rethinking the global carbon cycle with a large, dynamic ... · dn ex dt ¼ f in m ex ðn in3n...

Frontiers

Rethinking the global carbon cycle with a large,dynamic and microbially mediated gas hydrate capacitor

Gerald R. Dickens �

Department of Earth Sciences, Rice University, Houston, TX, USAShell Center for Sustainability, Rice University, Houston, TX, USA

Received 17 December 2002; received in revised form 20 May 2003; accepted 10 June 2003

Abstract

Prominent negative N13C excursions characterize several past intervals of abrupt (6 100 kyr) environmental change.

These anomalies, best exemplified by the s 2.5x drop across the Paleocene/Eocene thermal maximum (PETM) ca.55.5 Ma, command our attention because they lack explanation with conventional models for global carbon cycling.Increasingly, Earth scientists have argued that they signify massive release of CH4 from marine gas hydrates, althoughtypically without considering the underlying process or the ensuing ramifications of such an interpretation. At themost basic level, a large, dynamic ‘gas hydrate capacitor’ stores and releases 13C-depleted carbon at rates linked toexternal conditions such as deep ocean temperature. The capacitor contains three internal reservoirs: dissolved gas,gas hydrate, and free gas. Carbon enters and leaves these reservoirs through microbial decomposition of organicmatter, anaerobic oxidation of CH4 in shallow sediment, and seafloor gas venting; carbon cycles between thesereservoirs through several processes, including fluid flow, precipitation and dissolution of gas hydrate, and burial.Numerical simulations show that simple gas hydrate capacitors driven by inferred changes in bottom water warmingduring the PETM can generate a global N13C excursion that mimics observations. The same modeling extended overlonger time demonstrates that variable CH4 fluxes to and from gas hydrates can partly explain other N13C excursions,rapid and slow, large and small, negative and positive. Although such modeling is rudimentary (because processes andvariables in modern and ancient gas hydrate systems remain poorly constrained), acceptance of a vast, externallyregulated gas hydrate capacitor forces us to rethink N

13C records and the operation of the global carbon cyclethroughout time.8 2003 Elsevier B.V. All rights reserved.

Keywords: gashydrates; carbon cycle; methane; carbon isotopes; global change; Paleocene^Ecocene; Phanerozoic

1. Introduction

Secular changes in the stable carbon isotopic

composition of primary carbonate and organicmatter, as exempli¢ed in newly compiled Cenozo-ic benthic foraminiferal N13C records (Fig. 1), lieat the heart of paleo-environmental reconstruc-tions [1]. For decades, a common template forcarbon cycling on Earth’s surface (Fig. 2) hasbeen used to understand these variations. Accord-ing to this framework, widespread N

13C changes

0012-821X / 03 / $ ^ see front matter 8 2003 Elsevier B.V. All rights reserved.doi:10.1016/S0012-821X(03)00325-X

* Tel. : +1-713-348-5130; Fax: +1-713-348-5214.E-mail address: [email protected] (G.R. Dickens).

EPSL 6724 5-8-03 Cyaan Magenta Geel Zwart

Earth and Planetary Science Letters 213 (2003) 169^183

R

Available online at www.sciencedirect.com

www.elsevier.com/locate/epsl

mailto:[email protected]

http://www.elsevier.com/locate/epsl

re£ect di¡erences in the £uxes or isotopic compo-sitions of terrestrial carbon inputs (weatheringand volcanism) or marine carbon outputs (car-bonate and organic matter) [2^4].

Two wondrous discoveries challenge widelyheld views of global carbon cycling. First, N

13Crecords display extraordinary s 2x drops acrossseveral intervals of major environmental change[5^11]. For at least one of these events, the Paleo-cene/Eocene thermal maximum (PETM) ca. 55Ma (Fig. 1), the N

13C anomaly indicates an im-mense, 6 20 kyr input of 13C-depleted carbon tothe ocean and atmosphere, an addition impossibleto explain with conventional carbon cycle modelsbut analogous in several regards to anthropogenicfossil fuel emissions [12]. Second, sediment se-

quences along continental margins hold a tremen-dous amount of 13C-depleted CH4 as gas hydrate[13,14]. Current carbon cycle models neglect thisCH4 [2^4], although its distribution depends on£uxes to and from the ocean [15^20], and externalconditions, especially deep ocean temperature[14].

Coupling these ¢ndings, numerous recent pa-pers have argued that abrupt, negative N

13C ex-cursions signify bursts of CH4 from marine gashydrates [5^12,21^23]. But this literature almostinvariably invokes CH4 escape without account-ing for its formation, storage and release, leavingus with a series of profound questions, namelywhy, how, where, and when should we incorpo-rate gas hydrates into the global carbon cycle?

Fig. 1. The carbon isotopic composition (N13C) of benthic foraminifera over the Cenozoic [1] showing secular changes in the exo-genic carbon cycle (NEx), including the abrupt negative excursion at the PETM. Note that N

13C data have been averaged over5000 yr increments so that the magnitude of change across the PETM is dampened. Also note that foraminiferal carbonate has aV1.5x o¡set from NEx (Fig. 2).

Fig. 2. A basic, steady-state model for the Phanerozoic exogenic carbon cycle (ocean, atmosphere and biomass) [2] with conven-tional (C) £uxes, and with postulated connections to gas hydrate capacitors (M1, M2, Fig. 7). Masses are in gigatons of carbon(Gt C); £uxes are in Gt C/yr N values are in per mil.


G.R. Dickens / Earth and Planetary Science Letters 213 (2003) 169^183170

Building on recent work [14,24], this paper ex-plores these issues by attempting to satisfactorilyexplain the PETM N

13C excursion within the con-text of existing carbon cycle models and gas hy-drate knowledge. The assumptions, speculations,and unknowns inherent in such an exercise imme-diately highlight the frontier nature of gas hydratestudies and major research challenges.

2. Rapid, negative carbon isotope excursions andthe PETM

Events of extreme environmental change punc-tuate Earth’s history. Across some of these aber-rations, carbonate and organic matter fromwidely separated locations display prominent(s 2x) drops in N

13C, indicating sudden per-turbations in global carbon cycling. In the Pha-nerozoic, intervals include the Permian/TriassicBoundary ca. 252 Ma [11,21], Jurassic episodesca. 183 and 157 Ma [9,10], the Aptian ‘Selli’ eventca. 120 Ma [7,8], and the PETM [5,6,25^28].

