The future of the carbon cycle: review, calciﬁcation ...carboncycle.princeton.edu/sites/default/files/pdfs/BarkerEtal_RSTA... · 22.07.2003 · The future of the carbon cycle:

10.1098/rsta.2003.1238

The future of the carbon cycle: review,calcification response, ballast and

feedback on atmospheric CO2

By S. Barker, J. A. Higgins and H. Elderfield

Department of Earth Sciences, University of Cambridge,Downing Street, Cambridge CB2 3EQ, UK

Published online 22 July 2003

The operation of the carbon cycle forms an important part of the processes rel-evant to future changes in atmospheric carbon dioxide. The balance of carbonbetween terrestrial and oceanic reservoirs is an important factor and here wefocus in particular on the oceans. Future changes in the carbon cycle that mayaffect air–sea partitioning of CO2 are difficult to quantify but the palaeoceano-graphic record and modern observational studies provide important evidence ofwhat variations might occur. These include changes in surface nutrient use, theoceanic inventory of nutrients, and the elemental composition and rain-rate ratioof marine particles. Recent work has identified two inter-linked processes of poten-tial importance that we consider in some detail: the response of marine calcifica-tion to changes in surface water CO2 and the association of particulate organiccarbon with ballast minerals, in particular biogenic calcite. We review evidencefrom corals, coccolithophores and foraminifera, which suggests that the responseof reduced calcification provides a negative feedback on rising atmospheric CO2. Wethen use a box model to demonstrate how the calcification response may affect theorganic carbon rain rate through the ballast effect. The ballast effect on exportfluxes of organic and inorganic carbon acts to counteract the negative calcifica-tion response to increased CO2. Thus, two oceanic buffers exert a significant con-trol on ocean–atmosphere carbonate chemistry: the thermodynamic CO2 buffer;and the ballast/calcification buffer. Just how tightly coupled the rain-rate ratioof CaCO3/Corg is to fluxes of ballast minerals is an important question for futureresearch.

Keywords: climate change; carbon dioxide; oceans; calcification

1. Introduction

Evaluation of future changes in the carbon cycle forms a crucial part of the effortsto determine how climate will respond to changes in greenhouse gases. Within thiswork, assessing future changes in oceanic carbon uptake is of great importance. Thisis because of the quantitative importance of the oceanic carbon reservoir compared

One contribution of 14 to a Discussion Meeting ‘Abrupt climate change: evidence, mechanisms andimplications’.

Phil. Trans. R. Soc. Lond. A (2003) 361, 1977–19991977

c© 2003 The Royal Society

on March 11, 2015http://rsta.royalsocietypublishing.org/Downloaded from

http://rsta.royalsocietypublishing.org/

1978 S. Barker, J. A. Higgins and H. Elderfield

310

320

330

340

350

360

370

380

1960 1970 1980 1990 2000year

250

260

270

280

290

300

310

320

02000400060008000years (before present)

(a) (b)

atm

osph

eric

CO

2 (p

pmv)

Figure 1. (a) Monthly average CO2 measurements from the SIO air sampling network (Keeling& Whorf 2002). Variations in amplitude between the records result from regional differencesin vegetation and anthropogenic influences. (b) Record of Holocene atmospheric CO2 evolutionfrom a combination of ice-core measurements and direct sampling (Indermuhle et al . 1999;Etheridge et al . 1998; Keeling & Whorf 2002). The sharp increase in CO2 over the last 200 yearsor so strongly contrasts with the gradual increase since ca. 8000 years ago.

with the other reservoirs with which carbon exchange occurs on human time-scales(the oceanic reservoir of carbon is about 60 times larger than that of the atmosphere).There are two fundamental questions. How will increased atmospheric CO2 alter thebalance between terrestrial and oceanic carbon reservoirs? How well known are thefactors affecting oceanic CO2 uptake and their possible responses to future changes?

The most recent assessment of the Intergovernmental Panel on Climate Change(IPCC) (Houghton et al . 2001) has provided estimates of the global climate of thetwenty-first century. Using a variety of scenarios, estimated CO2 levels in year 2100are between ca. 500 and 1000 ppm compared with annual average levels which haverecently reached 370 ppm (figure 1a). The estimates of the associated temperaturechange are between 1.5 and 5.5 C and those of sea-level change between 0.1 and0.9 m. The uncertainties in the estimates reflect, in part, incomplete understandingof the carbon cycle.

Records of changes in atmospheric CO2 over the last 420 000 years have beenobtained from ice cores. Although some doubts exist as to the extent to which thepast, or indeed the present, is a good analogue for future climate change (e.g. Mitchell1990), there is no doubt that the mechanisms causing past changes in atmosphericCO2 are extremely relevant. The message from the body of work on, for example,the causes of glacial–interglacial cycles in carbon dioxide is, according to the IPCC(Houghton et al . 2001), that ‘given the complex timing of changes between climatechanges and atmospheric CO2, it is plausible that more than one mechanism hasbeen in operation; and indeed most or all of the hypotheses encounter difficulties ifcalled upon individually to explain the full magnitude of the change’. In this paper,we shall comment on some ‘clues from the past’ as well as from the modern oceans,including laboratory and model simulations, to address some factors affecting thefuture of the carbon cycle.

Phil. Trans. R. Soc. Lond. A (2003)



The future of the carbon cycle 1979

2. The balance between terrestrial and oceanic carbon reservoirs

The pre-industrial oceanic carbon reservoir has been estimated at 38 240 Gt (1 Gt =1015 g), as compared with 600 Gt in the atmosphere and 1930 Gt in the terrestrial bio-sphere (850 Gt as biomass and 1080 Gt as soil) (Brovkin et al . 2002). There appearsto be a consensus that the availability of extra CO2 and fixed nitrogen in the futurewill result in an increase in carbon storage but that soil warming will decrease carbonstorage. However, the balance between these two factors is unknown. Here, we turnto the past and review the evidence as to what caused the atmosphere’s CO2 to riseduring the past 8000 years (figure 1b).

Indermuhle et al . (1999) proposed that the 20 ppm increase in CO2 since the mid-Holocene (ca. 8000 years ago) was caused by a ca. 200 Gt decrease in the terrestrialbiomass over this period. This is very large and equivalent to ca. 30 years of fossil fuelCO2 increase at the current rate. Broecker et al . (2001) have expressed surprise thatsuch a large change in storage might reflect a small change in climate. It is thereforepertinent that the reasons for this rise in CO2 are determined. In support of theinterpretation for a massive biomass change is the observation that the increase inCO2 is accompanied by a decrease in δ13C of atmospheric CO2 of ca. 0.3 ± 0.1 %%(Indermuhle et al . 1999). Additionally, it is consistent with a decline in deep-oceancarbonate ion concentration over the same period (Broecker et al . 1999, 2001).

Broecker has proposed an alternative explanation for the increase in CO2 sincethe mid-Holocene (Broecker et al . 1999, 2001): that of a long-term response of theoceanic carbonate system to a ca. 500 Gt increase in terrestrial biomass during theEarly Holocene. This mechanism involves four stages:

(i) an ‘instantaneous’ increase in deep water [CO2−3 ] as a response to the biomass

increase;

(ii) a deepening of the calcite lysocline resulting in an imbalance between carbonateinput and output (see § 5 a);

(iii) calcite preservation to restore balance (a preservation spike is seen in NorthAtlantic sediments at ca. 9000 14C yr BP (Broecker et al . 1993; Chapman etal . 1996));

(iv) leading to a decrease in [CO2−3 ] and long-term (of the order of 5 kyr) increase

in atmospheric CO2.

