Event-dominated transport, provenance, and burial of organic carbon in the Japan Trench

(1)

Research Collection

Journal Article

Event-dominated transport, provenance, and burial of organic carbon in the Japan Trench

Author(s):

Schwestermann, Tobias; Eglinton, Timothy I.; Haghipour, Negar; McNichol, Ann P.; Ikehara, Ken; Strasser, Michael

Publication Date:

2021-06-01 Permanent Link:

https://doi.org/10.3929/ethz-b-000477351

Originally published in:

Earth and Planetary Science Letters 563, http://doi.org/10.1016/j.epsl.2021.116870

Rights / License:

Creative Commons Attribution 4.0 International

This page was generated automatically upon download from the ETH Zurich Research Collection. For more information please consult the Terms of use.

ETH Library

(2)

Earth and Planetary Science Letters 563 (2021) 116870

Contents lists available atScienceDirect

Earth and Planetary Science Letters

www.elsevier.com/locate/epsl

Event-dominated transport, provenance, and burial of organic carbon in the Japan Trench

T. Schwestermann

^a^,^∗

, T.I. Eglinton

^b

, N. Haghipour

^b^,^c

, A.P. McNichol

^d

, K. Ikehara

^e

, M. Strasser

^a^,^f

aInstituteofGeology,UniversityofInnsbruck,Innrain52,6020Innsbruck,Austria bGeologicalInstitute,ETHZürich,Sonneggstrasse5,8092Zürich,Switzerland

cLaboratoryofIonBeamPhysics,ETHZürich,Otto-Stern-Weg5,8093Zürich,Switzerland

dWoodsHoleOceanographicInstitution,NationalOceanSciencesAcceleratorMassSpectrometryFacility,WoodsHole,MA02543-1539,UnitedStates eGeologicalSurveyofJapan,NationalInstituteofAdvancedIndustrialScienceandTechnology(AIST),TsukubaCentral7,1-1-1Higashi,Tsukuba305-8567,Japan fMARUM–CenterforMarineEnvironmentalSciences,UniversityofBremen,LeobenerStr.8,28359Bremen,Germany

a rt i c l e i n f o a b s t r a c t

Articlehistory:

Received10June2020

Receivedinrevisedform7November2020 Accepted1March2021

Availableonline24March2021 Editor:Y.Asmerom

Keywords:

carbonisotopes carbonprovenance hadalzoneevent-stratigraphy carbontransfer

JapanTrench rampedPyr/Ox

The delivery of organic carbon (OC) to the ocean’s deepest trenches in the hadal zone is poorly understood, but may be important for the carbon cycle, contain crucial information on sediment provenanceandevent-relatedtransportprocesses,andprovideageconstraintsonstratigraphicsequences inthis terminal sink. Inthis study, wesystematicallycharacterize bulk organicmatter (OM)and OC signatures(TOC/TN, δ¹³C, ¹⁴C),as well as thosefromapplication of serialthermal oxidation(ramped pyrolysis/oxidation) of sediment cores recovered along an entire hadal trench encompassing high stratigraphicresolutionrecordsspanning nearly2000years ofdeposition.We analyzetwocoresfrom thesouthern andnorthernJapan Trench,wheresubmarinecanyon systemslinkshelfwithtrench.We compareresultswithpreviouslypublisheddatafromthecentralJapanTrench,wherecanyonsystemsare absent.Ouranalysesenablerefineddatingofthestratigraphicrecordand indicatethat eventdeposits arisefrom remobilizationofrelatively surficialsediment coupledwith deepererosion along turbidity currentpathwaysinthesouthernandcentralstudysiteandfromcanyonflushingeventsinthenorthern studysite.Furthermore,ourfindingsindicatedepositionofpredominantlymarineOCwithinhemipelagic backgroundsedimentaswellasassociatedwitheventdepositsalongtheentiretrenchaxis.Thisimplies thatcanyonsystemsflankingtheJapanTrenchdonotserveasashort-circuitforinjectionofterrestrial OCtothehadalzone,andthattropicalcyclonesarenotmajoragentsforsedimentandcarbontransfer intothistrenchsystem.ThesefindingsfurthersupportpreviousJapan Trenchstudies interpretingthat event deposits originate from the landward trenchslope and are earthquake-triggered. Thevery low terrestrialOCinputintotheJapanTrenchcanbeexplainedbythesignificantdistancebetweentrench andhinterland(>180km),andthephysiographyofthecanyonsthatdonotconnecttocoastandriver systems.WesuggestthatdetailedanalyzesoflongsedimentaryrecordsareessentialtounderstandOC transfer,depositionandburialinhadaltrenches.

©²⁰²¹^TheÂuthor(s).^Published^byÊlsevier^B.V.^Thisîsânôpenâccessârticleûnder^the^CC^BY^license (http://creativecommons.org/licenses/by/4.0/).

1. Introduction

Although it has been more than half a century since the pi- oneering work of sampling and analyzing organic carbon (OC) contentofsurfacesedimentsindeep-waterhadaltrenches (water depths of>6000 m; Jumars andHessler (1976); Bartlett (2009)), the processes involved in the transport, deposition and burial of OC in oceanic trenches remain poorly understood. Nevertheless,

*

Correspondingauthor.

E-mailaddress:tobisch91@bluewin.ch(T. Schwestermann).

thesupply ofOCto the hadal zoneis considered to playan important role in the deep-marine carbon cycle and in supporting hadalecosystems (e.g.,Jamiesonet al., 2010; Kiokaet al.,2019a;

Nunoura et al., 2015). Detailed investigation on the quantiﬁca- tion ofOC mineralization andburial in the typically narrow and elongated oceanictrenches that comprise thisterminal sedimentary sinkhave onlyrecently emerged, owingto latest technolog- ical advances that allow for samplingandstudying the sea- and subseaﬂoorinsuch extremewaterdepths (e.g.,Baoet al., 2018b, 2019;Glud et al.,2013,2021;Leduc et al.,2016;Luoet al.,2017, 2018; Wenzhöfer et al.,2016). It has beendocumented that mi- https://doi.org/10.1016/j.epsl.2021.116870

0012-821X/©²⁰²¹^TheÂuthor(s).^Published^byÊlsevier^B.V.^Thisîsânôpenâccessârticleûnder^the^CC^BY^license(http://creativecommons.org/licenses/by/4.0/).

(3)

crobialbiomass,phytopigmentconcentrations,andinfaunalabun- dances aswell asbenthiccarbon mineralization rates are higher in the trench than their adjacent abyssal plain (e.g., Glud et al., 2013, 2021; Leduc et al., 2016; Luo et al., 2019). However, the remobilization processes and sediment and OC ﬂuxes to trench environments that sustain hadal ecosystemsare currentlypoorly constrained.

