Ecology and Evolution. 2019;1–14.  | 1www.ecolevol.org

Received:9October2018  |  Revised:18December2018  |  Accepted:27December2018DOI:10.1002/ece3.4923

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

Long‐term sound and movement recording tags to study natural behavior and reaction to ship noise of seals

Lonnie Mikkelsen1 | Mark Johnson2,3 | Danuta Maria Wisniewska1,4 |  Abbo van Neer5 | Ursula Siebert5 | Peter Teglberg Madsen3,6 | Jonas Teilmann1

ThisisanopenaccessarticleunderthetermsoftheCreativeCommonsAttributionLicense,whichpermitsuse,distributionandreproductioninanymedium,providedtheoriginalworkisproperlycited.©2019TheAuthors.Ecology and EvolutionpublishedbyJohnWiley&SonsLtd.

1DepartmentofBioscience,AarhusUniversity,Roskilde,Denmark2SeaMammalResearchUnit,UniversityofSt.Andrews,St.Andrews,UK3DepartmentofBioscience,AarhusUniversity,AarhusC,Denmark4HopkinsMarineStation,StanfordUniversity,Stanford,California5InstituteforTerrestrialandAquaticWildlifeResearch(ITAW),UniversityofVeterinaryMedicineHannover,Foundation,Germany6AarhusInstituteforAdvancedStudies,AarhusUniversity,AarhusC,Denmark

CorrespondenceJonasTeilmann,DepartmentofBioscience,AarhusUniversity,Roskilde,Denmark.Email:jte@bios.au.dk

Funding informationMarieSklodowska‐Curiecareerintegrationgrant;MarineAllianceforScienceandTechnologyScotland;GermanFederalAgencyofNatureConservation,Grant/AwardNumber:FKZ3515822000andZ1.2‐53302/2010/14;AarhusUniversity,Grant/AwardNumber:2015‐15‐0201‐00549;DanishNationalResearchCouncil;OfficeofNavalResearch;CoastalProtection,NationalParksandOceanProtection

Abstract1. Theimpactofanthropogenicnoiseonmarinefaunaisofincreasingconservationconcernwithvesselnoisebeingoneofthemajorcontributors.Animalsthatrelyon shallow coastal habitats may be especially vulnerable to this form ofpollution.

2. Verylimitedinformationisavailableonhowmuchnoisefromshiptrafficindivid‐ual animals experience, andhow theymay react to it due to a lackof suitablemethods.Toaddressthis,wedevelopedlong‐durationaudioand3D‐movementtags(DTAGs)anddeployedthemonthreeharborsealsandtwograysealsintheNorthSeaduring2015–2016.

3. Thesetagsrecordedsound,accelerometry,magnetometry,andpressurecontinu‐ouslyforupto21days.GPSpositionswerealsosampledforonesealcontinuouslythroughouttherecordingperiod.Aseparatetag,combiningacameraandanac‐celerometerlogger,wasdeployedontwoharborsealstovisualizespecificbehav‐iorsthathelpedinterpretaccelerometersignalsintheDTAGdata.

4. Combiningdatafromdepth,accelerometer,andaudiosensors,wefoundthatani‐mals spent 6.6%–42.3% of the time hauled out (either on land or partly sub‐merged),and5.3%–12.4%oftheirat‐seatimerestingattheseabottom,whiletheremainingtimewasusedfortraveling,restingatsurface,andforaging.Animalswereexposedtoaudiblevesselnoise2.2%–20.5%oftheir timewhen inwater,andwe demonstrate that interruption of functional behaviors (e.g., resting) insomecasescoincideswithhigh‐levelvesselnoise.Two‐thirdsof theshipnoiseeventswere traceableby theAISvessel trackingsystem,whileone‐thirdcom‐prisedvesselswithoutAIS.

5. This preliminary study demonstrates how concomitant long‐term continuousbroadbandon‐animalsoundandmovementrecordingsmaybeanimportanttoolinfuturequantificationofdisturbanceeffectsofanthropogenicactivitiesatseaandassessmentoflong‐termpopulationimpactsonpinnipeds.

K E Y WO RD S

anthropogenicnoise,behavioralresponse,biologging,DTAG,exposurerates,grayseal,harborseal,long‐durationacousticdataloggers

2  |     MIKKELSEN Et aL.

1  | INTRODUCTION

Growing industrializationof themarineenvironment is resultinginhabitatchangesand increasingmarinedefaunation (McCauleyet al., 2015; Richardson, Greene, Malme, & Thomson, 1995). Agreater awareness of increasing levels of anthropogenic noisehaspromptedstudies tounderstandandmitigate theirpotentialnegative impactsonmarine life (Hildebrand,2005).Whilemanyrecentstudieshavesoughttoaddresstheeffectsofunderwaternoiseoncetaceans(Nowacek,Thorne,Johnston,&Tyack,2007),comparativelylessisknownaboutexposureandreactionstonoisein pinnipedswhile at sea. Like cetaceans, pinnipeds have sensi‐tiveunderwaterhearing; their fullhearing rangeextends fromafewhundredHzto70–80kHz(Cunningham&Reichmuth,2016;Hemilä, Nummela, Berta, & Reuter, 2006). They rely on soundfor communication (Mathevon, Casey, Reichmuth, & Charrier,2017;VanParijs,Hastie,&Thompson,1999),predatordetection(Deecke, Slater, & Ford, 2002), and possibly also for navigationand listening for prey (Schusterman, Levenson, Reichmuth, &Southall, 2000).Pinnipedshavebeen found to respond stronglytounderwatertonepulsesat8–45kHzincaptivity(Götz&Janik,2010;Kasteleinetal.,2015;Kastelein,Heul,Terhune,Verboom,& Triesscheijn, 2006a; Kastelein, Heul, Verboom, Triesscheijn,& Jennings, 2006b) and to sounds from seismic surveys (Harris,Miller,&Richardson,2001)andpiledriving (Russelletal.,2016)inthewild.

Amajortechnicalchallengeinassessingtheimpactofnoiseonmarine fauna is thatof sampling thenoise levels routinelyexperi‐encedbyanimalsinthewildandsimultaneouslytheanimals’natu‐ralbehavior.While controlledexperimentshave led to substantialprogressinevaluatingresponsesoffree‐rangingmarinemammalstoimpulsivenoisesourcessuchassonarorairguns(Milleretal.,2009;Tyacketal.,2011;vanBeestetal.,2018),fewstudieshaveexam‐inedtheeffectsofcontinuousnoise,includingshipnoise,whichmaydominatetheambientnoiselevelincoastalareasandnearshipping

lanes (Wisniewskaetal.,2018).Shipnoisemaybeespecially rele‐vanttocoastalsealsthatrelyonperiodicland‐basedresting(haulingout)andthereforespendmuchoftheirlivesincoastalhabitatsthatstronglyoverlapwithmarinetraffic.

Pinniped at‐sea behavior has been studied extensively using avarietyofbiologgingtechnologies.TrackingdevicesbasedonArgos,GPS,orVHFhaveprovided insight intohorizontalmovementpat‐terns,whereastime–depthrecorders,aloneorintegratedinposition‐ingdevices,havebeenusedtostudydivepattern(Carter,Bennett,Embling,Hosegood,&Russell,2016).Incoastalspecies,suchasgrayseals(Halichoerus grypus, Figure1)andharborseals(Phoca vitulina),at‐seabehaviorhasmainlybeendescribedbasedon2Ddivepro‐files,wheredivebehaviorsareclassifiedastravelingdives(V‐shapeddives), foraging (U‐shaped dives), or resting dives (skewed dives),as well as resting at the surface (Russell et al., 2015; Thompson,Hammond,Niceolas,&Fedak,1991),whichinsomecaseshasbeenvalidatedthroughcamerause(Heaslip,Bowen,&Iverson,2014).

Low‐sampling‐rate position and dive data can generally betransmittedbyradio,throughmobilephonenetworksorsatellite,enablinglong‐termdatacollectionwithtagsthatdonotneedtoberetrieved(McConnelletal.,2004).However,therateatwhichdataarecollectedneedstomatchthescaleofmovementofaparticularspeciesandthebehaviorofinterest.Usingdiveshapestodistin‐guishsearch,restingandforagingreliesonassumptionsaboutbe‐haviorwhichmaybetoosimple(Ramasco,Biuw,&Nilssen,2014).Whilepositiondatahavebeenusedtoinfertravelingandforagingusingarea‐restrictedsearch(ARS)orfirstpassagetime(FPT)anal‐yses,theseinferenceshavereceivedlimitedvalidation.Newstate‐spacemodels(SSM),likehiddenMarkovmodels(HMM),havethepotentialofintegrating3Dmovementsandenvironmentalparam‐eters, increasing the power to distinguish behaviors. However,theaccuracyofsuchmodelstoquantifyvariousstatesofbehav‐ior is highly influenced by data resolution (Carter et al., 2016).Higher‐sampling‐rate sensors, and in particular accelerometers,haveproventobeusefulforinterpretingfine‐scaledivebehaviors

F I G U R E 1  SleepinggraysealwithaDTAG3onHelgolandMay2015.Photo:SabineSchwarz

     |  3MIKKELSEN Et aL.

suchaspreycaptureevents(Gallonetal.,2013;Heerah,Hindell,Guinet, & Charrassin, 2014; Volpov et al., 2015). The challengeinhigh‐resolutiondata (fromsound,cameras,oraccelerometers)is the largeamountofdata thatcannotbe transmittedby radio,requiring instead that data are stored on board the tag whichmustbephysically recovered. It is, therefore, challenging toob‐tainhigh‐resolutiondataoverlongperiodsoftimeandinremoteareas.Moreover,evenwhentagscanberecovered,newmethodsareneededfortheefficientanalysisofthecomplexmulti‐sensordatasetsreturnedbythesedevices.