Brovkin et al . (2002) have recently used the CLIMBER-2 model in a transientHolocene simulation. Their simulation suggests that terrestrial carbon storage at8 ka may have been ca. 90 GtC higher than the pre-industrial value (less than the200 Gt estimated by Indermuhle et al . (1999)) but an additional oceanic source isalso required to explain the observed change of 20 ppm CO2. Brovkin et al . (2002)also point out that the change in atmospheric CO2 may be simulated solely by anincrease in oceanic carbonate sedimentation, leading to lower deep-water carbonateion concentrations and higher surface-ocean pCO2. This finding is in line with thehypothesis posited by Broecker et al . (2001). As yet the available data are insuffi-cient to discriminate between a terrestrial biosphere or deep oceanic source for theobserved Holocene increase in atmospheric CO2. Both higher precision δ13CO2 dataand a better understanding of oceanic carbonate preservation are needed.





Table 1. Factors potentially important for the future oceanic uptake of CO2

(after Houghton et al. 2001)

physical and chemical processes biologically linked processes

buffering changes changes in marine ecosystemsemission rates changes in nutrient utilizationwarming changes in oceanic content of major nutrientsvertical mixing and stratification changes in elemental composition of

biogenic materialschanges in rain-rate ratio

3. Factors affecting oceanic CO2 uptake andtheir response to future changes

Maps of the distribution of pCO2 in surface waters of the world’s oceans (e.g. Taka-hashi et al . 2002) demonstrate powerfully how complex the factors are that contributeto the exchange of CO2 between atmosphere and oceans. According to the IPCC(Houghton et al . 2001), the estimated flux of CO2 from the atmosphere to the oceanswill be between ca. 4 and 6 Gt yr−1 in the year 2100 compared with ca. 2 Gt yr−1

today. A number of physical, chemical and biological factors (table 1) have the poten-tial to affect future changes in the oceanic uptake of carbon dioxide (Houghton etal . 2001). Physical factors include changes in vertical mixing versus stratification ofthe oceans. Increasing vertical stratification (decreased mixing) would reduce CO2uptake by, in effect, reducing the oceanic volume available to CO2 absorption fromthe atmosphere. Stratification may also affect the biological carbon cycle by reducingthe available nutrients necessary for primary production in the euphotic zone. Chem-ical processes that may contribute to CO2-uptake variability include the CO2 bufferor Revelle factor (see § 4 b) and the effects of temperature on CO2 solubility. BecauseCO2 is less soluble in warm water than in cold water, pCO2 in seawater increasesby 10–20 ppm per degree increase in temperature, thereby increasing the sea-to-airflux and potentially increasing atmospheric CO2. Biologically linked processes areperhaps the most difficult to evaluate, although evidence from palaeoclimate recordssuggests that changes are likely to have occurred over glacial–interglacial time-scales.Here, we focus on the relationships between several of these factors: changes in theeffectiveness of the ocean CO2 buffer, changes in marine ecosystems, specifically thepotential fate of calcareous organisms and changes in the rain-rate ratio (the ratioof CaCO3 flux to Corg flux to the sea floor).

4. Response of marine calcification to changes in atmospheric CO2

(a) Production of calcium carbonate as a source of CO2

The major oceanic producers of calcium carbonate include coral reef communitiesand pelagic organisms such as foraminifera and coccolithophores (quantitatively themost important (Westbroek et al . 1993)). The production and burial of both organiccarbon and biogenic carbonate within the marine system provide a potential sinkfor carbon and therefore play a role in the global carbon cycle. However, while theformation of organic matter provides a sink for atmospheric CO2, the production





0.001

0.01

0.1

1

[CO2]

[HCO3]

[CO3 ]frac

tiona

l con

trib

utio

npH

4 10 115 6 7 8 9

2−

−

Figure 2. Bjerrum plot showing the relative contributions of CO2, bicarbonate (HCO−3 ) and

carbonate (CO2−3 ) ions to total dissolved inorganic carbon as a function of pH. Typical surface

seawater has a pH of approximately 8.2 and is therefore dominated by HCO−3 .

of carbonate causes an increase in pCO2 of surface waters and therefore acts as apotential source of CO2 to the atmosphere. The two processes may be representedby the following simplified equations.

Formation of organic matter:

CO2 + H2O = CH2O + O2. (4.1)

Formation of biogenic carbonate:

Ca2+ + 2HCO−3 = CaCO3 + CO2 + H2O. (4.2)

Although the formation and burial of CaCO3 provides a net sink for carbon, itis the production of CO2 during calcification that is of interest here. The interplaybetween the formation of organic matter and calcium carbonate and their opposingeffects on surface-ocean pCO2 has led to considerable debate on the role of marinecalcifying organisms in atmospheric CO2 variability, i.e. do they act as sources orsinks of atmospheric CO2? In particular, coral reefs have received much attention(e.g. Ware et al . 1992; Gattuso et al . 1993; Kayanne et al . 1995; Kawahata et al . 1997;Ohde & van Woesik 1999). The bloom episodes of coccolithophore production havealso been investigated for their effects on the CO2 system in surface waters (Holliganet al . 1993; Bates et al . 1996; Crawford & Purdie 1997; Murata & Takizawa 2002).Although the ‘source/sink’ debate is still active, it is the general consensus that coralcommunities, at least, provide a net source of CO2 to the atmosphere. Ware et al .(1992) provide an estimate of the net global coral reef CO2 source as 0.02–0.08 GtCeach year. This may be compared with the estimated annual emission of 7 GtC fromfossil fuel burning (Houghton et al . 1995).

(b) Marine calcification as a positive feedback to rising atmospheric CO2

The production of one mole of calcium carbonate does not result in the releaseof one mole of CO2. This is due to the carbonate buffering capacity of seawater(figure 2). An increase in dissolved CO2 will tend to lower the pH of seawater butthis will be partly compensated for by the reaction of CO2 with CO2−

3 :

CO2 + CO2−3 + H2O = 2HCO−

3 . (4.3)





In the modern ocean, the ratio of released CO2 to precipitated carbonate, Ψ , isapproximately 0.6 (Ware et al . 1992). Frankignoulle et al . (1994) proposed that Ψincreases as a result of increasing atmospheric CO2 concentration.† Specifically, adoubling of atmospheric CO2 would cause Ψ to increase to 0.76. This could providea net CO2 source of ca. 5 GtC between 1880 and 2050 (Frankignoulle et al . 1994).Therefore, if all else remained equal, marine calcification could be expected to providea positive feedback to rising atmospheric CO2. The CO2 produced by formation ofcalcium carbonate does not necessarily evade into the atmosphere. It may be usedduring the formation of organic matter and therefore may not provide a direct sourceof CO2 to the atmosphere. However, if the global production rate of organic matterand inorganic carbonate are considered to be constant, then an increase in Ψ willprovide a net source of CO2 to the ocean atmosphere system.

(c) Correlation between calcification rate and pCO2

The saturation state of seawater with respect to calcium carbonate may be definedas

Ω = [Ca2+][CO2−3 ]/Ksp, (4.4)

where Ksp is the stoichiometric solubility product for CaCO3 (a function of temper-ature and pressure). When Ω = 1, the solution is considered saturated with respectto carbonate; a value less than unity represents undersaturation and dissolutionmay occur, while a value greater than unity implies supersaturation. Since [Ca2+]is approximately conservative in seawater, [CO2−

3 ] may be considered as the domi-nant control over Ω. On dissolution in seawater, CO2 dissociates to form HCO−

3 andCO2−

3 ions. At average pH values, ca. 90% of the total dissolved inorganic carbon(∑

CO2) in seawater is in the form HCO−3 , the remainder is mostly CO2−

3 (figure 2).CO2 represents less than 1% of dissolved inorganic carbon in average surface-oceanwaters.‡ The partitioning between inorganic carbon species in seawater is such that,for a given temperature and salinity, pCO2 is approximately inversely proportionalto [CO2−

3 ]. A similar relation exists between pCO2 and Ω.

