The 13C anomaly in the northeastern Atlantic - OceanRep - Geomar

GLOBAL BIOGEOCHEMICAL

CYCLES, VOL. 12, NO. 3, PAGES 467-477, SEPTEMBER 1998

The •13C anomaly in the northeasternAtlantic Robin Keir, Gregor Rehder,and Erwin Suess •J,•,,;•,ooe,,sch,,fLen Geomar,•.................... .... h.,,,g..... r"'• f•r Marine c, .... •.... ,

Helmut

der rß •' m • T .... • Kiel, ,•hr,,sdan-A.,,rechts-,•mvers•ta,,

Germany

Erlenkeuser

Leibniz-Laborfor Altersbestimmung und Isotopenforschung der Chfistian-Albrechts-Universitiit, Kiel, Germany

Abstract.The •5•3C of dissolved inorganic carbonwasmeasured onsamples collected at 49øNin thenortheast Atlanticin January 1994. Deeperthan2000m, •5•3C exhibits thesamenegative correlationversusdissolvedphosphate thatis observedelsewherein thedeepAtlantic. Upward

from2000m to about600m, •5•3C shiftsto valuesmorenegative thanexpected fromthe correlationwith nutrientsat depth,whichis likely dueto penetrationof anthropogenic CO2. From

thesedata,theprofileof theanthropogenic •5•3C decrease iscalculated byusingeitherdissolved phosphate or apparent oxygenutilization asa proxyforthepreanthropogenic •5•3C distribution. The shapeof the anthropogenic anomalyprofile derivedfrom phosphateis similarto that of the increasein dissolvedinorganiccarbonderivedby othersin the samearea. The reconstruction from

oxygenutilization results in a lowerestimate of theanthropogenic •5•3C decrease in theupperwater column,and the verticalanomalyprofile is lesssimilarto that of the dissolvedinorganiccarbon

increase. A •3Cbudgetfortheatmosphere, ocean,andterrestrial biosphere indicates thatwithinthe rangeof probable oceanCO2uptake theratioof •5•3C to inorganic carbon change should bemostly influenced by the•3Cinventory change of thebiosphere. However,theuncertainty in theratiowe derivepreventsa strongcontrainton the sizeof the exchangeable biosphere.

1. Introduction

carbon increase(ACt) in the globaloceanmight be obtained from a limited setof locationsif it is possibleto reconstruct

The •-•C/•2C ratiosof atmospheric CO2 and dissolved the cumulative•5•-•Cand inorganiccarbonconcentration inorganiccarbonin the upperoceanhave been decreasing becauseanthropogenic carbondioxideproducedby fossil fuel burninganddeforestation is enrichedin •2C. The cumulative decrease observedin the atmospheric •5•-•C(about-1.3%o [Friedliet al., 1986; Keelinget al., 1989]) is considerably less than

would have

occurred if

the

27%

increase

changes.As discussed in section4, the ratioof thesechanges should provide an indication of the carbon mass of the

terrestrialbiospherethat exchanges readily with atmospheric CO2. The problem is to determinethe relationship with sufficientprecision. In the first part of this paper,we focus on the area of the northeastern Atlantic

in

at 49øN near the

atmosphericCO2over the last 200 yearswerecausedsolely by additionof this amountof anthropogenicCO2to a closed atmospheric gasvolume(about-4%0).The discrepancyis due to the rapidexchangeof atmosphericCO2 with the oceanand

Europeancontinental margin,andwe beginby reconstructing

decreasein the sea,Heimannand Maier-Reimer [ 1996] reason that the spatial distribution of these anomaliesin the ocean

must be similar, since they have almost the same shape functionin their atmospherichistories. This propositionis supportedby experimentswith their ocean carbon cycle model,which is forcedby the historiesof atmosphericCO2

contemporaryrate of anthropogenic CO2 uptakeinto the sea, either from observedtemporalchangesin the oceanic•5•-•C [Quay et al., 1992; Heimann and Maier-Reimer, 1996; Bacastowet al., 1996] or from the isotopicdisequilibrium betweenthe air and sea surface[Tanset al., 1993]. In their evaluationof the first of thesemethods,BroeckerandPeng

and•5•3C. Thetemporal changes in theverticalinventoryof dissolved inorganiccarbonandof •5•3Cvarywidelyoverthe

in the atmosphere,terrestrial biosphere, and ocean to

the vertical distribution of the decrease in •5•-•Cthat has

occurredthere. These results are then comparedto the

increasein dissolvedinorganic carbon estimatedby terrestrialbiosphere,which dilutesthe atmospheric•51-•C KOrtzingeret al. [1998] in the sameregion. Various•'•Cflux balanceshave beenusedto estimatethe decrease.In regardto anthropogenic CO2 uptakeand•5•-•C

surfaceof the ocean,but theratio of theseinventorychanges at any particularlocation appearsto be relatively constant [Heimannand Maier-Reimer,1996]. This resultsuggests that