To assess the effects of anthropogenic noise sources, the re‐ceivednoiselevelsanddetailedanimalbehaviormustbeestimatedsimultaneously (Nowacek et al., 2007). Studies of pinniped re‐sponsestonoisehavelargelyreliedonvisualorvideosurveillanceofanimalsatthesurfaceorwhenhauledout(Andersen,Teilmann,Dietz,Schmidt,&Miller,2012;Blackwell,Lawson,&Williams,2004;Harrisetal.,2001)combinedwithnoisepropagationmodelingwithnodirectmeasurementofnoiseexposure.Amoredirectapproachistosampletheacousticenvironmentoftheanimalusingasoundrecording tag (Johnson, Aguilar Soto, & Madsen, 2009). Despitebeingfirstdevelopedtostudytheeffectsofsoundexposuresonthedivebehaviorofdeep‐divingelephantseals(Burgess,Tyack,Boeuf,&Costa,1998;Costaetal.,2003;Fletcher,Boeuf,Costa,Tyack,&Blackwell,1996),soundandmovementrecordingtagshavenotbeenwidelyusedonsmallerpinnipeds,dueperhapstothe largesizeofearlierversionsofthesetags.Morerecently,compactsoundrecord‐ingtags,suchastheDTAG(Johnson&Tyack,2003),thatcombinesound recordings with high‐bandwidth movement sensors havebeenwidelyusedtostudynoiseimpactsoncetaceans.Thesetags,whicharetypicallyattachedwithsuctioncups,havebeenusedtolinkshort‐termbehavioralresponsestospecificnoisesourcessuchasmilitarysonar(DeRuiteretal.,2013),airgunsusedinoilprospec‐tion(Madsenetal.,2006;Milleretal.,2009),andshipnoise(AguilarSotoet al., 2006;Nowacek, Johnson,&Tyack,2004;Wisniewskaetal.,2018).DTAGdeploymentsoncetaceanshavesofarbeenlim‐itedby thedurationof suction cup attachments, aswell asmem‐oryand/orbatterycapacityconstraintsrelatedtothesmalltagsize,withatypicalmaximumrecordingtimeoflessthantwodaysathighsamplingrates(Johnsonetal.,2009).Suchshortdurationsarenotidealforassessingexposuresandresponsestoopportunisticnoisesources such as vessel passes. However, advances in low‐powerelectronictechnologynowallowforincreasedbatteryandmemorycapacity, resulting inextendedperiodsof continuous recordingofthreeormoreweeks,whilesimultaneouslyreducingtagsize.Sealsareidealcandidatesfortheselonger‐termdevices,asthetagscanbegluedtothefuroftheanimal.

Here,wepresentinitialresultsfromnewlydeveloped,long‐termhigh‐resolution sound andmovement DTAGs deployed on harborandgrayseals,representingthefirstmulti‐week,continuousbroad‐bandsoundrecordingsfromanymarineanimal.Wedemonstratethepotentialof suchdata forquantifying individualnoiseexposure insynchronywiththefine‐scalebehaviorsoftheanimal,enablingtheidentificationofnoise‐inducedbehavioralalterationstogetherwith

their consequences for individual time–energybudgets.Forvisualverification,wefurthermorecombinedsimultaneousvideoandac‐celerometryrecordingsonwildseals.

2  | MATERIALS AND METHODS

2.1 | DTAG deployment

During2015–2016,threeharborsealsandtwograysealswerecap‐turedfortagattachmentatLorenzensplate(N54.4393,E8.6419)intheGermanWaddenSeaandatHelgoland(N54.1886,E7.9117)inGermany,respectively(Table1).Thestudywasapprovedunderper‐mitnumberAzV312‐72241.121‐19(70‐6/07)andV244‐3986/2017(17‐3/14) of theMinistry of Energy,Agriculture, Environment andRuralAreasofSchleswig‐Holstein,Germany.

Harbor sealswere caughtwhen hauled out on sand banks bysurroundingthesealsusingalargenet(3m×200m)deployedfromtwoboatsandthendraggingthenetmanuallyonshore,wherethesealswere transferred into tube nets andmanually restrained forhandlingandtagging (Jeffries,Brown,&Harvey,1993).Graysealswerecaughtusingalightweightpole‐netmadeofcarbonfiberwindsurfermastsandnylonnet(meshsize:2×2cm).Sealswerecaughtatlowtidewhentheyweretoofarupthebeachtoreachthewaterbeforebeingenclosed in thepole‐net (Arcalís‐Planasetal.,2015).Animals were restrained in the nets for approx. 30min to deter‐minesex,weight,andlengthandtoattachthetags.Abloodsamplewasalsotakenforhealthassessment.Twoversionsofthetagwereused.The2015tag(DTAG‐3)containedasyntacticfoamfloatandhadan integratedArgostransmitterandVHFbeacon(Figure1).Amorecompactversionof thetag (DTAG‐4)wasused in2016,andthiswasattachedtoahigh‐pressureclosedcellfoamfloatcontain‐inganArgostransmitterwithanintegratedlow‐powerUHFbeacon(SPOT6,WildlifeComputers,Seattle,USA).ThecompleteDTAG‐3packagemeasured55×37×205mmandweighed325ginair,whiletheDTAG‐4measured40×33×180mmandweighed206g.Bothtagswereapprox.20gbuoyantinwater.Thetagsweremountedonanaluminumreleaseplate(1‐mmthick),whichinturnwasgluedtothefurontheseal'supperbackwithtwo‐componentepoxy (Ergo7211).FortheDTAG‐3,a2.5mmmagnesiumnut,astainlesssteelwasher, and a 5mmnylon pin (glued to the float) held the tag tothealuminumplate.Thetagwasreleasedfromtheplatewhenthenutcorrodedafterapprox.3–4weeks.InDTAG‐4,nickel–chromiumwiresoneachsideofthetagsecuredthetagtothealuminumplate.Afterapreprogrammedtime,thewiresweremadeanodicbyacur‐rentfromthebatterytocausetheirrapidcorrosion.OneofthefourDTAG‐3 and the DTAG‐4 deployment did not release as plannedandremainedattachedtotheanimalsforseveralweeks.Thesetagswerebroughttothecoastbywatercurrentswheretheywerefoundbylocalresidents.TheremainingtagswererecoveredbyboatusingArgostogetanapproximateposition(i.e.,withinaradiusofafewkilometers)followedbyVHFtracking.

Bothversionsof theDTAGcontainedsensors forsound,pres‐sure (depth), acceleration, magnetic field, and GPS. Sound was

4  |     MIKKELSEN Et aL.

sampled from a single 10‐mm‐diameter end‐capped cylindricalpiezo‐ceramichydrophoneatarateof240kHz(DTAG‐3)or192kHz(DTAG‐4)with16‐bitresolution.Sounddataweredecimatedwithinthetagbyafactorof2(DTAG‐3)or3(DTAG‐4)followedbyloss‐lesscompression(Johnson,Partan,&Hurst,2013)beforebeingstoredinmemory,resultinginastoredsamplingrateof120kHz(DTAG‐3)or64kHz (DTAG‐4).Toreducethechanceofclippingduetoflownoisewhenthesealisswimming,aone‐polehigh‐passfilterwasin‐cludedwithacut‐offfrequencyof100Hz,resultinginarecordingbandwidth (−3‐dB bandwidth) from100Hz to 51kHz and 100Hzto27kHzfortheDTAG‐3andDTAG‐4,respectively.Sensorscom‐prisingatriaxialaccelerometer,atriaxialmagnetometer,andapres‐suretransducerweresampledwith16‐bitresolutionat200HzperchannelinDTAG‐3,andat200Hz(acceleration)and50Hz(magne‐tometerandpressure)inDTAG‐4.ThetagsalsoincludedasnapshotGPSwhichmade64ms acquisitionsof availableGPS satellite sig‐nalswhentheanimalsurfaced.Thissensor functionedpoorlyduetoelectricalinterferenceontheDTAG‐3butwasoperationalontheDTAG‐4.GPSprocessingwasperformedafterrecoveryof thetagusing custom software and Internet‐published satellite almanacs.Thesound,sensor,andGPSdatawerestoredona64‐GBflashmem‐oryarrayinthetags.