Of these events, the PETM warrants specialattention because, unlike older intervals, strati-graphic relationships unequivocally demonstratethat its N

13C excursion was global and rapid. Atleast 40 di¡erent Paleogene stable isotope records,

constructed using deep and shallow marine car-bonate, and terrestrial carbonate and organicmatter, show a 2.5x or greater drop in N

13Cacross the PETM [1,5,6,25^28]. In records fromopen ocean locations (Fig. 3), this decrease occursover 10^40 cm and returns to near initial values ina roughly logarithmic pattern over 100^400 cm,depending on sedimentation rate. The exact shapeand timing of the PETM N

13C excursion remainopen issues, even at the same location [5,6,29,30].Nevertheless, on the basis of sedimentation rates,cyclostratigraphy, or He isotope accumulation[5,6,29^31], the decrease and return in N

13Cspanned 6 20 kyr and 6 220 kyr, respectively(Fig. 3). In striking contrast to the Cretaceous/Tertiary Boundary V10 Myr before, the PETMprecisely correlates to a prominent benthic fora-minifera extinction [25] and an extraordinary ter-restrial mammal diversi¢cation [27], suggesting anentirely di¡erent mechanism for extreme environ-mental change.

3. A basic model for the global carbon cycle

Straightforward mass balance equations under-pin any interpretation of ‘global’ N

13C records(and, ultimately, any placement of sea£oor CH4

into the global carbon cycle). The exogenic car-bon cycle includes all carbon in the ocean, atmo-sphere and biomass (Fig. 2). Carbon exchangesbetween these internal reservoirs within 2000years (at least at the present day). Consequently,over longer time scales, the exogenic carbon cyclecan be considered a single entity whose mass(MEx) and isotopic composition (NEx) change be-cause of variations in the £ux or isotopic compo-sition of external inputs or outputs [2^4].

The simplest useful expression for modeling NEx

over time is [2] :

dNEx

dt¼ F In

MExðNIn3NExÞ3

FOut

MExðNOut3NExÞ ð1Þ

where FIn, FOut, NIn and NOut are the £uxes andN

13C of external carbon inputs and outputs, and tis time. Both terms on the right can be expandedto include component £uxes. In particular, andfor reasons clari¢ed later, carbonate and organic

Fig. 3. Carbon isotope records across the PETM in di¡erentphases at three deep-sea locations. Original records havebeen placed on a common depth scale with the N

13C mini-mum at 0.0 m [24]. Note that the sedimentation rates varybetween sites, giving di¡erent shapes to the excursion.


G.R. Dickens / Earth and Planetary Science Letters 213 (2003) 169^183 171

matter outputs should be considered separately[2] :

dNEx

dt¼ F In

MExðNIn3NExÞ3

FCarb

MExðNCarb3NExÞ3

FOrg

MExðNOrg3NExÞ ð2Þ

where FCarb, FOrg, NCarb, and NOrg are the £uxesand N

13C of carbonate and organic matter out-puts.

Carbonate and organic matter precipitate fromreservoirs within the exogenic carbon cycle, sotheir £uxes and isotopic compositions relate toMEx and NEx. At the simplest level, the mass^composition relationships are: FCarb = kCarbMEx,FOrg = kOrgMEx, NCarb = NEx+vCarb, NOrg = NEx+vOrg, where kCarb and kOrg are reciprocal e-folding(residence) times, and vCarb and vOrg are thebroad isotopic fractionations between the exogen-ic carbon cycle and carbonate or organic matter.These relationships can be incorporated into Eq. 2as follows [2] :

dNEx

dt¼ F In

MExðNIn3NExÞ3kCarbðvCarbÞ3kOrgðvOrgÞ ð3Þ

Secular changes in NEx can be simulated usingthis equation if appropriate parameters areknown. A generic Phanerozoic exogenic carboncycle with quanti¢ed masses and £uxes atsteady-state conditions has been presented [2].After modi¢cations to include biomass and iso-topic fractionation during carbonate precipitation[24], this carbon cycle (Fig. 2) provides a ¢rst-order foundation for understanding past globalchanges in stable carbon isotopes.

4. A carbon mass balance problem

The sharp drop and gradual recovery in NEx

across the PETM (Fig. 3) implies a massive injec-tion of 13C-depleted carbon to the ocean or atmo-sphere [1,12,31,32]. Such an input should havedissolved signi¢cant amounts of CaCO3 on thedeep sea£oor [32], which indeed occurred duringthe PETM [25,26,28]. However, its source be-comes hugely problematic if one considers the

mass balance equations above and conventionalcarbon cycle models.

Theoretical carbon injections needed to cause agiven global N13C excursion can be evaluated bymodifying Eq. 3 [24] :

dNEx

dt¼ FAdd

MExðNAdd3NExÞþ

F In

MExðNIn3NExÞ3kCarbðvCarbÞ3kOrgðvOrgÞ ð4Þ

where FAdd and NAdd are the £ux and N13C of the

additional carbon. Solutions to this equationshow that only implausibly large carbon inputsfrom rivers, volcanoes, or weathering can accountfor a global 32.5x excursion in NEx within 20kyr [24]. For example, average Phanerozoic vol-canism (FVol) would have to increase s 100 times[24], and no evidence supports such volcanismduring the PETM. Excluding protracted anthro-pogenic emissions, conventional carbon cyclemodels cannot explain rapid, global 32.0xN

13C excursions [12,24].When considering possible sources for carbon

input at the PETM, one observation seems partic-ularly relevant: benthic foraminifera N

18O recordsindicate a sudden 5^7‡C warming of deep oceanwater [25,26], an environmental change unparal-leled in the last 70 Myr [1]. Massive carbon inputat the PETM appears causally related to a rapidrise in sea£oor temperatures.

5. Marine gas hydrates and free gas

At high gas concentration, certain low-molecu-lar-weight gases (e.g., CH4, CO2, Xe) can com-bine with water to form crystalline clathrate hy-drates of gas, or ‘gas hydrates’. The stability ofthese compounds depends on gas composition,pressure, temperature, and the activity of water,a parameter inversely related to salinity. For agiven gas composition, gas hydrates can dissociateto water and gas with a decrease in pressure orincrease in temperature or salinity.

Gas hydrates can naturally occur in marinesediment when gas saturates pore water in theregion of appropriate stability conditions, or gas



hydrate stability zone (GHSZ) [13]. At a singlesite, the GHSZ extends from the sea£oor to adepth where pressure and temperature on the geo-therm approximate those on a pore water^gas hy-

drate^free gas equilibrium curve (Fig. 4). Whenextrapolated across a continental margin, theGHSZ forms a lens starting at some shallowwater depth (nominally 250^500 m at the present

Fig. 5. Methane cycling in a modern gas hydrate reservoir, Ocean Drilling Program (ODP) Site 997, Blake Ridge, southeastUSA margin (adapted from [65]). (Left) Schematic of seismic re£ection pro¢le showing the borehole and the bottom-simulatingre£ector (BSR), an acoustic interface between gas hydrate and free gas. (Middle) Methane solubility curves, in situ gas concentra-tions and the sulfate reduction zone (SRZ) at Site 997. (Right) Inferred ‘high £ux’ CH4 cycling at Site 997. Note that the gross£uxes between gas hydrate and free gas (gray arrows) greatly exceed the net £uxes (black arrows) and no venting occurs at ornear this site.