[1993]employed anoverallbudgetof •-•Cinventorychanges investigatetwo effectson the average•5•-•C changein the system:(1) incremental additionof anthropogenic CO2 to the oceanand (2) the massesof exchangeable carbonin the ocean

the ratio of •13C decrease (Ab•-•C)to dissolvedinorganic andbiosphere. We use a similar procedureto examinethe effectof thesetwo factorson the A•5•C/ACrratio in the ocean. One might expect that the decreasein the •5•C of dissolvedinorganiccarbonin the oceanshouldbe greater than wouldbe produced by the net uptakeof anthropogenic

Copyfight 1998by theAmerican Geophysical Union. Papernumber98GB02054. 0886-6236/98/98GB-02054512.00

CO2 directly into a closed seawatervolume, since the air-sea 467

468

KEIR ET AL.: •13CANOMALY IN THE NORTHEASTERN ATLANTIC

CO2exchange that dilutesthe changein atmospheric 8•3C

Meteor cruiseM27/1 (Figure 1). This area is marked by a westwardmorphological protrusion of the margin known as the GobanSpur. The sampleswere poisonedwith HgC12,and

should enhancethe ocean •13C decrease. However, this dependsupon how much of the atmospheric dilution can be accountedfor by exchangewith the terrestrial biosphere. As discussedin section 4.2, it appearsthat if the exchangeable terrestrialcarbonmasswere large enough (about 2000 Gt C),

analysis of the•13Cof thetotaldissolved CO2wascarriedout within the following year at the Leibniz Laboratory of the University of Kiel. From thesedata (Table 1), we reconstruct

theratioof A8•3Cto ACr in the oceanwouldapproach the

the verticaldistribution of the 8•3Cdecrease usingeither

closed system value and would not dependon the amount of anthropogenicCO2 that has been taken up. This is likely not the case, since this would imply that virtually all of the terrestrial biosphere is exchanging with the atmosphere.

dissolvedphosphateor apparentoxygen utilization (AOU)as

proxiesfor the preanthropogenic •3C distribution. The

Thusonewouldexpectthattheratioof A•3C to AC, should

dissolved phosphate reportedhere and usedfor this purpose was analyzed on board cruise M27/1 by the University of Hamburg. For AOU, we havemadeuseof the datacollectedo n

be greater than that for uptake of anthropogenicCO2 in a

Transient Tracers in the Ocean (TTO) stations 116 and 1 17

closed ocean system.

[Scripps Institution of Oceanography, 1986] that are very close to our OMEX stations. Dissolved oxygen measurements made on M27/1 generally have about the same averagevalue as the TIO measurementsfor any given depth,

2. Methods

but

As part of the Ocean Margin Exchange Experiment (OMEX), samplesfor carbon isotope analysis were collected during January 1994 from a series of hydrocast stations acrossthe Europeancontinentalmargin at about49øN during

15øW

the

former

exhibit

considerable

scatter

and

therefore

cannotbe used for the reconstruction.Later in the same year, K6rtzinger et al. [1998] measureddissolvedinorganic carbon on an east-westsection (Meteor M30/2) that passedclose to

10øW

5øW

I

....... •:••...••• ..•.••••. -..-,----•,...-•-•.,,,,......,• •j•?• ................ :....••••.:•i•, .•..,•.••, •, (-

•

• ':,,,•;•

•

•

-•

•..••...................... •••••••,•.,•-•.•... .... •.• .

•

50øN TTO

117

ß

TO 11•' 15øW

. .o

• •• s, 10øW

50øN

o

.........................

5ow

Figure 1. Location of OceanMargin ExchangeExperiment(OMEX I) hydrocaststations, taken duringF/S Meteor cruise27/1, January1994, andTransientTracersin the Ocean(TTO)stations 116 and 117, taken in June

1981.

KEIR ET AL.: •5•3C ANOMALY IN 'H-IENORTHEASTERN ATLANTIC

Table 1. Meteor 27/1 Phosphate and8•3C

Table 1. (continued)

Depth,m

Depth, m

PO4,gmolkg4

fi•3C,%0

Station B, Jan. 3, 1994, 49ø42.07V, 9ø40.0'W 9

0.48

19

n•

28 43 74 117 159 180

0.47 0.47 0.47 0.48 0.48 0.48

207

0.47 Station C, Jan. 4, 1994, 49•8.4'N,

698 801

0.96

901 1001

0.73

1.18 1.19

0.76 0.84

1500 1600

(0.69)

1690

1.17 1.16 1.15 1.13 1.16 1.26 1.35 1.41

0.84 0.90 0.83 1.08 1.00 0.98 0.97 0.93

1800

0.91

2105

52

(0.66)

1.13

101 152 202 301 400 451 501 553 630

0.49 0.49 0.49 0.49 0.72 0.76 0.81 0.83 0.86

1.15 1.14 0.92 0.95 0.83 0.94 0.88 0.81 0.88

2404 2704 3000

SmtionG, Jan. 14, 1994,49ø01'N, 13ø46'W

Station D, Jan. 5, 1994, 49ø11.0'N, 12ø48.2'W 1.07 1.02 1.01 0.76 0.84 0.83 0.85 0.83 0.88 0.88