2.2 | Camera tag deployment

Thecameratagcomprisedatriaxialaccelerometeranddivelog‐ger(200Hzsamplingrate,“OpenTag,”LoggerheadInstruments,Sarasota,FL,USA),asmalldigitalcamera(30fps,DVL200,LittleLeonardo,Tokyo,Japan),anArgossatellitetransmitter(SPOT5),andaVHFtransmitter,allenclosedinahigh‐pressureclosedcellfoam. The accelerometer/dive logger and the videowere syn‐chronizedprior todeployment,andtimingsynchronywassub‐sequentlyverifiedbycomparingthetimingofsurfaceintervalsinthediveandvideodata.Thistag(approx.55×30×150mm,weight inair230g)wasdeployedonanadultmaleharborsealat “Bosserne” (N 55.9367; E 10.7757), Kattegat, Denmark. Asecond deployment on a female harbor sealwas conducted atthesamelocationbutwiththetagpackagereducedinsizeandtheArgosandVHFtransmittersreplacedwithaSPOT6Argostransmitter (tag dimensions approx. 35×35×160mm, weightinair170g).Bothtagswereslightlybuoyantinwatertoenablerecovery. On both occasions, the camera was set with a timedelay to start recording on the morning following tag attach‐ment inaneffort toavoidsamplingdisturbedbehavior relatedtothehandlingoftheanimal.Thecameratagwasmountedtoanaluminumplateprecoatedwithstandardconstructionadhesive(SMP‐38) which was in turn attached to the back of the seal.The coating was necessary to facilitate the use of super glue(Loctite422)betweenthesealandtheplate.Thetagwasheldontothealuminumplatebya1mmmagnesiumnut,astainlesssteelwasher,anda5mmnylonpinthatwasgluedtothefloat.Thenutcorrodedafter2dayspermittingthetagtoreleasefromtheanimal. TA

BLE

1 Tagdeploymentsummary.Ageclasswasasubjectivejudgementbasedonsizeoftheanimal

Animal ID

hs15

_069

ags

15_1

39a

gs15

_139

bhs

15_2

78a

hs16

_265

c20

16–1

3807

020

17–6

421

Species

Harborseal

Grayseal

Grayseal

Harborseal

Harborseal

Harborseal

Harborseal

Tagversion

DTAG3

DTAG3

DTAG3

DTAG3

DTAG4

Cameratag

Cameratag

Sex

Male

Female

Male

Female

Female

Male

Female

Weight(kg)

8510

014

466

6271

58

Std.length(cm)

149

176

144

126

148

138

121

Ageclass

Adult

Juvenile

Subadult

Adult

Adult

Adult

Adult

Deploymentlocation

Lorenzensplate

Helgoland

Helgoland

Lorenzensplate

Lorenzensplate

Bosserne

Bosserne

Taggingdate

10‐03‐2015

19‐05‐2015

19‐05‐2015

5‐10‐2015

21‐09‐2016

12‐08‐2016

27‐07‐2017

Recordingperiodontheseal

10‐03‐2015‐

24‐03‐2015

19‐05‐2015‐

02‐06‐2015

19‐05‐2015‐

04‐06‐2015

5‐10‐2015‐

19‐10‐2015

21‐09‐2016‐

11‐10‐2016

13‐08‐2016

06:00‐08:30

28‐07‐2017

08:2

0–10

:50

Durationofusabledata

336hr(14days)

336hr(14days)

377hr(15.7days)

336hr(14days)

521hr(21.7days)

2.5hrofvideo

2.5hrofvideo

Proportionoftimespenthauledout(onland/partly

submerged)

42.3%(28.7/13.6%)

24%(19.4/4.6%)

36%(29.2/6.8%)

6.6%(6.4/0.1%)

Datanotavailableyet

––

Proportionofat‐seatimespentresting(mean

submersiontime/

SD)

8.1%(5.3/1.4min)

7.2%(6.2/2.2min)

5.3%(5.2/1.4min)

12.4%(4.0/1.2min)

Datanotavailableyet

––

Proportionofat‐seatimeexposedtovesselnoise

10.6%

2.2%

20.5%

6.6%

Datanotavailableyet

––

Vesselnoiseevents(min/maxduration)

30(1.8/330min)

17(1/81min)

74(1.9/170min)

34(1.2–173min)

41

     |  5MIKKELSEN Et aL.

2.3 | Data analysis

Data analyseswere performed inMatlab R2013b (MathWorksInc.). Sound exposurewas quantified as one‐third octave bandlevels (TOLs), that is the rootmean square (RMS) sound pres‐sure level inone‐thirdoctavebands roughlyapproximating thefilterbankofthemammalianauditorysystem(Richardsonetal.,1995).Toreducethecontributionofnoisesmadebythetaggedanimal(e.g.,duetomovement),theTOLswerecomputedinthreesteps similar to themethod inWisniewska et al. (2018). First,successive4,096(DTAG‐3)or2048(DTAG‐4)pointFFTs(Hannwindow, 50% overlap) were computed of the sound recordinggivinga frequencyresolutionof29Hzand31Hz, respectively.The power in the FFT bins falling between 3 and 20kHz wassummedtogiveabroadbandnoiseestimatethat is largelyfreeoflow‐frequencyflownoisewithasamplingrateof58or62Hz(i.e.,oneestimateper50%overlappedFFT).Thelower10thper‐centileofthisbroadbandnoiseestimateover30sintervalswascomputed, and FFTs that had a noise estimate below the 10thpercentilewereidentified.ThespectralpoweroftheseFFTswasaveraged to give a 30s ambient noise spectrum that is robusttosoundtransients.Finally,TOLswereestimatedfromthe30saveragespectrabycombining thepower inFFTbinswhich fellwithineachthird‐octavebandandconvertingtoareceivedlevelindBre1µPa(RMS)bycorrectingforthesensitivityofthetagsof−176dBreV/µPa.

DailyplotswereconstructedoftheTOLs,alongwith20Hzdec‐imateddepth,three‐axisacceleration,andRMSjerk(Figure2).Jerk,J,isanindicationofrapidmovementofthetagandwascomputedbasedonthenormofthedifferentialofthetriaxialaccelerationfol‐lowingYdesenetal.(2014),thatis:

where Ax,tisthetriaxialaccelerationinm/s2atsampletimetinthe

x‐axis,andfsisthesamplingrate.TheRMSofJwascomputedover0.4sintervalswith50%overlaptoproduceatimeserieswith20Hzsamplingrate.

Depth,acceleration,andjerkplotswerescreenedmanuallyforhauloutandrestingperiodsatsea.Restingperiodsatseawereiden‐tifiedastwoormoresequentialU‐shapeddives,inwhichtheanimalshowedverylittleactivityinthedescendingandbottomphases(asassessedfromthejerkandaccelerometersignal),andalackofmove‐mentat thebottom (i.e., absenceof flownoise in theaudiodata).Thisbehaviorwasverifiedasrestingontheseafloorbyinspectionofthecameratagdata.Precisestartandendtimesofallrestingperiods(i.e.,thetotaltimefromthestartofthefirstdiveinarestingbouttotheendofthelastdiveincludinginterveningsurfacetimes)weremarked inthedata.Averagesubmersiontimeduringtheserestingperiodswassubsequentlydetermined.

TOLplots (Figure 2)were screened visually for noise eventsabove approx. 70dB re 1µPaRMS in oneormore third‐octavebands≥1kHz.Foreachevent,approx.10softherecordingwas

examinedby listening to identify thesoundsource. If shipnoisewas encountered, the start and end times of audibility wereidentified.

For the DTAG‐4, GPS positions were obtained at 2–3min in‐tervalswhenthesealwasatthesurface.Afterprocessing,positionestimateswithanRMSpseudo‐range residual greater than200morwithanestimatedaltitudemorethan150maboveorbelowtheWGS‐84geoidwererejected.Thetypicalpseudo‐rangeresidualofacceptedpositionswasabout20m.Noothertrackfilteringwasap‐plied.AutomaticIdentificationSystem(AIS,i.e.,mandatorytrackingof all larger ships >300 gross tonnes) records, obtained fromTheGermanFederalWaterwaysandShippingAdministration,wereex‐aminedtoidentifyallregisteredvesselswithin5kmofeachposition.Thetimeandrangeoftheclosestpassageofeachofthesevesselswererecorded.Toproducesoundexposureplots,theGPSpositionswereinterpolatedto30sintervalsandthenplottedasatrackcol‐oredbythecorresponding1kHzTOL.Thisthird‐octavebandwaschosenasbeingthelowestbandthatwaslargelyunaffectedbylow‐frequencyflownoise.GiventherelativelyfrequentGPSpositions,linearinterpolationwasusedtoestimatethepositionsat30sinter‐vals.OutagesinGPSmeasurementslastingmorethan10minwerenotinterpolated.

Videos from the camera tags were viewed in slow motion toclassifybehaviorsintoswimminginthewatercolumnoralongthebottom(possiblesearchbehavior),aswellassurfacingandrestingperiodsatthebottom.Theaccelerationdataassociatedwiththesebehaviorswereextractedtohelpinterpretsimilaraccelerometerre‐cordingsfromtheDTAGs.

3  | RESULTS

DTAG deployments yielded 14–21days of continuous sound andmovementdatafromtwograyandthreeharborseals,whilecameradeploymentsprovided2.5hreachofcombinedvideoandaccelera‐tiondataontwoharborseals(Table1).

The audio recordings contained frequent episodes of noiselevelswithTOLsabove1kHzexceeding70dBre1µPa.Figure2showsthethird‐octavelevelsfortheentire15.7‐dayDTAG‐3de‐ploymentonagrayseal(gs15_139b)inpanela,andinmoredetailfor one day in panel c,which demonstrate frequent fluctuationsinnoiselevel.Alargeproportionoftherecordednoiseatlowfre‐quencieswasduetowaterflowaroundthetag(flownoise).Highbroadband sound levels resulted from the seal breaking the sur‐face,bubblesbeing released fromaroundthe tagpackageor thefuroftheseal,fromrain,andalsocloseshippasses.Overall,shipnoisewasaudiblefor2.2%–20.5%ofthetimethatthefourseals,tagged in2015,spent inwater (excl.haulouttime,Table1).Thiswas spread over 17–74 events that lasted 1–330min, some ofwhichmay comprisemultipleoverlappingvessel passes.Anothertypeof low‐levelmechanical noisewas also audibleover severaldaysinthegs15_139bgraysealrecordingandmayhaveoriginatedfromanoffshorewindfarm,dredging,oranoilrig.