Fig. 4. Theoretical GHSZ on a typical margin during the late Paleocene (dashed) and after 5‡C warming (shaded) [14]. Alsoshown are locations of free gas before and after the thermal perturbation.



day and 900 m in the Paleocene) and generallythickening beneath deeper water (Fig. 4), depend-ing on local conditions [13,14,33].

Insu⁄cient concentrations of requisite gasespreclude gas hydrate formation in many deep-sea sediment sequences despite appropriate pres-sure, temperature and salinity. However, at nu-merous locations, particularly along continentalslopes, decomposition of organic matter throughmicrobial activity or to a lesser extent thermalcracking generates enough gas to saturate porewaters within portions of the GHSZ (Fig. 5). Be-cause microbes produce much of the gas, it isgenerally greatly enriched in CH4 (s 99%) andvery depleted in 13C (N13CW360x) [13].

An enormous volume of pore space can poten-tially host gas hydrates. By integrating cross-sec-tional areas (Fig. 4) along continental margins,this volume is likely 1^6U106 km3 at the presentday [14]. On average, pore space within the globalGHSZ probably contains 1^10% gas hydrate, im-plying that present-day oceanic gas hydrates hold1000^22 000 Gt C (Gt = 1015 g) [14], with currentliterature typically suggesting 10 000 Gt C [13,34].Not included in this estimate is free gas, whichexists when gas saturates pore waters beneaththe GHSZ [35^37]. Oceanic gas hydrates and as-sociated free gas constitute a substantial pool ofpressure^temperature-sensitive, 13C-depleted CH4

somehow connected to the exogenic carbon cycle(Fig. 2).

6. The gas hydrate dissociation hypothesis

Sudden deep ocean warming and release ofCH4 from marine gas hydrate systems (Fig. 6)provides a general explanation for the N

13C excur-sion and carbonate dissolution across the PETM[12,25,29,32] and perhaps older time intervals [9].Similar to the present day, gas hydrate and under-lying free gas stored enormous quantities of 13C-depleted CH4 in the upper few hundred meters ofsediment on Paleocene continental margins [14].However, some trigger (e.g., long-term late Paleo-cene global warming [1]) pushed ocean circulationpast a critical threshold, causing relatively warmsurface waters to sink into deep waters. This ther-

mal perturbation steepened geotherms on conti-nental margins, dissociating large amounts ofgas hydrate to free gas, and injecting CH4 intothe ocean or atmosphere, perhaps through sedi-ment failure (Figs. 4 and 6). In either reservoir,CH4 would oxidize to 13C-depleted CO2 [17,38],which would propagate throughout the exogeniccarbon cycle, causing a global negative N

13C ex-cursion and carbonate dissolution [32]. The exter-nal carbon inputs and outputs of conventionalmodels (Fig. 2) then £ushed the 13C-depleted car-bon from the exogenic carbon cycle over the nextV200 kyr [12,32].

Although other CH4 release mechanisms havebeen proposed for the PETM (e.g., slope failureor erosion [39,40]), thermal dissociation of gashydrates nicely explains the carbon cycle pertur-bation because the abrupt 5^7‡C rise in bottomwater temperature [1,25,26] should have shrunkthe global GHSZ signi¢cantly (Fig. 4), even con-sidering the time and energy required to heatdeeply buried gas hydrate [41]. Assuming thatlate Paleocene and present-day continental mar-gins were similar except for deep ocean temper-ature, which was V9‡C [1], the global GHSZdecreased s 50% during the PETM, from V1.5to 0.7U106 km3 [14]. Importantly, steady atmo-spheric warming can suddenly switch late Paleo-cene deep water sources from cold, southerly lat-itudes to relatively warm, northerly latitudes [42].Equally important, isotopic records of single fo-

Fig. 6. Schematic highlighting the gas hydrate dissociationhypothesis for the PETM [25]. Signi¢cant quantities of CH4

hydrate convert to free CH4 gas, which escapes through sedi-ment failure to form 13C-depleted CO2 in either the ocean oratmosphere.



raminifera specimens, albeit limited and contro-versial [43], indicate that abrupt surface warmingat high latitudes (i.e., environmental change) pre-ceded in part massive carbon input [29].

Barring a comet impact [43,44], which availablerecords of fossil turnover, the onset of warmth[29], cosmogenic He accumulation [30], and ma-rine osmium isotopic composition [45] do not sup-port, only oceanic gas hydrates can supplyenough 13C-depleted carbon to cause the PETMN

13C excursion. The same mass balance argumentapplies to other prominent negative N

13C excur-sions [21^23], assuming they also representrapid perturbations in NEx. For the generic Phan-erozoic exogenic carbon cycle (Fig. 2), release ofV2500 Gt C with a N

13C of 360x over 20 kyrwould cause a global 32.5x N

13C excursion.Such carbon transfer compels us to incorporategas hydrates into traditional carbon cycle models[24].

7. A large, dynamic gas hydrate capacitor

Marine gas hydrate systems probably serve as alarge capacitor containing dissolved gas, gas hy-drate and free gas, and exchanging carbon withthe exogenic carbon cycle through methanogene-sis and methanotrophy (Fig. 2). The capacitorconcept arises because, in order to explain nega-tive excursions in NEx, carbon £uxes to and fromgas hydrates must vary when external forcingchanges the dimensions of the global GHSZ[24,32]. We are just beginning to appreciate thecomplex, dynamic cycling of carbon within indi-vidual gas hydrate systems [46^51]. Nonetheless,essential components of this cycling (Fig. 5) canbe used to construct rudimentary global gas hy-drate capacitors (Fig. 7) of mass MGH and iso-topic composition NGH that transfer CH4 amongstthree reservoirs: dissolved gas (MDiss, NDiss), gashydrate (MHyd, NHyd), and free gas (MFree, NFree)[24]. Available evidence suggests minimal carbonisotopic fractionation as CH4 moves between dis-solved gas, gas hydrate and free gas [52,53].Hence, NGH = NDiss = NHyd = NFree, and only carbonmass transfer expressions are developed below.

Carbon slowly enters most gas hydrate systems

when archaea convert organic matter to dissolvedCH4 [54,55]. In established systems with signi¢-cant free gas at depth, CH4 generated in thepast can also migrate to shallow depth [16,18,56^58]. High CH4 concentrations at depthtypically drive an upward dissolved CH4 £ux,which encounters downward di¡using SO23

4[15,59,60]. Carbon slowly leaves most systemswhen archaea^bacteria consortia consume bothspecies via anaerobic oxidation of CH4 (AOM)across a sulfate/methane transition (SMT) [61,62], usually within 40 m of the sea£oor [15,60]. If CH4 input exceeds CH4 loss in shallowsediment, gas hydrate can eventually precipitate,provided ambient conditions lie within the GHSZ.The simplest equation to describe these processesis :

Fig. 7. Two plausible gas hydrate capacitors where 13C-de-pleted CH4 cycles through three reservoirs: dissolved gas,gas hydrate and free gas. Note the di¡erence in the directionof CH4 £ow between the internal reservoirs. Masses are ingigatons of carbon (Gt C); £uxes are in Gt C/yr and repre-sent net transport.