0.53

1.04

53

0.53

1.00

103

0.54

155

0.54

1.08 1.16

306

908

0.74 0.81 0.94 1.08 1.06

0.97 0.91 0.88 0.77 0.75

1108

1.12

0.68

1310 1509 1916

1.15

0.80

2520

1.12 1.13 1.16 1.26 1.26

0.73 0.87 1.00 0.90 1.04

2725

1.31

0.95

3029

1.38 1.40 1.45 1.45 1.47 1.47

2117 2317

0.99 1.07 1.06 1.05 0.88 0.81 0.69

the OMEX area,andfrom thesedata,ACt wasestimatedby

13

459 609 755

Station F, Jan. 7, 1994, 49ø03'N, 13•5'W 0.53 0.49 0.51 0.49 0.70 0.81 0.91

0.67 0.70 0.67 0.84

(0.88)

0.51

10 50 1O0 198 298 500 600

0.98 1.12 !.!4 1.17

1106 1200 1375

11ø12.2'W

0.44 0.43 0.45 0.72 0.87 0.95 0.97 1.04 1.05 1.07

fi•3C,%0

1.O7 0.95 1.06 1.00 1.02 1.02

10

9 103 200 499 705 903 990 1206 1308 1433

PO4,gmolkg4

469

3336 3641

3946 4240 4516

1.04 0.98

1.01 0.91 0.98 0.86

back calculationto the pre-formedconcentration of dissolved inorganiccarbon,basedon the methodsof Brewer [1978] and

below 1700 m, the profiles of phosphateandAOU appearto be similar,but a pronouncedmaximumin AOU at 900 m has no correspondencein the phosphate distribution, which is

Chen and Millero [1979].

approximately

constant from 900 m to 1800 m.

This

appearsto be dueto a transition in the preformedphosphate

3. Reconstruction of the Preanthropogenic b13C Distribution

The basis for estimatingthe preanthropogenic•5•3C distributionaccordingto eitherdissolvedphosphateor AOU will be discussed, in turn,separately. Although•5•3Ctendsto be correlatedwith both of these properties, differing reconstructions result.

This can be foreseen from the vertical

concentration froma valueof 0.5 gmolkg'• in the upper900 m to a valueof 0.8 gmolkg4 at 1700m (Figure3). 3.1.

Phosphate-Based

Reconstruction

From Geochemical Ocean Sections Study (GEOSECS) measurementsin 1972 [Kroopnick, 1985], it appears that

•5•3Canddissolved phosphate mayhaveexhibiteda single

distributionsof thesethree properties,which are shown in Figure2 accordingto the scalingusedin the reconstructions.

linear correlation in a large volume of the Atlantic Ocean which includesall deepwaters and those intermediate-depth waters north of about 35øS. Figure 4 shows Atlantic

Theprofile of 5•3Cappearsmoreuniformthan that of the

GEOSECS data in (1) waters north of 35øS with densities

othertwo properties,and the averagevalueof about 1.0%oin

corresponding to 26.8 < c•0< 27.7 and (2) all waterswith c•0>

theupper300m is low compared to previouslyreported •5•3C 27.7. The data(listedby Ostlundet al. [1987]) werenot for surfacewaters[Kroopnick, 1985]. In the water column

correctedas recommendedby Kroopnick [1985] becausethe

470

KEIR ET AL.: õ•3CANOMALY IN 7I-IE NORTHEASTERN ATLANTIC

AOU, gmolkg'1

õ]3C,%o 80 60 40 20 0I -20 -40 I I I I I I 0,6

-500

0,8 .-1.0__ 1,2

+l•m

Im,•1!1# e •i -•--

in PO4 [Broeckerand Maier-Reirner, 1992]. As shown in

expected from biological fractionation alone, apparently

{

ß

closedvolumeof water,the õ•3Cof the remainingdissolved carbon should increase by about1.1%o pergmolkg'• decrease

Figure4, theslopeof õ•3Cversus PO4of theintermediate and deepwaters(PO4> 0.7 mmol kg'•) is less negativethan

©i•mm m•

+aT

-1000

I

If dissolved inorganic carbon and nutrients are biologically fixed into biomass and then removed from a

because a light•3C/•2C signalhasentered the northernsurface

%_.

em

sourcesof these waters. Part of this signal may originate from anthropogenicCO2 that has already penetratedthese watersbut hasnot yet appreciablyaffectedolder Circumpolar

+m y

-1500

Deep.Waterin thesouth.In addition, therealsoappears to be a naturaltransferof •3C-deficient CO2 into northernAtlantic

{

ß

-2000 6•3C

surfacewatersby two processes:(1) one-way transferof CO2 through the atmosphere [Keir, 1993] and (2) poleward

-2500

transport of surface waterwithlow õ•3C[Lynch-Stieglitz et +

-3000

ß

al., 1995]. The one-waytransferof CO2from seato air carriesa lower

m m

õ•3Cthanin the total inorganiccarbonof the surface water, -3500

ß

but the isotopicfractionationof the returnflux of CO2 to the seais small [Siegenthalerand Mannich, 1981]. As a result,

m m m

-4000

theatmospheric õ•3Cis about9%0lowerthantheaverage õ•3C

m

ß

of surfacewaters. A region where the sea surfacepCO2 is undersaturated and representsa natural sink for atmospheric

•m

-4500

CO2, suchas the northernAtlanticappearsto be [Takahashiet

al., 1993],shouldreceivea one-way CO2influxwitha õ•3C -5000

•

1.6

I

1.2

i

•

0.8

i

I

0.4

significantlylessthan that of seasurfacedissolvedinorganic carbon.

