silicate weathering in anoxic marine sediments · silicate weathering in methanogenic sediments...

Available online at www.sciencedirect.com

www.elsevier.com/locate/gca

Geochimica et Cosmochimica Acta 72 (2008) 3067–3090

Silicate weathering in anoxic marine sediments

K. Wallmann a,*, G. Aloisi b, M. Haeckel a, P. Tishchenko c, G. Pavlova c, J. Greinert d,S. Kutterolf a, A. Eisenhauer a

a IFM-GEOMAR Leibniz Institute of Marine Sciences, Wischhofstrasse 1-3, 24148 Kiel, Germanyb Laboratoire de Paléoenvironnements et Paléobiosphère, Université Claude Bernard, Lyon I, 2, rue Dubois, F-69622 Villeurbanne cedex, France

c Pacific Oceanological Institute (POI), 43, Baltiyskaya Street, 690041 Vladivostok, Russiad Renard Centre of Marine Geology, Department of Geology and Soil Science, Ghent University, Krijgslaan 281 s.8, B-9000 Ghent, Belgium

Received 6 September 2007; accepted in revised form 17 March 2008

Abstract

Two sediment cores retrieved at the northern slope of Sakhalin Island, Sea of Okhotsk, were analyzed for biogenic opal,organic carbon, carbonate, sulfur, major element concentrations, mineral contents, and dissolved substances including nutri-ents, sulfate, methane, major cations, humic substances, and total alkalinity. Down-core trends in mineral abundance suggestthat plagioclase feldspars and other reactive silicate phases (olivine, pyroxene, volcanic ash) are transformed into smectite inthe methanogenic sediment sections. The element ratios Na/Al, Mg/Al, and Ca/Al in the solid phase decrease with sedimentdepth indicating a loss of mobile cations with depth and producing a significant down-core increase in the chemical index ofalteration. Pore waters separated from the sediment cores are highly enriched in dissolved magnesium, total alkalinity, humicsubstances, and boron. The high contents of dissolved organic carbon in the deeper methanogenic sediment sections (50–150 mg dm�3) may promote the dissolution of silicate phases through complexation of Al3+ and other structure-building cat-ions. A non-steady state transport-reaction model was developed and applied to evaluate the down-core trends observed inthe solid and dissolved phases. Dissolved Mg and total alkalinity were used to track the in-situ rates of marine silicate weath-ering since thermodynamic equilibrium calculations showed that these tracers are not affected by ion exchange processes withsediment surfaces. The modeling showed that silicate weathering is limited to the deeper methanogenic sediment sectionwhereas reverse weathering was the dominant process in the overlying surface sediments. Depth-integrated rates of marinesilicate weathering in methanogenic sediments derived from the model (81.4–99.2 mmol CO2 m

�2 year�1) are lower thanthe marine weathering rates calculated from the solid phase data (198–245 mmol CO2 m

�2 year�1) suggesting a decrease inmarine weathering over time. The production of CO2 through reverse weathering in surface sediments (4.22–15.0 mmolCO2 m

�2 year�1) is about one order of magnitude smaller than the weathering-induced CO2 consumption in the underlyingsediments. The evaluation of pore water data from other continental margin sites shows that silicate weathering is a commonprocess in methanogenic sediments. The global rate of CO2 consumption through marine silicate weathering estimated here as5–20 Tmol CO2 year

�1 is as high as the global rate of continental silicate weathering.� 2008 Elsevier Ltd. All rights reserved.

1. INTRODUCTION

The weathering of silicate minerals exposed on the con-tinents is the largest sink of atmospheric CO2 on geological

0016-7037/$ - see front matter � 2008 Elsevier Ltd. All rights reserved.doi:10.1016/j.gca.2008.03.026

* Corresponding author. Fax: +49 430 600 2827.E-mail address: [email protected] (K. Wallmann).

time scales (Wallmann, 2001). Clay minerals and dissolvedmetal ions are also produced during this reaction whichmay be summarized as:

Reactive silicatesþ CO2! clay mineralsþ dissolved cationsþ dissolved silicaþHCO3� ð1Þ

mailto:[email protected]

Fig. 1. Study area.

3068 K. Wallmann et al. / Geochimica et Cosmochimica Acta 72 (2008) 3067–3090

The rate of this process is positively correlated with globalmean temperature and atmospheric CO2 content, resultingin a negative feedback that stabilizes earths’ climate (Ber-ner, 2004).

Detrital silicates derived from the physical denudationof the continents and products of continental weatheringare major components of marine sediments. They affectthe composition of seawater and pore fluids through ion ex-change processes (Sayles and Mangelsdorf, 1977; Saylesand Mangelsdorf, 1979; von Breymann et al., 1990), chem-ical weathering (Maher et al., 2004), and reverse weathering(Michalopoulos and Aller, 1995; Michalopoulos et al.,2001). Maher et al. (2004) recently proposed that fine-grained clastic sediments deposited at the deep-sea flooroff Iceland may weather as rapidly as terrestrial soils of sim-ilar age. The effects of silicate weathering may, however, becompensated by reverse weathering. This process occurs inanoxic marine sediments that are rich in biogenic opal, me-tal hydroxides, and organic carbon (Michalopoulos and Al-ler, 1995; Michalopoulos et al., 2001). During reverseweathering, metal hydroxides and biogenic opal combineto form authigenic clay minerals:

Biogenic opalþmetal hydroxidesþ dissolved cationsþHCO3� ! clay mineralsþ CO2 ð2Þ

The metal oxides and hydroxides (Al(OH)3, Fe(OH)3) in-volved in this process are produced during continentalweathering under tropical conditions. The reaction istermed reverse weathering since dissolved cations are fixedin authigenic clays while bicarbonate is transformed intoCO2. It may serve as a major sink for seawater cations suchas K, Mg, and Li (Stoffyn-Egli and Mackenzie, 1984). Ratesof reverse weathering are, however, poorly constrained. Theoverall balance of marine silicate weathering, ion exchange,and reverse weathering in anoxic sediments is currentlyunknown.

Here, we present new pore water data from anoxic slopedeposits off Sakhalin Island showing unusually high con-centrations of dissolved Mg and total alkalinity (TA).Numerical modeling reveals that reverse weathering is lim-ited to the upper sulfate-bearing zone whereas silicateweathering dominates in the underlying methanogenic sed-iment section. We further show that the down-core increasein dissolved Mg and TA is not caused by cation exchangeon the surface of clay minerals but rather by the dissolutionof reactive silicate phases. The global rate of marine silicateweathering is estimated to be as high as the rate of conti-nental silicate weathering.

2. STUDY AREA

Sakhalin Island is situated at the north-western bound-ary of the Sea of Okhotsk, a large marginal sea located inthe north-western Pacific (Fig. 1). During the cold season,the Sea of Okhotsk is largely covered with sea-ice. Primaryproduction is low during winter and summer but very in-tense during spring and autumn (Broerse et al., 2000).The northern slope and shelf area of Sakhalin Island isstrongly influenced by fresh water and sediment input fromthe adjacent Amur River (Nakatsuka et al., 2004a). Here,

primary production is at its maximum and may reach val-ues of up to 250 g C m�2 year�1 during the warm season(Antoine et al., 1996). Primary production is dominatedby diatoms and sinking particles are mainly composed ofbiogenic opal and lithogenic material delivered by theAmur River. Most of these particles are initially depositedon the shelf and are subsequently eroded by bottom cur-rents (Nakatsuka et al., 2004a). The re-suspended particlesare transported down-slope by dense shelf waters producedduring sea-ice formation (Nakatsuka et al., 2004a). The lat-eral POC flux induced by down-slope transport is much lar-ger than the sinking POC flux from in-situ surface waterproduction (Nakatsuka et al., 2004b). Most of the sedimen-tary POC in slope sediments is thus derived from shelfdeposits. The CaCO3 flux to slope sediments is mainly com-posed of foraminifera and is one order of magnitude smal-ler than the biogenic opal flux (Nakatsuka et al., 2004a).Sediment particles deposited at the slope of Sakhalin Islandare thus mainly derived from the adjacent shelf area withthe Amur River serving as the ultimate source of the litho-genic fraction.

Slope sediments off Sakhalin are mainly composed ofbiogenic opal and terrigenous matter. The biogenic opalcontents are high in Holocene surface sediments and de-crease with sediment depth (Goldberg et al., 2005b). Thelow biogenic opal and biogenic barite contents in glacialsediments probably reflect reduced productivities in theslope and shelf area under glacial conditions (Goldberget al., 2005a). This decline in productivity might be relatedto the changing water discharge of the Amur River whichwas significantly reduced under glacial conditions (Nürn-berg and Tiedemann, 2004). Sediment cores taken in thecentral Sea of Okhotsk also suggest enhanced productivitiesduring the Holocene and strong productivity maxima dur-ing glacial terminations (Nürnberg and Tiedemann, 2004).Volcanic ashes produced in the Kamchatka–Kuriles regionare mainly deposited in the eastern and southern Sea ofOkhotsk but are also found in northern slope deposits(Goldberg et al., 2005b). Holocene sedimentation ratesare as high as 100 cm kyr�1 in the northern slope area

Silicate weathering in anoxic marine sediments 3069

and decrease towards the south by one order of magnitude(Wong et al., 2003; Wallmann et al., 2006).

The Amur River is one of the largest rivers of easternEurasia. It drains an area of 1.855 � 10+6 km2 (Gaillardetet al., 1999) located in the eastern part of the Central AsianOrogenic Belt (CAOB). Half of the drainage basin is in NEChina (Heilongliang Province) and half lies in Russia. Themean annual fresh water flux from the Amur Riveramounts to 344 km3 year�1 (Gaillardet et al., 1999; Ogiand Tachibana, 2006). The chemistry of the Amur Riverwater is dominated by dissolved Ca (223 lM) and bicar-bonate (477 lM) that are released into solution during car-bonate and silicate weathering (Gaillardet et al., 1999). TheCO2 consumption via silicate weathering amounts to33 � 10+9 mol year�1 over the entire Amur drainage area(Gaillardet et al., 1999). The corresponding area-weightedsilicate weathering rate (18 � 10+3 mol CO2 km�2 year�1)is significantly lower than the global average silicate weath-ering rate of 98 � 10+3 mol CO2 km�2 year�1 (Gaillardetet al., 1999). The ratio between total suspended and totaldissolved solids is larger than unity (Gaillardet et al.,1999) indicating a rather low efficiency of chemical weath-ering which is probably related to low temperatures andpermafrost conditions prevailing in the Amur Riverbasin.