Jt= fs√

�

�

Ax,t−Ax,t−1

�2+

�

Ay,t−Ay,t−1

�2+

�

Az,t−Az,t−1

�2�

6  |     MIKKELSEN Et aL.

Inthe2016harborsealdata(hs16_265c),GPSwasrecordedsuc‐cessfullyalongwithaudioandsensordataprovidinganopportunitytocomparenoiseexposureeventswithAISshiptracks.Theaverage

timebetweenGPSpositionswasalittleovertwominsthroughouttherecordingtime,exceptforafewlongerperiodswithoutsatellitecontact,resultingin12,972positionsover21days.Figure3shows

F I G U R E 2   (a)AudiofromthefullDTAG‐3deploymentongraysealgs15_139b,displayedasRMSsoundpressurelevelinthird‐octavebands(TOL,1–32kHz)averagedover30s.Thetagdidnotcollectdataduringthethreeuniformwhiteintervalsduetotechnicalissues.(b–f)Datafromasingleday,where(b)indicateswhenshipnoiseisaudible;(c)third‐octavelevels(TOLs);(d)diveprofile;(e)accelerationalongtheanimal'sx‐(surge,forwards‐backwards),y‐(sway,sidetoside),andz‐(heave,upanddown)axis;(f)jerkcalculatedasthedifferentialofthethreeaccelerationaxes.Periodswhenthesealisrestingatsea,correspondingtolowjerklevels,areindicatedinthegraph.TimeisinUTC.

0

10

20

30

40

Dep

th (m

)

−30

−20

−10

0

10

20

30

Acc

eler

atio

n (m

/s2 )

axayaz

5 10 15 200

100

200

300

400

500

Jerk

(m/s

3 )

Time of day

TOL

frequ

ency

(kH

z)

24−May−2015

1

2

4

8

16

32

TOL

(dB

re 1

µPa

rms)

60

70

80

90

Shi

pno

ise

19−May−2015

1248

1632 20−May−2015 21−May−2015

TOL

frequ

ency

(kH

z)22−May−2015

1248

1632 23−May−2015 24−May−2015

25−May−2015

1248

1632 26−May−2015 27−May−2015

28−May−2015

1248

1632 29−May−2015 30−May−2015

31−May−2015

1248

1632 01−Jun−2015 02−Jun−2015

0 5 10 15 2003−Jun−2015

0 5 10 15 201248

1632

Time of day04−Jun−2015

0 5 10 15 20

Resting

(a)

(b)

(c)

(e)

(f)Resting

Figure 4

(d)

Rain Ship

     |  7MIKKELSEN Et aL.

thetracklinecolor‐codedbysoundlevel.Elevatednoiselevels(or‐ange/redcolors)wereinmanycasesduetoshipnoise(asconfirmedby listening to each high‐noise‐level event). A total of 41 vesselpasseswererecordedwith1kHzTOLabove90dBre1µPaRMS(peak30‐saverage),butonly in27ofthesecaseswasanAIS‐reg‐isteredvesselwithin5kmoftheanimal.ThisindicatesthatvesselswithoutAISwereresponsibleforaboutone‐thirdofthehigh‐levelvesselnoiseexposuresexperiencedbythisseal.

Sensordata fromthe2015deployments revealed that the twoharborsealsspent6.6%–42.3%andthetwograysealsspent24%–36%oftherecordingtimehauledout(Table1).Usingaudiodata,itwaspossibletodifferentiatebetweenthesealbeingcompletelyoutofthewaterorpartlysubmerged:Soundsfromwaterflushingoverthetagorthesealliftingitsheadclearofthewatertobreathewereassociated with partial submergence (Table 1). Resting periods atsea, identifiedby lowacceleration/jerk levelsatdepth (Figure2e‐f,Figure4),wereafterwardvalidatedbyinspectionofthecameratagvideo during intervals with similar low variability of acceleration(Figure 5). Both harbor seals with camera tags exhibited this kindofrestingbehaviorinwhichtheanimalslayattheseafloor,rockingslowlywiththecurrent(Figure5andVideoS1).Basedontheseob‐servations,wefoundthattheharborsealsspent8.1%–12.4%andthegraysealsspent5.3%–7.2%oftheirat‐seatimeexhibitingthisrestingbehavior.On average, harbor sealswere submerged for 4–5.3minandgraysealsfor5.2–6.2minduringrestingdives(Table1).

DTAGrecordingsfrombothsealspeciescontainedexamplesofbehavioralchangesthatcoincidedwithvesselencounters.For thegrayseal(gs15_139b)depictedinFigure2b‐c,severalshippassagesresulted inelevatednoise levels (35.5%of theday shown), andatleastoneofthevesselpassesoccurredwhenthesealwasinarest‐ingdive,whichwassubsequentlyinterruptedasshowninFigure4.In thisplot, theanimal initiallyexhibits at‐sea restingbehavior.At

around06:44:00,shipnoisebecomesaudibleand isclearlyvisiblein the spectral plot by 07:08:00, indicating that the vessel is ap‐proachingtheanimal.Shortlythereafter,at07:09:50(Figure4d–f),thesealbreaksoffanascentfromanapparentlynormalrestingdive,descentsbrieflytothenresumeascendingtothesurface.Duringthenext dive (07:12:30), the animal accelerates rapidly at the bottomandmakesaninterruptedascentbeforereturningtothesurfacetobreathe.Thevessel noise reached amaximumbroadband level of113dBre1µPaRMS(0.1–50kHz,1saverage)at07:12:10whenthesealwas still at the bottom. The depth and acceleration data fol‐lowingthisvesselencounter(Figure4b,c)indicatethatthesealter‐minateditsrestingbehaviorandbeganactiveswimmingsoonafter07:16:00.Vesselnoiseceasedtobeaudibleat09:35:19.

Anotherexampleinwhichthesameanimalaltersbehaviorpresum‐ablyduetovesseldisturbanceisseenwhilethesealappearstobeper‐formingtravelingdives(Figure6).Here,theanimalisdivingtoapprox.10–15m depth with continuous flipper stroke‐and‐glide swimming,evidentaslargeoscillationsinthex‐(surge)andy‐axis(sway,Figure6).Duringthedivestartingat18:55,theanimalsuddenlydescendstowhatispresumablytheseabottom(atapprox.27mdepth)whereitremainsstationaryjudgingbythelackofactivityintheaccelerometerdata.Thisbehaviorcoincideswithapeakinvesselnoisefromwhatappearstobeasmallfastboatgiventherapidriseandfallinnoiselevel.Inthiscase,theeffectseemstobelimitedtothisparticulardive.

An example of disturbance during haul out potentially duetotheoccurrenceofavesselwasfoundindatafromharborsealhs15_069a (Figure7).Here, the animal is hauledout in the firstpartofthefigure,butat11:58:57theanimalretreatsabruptlyintotheseawheretheaudiodatarevealhigh‐levelshipnoise(thismayhavebeenaudible to thesealwhilehauledout,butwasnotde‐tectableinthetagduetothelowsensitivityofthepiezo‐ceramichydrophoneinair).Onceinthewater,thesealswimsenergetically,

F I G U R E 3  Trackofharborsealhs16_265c,color‐codedwith30‐saveragesofthird‐octavelevels(TOL)inthe1‐kHz‐centeredband(seeMaterialsandMethods).GraydottedlinesindicateAutomaticIdentificationSystem(AIS)tracksofshipsthatatsomepointpasswithin5kmoftheseal(roughlycorrespondingtotheexpectedrangeofaudibility)andreddotsmarkthepositionsofwindfarms

1kH

z TO

L (d

B re

1 µ

Pa R

MS)

8  |     MIKKELSEN Et aL.

asevidencedbylargeoscillationsintheaccelerometersignalandhighflownoiseintheaudio,untilitsurfacesat12:00:15.Theves‐selnoisedisappearswhenthetagisoutofthewaterbutisonceagainaudiblewhentheseal resumesdiving (theaudio recordingcanbefoundinonlineSupportingInformationAudioS1).

4  | DISCUSSION

Vesselnoiseisthedominantsourceofnoisepollutionintheocean(Hildebrand,2009)andmayhaveaparticularimpactoncoastalma‐rineanimalswhosehomerangesoverlapstronglywithbothship‐pinglanesandrecreationalboatuse.Amajorchallengeinassessingthe impactofanthropogenicnoiseonmarinefauna is toquantify

boththenoiseexperiencedbyindividualanimalsandhowtheyre‐spond to it.Here,we demonstrate amethod for obtaining thesedata: High‐resolution multi‐sensor tags (DTAGs) were deployedonharbor andgray seals, providing continuous, broadbandaudiorecordings along with synchronous high‐resolution movementdata for up to 21days.An analysis of these data can be used toestablishnormalbehavioralstatesandtoinferchangesinbehaviorin thecontextofnaturalandanthropogenicnoise in theenviron‐ment.TheadditionofGPSlocationsinthenewestversionofthetagprovidesinformationonwhereanimalsfindresources,wheretheyencounterhigher levelsofnoisedisturbance,andcanaid in iden‐tifyingspecificnoisesourcessuchasoilrigs,windfarms,bridges,andships(i.e.,byassociatingGPSlocationswithvesseltracksfromAIStransmissions).