† This decrease in the effective buffering capacity of seawater is analogous to the Revelle factor,which predicts that the capacity of ocean water to absorb atmospheric CO2 by reaction with CO2−

3 willdecrease as atmospheric CO2 rises and surface-ocean [CO2−

3 ] decreases.

‡ The partitioning of marine inorganic carbon species between HCO−3 , CO2−

3 and CO2 provides theexplanation for the significantly larger capacity for carbon storage in the oceans than in the atmosphereand is the basis for assuming that the oceans ultimately control atmospheric CO2 concentrations.

Figure 3. The effect of carbonate saturation on marine calcification. (a) Compilation of culture data fromthe BIOSPHERE-2 facility, showing the relation between saturation state and calcification rate for a coralreef mesocosm (Langdon et al . 2000; Langdon 2001). Although the data shown here represent severalexperiments involving a variety of chemical perturbations, there is a strong positive relation betweencalcification rate and saturation state. (b) Culturing experiments on two species of coccolithophore(Emiliania huxleyi, circles; Gephyrocapsa oceanica, squares) demonstrate a negative relation betweencalcification rate and pCO2 (reproduced from Riebesell et al . (2000), with permission from the NaturePublishing Group). This is synonymous with a positive relation between calcification rate and [CO2−

3 ].Note also that organic matter production shows a slight positive response to increased pCO2. (c) Datafrom Barker & Elderfield (2002) showing the relation between foraminiferal shell weight and [CO2−

3 ] (themain controller of calcite saturation).





aragonite

calc

ific

atio

n(1

0−13

mol

C c

ell−1

d−1

)

pH

0

2

4

6

8

10

0

2

4

6

8

10

0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

calc

ite/P

OC

280

365

750

8.4

200

300 200 150 100 75

400 600 800

8.2 8.1 8.0 7.9 7.8

180 200 220 240 260

G. bulloidesG. inflataN. pachyderma (s)

shel

l wei

ght (

µg)

0

40

80

120(a)

(b)

grow

th r

ate

(mm

ol C

aCO

3 m

−2 d

−1)

4 5 61 2 3

(c)

5

15

25

35

[CO3 ]surface (mol kg−1)2−

[CO3 ]surface (µmol l−1)2−

5 10 20 25 30 3515[CO2] (µmol l−1)

POC

pro

duct

ion

(10−1

3 m

ol C

cel

l−1 d

−1)

pCO2 (ppmv)

Figure 3. See opposite for description.

Most of the surface ocean is supersaturated with respect to both the calcite andaragonite forms of calcium carbonate. Therefore, it may be assumed that smallchanges in surface-ocean carbonate chemistry will not have a perceptible effect onmarine calcification. However, a range of studies has shown a strong dependenceof carbonate production rates on the degree of supersaturation. Several studies on





1.0

1.5

2.0

2.5

3.0

3.5 10

14

18

22

0 10 30 40 50

18O

(‰

)

age (ka)

180

200

220

240

260

280

150

200

250[CO2] (atmos)

shel

l wei

ght (

µg)

(a)

(b)

[CO

2] (

ppm

v)

20

shell weight

0 10 30 40 50age (ka)20

pCO

2 (a

tm)

pCO2 (aq)18Ο

Figure 4. (a) Foraminiferal shell weight versus δ18O and (b) calculated surface ocean pCO2

versus atmospheric CO2 from the VOSTOK ice core (Barnola et al . 1999) from a sedimentcore in the North Atlantic. As predicted from thermodynamic considerations, shell weights wereheavier during the Glacial period, when surface ocean [CO2−

3 ] would have been higher, reflectinglower values of atmospheric CO2. (Reprinted with permission from Barker & Elderfield (2002).Copyright 2002 American Association for the Advancement of Science.)

coralline calcification have concluded that rates of carbonate production are pos-itively correlated with the carbonate saturation state of seawater (Gattuso et al .1998; Langdon et al . 2000; Marubini et al . 2001; Leclercq et al . 2002) (figure 3a).Similar findings (figure 3b) have been reported for calcification rates in coccoliths(Riebesell et al . 2000). The shell weights of planktonic foraminifera have also beenshown to vary in response to increasing saturation state (figure 3c) of seawater duringgrowth (Spero et al . 1997; Bijma et al . 1999; Barker & Elderfield 2002). An increasein shell weight is thought to reflect higher rates of calcification. All of these studiesare in line with inorganic precipitation experiments that demonstrate a positive rela-tion between saturation state and precipitation or calcification rate (Zhong & Mucci1989; Zuddas & Mucci 1998). Since the carbonate saturation state of the surfaceocean is intimately linked to pCO2, it follows that rates of marine calcification willbe sensitive to changes in atmospheric CO2.

(d) Past evidence of marine calcification response to changing atmospheric CO2

Evidence from ice cores suggests that atmospheric CO2 during the last glacialmaximum (LGM) was ca. 30% lower than pre-industrial values. From our knowledgeof the marine carbonate system, we can predict that surface ocean [CO2−

3 ] wouldhave been correspondingly higher during glacial times. This proposition is supportedby isotopic evidence for a lower pH in the glacial oceans (Sanyal et al . 1995). Arecord of planktonic foraminiferal shell weights from the North Atlantic reveals thatsimilarly sized shells of Globigerina bulloides were ca. 70% heavier (figure 4a) dur-ing the LGM than during the Holocene (Barker & Elderfield 2002). By using anempirical calibration of shell weight to [CO2−

3 ] and combining chemical proxies toattain estimates of temperature and salinity, Barker & Elderfield (2002) showed that





(a) (b) (c) 1880

1990

2065

2100

1880

1990

2065

2100

2 3 4 5 50 70 9060 80 100 110

calcification (% of 1880 avg)aragoniteΩ

Figure 5. Projected future evolution of surface-ocean saturation state with respect to aragoniteas determined by (a) thermodynamic calculations and (b) the HAMOCC model. (c) Projectedchanges in reef calcification as a response to lowered saturation states. (Reprinted with permis-sion from Kleypas et al . (1999). Copyright 1999 American Association for the Advancement ofScience.)

the observed weight difference was appropriate for the observed decrease in glacialatmospheric CO2 (figure 4b).

(e) Marine calcification as a negative feedback to rising atmospheric CO2

The continued increase in atmospheric CO2 will cause a lowering of surface ocean[CO2−

3 ] and a lowering of the saturation state with respect to calcite and arago-nite. Due to the observed sensitivity of calcification rates in corals, coccoliths andforaminifera to changes in carbonate saturation, it may be predicted that carbon-ate production by these groups will decrease into the future. This phenomenon issignificant not only to the ecological stability of these types of organism but also totheir role in the global carbon cycle. On the time-scales relevant to modern society,the immediate impact will be a decrease in evolved CO2 due to reduced CaCO3formation.

Kleypas et al . (1999) presented future projections of surface-ocean Ωaragonite dis-tributions based on thermodynamic constraints and results from the HAMOCC(Hamburg Ocean Carbon Cycle) global model (Maier-Reimer 1993) (figure 5a, b).They also predicted consequent global changes in reef calcification rates over thenext 100 years, based on average calcification responses of tropical marine algae andcorals (figure 5c). It is clear from their estimates that the possible impact of increas-ing atmospheric CO2 on coral communities is significant. In a recent review of theexperimental evidence for effects of CO2 on calcification of reef-building organisms,





1850 1900 1950 2000 2050 2100 21500.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

year

pote

ntia

l CO

2 re

leas

e (G

tC y

r−1)

1→

←23→

4→

Figure 6. Modelled results of the calcification effect on potential CO2 release by marine calcifi-cation. Line 1 is the control run with no calcification response (i.e. constant CaCO3 production)and demonstrates the effect of reduced carbonate buffering (increased values of Ψ) with increas-ing pCO2. Lines 2–4 represent model runs with various calcification sensitivities. (Reproducedby permission from Zondervan et al . (2001). Copyright 2001 American Geophysical Union.)