3. SAMPLING AND ANALYTICAL TECHNIQUES

Sediments were taken with a piston corer (KL) duringcruise SO178 with RV SONNE in July–September 2004at the northern slope of Sakhalin Island (Fig. 1). CoreKL-13-6 was retrieved at a water depth of 713 m (position:52�43.880N; 144�42,650E) while core KL-29-2 was taken fur-ther north (53�50.000N; 144�14,230E) at 771 m water depth.Sediment samples and pore fluids were processed as previ-ously described (Wallmann et al., 2006). Pore fluids wereanalyzed on board for dissolved Ca, Mg, total alkalinity(TA), total dissolved sulfide (TH2S), silica (H4SiO4), ammo-nium (NH4), chloride (Cl), sulfate (SO4), and methane(CH4) applying complexometric and direct titration (Ca,Mg, TA), photometry (TH2S, H4SiO4, NH4), ion-chroma-tography (Cl, SO4), and gas chromatography (CH4) (Wall-mann et al., 2006). The concentration of dissolved humicsubstances was measured by UV-absorption spectroscopyat a wavelength of 254 nm using humic acid as a standardsolution which was extracted from shallow marine sedi-ments taken at Armur Bay, Vladivostok. Linear and wellreproducible calibration curves were obtained in the con-centration range of 0–50 mg/l. The standard deviation ofthis method was found to be 2%. The UV absorption ofpore water samples at 254 nm is, however, not very specificand sensitive towards artifacts related to oxidation pro-cesses. The absorption may be enhanced by ferric ironhydroxides produced by the oxidation of dissolved ferrousiron as well as by poly-sulfides and elemental sulfur formedby oxidation of dissolved sulfide. Samples were, hence, de-gassed and bubbled with argon gas to remove dissolved sul-fide and to minimize the contribution of oxidized sulfur andiron species to the measured UV absorption. Dissolved Na,K, Li, B, Mn, and Fe were determined on-shore in acidified

sub-samples by ICP-AES (http://www.ifm-geomar.de/index.php?id=1858&L=1).

Sediments were analyzed for particulate organic carbon(POC), carbonate (CaCO3), and total sulfur (S) with an ele-ment analyzer (Wallmann et al., 2006). A re-evaluation ofthe POC and S data revealed that the POC and S data pre-viously reported for core KL-29-2 (Wallmann et al., 2006)were erroneous. The error occurred in the data manage-ment and has no effect on the model results and conclusionsof this previous paper. Biogenic opal contents were mea-sured with an automated wet-leaching method (Müllerand Schneider, 1993). The relative precision of this methodis better than 2% for opal-rich samples and 4–10% for sam-ples with an opal content less than 10 wt%. Volcanic ashesare also dissolved by this method. The biogenic opal valuesare reported as SiO2 concentrations in wt% without any as-sumed water content for biogenic opal. The lithogenic con-tents were calculated as (Haake et al., 1993):

Lithogenics ðwt�%Þ ¼ 100� CaCO3 � 1:8� POC� 1:1� biogenic opal ð3Þ

Element concentrations in the solid phase (Na, K, Mg, Ca,Al, Ti) were determined by X-ray fluorescence spectros-copy. Sediment crushed to


analyses by using a 0.5 mm grid and by counting at least300 points. The following minerals and phases are detectedand quantified (vol.%) by this method: clay/silt matrix, bio-genetic clasts, quartz, lithoclasts, plagioclase, orthoclase, al-tered feldspars, olivine, pyroxene, amphibole, glass shards,chlorite, tourmaline, garnet, and zircon.

The grain size distribution was determined by wet siev-ing (fraction >63lm) and by measurements with a laser-based automatic grain size analyzer (Sedigraph 5100). Thefraction 5% silt) sediments(Bianchi et al., 1999).

4. NUMERICAL MODELING PROCEDURE

A numerical transport-reaction model was developed tosimulate the degradation of particulate organic matter(POM) and the weathering of silicates in anoxic marine sed-iments. The model calculates the concentration depth pro-files of 10 solid species (particulate organic carbon,particulate nitrogen, adsorbed ammonium, CaCO3, S, Ti,Na, K, Mg, terrigenous Ca) and 11 dissolved species (sul-fate, methane, ammonium, total dissolved sulfide (TH2S),total dissolved inorganic carbon (DIC), total alkalinity(TA), dissolved silica (H4SiO4), Na, K, Mg, Ca). Majorprocesses considered in the model are POM degradationvia sulfate reduction, methanogenesis, anaerobic oxidationof methane (AOM), carbonate precipitation and dissolu-tion, ammonium adsorption, and silicate weathering. A de-tailed description of the model is given in Appendix A.

Partial differential equations for solids and solutes wereset-up following the classical approach used in early diagen-esis modeling (Berner, 1980):

Solutes : U �oCot¼

o U �DS � oCox� �

ox�oðU �v �CÞ

oxþU �R ð4Þ

Solids : ð1�UÞ �oGot¼�oðð1�UÞ �w �GÞ

oxþð1�UÞ �R ð5Þ

where x is depth, t is time, U is porosity, C is the concentra-tion of dissolved species in pore water, v is the burial veloc-ity of solutes, G is the concentration of solids in drysediments, w gives the burial velocity of solids, and R de-fines the reactions occurring in the simulated sediment do-main. The model considers the decrease in porosity withsediment depth, advective transport of solutes and solidsvia burial and steady state compaction, molecular diffusionof dissolved species and various microbially mediated andchemical reactions. The burial velocities of solutes and sol-ids (v and w) were derived from a nitrogen mass balance atsteady state (Wallmann et al., 2006).

Total alkalinity (TA) was approximated by the follow-ing equation:

TA ¼ HCO3� þ 2 � CO32� þHS� ð6Þ

since dissolved bicarbonate ðHCO3�Þ, carbonate ðCO32�Þ,and bisulfide (HS�) are the major weak bases in anoxicpore fluids. The contribution of other minor pore waterconstituents ðBðOHÞ4�;H

þ;OH�;NH3; . . .Þ was neglectedfor simplicity.

Diffusive transport of TA, total dissolved inorganic car-bon ðDIC ¼ CO2 þHCO3�;CO32�Þ and total dissolvedsulfide (TH2S = H2S + HS

�) is governed by the diffusioncoefficients and concentration gradients of the individualspecies contributing to the total concentrations (Luffet al., 2001). The following differential equation was, thus,used to calculate the vertical distribution of TH2S:

U � oCðTH2SÞot

¼o U � DSðHS�Þ � oCðHS

�Þox þ �U � DSðH2SÞ �

oCðH2SÞox

� �ox

� oðU � v � CðTH2SÞÞox

þ U � R ð7Þ

Analogous equations were applied for TA and DIC (VanCappellen and Wang, 1996). Concentrations of CO2,HCO3

�;CO32�, H2S, and HS

� as well as pH values werecalculated for each time step and depth interval as a func-tion of ambient TA, DIC, and TH2S applying analyticalsolutions of the carbonate system. Dissociation constantsof CO2, HCO3

�, and H2S appearing in these analyticalsolutions were corrected for ambient salinity, temperature,and pressure conditions and were treated as depth-depen-dent variables considering the increase in pressure and tem-perature with sediment depth (Zeebe and Wolf-Gladrow,2001).

Degradation of particulate organic carbon (POC) wasdescribed by the following kinetic rate law (Middelburg,1989; Wallmann et al., 2006):

RPOC ¼KC

CðDICÞ þ CðCH4Þ þ KC� 0:16 � ða0 þ axÞ�0:95

� GðPOCÞ ð8Þ

where RPOC is the POC degradation rate, C(DIC) is theconcentration of dissolved inorganic carbon, C(CH4) isthe methane concentration in pore waters, a0 is the initialage of the deposited organic matter, ax is the depth-depen-dent age of sedimentary organic matter, G(POC) is the POCconcentration and KC is a Monod constant describing theinhibition of POC degradation by DIC and CH4 (Wall-mann et al., 2006).

Silicate weathering consumes CO2 and induces the re-lease of dissolved cations and TA. The overall stoichiome-try of this process may, thus, be given as:

reactive silicatesþ ðaþ bþ cþ dÞCO2

! authigenic silicatesþ a2

Mg2þ þ b2

Ca2þ þ cNaþ

þ dKþ þ ðaþ bþ cþ dÞHCO�3 ð9Þ

Silicate weathering rates were mainly constrained by Mgand TA data. Dissolved Mg was used to trace the depth-dependent weathering rate since terrigenous particles werethe only significant source of dissolved Mg in the studied


sediments. For this purpose empirical functions are fittedthrough the Mg pore water data to obtain representativeconcentrations for each depth interval (C(Mg)OBS). TheMg release rate (RMg) is set proportional to the differencebetween measured (C(Mg)OBS) and modeled (C(Mg))concentrations:

RMg ¼ kMg � ðCðMgÞOBS � CðMgÞÞ ð10Þ

It should be noted that Eq. (10) is not a mechanistic kineticrate law. It is rather applied as an inverse modeling tool.The rate of Mg release (RMg) is calculated from the diver-gence between observed and modeled Mg concentrations.A large value is assigned to the constant kMg to maintainthe divergence at a very low level and to ensure that themodel closely tracks the measured Mg values(kMg P 0.05 year

�1). With this approach, observed Mgconcentrations are applied as external forcing to the modeland the RMg values are determined without specifying thekinetic rate law for Mg release and silicate weathering. Sen-sitivity tests showed that the resulting RMg values are notdepending on kMg if kMg P 0.05 year�1. They only depen-dent on measured Mg concentrations and the diffusiveand advective transport of dissolved Mg in the model col-umn. The release of other cations and species is set propor-tional to the Mg release rate. The weathering rate (RWE) interms of total alkalinity production and CO2 consumptionwas thus calculated as:

RWE ¼ 2 � ðRMg þ RCaÞ þ RNa þ RK¼ ð2þ 2 � rCa=Mg þ rNa=Mg þ rK=MgÞ � RMg¼ rTA=Mg � RMg ð11Þ

where rCa/Mg, rNa/Mg, rK/Mg, and rTA/Mg give the release ofCa, Na, K, and TA normalized to RMg.Total alkalinity isnot only produced by silicate weathering but is also affectedby sulfate reduction, ammonium release during POM deg-radation, AOM, and carbonate precipitation. Consideringthe following stoichiometries:

Sulfate reduction ðRSRÞ : 2CðH2OÞþSO42�!H2Sþ2HCO3�

Ammonium release ðRNH4 Þ : �NH3þ Hþ !NHþ4AOM ðRAOMÞ : CH4þSO42� !HS

� þHCO3�þH2OCaCO3 precipitation ðRCaCO3 Þ : Ca

2þ þHCO3� !CaCO3þHþ

ð12Þ

the rate of TA production (R(TA)) is calculated as:

RðTAÞ ¼ RWE þ 2 � RSR þ RNH4 þ 2 � RAOM � 2 � RCaCO3ð13Þ

Rate laws for the TA producing reactions are given in TableA3.