Ithasbeenspeculatedthatbecauseoftheirgeometryandprop- agating internal tides, OCis funneled towards thetrench bottom (Turnewitsch et al., 2014). However, recent observations in isolated trench basins or in canyons that connect coastal areas to deep-water trenches indicate that event-related sediment-gravity ﬂows triggered by tropical cyclones orearthquakes have the po- tentialtotransfer,depositandeventuallyburyvastamountsofOC inoceanictrenches (e.g.,triggered by the2008TyphoonMorakot offshore Taiwan (e.g.,Kaoet al., 2010), the 2016Kaikoura earth- quakeoffshore NewZealand (Mountjoyet al.,2018);orthe2011 Tohoku-okiearthquakeoffshoreJapan (Kiokaet al.,2019a)).There isa needforimprovedunderstanding ofcarbontransfertohadal trenches,includingpossibletransportpathways(e.g.,canyons,currents, gravitationalslope processes) and triggermechanisms that facilitate episodic OC transfer from the slope, shelf, or the terrestrial hinterland. An important andfrequently proposed trigger mechanism along subduction trenches is earthquake, which can remobilizeclastic materials andOCfromthe open adjacentslope andalongcanyonsincisingtheslope(e.g.,Baoet al.,2018b;Kioka et al., 2019a,b; Migeon et al.,2017). Theresultingearthquake re- latedevent-deposits in the trenchoftenconsist ofreworked ma- rinesedimentswithvaryingcontributionsofterrestrialandvolcan- oclasticcomponentsandOC,dependingonthegeologyandphys- iography of the subduction margin (e.g., Pouderoux et al., 2014;

Ikeharaet al.,2016;Goldﬁnger,2009;Schwestermannet al.,2020).

Tropical cyclones also serve as important trigger mechanisms to initiate sediment gravity flow and associatedOC transport, espe- cially when canyons connect trench basins with the coast and fluvial systems (e.g., Kao et al., 2010; Pope et al., 2017). Storm- and flood-induced eventdeposits in the deep-seahave beende- tectedby thepresenceofterrestrialplantremains,suchasleaves or woodfragments(Kao et al., 2010; Migeon et al., 2017). How- ever,ifmacroscopicplantremainsareabsent,furthergeochemical analysis, suchasdeterminingtheratiosoftotal OCtototalnitro- gen(TOC/TN)andcarbonisotopiccompositionsofOC,isnecessary todeterminetheprovenanceofOCandinferthedominantdriving mechanism.Kaoet al.(2010),forinstance,reportedfortheManila Trench (MT) that a deep-marine event deposit from a typhoon- floodtriggeredhyperpycnalflowhasverysimilarisotopicvaluesas the riversediments.Studiesinthe NewBritain Trench(NBT;Luo et al., 2019; Xiao et al.,2020) show overall a high deposition of terrestrialOCinthetrench(comprising61±^1%ôf^the^TOC)^due^to thesteep landwardtrenchslope(∼⁸^◦⁾ ând^the^proximity^to ^land (∼^{60 km).}

Above-mentionedstudiescoverdifferenttime-scalesastheyei- ther examine known events captured by instrumental data (e.g., sedimenttraps:Kaoet al.,2010;differentialbathymetry:Mountjoy et al.,2018)orintegratedatafromdatedsurfacecoresencompass- ingtherecentpast(<100years;Luoet al.,2019;Xiaoet al.,2020).

However, littleisknownabouthowtheprovenanceandﬂuxesof OCvaryoverlongertimescalesthat potentiallyspandepositional episodesin responseto differenttriggerprocessesthat mayhave differentrecurrencerates(e.g.,subduction-zoneearthquakesoccur atalowerfrequencythantropicalcyclones).Therefore,down-core studiescoveringeventdepositsthathaveoccurredoverseveralcy- cles ofpotential triggers (earthquakes,typhoons)are needed. For timescalesbeyond²¹⁰Pbdating,however,itischallengingtodate hadal stratigraphic records ofsediments deposited far below the carbonatecompensationdepth(CCD)(Jamiesonet al.,2010;Berger

et al.,1976),whichleadstothedissolutionofdateablecalcareous fossils.Inareaswherecalcareousfossilsare absent,¹⁴Cdating of bulkOCand,morerecently,thermalfractionsfromserialthermal oxidation (so-calledramped pyrolysis/oxidation,RPO) of OC have beenappliedtoprovideageconstraintsonstratigraphicsequences (e.g.,Baoet al.,2018b;Subtet al.,2016;Rosenheimet al.,2008).

PioneeringhadalzoneresearchinvestigatingOC¹⁴Csignatures hasbeenundertakeninthecentralJapanTrench(JT;7-8kmdeep, offshore NEJapan; Fig. 1), whereno canyons reaching the hadal zone incise the shelf or connect to rivers draining the hinterland. Extensiveeffort has been invested to understand sediment andcarbon delivery to the trenchlinked to the 2011 Magnitude 9Tohoku-oki earthquake(Bao et al.,2018b; Kioka et al.,2019a).

Also,olderearthquake-triggeredeventdepositsthatcorrelatetoa 1300-yearlong historyof megathrust earthquakes (Ikeharaet al., 2016)exhibitbulkOC¹⁴Cagesthatare3-6kyroldercomparedto therespectiveunder- andoverlyinghemipelagicbackgroundsedi- ment(Baoet al.,2018b).Thehemipelagicbackgroundstratashow arobustlinearrelationshipwithsedimentdepth,althoughtheage offsetto knownmarker bedsis ∼^1.6^kyr. ^Similar ^linear^relation- ship and age-offset for intra-event stratigraphic successions was alsoobserved incores fromthesouthern andnorthern JT (Kioka et al.,2019b),wherethehadaltrench,incontrasttothecentralJT, isconnectedtoshallowerwatersbytheNakaminatoandOgawara canyonsystems,respectively.

TheNakaminatocanyonispotentiallyfedbytheNakaandTone rivers,thelattercomprisingthelargestdrainageareainJapan.The catchmentareasofbothriverscanbeimpactedbyseveretropical cyclones,asoccurredbythe2019FaxaiandHagibistyphoonsand theAD 1856Edo-Ansei typhoonthat isconsidered asone ofthe strongesttyphoonsinJapanesehistory(Sakazakiet al.,2015).

Thus, the hadal JT provides an ideal study site to investigate temporal andspatial variabilityin carbon provenance andtrans- fer processes associated with event-related deposition in hadal trenches ona margin-widescale. Inthis study,we analyzed two 10m longsedimentcoresfromterminalsinks oftheNakaminato andOgawaracanyonsinthesouthernandnorthernJT,respectively.

Inordertoassesscarbon transferthroughcanyon systemsversus open slope lateralremobilizationand translocationprocesses, we compare results fromthese two cores with published datafrom thecentral JT (Baoet al., 2018b), where canyons are absent. We furtherexamined thecharacteristicsofsurficial sedimentsonthe adjacentlandwardslope.Throughthiscomparisonbetweendiffer- enttrench-fillbasinsandbetweentrenchbottomandtrenchslope settings,weseekto (1)constrain theprovenanceofOCalongthe entireJT(marinevsterrestrialOC),(2)toinfertriggermechanisms of event deposits (e.g., tropical cyclones vs. earthquakes), (3) to investigatethe influenceof sediment remobilizationandredistri- bution processes on bulk OC ¹⁴C characteristics of hadal trench sediments (e.g., hydraulic particle sorting; cf. Bao et al., 2019;

Hemingway et al., 2019, and (4) to reﬁne ¹⁴C-based age models forJTsedimentaryrecords.