Langdon (2001) states that based on recent studies, calcification rates on coral reefscould decline by 11–44% by the middle of this century. Further, due to a long-termreduction in skeletal growth it is likely that corals will be more susceptible to bio-erosion, storm damage and possibly disease (Langdon 2001). Other factors related toelevated atmospheric CO2 may also influence coral reef ecology, notably the ‘bleach-ing’ effect of increased temperatures.

As discussed previously, increasing the atmospheric concentration of CO2 willdecrease the carbonate buffering capacity of the surface ocean. This means that moreCO2 will be produced per mole of calcium carbonate precipitated and therefore con-stitutes a positive feedback to rising CO2. However, due to the observed negativeimpact of increased atmospheric CO2 on marine calcifying organisms, it is probablethat global marine biogenic carbonate production will decrease into the future. Thiswill counter the positive feedback effect of a reduced buffering capacity and mayprovide a net negative feedback to rising CO2. Zondervan et al . (2001) present amodel for this scenario based on experimental evidence for coccolith-calcification-rate dependence on pCO2. They predict that global production of CO2 by precip-itation of CaCO3 may decrease by up to 0.4–0.5 GtC yr−1 by the year 2150 (evenaccounting for the decreased buffering capacity), given a modern global carbonateproduction rate of ca. 1 GtC yr−1 (figure 6).

5. The ballast hypothesis

(a) Rain-rate ratio

Considerable dissolution of marine carbonate occurs within the upper few centi-metres of sediment at the sea floor. This dissolution is driven by a lowering of [CO2−

3 ]within sediment pore waters as a result of increased CO2 released during respirationof organic matter within the sediments (Emerson & Bender 1981). Obviously, theamount of organic matter entering the sediments and therefore available to respira-tion dictates the amount of dissolution that may occur in this way. A higher ratio





of organic matter to inorganic carbonate will cause an increase in the proportionof carbonate that dissolves. The rain-rate ratio or rain ratio may be considered asthe molar ratio of particulate organic carbon to particulate inorganic carbon (i.e.carbonate) arriving at the sea floor. Changes in the rain ratio may alter the extentof carbonate dissolution or preservation on a global scale due to its influence onsediment pore-water chemistry (Archer 1991).

On time-scales of thousands of years, the supply rate of calcium carbonate to theoceans by terrestrial weathering is balanced by deposition and burial of carbonateat the sea floor. If this steady-state situation is perturbed, for example by a decreasein carbonate burial rates, the system will tend to respond in such a way as to regainthe balance (Broecker & Peng 1987). In this example, reduced burial of carbonateat the sea floor will cause an increase in deep-water [CO2−

3 ]. This in turn will leadto an increase in the preservation potential of carbonate sediments and will result inan increase in carbonate output to match the original input. This process is knownas carbonate compensation (Broecker & Peng 1987).

Archer & Maier-Reimer (1994) presented a model that combined changes in therain ratio (and subsequent adjustment of pore-water calcite dissolution) with carbon-ate compensation to explain the observed decrease in atmospheric CO2 during glacialtimes. They suggested that a 40% decrease in carbonate production with respect toorganic production in the surface ocean would cause a decrease in carbonate burialand an increase in deep ocean [CO2−

3 ]. This in turn would cause a lowering of atmo-spheric CO2 to near glacial values due to the inverse relation between [CO2−

3 ] andpCO2.

As discussed earlier, the effect of continued atmospheric CO2 increase in the futurewill probably lead to reduced production of marine carbonate. It may seem likelythat we should expect a corresponding decrease in the ratio of carbonate to organicmatter falling to the deep sea. This would lead to changes in carbonate burial ratessimilar to those described by Archer & Maier-Reimer (1994) and possibly to long-term changes in deep-sea [CO2−

3 ] and atmospheric CO2.

(b) The ballast hypothesis

Quantifying the vertical flux of organic matter from the surface to the deep oceanis important for investigations into the role of the oceanic biological pump in theglobal carbon cycle. Models attempting to represent the cycling of organic matterand nutrients in the upper ocean (e.g. Sarmiento et al . 1993) have used power-lawfunctions, based on empirical observations (e.g. Martin et al . 1987), to representvariations in the sinking flux of organic matter with depth. These models implyan independence between organic and inorganic material fluxes to the deep sea.However, recent work by Armstrong et al . (2002) and Klaas & Archer (2002) suggeststhat fluxes of organic and inorganic matter are in fact intimately related to oneanother. This interdependence implies that deep-sea rain ratios may ultimately befixed at least on regional scales.

The dominant role of large particles and aggregates in the flux of material tothe ocean floor is well documented (e.g. McCave 1975; Honjo 1980). Apart froman organic component, inorganic mineral phases such as opaline silica, calcium car-bonate and lithogenic dust (collectively termed ‘ballast’ minerals) represent a majorconstituent of these particles (Honjo 1980) and presumably play an important role





POC

flu

x (m

g m

−2 d

−1)

CaCO3 flux (mg m−2 d−1)0 100 200 300 400 0 200 400 600 800 0−50 50 100 150 200 250

80

60

40

20

0

opal flux (mg m−2 d−1) lithogenic flux (mg m−2 d−1)

(a) (b) (c)

Figure 7. Scatter plots showing the correlation between POC flux and ballast flux below 1000 mfrom a global coverage of sediment traps. CaCO3 is thought to be the most important transporterof POC to the deep sea. (a) r = 0.829, (b) r = 0.595, (c) r = 0.536; P < 0.0001. (Reproducedby permission from Klaas & Archer (2002). Copyright 2002 American Geophysical Union.)

in the rate of sinking due to their relatively high densities (Smayda 1970; McCave1975; Ittekkot & Haake 1990). Armstrong et al . (2002) present a new model fordescribing the flux of sinking particulate organic carbon (POC) through the watercolumn. They suggest that below ca. 1800 m the flux of organic matter is in directquantitative proportion to the fluxes of ballast minerals. This may be illustratedby the relative constancy of POC flux normalized to total mass flux (i.e. the massproportion of organic matter to total material flux) as compared to the absolute fluxof POC. Klaas & Archer (2002) present a compilation of sediment trap data fromaround the globe from which they conclude that most (80–83%) of the organic car-bon rain to the deep sea is carried by calcium carbonate (figure 7). They argue thatthis is because carbonate is denser than opal (and therefore has a greater carryingcoefficient) and more abundant than terrigenous material.

6. Modelling the effects of POC ballasting: a positive feedbackto rising atmospheric CO2

The intimate relation between deep vertical fluxes of carbonate and organic mattersuggested by Armstrong et al . (2002) and Klaas & Archer (2002) has implicationsnot only for theories involving past changes in the rain-rate ratio (e.g. Archer &Maier-Reimer 1994) but also for atmospheric CO2 uptake related to the burial oforganic matter. The formation of organic carbon in the surface ocean involves anuptake of CO2 (equation (4.1)). As POC sinks through the water column, most isremineralized back to inorganic CO2 in the upper 1–2 km (Armstrong et al . 2002).The rest is transported to the deep sea and sediments and provides a net sink foratmospheric CO2 on centennial time-scales. A reduction in the proportion of POCsinking to deeper waters will reduce the effective capacity of organic matter formationas an atmospheric CO2 sink. We propose that as a result of decreased carbonateproduction in response to rising atmospheric CO2, the flux of POC to the deepocean will also be reduced due to the ballast effect. The resulting redistribution ofPOC remineralization with depth will then provide a positive feedback for increasingatmospheric CO2.