Dissolved sulfide is removed from pore waters by theprecipitation of iron sulfide minerals and other solids.The following stoichiometry is assumed for sulfide precipi-tation processes:

H2Sþ 2=5Fe2O3 ! 2=5FeS2 þ 1=5FeSþ 1=5FeOþH2Oð14Þ

Sulfide is, thus, fixed in pyrite (FeS2) and iron mono sulfides(FeS) while ferric iron oxides (Fe2O3) are reduced to ferrous

iron minerals. Eq. (14) also considers the reduction of ironin clay minerals, e.g., the conversion of Fe2O3 into FeO, in-duced by dissolved sulfide (Raiswell and Canfield, 1996). Inthis formulation, the TA balance of the pore fluids is not af-fected by sulfide precipitation processes.

Pathways of fermentation and methanogenesis are verydiverse. Charge balance constraints, however, guaranteethat uncharged organic carbon molecules are convertedinto a mixture of metabolites having no net effect on totalalkalinity (see for example Eq. (15)):

Fermentation : 2CðH2OÞ!CH3COO�þHþ orCðH2OÞþH2O!CO2þ2H2

Methanogenesis : CO2þ4H2!CH4þ2H2O orCH3COO

�þHþ !CH4þCO2

ð15Þ

In methanogenic sediments TA is thus only produced bythe degradation of organic nitrogen compounds and by sil-icate weathering.

Constant concentrations of dissolved species were pre-scribed at the upper and lower boundary of the model col-umn (Dirichlet boundary conditions). The model domainexcludes the thin oxic and suboxic surface layers and is re-stricted to the underlying anoxic sediment column. Biotur-bation and bioirrigation as well as oxic and suboxicdiagenesis are thus not included in the model. Concentra-tions of dissolved species used as boundary values were ta-ken from our data set using samples taken at theuppermost and deepest points of the studied sedimentcores. Reliable methane and DIC concentration measure-ments are not available since most of the dissolved meth-ane and some of the dissolved CO2 are lost duringsampling. Methane concentrations at the lower boundarywere, thus, varied until the resulting AOM rates producedsulfate penetration depths consistent with the data. DICconcentrations at the base of the two studied sedimentcores were set to high values to approach saturation ofthe pore fluids with respect to calcite. Solid phase tracers(POC and Ti) indicate non-steady state conditions at thetwo sampling sites. To mimic this situation, POC and Ticoncentrations at the top of the model column were al-lowed to change over time. The time-dependent interfacialPOC and Ti concentrations applied in the modeling(Fig. 2) were derived from the data considering sedimentages and compositions. In the modeling, Ti is treated asan inert tracer. The calculated Ti concentration depth pro-files are thus only determined by burial, compaction, andthe time-dependent concentration applied at the upperboundary of the model column. The interfacial concentra-tions of terrigenous Mg, Ca, Na, and K were assumed totrack the corresponding Ti concentrations. Down-corechanges in the calculated element/Ti ratios are thus en-tirely due to the turnover of alkaline and alkaline earthmetals during silicate weathering and reverse weatheringprocesses. Prior to each non-steady state simulation, themodel column was equilibrated over a model period of30 kyr applying constant upper boundary values corre-sponding to the POC and Ti concentrations measured atthe base of the sediment cores. The solid phase and porewater distributions generated by these steady state model

Fig. 2. Upper boundary values for particulate organic carbon(POC) and Ti applied in the model runs.

Table 1Grain size distribution and mineral contents

Parameter KL-13-6 Kl-29-2

Clay fraction (63 lm) in wt% 4 ± 1 3 ± 1Mean diameter of the silt fraction in lm 8 ± 1 10 ± 1Smectite in % of clay 50 ± 8 49 ± 10Illite in % of clay 30 ± 6 23 ± 10Kaolinite in % of clay 19 ± 5 28 ± 10Quartz in % of total 6.2 ± 1.9 3.4 ± 0.9Feldspars in % of total 10.0 ± 0.8 7.9 ± 1.5Pyroxene in % of total 2.9 ± 0.5 4.0 ± 0.5Olivine in % of total 0.9 ± 0.5 0.3 ± 0.3Volcanic glass in % of total 0.4 ± 0.4 2.5 ± 0.5

Numbers indicate mean values averaged over the entire core lengthand their corresponding standard deviations.


runs were applied as initial values for the non-steady statesimulations.

Finite difference techniques (the method-of-lines code)which have been successfully applied in previous modelsof early diagenesis (Boudreau, 1996b; Luff et al., 2000; Luffand Wallmann, 2003) were used to solve the model. The setof 21 partial differential equations defining the model (onefor each species) is converted into a large number of ordin-ary differential equations (ODE) giving the temporalchange of species concentrations at each depth interval. Acentered finite difference scheme was used for dissolved spe-cies while an upward scheme was applied for the transportof solids. The ODE system was set-up on an uneven gridwith higher resolution at the surface. It was solved usingthe NDSolve object of MATHEMATICA Version 5.2.

5. RESULTS AND DISCUSSION

The data discussed in the following sections clearly indi-cate that Sakhalin slope sediments and pore fluids arestrongly altered by in-situ silicate weathering. The solidphase data (Section 5.1) suggest a down-core decrease inreactive silicate phases (plagioclase) and cation abundanceaccompanied by an increase in smectite contents and chem-ical alteration. Pore waters are highly enriched in dissolvedcations, total alkalinity, and dissolved humic substances(Section 5.2). The down-core decrease in cation concentra-tions observed in the solid phase is matched by a corre-sponding increase in dissolved cation concentrationssuggesting that the cation distributions are controlled byin-situ weathering processes. The dissolution of silicateminerals may be promoted by complex formation betweenlattice-forming cations and dissolved humic and fulvic acidanions. Numerical transport-reaction modeling (Section5.3) suggests that the release of dissolved cations and the

production of total alkalinity can only be explained bymarine silicate weathering. This interpretation is supportedby thermodynamic equilibrium calculations showing thatcation exchange reactions between sediments and porewaters have only a minor effect on observed cation distribu-tions (Section 5.4). Rates of marine silicate weathering inSakhalin slope sediments are shown to be higher than therates of continental silicate weathering in the hinterlanddrained by the Amur River (Section 5.5). Reverse weather-ing (Section 5.6) is limited to the upper sulfate-bearing zoneof the studied sediments and proceeds at a much lower ratethan silicate weathering in the underlying methanogenicsediments. Global rates of marine silicate weathering (Sec-tion 5.7) may be as high as the rates of continental silicateweathering since marine silicate weathering is a commonprocess in methanogenic sediments. Anoxic diagenesisincluding silicate weathering is finally shown to have a pro-found effect on global and marine carbon cycling (Section5.8).

5.1. Solid phase composition

The sediments are composed of a mixture of silt and claywith minor amounts of coarse material (Table 1). In coreKl-13-6, the mean diameter within the silt fraction (2–63 lm) decreases with sediment depth from 9 lm at the coretop to 6 lm at the core base. The relative abundance of clayminerals as determined by XRD measurements shows dis-tinct variations between the two cores (Table 1) and withsediment depths (Fig. 3). Core KL-29-2 has higher kaoliniteand lower illite contents than core KL-13-6. Smectite is themost abundant clay mineral at both sites and the relativeabundance of smectite increases with sediment depth inboth cores. Smectite is often formed by the weathering ofigneous minerals and volcanic glasses in soils and marinesediments (Schulz and Zabel, 2006). The down-core in-crease could thus either reflect changing weathering condi-tions on land or the in-situ weathering and alteration ofreactive silicate phases within the studied sediment cores.Quartz contents determined by XRD (Table 1) show astrong increase with sediment depth in core KL-13-6 andrather constant values in KL-29-2 (Fig. 4). Microscopic

Fig. 3. Smectite, illite and kaolinite contents. Values are given as fractions of total clay content.

Fig. 4. Quartz contents.


investigations revealed that feldspars are a major compo-nent of Sakhalin slope sediments (Table 1). They are mainlycomposed of plagioclase whereas potassium-bearing feld-spars are only a minor component (

Fig. 5. Chemical composition of sediments. Si* and Ca* reflect the element fractions bound in terrigenous phases. They are calculated asdifference between the total element contents determined by XRF measurements and the element fractions bound in biogenic phases (biogenicopal (BOpal) and CaCO3). Element ratios refer to concentrations given in wt%. The content of lithogenic phases and the chemical index ofalteration (CIA) are calculated as explained in the text.


5.2. Pore water composition

Dissolved chloride showed no significant down-coretrend and very little variation between the two samplingsites (data not shown). Mean chloride concentrations incore KL-13-6 and core KL-29-2 are 553 ± 4 and556 ± 4 mM, respectively. Dissolved sulfate decreases rap-idly with depth and is almost completely consumed at sed-iment depths of 3 m (core KL-13-6, Fig. 8) and 5 m (coreKL-29-2, Fig. 9). The shallow sulfate penetration clearlyindicates high rates of organic matter degradation in thestudied sediment cores (Wallmann et al., 2006). Organicmatter is microbially converted into dissolved methaneand CO2 below the sulfate penetration depth. Upwards dif-fusing methane is consumed within the AOM zone markedby a strong TH2S maximum located at 3–6 m sedimentdepth. Dissolved ammonium is produced by the anaerobicdegradation of organic matter and shows a regular down-core increase. The ammonium increase is accompanied bya strong increase in total alkalinity (TA). The steep TA in-

crease within the sulfate reduction and AOM zone towardsa concentration of about 60 mM is expected since largeamounts of TA are produced via sulfate reduction andAOM (Eq. (12)). The steady TA increase within the under-lying methanogenic zone is, however, not explained by con-ventional diagenetic theory since CO2 rather thancarbonate alkalinity is produced during methanogenesis(Eq. (15)). The high TA values clearly show that additionalprocesses such as marine silicate weathering contribute tothe observed TA accumulation. Ca concentrations decreasetowards the AOM zone where TA production induces theprecipitation of authigenic carbonates (Wallmann et al.,2006). Concentrations are further diminished within theunderlying methanogenic sediment section even thoughmicrobial CO2 produced in this zone should induce carbon-ate dissolution rather than precipitation. This observationindicates that additional TA is produced within the metha-nogenic zone that does not accumulate in the pore fluidsbut is removed during CaCO3 precipitation. An increasein dissolved Ca concentrations is only observed at the base

Fig. 6. Solid phase composition of core KL-13-6. Squares indicate data and model results are plotted as solid lines. Dotted lines representelement ratios applied as upper boundary values and the CaCO3 contents calculated in a model run without silicate weathering.


of the sediment cores and reflects Ca mobilization in deepersediment layers.