PO4, lamolkg-1

Figure 2. Verticalprofilesof õ•3Canddissolved phosphate at the OMEX stationsand apparentoxygen utilization (AOU) from the TID stations. Different symbols distinguish the individual

stations.

The natureof the upperoceancirculation in the Atlantic

may also contributeto the loweringof preformed õ•3C. Thermodynamically, the carbon isotopic fractionation between the sea and air increases as surface temperature decreases.For this reason,balancedair-seaexchangesof CO2

tendto reducethe6•3Cof warmsurface waters,andsubtropical correction factors (which are based on the deviation from a

global regression against oxygen utilization) for the northern Atlantic stations are systematically positive, and this may be dueto real anthropogeniceffects. The density layer, 26.8 < cy0 < 27.7, lies at an intermediate-depth rangeof about800- 1800 m in most of the Atlantic, except whereit outcrops in high latitudes [Zahn and Keir, 1994]. In the North Atlantic in winter, it appearsthat this density range surfacesover an annulararea stretchingfrom the LabradorSea eastwardover the EuropeanBasin [McCartney and Talley, 1982]. At phosphateconcentrationsgreaterthan about 0.7

gmolkg'•, thetrendof 6•3Cversus PO4in bothdensityranges coincides,and the covariationappearsto be linear (Figure4). Surface phosphateconcentrationsin winter vary between0.9

and 0.65 gmol kg'• from 63øN to 47øN along 20øW [Takahashiet al., 1993], and the concentrationof 0.7 gmol

kg'• is approximatelythat of the averagepreformed

1,6

1,41,2•

o

of 6•3Cversus PO•in intermediate anddeepwatersis mostly influenced by mixing between end-membercharacteristics foundin the northern winter surfaceand at depthin southern waters of the Atlantic

Ocean.

/

ß.½97/

1,0-

1700

..' a • _

E

=0,80

a. 0,60,4-

;#

i •t--26.8

I' vrø •6 I

0,2I •

[BroeckerandPeng,1982]. ThustheGEOSECSdatain Figure 4 imply that northof 35øS in the Atlantic the naturalpattern

4800

..•]

U,U -40

-20

0

20

40

60

80

100

AOU, I.tmolkg Figure 3. Phosphateversus apparentoxygen utilization at TID stations 116 and 117. Here the phosphate concentrations are thoseanalyzedon the TI'O expedition.

KEIR ET AL.: •5a3C ANOMALY IN '[I-IE NORTHEASTERN ATLANTIC

2.5 2.0

shown in Figure 5, we define the preanthropogenictrendby assuming(1) that anthropogenicCO2has negligibly affected

squares:oe > 27.7

016 01•

circles:26.8 < oe < 27.7

the •5•3Cof the Circumpolar DeepWaterthat entersthe Atlanticthrough theVemaPassage and(2) thatthe•5•3C of the

(northof 35øS)

50e• 7 012 "•O&. 42'• "•'00•. , o

winter mixed layer at the Goban Spurhad decreased by 0.8%0 at the

!:5

o

1.0-

o B

0.5-

time

0

[]

radiocarbon

0.0[]

0.5

1.0

1.5

2.0

the

measurements

in

1994.

The

first

found

in

WSBW

indicate

that

even

this

component is poorly ventilated relative to carbon isotopes during its formation [Weisset al., 1979]. In regardto the secondassumption,0.8%0is the average

-0.5 0.0

of

assumptionwould seem reasonablesince CircumpolarDeep Water containsa large componentof old deepwaterreturning from the Pacific, a small component of not-so-recentlyformed North Atlantic Deep Water (NADW), and a small quantity of more recently formed Weddell Sea Bottom Water (WSBW) [Lynch-Stieglitz et al., 1995]. The low values of

o

B

0

471

2.5

3.0

PO4,[zmolkg

•5•3Cdecrease recorded in demosponges collectedfrom the Caribbeanand Coral seas(0.9 and 0.7%0,respectively [BOhrn et al., 1996]). These recordsspanthe last two centuriesup to

the endof 1992, andwe assume that •5•3Cdecreased by this Figure 4. Atlantic1972 Geochemical OceanSectionsStudy

(GEOSECS) •5•3C versus PO4withinthedensityrangeof 26.8 < c•e< 27.7 north of 35øS(circles)andall waterswith c•e> 27.7 (squares).Numbersadjacentto thesamples with low PO4 concentrationindicate depth in meters below surface.Lines

amount in surface waters over the Goban Spur as well.