dMDiss

dt¼ FMeth þ FDeep3FAOM3FHyd ð5Þ

where FMeth is methanogenesis, FDeep is net CH4

supply from depth, FAOM is CH4 £ow to theSMT, and FHyd is net gas hydrate formation.Methanogenesis must depend on the organic car-bon supply to sediment, so, at the most basiclevel, FMeth = kMethFOrg, where kMeth representsthe fraction of organic matter converted to CH4.The CH4 £ux to the SMT probably depends on£uid £ow and gas concentration at depth [15,46].Hence, FAOM = vMDiss, where v is parameterized£uid £ow. Net carbon transfer between dissolvedCH4 and the other reservoirs is more problematicbecause it depends on the overall £ux regime (Fig.7), and whether CH4 dominantly moves from gashydrate to free gas via sediment burial (‘low£ux’), or from free gas to gas hydrate via upward£uid £ow (‘high £ux’)[46^51]. In the ¢rst case,FHyd = 0 when MDiss is less than a critical massrepresenting saturation, and FHyd =FMeth+FDeep3

FAOM when MDiss exceeds this mass [24]. In thesecond case, FHyd =3rMHyd where r is a parame-terized rate for gas hydrate dissolution in shallowsediment. Both end-member regimes probably op-erate in various locations at di¡erent scales but itis unclear which better characterizes the globalsea£oor CH4 cycle.

Once in gas hydrate, carbon moves through gashydrate systems by three general processes, sedi-ment burial and dissolution as noted above, anddissociation when the dimensions of the GHSZchange [14,41]. These processes can be describedat the simplest level by:

dMHyd

dt¼ FHyd3FBGH3FConv ð6Þ

where FBGH is net gas hydrate burial, and FConv isnet carbon transfer between gas hydrate and freegas when the GHSZ changes (Figs. 4 and 6). Buri-al of solid methane carbon at steady-state condi-tions probably depends on the sedimentation rate(SR) and the amount of gas hydrate in porespace [46,50,51]. Thus, in the ‘low £ux’ model,FBGH = sMHyd, where s is parameterized SR. The‘high £ux’ alternative is discussed below. The oth-er £ux, FConv, which can be positive or negative,

can be calculated from average amounts of freegas and gas hydrate, and GHSZ dimensions be-fore and after an environmental perturbation[14,24].

In several places, particularly where faults in-tersect the sea£oor, free CH4 gas vents to thewater column [16^20]. Mass balance closure ingas hydrate systems occurs through the free gasreservoir, after including this venting. Consideringthe above, the appropriate equation is:

dMFree

dt¼ FBGH þ FConv3FDeep3FVent ð7Þ

where FVent is sea£oor venting. Carbon likely mi-grates between free gas and other sediment reser-voirs via faults and gas chimneys [18,56^58]. Inthe simplest case, this transfer depends on £uid£ow and free gas abundance, so that FDeep =vMFree in the ‘low £ux’ model, or FBGH =3vMFree

in the ‘high £ux’ model. In addition to sea£oorventing above faults, theoretical considerationsand seismic studies suggest that CH4 outgassingmight occur through sediment slumping [63,64].Thus, FVent depends on the amount of free gasand channeled £uid £ow, which may be linkedbecause excessive free gas could induce £owthrough fracturing or slumping [37,64]. A simpleexpression for this behavior is FVent =c(MFree3MCrit) where c is directed £ow, andMCrit is the critical mass of free CH4 above whichfree gas can escape.

8. Connecting the capacitor to the exogenic carboncycle

Temporal changes in carbon masses within andfrom the global gas hydrate capacitor can be cal-culated from the above equations with four items:(1) initial amounts of CH4 in dissolved gas, gashydrate and free gas, (2) values of the rate param-eters (e.g., r, v), (3) history of organic matter buri-al, and (4) evolution of the GHSZ. None are wellconstrained; indeed, they barely have been dis-cussed in the literature from a perspective usefulfor modeling the sea£oor CH4 cycle. Assuming10 000 Gt C in gas hydrate [34] within 3.5U106

km3 of pore space [14], 1000 Gt C in dissolved



gas, and 500 Gt C in free gas [24,37], speculativegas hydrate capacitors can be o¡ered for thepresent day, depending on how carbon £owsglobally. These capacitors (Fig. 7) transfer 13C-depleted CH4 between the exogenic carbon cycleand gas hydrates through methanogenesis, AOM,and CH4 venting, which can be bracketed crudelyfrom the few available regional extrapolations[15,17,19,60] and an assumption of steady state,FMeth =FAOM+FVent [24].

Even at steady-state conditions ^ itself an openissue for the present day [60,65,66] ^ mass balancecoupling between gas hydrates and the exogeniccarbon cycle raises some intriguing problemsmostly circumvented in the literature. First, car-bon £ux estimates for weathering and volcanoesneed revision to account for the small but highly13C-depleted CH4 inputs (Fig. 2). Second, somesedimentary component must remove signi¢cantquantities of relatively 13C-enriched carbon fromthe exogenic carbon cycle to o¡set production andstorage of CH4 (Figs. 2 and 7). The 13C-enrichedauthigenic carbonates found within gas hydratesystems [67] may provide this sink. Third, the res-idence time of CH4 within the global gas hydratecapacitor must be fairly long, 1.5^7 Myr accord-ing to models presented here. This inferenceagrees with recent studies of individual gas hy-drate systems [50,51]. Two conundrums also arisewhen connecting a present-day gas hydrate ca-pacitor to a Phanerozoic carbon cycle. First,warmer bottom water temperatures and shortercontinental margins would reduce the globalGHSZ. In particular, the late Paleocene GHSZwas nominally half that of the present day, sothat a gas hydrate capacitor with 11 500 Gt Cpresumes that gas hydrate occupied V8.5% ofpore space within the GHSZ [14]. This occupancyseems high but could re£ect greater organic car-bon burial in the past (and hence, increased FMeth)as perhaps suggested by the N

13C maximum ca. 58Ma (Fig. 1). Second, the N

13C of gas hydrates orthe isotopic fractionation during methanogenesismust have been less in the past relative to thepresent day. The 360x CH4 produced in mod-ern gas hydrate systems derives dominantly frommarine organic matter with a N

13C of 322 to325x. However, a N

13C of 327 to 330x char-

acterizes marine organic matter for much of thePhanerozoic [68].

9. Simulating the PETM and the surroundingpaleogene

The carbon cycle presented here (Figs. 2 and 7)transfers carbon between conventional reservoirsand a large, dynamic and microbially mediatedsea£oor CH4 cycle. For generic Phanerozoic con-ditions with 10‡C deep ocean temperatures (andcaveats noted above), 1.65^7.0U1012 g C/yr of

Fig. 8. (A) The predicted mass perturbation of the two gashydrate capacitors (Fig. 7) with a 5‡C warming of deepocean water over 20 kyr. (B) The ensuing N

13C excursions inthe Phanerozoic exogenic carbon cycle (Fig. 2). Model 1 (sol-id); model 2 (dashed).