F I G U R E 4  Behaviorofgs15_139bbefore,during,andaftertheshipencounteron24May2015indicatedinFigure2f.(a–c)dataovera75‐minperiodcenteredonthevesselpass.(a)Receivedpowerspectrumdensitylevel;(b)diveprofile;(c)accelerationprofile;(d–f)zoomed‐inviewofthesamedataduring9minofthevesselpass.Noticetheregulardive/restingpatternthatisinterruptedwhenshipnoiseincreases

Freq

uenc

y (k

Hz)

0

10

20

30

40

60

80

0

10

20

30

Dep

th (m

)

07:08:24 07:09:36 07:10:48 07:12:00 07:13:12 07:14:24 07:15:36 07:16:48

−20

0

20

Time of day

Acc

eler

atio

n (m

/s2 )

ax ay az

Freq

uenc

y (k

Hz)

0

10

20

30

Rec

eive

d le

vel

(dB

re 1

µPa2 H

z–1)

Rec

eive

d le

vel

(dB

re 1

µPa2 H

z–1)

40

60

80

0

10

20

30

Dep

th (m

)

06:36:00 06:48:00 07:00:00 07:12:00 07:24:00 07:36:00

−20

0

20

Acc

eler

atio

n (m

/s2 )

ax ay az

(b)

(c)

(e)

(f)

(d)

(a)

     |  9MIKKELSEN Et aL.

Usingtheaudiorecordings, itwaspossibletodistinguishnatu‐ralnoisesourcessuchasrainfromvesselnoiseandtomakeapre‐liminaryassessmentof theamountof timeeachnoisesourcewas

audible.Theproportionof time thatvesselnoisewasaudiblevar‐iedwidelybetweenanimals (2%–20%) likely reflectingdifferencesinthetrafficdensityinthelocationsvisitedbyanimals,withsome

F I G U R E 5   (a,c)Framesfromthevideorecordingsobtainedwithcameratagsontwoharborseals(partofthesealsareseeninthelowerpartofeachpicture);bandd)theassociatedaccelerationprofiles,wherethearrowsindicatethespecifictimeofthesnap‐shots.Bothimagesweretakenfromperiodswherethesealswererestingontheseafloor.Theanimalsarerollingfromsidetosidepresumablywiththewavecycles(mostobviousindassmalloscillationsinthey(sway)axis).Seetheassociatedvideo(SupportinginformationAudioS1)toimage(a)intheSupportingInformation

09:42:00 09:43:00 09:44:00 09:45:00 09:46:00 09:47:00 09:48:00 09:49:00

Time of day

ax ay az

06:52:00 06:53:00 06:54:00 06:55:00 06:56:00 06:57:00 06:58:00 06:59:00−20

−10

0

10

20

Time of day

Acc

eler

atio

n (m

/s2 )

(a)

(b)

(c)

(d)

2016/08/13 06:54:26 2017/07/28 09:45:30

F I G U R E 6  Anexampleofachangeinbehaviorofgraysealgs15_139bpresumablycausedbyvesseldisturbance.Inthedivebeginningat18:55,thesealdescendsdeeperthaninpreviousdives,presumablytothebottomwherethesealremainsduringthevesselpass.Afterthisfastmovingboatpasses,thesealresumesitspreviousdivingstyle.(a)Receivedpowerspectrumdensitylevel;(b)depthprofile;(c)accelerationprofile

0

10

20

30

40

60

80

0

10

20

30

18:45:00 18:48:00 18:51:00 18:54:00 18:57:00 19:00:00 19:03:00−20

−10

0

10

20

Time of day

ax ay az

Freq

uenc

y (k

Hz)

Rec

eive

d le

vel

(dB

re 1

µPa2 H

z–1)

Dep

th (m

)A

ccel

erat

ion

(m/s

2 )

(b)

(c)

(a)

10  |     MIKKELSEN Et aL.

animalsbychanceoractivelyforaging,traveling,orrestinginareaswithdenseshiptraffic,forexample,commercialfishinggrounds.IntheexampleofFigure4,thevesselpassappearedtoprovokeanen‐ergeticresponsethatterminatedanaturalrestingcycleandprecip‐itatedabehavioralchangethatcontinuedwellafterthevesselhadpassed.Similardive responses tosoundstimulihavebeenseen inelephantsealsthatchangedfromascenttodescentandvice‐versawhenexposedtonoiseplaybacksmadebyanacoustictagattachedtotheirbacks(Fregosietal.,2013).IntheexampleofFigure7,theanimalmayhaverespondedtotheacousticorvisualcueoftheshipwhile hauled out on land, and again the consequencewas a pro‐longeddisruptionofrestingbehavior.Disturbancesofsealsrestingonlandhavebeendemonstratedusingvisualobservationsandte‐lemetry(Andersenetal.,2012;Edrénetal.,2010;Henry&Hammill,2001;Jansen,Boveng,Dahle,&Bengtson,2010;Osinga,Nussbaum,Brakefield,&UdodeHaes,2012).However,visualcontactisinevita‐blylostsoonaftersealsenterthewatermakingitdifficulttoassessthedurationof response.Oneof fewstudiestoaddressthisusedtelemetrytomeasurethedurationanddistancecoveredbysealsbe‐forereturningtoahauloutsiteafterbeingdisturbedwhilehauled‐out(Andersen,Teilmann,Dietz,Schmidt,&Miller,2014).Withthefine‐scale3DmovementdataandlongdurationoftheDTAGrecord‐ings,itisnowpossibletostudytheeffectsofdisturbancesinmuchfiner detail alongwith their frequencyof occurrenceover severalweeks. Thus, the combination of high‐resolution movement andsoundrecordingsprovidesapowerfullinkbetweennoiseexposures,specificchanges inbehavior,andtheirpotentialenergeticcosts intermsofswimmingeffortortimedivertedfromforaging.

The potential significance of cumulative effects of behavioralresponsestodisturbancedependsuponhowoftentheseoccurandthis has been difficult to assess for any marine mammal species.AlthoughitisattractivetoconsiderAISshiprecordsasawaytoinfervesselnoiseexposure,closevesselpassesdeducedfromAISrecords

donotaccountforallhigh‐levelboatnoiseevents,atleastforani‐malsclosetothecoastwheresmallvesseltrafficismoreprominent.Inthethree‐weekdatasetreportedhere,only66%ofvesselpassesforwhichthe1kHzTOLexceeded90dBre1µPaRMSatthesealweretraceabletoanAISreportedvesselwithin5km.Thishighlightsthe potentially significant, but difficult to predict, contribution ofvesselnoisefromsmallboats.Toaccountforthis,itisimportanttoobtain in situ informationaboutnoiseexposure for individual ani‐malsinsteadofrelyingonAISdata,atleastforanimalsclosetothecoastwheresmallvesseltrafficisabundant.

Soundrecordingtagsusedoncetaceanshavesofarbeenlimitedbysuctioncupattachmentstoabout2days.Longerduty‐cycledre‐cordingdurations(upto8and12days)havebeenobtainedonele‐phantsealsforwhichamoresecureattachmenttothefurispossible(e.g.,Fregosietal.,2013;Géninetal.,2015).However,thebandwidthoftheserecordingswaslimitedto0.6–16kHz,thelowerendofwhichoverlapswiththefrequencyrangeofflownoisemakingthecontribu‐tionofvesselnoisetothetotalreceivedleveldifficulttoassess.Anar‐rowbandwidthalsolimitsaccuratemeasurementofnoiselevelsfromsmallboatsandlargervesselstravelingathighspeedwhichcanpro‐ducesubstantialenergyathighfrequencies(Hermannsen,Beedholm,Tougaard,&Madsen,2014; Jensenetal.,2009;Wisniewskaetal.,2018)thatoverlapwithmarinemammalpeakhearingsensitivity.Asdemonstratedhere,increasingmemorydensityandnewlow‐powerelectronic systemsmake it possible to achieve both long‐durationand wider recording bandwidth in a miniature biologging tag, de‐ployable on even the smallest pinnipeds. The extended recordingtimeachievedhereprovides insight into the long‐termbehavioroftheanimalsandenablesquantificationofwhere,howoften,andatwhat levelanimalsencounteranthropogenicnoisesources.Theex‐tended recording duration also expands the potential to performcontrolledexposureexperiments (CEEs) at sea.Theseexperimentshavebeenperformedsuccessfullyoncetaceanstaggedwithsound

F I G U R E 7  Exampleofdisturbanceduringhauloutofaharborseal(hs15_069a)on18March2015.(a)Receivedpowerspectrumdensitylevel;(b)accelerationprofile

0

10

20

30

40

60

80

11:57:09 11:57:45 11:58:21 11:58:57 11:59:33 12:00:09 12:00:45 12:01:21−20

−10

0

10

20

Time of day

ax ay az

Freq

uenc

y (k

Hz)

Acc

eler

atio

n (m

/s2 )

Rec

eive

d le

vel

(dB

re 1

µPa2 H

z–1)

(b)

(a)seal is on land enters the water

     |  11MIKKELSEN Et aL.

andmovementrecordingtagstoassesstheimpactofspecificnoisesources (Kvadsheimet al., 2017;Madsen et al., 2006;Miller et al.,2009;Sivleetal.,2012).Still,theshortrecordingdurationofthesedevicesmeantthatlittletimecouldbeallowedaftertagdeploymentforanimalstoreturntonaturalbehaviorandtomeasurebaselinebe‐havior. Longer tag deployments facilitate exclusionof post‐taggingeffects,whileprovidingextendedbaselinedatapriortoexposure,aswellasthepossibilityofdetectingprolongedresponsestoexposures.Thelongrecordingtimealsoopensthepotentialtoperformmultiple,widelyspacedexposurestothesameindividuals,whichmayallowamorethoroughevaluationofindividualdose‐responsefunctionsandpotentialhabituation(Tyack,Gordon,&Thompson,2004).

