theoretical basis of an ac-cess site angler survey design Biometrics 45 83ndash98 httpsdoiorg1023072532036

Rocklin D Levrel H Drogou M Herfaut J amp Veron G (2014)Combining telephone surveys and fishing catches self-reportTheFrenchseabass recreational fisheryassessmentPLoS ONE9e87271httpsdoiorg101371journalpone0087271

Smallwood C B Pollock K HWise B S Hall N G amp GaughanD J (2012) Expanding aerial-roving surveys to include counts ofshore-based recreational fishers from remotely operated camerasBenefitslimitationsandcosteffectivenessNorth American Journal of Fisheries Management321265ndash1276httpsdoiorg101080027559472012728181

Teixeira D ZischkeM T ampWebley J A C (2016) Investigatingbias in recreational fishing surveys Fishers listed in public tele-phone directories fish similarly to their unlisted counterpartsFisheries Research 181 127ndash136 httpsdoiorg101016jfishres201604012

van Poorten B T amp Brydle S (2018) Estimating fishing effortfrom remote traffic counters Opportunities and challengesFisheries Research 204 231ndash238 httpsdoiorg101016jfishres201802024

van Poorten B T Carruthers T RWardHGM ampVarkeyD A(2015)Imputingrecreationalanglingeffortfromtime-lapsecamerasusinganhierarchicalBayesianmodelFisheries Research172265ndash273httpsdoiorg101016jfishres201507032

12emsp |emsp emspensp FLYNN et aL

VenturelliPAHyderKampSkovC(2017)AnglerappsasasourceofrecreationalfisheriesdataOpportunitieschallengesandproposedstandardsFish and Fisheries18578ndash595httpsdoiorg101111faf12189

WiseBSTelferCFLaiEKMHallNGampJacksonG(2012)Long-termmonitoringofboat-basedrecreational fishing inSharkBayWesternAustraliaProvidingscientificadviceforsustainablemanagement in a World Heritage Area Marine and Freshwater Research631129ndash1141httpsdoiorg101071MF12054

WoodGLynchTPDevineCKellerKampFigueiraW(2016)High-resolutionphoto-mosaictime-seriesimageryformonitoringhumanuseofanartificialreefEcology and Evolution66963ndash6968httpsdoiorg101002ece32342

YuFRupertACCDiTizianaMNishikantaKRandySIllahNhellipJeffreyG(2011)TerapixelimagingofcosmologicalsimulationsThe Astrophysical Journal Supplement Series19718

Zischke M T Griffiths S P amp Tibbetts I R (2012) Catch and ef-fort from a specialised recreational pelagic sport fishery off

eastern Australia Fisheries Research 127ndash128 61ndash72 httpsdoiorg101016jfishres201204011

SUPPORTING INFORMATION

Additional supporting information may be found online in theSupportingInformationsectionattheendofthearticle

How to cite this articleFlynnDJHLynchTPBarrettNSWongLSCDevineCHughesDGigapixelbigdatamoviesprovidecost-effectiveseascapescaledirectmeasurementsofopen-accesscoastalhumanusesuchasrecreationalfisheriesEcol Evol 2018001ndash12 httpsdoiorg101002ece34301

Graphical AbstractThecontentsofthispagewillbeusedaspartofthegraphicalabstractofhtmlonly

Itwillnotbepublishedaspartofmainarticle

Wetrailedaremoteautonomousultra-high-resolutionphotosamplingmethodasacosteffectsolutionfordirectseascape-scaledmeasure-mentsOursequentialphotosamplingwasbatchedprocessesusingnovelTimeMachinedatahandlingsoftwaretoproduceldquobigdatardquotimeseriesmoviesfromaspatialsubsetofanoffshorefisherywhichwecomparedtoaregionalbus-routesurveyofboattrailersandinterviewswithparticipantsOurcamerasamplescloselytrackedlargerscaleregionaltrendsbutwithlaborcostsaquarterofthebus-routesurveycostsforanindividualsample25ofthebusroutes

Page 12: Gigapixel big data movies provide cost-effective seascape

12emsp |emsp emspensp FLYNN et aL

VenturelliPAHyderKampSkovC(2017)AnglerappsasasourceofrecreationalfisheriesdataOpportunitieschallengesandproposedstandardsFish and Fisheries18578ndash595httpsdoiorg101111faf12189

WiseBSTelferCFLaiEKMHallNGampJacksonG(2012)Long-termmonitoringofboat-basedrecreational fishing inSharkBayWesternAustraliaProvidingscientificadviceforsustainablemanagement in a World Heritage Area Marine and Freshwater Research631129ndash1141httpsdoiorg101071MF12054

WoodGLynchTPDevineCKellerKampFigueiraW(2016)High-resolutionphoto-mosaictime-seriesimageryformonitoringhumanuseofanartificialreefEcology and Evolution66963ndash6968httpsdoiorg101002ece32342

YuFRupertACCDiTizianaMNishikantaKRandySIllahNhellipJeffreyG(2011)TerapixelimagingofcosmologicalsimulationsThe Astrophysical Journal Supplement Series19718

Zischke M T Griffiths S P amp Tibbetts I R (2012) Catch and ef-fort from a specialised recreational pelagic sport fishery off

eastern Australia Fisheries Research 127ndash128 61ndash72 httpsdoiorg101016jfishres201204011

SUPPORTING INFORMATION

Additional supporting information may be found online in theSupportingInformationsectionattheendofthearticle

How to cite this articleFlynnDJHLynchTPBarrettNSWongLSCDevineCHughesDGigapixelbigdatamoviesprovidecost-effectiveseascapescaledirectmeasurementsofopen-accesscoastalhumanusesuchasrecreationalfisheriesEcol Evol 2018001ndash12 httpsdoiorg101002ece34301

Graphical AbstractThecontentsofthispagewillbeusedaspartofthegraphicalabstractofhtmlonly

Itwillnotbepublishedaspartofmainarticle

Wetrailedaremoteautonomousultra-high-resolutionphotosamplingmethodasacosteffectsolutionfordirectseascape-scaledmeasure-mentsOursequentialphotosamplingwasbatchedprocessesusingnovelTimeMachinedatahandlingsoftwaretoproduceldquobigdatardquotimeseriesmoviesfromaspatialsubsetofanoffshorefisherywhichwecomparedtoaregionalbus-routesurveyofboattrailersandinterviewswithparticipantsOurcamerasamplescloselytrackedlargerscaleregionaltrendsbutwithlaborcostsaquarterofthebus-routesurveycostsforanindividualsample25ofthebusroutes

Page 13: Gigapixel big data movies provide cost-effective seascape

Graphical AbstractThecontentsofthispagewillbeusedaspartofthegraphicalabstractofhtmlonly

Itwillnotbepublishedaspartofmainarticle

Wetrailedaremoteautonomousultra-high-resolutionphotosamplingmethodasacosteffectsolutionfordirectseascape-scaledmeasure-mentsOursequentialphotosamplingwasbatchedprocessesusingnovelTimeMachinedatahandlingsoftwaretoproduceldquobigdatardquotimeseriesmoviesfromaspatialsubsetofanoffshorefisherywhichwecomparedtoaregionalbus-routesurveyofboattrailersandinterviewswithparticipantsOurcamerasamplescloselytrackedlargerscaleregionaltrendsbutwithlaborcostsaquarterofthebus-routesurveycostsforanindividualsample25ofthebusroutes