Mg concentrations are low in sulfate-bearing surfacesediments and increase rapidly within the methanogeniczone. The Mg enrichment is probably caused by the disso-lution of Mg-bearing silicate phases. Mg-rich mineralsoccurring in the studied sediments (olivine, pyroxenes) arehighly reactive and could release large amounts of dissolvedMg into the pore water. Smectite alteration could also con-tribute to the observed Mg release since Mg is preferentiallyreleased into solution during the incongruent dissolution ofthis clay mineral at circum-neutral pH (Golubev et al.,2006). Ca could be released into solution by the dissolutionof plagioclase and pyroxene. The release of Ca via silicateweathering is, however, masked by the coeval precipitationof authigenic carbonates. Precipitation of authigenic car-bonates could also remove dissolved Mg from solutionvia dolomite and high-Mg calcite formation (Mooreet al., 2004). Solid phase analysis (smear slide observationsand XRD), however, failed to detect any dolomite and thedissolved Mg profile showed no indication for Mg removal.We, thus, suggest that Mg-poor carbonates such as arago-nite and low-Mg calcite are the major authigenic carbonates

formed in Sakhalin slope sediments and assume that Mg re-moval via carbonate precipitation is of only minor impor-tance at the investigated sites. Dissolved Na and Kconcentrations measured by optical ICP show significantlymore scatter than the Ca and Mg values determined bytitration procedures. The down-core increase in Na and Kcould also be related to silicate weathering processes. Dis-solved silica is almost constant in core KL-13-6 and de-creases towards the base of core Kl-29-2. It is controlledthrough the dissolution of biogenic opal and other reactivesilicate phases and the precipitation of authigenic silicates.The lack of significant silica enrichments in the deeper sed-iment layers suggests that cation-rich silicate phases areconverted into cation-depleted authigenic silicates duringmarine weathering.

Dissolved humics are highly enriched in methanogenicsediments. Assuming a carbon content of 50% in the humicsubstances, the values measured at the base of cores KL-13-6 and Kl-29-2 correspond to dissolved organic carbon(DOC) concentrations of 60–150 mg l�1. The contributionof humic substances to total alkalinity was determined bytwo independent titration methods (Tishchenko et al.,2006). The highest contribution was found at the base of

Fig. 7. Solid phase composition of core KL-29-2. See legend of Fig. 6 for further information.


core KL 13-6 where up to 3 meq of protons dm�3 were ta-ken-up by dissolved organic acid anions during the titrationcorresponding to less than 3% of the measured total alka-linity value (Tishchenko et al., 2006). DOC concentrationsin anoxic marine pore waters are 1–2 orders of magnitudehigher than DOC values in oxic pore fluids and show a stea-dy increase with sediment depth (Starikova, 1970; Nissen-baum, 1972; Krom and Sholkovitz, 1977; Barcelona,1980; Chen et al., 1993). DOC concentrations in the rangeof 50–150 mg l�1 were previously observed in reducing sed-iments with high particulate organic matter contents (Nis-senbaum, 1972; Krom and Sholkovitz, 1977; Barcelona,1980; Chen et al., 1993). Spectroscopic measurements indi-cate that most of the DOC in anoxic pore fluids is com-posed of dissolved humic substances (Krom andSholkovitz, 1977; Chen et al., 1993; Komada et al., 2004).Humic acids and fulvic acid anions are strong ligands form-ing stable complexes with dissolved metal cations such as Aland Fe (Stevenson, 1982). This complex formation maysupport the disintegration of Al- and Fe-bearing silicatephases. Studies on the continental weathering of silicatesshow that weathering rates are enhanced by dissolved hu-mic substances (Viers et al., 1997). It is quite likely thatthe weathering of silicates in anoxic marine sediments is

also promoted by high contents of dissolved humics in asso-ciated pore fluids.

Trace element concentrations confirm that silicateweathering may occur in the deep methanogenic sedi-ment layers (Fig. 10). Boron concentrations increase stea-dily with sediment depths while lithium shows apronounced decrease in sulfate-bearing surface sedimentsand a minimum within the underlying AOM zone. Li re-moval as indicated by the pore water data is usually as-cribed to the uptake of Li in authigenic silicate phasesformed by volcanic ash alteration (Zhang et al., 1998)or reverse weathering (Stoffyn-Egli and Mackenzie,1984) and to the adsorption of Li on sediment particlesurfaces (James and Palmer, 2000). The Li increase to-wards the base of the cores may again be explained bysilicate weathering (Zhang et al., 1998) or changes inadsorption equilibria. The steady increase in dissolvedB could be caused by the dissolution of B-bearing sili-cate phases and may be supported by the release of bo-rate anions from sediment surfaces. The concentration ofborate anions is controlled by ambient pH values. Ther-modynamic equilibrium calculations following Zeebe andWolf-Gladrow (2001) show that the borate concentrationdecrease from 26 lM at the surface (pH 7.5,

Fig. 8. Pore water composition of core KL-13-6. Squares indicate data and model results are plotted as solid lines. Dotted lines represent theresults of a model run without silicate weathering. The saturation state of the pore fluids with respect to calcite (Sat.) and rates of silicateweathering (RWE) are calculated as outlined in the text.


B = 0.4 mM) to only 2 lM at the base of the cores (pH6.0, B = 1.0 mM) under in-situ pressure and temperatureconditions (80 bar, 2 �C). The steady increase in dis-solved B may thus be related to the strong down-coredecrease in pH promoting the release of borate anionsfrom sediment surfaces. Iron and manganese concentra-tions are low within the upper 15 m of the sediment col-umn and increase towards the base of the cores. Theincrease in Mn and Fe (Fig. 10) is accompanied by acorresponding rise in dissolved Ca concentrations (Figs.8 and 9). Low Mn and Fe values occur in pore fluidsoversaturated with respect to calcite and aragonite (Figs.8 and 9) whereas high values are attained in less satu-rated core base fluids. Dissolved Mn and Fe concentra-tions are, thus, probably controlled by the precipitationand dissolution of Mn- and Fe-bearing carbonate phases.

5.3. Model results

A non-steady state approach was applied in the model-ing since the down-core changes in the solid tracers POCand Ti suggest significant changes in the delivery of organicand terrigenous matter to the sediment surface over time.The concentrations of both solid phase species at the sedi-ment surface were thus varied over the model period(Fig. 2) to fit the depth profiles of the reactive tracer POCand the inert tracer Ti (Figs. 6 and 7). The cation/Ti ratiosat the sediment surface were maintained at a constant valuesuch that the delivery of solid phase cations to the sedimentsurface tracked the changing interfacial Ti fluxes. Down-core changes in calculated cation/Ti ratios (Figs. 6 and 7)are thus entirely due to reactions occurring within the mod-el column.

Fig. 9. Pore water composition of core KL-29-2. See legend of Fig. 8 for further information.


Organic matter degradation processes as driven by thechanging POC input affect the depth profiles of POC, dis-solved sulfate, sulfide, methane, DIC, TA and ammonium.The good fit to observations obtained for both cores (Figs.8 and 9) shows that the microbial turnover of POC in thestudied sediments is appropriately described by the kineticmodel (Eq. (8)) applied in the simulations (Wallmannet al., 2006). Dissolved sulfate and sulfide concentrationsare also affected by the anaerobic oxidation of methane(AOM). Dissolved methane concentrations at the core basewere set to high values approaching the solubility of meth-ane hydrates to match the observed sulfate penetrationdepths. Dissolved sulfide formed during sulfate reductionand AOM was partly removed from the pore water (Eq.(14)) to approach the measured concentrations. The result-ing increase in solid phase S concentrations is consistentwith the measured values (Figs. 6 and 7). Depth-integratedrates of POC degradation, sulfate reduction and AOM

(Table 2) differ slightly from the corresponding rates re-ported in Wallmann et al. (2006) because these previousestimates were derived with a steady state model consider-ing only the upper section of the sediment cores. The ratesobtained in the new non-steady state model covering the en-tire depth of the sediment cores confirm the high reactivityof Sakhalin slope sediments (Table 2).

Rates of Mg release and silicate weathering were derivedfrom the down-core change in dissolved Mg (Eqs. (10) and(11)). Mg/Ti solid phase profiles provide independent con-trol for this approach because they are entirely determinedby the Mg release rates derived from the pore water data.The close match between modeled and measured Mg/Ti ra-tios (Figs. 6 and 7) confirms that down-core changes in dis-solved Mg are caused by weathering reactions occurringwithin the sediment column sampled at stations 13-6 and29-2. An additional model run without silicate weatheringyields the curvature and shape of the dissolved element pro-

Fig. 10. Concentrations of dissolved trace elements measured inthe pore water of cores KL-13-6 and KL-29-2.

Table 2Depth-integrated turnover rates (in mmol m�2 year�1) calculatedin the transport-reaction model

Rate KL-13-6 KL-29-2

POC degradation 340 304Sulfate reduction via POC degradation 107 107Anaerobic oxidation of methane 66.4 52.4Sulfide removal 78.9 93.4Methane and CO2 production

during methanogenesis63.3 45.3

Ammonium adsorption 10.6 7.93Carbonate precipitation 32.6 27.2Mg uptake via reverse weathering 1.00 2.46CO2 release and TA consumption

via reverse weathering4.22 15.0

Mg release via silicate weathering 19.4 16.3CO2 consumption and TA production

via silicate weathering81.4 99.2

Depth of reverse weathering zone (m) 0–4.92 0–9.10Depth of silicate weathering zone (m) 4.92–23.5 9.10–25.0Length of model column (m) 23.5 25.0


files controlled by boundary values and transport processes,only (dotted lines in Figs. 8 and 9). The nonlinear shape ofthese profiles is caused by the strong down-core decrease inporosity (Figs. 6 and 7). The resulting Mg model profilesare very different from the data confirming that Mg is a sen-sitive tracer for silicate weathering.

The release of other cations is coupled to the Mg releasethrough constant cation/Mg ratios such that the release ofCa, Na, and K closely tracks the down-core profile of Mgrelease. In the modeling procedure, dissolved Ca is removedfrom the pore fluids via CaCO3 precipitation until the cal-culated Ca concentrations closely approach the measuredvalues. The atomic ratio between Mg and Ca release (Ca/Mg = 1) is, thus, not constrained through dissolved Ca val-ues but rather by solid phase CaCO3 concentrations. Thecalculated authigenic CaCO3 concentrations fall into therange of measured values (Figs. 6 and 7). It should, how-

ever, be noted that the calculated decrease in silicate-boundCa (i.e., Ca*) is much less pronounced than the down-corechange documented by the solid phase data (Figs. 6 and 7).A better fit to these data would be attained by choosing alarger Ca/Mg weathering ratio which would, however, re-sult in CaCO3 concentrations outside the measured range.Moreover, an unknown but significant fraction of the mea-sured CaCO3 content is of pelagic rather than authigenicorigin. Higher Ca/Mg ratios would therefore not be consis-tent with the data. The release of Na during weathering isconstrained by the down-core changes in dissolved cationconcentrations and particulate Na/Ti ratios. A good fit tothe pore water data (Figs. 8 and 9) was obtained applyingatomic weathering ratios of Na/Mg = 0.1 for station 13-6and 2.0 for station 29-2 (Table A1). The down-core de-crease in Na/Ti observed in core KL-29-2 (Fig. 7) is quitewell reproduced by the model whereas the irregular Na/Tiprofile of core KL-13-6 (Fig. 6) can not be explained bymarine weathering processes. Potassium release during sili-cate weathering is only constrained by the pore water datasince the K/Ti ratios show either an increase at the corebase (station 13-6, Fig. 6) or no significant down-corechange (station 29-2, Fig. 7). The K pore water data suggestan atomic weathering ratio of only 0.1 for both stations(Table A1). Thus, the data and the modeling indicate thatonly very little K is released during the marine weatheringof silicates.