360x CH4 enter and leave the exogenic carboncycle by moving through dissolved gas, gas hy-drate and free gas (Figs. 2 and 7). The enormousgas hydrate reservoir holds 10 000 Gt C within alens that expands downward along continentalmargins starting at V900 m water depth (Fig.5). However, changes in external conditions, espe-cially bottom water temperature, will recon¢gurethis volume, rearranging masses and £uxes withinthe circuit.

Benthic foraminifera N18O records across the

PETM support a 5^7‡C rise in deep ocean tem-perature within V20 kyr, followed by a 5^7‡Ccooling over V50 kyr [1,25,26]. This perturbationwould profoundly impact carbon £uxes from anygas hydrate capacitor if AOM and CH4 ventingdepend on the abundance of dissolved gas or freegas (Fig. 8). With models presented here, theGHSZ shrinks during initial warming (Fig. 4),and V5500 Gt C of gas hydrate converts tofree gas, of which s 2600 Gt C escape from thesea£oor, quickly depleting the capacitor and add-ing 13C-depleted carbon to the exogenic carboncycle. Once cooling begins, the GHSZ growsand remaining free gas reforms gas hydrate. Be-cause so much carbon escaped, however, free gasabundance drops below steady-state conditions,and CH4 discharge (FAOM+FVent) drops belowCH4 production (FMeth). The capacitor then re-charges, e¡ectively removing 13C-depleted carbonfrom the exogenic carbon cycle. Consideration ofa signi¢cant gas hydrate capacitor during thePETM necessarily results in a global N13C excur-sion that resembles isotopic records (Fig. 3),although the shape and magnitude of this excur-sion depend on parameters of the capacitor, suchas initial masses and £ow of CH4 (Fig. 8).

With the models discussed here, only sudden,massive gas hydrate dissociation and free gasventing to the exogenic carbon cycle can explainrapid, negative excursions in NEx. Such carbontransfer probably requires widespread sedimentfailure on continental slopes [63,64], which musthave occurred at water depths s 900 m duringthe PETM (Figs. 4 and 6). Slumping at the ap-propriate paleodepths during the PETM has beendocumented on the eastern USA margin [25,39].

Most current literature has discussed prominent

negative N13C excursions as isolated CH4 bursts

[5^11,21^24]. However, an integrated carbon cyclemodel allows us to appropriately assess these per-turbations within the context of surrounding time.For example, consider the broad Paleocene^Eo-cene transition from 57 to 50 Ma (Fig. 1). Usingbenthic foraminifera N

18O records as a proxy fordeep ocean temperature [1] (and assuming other

Fig. 9. (A) Inferred bottom water temperature changes be-tween 57 and 50 Ma [1] and the predicted evolution of gashydrate mass for the two gas hydrate capacitors (Fig. 7).(B) The modeled N

13C record of the Phanerozoic exogeniccarbon cycle assuming conventional carbon £uxes (Fig. 2) re-main constant compared to the measured N

13C of benthic fo-raminifera [1]. Model 1 (solid); model 2 (dashed). Note thattemperature and N

13C data have been averaged over 5000 yrincrements [1] so that changes in mass and N

13C are damp-ened across the PETM.



factors controlling gas hydrate distribution [14]remain constant), numerous brief intervals of ac-celerated CH4 outgassing and storage occurredaccording to models presented here, collectivelyreducing the amount of gas hydrate from an as-sumed 10 000 Gt at 57 Ma to 6 7000 Gt at 53 Ma(Fig. 9). With this view, the PETM N

13C excursionno longer represents an aberration requiring anextraordinary mechanism, but instead an out-standing example of a common phenomenonwhere sea£oor CH4 discharge exceeds CH4 pro-duction. Paleogene benthic foraminifera assem-blages, which appear to indicate multiple pulsesof enhanced chemosynthetic activity [69], maysupport this interpretation.

10. Ten immediate challenges

A fundamental reconstruction of the global car-bon cycle has been o¡ered here to account forvast sea£oor CH4 reservoirs at the present day,and prominent negative N

13C excursions in thepast (Figs. 2 and 7). The most crucial insightfrom such a model is that CH4 £uxes to andfrom gas hydrates continually vary, always con-tributing to changes in NEx, large and small, neg-ative and positive (Fig. 9). With the presentedmodel, variable CH4 £uxes adequately explainmany short-term excursions in NEx, but cannotcause long-term (s 500 kyr) changes in NEx (e.g.,the 2.5x drop ca. 57^53 Ma, Fig. 1) because gashydrates cannot excessively furnish CH4 withoutrecharge, and methanogenesis occurs slowly (Fig.9). However, if past conditions (e.g., warmerwater) accelerated CH4 production, sea£oorCH4 £uxes could drive longer changes in NEx.

The modi¢ed carbon cycle model of this paperis necessarily speculative, deterministic and sim-ple. Indeed, the idea of an enormous marine gashydrate capacitor, collecting and discharging var-iable amounts of CH4 over time, should be highlycontroversial with available information. So, howcan we build a more vinous model, su⁄cientlyrobust and complex to test?1. Quantify the abundance of carbon residing in

gas hydrate and free gas. Though widely cited,the 10 000 Gt estimate for the present-day mass

of gas hydrates is wildly unconstrained [14,34].No rigorous evaluation for the amount of freegas has been published [24,37].

2. Establish the £ow of carbon through gas hy-drate systems. Recent e¡orts to balance massesand £uxes within gas hydrate systems have fo-cused on CH4 [50,51]. No study has appropri-ately traced the £ow of carbon, including inorganic matter, dissolved phases, and authigen-ic carbonate.

3. Determine carbon outputs from modern gashydrate systems. Quanti¢ed regional to globalestimates for AOM and CH4 venting are justsurfacing [15,17,19,60]. These estimates varysigni¢cantly and it is not clear how much ofthe total £uxes apply to gas hydrate systems.

4. Connect other sea£oor gas reservoirs. In addi-tion to gas hydrate systems, large amounts ofgas occur in conventional hydrocarbon playsand on continental shelves. These reservoirsand their £uxes might also vary considerablyover time [33,66,70].

5. Incorporate triggers for variable CH4 outputs.A complete model needs coupling to tectonism,Earth surface temperature, and environmentalchange (i.e., the underlying reasons for chang-ing external conditions [42]) rather than simplyinputting £uctuations in the GHSZ.

6. Track the oxidation of escaping CH4. No sig-ni¢cance has been placed on whether ventingCH4 is oxidized in the ocean or atmosphere(Fig. 6), although any consequential atmo-spheric warming, dissolved O2 de¢ciency, andCaCO3 dissolution strongly depend on this is-sue [32].

7. Include appropriate carbon cycle feedbacks.An advanced model must consider that CH4

addition will increase CaCO3 dissolution andweathering [32,45], which might enhance ma-rine productivity [5], organic matter burial,and gas hydrate recharge [24].