Althoughthetemporal decrease of surface ocean•5•3C dueto anthropogenicCO2is not expectedto be uniform spatially, the decreasesimulatedby the Hamburg model at the Goban

decreases at thelocations wherethe indicatethe linearregression of the deepwater•5•3Cversus Spuris similarto the•5•3C PO4 (squares)and the slope expected from biological fractionationin a closedsystem.

sponges were collected [Bacastowet al., 1996]. We refer specifically to their Figure 9a, which showscontoursof the

rate of changein 8•3Cduringthe period1983-1995. The Caribbean and the northeast Europeanmargin both exhibit

gyre watershavelower•5•3Cthanexpected fromtheir low

ratesof-0.018%oyear'l, while the Coral Sea has about-

nutrient values [Lynch-Stieglitz et al., 1995]. Transport of warm waters northward followed by cooling feeds the

\

formation of subsurfacewaters in the North Atlantic, and if

\ \ \

the warmto cold waterconversion rate is morerapid than the isotopic reequilibration of surface waters (timescale -10 years)dueto the cooling, this transportmay also contribute

\

2.0

\ \

to a natural reduction in thepreformed 15•3C of subpolar mode anddeepwaters. We note parentheticallythat in Figure 4 the phosphate

1.5o

concentrations lowerthan0.7 gmolkg'• werefoundduring summerwithin the surfacemixed layer in the high northern

/

•

NADW

'•.

1.0-

latitudes, andthegreaternegative slopeof •5•3Cagainstthese lower PO4 valuesis similar to that expectedfrom biological fractionation alone. This could occur becauseof rapid removal of carbonandphosphateduringthe spring bloom, starting from prebloom phosphateconcentrationsof about

0.8 gmolkg'•. Althoughthe entire watercolumnnearthe GobanSpuris

denser thanc• = 26.8 in winter,the 8•3Cvalueswemeasured do not showa linear relationto phosphate(Figure5). The

$•3C-PO4 trendin thedeeper partof thewatercolumnfollows the mixingline betweenNorthAtlanticandCircumpolar Deep Waterobservedelsewherein the Atlantic [Lynch-Stieglitzet

Gob• S••

0.5-

•

•

-•'eo•,

1994'

"e•4• • CPDW

0.0 0,0

0.5

1.0

1.5

2.0

2.5

PO4, !mol kg Figure 5. •5•3Cversus PO4 measured fromOMEXhydrocast

correlatedwith phosphatethroughoutthe water column in winterin the northeastern Atlantic. This is implied by the

stationsshownin Figure 1. Startingfrom the compositionof CircumpolarDeepWater (CPDW), lines show mixing trend with North Atlantic Deep Water (NADW) [Lynch-Sieglitz et al., 1995], effect of closed system biological fixation (dashed),and our assumedpreanthropogenic trendin the study area. The preanthropogenic trend has a lower negative slope than producedby biological fixation, indicating that the preformedNADW is a naturalsink for atmosphericCO2 (see

GEOSECS observations

text). Equation for thislineis •3C(pre)= 2.20- 0.803[PO4].

al., 1995], but $•3Cappearsto shift to noticeablylower valuesat phosphateconcentrations less than about1.2 •mol

kg4. In orderto estimatethe decrease in $•'•Covertime, we assumehere that the preanthropogenic $•3C waslinearly

in the western North Atlantic.

As

472

KEIR ET AL.: •3C ANOMALY IN 'ITIE NORTHEASTERN ATLANTIC

0.020%0year'l. Otherareashavequitedifferentrates,aslow as -0.004%0year'• nearAntarctica, for example. The 5•3C of newly formedNADW is expectedto be slightly greaterin the preindustrialera than in 1972 (at the time of GEOSECS) but not ashigh as wouldbe predictedfrom biologicaluptakefrom the compositionof CircumpolarDeep.

Water. Ourestimated preanthropogenic •513C-PO4 trendfalls inbetween these two limiting extremes (Figure 5).

The

predicted preanthropogenic •3C shownin Figure5 is on average 0.09%0greaterthanthe five measured •3C values below3300 m depth(PO4-- 1.45 gmolkg'l). Sincethese samplesshouldnot have been influencedby anthropogenic CO2, the assumed linear pre-anthropogenic correlation

between•13CandPO4appearsto slightly overestimate the

The difference between our measured 5•3C and that

predictedfrom the phosphateconcentration (A8•3C)of correspondingsamplesis plotted versusdepthin Figure 6. The distribution indicates that the carbon isotope anomaly haspenetrated ratherdeepin the watercolumnat 49øN in the eastern Atlantic. This is to be expected from the active waters in

the

North

Atlantic

that of tritium, but the isotope signal appearsto penetrate deeperthanthe bomb-produced tracer. This is to be expected becauseanthropogenicCO: hasbeen addedto the oceanover a longer time period than tritium. The shape of the vertical profile of the phosphate-

calculated A5•3Canomalypatternis similarto theprofileof excessCO2 (Figure 6) calculatedfrom total dissolvedCO: measurements during Meteor cruise30/2 in the samegeneral

thewatercolumninventory, we averaged theA5•3Cvaluesin

3.2. •3C Anomaly (PO4-Based) versus Depth

of subsurface

theEuropean margin[Ostlund andGrail,1987]. As shownin Figure6, theshape of theprofileof A5•3Cis fairlysimilarto

area during October 1994 [KOrtzingeret al., 1998]. To obtain

temporalchangein the deepestpart of the water column.

formation

the core of the cyclonic circulationof LabradorSea Water eastwardacrossthe Atlantic [Talley and McCartney, 1982]. Tritium measurements duringTID show that these transient tracershavebegunto appearat the 2000 m level adjacentto

in

general. The 1800 m depthat the Goban Spur correspondst o

the upper300 m (the depth of the mixed layer at the time of collection) and in 200 m intervals below the mixed layer.