Independent control on the total weathering rate is pro-vided by total alkalinity (TA) profiles since the cation re-lease is accompanied by a corresponding alkalinityproduction (Eqs. (9) and (11)). Model runs without silicateweathering (dotted lines in Figs. 8 and 9) show that the TAprofiles within the methanogenic sediment sections arestrongly affected by silicate weathering while the TA in-crease in the overlying sediments is easily explained by sul-fate reduction and AOM. The saturation state with respectto calcite and the pH values are also strongly enhanced bysilicate weathering (see solid and dotted lines in Figs. 8 and9). The good match between measured and calculated TAvalues (Figs. 8 and 9) demonstrates that the overall ratesof silicate weathering derived by the pore water modelingare well constrained (Table 2).

5.4. Cation exchange

In the previous sections it was assumed that the releaseof cations observed in the data is due to silicate weatheringprocesses. It may, however, also be argued that the cationrelease is caused by ion exchange processes. Thus, Mgenrichments in anoxic pore fluids were generally attributedto ion exchange with dissolved ammonium (Cook, 1974;Gieskes et al., 1982; Suess et al., 1987; von Breymannet al., 1990). The cation exchange capacity (CEC) of rapidlyaccumulating diatomaceous hemipelagic mud determinedat various sites falls into the range of 150–250 meq/kg(von Breymann et al., 1990). Experimental data (Saylesand Mangelsdorf, 1977; Sayles and Mangelsdorf, 1979;von Breymann et al., 1990) and numerical modeling (Char-let and Tornassat, 2005) show that the surface sites of clayminerals, river particles and sediments suspended in seawa-

Table 3Ion exchange equilibria

Surface sediments Core base sediments DA(mmolkg�1)

C(mM)

A(mmol kg�1)

C(mM)

A(mmol kg�1)

Na 460 79.8 475 77.8 �2.0K 10 5.1 12 5.3 +0.2NH4 0.1 0.05 10 4.38 +4.33Mg 53 42.5 75 42.4 �0.1Ca 10 15.0 6 13.8 �1.2H pH 7.5 2 � 10�4 pH 6.0 6 � 10�3 +5.8 � 10�3

Concentrations of dissolved species (C) represent average valuesobserved at the surface and base of cores KL-13-6 and KL-29-2.Corresponding concentrations of adsorbed species (A) are calcu-lated as described in the text. The change in adsorbed speciesconcentrations (DA) is defined as difference between core base andcore top values.


ter are occupied by Na, Mg, Ca, and K with relative pro-portions of 30–50% Na, 30–50% Mg, 10–20% Ca, and 2–10% K on equivalent base. These proportions translate intoexchangeable cation concentrations of 60–100 mmolNa kg�1, 30–50 mmol Mg kg�1, 10–20 mmol Ca kg�1, and4–20 mmol K kg�1 assuming a CEC value of 200 meq/kgfor Sakhalin slope sediments. The exchangeable fractionscontribute significantly to the total element concentrationsmeasured in the studied surface sediments. Ion exchangeprocesses could, thus, affect the dissolved and particulatemetal distributions observed in Sakhalin slope sedimentsif the ion exchange equilibria would be significantly shiftedby diagenetic reactions affecting the pore water composi-tion. The release of ammonium during the break-down oforganic matter induces ion exchange reactions (von Brey-mann et al., 1990) since ammonium is readily adsorbedon sediment particle surfaces (Mackin and Aller, 1984)while adsorbed Ca ions could be removed from ion ex-change sites through CaCO3 precipitation processes.

The distribution of cations between the dissolvedphase and ion exchange sites at the surface of solids isaffected by numerous parameters including the chemicalcomposition of fluids and solid surfaces, pressure, tem-perature, and CEC (Stumm and Morgan, 1996). The spe-ciation of dissolved metal cations has also a significanteffect on the adsorption equilibria since ion exchange siteshave different affinities to free ions, ion pairs and com-plexes. Thus, Charlet and Tornassat (2005) argue thatalkaline earth ions (Ca, Mg) are not adsorbed as freedivalent cations but rather as chloride ion pairs (CaCl+,MgCl+) at the surface of clay minerals dispersed in sea-water whereas von Breymann et al. (1990) propose thatthe formation of dissolved Mg-CO3 ion pairs in methano-genic sediments promotes the release of surface-boundMg into the pore water. Therefore, we applied thermody-namic equilibrium calculations to further investigate thespeciation of dissolved Mg in our sediment cores. Stoichi-ometric equilibrium constants valid for seawater (Millero,1996) were used to calculate the concentration of freeions considering the ion pairs Na-SO4, Na-HCO3, Na-CO3, K-SO4, Mg-SO4, Mg-HCO3, Mg-CO3, Ca-SO4,Ca-HCO3, and Ca-CO3. The resulting concentrations offree ions include the contribution of chloride ion pairsnot explicitly considered in the calculations. The equilib-rium model shows that the contribution of free Mg ionsto the total dissolved Mg concentration decreases from89% at the surface to 83% at the base of the cores dueto the formation of carbonate and bicarbonate ion pairs(Mg-HCO3, Mg-CO3). von Breymann et al. (1990) founda much higher contribution of Mg-CO3 ion pairs investi-gating the speciation of dissolved Mg in methanogenicsediments from the Peruvian margin. They applied pHvalues measured in the cores to derive CO3 and Mg-CO3 concentrations. These values are, however, erroneousbecause CO2 is lost during core retrieval and sedimentprocessing such that the measured pH values and derivedCO3 concentrations are grossly overestimated. In our cal-culations, CO3 ion concentrations are calculated frommeasured TA and modeled DIC values to avoid the useof erroneous pH data. The resulting CO3 concentrations

are more realistic and indicate that the role of Mg-CO3ion pairing in anoxic sediments is not as important aspreviously postulated.

In the following, equilibrium calculations are applied toestimate the shift in ion exchange equilibria induced by thedown-core change in pore water composition. Apparentselectivity coefficients for ion exchange between metal ionsand protons used in these calculations were taken fromMotellier et al. (2003). They describe the ion exchange equi-libria between adsorbed metal cations ðMnþADSÞ, dissolvedprotons (H+), dissolved metal cations (Mn+) and adsorbedprotons ðHþADSÞ:MnþADS þ nHþ $Mnþ þ nHþADS ð17Þ

Motellier et al. (2003) identified three different types ofadsorption sites on clays and sedimentary rocks such thatthe CEC is defined as:

CEC ¼X3i¼1

CECðiÞ

¼X3i¼1

HþADSðiÞ þNaþADSðiÞ þKþADSðiÞ

þ 2Mg2þADSðiÞ þ 2Ca2þ

ADSðiÞ ð18Þ

They derived selectivity coefficients for ion exchange be-tween protons and the cations Na+, K+, Mg2+, and Ca2+

for each type of site through experiments with clay-rich sed-imentary rocks from the Calovo-Oxfordian formation. Weapplied these coefficients and included NH4

þ as an addi-tional cation assuming that ammonium has the sameadsorption properties as potassium. The model resultslisted in Table 3 were obtained with a total CEC of200 meq kg�1 distributed over three different types of ionexchange sites (CEC(1) = 89 meq kg�1, CEC(2) =79 meq kg�1, CEC(3) = 32 meq kg�1). With this choice ofCEC(i) values, the calculated loading of surface sedimentswith cations falls into the range of values observed in exper-imental studies with marine surface sediments and clays(Sayles and Mangelsdorf, 1977; Sayles and Mangelsdorf,1979; von Breymann et al., 1990). The calculations for corebase sediments show small but significant changes in


adsorbed ion concentrations. Adsorbed ammonium in-creases in response to the release of dissolved ammoniumduring the break-down of organic matter while adsorbedCa is diminished since dissolved Ca is removed from thepore water by carbonate precipitation processes. Moreover,Na is desorbed from sediment surfaces and replaced by ad-sorbed ammonium (Table 3). The overall stoichiometry ofion exchange in the studied cores may be approximated as:

NHþ4 þ 0:3Ca2þADS þ 0:4NaþADS

! NHþ4 ADS þ 0:3Ca2þ þ 0:4Naþ ð19Þ

Changes in adsorbed metal ion concentrations (Table 3) arevery small compared to the total metal concentrations insurface sediments (Table 4). Ammonium adsorption wasconsidered in the transport-reaction model whereas theassociated release of cations was not taken into account.The depth-integrated rates of ammonium adsorption (7.9–10.6 mmol m�2 year�1, Table 2) indicate that about 4 mmolNa m�2 year�1 and up to 3 mmol Ca m�2 year�1 could bereleased through ion exchange with ammonium. The corre-sponding Na and Ca release rates derived from solid phasedata (42–104 mmol m�2 year�1, Table 4) and throughtransport-reaction modeling (1.9–32.6 mmol m�2 year�1,Tables 2 and A1) are larger showing that most of the Naand Ca release is caused by marine silicate weathering.

Adsorbed Mg remains almost constant over the entiresediment column (Table 3). The previous hypothesis thatthe increase in dissolved Mg observed in anoxic sedimentsis caused by ion exchange with ammonium (Cook, 1974;Gieskes et al., 1982; Suess et al., 1987) is thus not supportedby the modeling. The results listed in Table 3 rather suggestthat other processes such as silicate weathering have to beinvoked to explain the dissolved Mg increase. Protonadsorption is not significantly enhanced by pore water acid-ification within core base sediments (Table 3) such that thetotal alkalinity of pore fluids is not affected by protonadsorption. The use of dissolved Mg and total alkalinityas tracers for silicate weathering in anoxic sediments istherefore confirmed by ion exchange modeling.