8. Evaluate the impact on other geochemicalcycles. AOM and CH4 venting necessarily af-fect the marine sulfur, oxygen and perhaps ba-rium cycles [17,32,60,71] but it remains unclearwhether past perturbations in these othercycles are consistent with highly variable CH4

release.



9. Develop tracers other than carbon isotopesfor CH4 £uxing. Oxidation of CH4 shouldform authigenic minerals (e.g., calcite, barite)and organic compounds (e.g., hopanoids). Wehave only begun examining these phases inrecent sediment to constrain past CH4 £uxes[65,67,72].

10. Extend the model over the geological record.The ultimate challenge lies in building a glob-al carbon cycle model that successfully ac-counts for the present-day mass of gas hy-drate and agrees with the global N13C recordafter running from the Neoproterozoic, whenconditions a¡ecting gas hydrate distributionwere perhaps the most dramatic [73], throughthe Neogene and Quaternary, when recordsfor assessing sea£oor CH4 £uxes becomemost complete [72,74].

Acknowledgements

An NSF MESH sponsored workshop on gashydrates and climate change provided the impetusfor this paper. I thank the workshop participantsfor stimulating ideas, and K. Kvenvolden, A. Mil-kov, E. Thomas, and J. Zachos who, at varioustimes, including as reviewers of this paper, haveforced me to address key issues. I also thank A.Halliday for the challenge to put my thoughts on10(+) pages! This writing was supported throughU.S. NSF Grants (OCE 0117950, EAR 0120727).[AH]

References

[1] J. Zachos, M. Pagani, L. Sloan, E. Thomas, K. Billups,Trends, rhythms, and aberrations in global climate 65 Mato present, Science 292 (2001) 686^693.

[2] L.R. Kump, M.A. Arthur, Interpreting carbon-isotopeexcursions: Carbonates and organic matter, Chem.Geol. 161 (1999) 181^198.

[3] Y. Godde¤ris, L.M. Franc[ois, J. Veizer, The early Paleo-zoic carbon cycle, Earth Planet. Sci. Lett. 190 (2001) 181^196.

[4] K. Wallmann, Controls on the Cretaceous and Cenozoicevolution of seawater composition, atmospheric CO2 andclimate, Geochim. Cosmochim. Acta 65 (2001) 3005^3025.

[5] S. Bains, R.M. Cor¢eld, R.D. Norris, Mechanisms ofclimate warming at the end of the Paleocene, Science285 (1999) 724^727.

[6] U. Ro«hl, T.J. Bralower, R.D. Norris, G. Wefer, Newchronology for the late Paleocene thermal maximumand its environmental implications, Geology 28 (2000)927^930.

[7] H.C. Jenkyns, P.A. Wilson, Stratigraphy, paleoceanogra-phy, and evolution of Cretaceous Paci¢c guyots: Relicsfrom a greenhouse earth, Am. J. Sci. 299 (1999) 341^392.

[8] A.H. Jahren, N.C. Arens, G. Sarmiento, J. Guerrero, R.Amundson, Terrestrial record of methane hydrate disso-ciation in the Early Cretaceous, Geology 29 (2001) 159^162.

[9] S.P. Hesselbo, D.R. Gro«cke, H.C. Jenkyns, C.J. Bjerrum,P. Farrimond, H.S.M. Bell, O.R. Green, Massive dissoci-ation of gas hydrate during a Jurassic oceanic anoxicevent, Nature 406 (2000) 392^395.

[10] M. Padden, H. Weissert, M. deRafelis, Evidence for LateJurassic release of methane from gas hydrate, Geology 29(2001) 223^226.

[11] E.S. Krull, G.J. Retallack, 13C depth pro¢les from paleo-sols across the Permian-Triassic boundary: Evidence formethane release, Bull. Geol. Soc. Am. 112 (2000) 1459^1472.

[12] G.R. Dickens, The blast in the past, Nature 401 (1999)752^755.

[13] K.A. Kvenvolden, T.D. Lorenson, The global occurrenceof natural gas hydrate, in: C.K. Paull, W.P. Dillon (Eds.),Natural Gas Hydrates: Occurrence, Distribution andDetection, AGU Geophys. Monogr. Ser. 124 (2001) 3^18.

[14] G.R. Dickens, The potential volume of oceanic methanehydrates with variable external conditions, Org. Geochem.32 (2001) 1132^1193.

[15] W.S. Borowski, C.K. Paull, W. Ussler III, Global andlocal variations of interstitial sulfate gradients in deep-water, continental margin sediments: Sensitivity to under-lying methane and gas hydrates, Mar. Geol. 159 (1999)131^154.

[16] E. Suess, M.E. Torres, G. Bohrmann, R.W. Collier, J.Greinerft, P. Linke, G. Rehder, A. Tre¤hu, K. Wallman,G. Winckler, E. Zuleger, Gas hydrate destabilization: En-hanced dewatering, benthic material turnover and largemethane plumes at the Cascadia Margin, Earth Planet.Sci. Lett. 170 (1999) 1^15.

[17] D.L. Valentine, D.C. Blanton, W.S. Reeburgh, M. Kast-ner, Water column methane oxidation adjacent to an areaof active hydrate dissociation, Eel River Basin, Geochim.Cosmochim. Acta 65 (2001) 2633^2640.

[18] M. Wiedicke, H. Sahling, G. Delisle, E. Faber, S. Neben,H. Beiersdorf, V. Marchig, W. Weiss, N. von Mirbach, A.A¢at, Characteristics of an active vent in the fore-arcbasin of the Sunda Arc, Indonesia, Mar. Geol. 184(2002) 121^141.

[19] M.E. Torres, J. McManus, D.E. Hammond, M.A. de



Angelis, K.U. Heeschen, S.L. Colbert, M.D. Tryon, K.M.Brown, E. Suess, Fluid and chemical £uxes in and out ofsediments hosting methane hydrate deposits on HydrateRidge, OR, I: Hydrological provinces, Earth Planet. Sci.Lett. 201 (2002) 525^540.

[20] M.D. Tryon, K.M. Brown, M.E. Torres, Fluid and chem-ical £ux in and out of sediments hosting methane hydratedeposits on Hydrate Ridge, OR, II: Hydrological process-es, Earth Planet. Sci. Lett. 201 (2002) 541^557.

[21] M.J. de Wit, J.G. Ghosh, S. de Villiers, Multiple organiccarbon isotope reversals across the Permo-Triassic bound-ary of terrestrial Gondwana sequences: Clues to extinc-tion patterns and delayed ecosystem recovery, J. Geol.110 (2002) 227^246.

[22] R.A. Berner, Examination of hypotheses for the Permo-Triassic boundary extinction by carbon cycle modeling,Proc. Natl. Acad. Sci. USA 99 (2002) 4172^4177.