The combination of audio, depth, and accelerometer data col‐lected by these long‐duration tags provides amore complete de‐scriptionofbehaviorthanobtainedwithsingle‐sensortagsenablingbehavioralstatesandtransitionstobedetectedmoreprecisely.Forexample, although haul out periods can be detected using just adepthsensor, theaccelerometerandaudiodataallowedus toad‐ditionally identify time intervals when the animal was resting onthebeachbutwaspartly submergedor flushedduring rising tide.Quantifyinghaulout inthisway is lesspronetobiascomparedtousing a salt‐water switch in satellite flipper tags, inwhich the tailmustbedryforaminimumtimeintervaltoberegisteredasahaul‐out (Lonergan,Duck,Moss,Morris,&Thompson,2013).Accuratehauloutdurationsareessentialwhencorrectingabundancesurveysfortheamountoftimeanimalsspendonlandandthusavailableforvisualdetection(Harvey&Goley,2011;Lonerganetal.,2013).

It has beenwidely assumed that harbor and gray sealsmainlyrest at or near the haul out, while floating at the sea surface orduringshallowdives(approx.8m,Thompsonetal.,1991;Russelletal.,2015),thesetypesofrestingwerealsofoundintheDTAGdata.InspectionoftheDTAGandvideo/accelerometrydataalsorevealedfrequent restingbehavior faroffshoreat thebottomofU‐shapeddivesdownto35m(Figure2)asrecentlysuggestedfromdivepro‐filesbyRamascoetal.(2014).ThistypicallylookedlikeaU‐shapeddivebutwith lowactivityduring thebaseof thedive.That thesedivescompriserestingbehaviorinharborsealswasconfirmedwiththevideorecordings,inwhichthesealwasseentobelyingstillorrocking inthecurrentonthebottom.Breath‐holdrestinghasalsobeen observed in fur seals (Arctocephalinae) (Jeanniard‐du‐Dot,Trites,Arnould,Speakman,&Guinet,2017)andasymmetricaldiveprofiles (so‐calleddriftdives)performedbyelephant seals are as‐sumedtorepresentrestingbehavior(Crocker,Boeuf,&Costa,1997;Watanabe,Baranov,&Miyazaki,2015).Forharborandgrayseals,U‐shapeddivesaretypicallyassociatedwithforagingbehavior(Russelletal.,2015;Thompsonetal.,1991)andastandard2Ddiveprofileanalysismayhavecategorizedtherestingdivesfoundhere(e.g.,inFigure4)asU‐shapedforagingdivesduetolackofdataonfine‐scalemovements (Carteretal.,2016).Thiscould leadtooverestimationof the number of foraging dives. Although the low activity levelsprovideclearevidenceforrestingwithindives,thesedivesalsohaduniformlyasymmetricalprofileswithslowdescentandfastascentrates inourdata forboth species.This suggests thepossibilityof

identifyingthisbehaviorfromstandard2Ddiveprofilesalonefromwhichtheycouldbeexcludedfromanalysesofforaging.

Volpovetal. (2015)deployedhead‐mounted3‐axisaccelerom‐eters in free‐rangingAustralian fur seals (Arctocephalus pusillus) toidentify individual attempted prey captures, which was validatedusingon‐animalvideocameras.Preycaptureattemptswereidenti‐fiedbypeaksinthevarianceofaccelerationineitherthex‐(surge),y‐ (sway),orz‐(heave)axis.Asimilarapproachusing jerkhasbeendemonstrated with captive harbor seals (Ydesen et al., 2014).However,automaticdetectionofpreycaptureattemptsfromaccel‐eration iscomplicatedbyotherbehaviorsthatmayresult inpeaksinacceleration (e.g.,Figure4f).Fromourcamerarecordings, itap‐pearedthatharborsealsmainlysearchedforpreyalongtheseabot‐tom, and, as inVolpov et al. (2015),moved their heads from sidetoside inaswayingmotionwhileswimmingenergeticallyclosetothebottom.Unfortunately, itwasdifficulttoseewhetherthesealcaughtanypreyasthecamerawasmountedonthebackoftheseal.Hence, future camera/accelerometer deployments with the cam‐eraplacedfurtherforwardwillhelpdetermineactualpreycaptureevents.

Biologgingtechnologyhasadvanceddramaticallyoverthepastdecadestakingadvantageoftechnologicaldevelopmentsincom‐puters,datastorage,andbatterycapacity,andtheminiaturizationofmobiledevicesandsensors.Thesemethodsarerevolutionizingmarinemammalscience,sheddinglightonthebehaviorsofanimalsandtheenvironmentstheyencounterwhenoutofsight.Withthisstudy,wedemonstratethepotentialuseofmulti‐weeksoundandmovementrecordingtagsonsealstostudyexposuretonoise,nat‐ural undisturbedbehavior, andhow this behaviormay change inresponse to anthropogenic activities at sea. The combination oflong‐duration datawith high temporal resolution sensorsmakesit possible to quantify the frequency and duration of anthropo‐genicdisturbancesandhowtheymayaffectcommunicationspaceandessentialbehaviorslikerestingandforaging.Conservationofpinnipedshasmainlybeenfocusedontheirhauloutsites,whichis due to the limited knowledge of how underwater noise mayaffect the animals at sea. New biologging techniques allow thequantificationoftimebudgets,andpotentiallyenergybudgets,forindividual seals,which can then feed intomodels for populationconsequencesofdisturbance(Nabe‐Nielsenetal.,2018).Thiscom‐binationofhigh‐qualitybiologgingdataandmodelingisessentialfor identifying andmanaging disturbing anthropogenic activitiesbothintimeandspace,whichmaycompromisethelong‐termsur‐vivalanddistributionofmarinemammalpopulations.Managementinterventionscould include reducing impactofvesselsby reduc‐ing speed, redirecting ship routes,or setting standards fornoiseemission.

ACKNOWLEDGMENTS

Wewouldliketothankthelargeteamofhelpersinvolvedinthesealcatchesandtag‐retrieval,especiallyAndersGalatius,RuneDietzandIlektraAmiralifromAarhusUniversity,aswellasHeatherM.Vance

12  |     MIKKELSEN Et aL.

from University of St. Andrews for aid in preparations of equip‐ment and data evaluation. René Swift and Mikkel Villum Jensenhelped with manufacturing of the tags. This study was fundedby the German Federal Agency of Nature Conservation underthe project “Effects of underwater noise on marine vertebrates”(Cluster 7, Z1.2‐53302/2010/14) and “UnderWaterNoise Effects– UWE” (Project numbers FKZ 3515822000). The catches werefunded and supported by the Schleswig‐Holstein's Government‐OwnedCompanyforCoastalProtection,NationalParksandOceanProtection. The Danish seals were tagged under the permissionfromtheDanishNatureAgency(SN‐0005)andtheAnimalWelfareDivision (Ministry of Justice, 2015‐15‐0201‐00549).MJwas sup‐ported for developmentof the tagsby aMarie Sklodowska‐CuriecareerintegrationgrantandbytheMarineAllianceforScienceandTechnologyScotland.PTMandDMWwerepartly supportedbyalargeframegrantfromtheDanishNationalResearchCouncil.DMWwasalsosupportedbyanOfficeofNavalResearchgranttoJeremyGoldbogenatStanfordUniversity.

CONFLICT OF INTEREST

Nonedeclared.

AUTHORS’  CONTRIBUTIONS

LM,MJ, PTM, and JT conceived the ideas.MJ designed the tagswhichwereadaptedforrecoverybyLMandJT.LM,DMW,AvN,US,andJTcollectedthedata;LM,MJ,andDMWanalyzedthedata.Allauthorscontributedcritically tothedraftsandgavefinalapprovalforpublication.

DATA ACCESSIBILITY

Datasetsgeneratedduring the current study is available atDryadDigitalRepository:https://doi.org/10.5061/dryad.8s75sg6.

ORCID

Jonas Teilmann https://orcid.org/0000‐0002‐4376‐4700

REFERENCES

AguilarSoto,N.,Johnson,M.,Madsen,P.T.,Tyack,P.L.,Bocconcelli,A.,&FabrizioBorsani,J.(2006).Doesintenseshipnoisedisruptforagingin deep‐diving Cuvier's beakedwhales (Ziphius cavirostris)?Marine Mammal Science,22,690–699.

Andersen, S. M., Teilmann, J., Dietz, R., Schmidt, N. M., & Miller, L.A. (2012). Behavioural responses of harbour seals to human‐in‐duced disturbances. Aquatic Conservation: Marine and Freshwater Ecosystems,22,113–121.https://doi.org/10.1002/aqc.1244

Andersen, S. M., Teilmann, J., Dietz, R., Schmidt, N. M., & Miller, L.A. (2014). Disturbance‐induced responses of VHF and satellitetagged harbour seals.Aquatic Conservation: Marine and Freshwater Ecosystems,24,712–723.https://doi.org/10.1002/aqc.2393

Arcalís‐Planas,A.,Sveegaard,S.,Karlsson,O.,Harding,K.C.,Wåhlin,A.,Harkonen,T.,&Teilmann,J.(2015).LimiteduseofseaicebytheRossseal(Ommatophoca rossii),inAmundsenSea,Antarctica,usingtelem‐etryandremotesensingdata.Polar Biology,38,445–461.https://doi.org/10.1007/s00300‐014‐1602‐y

Blackwell, S.B., Lawson, J.W.,&Williams,M.T. (2004).Tolerancebyringed seals (Phoca hispida) to impact pipe‐driving and construc‐tionsoundsatanoilproductionisland.The Journal of the Acoustical Society of America,115,2346–2357.