Table 4Depth-integrated rates of element release and silicate weathering determ

KL-13-6

Surface sediments Co

Depth interval (m) 0–5 20Depth (m) 2.5 21Age (kyr) 1.2 14Porosity 0.826 0.TiO2 concentration (mmol kg

�1) 57 ± 1 88Na2O concentration (mmol kg

�1) 289 ± 5 39K2O concentration (mmol kg

�1) 156 ± 2 31MgO concentration (mmol kg�1) 368 ± 8 52CaO* concentration (mmol kg�1) 151 ± 15 98Na release rate (mmol m�2 year�1) +77 (+42 to +112)K release rate (mmol m�2 year�1) �108 (�143 to �73)Mg release rate (mmol m�2 year�1) +34 (+9.8 to +58)Ca release rate (mmol m�2 year�1) +104 (+72 to +135)CO2 consumption rate (mmol m

�2 year�1) +245 (+63 to +425)

5.5. Marine silicate weathering

The stoichiometry of silicate weathering in Sakhalinslope sediments derived through transport-reaction model-ing may be represented as (Tables 2 and A1):

cation-rich silicatesþ 5:1CO2! cation-depleted silicatesþ 5:1HCO3� þNaþ

þ 0:1Kþ þMg2þ þ Ca2þ þ 0:01SiðOHÞ4 ð20Þ

The stoichiometric coefficients in Eq. (20) correspond to theaverage element/Mg ratios listed in Table A1. The amountof CO2 being neutralized is constrained by the charge bal-ance. The stoichiometric coefficients derived from the mod-eling indicate that Si(OH)4 released during the weatheringof primary silicates is almost completely bound in authigen-ic silicate phases that are strongly depleted in Na, Mg, andCa with respect to the primary silicates. Igneous mineralsenriched in Na, Mg, and Ca (plagioclase feldspar, pyrox-ene, olivine) and volcanic glass (Table 1) may dissolve toform cation-depleted authigenic clays. The formation ofcation-leached surface layers during the incongruent disso-lution of primary silicates (Casey et al., 1993) and smectites(Golubev et al., 2006) may also contribute to the observedcation release and alkalinity production.

The pore water data clearly show that Ca2+ ions beingreleased during silicate weathering and diffusing into thesediments from the overlying bottom water are precipitatedas authigenic carbonate minerals. Depth-integrated rates ofCO2 production via methanogenesis (45.3–63.3 mmol m�2 year�1, Table 2) are smaller than thedepth-integrated CO2 consumption rates through silicateweathering (81.4–99.2 mmol m�2 year�1, Table 2). Themodeling thus suggests that silicate weathering does notonly consume methanogenic CO2 but also CO2 producedduring CaCO3 precipitation (27.2–32.6 mmol m

�2 year�1,Table 2) and additional CO2 that is infiltrating the sedi-ments from below through molecular diffusion. DissolvedCa released during silicate weathering (16.3–19.4 mmol m�2 year�1, Tables 2 and A1) is completely re-

ined from solid phase data

KL-29-2

re base sediments Surface sediments Core base sediments

–23 0–5.5 21–24.5 2.75 22.5.9 1.73 16.6658 0.823 0.763± 3 55 ± 2 68 ± 16 ± 15 325 ± 36 346 ± 91 ± 18 154 ± 6 183 ± 34 ± 19 347 ± 13 403 ± 7± 18 230 ± 35 219 ± 20

+71 (+3 to +139)+9 (�4 to +23)+17 (+2 to +31)+42 (+1 to +82)+198 (+5 to +388)


moved by CaCO3 precipitation. Considering that twoequivalents of TA are consumed by each Mol of Ca2+ beingprecipitated as CaCO3, 33% (core KL-29-2) to 48% (coreKL-13-6) of the TA generated by silicate weathering are re-moved by the precipitation of silicate-derived Ca2+ (Table2).

Depth-integrated rates of element release during silicateweathering (RRE in mmol cm

�2 year�1) can also be calcu-lated from solid phase data considering the depth and ageof surface sediments (xSur, ageSur) and core base sediments(xBase, ageBase):

RRE ¼xBase � xSur

ageBase � ageSur� dS � ð1� UBaseÞ

� GSur � GBase �TiSurTiBase

� �ð21Þ

where dS is the density of dry solids (2.5 g cm�3), G is the

concentration of the considered element in mmol kg�1 atthe base and surface of the sediment column and U is thecorresponding porosity. Concentrations at the base ofthe column are multiplied with the Ti concentration ratio(TiSur/TiBase) to account for variable contents of biogenicopal diluting the lithogenic phases.

The resulting element release rates are high and variable(Table 4). The element concentrations in surface and corebase layers represent mean values and their standard devi-ations calculated over the depth intervals listed in the firstrow of Table 4. Average release rates and the correspondingrange of possible values were calculated from mean concen-trations and their standard deviations. Subsequently, therelease of total alkalinity and the consumption of CO2 weredetermined applying the overall stoichiometry of the weath-ering reaction (Eq. (9)). It should, however, be consideredthat the composition of terrigenous phases raining to theseafloor may have changed over the investigated time inter-val. Thus, the high potassium concentrations at the base ofcore KL-13-6 probably reflect elevated inputs of terrige-nous K under glacial conditions rather than reverseweathering.

The solid phase data (Table 4) indicate the followingoverall stoichiometry:

cation� rich silicatesþ 8:5CO2 þ 1:9Kþ

! cation� depleted silicatesþ 8:5HCO3�

þ 2:8Naþ þMg2þ þ 2:8Ca2þ ð22Þ

The stoichiometry above (Eq. (22)) implies that 66% of TAis consumed by authigenic CaCO3 precipitation consideringthat Ca released during weathering is fixed in authigeniccarbonate minerals. Dissolved pore water profiles aremainly determined by rates of ongoing reactions whereassolid phase data mirror also past reaction rates. The ele-vated weathering rates derived from solid phase data (Table4) may thus point towards a more intense weathering re-gime prevailing in the past. Since the glacial climatic condi-tions were less favorable for continental weathering, thedifference between the rates derived from pore water andsolid phase data (Tables 2 and 4) suggest that the rate ofmarine weathering in Sakhalin slope sediments may havedeclined over time. Depth-integrated CO2 consumption

rates derived via transport-reaction modeling (81.4–99.2 mmol m�2 year�1, Table 2) and from solid phase data(198–245 mmol m�2 year�1) suggest, however, that marineweathering in Sakhalin slope sediments proceeds faster thancontinental silicate weathering in the hinterland drained bythe Amur River (18 mmol m�2 year�1; Gaillardet et al.,1999).

Maher et al. (2004) recently showed that fine-grainedclastic sediments deposited at the deep-sea floor off Icelandare subject to marine weathering. Plagioclase, the maincompound of these sediments, weathered at a rate of6.9 � 10�18 mol g�1 s�1 (Maher et al., 2006). Consideringthat approximately 6 mol of CO2 are neutralized duringthe dissolution of 1 mol plagioclase (Maher et al., 2006),up to 1.3 � 10�9 mol CO2 g�1 year�1 may be consumedby this process. Applying an average porosity of 0.7, a pla-gioclase content of 50 wt% and a mineral density of2.7 g cm�3, this value translates into a bulk CO2 consump-tion rate of 5 � 10�10 mol year�1 (cm3 wet sediment)�1.The maximum weathering rates observed in Sakhalin slopesediments (10–15 lmol CO2 consumption year

�1 (cm3 porewater)�1; Figs. 8 and 9) are about 4 orders of magnitudehigher than the rates derived by Maher et al. (2006). Sedi-ments studied by Maher et al. (2006) were not significantlyenriched in Mg and alkalinity and are characterized by onlymoderate rates of organic matter degradation and low ratesof methanogenesis.

5.6. Reverse weathering

The pore water data show that dissolved Mg and othercations are removed from solution in the upper section ofthe sediment cores (0–4.92 m in core KL-13-6 and 0–9.10 m in core KL-29-2, see Table 2). The depth-integratedrates of reverse weathering derived through transport-reac-tion modeling suggest the following overall stoichiometry:

Biogenic opalþmetal hydroxidesþNaþ þ 0:1Kþ

þMg2þ þ Ca2þ þ 5:1HCO3�

! authigenic silicate phasesþ 5:1CO2 ð23Þ

The depletion of dissolved K+ in the upper section of thesediment cores is clearly seen in the data (Figs. 8 and 9)but is not well reproduced by the model. It is, thus, likelythat more dissolved K+ is removed through reverse weath-ering than indicated by the model. Dissolved Li+ is alsostrongly depleted in the upper core sections (Fig. 10) indi-cating that Li+ ions are taken-up in authigenic silicatephases formed during reverse weathering. Previous studiesof reverse weathering of opal-rich sediments (Michalopou-los and Aller, 1995; Michalopoulos et al., 2001; Michalop-oulos and Aller, 2004) also show that substantial amountsof potassium, magnesium, lithium, and sodium are boundin authigenic silicate phases formed within the upper fewmeter of the sediment column. Michalopoulos and Aller(2004) estimated uptake rates of 2.8 lmol K g�1 year�1

and �16 lmol SiO2 g�1 year�1 for Amazon delta sedi-ments. These rates are at least three orders of magnitudehigher than the maximum TA consumption rates observedin the upper sections of core KL-13-6 (0.7 nmol g�1 year�1,


Fig. 8) and core KL-29-2 (1.3 nmol g�1 year�1, Fig. 9).Sakhalin slope sediments are, thus, less reactive in termsof reverse weathering than Amazon delta sediments. Ratesof reverse weathering are probably limited by the deliveryof Al- and Fe-hydroxides. These essential reagents for re-verse weathering (Eq. (23)) are rapidly formed in the Ama-zon drainage area because of the intense tropicalweathering regime. The formation of metal hydroxides inthe Amur River basin is, however, restricted by low temper-atures and permafrost conditions. The harsh climatic condi-tions in the hinterland are, thus, the ultimate reason for thereduced rates of reverse weathering in Sakhalin slope sedi-ments. The depth-integrated rates of reverse weathering(4.22–15.0 mmol CO2 m

�2 year�1, Table 2) are one orderof magnitude smaller than the corresponding weatheringrates (Table 2). The net result of the various interactions be-tween pore fluids and silicate phases in Sakhalin slope sed-iments is, thus, CO2 consumption and the release ofdissolved Mg, Ca, and Na through marine silicateweathering.

5.7. Global rates of marine silicate weathering

Total alkalinity values >60 mM indicative for marine sil-icate weathering were previously observed at other produc-tive continental margins where organic-rich sedimentsaccumulate at high rates (Table 5). These high TA valuesoccur in methanogenic sediments, only, and are accompa-nied by elevated concentrations of dissolved ammoniumimplying high rates of organic matter degradation (Table

Table 5Pore water composition at productive continental margin sites

Leg/Site Study area Depth (mbsf) TA (mM)

112/683A Peru Margin 127.1 92.32112/685A Peru Margin 132.3 156.37112/688A Peru Margin 192.88 265.7117/723A Oman Margin 332.04 106.8128/798B Japan Sea 108.81 76.34141/861C Chile Margin 64.4 62.41164/994C Blake Ridge 351.98 104.25164/995A Blake Ridge 295 114.01164/997A Blake Ridge 315.35 126.26166/1004A Bahama Bank 92.72 73.41167/1014A California Margin 164.85 113.08167/1019C California Margin 108.15 103.3172/1054A Carolina Slope 195.7 67.77175/1078C Angola Margin 150.3 86.48175/1084A Namibia Margin 93.07 171.67182/1127B Australia Margin 133.06 105.1182/1131A Australia Margin 136.3 137201/1230A Peru Margin 132.15 157.17202/1234A Chile Margin 27.81 72.84204/1244B Cascadia Margin 46.78 68.23204/1245B Cascadia Margin 130.66 72.65204/1246B Cascadia Margin 128.15 83.14204/1251B Cascadia Margin 40.41 122.4204/1252A Cascadia Margin 45.75 108.24

The pore water data are taken from the ODP data base (Legs 100–2includes only those sites where elevated total alkalinity contents (TA > 6having the highest TA value.