Table 2. Input parameters for the ocean–atmosphere box model used in the study

temperature volumeregion (C) salinity (kg)

l 21.5 34.7 5.111 × 1016

n 4.0 35.5 5.518 × 1015

s 8.0 35.0 5.518 × 1015

p 1.0 34.0 4.259 × 1015

pi 1.0 35.0 4.685 × 1016

m 5.0 35.0 2.896 × 1017

a 2.0 35.0 2.358 × 1017

d 1.0 35.0 6.474 × 1017

global CO2 3.0379 × 1018 mol volume of ocean 1.292 × 1018 kgglobal alkalinity 3.14 × 1018 mol area of ocean 3.49 × 1014 m2

global phosphate 2.769 × 1015 mol volume of the atmosphere 1.773 × 1020 mol

POC productiona 12 GtC yr−1 benthic flux POCa 0.57 GtC yr−1

CaCO3 productiona 3 GtC yr−1 benthic flux CaCOa3 0.95 GtC yr−1

ballast parametersb

rain rate (ρ) 0.6fraction CaCO3 burial 0.3length-scale of remineralization (δB, δE) 520 m

exchange term water flux (Sv) region gas flux (m d−1)

T 30 l 3flm 80 n 5fppi 80 s 5fpid 80 p 5fpia 30

aArcher et al . (2000); bArmstrong et al . (2002).

(a) Model description

We simulate the effect of increased atmospheric CO2 on calcification and mineralballasting of POC using a modified version of the eight-box ocean–atmosphere modelof Toggweiler (1999). The ocean is represented as an eight-box reservoir consisting offour surface boxes, three intermediate boxes, and a deep box (table 2, figure 8). Thefour surface-ocean boxes represent the sinking region of the North Atlantic (n), thelow-latitude surface-ocean (l), the Subantarctic (s), and the upwelling region of thepolar ocean (p). The intermediate ocean is divided into a polar-intermediate region(pi), a region corresponding to the thermocline (m), and the rest of the intermedi-ate depth water (a). Oceanic conveyor circulation (denoted T), originates with deepwater formation in the North Atlantic, which upwells to the surface in the SouthernOcean. Exchange terms (fppi, fpid, flm and fpia) simulate bi-directional exchangebetween boxes, fppi and fpid representing specifically the formation of Antarcticbottom water. Values for exchange and circulation fluxes were set so as to reproduce





atmosphere

polarsub-polar

polarinter-

mediate

deep

intermediate

low latitude

thermocline

NorthAtlantic

Figure 8. Nine-box model of the ocean–atmosphere system after Toggweiler (1999). Bi-directionalarrows indicate exchange fluxes between reservoirs. Dashed arrows represent the particulate fluxout of the surface ocean. Large-headed arrows represent oceanic conveyer circulation (T).

the observed distribution of phosphorus in the surface, intermediate, and deep ocean.The magnitudes of the fluxes are significantly larger than similar box-model studies,owing to the high surface-ocean production values (14 GtC yr−1 and 3.5 GtC yr−1

for POC and CaCO3, respectively). We attribute this apparent inconsistency withinour model as a result of the equations of Armstrong et al . (2002), which may under-estimate dissolution and remineralization of CaCO3 and POC in the upper watercolumn.

A full set of input parameters for the model are given in table 2. Carbon spe-ciation in all the oceanic boxes and the surface-ocean partial pressure of CO2 arecalculated after Zeebe & Wolf-Gladrow (2001). The Redfield ratio Corg:N:P, linkingthe production and remineralization of POC to changes in phosphorus and alkalinityis set to 106:16:1 (Redfield et al . 1963). Production rates of POC and CaCO3 wereset at 14 GtC yr−1 and 3.5 GtC yr−1, respectively, using the equations of Armstronget al . (2002), the ballast parametrizations listed in table 1, and the estimates forbenthic CaCO3 and associated POC fluxes (0.9 GtC yr−1, 0.6 GtC yr−1) of Archeret al . (2000). Sediment burial is not simulated in the model and all of the sinkingPOC and CaCO3 is remineralized within the water column.

Primary productivity and the rain of both organic and inorganic carbon out of thesurface ocean are defined by the numerical model of Armstrong et al . (2002), whichaccounts for the observed close correlation between POC fluxes and ballast mineralfluxes (calcium carbonate, opal and lithogenic material). For simplicity, we consideronly calcium carbonate (CaCO3) in our calculations of POC fluxes: a reasonableassumption given the results presented by Klaas & Archer (2002), which suggestthat the majority of the benthic flux of organic carbon (80–83%) is associated withCaCO3. According to Armstrong et al . (2002), the total flux of POC at depth is par-titioned between quantitatively associated POC and the remaining or excess POC.





Assuming that excess and ballasted POC decay exponentially to an asymptote cor-responding to the observed benthic ratio of POC to ballast, Armstrong et al . (2002)defined the flux of POC at depth (FOC(z)) as

FOC(z) = FOC(∞) + FOC(∞)[FB(z0)/FB(∞) − 1] exp−(z − z0)/δB+ [FOC(z0) − FOC(∞)[FB(z0)/FB(∞)] exp−(z − z0)/δE, (6.1)

where δB and δE are the remineralization length-scales for ballast (CaCO3) and excessPOC. In this preliminary study, we used the additional simplification of Armstronget al . (2002) that δB = δE, reducing equation (6.1) to

FOC(z) = FOC(∞) + [FOC(z0) − FOC(∞)] exp−(z − z0)/δE, (6.2)

where FOC(∞) is evaluated from a prediction of FB(z0) by

FOC(∞) = ρFB(∞) = ρFB(z0)[FB(∞)/FB(z0)], (6.3)

and ρ is the asymptotic POC-to-ballast ratio (the rain ratio), i.e. FOC(∞)/FB(∞).Similarly, the dissolution of CaCO3 is described as an exponential decay to the burialflux of CaCO3:

FB(z)/FB(z0) = FB(∞)/FB(z0) + [1 − FB(∞)/FB(z0)] exp−(z − z0)/δB. (6.4)

The inverse relationship between calcification and pCO2 is modelled using an expo-nential equation of Barker (2002),

FB(z0, t) = FB0(z0, t0) exp0.0083([CO2−3 ]t − [CO2−

3 ]t0), (6.5)

where [CO2−3 ]t0 is the initial steady-state carbonate ion concentration. Over a range

of concentrations of atmospheric CO2 (300–700 ppm), equation (6.5) gives a decreasein marine calcification of 40%, consistent with a predicted decrease in calcificationover a comparable range of CO2 in coccoliths (16–45%) (Zondervan et al . 2001),corals (40%) (Langdon et al . 2000) and natural planktonic assemblages (36–83%)(Riebesell et al . 2000).

(b) Results

The model is initialized with steady-state partitioning of total CO2, alkalinity andphosphorus. At each model time-step, CO2 is added to the atmosphere at a constantrate. This increase in atmospheric CO2 lowers the concentration of CO2−

3 in thesurface ocean in accordance with the inverse relationship between CO2 and CO2−

3 .The decrease in CO2−

3 results in a reduction in marine planktonic calcification, inaccordance with equation (6.5), from an initial production rate of 3 GtC yr−1 to avalue of 1.8 GtC yr−1. This decrease in CaCO3 production lowers the amount of CO2produced via calcification. Alternatively, the decrease in CaCO3 production can beinterpreted as a relative increase in surface-ocean alkalinity which is equivalent toan increase in the capacity of the surface ocean for CO2 uptake. As a result, thecalcification response acts as a negative feedback on increasing atmospheric CO2(figure 9a). The net draw-down of atmospheric CO2 resulting from the calcificationresponse is small (17.4 ppm or ca. 5% of the total atmospheric CO2 increase), butshould not be neglected when evaluating the role of the ocean in mitigating changesin atmospheric CO2.