Burgess,W. C., Tyack, P. L., Le Boeuf, B. J., & Costa, D. P. (1998). Aprogrammable acoustic recording tag and first results from free‐ranging northern elephant seals. Deep‐Sea Research Part Ii‐Topical Studies in Oceanography, 45, 1327–1351. https://doi.org/10.1016/S0967‐0645(98)00032‐0

Carter, M. I. D., Bennett, K. A., Embling, C. B., Hosegood, P. J., &Russell, D. J. F. (2016). Navigating uncertain waters: A critical re‐view of inferring foraging behaviour from location and dive datain pinnipeds. Movement Ecology, 4, 25. https://doi.org/10.1186/s40462‐016‐0090‐9

Costa,D.P.,Crocker,D.E.,Gedamke,J.,Webb,P.M.,Houser,D.,Blackwell,S.B.,…LeBoeuf,B.J.(2003).Theeffectofalow‐frequencysoundsource(AcousticThermometryofOceanClimate)onthedivingbe‐havior of juvenile northern elephant seals,Mirounga angustirostris. Journal of the Acoustical Society of America,113,1155–1165.

Crocker,D.E.,Boeuf,B.J.L.,&Costa,D.P.(1997).Driftdivinginfemalenorthernelephantseals:Implicationsforfoodprocessing.Canadian Journal of Zoology,75,27–39.https://doi.org/10.1139/z97‐004

Cunningham, K. A., & Reichmuth, C. (2016). High‐frequency hearingin seals and sea lions. Hearing Research, 331, 83–91. https://doi.org/10.1016/j.heares.2015.10.002

Deecke,V.B.,Slater,P.J.B.,&Ford,J.K.B.(2002).Selectivehabituationshapesacousticpredatorrecognition inharbourseals.Nature,420,171.https://doi.org/10.1038/nature01030

DeRuiter,S.,Southall,B.L.,Calambokidis,J.,Zimmer,W.M.X.,Sadykova,D.,Falcone,E.A.,…Tyack,P.L.(2013).FirstdirectmeasurementsofbehaviouralresponsesbyCuvier’sbeakedwhalestomid‐frequencyactivesonar.Biological Letters,9,20130223.https://doi.org/10.1098/rsbl.2013.0223

Edrén,S.M.E.,Andersen,S.M.,Teilmann,J.,Carstensen,J.,Harders,P.B.,Dietz,R.,&Miller,L.A.(2010).TheeffectofalargeDanishoff‐shorewindfarmonharborandgraysealhaul‐outbehavior.Marine Mammal Science,26,614–634.

Fletcher,S.,LeBoeuf,B.J.,Costa,D.P.,Tyack,P.L.,&Blackwell,S.B.(1996).Onboard acoustic recording fromdivingnorthern elephantseals. The Journal of the Acoustical Society of America, 100, 2531–2539.https://doi.org/10.1121/1.417361

Fregosi,S.,Klinck,H.,Horning,M.,Mellinger,D.K.,Costa,D.P.,Mann,D.A.,…Huckstadt,L.(2013).Useofananimal‐borneactiveacoustictagtoconductminimally‐invasivebehavioralresponsestudies.The Journal of the Acoustical Society of America,134,4044–4044.https://doi.org/10.1121/1.4830759

Gallon,S.,Bailleul,F.,Charrassin,J.B.,Guinet,C.,Bost,C.A.,Handrich,Y.,&Hindell,M. (2013). Identifying foragingevents indeepdivingsouthernelephantseals,Mirounga leonina,usingaccelerationdataloggers.Deep Sea Research Part II: Topical Studies in Oceanography,88,14–22.https://doi.org/10.1016/j.dsr2.2012.09.002

Génin,A.,Richard,G.,Jouma'a,J.,Picard,B.,ElKsabi,N.,VacquiéGarcia,J.,&Guinet,C.(2015).Characterizationofpostdiverecoveryusingsoundrecordingsanditsrelationshiptodiveduration,exertion,andforagingeffortofsouthernelephantseals(Miroungaleonina).Marine Mammal Science,31,1452–1470.

Götz,T.,&Janik,V.M. (2010).Aversivenessofsounds inphocidseals:Psycho‐physiological factors, learning processes and motiva‐tion. Journal of Experimental Biology, 213, 1536–1548. https://doi.org/10.1242/jeb.035535

     |  13MIKKELSEN Et aL.

Harris,R. E.,Miller,G.W.,&Richardson,W. J. (2001). Seal responsesto airgun sounds during summer seismic surveys in the AlaskanBeaufort Sea. Marine Mammal Science, 17, 795–812. https://doi.org/10.1111/j.1748‐7692.2001.tb01299.x

Harvey,J.T.,&Goley,D.(2011).Determiningacorrectionfactorforae‐rialsurveysofharborsealsinCalifornia.Marine Mammal Science,27,719–735.https://doi.org/10.1111/j.1748‐7692.2010.00446.x

Heaslip,S.G.,Bowen,W.D.,&Iverson,S.J.(2014).Testingpredictionsofoptimaldivingtheoryusinganimal‐bornevideofromharbourseals(Phocavitulinaconcolor).Canadian Journal of Zoology,92,309–318.

Heerah,K.,Hindell,M.,Guinet,C.,&Charrassin,J.‐B.(2014).Anewmethodtoquantifywithindiveforagingbehaviourinmarinepredators.PLoS ONE,9,e99329.https://doi.org/10.1371/journal.pone.0099329

Hemilä,S.,Nummela,S.,Berta,A.,&Reuter,T.(2006).High‐frequencyhearinginphocidandotariidpinnipeds:Aninterpretationbasedoninertialandcochlearconstraints.Journal of the Acoustical Society of America,120,3463–3466.https://doi.org/10.1121/1.2372712

Henry,E.,&Hammill,M.O.(2001).Impactofsmallboatsonthehauloutactivityofharbour seals (Phoca vitulina) inMétisBay,StLawrenceEstuary,Québec,Canada.Aquatic Mammals,27,140–148.

Hermannsen,L.,Beedholm,K.,Tougaard,J.,&Madsen,P.T.(2014).Highfrequencycomponentsofshipnoise inshallowwater: Implicationsforharborporpoises (Phocoena phocoena). Journal of the Acoustical Society of America,136,1640–1653.

Hildebrand,J.(2005).Impactsofanthropogenicsound.InJ.E.Reynolds,W.F.Perrin,R.R.Reeves,S.Montgomery,&T.J.Ragen(Eds.),Marine mammal research: Conservation beyond Crisis(pp.101–124).Baltimore,MD:TheJohnsHopkinsUniversityPress.

Hildebrand,J.A.(2009).Anthropogenicandnaturalsourcesofambientnoiseintheocean.Marine Ecology Progress Series,395,5–26.https://doi.org/10.3354/meps08353

Jansen,J.K.,Boveng,P.L.,Dahle,S.P.,&Bengtson,J.L.(2010).ReactionofHarborSealstoCruiseShips.Journal of Wildlife Management,74,1186–1194.https://doi.org/10.1111/j.1937‐2817.2010.tb01239.x

Jeanniard‐du‐Dot,T.,Trites,A.W.,Arnould,J.P.Y.,Speakman,J.R.,&Guinet,C.(2017).Activity‐specificmetabolicratesfordiving,transit‐ing,andrestingatseacanbeestimatedfromtime–activitybudgetsinfree‐rangingmarinemammals.Ecology and Evolution,7,2969–2976.https://doi.org/10.1002/ece3.2546

Jeffries,S.J.,Brown,R.F.,&Harvey,J.T.(1993).Techniquesforcaptur‐ing,handlingandmarkingharbourseals.Aquatic Mammals,19,21–25.

Jensen, F.H., Bejder, L.,Wahlberg,M., Aguilar Soto,N., Johnson,M.,&Madsen, P. T. (2009).Vessel noise effects ondelphinid commu‐nication.Marine Ecology Progress Series, 395, 161–175. https://doi.org/10.3354/meps08204

Johnson,M.,AguilarSoto,N.,&Madsen,P.T.(2009).Studyingthebe‐haviourandsensoryecologyofmarinemammalsusingacousticre‐cording tags: A review.Marine Ecology Progress Series,395, 55–73.https://doi.org/10.3354/meps08255

Johnson, M., Partan, J., & Hurst, T. (2013). Low complexity loss‐less compression of underwater sound recordings. The Journal of the Acoustical Society of America, 133, 1387–1398. https://doi.org/10.1121/1.4776206

Johnson,M.,&Tyack,P.L. (2003).Adigitalacoustic recording tag formeasuring the response of wild marine mammals to sound. IEEE Journal of Oceanic Engineering, 28, 3–12. https://doi.org/10.1109/JOE.2002.808212

Kastelein,R.,Helder‐Hoek,L.,Gransier,R.,Terhune,J.M.,Jennings,N.,&DeJong,C.A.F.(2015).Hearingthresholdsofharborseals(Phoca vitulina)forplaybacksofsealscarersignals,andeffectsofthesignalsonbehaviour.Hydrobiologia,756,75–88.

Kastelein, R. A., van der Heul, S., Terhune, J. M., Verboom,W. C., &Triesscheijn, R. J. V. (2006a). Deterring effects of 8–45 kHz tonepulses on harbour seals (Phoca vitulina) in a large pool. Marine Environmental Research,62,356–373.