5). The wide-spread occurrence of elevated TA values inmethanogenic sediments indicates that silicate weatheringoccurs at many sites around the world where rapid burialof organic matter fuels high rates of methanogenesis. HighTA values often occur at low and mid latitudes even thoughmost of the primary silicates should already be consumedon land by intense weathering before entering the ocean(Table 5). In these settings, other cation-rich silicates haveto account for the observed TA release. At active continen-tal margins (Peru, Chile, Cascadia, California and JapanSea) the marine weathering of volcanic ashes delivered bythe adjacent volcanic arc could be responsible for the ob-served TA enrichment. At other sites, the incongruent dis-solution and leaching of ubiquitous smectite maycontribute significantly to the observed cation and TA re-lease (Golubev et al., 2006). Dissolved Ca concentrationsin methanogenic pore waters are always smaller than theseawater value (�10 mM) indicating that seawater Ca andCa being released during silicate weathering are removedfrom solution via carbonate precipitation (Table 5). SinceCO2 should induce carbonate dissolution rather thanCaCO3 precipitation, the depleted Ca values suggest thatmost of the metabolic CO2 is neutralized during marine sil-icate weathering. Dissolved Mg is enriched beyond the sea-water value in a number of cores (Table 5). Mg depletionsare, however, more common and could either be explainedby dolomite formation in the deeper sections of the sedi-ment column or by Mg uptake in the underlying oceaniccrust. Dissolved ammonium and DOC are highly correlatedin anoxic marine sediments (Komada et al., 2004). The high

NH4 (mM) Mg (mM) Ca (mM) Cl (mM)

19.9 47.9 8.9 51231.8 68.25 5.5 53063.0 96.78 7.4 52240.4 19.01 2.7 5337.89 53.4 5.5 542

n.d. 53.88 2.8 56621.7 15.09 2.2 5.118.5 16.71 2.0 50518.7 16.75 2.5 50513.8 43.1 6.7 6.8n.d. 67.11 6.8 549n.d. 38.3 2.2 3696.05 36.36 4.2 555

14.4 39.53 4.3 54638.4 75.38 2.9 53718.5 29.8 3.7 108725.1 11.4 2.7 90637.4 n.d. n.d. 5509.9 63.9 1.7 563

12.7 49.8 2.9 54616.9 35.5 4.3 55122.1 43.8 5.0 55311.4 65.3 2.2 54312.3 n.d. n.d. 537

10, http://iodp.tamu.edu/janusweb/general/dbtable.cgi). The table0 mM) have been observed. Each site is represented by the sample

http://iodp.tamu.edu/janusweb/general/dbtable.cgi


ammonium values observed in the methanogenic sedimentsthus imply high DOC concentrations. These elevated DOCvalues could be the ultimate reason for the enhanced reac-tivity of silicate phases in methanogenic sediments sinceDOC is known to promote silicate weathering processes(Viers et al., 1997).

A linear correlation between ammonium and TA valuesfrom ODP cores reveals a slope of 3.0 ± 0.4 (Fig. 11). ThisTA/NH4 ratio is very close to the ratio of DIC and NH4 re-lease during methanogenesis (DIC/NH4 = 3.3):

ðCðH2OÞÞ106ðNH3Þ16ðH3PO4Þ! 53CH4 þ 38CO2 þ 16NH4þ þH2PO4� þ 15HCO3�

ð24Þ

The close match between the measured TA/NH4 values andthe DIC/NH4 ratios predicted by Redfield stoichiometry(Eq. (24)) suggests that most of the CO2 produced in meth-anogenic sediments is converted into TA. It should, how-ever, be considered that the DIC/NH4 ratio duringmethanogenesis is significantly enhanced by ammoniumadsorption on sediment surfaces. On the other hand, the de-pleted Ca and Mg values in methanogenic sediments (Table5) clearly show that most of the alkaline earth cations andsubstantial amounts of the alkalinity produced during sili-cate weathering are removed from solution via carbonateprecipitation. Carbonate precipitation, thus, induces a de-crease in the DIC/NH4 ratio that may effectively compen-sate for the DIC/NH4 increase caused by ammoniumadsorption. Overall, the available ODP data suggest thatmost of the CO2 produced in methanogenic sediments isconverted into carbonate alkalinity through silicate weath-ering processes not only at the Sakhalin slope but also atother productive continental margins.

The global rate of silicate weathering in anoxic sedi-ments can be derived from estimates of global sedimentarymethane production assuming that marine silicate weather-ing is mostly occurring in methanogenic sediments. Therates of methane formation and consumption in anoxic

Fig. 11. Concentrations of dissolved ammonium and total alka-linity (TA) measured in pore fluids retrieved during ODP drilling.The data plotted as solid squares are listed in Table 5. The solid linerepresents the data trend and is calculated by linear regression(TA = 47 ± 10 + 3.0 ± 0.4 NH4). The dotted line indicates the TAincrease in methanogenic sediments caused by the degradation ofN-bearing organic matter (Eq. (24)). The broken line shows thealkalinity increase that would result upon a complete conversion ofmetabolic CO2 into bicarbonate.

marine sediments have been estimated as 5 TmolCH4 year

�1 (Reeburgh et al., 1993) to 20 Tmol CH4 year�1

(Hinrichs and Boetius, 2002). CO2 production rates are ashigh as the corresponding methane values since CO2 andCH4 are produced at a one-to-one molar ratio during thebreak-down of organic matter in methanogenic sediments.Assuming that methanogenic CO2 is completely neutralizedthrough silicate weathering, the global rate of silicateweathering in methanogenic marine sediments results as5–20 Tmol CO2 year

�1. The global rate of continental sili-cate weathering is 11.7 Tmol CO2 year

�1 (Gaillardetet al., 1999). Thus, off-shore weathering of silicates in mar-ine sediments may consume as much CO2 as on-shoreweathering.

5.8. Effects of anoxic sediment diagenesis on marine and

global carbon cycling

Organic matter is microbially degraded via sulfatereduction and methanogenesis in anoxic sediments. Sulfideformed during sulfate reduction is either oxidized at thesediment surface or buried in sediments. Sulfate reductionleading to sulfide burial can be represented by the followingstoichiometry using Eqs. (12) and (14):

2CðH2OÞ þ SO42� þ 2=5Fe2O3! 2HCO3� þ 2=5FeS2 þ 1=5FeSþ 1=5FeOþH2O

ð25Þ

The organic matter (C(H2O)) being degraded via sulfatereduction was previously formed by photosynthesis:

CO2 þH2O! CðH2OÞ þO2 ð26Þ

Thus, sulfate reduction induces the transformation of CO2into HCO3

� if the sulfide formed during this process is bur-ied in sediments. CO2 rather than HCO3

� is, however,formed when sulfide is reoxidized by oxygen and other elec-tron acceptors. Sulfate reduction without sulfide burial has,thus, no significant effect on the global carbon cycle. Theglobal rate of sulfide burial at the modern seafloor has beenestimated to be 0.5–1.1 Tmol year�1 (Hansen and Wall-mann, 2003) and the stoichiometry of Eq. (25) implies that1.0–2.2 Tmol CO2 year

�1 are converted into HCO3�

through this process.CO2 formed during the break-down of organic matter in

methanogenic sediments is converted into carbonate alka-linity through silicate weathering (Section 5.7). About onethird of the CO2 is removed by authigenic carbonate forma-tion while the remaining two thirds are returned into theocean as dissolved bicarbonate. The overall reaction may,thus, be summarized as:

cation� rich silicatesþ 2CðH2OÞ þH2O! cation� depleted silicatesþ 2=3HCO�3þ cationsþ 1=3CaCO3 þ CH4 ð27Þ

The charge balance of this reaction is maintained by the re-lease of an equivalent amount of dissolved cations (Mg2+,Na+). Considering the global rate of methanogenesis (5–20 Tmol year�1), marine silicate weathering induces a CO2consumption of 5–20 Tmol year�1, an alkalinity flux into


the ocean of 3.3–13.3 Tmol year�1, and authigenic carbon-ate burial at a rate of 1.7–6.7 Tmol year�1.

Methane formed during organic matter degradation inanoxic sediments is almost completely reoxidized intoHCO�3 through AOM and one third of the DIC formedthrough AOM is precipitated as authigenic CaCO3 (Luffand Wallmann, 2003; Luff et al., 2004; Wallmann et al.,2006). The overall carbon turnover induced by AOM andcarbonate precipitation may, thus, be represented by thefollowing stoichiometry:

CH4 þ SO42� þ 1=3Ca2þ ! HS� þ 1=3HCO3�

þ 4=3H2Oþ 1=3CaCO3þ 1=3CO2 ð28Þ

Most of the sulfide being formed during AOM is reoxidizedto generate large quantities of acidity. HCO3

� formedthrough AOM is thus largely converted into CO2 beforeit can enter the ocean. A significant fraction of theHCO3

� is, however, permanently buried as authigenicCaCO3. Considering the global rate of methane productionand oxidation (5–20 Tmol year�1, Section 5.7) and the stoi-chiometry given in Eq. (28), it can be estimated that 1.7–6.7 Tmol CaCO3 year

�1 are buried in anoxic marine sedi-ments as a consequence of AOM.

CO2 bound in organic matter via photosynthesis is usu-ally returned into the ocean and atmosphere via aerobic res-piration processes. Anoxic sediments show, however, a verydifferent behavior since dissolved alkalinity and authigeniccarbonates rather than CO2 are formed in these systems(Aller et al., 1996). Anoxic diagenesis including sulfide bur-ial, AOM, and silicate weathering induces a global burialflux of authigenic carbonates on the order of 3.3–13.3 Tmol year�1 and an alkalinity flux into the ocean of4.3–15.5 Tmol year�1. These major carbon fluxes constitutesignificant CO2 sinks and alkalinity sources that have notbeen accounted for by previous mass balances and modelsof the global and marine carbon cycles.