[23] D.J. Beerling, M.R. Lomas, D.R. Gro«cke, On the natureof methane gas-hydrate dissociation during the Toarcianand Aptian oceanic anoxic events, Am. J. Sci. 302 (2002)28^49.

[24] G.R. Dickens, Modeling the global carbon cycle with agas hydrate capacitor: Signi¢cance for the Latest Paleo-cene thermal maximum, in: C.K. Paull, W.P. Dillon(Eds.), Natural Gas Hydrates, Occurrence, Distributionand Detection, AGU Geophys. Monogr. Ser. 124 (2001)19^38.

[25] M.E. Katz, D.K. Pak, G.R. Dickens, K.G. Miller, Thesource and fate of massive carbon input during the LatestPaleocene Thermal Maximum, Science 286 (1999) 1531^1533.

[26] D.J. Thomas, T.J. Bralower, J.C. Zachos, New evidencefor subtropical warming during the late Paleocene thermalmaximum Stable isotopes from Deep Sea Drilling ProjectSite 527, Walvis Ridge, Paleoceanography 14 (1999) 561^570.

[27] G.J. Bowen, W.C. Clyde, P.L. Koch, S.Y. Ting, J. Alroy,T. Tsubamoto, Y.Q. Wang, Y. Wang, Mammalian dis-persal at the Paleocene/Eocene boundary, Science 295(2002) 2062^2065.

[28] T.J. Bralower, I. Premoli-Silva, M.J. Malone, Scienti¢cParticipants of Leg 198, New evidence for abrupt climatechange in the Cretaceous and Paleogene: An Ocean Drill-ing Program expedition to Shatsky Rise, northwest Pacif-ic, GSA Today 12 (2002) 4^10.

[29] D.J. Thomas, J.C. Zachos, T.J. Bralower, E. Thomas, S.Bohaty, Warming the fuel for the ¢re: Evidence for thethermal dissocation of methane hydrate during the Paleo-cene-Eocene thermal maximum, Geology 30 (2002) 1067^1070.

[30] K.A. Farley, S.F. Eltgroth, An alternative age model forthe Paleocene-Eocene thermal maximum using extrater-restrial 3He, Earth Planet. Sci. Lett. 208 (2003) 135^148.

[31] R.D. Norris, U. Ro«hl, Carbon cycling and chronology ofclimate warming during the Palaeocene/Eocene transition,Nature 401 (1999) 775^778.

[32] G.R. Dickens, Methane oxidation during the late Palaeo-cene thermal maximum, Bull. Soc. Ge¤ol. Fr. 171 (2000)37^49.

[33] A.V. Milkov, R. Sassen, Two-dimensional modeling ofgas hydrate decomposition in the northwestern Gulf ofMexico: signi¢cance to global change assessment, GlobalPlanet. Change 36 (2003) 31^46.

[34] K.A. Kvenvolden, Potential e¡ects of gas hydrate on hu-man welfare, Proc. Natl. Acad. Sci. USA 96 (1999) 3420^3426.

[35] I. Grevemeyer, A. Rosenberger, H. Villinger, Natural gashydrate on the continental slope o¡ Pakistan: Constraintsfrom seismic techniques, Geophys. J. Int. 140 (2000) 295^310.

[36] A.M. Tre¤hu, E.R. Flueh, Estimating the thickness of thefree gas zone beneath Hydrate Ridge, Oregon continentalmargin, from seismic velocities and attenuation, J. Geo-phys. Res. 106 (2001) 2035^2045.

[37] M. Hornbach, S. Holbrook, D. Sa¡er, A global, criticallypressured free gas zone beneath methane hydrate provin-ces, EGS-AGU-EUG Assembly (2003) Abstract EAE03-A-08139.

[38] D.J. Wuebbles, K. Hayhoe, Atmospheric methane andglobal change, Earth-Sci. Rev. 57 (2002) 177^210.

[39] M.E. Katz, B.S. Cramer, G.S. Mountain, S. Katz, K.G.Miller, Uncorking the bottle: What triggered the Paleo-cene/Eocene thermal maximum methane release?, Pale-oceanography 16 (2001) 549^562.

[40] W.S. Holbrook, D. Lizarralde, I.A. Pecher, A.R. Gor-man, K.L. Hackwith, M. Hornbach, D. Sa¡er, Escapeof methane gas through sediment waves in a large meth-ane hydrate province, Geology 30 (2002) 467^470.

[41] W. Xu, R.P. Lowell, E.T. Peltzer, E¡ect of sea£oor tem-perature and pressure variations on methane £ux from agas hydrate layer: Comparison between current and latePaleocene climate conditions, J. Geophys. Res. 106 (2001)26413^26423.

[42] K.L. Bice, J. Marotzke, Could changing ocean circulationhave destabilized methane hydrate at the Paleocene/Eo-cene boundary? Paleoceanography 17 (2002) 10.1029/2001PA000678.

[43] D.V. Kent, B.S. Cramer, L. Lanci, D. Wang, J.D. Wright,R. van der Voo, A case for a comet trigger for the Paleo-cene/Eocene thermal maximum and carbon isotope excur-sion, Earth Planet. Sci. Lett. 211 (2003) 13^26.

[44] D. Deming, On the possible in£uence of extraterrestrialvolatiles on Earth’s climate, the origin of the oceans,Palaeogeogr. Palaeoclimatol. Palaeoecol. 146 (1999) 33^51.

[45] G. Ravizza, R.N. Norris, J. Blusztajn, M.P. Aubry, Anosmium isotope excursion associated with the late Paleo-cene thermal maximum: Evidence of intensi¢ed chemicalweathering, Paleoceanography 16 (2001) 155^163.

[46] P.K. Egeberg, G.R. Dickens, Thermodynamic and porewater halogen constraints on gas hydrate distribution atODP Site 997 (Blake Ridge), Chem. Geol. 153 (1999) 53^79.



[47] W. Xu, C. Ruppel, Predicting the occurrence, distribu-tion, and evolution of methane gas hydrate in porousmarine sediments, J. Geophys. Res. 104 (1999) 5081^5095.

[48] M.K. Davie, B.A. Bu¡ett, A numerical model for theformation of gas hydrate below the sea£oor, J. Geophys.Res. 106 (2001) 497^514.

[49] I.A. Pecher, N. Kukowski, C. Huebscher, J. Greinert, J.Bialas, GEOPECO Working Group, The link betweenbottom-simulating re£ections and methane £ux into thegas hydrate stability zone ^ new evidence from Lima Ba-sin, Peru Margin, Earth Planet. Sci. Lett. 185 (2001) 343^354.

[50] M.K. Davie, B.A. Bu¡ett, Sources of methane for marinegas hydrate: inferences from a comparison of observa-tions and numerical models, Earth Planet. Sci. Lett. 206(2003) 51^63.

[51] K.L. Gering, Simulations of methane hydrate phenomenaover geologic timescales. Part I E¡ect of sediment com-paction rates on methane hydrate and free gas accumu-lations, Earth Planet. Sci. Lett. 206 (2003) 65^81.