Kastelein,R.A.,vanderHeul,S.,Verboom,W.C.,Triesscheijn,R.J.V.,&Jennings,N.V.(2006b).Theinfluenceofunderwaterdatatransmis‐sionsoundsonthedisplacementbehaviourofcaptiveharbourseals(Phoca vitulina).Marine Environmental Research,61,19–39.

Kvadsheim,P.H.,DeRuiter,S.,Sivle,L.D.,Goldbogen,J.,Roland‐Hansen,R.,Miller,P.J.O.,…Southall,B.(2017).Avoidanceresponsesofminkewhalesto1–4kHznavalsonar.Marine Pollution Bulletin,121,60–68.https://doi.org/10.1016/j.marpolbul.2017.05.037

Lonergan,M., Duck, C.,Moss, S.,Morris, C., & Thompson, D. (2013).Rescalingofaerialsurveydatawithinformationfromsmallnumbersoftelemetrytagstoestimatethesizeofadecliningharboursealpop‐ulation.Aquatic Conservation: Marine and Freshwater Ecosystems,23,135–144.https://doi.org/10.1002/aqc.2277

Madsen,P.T.,Johnson,M.,Miller,P.J.O.,AguilarSoto,N.,Lynch,J.,&Tyack,P. (2006).Quantitativemeasuresofair‐gunpulses recordedonspermwhales(Physeter macrocephalus)usingacoustictagsduringcontrolledexposureexperiments.The Journal of the Acoustical Society of America,120,2366–2379.

Mathevon,N.,Casey,C.,Reichmuth,C.,&Charrier,I. (2017).NorthernElephant Seals Memorize the Rhythm and Timbre of TheirRivals' Voices. Current Biology, 27, 2352–2356.e2352. https://doi.org/10.1016/j.cub.2017.06.035

McCauley,D. J.,Pinsky,M.L.,Palumbi,S.R.,Estes, J.A., Joyce,F.H.,& Warner, R. R. (2015). Marine defaunation: Animal loss in theglobal ocean. Science, 347, 1255641. https://doi.org/10.1126/science.1255641

McConnell, B., Beaton, R., Bryant, E.,Hunter, C., Lovell, P., &Hall, A.(2004).Phoninghome‐anewGSMmobilephonetelemetrysystemtocollectmark‐recapturedata.Marine Mammal Science,20,274–283.https://doi.org/10.1111/j.1748‐7692.2004.tb01156.x

Miller, P. J.O., Johnson,M. P.,Madsen, P. T., Biassoni,N.,Quero,M.,&Tyack,P.L.(2009).Usingat‐seaexperimentstostudytheeffectsofairgunsontheforagingbehaviorofspermwhales intheGulfofMexico.Deep Sea Research Part I: Oceanographic Research Papers,56,1168–1181.https://doi.org/10.1016/j.dsr.2009.02.008

Nabe‐Nielsen, J., Beest, F.M., Grimm, V., Sibly, R.M., Teilmann, J., &Thompson, P.M. (2018). Predicting the impacts of anthropogenicdisturbances on marine populations. Conservation Letters, 11(5),e12563.https://doi.org/10.1111/conl.12563

Nowacek, D. P., Johnson,M. P., & Tyack, P. L. (2004). North Atlanticrightwhales(Eubalaena glacialis)ignoreshipsbutrespondtoalertingstimuli.Proceedings of the Royal Society of London. Series B: Biological Sciences,271,227–231.

Nowacek,D. P., Thorne, L.H., Johnston,D.W.,&Tyack, P. L. (2007).Responsesofcetaceanstoanthropogenicnoise.Mammal Review,37,81–115.https://doi.org/10.1111/j.1365‐2907.2007.00104.x

Osinga, N., Nussbaum, S. B., Brakefield, P.M., & Udo deHaes, H. A.(2012). Response of common seals (Phoca vitulina) to human dis‐turbances in the Dollard estuary of theWadden Sea.Mammalian Biology – Zeitschrift Für Säugetierkunde, 77, 281–287. https://doi.org/10.1016/j.mambio.2012.02.005

Ramasco,V.,Biuw,M.,&Nilssen,K.T.(2014).Improvingtimebudgetes‐timatesthroughthebehaviouralinterpretationofdiveboutsinhar‐bourseals.Animal Behaviour,94,117–134.https://doi.org/10.1016/j.anbehav.2014.05.015

Richardson,W.J.,Greene,C.R.,Malme,C.I.,&Thomson,D.H.(1995).Marine mammals and noise.SanDiego:AcademicPress.

Russell,D.J.F.,Hastie,G.D.,Thompson,D.,Janik,V.M.,Hammond,P.S.,Scott‐Hayward,L.A.S.,…McConnell,B.J.(2016).Avoidanceofwindfarmsbyharboursealsislimitedtopiledrivingactivities.Journal of Applied Ecology,56,1642–1652.https://doi.org/10.1111/1365‐2664.12678

Russell,D. J. F.,McClintock,B. T.,Matthiopoulos, J., Thompson, P.M.,Thompson,D.,Hammond,P.S.,…McConnell,B. J. (2015). Intrinsicandextrinsicdriversofactivitybudgetsinsympatricgreyandharbourseals.Oikos,124,1462–1472.https://doi.org/10.1111/oik.01810

14  |     MIKKELSEN Et aL.

Schusterman,R.J.,Levenson,D.H.,Reichmuth,C.J.,&Southall,B.L.(2000).Whypinnipedsdon'techolocate.Journal of the Acoustical Society of America,107,2256–2264.https://doi.org/10.1121/1.428506

Sivle,L.D.,Kvadsheim,P.H.,Fahlman,A.,Lam,F.P.A.,Tyack,P.L.,&Miller,P.J.O.(2012)Changesindivebehaviorduringnavalsonarex‐posureinkillerwhales,long‐finnedpilotwhales,andspermwhales.Frontiers in Physiology,3,400.

Thompson,D.,Hammond,P.S.,Niceolas,K.S.,&Fedak,M.A. (1991).Movements,divingandforagingbehaviourofgreyseals(Halichoerus grypus).Journal of Zoology,224,223–232.

Tyack, P. L., Gordon, J., & Thompson, D. (2004). Controlled exposureexperiments to determine the effects of noise on marine mam‐mals. Marine Technology Society Journal, 37, 41–53. https://doi.org/10.4031/002533203787537087

Tyack,P.L.,Zimmer,W.M.X.,Moretti,D.,Southall,B.L.,Claridge,D.E., Durban, J.W., … Boyd, I. L. (2011). Beakedwhales respond tosimulatedandactualnavysonar.PLoS ONE,6,e17009.https://doi.org/10.1371/journal.pone.0017009

vanBeest,F.M.,Teilmann,J.,Hermannsen,L.,Galatius,A.,Mikkelsen,L., Sveegaard, S., …Nabe‐Nielsen, J. (2018). Fine‐scalemovementresponses of free‐ranging harbour porpoises to capture, taggingandshort‐termnoisepulsesfromasingleairgun.Royal Society Open Science,5,170110.https://doi.org/10.1098/rsos.170110

VanParijs,S.M.,Hastie,G.D.,&Thompson,P.M.(1999).Geographicalvariation in temporalandspatialvocalizationpatternsofmalehar‐bour seals in themating season.Animal Behaviour,58, 1231–1239.https://doi.org/10.1006/anbe.1999.1258

Volpov,B.L.,Hoskins,A.J.,Battaile,B.C.,Viviant,M.,Wheatley,K.E.,Marshall,G.,…Arnould,J.P.Y.(2015).IdentificationofPreyCapturesinAustralianFurSeals(Arctocephalus pusillusdoriferus)UsingHead‐MountedAccelerometers:FieldValidationwithAnimal‐BorneVideoCameras.PLoS ONE,10,e0128789.https://doi.org/10.1371/journal.pone.0128789

Watanabe,Y.Y.,Baranov,E.A.,&Miyazaki,N. (2015).Driftdivesandprolonged surfacing periods in Baikal seals: Resting strategies inopenwaters?The Journal of Experimental Biology,218,2793.

Wisniewska,D.M., Johnson,M.,Teilmann, J., Siebert,U.,Galatius,A.,Dietz,R.,&Madsen,P.T.(2018).Highratesofvesselnoisedisruptforaginginwildharbourporpoises(Phocoena phocoena).Proceedings of the Royal Society B: Biological Sciences,285.

Wisniewska,D.M.,Johnson,M.,Teilmann,J.,Rojano‐Donate,L.,Shearer,J., Sveegaard, S.,…Madsen, P. T. (2016).Ultra‐high foraging ratesofharborporpoisesmakethemvulnerabletoanthropogenicdistur‐bance. Current Biology, 26, 1441–1446. https://doi.org/10.1016/j.cub.2016.03.069

Ydesen,K.,Wisniewska,D.M.,Hansen, J.D., Beedholm,K., Johnson,M.,&Madsen,P.T. (2014).Whata jerk:Preyengulfment revealedbyhigh‐rate,super‐cranialaccelerometryonaharbourseal (Phocavitulina).Journal of Experimental Biology,217,2239–2243.https://doi.org/10.1242/jeb.100016

SUPPORTING INFORMATION

Additional supporting information may be found online in theSupportingInformationsectionattheendofthearticle.

How to cite this article:MikkelsenL,JohnsonM,WisniewskaDM,etal.Long‐termsoundandmovementrecordingtagstostudynaturalbehaviorandreactiontoshipnoiseofseals.Ecol Evol. 2019;00:1–14. https://doi.org/10.1002/ece3.4923