The AS•3C/ACrratio obtainedfrom the depth-integrated inventories of both quantities is -0.016%o(!.tmolC kg'•)'•. This result is not purely observational, as it rests on the

assumption that the 5•3C of GobanSpur surfacewaters decreasedby 0.8%0.The uncertaintyin this ratio appearsto be

+0.005%0(gmol C kg'•)'•.

This is basedon a +0.1%o

uncertaintyboth in the measurements within the upper 300 m

andin theproxyrecords, anda _+10gmolkg'• uncertainty in the ACt reconstruction.

a5•3C(points)andaCr (line) 0 I

0

I

I

-0.2

25 I I

-0.4

I

I

I

-0.6

TheA5•3C/ACr ratio we obtaincanbe compared to the plotof vertically integrated ratesof change in •3C/•:Cversus

50 I.tmolkg" I

-0.8

I

dissolvedinorganic carbon simulatedby the Hamburgocean general circulation model [Helmann and Maier-Reimer, 1996]. The model uses prescribed histories of the

%o

-500

atmospheric CO: andits •3C togetherwith one of two

-1000

alternate formulations of the gas exchange over the ocean surface. In both cases, the plot shows generally a good correlation between the two inventory changes, the slope

-1500

-

ø•.•a

_

I1

/-o• []

,

o o

o/,ø.o

-2000

being equivalentto AS•3C/ACr. A close inspectionof 3H

.-

Heimann and Maier-Reimer's Figure 6 indicates that the relationshipbetweenthesetwo inventorychangestendsto be split into two trendswith slopesof about-0.019%o(gmol C

kg'•)'• and-0.016%o (gmolC kg'•)'•, the lattercorresponding

-2500

to the value we derivefrom the phosphatereconstruction. In the model, all of the points that come from areas of greater accumulationof anthropogenic CO:, which shouldincludethe northeastenAtlantic, fall on the line with the lower negative slope.

-3000

&

-3500

I -4000

-4500

-5000

[]

I

3.3.

Calculation

of A5•3C

As an alternative to the method above, we apply a

[]

procedure for estimating A5•3CfromAOUthatis essentially I I I I I I I I I I

0.0

1.0

2.0

3.0

4.0

5.0

TU

Figure 6.

AOU-Based

Phosphate-based A5•3Cversusdepthin the

OMEX area (symbols left side). The solid line indicates the averageACr in the easternAtlantic (47ø - 49øN) accordingto KSrtzingeret al. [in press]. The axesare scaledto the ratio of

theirwatercolumninventories (-0.016%o per pmolC kg'•).

equivalentto that of Kroopnick[1985]. Our calculationdoes not include corrections for changesin alkalinity and preformed inorganiccarbonconcentration,but these corrections .

ß 11C

•uua•

•

slope(-0.0074%o per pmolO: kg'•) closeto that expected from Redfield-ratio biological recycling in a closed ocean

system.Oneassumes thatthe•3C of a waterparcelat depth haschanged fromthepreformed 5•3Cin thisproportion to the

Squaresymbolsat right show the vertical profile of tritium at

apparent oxygen utilization.

TFO Station117 [Ostlundand Grail, 1987].

distribution of thecalculated preformed •3C arethenascribed

Changes in the spatial

KEIR ET AL.' •5•3CANOMALY IN THE NORTHEASTERN ATLANTIC

to the effects of anthropogenic CO2.

Thus the spatial

The core of outflow

water from the Mediterranean

473 Sea is

distribution of preanthropogenic, preformed •5•3Cis assumed found at 900 m depth at the Goban Spur, and one could ask to be constant

for

the

water

volume

under

consideration.

Kroopnick applied this procedureto individual water masses that are formed in the same region. Here this assumption appliesto different water massesfound in the water column.