2. JapanTrench(JT)

TheJTsubductionzoneislocatedEastofHonshu,Japan,where thePacificPlatesubductsbeneaththeOkhotskPlate.Dueto rela- tivelyhighconvergence ratesof∼86mm/a(DeMetset al., 2010) the area is frequently affected by strong earthquakes (e.g., the 2011Mw9.0Tohoku-okiearthquake). The trenchis alignedalong a NNE-SSW directionand bordersin the south onthe Izu-Bonin Trench. In the north, the JT intersects with the Kuril Trench at the Erimo Seamount (Fig. 1). The incoming Pacific plateis char- acterizedby NNE-SSW trendinghorst-graben structures linked to extension on the downward bending subducting plate, resulting inroughtopographywithtypical graben-fillandtrench-fillbasins

(4)

T. Schwestermann, T.I. Eglinton, N. Haghipour et al. Earth and Planetary Science Letters 563 (2021) 116870

Fig. 1.a)BathymetricoverviewmapoftheJapanTrench(Strasseret al.,2017;Kiokaet al.,2019a)betweentheDaiichiSeamountintheSouthandtheErimoSeamount intheNorthwith1000-mcontourlines.Redcrossesmarkcorelocationsontheslope,whereasgreencrossesmarkcorelocationsinthetrench.Blackboldlinesmarkthe NakaminatoandOgawaracanyonintheSouthandNorth,respectively.YellowshadedareasoffCapeErimoindicatescarpsofsubmarinelandslidesofunknownage.(note clockwise-rotationofthemapby90degrees)b)Bathymetricmapofthesouthernbasinwith50-mcontourlinesandcorelocationGeoB21804-1;c)Bathymetricmapofthe centralbasinwith50-mcontourlinesandcorelocationGeoB16431;d)Bathymetricmapofthenorthernbasinwith50-mcontourlinesandcorelocationGeoB21817c.(For interpretationofthecolorsintheﬁgure(s),thereaderisreferredtothewebversionofthisarticle.)

(Kioka et al.,2019a).Waterdepthsoftrenchbasinsare ∼^7,400 ^m in thenorth(Fig. 1b), 7,600 min thecentral basin (Fig. 1c),and 8,030minthesouthernbasin,northofDaiichi-Kashimaseamount (Fig. 1d)(Kiokaet al., 2019a). Thelandwardslope isdivided into a gently dipping upper slope with occurrence of isolated basins (e.g., Arai et al., 2014), and a more steeply dipping lower slope with ∼⁵^◦ ^(Kawamura ^{et al.,}^2012). ^The ^lower ^slope ^is ^character- ized by trench-parallel lineaments, which are related to reverse andnormalfaults(Tsujiet al.,2011).Theexistenceofamid-slope terrace,extendingfromthecentralJT(∼^5,500^m)^to^the^northern JT(∼^4,000^m),^might^relate^to^thedevelopmentofanaccretionary prism(e.g.,Tsujiet al.,2011).

On eastern Honshu, major river systems, such as the Tone and Naka rivers (SE), the Abukuma and Kitakami rivers (E), and the Mabechi river (NE) (Fig. 1) have the capability to transport terrestrial carbon fromtheir respective watersheds to the ocean.

Two canyon systems – the Ogawara canyon in the north and the Nakaminato in the south – connect the shelf area withthe trench(Fig.1).ThesecanyonsareclassiﬁedasTypeIIshelf-incising canyons (Harris and Whiteway, 2011) and potentially facilitate transfer of terrestrialOCto the hadal realm. Severalsmallchan- nel structures conﬁned to the lower and upper slope (i.e., Type IIIcanyons;HarrisandWhiteway, 2011)occur betweenthesetwo

TypeII canyons(e.g.,Kawamura et al.,2012). Calculated ﬂowac- cumulations within the JT (Kioka et al., 2019b) indicate that the centralbasin(Fig.1c)isnotaffectedbyacanyonsystem.

Sedimentalong the trenchandlowerslope consistsmainlyof diatomfrustulesanddiatomaceousmudduetohighregionalma- rine primary production supported by upwelling of nutrient-rich watersresultingfromthemixingofthewarmKuroshioandTsug- aru current withthe cold Oyashio current (Ikehara et al., 2016).

Ontheupperslope,abovethecarbonatecompensationdepth(CCD

∼^4,500 ^m,^Berger^{et al.,}^1976),^calcareous^anddiatomaceousmud with clastic and volcanic grains were found (e.g., Arita and Ki- noshita,1984; Strasser et al., 2017). Sediment intheNakaminato Canyonconsistsmainlyofclayey-siltfrequently interbeddedwith sandlayers and sandpatches, based on coresNT15-07-PL06 and PC06(JAMSTEC,2015).

The sedimentary succession in the entire JT is heavily af- fectedbyeventdeposits(e.g.,turbidites).Thesedepositshavebeen mappedandcorrelatedalongthetrenchusinghydro-acousticsub- bottom proﬁles and sediment cores (Ikehara et al., 2016; Kioka et al., 2019b). Several of the event layers have been dated by meansof tephrochronology andradiocarbon ofbulk OC andcor- relatedto major historic earthquakes, such as the 2011 Tohoku- okiearthquake, theAD1677 Empo-Boso-okiearthquake,AD 1454

(5)

Kyotoku earthquake, the AD 869 Jogan earthquake, and an older earthquakeinthe2^ndto3^rdcenturyBP(Ikeharaet al.,2016;Kioka et al.,2019b).Oneevent,however,potentiallycorrelatestotheAD 1856Edo-Anseityphoon(Kiokaet al.,2019b).

Event deposits triggered by the 2011 Tohoku-oki earthquake containhighexcess²¹⁰Pb(Ikeharaet al.,2016;Kiokaet al.,2019a;

McHughet al.,2016,2020)andweredescribedasproductsofsur- ﬁcialsedimentremobilization(Kioka et al.,2019a;McHughet al., 2016,2020).Thisearthquake-shakinginducedprocessremobilizes over wide areasonly a few cm of organic-rich surface sediment (e.g., Molenaar et al., 2019), which is then mobilized downslope as dilute ﬁne-grained turbidity currents into deeper basins. This remobilizationprocesscantransferrelativelylargeamountsofcar- bontothehadaltrench(Kiokaet al.,2019a).

3. Methods

3.1. Samplingmaterialandstrategy

We primarilyanalyzed piston coreGeoB21804-1(36.07093^◦N/

142.73408^◦E, 8025 m water depth) from the southern Japan Trench (JT), composite core GeoB21817c established from cores GeoB21817-1/-2(supplementarytextS1andﬁgureS1;40.39558^◦N/

144.42093^◦E, 7607 m water depth) from the northern JT and twosamplesfromcoreGeoB21818-1/-2(40.24648^◦N/143.81345^◦E, 3138 mwaterdepth),retrievedonthenorthernslope(Fig.1a).All threecoreswerecollectedduringR/VSonnecruiseSO251ain2016 (Strasseret al., 2017). Cores wereopened anddescribed onboard and were shipped to MARUM–Center for Marine Environmental Sciences, where they were stored at4^◦C andsubsequently sam- pled. Samplesused forfurther analysiswere storedat −²⁰^◦^C ^at ETHZürich.