−3.0

−1.5

0

300 400 500 600 700

calcificationeffect

0

0.5

300 400 500 600 700

ballasteffect

1.0

pCO2 (ppm)pCO2 (ppm)

(a) (b)

CO

2 up

take

(G

t C)

Figure 9. (a) Negative feedback of calcification response to increasing atmospheric CO2. Thedashed line is the control run with no calcification effect. The decrease in effective CO2 uptakewith increasing pCO2 in the control run is due to the positive-feedback effect of lowering thecarbonate-buffering capacity (Frankignoulle et al . 1994). CO2 fluxes were calculated using onlyCaCO3 production; the effect of CaCO3 dissolution on subsequent CO2 uptake was neglected.(b) Positive feedback of mineral ballasting of POC as a consequence of the calcification response.CO2 fluxes were calculated using POC export to the deep sea. The dashed line represents CO2

uptake by POC export with no ballast effect.

Mineral ballasting of POC adds a further complication to the response of calcifi-cation to changes in atmospheric CO2 as changes in ballast production drive simul-taneous changes in the POC flux (equation (6.3)). A decrease in CaCO3 productionresults in an upward shift in the locus of POC remineralization, thereby lowering theflux of POC to the deep sea (figure 10). As illustrated in figure 10b, this decrease incalcification in response to increasing atmospheric CO2 results in a significant reduc-tion in the POC flux (10–50%) below ca. 1500 m. Reducing the deep-sea POC fluxeffectively increases the reservoir of CO2 in the surface and intermediate ocean. Thisincrease in CO2 in the surface ocean acts as a positive feedback on increasing atmo-spheric CO2 (figure 9b). Adding the effect of mineral ballasting to the model causesthe amount of surface ocean CO2 and atmospheric CO2 to increase by 3 µmol kg−1

and 3 ppm, respectively.The combined model results of calcification response and ballast effect on atmo-

spheric CO2 are shown in figure 11a. The negative feedback effect of reduced calcifi-cation acts to lower predicted atmospheric CO2 values by ca. 2.5% (by 2250) below acontrol scenario with neither calcification nor ballast effects included. Adding in theballast effect goes some way to countering the calcification effect, but is insufficientto completely reverse it. In response to this, we considered the effect of changingthe rates of dissolution and remineralization of CaCO3 and POC. When the dis-solution and remineralization rates are increased (by using a shorter length-scale ofdissolution/remineralization), less CO2 is exported to the deep ocean, resulting inhigher concentrations of CO2 in the surface and intermediate ocean. Lowering thelength-scale of dissolution by 20% increases surface-ocean CO2, alkalinity and atmo-spheric CO2 by 21 µmol kg−1, 13 µmol kg−1 and 20 ppm, respectively. This enablesthe ballast effect to completely counter the negative feedback of reduced calcifica-





remineralization remineralization

export production

calcificationpCO2

0

1000

2000

3000

4000

40% 60% 80% 100% 120%

2150925 ppm

2050498 ppm

2000370 ppm

5000

dept

h (m

)

flux of organic carbon (T)

export production

export production(a)

(b)

flux of organic carbon (2000)

Figure 10. (a) The effect of the calcification response on mineral ballasting of POC to the deepsea. Including the effect causes an increase in the amount of POC remineralized before sinkingto the deep sea. (b) Modelled future POC fluxes based on IPCC estimates of pCO2 increase andthe response of marine calcification.

tion and provides a net positive feedback to rising CO2 (figure 11b). Similarly, anincrease in the length-scale of dissolution yields greater export of POC and CaCO3out of the surface and intermediate ocean. Increasing the length-scale of dissolutionby 20% lowers surface-ocean CO2, alkalinity and atmospheric CO2 by 17 µmol kg−1,9 µmol kg−1 and 16 ppm, respectively.

In addition to CaCO3 and POC dissolution and remineralization rates, weinvestigated the sensitivity of the model to changes in the burial flux of CaCO3(FB(z0)/FB(∞)) and the asymptotic transport ratio, ρ, equivalent to the rain-rateratio (FOC(∞)/FB(∞)). The asymptotic transport ratio is a measure of the POCcarrying capacity of CaCO3. As a result, changes in ρ are important with respectto the flux of POC at depth. A change in the rain-rate ratio from 0.6 to 0.7(mol POC/mol CaCO3) yields a 16% increase in the POC flux at 3000 m. How-





2000 2250year2125

0.97

0.98

0.99

1.00

1.01

2000 2250year2125

calcificationeffect

ballasteffect

pCO

2

(a) (b)

∆ballasteffect

calcificationeffect

(mod

elle

d/bu

sine

ss a

s us

ual)

Figure 11. (a) ∆pCO2 curves for calcification response and mineral ballasting of POC. (b)∆pCO2 curves for calcification response and mineral ballasting with increased dissolution/remineralization rates of CaCO3 and POC. Remineralization length-scales are shortened by 20%as a consequence of a reduction in the saturation state of the ocean with increasing atmosphericCO2.

ever, given that the POC flux at 3000 m is only ca. 5% of the surface POC exportflux, the partitioning of CO2 in the ocean–atmosphere system is relatively insensi-tive to changes in the rain-rate ratio on short time-scales. There is a small responseof atmospheric CO2 to changes in the CaCO3 flux to the sediment. On time-scalescorresponding to the anthropogenic rise in atmospheric CO2, changes in the CaCO3flux to the sediment yield a significant shift in the distribution of CO2 and alkalinityin the oceanic reservoir. However, this shift results in only a trivial change (less than1 ppm) in atmospheric CO2.

(c) Discussion

The model calculations suggest that mineral ballasting of POC acts as a positivefeedback on increasing atmospheric CO2. The magnitude of this positive feedbackis sensitive to the dissolution/remineralization rates of CaCO3 and POC. Assuminga 20% increase in the dissolution/remineralization rates of CaCO3 and POC, wecalculate a small net increase in atmospheric CO2 (positive feedback on increasingatmospheric CO2) related to the combined effects of mineral ballasting and reducedcalcification (figure 11b). While these results depend critically on an increase indissolution/remineralization rates, we suggest that such a response is plausible giventhe relationship between increasing atmospheric CO2 and the saturation state of thesurface ocean with respect to CaCO3.

While our model does reproduce the expected changes in oceanic CO2 partitioning,there are a number of further experiments which will contribute to the future analysisof the ballast hypothesis on the global carbon cycle. Of particular interest is the effectof mineral ballasting on the regulation of atmospheric CO2 on glacial–interglacialtime-scales. Simulating the effect of mineral ballasting of POC on glacial–interglacialtime-scales will require a model that includes all three types of ballast minerals(opal, calcium carbonate and lithogenic material), weathering/sediment burial andcarbonate compensation. Additionally, a more mechanistic understanding of mineralballasting is required. Given the apparent importance of biologically mediated dis-solution of CaCO3 (e.g. Milliman et al . 1999), it is clear that ballast mineral fluxes





are not independent variables, but themselves depend on the flux of POC. From theperspective of changes in the global carbon cycle, oceanic productivity, and on longertime-scales POC and CaCO3, burial fluxes are considered important regulators ofatmospheric CO2. However, scant attention has been given to the underlying mecha-nisms governing POC and CaCO3 remineralization/dissolution. As these mechanismsultimately determine the sediment fluxes of both organic carbon and calcium carbon-ate, a more quantitative treatment may shed light on additional important feedbackswithin the global carbon cycle.