Usingthe global•5•3CversusAOUslope,the •5•3Cvalues observedin our deepestsamples are predictedclosely by a

•5•3C intercept of 1.60%o at oxygensaturation.Thisvalueis slightly greater than the value of 1.47%o found in the regressionof all deep water data [Kroopnick, 1985]. We

assume herethat the •5•3Canomalyin the watercolumnis given by the deviation of the measuredvalue from that predictedby 1.6 - 0.0074 x AOU. Here we haveusedthe AOU from TTO stations116 and 117 below 300 m (Figure3) and an assumedAOU = 0 in the upper300 m correspondingto the winter mixed layer.

whetherthe AOU maximumat this level is partly dueto slow ventilation. However, from the tritium distribution at TrO 117, this would not seem to be the case, as its concentration

at 900 m is greaterthan that foundat the 1700 m level (Figure

6), wherethe AOU-based A•5•3Cgoes througha relative maximum. Since the isotope anomaly is being introducedto the ocean over a longer time scale than tritium, it would not seem possible to simultaneously producethe tritium and

AOU-based A•5•3C profiles. The PO4-based A•5•3Cappears more similar to both the tritium and ACt profiles, and we believe that this estimateof the penetration of the isotope anomaly is more realistic.

In the case of the AOU-derived

A•5•3C, its verticalinventory relativeto thatof ACt is about -0.010%o(gmol C kg'•)'•, about60% of the PO4-derived estimate. As discussedin section 4.2, lower values of this

Theprofilesof A•13Cobtainedby the AOUmethodare

ratio imply greater exchange capacity of the terrestrial

comparedto the averageprofile obtainedfrom the phosphate methodin Figure 7. Below 1700 m, the two methodsyield coherentprofiles, but there is a noticeable divergencein the

biosphere. The AOU-derived A•5•3C/ACr wouldimply an

profilesabovethis level. The AOU-predicted Afi•3Cat the surfaceis -0.6%0 as comparedto the -0.8%0indicatedby the sponge proxy records [B6hm et al., 1996]. The largest discrepancyoccursat 900 m depth, where the AOU-based

A•5•3Cgoes througha minimumof about -0.2%o. The phosphate-based A•5•3C, in contrast, indicates no minimumin the vertical distributionand gives a value of about -0.55%o at this depth.

-0.

concerning thesurfacedistribution of preformed •5•3Candthe impact of the natural biogeochemical fractionation in the upper water column. The phosphatemethodassumesthat endmember mixing dominatesall the way up to the winter surface sourcesfor the water column. It also implicitly assumesthat

thespatialdistribution of preformed •5•3Cwascorrelated with that of preformed phosphate. Thus, corresponding to the preformed phosphate difference indicated by Figure 3,

A•513C, %0 0

exchangeable carbon mass in the terrestrial biosphere of about3000 Gt C, which is about 50% greaterthan the entire mass of the biosphere. The two methodologies carry different implications

8

preformed •5•3C is implicitlyassumed to havebeenon average 0.24%0 greaterin surfacesourcesfor the upper 900 m than

from AOU -500

those for 1700 m.

ß ß

,-

In contrast, it is assumed in the AOU

methodthat the preformed •5•3Cwasconstantfor the water column.

The phosphate assumption implies that the full effect of

- 1000 &e

fromPO4

localreoxidation of organicmatteron the•5•3C distribution is reducedby the combination of isopycnal mixing and gas-

- 1500

exchange effects in the regions of water mass formation.

This affectsthe calculationof A/5•3Cin the upper water column.For example,at 900 m, the naturaldecrease in •5•3C according to anAOU of 88 gmolO2kg'• is 0.65%o,equivalent to theobserved increase of 0.6 gmolP kg-• overits preformed

-2000

E œ

c). -2500

value. However, in the phosphatemethod, this is accorded -3000

onlyan 0.48%0decrease in •513C asthenaturaleffect,sincethe slopeof the •5•3Cvs. PO4 (-0.8%0pergmolP kg'•) is lower

I

ß

-3500=

than that due to biologicalfractionation. Thus, at 900 m, the discrepancybetween the two methodsresults from about a

0.2%odifference in the impliedpreformed/5•3C anda 0.15%o -4000

difference in the natural decreasedue to the biogeochemical cycle. The discrepancybetween the phosphate-and AOU-

ß

-4500

ß

basedA•5•3Cillustratesthat carbonisotope oceanography needs to be better understood.

-5000

4. 13CMass Balance

Solid points show AS13Cversusdepth as

In orderto illustrate howtheratioof Afi•3Cto ACrin the

calculatedaccordingto the AOU method. The curveshowsthe

oceandependson the amountof anthropogenicCO2addedto the oceanand the massof the exchangeable biosphere,we use a massbalanceapproachmodifiedfrom Broeckerand Peng

Figure 7.

trendof A•5•3C fromthephosphate method, drawnthrough the depth-averagedvalues (crosses).