Altogether, 24 samples fromGeoB21804-1 (south), 8 samples from GeoB21817c (north) and 2 samples from GeoB21818-1/- 2 were analyzed. The majority of these samples were collected within event deposits, to analyze bulk OC signatures. Sample preparationandmeasurementswere followingtheprocedures after Bao et al. (2018b) and Kioka et al. (2019b) (see details in sections3.2,3.4,3.5),inordertofacilitatecomparisonofpriorre- sultswithnewlyacquireddatafromthisstudy.

To investigatethe margin-wide bulk OC ¹⁴C ageoffset at the surface,surfacesediment(0-1cm)of5slopecores(Supplementary table S1)collected during Japanese andGermanresearch cruises, havebeensampledandanalyzed.Fortrenchcores,sampleswere chosen fromthinoxidizedlayerrightbelowthebaseofthe2011 Tohoku-oki earthquake event deposit TT1 representing the 2011 paleosurface(Ikeharaet al.,2016),sincebulkOCofcurrentsurface samples(0-1cm)mightbebiasedbythemixedcarbonwithinthe eventdeposit.

MolarTOC/TNratios(hereafter:TOC/TN)weremeasured onall bulk OC ¹⁴C samples (also the ones presented by Kioka et al.

(2019b) and Usami et al. (2021)). Stable carbon isotope compositions (δ¹³C)ofbulk OCwere determined on 20 samplescorre- spondingtohemipelagicbackgroundsedimentationandeventde- positsofcoresGeoB21804-1andGeoB21817c.Resultswere com- paredwithdatafromthe centralbasin(Baoet al.,2018b) tofur- therexaminethesedimentandcarbonﬂuxthroughthecanyons.

Fifteen selected samples of GeoB21804-1, GeoB21817c, and GeoB21818-1/-2weresubjecttoRPOtoderiveinformationonthe compositional and age heterogeneity within bulk organicmatter.

Sampleswereselectedtoenablecomparisonbetweenhemipelagic backgroundsedimentandindividualeventdeposits.Sampleprepa- ration and measurements followed procedures after Bao et al.

(2018b; details inthe following sections) to compare the results fromthecentralJT(Baoet al.,2018b)withourﬁndingsfromthe northernandsouthernJTandslope.

3.2.Samplepreparation1:HClfumigation

Samplesfor bulk OC ¹⁴C, TOC/TN, andbulk OC δ¹³C analyses were freeze-driedand powdered. 10-35mg (analysis dependent) of the powderedsamples were weighted into pre-combusted Ag capsules,placedon aceramic trayina desiccator,andfumigated for72h at 60^◦Cwith ∼³⁰ ^mlôf concentratedHCl (37%, metal- trace purity). Afterwards, the acidified samples were neutralized with∼²⁰^g^NaOH^pellets^forânother^72hât⁶⁰^◦^C.

3.3.Samplepreparation2:HClrinsing

Due to instrumental constraints at the time of δ¹³C analyses (see supplementary informationText S2), an additionalset of10 samplesofcoreGeoB21804-1were analyzed forbulkOC δ¹³Cto investigateinfluenceofthecanyon.Forthis,∼¹⁰^mgôf^powdered sampleswerefilledinto15-mlcentrifugaltubesandfilledwith∼³ mlof6molar HCl. Thereactionwasleft for∼¹²^hours. Âcidified sampleswere centrifuged for4 minuteswith3000 rotations/min to separate the supernatant liquid fromthe solid material. Sam- pleswere neutralizedwithdeionized waterandcentrifuged for5 minwith3000rotations/minforatleast3times,beforetheywere driedforthreedaysinan ovenat50^◦C.Oncedry,sampleswere groundwithanagatemortarandfilledintosmallvalves.

3.4.BulkOCanalysis

Fumigated samples were analyzed for TOC, TN, bulk OC ¹⁴C onthecoupledelementalanalyzerEA-IRMS-AMSonlinesystemat ETH Zürich (McIntyre et al., 2017). The 15 fumigated bulk sediment samples, processed on the RPO (section 3.5), were analyzed forbulkOCδ¹³Cutilizinganelementalanalyzercoupledto Precision-Isoprime.TensamplesrinsedwithHClwereanalyzedfor bulkOCδ¹³Cbymeans ofan elementalanalyzer(Thermo-Fisher- Delta-VwithFlashEA)oftheClimateGeologygroupatETHZürich.

(SeesupplementaryinformationTextS2fordetailedcalibrationof allmeasurements).All¹⁴Cresultsarereportedasuncalibrated¹⁴C bulkagesthroughoutthemanuscript.

3.5.Rampedpyrolysis/oxidation(RPO)

Inadditiontobulk-levelmeasurements,15bulksedimentsam- ples were subjected to ramped temperature pyrolysis/oxidation (RPO, Rosenheimet al., 2008) at theNational Ocean Science Ac- celerated Mass Spectrometer facility (NOSAMS) at Woods Hole OceanographicInstitution(WHOI).Fortheseanalyses,100-150 mg of acid-fumigated samples (as described in section 3.2) were loaded into a quartz reactor and were heated and oxidized at a constant temperature increase of 5^◦C min⁻¹ to ∼⁹⁹⁰^◦^C ^(Bao et al.,2018b;Rosenheimet al.,2008).Theresultinggaswastrans- ported out of the reactor by using a He-O₂-carrier gas (∼^8%

O2,totalflowrate 35 mL/min, oxidationmode)andpassed under isothermal condition a catalyst wire (∼⁸⁰⁰^◦^C), â ^chemical ^trap (∼⁴⁵⁰^◦^C; ^containing ^CuO, Âg, ând ^MnO ^granules) ând â ^water trap(−⁸⁶^◦^C) ⁱⁿ ôrder ^to ^purify ^the ^gas, ^to ^bind ôther ^combus- tion gases asCl2 and SO2, and to remove H2O, respectively. Af- terpassing a flow-throughinfrared CO2 analyzer (Sable-Systems- International-Inc.,CA-10a) tomeasure theCO2 concentration,the purifiedCO₂ of5temperatureintervals(T₁:170-320^◦C;T₂: 320- 391^◦C;T3:391-486^◦C;T4:486-570^◦C; andT5:570-915^◦C; after Baoet al.,2018b)was sequentiallyisolatedandtrappedintopre- combustedglasstubes. At thebeginning ofeach experimentand once within every temperature interval, leak checks were conducted. The purified CO₂ was measured as gas at the ETH AMS facility.

(6)

Fig. 2.a)CoreGeoB21804-1collectedinthesouthernbasinwithlitholog(afterStrasseret al.,2017),TOC/TN(triangles),bulkOC¹⁴Cages(circles),andinterpretationof eventdeposits(afterKiokaet al.,2019b).b)ReferencecoreGeob16431fromthecentralbasinwithlitholog(Ikeharaet al.,2016),bulkOCδ¹³CvaluesandbulkOC¹⁴Cages (dataofBaoet al.(2018b)).c)CoreGeoB21817cretrievedinthenorthernbasinwithlitholog(afterStrasseret al.,2017),TOC/TN,andbulkOC¹⁴Cages.Triangleswithred borderina)andc)werefurtheranalyzed forδ¹³C.BulkOC¹⁴Cageswithinthebackgroundsedimentina)andc)areafterKiokaet al.(2019b).FourbulkOC¹⁴Cages markedwithorangerectanglesareafterUsamiet al.(2021).Samplesmarkedwithred(a),green(b),andblue(c)arrowswerefurtherprocessedontheRPO(samplesin(b) byBaoet al.,2018b).