7. Conclusions

Understanding the role of the ocean in the global carbon cycle is critical to our abilityto predict the Earth’s response to the current anthropogenic rise in atmospheric CO2.Much of the uncertainty surrounding the role of the ocean in the global carbon cyclecan be attributed to a dearth of knowledge concerning the mechanisms involved inthe ocean’s biological carbon pump and how these might respond to changes in atmo-spheric CO2. Frankignoulle et al . (1994) first calculated that marine calcification actsas a positive feedback on increasing atmospheric CO2, resulting from the decreasedbuffer capacity of the surface ocean. However, Frankignoulle et al . (1994) assumedthat calcification rates remained constant over time. Subsequent studies on coccoliths(e.g. Zondervan et al . 2001), corals (Langdon et al . 2000) and foraminifera (Barker2002) have established an inverse relationship between calcification and atmosphericCO2. From simple calculations of projected marine calcification rates, Zondervan etal . (2001) and Barker (2002) demonstrated that the inverse relationship betweencalcification and atmospheric CO2 resulted in a net negative feedback on increasingatmospheric CO2. The appearance of the ballast hypothesis has provided additionalcomplication to the question of marine calcification as a feedback on changes in atmo-spheric CO2 because it explicitly relates POC export fluxes to CaCO3 export fluxes.We have shown that the ballast effect may act to counter the negative feedback effectof reduced calcification and may constitute a net positive feedback depending on thechoice of the dissolution/remineralization rates of CaCO3 and POC.

We acknowledge discussions with and help from Wally Broecker, Victor Brovkin and AndyRidgwell, and support from the Comer Foundation, the Cambridge–MIT Institute, the NaturalEnvironment Research Council (GR3/1310 and GR3/JIF/05a) and the European Commission(EVR1-CT-40018: CESOP).

References

Archer, D. 1991 Modelling the calcite lysocline. J. Geophys. Res. 96, 17 037–17 050.Archer, D. & Maier-Reimer, E. 1994 Effect of deep-sea sedimentary calcite preservation on

atmospheric CO2 concentration. Nature 367, 260–263.Archer, D., Winguth, A., Lea, D. & Mahowald, N. 2000 What caused the glacial/interglacial

atmospheric pCO2 cycles? Rev. Geophys. 38, 159–189.Armstrong, R. A., Lee, C., Hedges, J. I., Honjo, S. & Wakeham, S. G. 2002 A new, mechanistic

model for organic carbon fluxes in the ocean based on the quantitative association of POCwith ballast minerals. Deep-Sea Res. II 49, 219–236.

Barker, S. 2002 Planktonic foraminiferal proxies of temperature and pCO2. PhD thesis, Univer-sity of Cambridge, UK.





Barker, S. & Elderfield, H. 2002 Foraminiferal calcification response to glacial–interglacialchanges in atmospheric CO2. Science 297, 833–836.

Barnola, J. M., Raynaud, D., Lorius, C. & Barkov, N. I. 1999 Historical CO2 record fromthe Vostok ice core. In Trends: a compendium of data on global change. Carbon DioxideInformation Analysis Center, Oak Ridge National Laboratory, US Department of Energy.

Bates, N. R., Michaels, A. F. & Knap, A. H. 1996 Alkalinity changes in the Sargasso Sea:geochemical evidence of calcification? Mar. Chem. 51, 347–358.

Bijma, J., Spero, H. & Lea, D. W. 1999 Reassessing foraminiferal stable isotope geochemistry:impact of the oceanic carbonate system (experimental results). In Uses of proxies in pale-oceanography: examples from the South Atlantic (ed. G. Fischer & G. Wefer), pp. 489–512.Springer.

Broecker, W. S. & Peng, T. H. 1987 The role of CaCO3 compensation in the glacial to interglacialatmospheric CO2 change. Global Biogeochem. Cy. 1, 15–39.

Broecker, W. S., Lao, Y., Klas, M., Clark, E., Bonani, G., Ivy, S. & Chen, C. 1993 A search foran Early Holocene CaCO3 preservation event. Paleoceanography 8, 333–339.

Broecker, W. S., Clark, E., McCorkle, D. C., Peng, T. H., Hajdas, I. & Bonani, G. 1999 Evidencefor a reduction in the carbonate ion content of the deep sea during the course of the Holocene.Paleoceanography 14, 744–752.

Broecker, W. S., Lynch-Stieglitz, J., Clark, E., Hajdas, I. & Bonani, G. 2001 What caused theatmosphere’s CO2 content to rise during the last 8000 years? Geochem. Geophys. Geosyst. 2,U1–U13.

Brovkin, V., Bendtsen, J., Claussen, M., Ganopolski, A., Kubatzki, C., Petoukhov, V. &Andreev, A. 2002 Carbon cycle, vegetation and climate dynamics in the Holocene: exper-iments with the CLIMBER-2 model. Global Biogeochem. Cy. 16, 86 (DOI: 10.1029/2001GB001662).

Chapman, M. R., Shackleton, N. J., Zhao, M. & Eglinton, G. 1996 Faunal and alkenone recon-structions of subtropical North Atlantic surface hydrography and paleotemperature over thelast 28 kyr. Paleoceanography 11, 343–357.

Crawford, D. W. & Purdie, D. A. 1997 Increase of pCO2 during blooms of Emiliania huxleyi :theoretical considerations on the asymmetry between acquisition of HCO−

3 and respiration offree CO2. Limnol. Oceanogr. 42, 365–372.

Emerson, S. & Bender, M. 1981 Carbon fluxes at the sediment-water interface of the deep-sea:calcium carbonate preservation. J. Mar. Res. 39, 139–162.

Etheridge, D. M., Steele, R. L., Langenfelds, R. J., Francey, J.-M., Barnola, J. M. & Morgan,V. I. 1998 Historical CO2 records from the Law Dome DE08, DE08-2, and DSS ice cores.In Trends: a compendium of data on global change. Carbon Dioxide Information AnalysisCenter, Oak Ridge National Laboratory, US Department of Energy.

Frankignoulle, M., Canon, C. & Gattuso, J. P. 1994 Marine calcification as a source of carbondioxide: positive feedback of increasing atmospheric CO2. Limnol. Oceanogr. 39, 458–462.

Gattuso, J. P., Pichon, M., Delesalle, B. & Frankignoulle, M. 1993 Community metabolism andair–sea CO2 fluxes in a coral-reef ecosystem (Moorea, French-Polynesia). Mar. Ecol. Prog.Ser. 96, 259–267.

Gattuso, J. P., Frankignoulle, M., Bourge, I., Romaine, S. & Buddemeier, R. W. 1998 Effectof calcium carbonate saturation of seawater on coral calcification. Global Planet. Change 18,37–46.

Holligan, P. M. (and 14 others) 1993 A biogeochemical study of the coccolithophore, emiliania-huxleyi, in the North Atlantic. Global Biogeochem. Cy. 7, 879–900.

Honjo, S. 1980 Material fluxes and modes of sedimentation in the mesopelagic and bathypelagiczones. J. Mar. Res. 38, 53–97.





Houghton, J. T., Meira Filho, L. G., Lee, H., Callander, B. A., Haites, E., Harris, N. & Maskell,K. (eds) 1995 Climate change 1994: radiative forcing of climate change and an evaluation ofthe IPCC IS92 emissions scenarios. Cambridge University Press.

Houghton, J. T., Ding, Y., Griggs, D. J., Noguer, M., van der Linden, P. J., Dai, X., Maskell,K. & Johnson, C. A. (eds) 2001 Climate change 2001: the scientific basis. Contribution ofWorking Group I to the Third Assessment Report of the International Panel on ClimateChange. Cambridge University Press.

Indermuhle, A. (and 11 others) 1999 Holocene carbon-cycle dynamics based on CO2 trapped inice at Taylor Dome, Antarctica. Nature 398, 121–126.

Ittekkot, V. & Haake, B. 1990 The terrestrial link in the removal of organic carbon. In Facets ofmodern biogeochemistry (ed. V. Ittekkot, S. Kempe, W. Michaelis & A. Spitzy), pp. 319–325.Springer.

Kawahata, H., Suzuki, A. & Goto, K. 1997 Coral reef ecosystems as a source of atmosphericCO2: evidence from pCO2 measurements of surface waters. Coral Reefs 16, 261–266.

Kayanne, H., Suzuki, A. & Saito, H. 1995 Diurnal changes in the partial pressure of carbondioxide in coral-reef water. Science 269, 214–216.