6. CONCLUSIONS AND PERSPECTIVES

Rapidly accumulating sediments composed of biogenicopal, organic matter, plagioclase feldspars, olivine, pyrox-ene, volcanic glass, clay minerals, and other products of con-tinental weathering are deposited at the Sakhalin slope.Reverse weathering occurring in the upper section of thestudied sediment cores induces the fixation of dissolved cat-ions (Mg2+, Na+, K+, Li+) in authigenic clays and the con-sumption of carbonate alkalinity. Ammonium releasedduring the break-down of organic matter is adsorbed at sed-iment surfaces replacing adsorbed Ca2+ and Na+ ions thatare released into the pore water via cation exchange. Reactivesilicates (plagioclase feldspars, olivine, pyroxene, volcanicglass) are partly dissolved and transformed into cation-de-pleted silicate phases in the deeper methanogenic sedimentsections. The marine weathering of these silicates inducesthe release of Mg2+, Ca2+, and Na+ as well as the formationof authigenic carbonates. CO2 formed during carbonate pre-cipitation and the break-down of organic matter in methano-genic sediments is completely neutralized during silicate

weathering. Rates of marine silicate weathering are one orderof magnitude higher than the rates of reverse weathering andcation exchange in Sakhalin slope sediments and significantlyhigher than the rates of continental silicate weathering in thehinterland drained by the Amur River.

Pore water data obtained at other productive continen-tal margin sites demonstrate that silicate weathering is acommon process in methanogenic sediments with high met-abolic activity. Most of the CO2 being formed during theanaerobic break-down of organic matter is converted intoalkalinity and authigenic carbonates. Thus, silicate weath-ering in methanogenic marine sediments constitutes a sig-nificant sink for CO2 and a source for carbonatealkalinity that need to be included in future budgets ofthe global and marine carbon cycles. The global rate ofmarine silicate weathering estimated as 5–20 Tmol CO2 y-ear�1 may be as high as the global CO2 consumptionthrough continental weathering.

Marine sediments have previously been regarded asimportant CO2 sinks since large amounts of organic carbonformed though photosynthesis are ultimately buried at theseafloor. It has, however, been ignored that the degradationof organic matter in anoxic sediments works in a funda-mentally different way than aerobic degradation. While aer-obic respiration recycles CO2 into oceans and atmosphere,anaerobic degradation-coupled to marine silicate weather-ing-produces large amounts of carbonate alkalinity andauthigenic carbonates. Thus, CO2 fixed in organic matterduring primary production is not recycled but transformedinto other carbon compounds (HCO�3 , CaCO3) during an-oxic diagenesis. The global rate of carbon transformationin anoxic sediments is estimated here to be 7.6–28.8 TmolC year�1. This value is at least as high as the global rateof organic carbon burial in marine sediments (13 Tmol C y-ear�1; Hedges and Keil, 1995) and therefore this previouslyignored CO2 sink has to be considered in future budgets ofglobal carbon cycling.

The total alkalinity (TA) inventory of the oceans is be-lieved to be controlled by riverine HCO�3 inputs and throughcarbonate burial at the seafloor. The alkalinity budget of theHolocene ocean seems to be out of balance since TA removalthrough carbonate burial (48 Tmol year�1; Berner and Ber-ner, 1996) is larger than the riverine TA flux(38 Tmol year�1; Berner and Berner, 1996). Part of thisimbalance is caused by glacial to interglacial sea-levelchanges affecting carbonate burial on the continentalshelves. TA generated in anoxic marine sediments throughsilicate weathering and sulfur burial (4.3–15.5 Tmol year�1)has, however, never been considered and may in fact accountfor a major portion of the apparent TA imbalance.

Marginal seas are known to act as CO2 pumps taking upatmospheric CO2 and exporting carbonate alkalinity intothe open ocean (Tsunogai et al., 1999). This continentalshelf pump operating in the Sea of Okhotsk (Otsukiet al., 2003) and in other marginal seas (Thomas et al.,2004) may remove large quantities of anthropogenic CO2from the atmosphere. Anoxic sediments that are preferen-tially deposited at productive continental margins may con-tribute to the CO2 uptake through marine silicateweathering and sulfur burial.


The global rate of authigenic carbonate burial in anoxicmarine sediments estimated here as 3.3–13.3 TmolCaCO3 year

�1 approaches the modern rate of pelagic car-bonate burial (10 Tmol CaCO3 year

�1; Archer, 1996). Pro-ductive continental margins with high rates of terrigenoussedimentation are, thus, important and previously ignoredsites of carbonate accumulation. About 50% of the Ca bur-ied in authigenic carbonates originates from seawater suchthat the Ca balance of the ocean is also affected by anoxicdiagenesis.

In contrast to continental weathering, marine weather-ing is not directly coupled to average global surface tem-perature and atmospheric pCO2. Carbon transformationsin anoxic sediments are rather fueled by the depositionof particulate organic matter and reactive silicate phases.The consumption of CO2 in these sediments is, thus, con-trolled by continental erosion and marine productivity.The negative climate feedback established by the tempera-ture- and pCO2-dependent rate of continental weathering(Berner, 2004) is weakened by marine weathering pro-cesses since reactive silicate phases which are not con-sumed on land may be weathered in methanogenicsediments. Marine weathering might, thus, amplify climatechange on geological time scales and could, for example,contribute to the draw-down of atmospheric CO2 ob-served during the late Cenozoic and glacial periods ofthe Quaternary.

The results presented in this study have also majorimplications for the applied geosciences. Thus, naturalgas hydrates and gas reservoirs formed in marine sedi-ments contain large amounts of CH4 but usually muchless CO2 even though both gases are produced at equi-molar rates during the break-down of organic matter.

Table A1Parameter values applied in the modeling

Parameter (symbol)

Bottom water temperature (T0 in �C)Geothermal gradient (DT in �C/m)Pressure at the sediment water interface (P0 in bar)Pressure gradient (DP in bar/m)Sedimentation rate (wf in cm kyr

�1)Porosity at infinitive depth (Of)Density of dry solids (dS in g cm

�3)Initial age of POM (a0 in year)Inhibition constant for POC degradation (KC in mM)Inhibition constant for methane formation (KSO4 in mM)Kinetic constant for AOM (kAOM in cm

3 year�1 mmol�1)Equilibrium constant for ammonium adsorption (KADS in cm

3 g�1)Kinetic constant for ammonium adsorption (kADS in mM year

�1)Kinetic constant for sulfide removal (kSP in wt-S year

�1)Attenuation constant for sulfide removal (rSP in cm

�1)Monod constant for sulfide removal (KSP in mmol cm

�3)Kinetic constant for calcite precipitation (kCaCO3 in year

�1)Kinetic constant for Mg release (kMg in year

�1)Atomic C/N ratio during POM degradation (rN)Atomic Ca/Mg ratio during silicate weathering (rCa/Mg)Atomic Na/Mg ratio during silicate weathering (rNa/Mg)Atomic K/Mg ratio during silicate weathering (rK/Mg)Atomic Si/Mg ratio during silicate weathering (rSi/Mg)Molar TA/Mg ratio during silicate weathering (rTA/Mg)

Low CO2 contents in natural gas and hydrates may becaused by marine weathering processes. The CO2 con-sumption via marine silicate weathering should, thus,be considered in basin modeling to better predict theCO2 contents of economically important natural gasand hydrate reservoirs. Moreover, marine sediments areincreasingly used for the disposal of CO2 separated fromnatural gas and in coal power plants (House et al.,2006). The results presented in this study imply that ter-rigenous sediments with high contents of reactive silicatephases might be well suited sites for CO2 disposal sinceCO2 may be rapidly neutralized by marine silicateweathering.

ACKNOWLEDGMENTS

The captains and crew members of the research vessel Sonneprovided helpful assistance at sea, their work is greatly appreciated.Special thanks go to Bettina Domeyer, Anke Bleyer, and ReginaSurberg for having carried out the chemical analyses. The helpfulcomments of two anonymous reviewers and the associated editorR.C. Aller are greatly appreciated. This work was funded by theBMBF within the framework of the German-Russian KOMEXproject (Grant 03G0534A).

APPENDIX A

Model parameters and equations applied in the simula-tions are summarized in Tables A1–A4. Since the modelcolumns extend over more than 20 m, changes in tempera-ture and pressure were considered in the thermodynamicequilibrium calculations (Zeebe and Wolf-Gladrow, 2001)and the calculation of molecular diffusion coefficients

KL-13-6 KL-29-2

2.0 2.00.035 0.03571.3 77.10.1 0.193 1150.64 0.762.5 2.5500 50035 351 110 101.7 1.00.1 0.114 � 10�5 17 � 10�50.0025 0.00251 � 10�7 1 � 10�70.05 0.050.05 0.0516/106 16/1061 10.1 20.1 0.10.005 0.024.2 6.1

Table A2Depth-dependent constitutive equations used in the modeling

Parameter Constitutive equation

Porosity U ¼ p1 � e�p2�x þ p3�Uf1þe

x�p4p5

þ Uf

Temperature T = T0 + DT�xPressure P = P0 + DP�xMolecular diffusion in sediments DS ¼ DM1�2�lnðUÞBurial of solids w ¼ wf �ð1�Uf Þ1�UBurial of pore water v ¼ Uf �wfUSediment age ax ¼

Pxi¼1

DxðiÞw ¼

Pxi¼1

DxðiÞ�ð1�Uði�1=2ÞÞwf �ð1�Uf Þ

Factor converting G (wt%) in C (mmol cm�3) rðiÞ ¼ 10�dS �ð1�UÞMWðiÞ�U

Table A3Rate laws used in the modeling

Rate Kinetic rate law

POC degradation RPOC ¼ KCCðDICÞþCðCH4ÞþKC � 0:16 � ða0 þ axÞ�0:95 � GðPOCÞ

PON degradation RPON ¼ 1412 � 16106 � RPOCSulfate reduction RSR ¼ 0:5 � CðSO4ÞCðSO4ÞþKSO4 � rðPOCÞ � RPOCMethanogenesis RM ¼ 0:5 �

KSO4CðSO4ÞþKSO4

� rðPOCÞ � RPOCAmmonium production RNH4 ¼ rðPONÞ � RPONAnaerobic oxidation of methane RAOM = kAOM � C(SO4)�C(CH4)Sulfide precipitation RSP ¼ CðTH2SÞCðTH2SÞþKSP � kSP � e

�rSP �x

Ammonium adsorption RADS ¼ kADS � 1� ADSCðNH4Þ�KADS� �

CaCO3 precipitation RCaCO3 ¼ kCaCO3 � ðCðCaÞ � CðCaÞOBSÞMg release RMg = kMg�(C(Mg)OBS � C(Mg))Silicate weathering RWE = rTA/Mg�RMG

Table A4Rate expressions applied in the differential equations

Species Rates

Sulfate (SO4) R(SO4) = �RSR � RAOMMethane (CH4) R(CH4) = +RM � RAOMAmmonium (NH4) RðNH4Þ ¼ þRNH4 � RAD

silicate weathering in anoxic marine sediments · silicate weathering in methanogenic sediments...

Documents