ResultingdatafromRPOwasfurthersubjecttoaregularizedin- versemethodtoestimate thedistributionofOCactivationenergy, whichisaproxyformolecularbondstrengthandchemicalstruc- tureoftheOC(Hemingwayet al.,2017).ThedistributionoftheOC activationenergywascalculatedbyusingthedefaultsettingofthe opensourcePythonpackage“rampedpyrox”(Hemingway,2018).

4. Results

4.1. BulkOCanalysis 4.1.1. GeoB21804-1(south)

Bulk OC ¹⁴C measurements reveal a wide spectrum of ages, ranging from1288±⁶⁰ ^to⁵⁰⁰⁹±⁹⁰^yr ^BP ^(Fig. ^2a). În^the ûpper partofthecore(<720.0cm),wherealinearrelationshipbetween inter-event-depositbackgroundsedimentandbulkOC¹⁴Cageshas beendocumented by Kiokaet al.(2019b),ournewagedata rep- resent mainly older bulk ages of the eventdeposits. Thickevent deposits areabsent fromthe lower part(>720.0 cm) ofthecore, andinsteadthisintervalcontainsonlythindepositsresultingfrom minorsedimentremobilizationprocesses.Forthissection,mostof our new bulk OC ¹⁴C data define the downward projected linear age increase trend of background sediment, with only two samples indicating old ages within a small sediment remobilization deposit(red arrow747.3 cm inFig. 2a). The sandy bases of event deposits show generallythe oldest ages (e.g., 5009±⁹⁰ ^yr BP at 274.75 cm) whereas the overlying homogeneous diatoma- ceousmudusuallyhaveyoungerages.TOC/TNrevealsvaluesfrom 5.78-9.67(mean=7.59,standarddeviation(sd)=0.59,n=35),andTOC

content from 0.54%-2.41% (mean=1.42% sd=0.45%, n=35; Fig. 2a, supporting informationtable S2). The δ¹³C measurements (fumigated)range between −²¹.22 ^and −²⁰.10 ^(mean=−²⁰.75 ^,^sd=0.38^,ⁿ⁼⁸⁾ ^and^the ^rinsed^samples^between−²¹.85 and−²¹.47 (mean=−²¹.65 ,sd=0.13 , n=10; supporting informationtableS2).

4.1.2. GeoB21817c(north)

Bulk OC ¹⁴C ages of core GeoB21817c range from 4648±⁷⁸ yr BP at the transition of hemipelagic background sediment to eventdepositto7173±⁹¹^yr^BP^within^the^event^deposit^(Fig.^2c).

Theseagescorrespondtotheupper7mofanapproximately10 m thickeventdeposit(Kioka et al., 2019b). The8 samplescollected within the event deposit show very constant ages that vary be- tween6679±⁸⁷^and⁷¹⁷³±⁹¹^yr ^BP^(mean=6975 ^yr ^BP).^TOC/TN rangebetween6.76and9.14(mean=7.79,sd=0.72,n=24),TOCcon- tentbetween0.9%and1.55%(mean=1.27%,sd=0.13%,n=24;Fig.2c), andδ¹³Coffumigatedsamplesbetween−²¹.96^and−¹⁹.83 (mean=−²⁰.95 sd=0.76 ,n=4; supporting information table S3).

4.1.3. Surfacesedimentsamples

In order to assess age offsets at the sediment surface, core- top (0-1 cm) samples were analyzed from the slope cores. Core GeoB21817cinthenorthernbasin wassampledatadepth inter- valof4to5.5cm,whereascoresGeoB21823 (samecorelocation asGeoB16431) and GeoB21804-1 from the central and southern basin,respectively,were sampledbelowthebaseofthe AD2011 eventdeposit,representingthesedimentsurfaceat2011.Thesam-

(7)

Fig. 3.Activationenergydistributionsofbackgroundsediments(BS)fromthefourdifferentcorelocationsGeoB21804-1southernJapanTrench(red),GeoB16431centralJapan Trench(green),GeoB21817cnorthernJapanTrench(blue),GeoB21818-2northernslope(cyan)insolidlines,andofeventdepositsofthe2011Tohoku-okiearthquakefrom thesouthernTrench(reddottedline)andthecentralJapanTrench(greendottedline).Thethreevertical,graybarsillustratetherangeofthetwoprimarypeaksbetween

∼¹³⁸ând¹⁵²^kJ/molând∼¹⁶⁷ând¹⁷⁴^kJ/mol)ând^the^secondary^peak^between^(240-245^kJ/mol).

ples yieldbulk OC¹⁴C agesvaryingbetween1431±⁷⁵^yr ^BPând 1796±⁷⁵ ^yr ^BP ^with ân âverage ôf ¹⁵⁸¹±⁷⁴ ^yr ^BP (supporting information table S1). We note that the oldest age (1796±⁷⁵ ^yr BP) is froma surface sample fromthe Nakaminato canyon.Nev- ertheless,thedataindicateaconsistentlysmallerageoffsetatthe northernstudysite(1451±⁷⁴^yr^BP)^thanât^the^centralând^south- ernstudy sites(1623±⁷⁸^yr^BPând¹⁷¹⁴±⁶⁹^yr^BP,respectively).

TOC/TN range between 7.02 and8.24 (mean=7.55, sd=0.43, n=8), TOC between0.80% and1.74% (mean=1.50%, sd=0.29%,n=8). δ¹³C was measured on 3 samplesand range between−²¹.85 ^and

−²⁰.70^.

4.2. Rampedpyrolysis/oxidation

Fifteensamplesfromdifferentsedimentationphases(i.e.,hemipelagic background sediment and event deposits) and different corelocations(i.e.,northern andsouthern JT andnorthernslope) were subjectto RPO. Thermograms generallyindicate two peaks attemperatures of316-341^◦C (peak 1)and411-443^◦C(peak 2), respectively (SupportinginformationFigure S2). In some samples (e.g.,GeoB21804-1177.25cm,GeoB21817c4.75cm),athirdsmall peak can be observedatabout737-745^◦C. ¹⁴C measurements of evolvedcarbon fractionscorresponding to the individual temperature intervals (T1-T5) from the 15 samples exhibit overall in- creasing¹⁴Cageswithincreasingtemperature(i.e.,fromT1 toT5; supporting information Figure S2). Sample GeoB21818-1 6.75cm representstheonlyexception,whereT3 showstheoldest¹⁴Cage andT5 youngest.

Apparent activation energies calculatedfrom thermogramsin- dicateawidedistributionfrom∼¹⁰⁰^kJ/mol^to∼³⁰⁰^kJ/mol^with similar peaks amongsamples (supporting informationﬁgure S3).

The results reveal striking similar peak distributions for background sediment in the north (GeoB21817 4.75 cm, 183.25 cm) and in the south (GeoB21804 177.25 cm) of the trench (Fig. 3).

Event deposits differ only partly in activation energy peak distribution. The primary peaksbetween ∼¹⁶⁷ ând¹⁷⁴^kJ/mol ând between∼¹³⁸ând¹⁵²^kJ/molâre êvident ⁱⁿâll ^samples^(Fig. ^3, and supporting information Figure S3). Smaller, secondary peaks cannotbeconstantlytracedthroughallsamples.

Fig. 4.Diagram(modiﬁedafterGoniet al.,2008andLambet al.,2006,andrefer- encestherein)indicatingtheprovenanceofthesamplesbasedonTOC/TNandδ¹³C, indicatingmarineorigin(POC=particulateorganiccarbon,SOM=soilorganicmat- ter,DOC=dissolvedorganicmatter).TrianglesindicateJapanTrenchsediments.

5. Discussion

5.1. Trench-wideconstraintsonthenatureandprovenanceofOC Marine organic matter (MOM) generally consists of abundant nitrogen-rich compounds (e.g., proteins), whereas terrestrial organic matter contains carbon-rich structural polymers (e.g., cel- lulose, lignin, cutin and cutan (e.g., Hedges et al., 1997). These biochemicalvariations,togetherwithdifferentcarbonisotopicfrac- tionationcharacteristics,canresultindistinctTOC/TNandisotopic compositions(δ¹³Cvalues)ofmarineandterrestrialorganicmatter (Fig. 4). TOC/TN andδ¹³Care thus widely usedto determinethe provenanceoforganicmatter(e.g.,Goni et al.,2008;Lamb et al., 2006).TOC/TNofMOMtypicallyvariesbetween5.78and9.67and δ¹³Cvaluesbetween−¹⁹ând−²²^,^whereas^TOC/TNôf^ter- restrialorganicmatter (C3-plants)are≥²⁰ândδ¹³C valuesrange between−²⁵and−²⁸(e.g.,Goniet al.,2008;Hedgeset al., 1997). δ¹³C valuesfor C4 ecosystems wouldbe significantly less negative.

(8)

Fig. 5.a)Sequencefrombackgroundsediment(177.25cm)toturbidite(244.75cmand274.75cm)ofcoreGeoB21804-1presentanageincreaseinall5temperaturefractions (T1–T5),indicatingremobilizationofslightlyoldercarbon.ThethermogramsshowanoveralldecreaseinTOCcontenttowardsthebaseoftheturbidite,especiallyofpeak 1,whichdisappearsinsample274.75cm.b)ThermogramsofhemipelagicbackgroundsedimentsoftheslopecoreGeoB21818-2/-1presentdecreasedTOCcontentwithin oldersediment,whilepeak1and2remainsimilar.

TOC/TNoftheanalyzedJapanTrench(JT)samplesvarybetween 6and10andbulkOCδ¹³Crangebetween−¹⁹.83ând−²¹.96 ^,^suggesting^that ^the^bulk ÔC^mainly ^consistsôf^MOM^(marine particulateorganiccarbon(POC),andmarine algae;Fig.4).Given thegeographic location andthelow TOC/TNvaluessignificant C4 contributioncanbeexcluded.Ourobservationthatsedimentinthe JTmainlycontainsMOM,inparticularmarinealgaeisthusinline withpreviousstudiesbasedonmolecularanalysesandstableiso- topes(Ishiwatariet al.,2000)andthebulkOCδ¹³Cdata(Baoet al., 2018b).

Further,activationenergydistributions(inverselymodeledfrom thermal decay of OC) of background sediment from the southern,centralandnorthernJTshowstrikinglysimilarprofiles(Fig.3 and supplementary figure S3). Also, within turbidites, the primary peaks inthe activation energy distribution mirror those of hemipelagic background sediment (Fig. 5a). Peaks below ∼¹⁶⁰ kJ/mol,representingmainlytemperaturefractionsT₁andT₂within thefirstpeakofthethermogramat∼^316-341^◦^C,^lie ⁱⁿ^the^range

ofthermal decomposition ofbiopolymers (e.g., proteins andcar- bohydrates) of marine algaeand marine POC (e.g., Sanchez-Silva et al.,2012). The activationenergy peak between∼¹⁶⁵^and¹⁸⁰ kJ/mol, representing thermogram peak ∼^411-443^◦^C ^and ^mainly temperaturefractionT₃,isattributedtoaggregatedand/or encap- sulatedPOC(Baoet al.,2018a,2019)aswellascarbonheldwithin mineralpores (Hemingwayet al.,2019), whereslightlymore energy is necessaryfor thermal decomposition.The small peaksat 240-245kJ/molareexpectedtocorrespondtothethirdsmallpeak inthethermogramat737-745^◦C.Thishightemperaturepeaklies in the range of the thermal decomposition ofcalcite polymorph (Karunadasaet al.,2019)andis,thus,inferred tostemfromrem- nantIC, whichescaped removalvia fumigation. Alternatively,the thirdpeakcouldrepresentliberation ofcarbontightlyheldwithin clayminerals during the temperature-steppedcombustion (Wang et al.,2016) orcould be minorcontributions by rock-derivedOC (e.g.,graphite).

(9)

TheseaboveanalysessuggestthatOCdepositionovertheentire trench-slope system is similar, regardless of the presence or ab- sence ofcanyons.Moreover, the similarities betweenhemipelagic background sediment and event deposits in terms of activation energies, TOC/TN and δ¹³C reveal OC withsimilar bonding envi- ronment (i.e., a common OM composition and sediment fabric) thatisprimarilyofmarineorigin.Thistrench-widesimilarityinOC composition potentiallyreﬂectsthehighlevelsofmarineprimary productionoverthisregion,implyingthateventdepositsoriginate primarilyfromadjacentslopesandthatthecanyonsdonottrans- fersigniﬁcantamountsofterrestrialOCtothetrench.

5.2. Triggermechanism(s)ofevent-relatedsedimentandcarbon transporttothetrench

Basedonourδ¹³CandTOC/TNdata,weconcludethat,incontrast to e.g., theMT (Kao et al.,2010), hyperpycnalflows associ- atedwithfloodevents(e.g.,duringtyphoonsthatresultinintense discharge of terrestrialorganic matter) do not reach the JT. This alsois inlinewiththe lackoftyphoon-relatedbreaks ofsubma- rinetelecommunicationcablesinourstudyarea,whichhavebeen recordedelsewhere(e.g.,aroundTaiwan,northernPhilippines,and southwestern Japan; Popeet al., 2017). Hence, the eventdeposit datedbyKiokaet al.(2019b) toAD1846+22/-25mightnotbere- lated to the AD 1856 Edo-Ansei Typhoon, according to our new result. Nevertheless, our investigation, as well as other studies in high-seismicity areas (e.g., Hikurangi Trough; Mountjoy et al., 2018) show that earthquakes are key events leading to transfer of large amounts ofsediment andcarbon to thedeep sea.Since theeventdepositsandhemipelagicbackgroundsedimentsexhibit similar bulk OC signals (i.e., TOC/TN or δ¹³C), we infer that all analyzed event deposits –including the ∼ÂD ¹⁸⁴⁶ êvent^(Kioka et al., 2019b) – are earthquake-triggered,andthat they originate fromthe trenchslope. Thesefindings corroborateprevious inter- pretationsbasedonevent-depositcorrelationwithhistoricalearth- quakes and tsunamis (Ikehara et al., 2016; Kioka et al., 2019b) as well as sediment provenance data by Schwestermann et al.

(2020);McHughet al. (2020) indicatingearthquake-triggeredsur- ﬁcialslopesedimentremobilization.