474

KEIR ET AL.: •13CANOMALY IN THE NORTHEASTERN ATLANTIC

changes canbeobtained by subtracting •](Miø+ AMi)•iøfrom [1993]. This approach considers that certain preanthropogenic masses of carbonin theatmosphere (M,ø),sea both sides of (1), thus (Msø),andbiosphere(Mt,ø) haveundergone "characteristic" 1513Cdecreases up to the presenttime. Becausethe atmosphere is rapidlymixed,its 1513C changeis fairlyuniform i

spatially. In the terrestrialbiosphereand ocean, the decrease

in 15•3Cis not homogenous, as the atmospheric signal

i

The right side of this relationshipimplies that the magnitude

of the overall 1513C decrease dependson the amountof

anthropogenic CO2 added, the partitioning of this CO2 betweenthe reservoirs, and the exchangeablecarbon masses surfacewater. HereM.•. ø corresponds to the "penetration they contain. In the sections4.1 and 4.2, we impose one of depth"for the fossil fuel 13Canomaly,that is, M,.ø is the the following two conditionson equation(2): (1) the changes penetratesdifferently into these reservoirs. For the ocean, the characteristic 1513Cdecrease is taken to be that of the

andbiospheric 813Coccurin proportion to the carbonmasswhich,multipliedby the surface oceanAiS13C, in atmospheric yields the •3C inventory changein the water column ocean's isotopic change, or (2) the change in atmospheric /5•3Cis fixedat its proxy-observed value. In thefirst case,we [Broecker and Peng, 1993]. are interested in the expected behavior of the /5•3Cdecrease in For the terrestrialbiosphere, it is difficult to estimate an the ocean with variable amounts of anthropogenic CO2 averageA•13Crelativeto that of the atmosphere, in part becausethe size of the large amount of soil carbon and its turnover rate are both uncertain. Conceptually, we think of

uptake. In the secondcase,we are interestedin what couldbe deducedfrom the observedratio of thesetwo changes. Values common

to both calculations

the "characteristic" /5•3Cchange of thebiosphere (A•5 b) asthe averagechangein/513C thatoccursin the biosphericsystems 4.1. AiSs versus ACt actively exchanging with the atmosphere,i.e. vegetation, wood, and detritus. According to the ecosystem model of Emmanuelet al. [1984], these systems contain about 700 Gt C with an overall average turnover time of about 12 years. When such a box model is forced by an exponential change

withtime constantl/p, the A•13Cof the variousboxestend to respond exponentially.In time,theA•13Cin a box with residence time 1/k tends to approach a ratio of k/(k+p) relative to the A•13C of the influx.

Since the overall e-

foldingtime for the accelerating rateof ;5•3Cdecrease in the atmosphere hasbeenabout30 years,A;5b is roughlyestimated

to be 70% (i.e., 30/(30+12))of the atmospheric ;5•3Cchange (A/5,). For thisdiscussion, M•,ø,whichwelooselyreferto as

are listed in Table 2.

From ice core and air measurementsit appears that atmosphericCO2 has increasedby 75 ppm over the last 200 years, equivalentto an accumulationof 159 Gt C (as of 1992 [Keeling and Whorf, 1994]). First, we calculatethe changes

in surface water/513C (A/5•)thatoccurwhenvariousadditional amountsof anthropogenicCO2 have accumulatedin the ocean and in the biosphere. For this purpose,we assumethat the

/513C changes in eachreservoiroccurin proportionto each other. In the caseof the changein seawater/513C, the exact proportions do not significantly affect the result becausethe upperocean contains the largest exchangeablecarbon mass, roughly 3 times that of the combined atmosphere and

the exchangeable biospheric carbon, includes the active components above and the amount of the soil carbon

biosphere. We use AiS,,/AiS•:1.6 and AiS•,/AiS• = 1.1, the

deliveringthe equivalent of theiraverage ;5•3C change. For

core-atmosphereand the demospongeproxy records and the

example, in the model of Emmanuelet al. [1984], carbon

latterbeingbasedon the estimateof 0.7 for A/Sb/Ai5 a described

transfered

in section 4. Substitutinginto (2), one obtains:

from detritus and wood into soil carbon has about a

formercomingfrom the observed/5•3Cchangesin the ice

100 yearresidence time andthereforewouldexhibit a /5•3C changeabout25% (i.e. 30/(30+100))of the anomaly•13C suppliedfrom the "active" systems. Assuminga soil carbon

massof 1500 Gt C, M•,ø in this casewouldbe about700 + 1500x0.25•

l100GtC.

Given the

_

z i

(,s*-,s, ø)

ArSs - (1.6M. +1.1M,, +M•,)

(3)

whereM. = (M.ø + AM.), etc. From the amountof

preanthropogenic exchangeable carbon

masses,the budgetfor 13Ccanbe described in termsof the product of thecontemporary massand•13Cin eachreservoir,

anthropogenic CO2 addedto the ocean, AM,,, the change in inorganic carbon concentrationis given by

M i •ii [Tanset al., 1993]. Thus, Table 2. Values Employed in all Mass Balance i

Calculations

i

Parameter

where AMisthetotalamount ofanthropogenic CO2added, 5' is its average b•3C,andbiø is the pre-anthropogenic b•3C characteristicof eachreservoir. Isotopic fractionationsby

biosphereresultin naturaldifferencesin thesefi•3Cvalues.

As a result,the fil3c of anthropogenic CO2 is about20%0 lower than that of atmosphericCO2, is 30%0 lower than in surfacewater inorganiccarbon,but is not very different than the bulk of the biospherederivedfrom the C3 photosynthesis

pathway. Since •M = ••,

a relationshipfor the fil3c

Atmosphericcarbonmasses: Pre-Anthropogenic CUlIIUlatI¾(• 111Cl