[52] C.K. Paull, T.D. Lorenson, W.S. Borowski, W. Ussler III,K. Olsen, N.M. Rodriguez, Isotopic composition of CH4,CO2 species, and sedimentary organic matter within sam-ples from the Blake Ridge: Gas source implications, Proc.ODP Sci. Results. 164 (2000) 229^236.

[53] P.J. Wallace, G.R. Dickens, C.K. Paull, W. Ussler III,Carbon isotopic composition of methane in gas hydratesand free-gas-bearing sediments from the Blake Ridge: Ef-fects of degassing on isotopic compositions of gas releasedfrom pressurized core samples, Proc. ODP Sci. Results164 (2000) 101^112.

[54] K.A. Bidle, M. Kastner, D.H. Bartlett, A phylogeneticanalysis of microbial communities associated with meth-ane hydrate containing marine £uids and sediments in theCascadia margin (ODP site 892B), FEMS Microbiol.Lett. 177 (1999) 101^108.

[55] J.R. Marchesi, A.J. Weightman, B.A. Cragg, R.J. Parkes,J.C. Fry, Methanogen and bacterial diversity and distri-bution in deep gas hydrate sediments from the CascadiaMargin as revealed by 16S rRNA molecular analysis,FEMS Microbiol. Ecol. 34 (2001) 221^228.

[56] A.R. Gorman, W.S. Holbrook, M.J. Hornbach, K.L.Hackwith, D. Lizarralde, I. Pecher, Migration of methanegas through the hydrate stability zone in a low-£ux hy-drate province, Geology 30 (2002) 327^330.

[57] C. Ruppel, M. Kinoshita, Fluid, methane, and energy £uxin an active margin gas hydrate province, o¡shore CostaRica, Earth Planet. Sci. Lett. 179 (2000) 153^165.

[58] W.T. Wood, J.F. Gettrust, N.R. Chapman, G.D. Spence,R.D. Hyndman, Decreased stability of methane hydratesin marine sediments owing to phase-boundary roughness,Nature 420 (2002) 656^660.

[59] H. Fossing, T.G. Ferdelman, P. Berg, Sulfate reductionand methane oxidation in continental margin sedimentsin£uenced by irrigation (South-East Atlantic o¡ Nami-bia), Geochim. Cosmochim. Acta 64 (2000) 897^910.

[60] S. D’Hondt, S. Rutherford, A.J. Spivack, Metabolic ac-tivity of subsurface life in deep-sea sediments, Science 295(2002) 2067^2070.

[61] K.-U. Hinrichs, J.M. Hayes, S.P. Sylva, P.G. Brewer,E.F. DeLong, Methane-consuming archaebacteria in ma-rine sediments, Nature 398 (1999) 802^805.

[62] A. Boetius, K. Ravenschlag, C.J. Schubert, D. Rickert, F.Widdel, A. Gieseke, R. Amann, B.B. J]rgensen, U. Witte,O. Pfannkuche, A marine microbial consortium apparent-ly mediating anaerobic oxidation of methane, Nature 407(2000) 623^626.

[63] S. Bouriak, M. Vanneste, A. Saoutkine, Inferred gas hy-drates and clay diapirs near the Storegga Slide on thesouthern edge of the Voring Plateau, o¡shore Norway,Mar. Geol. 163 (2000) 125^148.

[64] P. Cochonat, J.-P. Cadet, S.J. Lallemant, S. Mazzottid,H. Nouze¤a, C. Foucheta, J.-P. Foucher, Slope instabilitiesand gravity processes in £uid migration and tectonicallyactive environment in the eastern Nankai accretionarywedge (KAIKO-Tokai ’96 cruise), Mar. Geol. 187(2002) 193^202.

[65] G.R. Dickens, Sulfate pro¢les and barium fronts in sedi-ment on the Blake Ridge: Present and past methane£uxes through a large gas hydrate reservoir, Geochim.Cosmochim. Acta 65 (2001) 529^543.

[66] R. Sassen, S.T. Sweet, A.V. Milkov, D.A. DeFreitas,M.C. Kennicutt, Thermogenic vent gas and gas hydratein the Gulf of Mexico slope: Is gas hydrate decompositionsigni¢cant?, Geology 29 (2001) 107^110.

[67] N.M. Rodriguez, C.K. Paull, W.S. Borowski, Zonation ofauthigenic carbonates within gas hydrate-bearing sedi-mentary sections on the Blake Ridge; o¡shore southeast-ern North America, Proc. ODP Sci. Results 164 (2000)301^312.

[68] J.M. Hayes, H. Strauss, A.J. Kaufman, The abundance of13C in marine organic matter and isotopic fractionation inthe global biogeochemical cycle of carbon during the past800 Ma, Chem. Geol. 161 (1999) 103^125.

[69] E. Thomas, J.C. Zachos, Was the late Paleocene thermalmaximum a unique event?, GFF 122 (2000) 169^170.

[70] M.J. Malone, G. Claypool, J.B. Martin, G.R. Dickens,Variable methane £uxes in shallow marine systems: Thecomposition and origin of pore waters and authigenic car-bonates on the New Jersey shelf, Mar. Geol. 189 (2002)175^196.

[71] G.R Dickens, T. Fewless, E. Thomas, T.J. Bralower, Ex-cess barite accumulation during the Paleocene-Eocenethermal maximum: Massive input of dissolved bariumfrom sea£oor gas hydrate reservoirs, in: S.L. Wing etal. (Eds.), Causes and Consequences of Globally WarmClimates of the Early Paleogene, Geol. Soc. Am. Spec.Pap. 369 (2003) 11^23.

[72] K.-U. Hinrichs, L.R. Hmelo, S.P. Sylva, Molecular fossilrecord of elevated methane levels in late Pleistocene coast-al waters, Science (2003) 1214^1217.

[73] M.J. Kennedy, N. Christie-Blick, L.E. Sohl, Are Protero-zoic cap carbonates and isotopic excursions a record of



gas hydrate destabilization following Earth’s coldest inter-vals?, Geology 29 (2001) 443^446.

[74] J.P. Kennett, K.G. Cannariato, I.L. Hendy, R.J. Behl,Carbon isotopic evidence for methane hydrate instabilityduring quaternary interstadials, Science 288 (2000) 128^133.

G.R. Dickens was raised in northernCalifornia. After obtaining a ChemistryBSc at the University of California, Da-vis, he honed his numerical skills dealingblackjack at Lake Tahoe. He then at-tended the University of Michigan, col-lecting an Oceanography PhD in 1996.Since, he has been a lecturer and seniorlecturer at James Cook University, Aus-tralia, and associate professor at RiceUniversity. His initial ideas for this pa-

per were scrawled upon wet napkins in Charly’s bar with J.R.O’Neil in 1994 after a summer DOE fellowship with M.S.Quinby-Hunt.



rethinking the global carbon cycle with a large, dynamic ... · dn ex dt ¼ f in m ex ðn in3n...

Documents