Theremaybeseveralreasonsforminorterrestrialcarbontrans- fer to theJT compared to other trenches (e.g.,MTandthe NBT).

Theseincludethephysiographyofthecanyonsandtheshelfwidth anddistancebetweenthecoastandtrench(NBT:∼⁶⁰^km^distance, average slopeangle∼⁸^◦^,^Xiao^{et al.,} ^2020; ^JT:^>180 ^km^distance, averageslopeangle<2.5^◦).Thecanyonsinourstudyareaarenei- therwelldeveloped,nordirectlyassociatedwithrivers(Harrisand Whiteway, 2011). Moreover, the head of theNakaminato Canyon is over 40 kmfrom the coast and, at a water depth of>400 m, does not incise the shelf, falling under the classiﬁcation of type III canyon(Harris andWhiteway,2011). Inthenorth, one branch of the Ogawara Canyon incises the shelf of north eastern Hon- shu. However, this canyon branch is interrupted by the Hidaka Trough,where turbiditycurrents mightlose theenergyand thus theabilityto injectsedimentandterrestrialcarbonintotheJT. A secondbranchoftheOgawaraCanyonconnectstothesouthcoast ofHokkaido(CapeErimo,Fig.1)andintersectstheHidakaTrough atitseasternend.Whilethisbranchmaysupportdirectsediment andcarbontransferto theJT, itbarely incisestheshelf, withthe canyonheadlocatedattheshelf edgemorethan40kmfromthe coastatawaterdepthof∼²⁰⁰^m.

5.3. ¹⁴CconstraintsonthesourceanddeliveryofOCtothehadaltrench Our new bulk OC ¹⁴C measurements on event deposits in thesouthern (coreGeoB21804-1) andnorthern (coreGeoB21817)

basin indicate overall older ages (≤³ ^kyr ^and ≤⁴ ^kyr, ^respectively)compared tothe respective bothoverlying andunderlying hemipelagic background sediment dated by Kioka et al. (2019b).

Suchagediscrepancies havepreviously beenfound inthecentral JT(∼³^kyr^and^<6^kyr^for^theKyotoku- andJoganEarthquakeevent deposits,respectively;Baoet al., 2018b), andindicate remobiliza- tionofpre-agedOC(Schwestermannet al.,2020).

BulkOC ¹⁴Cages ofcoreGeoB21804-1 (south)showsfourbig eventdeposits(marked incolor,afterKiokaet al.,2019b)andup to five smallerevent deposits (Fig. 2). However, the fine-grained topsofthelattereventsarebarelydistinguishablefromtheback- ground sediment. We therefore focus mainly on the four large event deposits. The first and most recent of these (from top to bottom), which corresponds to the 2011 Tohoku-oki earthquake (Kioka et al., 2019a), reveals the same bulk OC ¹⁴C age as that ofthe underlying sediment, with the exception of thevery base ofthis unit that exhibits an ∼^0.4 ^kyr âge^shift. ^Thisobservation indicatesthatfreshlydepositedsediment includingyoungOCwas remobilized,consistent withprior studies that found highexcess 210Pbinthiseventdeposit(Kiokaet al.,2019a).Incontrast,earlier (stratigraphicallylower)eventdepositsexhibitlarger¹⁴Cage-shifts (max.3ka),althoughthesandbasesoftheselayerstypicallyshow substantial age offsets of >1 kyr, whereas age offsets are gener- allylessthan0.5kyrfortheoverlyinghomogeneousdiatomaceous mud.

14C agesmeasured onCO2 forthe ﬁveindividual temperature fractions(T₁-T₅)fromRPOexhibitsimilar shiftsbetweensamples.

Forexample,inturbiditesamplesexhibitingashiftinbulkOC¹⁴C agesof1kyr,each corresponding temperaturefractionisalsoin- creasedby∼¹^kyr^(Fig.^5).^While^suchâge^shiftsⁱⁿ^bulkÔC^could be explained by preferential degradation of organic matter (e.g., Bao et al., 2019), this explanation is difficult to reconcile with concurrent shifts for all five temperature fractions. Alternatively, agediscrepancies,particularlyforcoarsersedimentsatthebaseof turbidites, mightresultfromhydraulic sorting.RPO thermograms show that peak 1 (316-341^◦C)decreases in the coarser basesof turbiditescompared topeak2(411-443^◦C)(Fig.5a,from177.25- 274.75 cm).Thisasymmetricdecrease furthersupports theinter- pretationof peak1 to reflectcontributions fromrelatively labile, low-densityPOCderivedfrommarineproductivity(see5.1above) that remains longer in suspension relative to coarser anddenser particles (e.g., sand and coarse silt) already deposited (Fig. 5a).

Thisisalsosupportedbythegrain-sizedistributionsandTOCcon- tent(supportinginformationFig.S4,tableS2).Peak2(411-443^◦C), in turn, is interpreted as more recalcitrant OC, or OC protected through close association with mineral grains, or encapsulation withinorgano-mineralaggregates(e.g.,Baoet al.,2018a;Heming- way et al., 2019, section 5.1). This could explain its persistence inthe sandy base, aswell asthe generallyolder¹⁴Cages ofT3- T5. This hypothesis is also corroborated by two thermogramsof hemipelagic background sediment with comparable OC composi- tionfromthenorthernslope(Fig.5b)whichshowamoreuniform decrease in TOC content with ¹⁴C ages of up to ∼^15,000 ^years.

Also, homogenous diatomaceousmud of the thick eventdeposit inthenorth(GeoB21817c,supporting informationFig.S3)clearly showa highercontributionofpeak1thanevident inthecoarser baseoftheturbiditeunitinthesouth(GeoB21804-1274.75 cm).

Whilecoarsergrain-sizecan thuspartlyexplainthelargerage shifts in bases of turbidites (e.g., GeoB21804-1 274 cm), it can- notexplainthesmalleragediscrepanciesintheﬁne-grainedupper partofturbidites. Theincorporationofslightlyoldercarbon,such asvia deepererosionofsandy turbiditycurrentsmightthus bea thirdexplanation.FollowingtheinterpretationsofSchwestermann et al.(2020) forthecentralJT,smallageshiftsdocumentedforthe eventdeposits of historic earthquakes may reﬂectremobilization

(10)

of surﬁcialslope sediments,from whichsandy turbidity currents canevolvethaterodeintodeeperstrata.

Core GeoB21817c(north) comprises asingle, massive(∼10 m thick)eventdepositwithbulk OC¹⁴Cages thatare∼⁴^kyr^older than the overlaying background sediment (Kioka et al., 2019b).

This age offset is near constant over the entire 7 meters of the eventdeposit, indicatingvery well mixed,homogenoussediment.

Thebaseoftheturbiditehasnotbeencored,howeverasandbase can be expected based on the acoustic reflections (Kioka et al., 2019b).Basedonobservationsfromeventdepositsinthesouthern andcentralJT, wewouldexpectthe sandbaseto exhibitalarger offset inbulk OC ¹⁴Cages. This thick eventdeposit mightrelate to a canyon flushing eventsimilar to that observed at theHiku- rangi Margin (Mountjoy et al., 2018), and likely originates from submarinelandslidesontheupperslope(Usamiet al.,2021).From geomorphicanalysesofthebathymetrymap(Fig. 1),weinterpret two potentiallandslide-scars offCapeErimo (supplementarydata figure S5) asa putative source forsuch a canyon flushing event.

However,theageofthesetwolandslidesisunknown.Alternatively, large-scale(overbroadareas)remobilizationofsurfacesediments, as observed on the northern trench slope (Ikehara et al., 2020;

Molenaar etal., 2019), evolving into sandy and erosive turbidity currents in the Ogawara canyon, could be a second explanation.

In eithercase, the constant bulk ageof ∼⁷^kyr ôf^the êvent^deposit found in four small trench basins (Usami et al., 2021), as well astheuniformbulkOC¹⁴Cagedistributionoverthe5tem- peraturefractions(T1-T5,thisstudy,supportinginformationfigure S3)pointstowardaverywell-mixed,homogenouseventdeposit.

5.4. RefineddatingandcorrelationofeventdepositsintheJapanTrench ThebulkOC¹⁴Cagesofsurface sedimentsamplescollectedin thetrenchandonthetrenchslope reveala decreasingageoffset fromsoutherntothenorthernsectionsofthestudyarea(1715±⁷⁰ yrBPinthesouth,1623±⁷⁸^yr^BPⁱⁿ^the^centralând¹⁴⁵¹±⁷⁴^yr BPinthenorth,section4.1.3,supplementarytableS1).Withinthe samesectionofthestudyarea,trenchslopeandtrenchsediments indicategenerallysimilarages(exceptinthesouth).Thelargeage offset in the south is strongly influenced by the surface sample of core NT15-07-PL06 retrieved within the Nakaminato Canyon.

Thissamplemaynotberepresentative,however,duetooldersed- imentandcarbontrappedwithinthecanyon.Whenjustconsider- ingtheageoffsetfromthesurface sampleinthesoutherntrench (1633±⁶⁴ ^yr) ît îs ^very ^similar ^to ^the ône ⁱⁿ ^the ^central ^study area. In contrast, surface sediments in the northern study area have smaller(∼¹⁷⁵¹⁴^C ^years) ^bulk ÔCâge ôffsets. ^Weâttribute thisspatialvariabilitytodifferencesinoceanographicsettingwith higherprimaryproductivityalongtheHokkaidocoastresultingin relativelyenhanceddeliveryofyoungerOCtothenorthernJT(Us- amiet al.,2021) comparedtoolder,preagedOC.Thegenerallysim- ilarsurface ageoffsetbetweentrenchslopeandtrenchsediments explainsthateventdepositsderivedfromsurficialsedimentremo- bilization (e.g., the 2011 Tohoku-oki earthquake event deposits), havethevery similarageoffsetasthedirectunder- andoverlay- inghemipelagicbackgroundsediment.

If we apply the newly constrained mean age-offset for the southerntrenchsite(1633bulk¹⁴Cyr,insteadof∼¹⁶⁰⁰^bulk¹⁴^C yrasusedbyKiokaet al.(2019b)),thepreviouslypublishedevent ageforthe deposit(AD1846(+22/-25))forwhichan earthquake vs. typhoon trigger couldnot be resolved (Kioka et al.,2019b) is now reﬁned to be approximately 33bulk ¹⁴C yearsyounger. Be- ingawareoftheuncertainties,thisslightlyyounger“ﬂoating”bulk 14Cagerangebetteralignswitheitherthehistoricallydocumented AD 1896 January M7.3 Ibaraki-oki, or the AD 1897 August M7.7 Sanriku-okiearthquakes thanwiththeearlierAD 1856Edo-Ansei Typhoon.Anearthquakeratherthanatyphoontriggeriscorrobo-

ratedbythemarinecarbonprovenanceofOCinthiseventdeposit (sections5.1and5.2).Applying correctedmeanageoffsetsinthe northern study area (i.e., 149 years younger than estimated by Kioka et al. (2019b)), the reﬁned age of the thick eventdeposit is 1.77(+0.49/-0.31) bulk OC ¹⁴C kyr BP,potentially coeval with tsunami deposits along the Iwate coast (see Fig. 1 forlocations) that suggest large earthquake in this area during the 2^nd or 3^rd CenturyAD(Takadaet al.,2016).

6. Conclusion

Thisinvestigationis,tothebestofourknowledge, theﬁrstto systematically characterize OC signatures in sediments that have been deposited along the axis of an entire hadal oceanic trench system(Japan Trench(JT)) over the past 2 millennia. Thesenew data permit examination of links between event-related carbon transferto hadal trenches and margin geomorphologyaswell as potential trigger mechanisms with different rates of recurrence (i.e.,lowerfrequencysubduction zone earthquakes vs.higherfre- quencytropicalcyclones).New¹⁴Cdataemanatingfromthisstudy also serve to reﬁne dates of prior earthquake-related event de- positsintheJT.

BulkOC ¹⁴C age differencesbetweenhemipelagic background sedimentandeventdepositsrangefrom≤³^kyrⁱⁿ^the^southern^JT,

<6kyrinthecentralJT,to≤⁴^kyrⁱⁿ^the^northern^JT.^Theseâgeôff- sets,whichareusually largestinthelower, sandylayers ofevent deposits,cannot be solely explained by hydraulic particlesorting or selective preservation of OC. For the southern andcentral JT, weattribute the offsetstoearthquake-triggered remobilizationof surficialsediments (includingtephra layers)that evolve intoero- siveturbiditycurrentsentrainingoldercarbon.ForthenorthernJT, sporadic canyon flushing events (potentially initiated by subma- rinelandslidesinupslopeHidakaTrough)resultinremobilization andtranslocationofvastamountsofsedimentandOCtothehadal zone.

DepositionofmarineOCpredominatesalongtheentireJTsys- tem,includingwithineventdeposits.Thissuggests(i)thatcanyon systemsdonot efficientlyfunnel terrestrialOCto thehadal zone and(ii)that typhoonsdonotserveasimportanttriggersforsed- imentandcarbon transfer tothe JT. These findings lend support to prior observations from the JT indicating that event deposits originatedfromthe landwardtrenchslopeandweretriggered by earthquakeshaking.Thesefindingscontrastwiththosefromother trenches (e.g., NBT, MT) where substantial delivery of terrestrial OCtothehadalzonehasbeenobserved.Thesedifferencescanbe explainedby the largedistance betweenthetrenchandadjacent landmasses (>180 km), and by the physiography of the canyons feedingtheJT.Basedonthesefindings,wesuggestthatsedimenta- tionprocessesinhadaltrenchsystemsmayvaryspatiallyandtem- porally,andthatthesource,transfer,anddepositionofOCinhadal trenches should not be generalized. High spatial and temporal- resolutionsamplingofhadaltrenchsedimentarysuccessions,cou- pledwithadvanced carbonisotopic measurements,are necessary for detailed characterization of the mechanics and dynamics of trenchsystems. Applying such approaches to other trenches and comparingknowledge gainedacrossdifferenthadal environments wouldcontributetowards aholisticunderstanding ofOCtransfer, deposition,andburialintrenchsystemsandtheir roleinthecar- boncycleandecologyofthedeepestoceanrealms.

CRediTauthorshipcontributionstatement

TSandNHanalyzedOCdata.

TS and APM conducted ramped pyrolysis/oxidation measurements.

MS,TIE,andTSprojectedthestudy.