Keeling, C. D. & Whorf, T. P. 2002 Atmospheric CO2 records from sites in the SIO air samplingnetwork. In Trends: a compendium of data on global change. Carbon Dioxide InformationAnalysis Center, Oak Ridge National Laboratory, US Department of Energy.

Klaas, C. & Archer, D. 2002 Association of sinking organic matter with various types of mineralballast in the deep sea: implications for the rain ratio. Global Biogeochem. Cy. 16, 63. (DOI:10.1029/2001GB001765).

Kleypas, J. A., Buddemeier, R. W., Archer, D., Gattuso, J. P., Langdon, C. & Opdyke, B. N.1999 Geochemical consequences of increased atmospheric carbon dioxide on coral reefs. Sci-ence 284, 118–120.

Langdon, C. 2001 Review of experimental evidence for effects of CO2 on calcification of reefbuilders. In Proc. 9th Int. Coral Reef Symp. Bali, Indonesia (ed. M. K. Kasim Moosa, S.Soemodihardjo, A. Nontji, A. Soegiarto, K. Romimohtarto, Sukarno & Suharsono), vol. 2,pp. 1091–1098. Bali: The Ministry of Environment, The Indonesian Institute of Sciences andThe International Society for Reef Studies.

Langdon, C., Takahashi, T., Sweeney, C., Chipman, D., Goddard, J., Marubini, F., Aceves, H.,Barnett, H. & Atkinson, M. J. 2000 Effect of calcium carbonate saturation state on thecalcification rate of an experimental coral reef. Global Biogeochem. Cy. 14, 639–654.

Leclercq, N., Gattuso, J. P. & Jaubert, J. 2002 Primary production, respiration, and calcificationof a coral-reef mesocosm under increased CO2 partial pressure. Limnol. Oceanogr. 47, 558–564.

McCave, I. N. 1975 Vertical flux of particles in the ocean. Deep-Sea Res. 22, 491–502.Maier-Reimer, E. 1993 Geochemical cycles in an ocean general-circulation model: preindustrial

tracer distributions. Global Biogeochem. Cy. 7, 645–677.Martin, J. H., Knauer, G. A., Karl, D. M. & Broenkow, W. W. 1987 VERTEX: carbon cycling

in the northeast Pacific. Deep-Sea Res. 34, 267–285.Marubini, F., Barnett, H., Langdon, C. & Atkinson, M. J. 2001 Dependence of calcification on

light and carbonate ion concentration for the hermatypic coral Porites compressa. Mar. Ecol.Prog. Ser. 220, 153–162.

Milliman, J. D., Troy, P. J., Balch, W. M., Adams, A. K., Li, Y. H. & Mackenzie, F. T. 1999Biologically mediated dissolution of calcium carbonate above the chemical lysocline? Deep-SeaRes. I 46, 1653–1669.

Mitchell, J. F. B. 1990 Is the Holocene a good analogue for greenhouse warming? J. Clim. 3,1177–1192.





Murata, A. & Takizawa, T. 2002 Impact of a coccolithophorid bloom on the CO2 system insurface waters of the eastern Bering Sea shelf. Geophys. Res. Lett. 29, 42 (DOI: 10.1029/2001GL013906).

Ohde, S. & van Woesik, R. 1999 Carbon dioxide flux and metabolic processes of a coral reef,Okinawa. Bull. Mar. Sci. 65, 559–576.

Redfield, A. C., Ketchum, B. H. & Richards, F. A. 1963 The influence of organisms on thecomposition of seawater. In The sea (ed. M. N. Hill), vol. 2, pp. 26–77. Wiley.

Riebesell, U., Zondervan, I., Rost, B., Tortell, P. D., Zeebe, R. E. & Morel, F. M. M. 2000Reduced calcification of marine plankton in response to increased atmospheric CO2. Nature407, 364–367.

Sanyal, A., Hemming, N. G., Hanson, G. N. & Broecker, W. S. 1995 Evidence for a higher pHin the Glacial ocean from boron isotopes in foraminifera. Nature 373, 234–236.

Sarmiento, J. L., Slater, R. D., Fasham, M. J. R., Ducklow, H. W., Toggweiler, J. R. & Evans,G. T. 1993 A seasonal 3-dimensional ecosystem model of nitrogen cycling in the North-Atlantic euphotic zone. Global Biogeochem. Cy. 7, 417–450.

Smayda, T. J. 1970 Normal and accelerated sinking of phytoplankton in the sea. Oceanogr. Mar.Biol. 8, 353–414.

Spero, H. J., Bijma, J., Lea, D. W. & Bemis, B. E. 1997 Effect of seawater carbonate concen-tration on foraminiferal carbon and oxygen isotopes. Nature 390, 497–500.

Takahashi, T. (and 11 others) 2002 Global sea–air CO2 flux based on climatological surface oceanpCO2, and seasonal biological and temperature effects. Deep-Sea Res. II 49, 1601–1622.

Toggweiler, J. R. 1999 Variation of atmospheric CO2 by ventilation of the ocean’s deepest water.Paleoceanography 14, 571–588.

Ware, J. R., Smith, S. V. & Reakakudla, M. L. 1992 Coral reefs: sources or sinks of atmosphericCO2? Coral Reefs 11, 127–130.

Westbroek, P. (and 13 others) 1993 A model system approach to biological climate forcing: theexample of Emiliania huxleyi. Global Planet. Change 8, 27–46.

Zeebe, R. E. & Wolf-Gladrow, D. 2001 CO2 in seawater: equilibrium, kinetics, isotopes. ElsevierOceanography Series. Elsevier.

Zhong, S. J. & Mucci, A. 1989 Calcite and aragonite precipitation from seawater solutions ofvarious salinities: precipitation rates and overgrowth compositions. Chem. Geol. 78, 283–299.

Zondervan, I., Zeebe, R. E., Rost, B. & Riebesell, U. 2001 Decreasing marine biogenic calcifica-tion: a negative feedback on rising atmospheric pCO2. Global Biogeochem. Cy. 15, 507–516.

Zuddas, P. & Mucci, A. 1998 Kinetics of calcite precipitation from seawater. II. The influenceof the ionic strength. Geochim. Cosmochim. Acta 62, 757–766.

Discussion

I. N. McCave (Department of Earth Sciences, University of Cambridge, UK ). Haveyou included, or can you include, in your box model the consequences of glacial–interglacial and stadial–interstadial carbon-system variations for δ13C in bottom seawater (and benthic foraminifera)?

H. Elderfield. The consequences have not been included as we were concernedwith the response only to future CO2 change. This is straightforward to do, althoughin the context of the approach taken we would have to decide whether the oceanconfiguration in the model is appropriate, for example, the depth and vigour ofNADW production, for past times.

J. G. Shepherd (School of Ocean & Earth Science, University of Southampton,UK ). How constant do you think the rain ratio may be?





H. Elderfield. The implication of the ballast model is that the rain ratio may bemore constant than has been surmised. It is important, though, to make it perfectlyclear what is meant by this term. It was used by Archer & Maier-Reimer explicitlyto refer to the ratio of inorganic to organic carbon at the sea bed (disregardingnepheloid layer resuspension), because this ratio affects sediment respiration andhas the potential to separate the depths of the lysocline and saturation horizon.However, it has also been used to refer to the ratio of the export flux and the ratioof the production flux, and it is perhaps here that misunderstanding occurs. Clearly,there are large regional (and probably temporal) differences in production and exportratios (the diatom/phaeocystis ratio in the Southern Ocean, for example) and thesewill affect atmospheric CO2 on 102-year time-scales. Interaction with sediments leadsto carbonate compensation, which also changes CO2 but on 103-year time-scales.






Documents

The future of the carbon cycle: review, calciﬁcation ...carboncycle.princeton.edu/sites/default/files/pdfs/BarkerEtal_RSTA... · 22.07.2003 · The future of the carbon cycle: