The Global Carbon Cycle: A Test of Our Knowledge of Earth as a System
Author(s): P. Falkowski, R. J. Scholes, E. Boyle, J. Canadell, D. Canfield, J. Elser, N. Gruber,
K. Hibbard, P. Högberg, S. Linder, F. T. Mackenzie, B. Moore III, T. Pedersen, Y. Rosenthal, S.
Seitzinger, V. Smetacek, W. Steffen
Source: Science, New Series, Vol. 290, No. 5490 (Oct. 13, 2000), pp. 291-296
Published by: American Association for the Advancement of Science
Stable URL: http://www.jstor.org/stable/3078125
Accessed: 08/01/2010 16:49
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.
Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at
http://www.jstor.org/action/showPublisher?publisherCode=aaas.
Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact [email protected].
American Association for the Advancement of Science is collaborating with JSTOR to digitize, preserve and
extend access to Science.
http://www.jstor.org
REVIEW:
CLIMATE
The
CHANGE
Global
Carbon
Knowledge of
Cycle:
Earth
A
as
Test
a
of
Our
System
P. Falkowski,'*t R. J. Scholes,2* E. Boyle,3t J. Canadell,4t D. Canfield,5t J. Elser,6t N. Gruber,:t K. Hibbard,8:
P. Hogberg,9t S. Linder,'?t F. T. Mackenzie,"1 B. Moore 111,8:T. Pedersen,'2t Y. Rosenthal,1
S. Seitzinger,': V. Smetacek,'13 W. Steffen'4:
over the past 420,000 years, the climate system has operated within a relatively conMotivated by the rapid increase in atmospheric CO2due to human activities since the
strained domain of atmospheric CO2 and
IndustrialRevolution,several international scientific research programshave analyzed
the role of individualcomponents of the Earthsystem in the global carbon cycle. Our
temperature(7, 8) (Fig. 1). In the CO2-temknowledge of the carbon cycle within the oceans, terrestrial ecosystems, and the
peraturephase space that characterizedthe
atmosphere is sufficiently extensive to permit us to conclude that although natural
preindustrial world, CO2 oscillated in
processes can potentially slow the rate of increase in atmospheric COz, there is no
100,000-year cycles by approximately 100
natural "savior"waiting to assimilate all the anthropogenically produced CO2 in the
partsper million by volume (ppmv), between
coming century. Our knowledge is insufficient to describe the interactions between
about 180 and 280 ppmv. On millennial time
the components of the Earthsystem and the relationship between the carbon cycle
scales, changes in CO2 recordedin ice cores
and other biogeochemical and climatological processes. Overcoming this limitation
are highly correlatedwith changes in temperrequires a systems approach.
ature (9). Although high-resolution analysis
of ice cores suggests that there are periods in
Earth'shistory when temperaturecan change
ver the past 200 years, human activ- generations. Given present trends in energy relatively sharply without a discernible
ities have altered the global carbon demands,ample fossil fuel reserves, a lack of
change in CO2 (7), the converse does not
alternative
concerted,
global,
cycle significantly. Understanding
energy produc- appearto be true.
the consequences of these activities in the tion strategies,andprojectionsof humanpopComparison of the present atmospheric
coming decades is critical for formulating ulation growth, atmosphericCO2 concentra- concentrationof CO2with the ice core record
tions appearfated to increase throughoutthe reveals that we have left the domain that
economic, energy, technology, trade, and security policies that will affect civilization for coming century (1, 2). The rate of change in defined the Earth system for the 420,000
atmosphericCO2depends,however, not only years before the IndustrialRevolution (10)
'Institute of Marineand Coastal Sciences, Rutgers on humanactivities but also on biogeochemi(Fig. 1). Atmospheric CO2 concentrationis
University, 71 Dudley Road, New Brunswick,NJ cal and climatological processes and their now nearly 100 ppmv higher, and has risen to
08901, USA.2Councilof Scientificand IndustrialReinteractionswith the carbon cycle. Here we
thatlevel at a rateat least 10 andpossibly 100
search,Environmental
Division,Post Office Box 395,
examine some of the changes in biogeo- times fasterthan at any other time in the past
Pretoria0001, SouthAfrica.3Earth,Atmospheric,and
PlanetaryScience Department,MassachusettsInsti- chemical and climatological processes con420,000 years. We have driven the Earth
tute of Technology,42 CarletonStreet, MailCode: comitant with alterations in the carbon and
system from the tightly bounded domain of
MA
USA.
4Global
E34-258,Cambridge, 02142-1324,
in
nutrient
the
world,
cycles
contemporary
glacial-interglacial
dynamics. Are we in a
Change and TerrestrialEcosystems International
ProjectOffice, CommonwealthScientificand Indus- and comparethese processes with our under- transitionperiod to a new, stable domain? If
trialResearchOrganisationWildlifeand Ecology,Post
so, what are the main forcing factors and
standing of the preceding 420,000 years of
OfficeBox284, Canberra,
AustralianCapitalTerritory, Earth's
feedbacksof this transition?What will be the
history.
2601, Australia.Slnstitute of Biological Sciences,
climatological features of a new domain?
Odense University,5230 Odense,Denmark.6DepartWhat will be the responses and feedbacks of
ment of Zoology,ArizonaState University,Temple, Entering Uncharted Waters
AZ 85287-1501, USA.7lnstituteof Geophysicsand Under the auspices of the InternationalGeoEarth's ecosystems? How and when can and
PlanetaryPhysics and Departmentof Atmospheric sphere-BiosphereProgramme(IGBP), severshould we returnto the preindustrialdomain?
Sciences,5853 SlichterHall,Universityof California, al
scientific
studies
have
The active carbon reservoirs and their
international
large
LosAngeles,CA90095-4996, USA.8Institutefor the
focused on elucidatingvarious aspects of the strengths. Atmospheric CO2 exchanges rapStudy of Earth,Oceans, Space, Universityof New
global carboncycle over the past decade (3).
idly with oceans and terrestrialecosystems
Hampshire,MorseHall,39 CollegeRoad,Durham,NH
03824, USA.9Departmentof ForestEcology,Swedish These programshave helped addresstwo ma(11). The ratio between the rate at which
Universityof AgriculturalSciences,S-901 83 Umea,
recurrentquestions in the currentdebate these two reservoirsabsorbatmosphericCO2
Sweden.'Department for ProductionEcology,Swed- jor
about
andthe rateof emissions determinesthe overglobal change: Can we distinguishbeish Universityof Agricultural
Sciences,PostOfficeBox
tween anthropogenicperturbationsand natu- all rate of change of atmosphericCO2. The
7042, S750 07 Uppsala,Sweden. "Department of
sink strengthof the reservoirsdeterminesthe
Oceanography,SOEST,Universityof Hawaii,Honolu- ral variability in biogeochemical cycles and
lu, HI,96822, USA.'2Departmentof Earthand Ocean climate? And what is the
of
to absorb excess or anthropogenic
sensitivity
capacity
Sciences,Universityof BritishColumbia,Vancouver,
Earth's climate to changes in atmospheric CO2. During glacial-interglacialtransitions,
BCV6T 1Z4, Canada.'3AlfredWegenerInstitutefor
CO2? We consider the two questions in the for example, the atmosphereacts to transfer
Polar and MarineResearch,Am Handelshafen12,
27570 Bremerhaven,Germany.'41GBPSecretariat, context of the relatively recent geological
carbonbetween terrestrialecosystems andthe
RoyalSwedishAcademyof Sciences,Box 50005 Lilia history of Earth, for which we have robust oceans. The remarkableconsistency of the
Frescativagen4, S-10405 Stockholm,Sweden.
paleoclimatologicalproxies.
upperand lower limits of the glacial-intergla*Co-chairsof the InternationalGeosphere-Biosphere
Arrheniusrecognized over 100 years ago
cial atmosphericCO2 concentrations,and the
Programme(IGBP)WorkingGroupand lead authors.
that atmosphericCO2plays a criticalrole apparentfine control over periods of many
tTo whom correspondenceshould be addressed.E- (4)
in regulating Earth's temperature (5, 6).
thousands of years aroundthose limits, sugmail:[email protected]
IMembersof the IGBPWorkingGroup.
Analyses of ice cores strongly suggest that gest strong feedbacks that constrainthe sink
www.sciencemag.org SCIENCE VOL290
13 OCTOBER
2000
291
SCIENCE'S
strengths in both the oceans and terrestrial
ecosystems. The relatively rapid transition
from glacial to interglacial states and the
initially steep, but eventually gradual,transition into glacial periods (8) suggests that the
ratesof absorptionand emission of CO2from
the oceans and terrestrial ecosystems are
asymmetrical.It should be noted thatbecause
of this asymmetry, the average atmospheric
CO2 concentrationduring the past 420,000
years was only -220 ppmv, not 280 ppmv as
usually ascribed (12).
How is atmospheric CO2 regulated? The
total of dissolved inorganic carbon in the
oceans is 50 times that of the atmosphere
(Table 1), and on time scales of millennia,
the oceans determineatmosphericCO2 concentrations,not vice versa. AtmosphericCO2
continuouslyexchanges with oceanic CO2 at
the surface.This exchange, which amountsto
-90 gigatons (Gt) of carbonper year in each
direction, leads to rapid equilibrationof the
atmospherewith the surfacewater.Upon dissolution in water,CO2forms a weak acid that
reacts with carbonate anions and water to
form bicarbonate.The capacityof the oceanic
carbonate system to buffer changes in CO2
concentration is finite and depends on the
addition of cations from the relatively slow
weatheringof rocks. Because the rate of anthropogenicCO2 emissions is several orders
of magnitudegreaterthan the supply of mineral cations, on time scales of millennia the.
ability of the surface oceans to absorb CO2
will inevitably decrease as the atmospheric
concentrationof the gas increases (13).
The concentrationof total dissolved inor-
COMPASS
ganic carbonin the ocean increasesmarkedly
below aboutthe upper300 m, whereit remains
significantly above the surface ocean-atmosphere equilibriumvalue in all ocean basins.
The higherconcentrationof inorganiccarbonin
the ocean interiorresultsfroma combinationof
two fundamental processes: the "solubility
pump"and "biologicalpumps"(14, 15).
The efficiency of the solubility pump depends on the thermohalinecirculationand on
latitudinal and seasonal changes in ocean
ventilation (16, 17). CO2 is more soluble in
cold, saline waters, and sequestrationof atmospheric CO2 in the ocean interioris therefore controlled by the formation of cold,
dense water masses at high latitudes, especially in the North Atlantic and in the Southern Ocean confluence. As these water masses
sink into the ocean interiorand are transported laterally, CO2 is effectively prevented
from re-equilibratingwith the atmosphereby
a cap of lighter overlying waters. Re-equilibration occurs only when waters from the
ocean interiorare broughtback to the surface,
decades to several hundredsof years later.
Coupled climate-ocean simulations (6,
18) suggest that CO2-inducedglobal warming will lead to increasedstratificationof the
water column. If this occurs, the transportof
carbon from the upper ocean to the deep
ocean will be reduced, with a resulting decrease in the rate of sequestrationof anthropogenic carbon in the ocean (19, 20). The
combined effects of progressive saturationof
the buffering capacity and increased stratification will weaken two importantnegative
feedbacks in the carbon-climate system,
Modern
350-
E 3000
=.
Qs
.O 250-
200
200-
13.U
-10
.
.
.
.
I
.
.I
-5
.I
I.
.
I ??
0
Deuterium-based TemperatureAnomalies, ?C
Fig. 1. A correlationbetween atmosphericpartial pressureof CO2 (pCO2) and isotopic (8D)
temperatureanomaliesas recordedin the Vostokice core.Thefigureshows that climatevariations
in the past 420,000 years operatedwithin a relativelyconstraineddomain.Data are from (8).
292
13 OCTOBER
2000 VOL290
therebyreducingthe rateof oceanic uptakeof
anthropogenicCO2. The magnitude of these
feedbacks is critically dependent on how
ocean circulationand mixing will respondto
the climatic forcing.
Biological processes also contributeto the
absorptionof atmosphericCO2 in the ocean.
Phytoplanktonphotosynthesislowers the partial pressure of CO2 in the upper ocean and
therebypromotesthe absorptionof CO2from
the atmosphere. Approximately 25% of the
carbonfixed in the upperocean sinks into the
interior(21, 22), where it is oxidized through
heterotrophicrespiration,raising the concentration of dissolved inorganic carbon (DIC).
The export of organic carbon from the surface to the ocean interiorpresently accounts
for - 11 to 16 Gt of carbonper year (23). This
process keeps atmospheric CO2 concentrations 150 to 200 ppmv lower thanthey would
be if all the phytoplanktonin the ocean were
to die (23, 24). In addition to the organic
biological pump, several phytoplanktonand
zooplankton species form CaCO3 shells that
sink into the interior of the ocean, where
some fraction dissolves. This inorganic carbon cycle leads to a reduction in surface
ocean DIC relative to the deep ocean and is
therefore sometimes called the "carbonate
pump." The process of precipitatingcarbonates, however, increases the partial pressure
of CO2 (25). Hence, on time scales of centuries, while the carbonatepump lowers DIC
concentrationsin the upperocean, it simultaneously leads to the evasion of CO2 from the
ocean to the atmosphere.
Coupled climate-biogeochemical models
suggest that the biological pumps tend to
counteractthe decrease in uptake caused by
the solubility pump (20, 26). If the biological
pumps are to absorb anthropogenicCO2 in
the coming century,their efficiency must increase. In principle,this can be accomplished
by any or all of four processes: (i) enhancing
utilization of excess nutrients in the upper
ocean, (ii) adding one or more nutrientsthat
limit primary production, (iii) changing the
elemental ratios of the organic matter in the
ocean, and (iv) increasingthe organiccarbon/
calcite ratio in the sinking flux (27). There
are significant gaps in our knowledge that
limit our ability to predict the magnitude of
changes in oceanic uptake, but the likely
changes in the biological pump are too small
to counteractthe projectedCO2 emissions in
the coming century. Almost certainly, however, changes in oceanic ecosystem structure
will accompany changes in physical circulation (and hence changes in nutrientsupplies),
lowered pH, and changes in the hydrological
cycle. Our present knowledge of the factors
thatdeterminethe abundanceand distribution
of key groups of marine organisms is so
limited that it is unlikely we will be able to
predict such changes within the next decade
SCIENCEwww.sciencemag.org
SCIENCE'S
with reasonable certainty (28). These uncertainties affect our ability to predict specific
responses, but not the sign of the changes in
atmosphericCO2or the impactof this change
on upperocean pH. If our currentunderstanding of the ocean carboncycle is borneout, the
sink strengthof the oceans will weaken, leaving a largerfractionof anthropogenicallyproduced CO2 in the atmosphere or to be absorbed by terrestrialecosystems.
Terrestrialecosystems also exchange CO2
rapidlywith the atmosphere,but unlike in the
oceans, there is no physicochemical pump.
CO2 is removed from the atmospherethrough
photosynthesis and stored in organic matter.
It is returnedto the atmospherevia a number
of respiratorypathways that operate on various time scales: (i) autotrophicrespirationby
the plants themselves; (ii) heterotrophicrespiration,in which plant-derivedorganic matter is oxidized primarily by soil microbes;
and (iii) disturbances,such as fire, in which
large amountsof organic matterare oxidized
in very short periods of time.
On a global basis, terrestrialcarbon storage primarilyoccurs in forests (29). The sum
of carbon in living terrestrialbiomass and
soils is approximately three times greater
thanthe CO2in the atmosphere(Table 1), but
the turnovertime of terrestrialcarbon is on
the orderof decades. Direct determinationof
changes in terrestrial carbon storage has
proven extremely difficult (30). Rather, the
contributionof terrestrialecosystems to carbon storage is inferred from changes in the
concentrations of atmospheric gases, especially CO2 and 02, their isotopic composition, inventories of land use change, and
models (31-33). The models requireaccurate
knowledge of the oceanic uptakeof CO2 (31,
34, 35).
Terrestrialnet primaryproduction (NPP)
(36) is not saturatedby present atmospheric
CO2 concentrations (37). Consequently, as
atmosphericCO2 increases, terrestrialplants
are a potential sink for anthropogeniccarbon.
The principalcarbon-fixingenzyme in plants
is ribulose 1,5-bisphosphatecarboxylase/oxygenase (rubisco) (38). In C3 plants, the activity of rubisco increases with increasing
CO2 concentrations,saturatingbetween 800
and 1000 ppmv CO2, a concentration that
will probably be reached early in the next
century at the present emissions rate (2). Because the saturation function decreases as
CO2 increases, terrestrialplants will become
less of a sink for CO2 in coming decades.
Some experimental evidence suggests that
because of nutrientlimitation (39), NPP may
level off at only 10 to 20% above current
rates, at an atmosphericCO2concentrationof
550 to 650 ppmv, or double preindustrial
concentrations (40). Furthermore,increased
temperaturewill probably lead to higher microbial heterotrophicrespiration,which may
COMPASS
counteractand even exceed the enhancement to about 5% of the present anthropogenic
of NPP (41). The combined effects of higher CO emissions.
There is no evidence that phosphorussigCO2concentrations,highertemperatures,and
changes in disturbanceand soil moisture re- nificantly limits primaryproductionin coastgimes lead to considerableuncertaintyabout al or open oceans on a global scale (51) or
the ability of terrestrialecosystems to miti- thatphosphorusloading of the coastal oceans
gate againstrising CO2in the coming decades has significantly altered global primarypro(42). However, recent results from long-term duction (46).
soil warming experiments in a boreal forest
Iron is a micronutrientthat limits both
contradictthe idea that the projected rise in primaryproduction(52) and nitrogenfixation
temperatureis likely to lead to forests that are in many areas of the ocean (53). Eolian
now carbon sinks becoming carbon sources (windbome) iron fluxes, a principalsource of
in the foreseeable future (43).
iron inputto the open ocean, are coupled both
to land use and the hydrological cycle (54).
Again, as in the case of marine ecosystems, we can predict that the negative feed- Episodic aridity affects eolian iron supplies,
back afforded by terrestrial ecosystems in and in the coming decades, iron fluxes to the
removing anthropogenic CO2 from atmo- ocean could therefore increase because of
sphere will continue; however, the sink increasedevaporationof soil moisture or destrength will almost certainly weaken. The crease because of increased precipitation
exact magnitude of the change in sink (55). Increases in evaporationand increases
in precipitationare expected in differentparts
strengthremains unclear.
Interactionof the carbon cycle with other of the land surface in response to increasing
biogeochemical cycles. All biotic sinks for global temperatures.Thus, although increasCO2 require other nutrients in addition to
ing temperatureand its potentialinfluence on
carbon.Humanshave affected virtuallyevery the availabilityof iron in the open ocean will
major biogeochemical cycle (Table 2), but affect the biological uptake of carbon in the
the effects of these impacts on the interac- ocean, at present we do not know the sign of
tions between these elemental cycles are the changes. However, even if iron fluxes
were to increase to such an extent that ocepoorly understood (44). The production of
synthetic fertilizers, the cultivation of nitro- anic nitrogen fixation were stimulatedto the
maximum, the maximum change in atmogen-fixing crops, and the deposition of fossil
fuel-associated nitrogen are collectively of
spheric CO2that could ensue would be about
the same order of magnitude as naturalbio- 40 ppmv. Although not insignificant,such an
logical nitrogen fixation (45). These inputs effect is unrealistic on time scales of centuwill continue to rise with the projected in- ries (56).
crease in human population (46). Similarly,
Eolian nitrogeninputscan also potentially
there has been an approximatelyfourfold in- enhanceboth terrestrialand marineuptakeof
crease in phosphorusinputs to the biosphere, anthropogenicCO2 (37). In terrestrialecoprimarilydue to mining of phosphoruscomsystems, the sink strengthresulting from eolian nitrogen deposition depends on the carpounds for fertilizer.
At first glance, one might conclude that bon: nitrogenratio of the stored organic matsimultaneous increases in nitrogen fixation
and phosphate production would stimulate Table 1. Carbonpools in the major reservoirson
the biological sequestrationof carbon in ter- Earth.
restrial and marine ecosystems. Will such
stimulation provide salvation from the conQuantity
Pools
(Gt)
tinued anthropogenicemissions of CO2to the
atmosphere?
720
Atmosphere
It is estimated that by 2050, the total
Oceans
38,400
transport of fixed inorganic nitrogen from
Total inorganic
37,400
land to the coastal zone will have increased
Surface layer
670
fromthe presentvalue of -20 teragrams(Tg)
36,730
Deep layer
Total organic
of nitrogen to - 40 Tg of nitrogen per year
1,000
(47), concomitantwith an increase in human Lithosphere
>60,000,000
population from 6 to 9 or 10 billion. AlSedimentarycarbonates
15,000,000
Kerogens
though nutrientloading has resultedin coastal eutrophicationon a global scale (48), deni- Terrestrialbiosphere (total)
2,000
trification presently removes virtually all
600-1,000
Livingbiomass
Dead biomass
1,200
land-derivednitrogen before it can reach the
ocean
Coastal
denitrification
1-2
(49, 50).
open
Aquatic biosphere
thus effectively decouples the terrestrialand Fossil fuels
4,130
Coal
3,510
oceanic nitrogencycles. However, even if no
Oil
230
denitrificationoccurred,the increasedflux of
Gas
140
land-derived nitrogen would sequester only
Other (peat)
250
0.4 Gt of carbonper year (48), corresponding
www.sciencemag.org
SCIENCE VOL 290
13 OCTOBER2000
293
SCIENCE'S
ter and the degree of nitrogen saturationof
soils (44, 57-59). Nitrogen deposition is predicted to continue and has the potential to
enhance the carbon sink in nitrogen-limited
ecosystems, but it will probablybecome decreasingly effective at doing so. Futurenitrogen deposition will largely occur on already
nitrogen-saturatedsoils, such as in the forests
of WesternEurope,China, and India, and on
agriculturallands in the tropics,whose capacity to sequester carbon is intrinsically small
COMPASS
and where soils are mostly limited by phosphate (57). In the context of the global carbon
cycle, the eolian input of nitrogen to marine
ecosystems is essentially irrelevant(59).
In addition to ecophysiological considerations, land use change plays a major role in
the carbon source/sink dynamics. The increased pressure in the developing world to
increase food and fiber production by converting forests to agriculturaluse effectively
increasesthe flux of carbonto the atmosphere
while simultaneouslyreducing the land area
available for active sinks. Abandonmentof
agricultural land and regrowth of forests,
largely in the temperate Northern Hemisphere, may be a significant terrestrialCO2
sink at present (34) but cannot be sustained
indefinitely.This sink can buy some time, but
unless CO2 emissions are reduced, it cannot
mitigate against continued accumulation of
the gas in Earth'satmospheregiven projected
emission scenarios.
The Need for an Integrated Systems
Table 2. Examplesof humaninterventionin the globalbiogeochemicalcycles of carbon,nitrogen, Approach
sulfur,water,andsediments.Dataarefor the mid-1900s.
phosphorus,
The global carboncycle is affected by human
activities and is coupled to other climatologof flux(millions
Magnitude
ical and biogeochemical processes. As disof metrictons peryear)
%changedue to
Element
Flux
humanactivities cussed above, we have considerableinformation about specific aspects of the carbon cyNatural Anthropogenic
cle, but many of the couplings and feedbacks
C
anddecayCO2
Terrestrial
61,000
respiration
are poorly understood. As we drift further
+13
Fossilfuel and landuse CO2
8,000
away from the domain that characterizedthe
130
N
Naturalbiologicalfixation
preindustrialEarth system, we severely test
140
+108
Fixationowingto ricecultivation,
the limits of our understandingof how the
combustionof fossilfuels,and
Earthsystem will respond.
of fertilizer
production
A look at the current understandingof
Chemicalweathering
3
P
CO2 changes illustrates
glacial-interglacial
12
+400
Mining
the problem.Perhapssurprisingly,there is no
80
S
Naturalemissionsto atmosphereat
consensus on the causes of these changes.
Earth'ssurface
There are at least 11 hypotheses (53, 60, 61),
90
+113
Fossilfuel andbiomassburning
which may be grouped into three basic
emissions
themes: (i) physical/chemical "reorganiza111 x 1012
overland
O and H Precipitation
tion" of the oceans, (ii) changes in the ocean
+16
18 x 1012
(as H20) Globalwaterusage
carbonatesystem, and (iii) changes in ocean
river
1 x 1010
SedimentsLong-term
preindustrial
nutrientinventories.Many of these hypothesuspendedload
ses are not mutually exclusive. The interac2 x 1010
+200
Modernriversuspendedload
tions between marine and terrestrialecosystems, changes in ocean circulation,radiative
forcing, and greenhouse gases all probably
.400
interactin a specific sequence to give rise to
the natural cyclic atmospheric and climatic
oscillations. These interactionsare not pres105 ently represented in detailed models of the
glacial-interglacialtransitions.
104 This example illustrates three points.
First, in the recent history of Earth,the carr'
iu
bon cycle did not operate in a vacuum and
I
I
I
I,DI
103 .........
50
--150
-400
-300
-250
-200
510 0
-4500
-350
Q.
was not constrainedto a specific reservoir.
Glacial-interglacial
Naturalchanges in the inventoriesof carbon,
a 102as inferredfrom the ice core records of glacial-interglacialtransitions,are linked to other biogeochemical and climatological pro< 100 cesses. Those linkages continue to the
present, but the quantitativeimpacts in the
coming centuryare obscuredby simultaneous
10-1alterations of numerous biogeochemical cycles through human activities. Second, the
10-2
,
scientific community has generally ap109 108 107 106 105 104 103 102 101 100 10-1 10-2 10-3
proached problems such as glacial-interglacial transitions from a disciplinary perspecPeriod [years]
tive. This approachhas not produced comFig.2. Schematicvariancespectrumfor CO2overthe courseof Earth'shistory.Note the impactof pletely satisfactory explanations for what is
humanperturbations
on the decade-to-centuryscale. (Inset) Changesin atmosphericCO2overthe
natural perturbationin the
a
past 420,000 years as recordedin the Vostokice, showingthat both the rapidrate of changeand clearly large
the increasein CO2concentrationsince the IndustrialRevolutionare unprecedentedin recent global carboncycle. Because of the disciplingeologicalhistory.
ary nature of research, interactionsbetween
.
..
294
-
13 OCTOBER
2000 VOL290
.,
SCIENCEwww.sciencemag.org
SCIENCE'S
components of the Earth system are not incorporated into present biogeochemical or
climate models. When changes in isolated
processes are considered, we usually understand the signs of feedbacks, if not the magnitudes of the responses. It is when processes
interactthat we have significant problems in
reproducing the phenomena quantitatively.
Clearly, a systems approachis needed. Third,
reconstructionsof the carbon cycle (for example, during glacial-interglacialtransitions)
provide testable hypotheses about the Earth
system. Consensus on how a 100-ppmv
change in atmosphericCO2 can occur naturally within a 100,000-year time frame (62)
would imply some understandingof the feedbacks within the Earthsystem. Knowledge of
these feedbacks does not give us predictive
capability for the coming decades or centuries, but it can help us develop the modeling
tools needed to integrate the detailed information gathered from component studies of
the contemporaryworld.
Our analysis above shows that although
naturalsinks can potentially slow the rate of
increase in atmosphericCO2, there is no natural savior waiting to assimilate all the anthropogenic CO2 in the coming century. Although on geological time scales the anthropogenic emission of CO2 is a transientphenomenon (Fig. 2), it will affect Earth's
biogeochemical cycles for hundredsof years
to come (20, 63). Ourpresentimperfectmodels suggest thatthe feedbacksbetween carbon
and other biogeochemical and climatological
processes will lead to weakened sink
strengths in the foreseeable future, and the
prospects of retrieving anthropogenic CO2
from the atmosphere by enhancing natural
sinks are small. This condition cannot persist
indefinitely. Potential remediationstrategies,
such as the purposeful manipulationof biological and chemical processes to accelerate
the sequestration of atmospheric CO2, are
being seriously considered by both governmental bodies and private enterprises.These
mitigationstrategieswill themselves have unknown consequences and must be carefully
assessed within the context of an integrated
systems approachbefore any action is taken.
As we rapidly enter a new Earth system
domain, the "Anthropocene"Era (64), the
debate about distinguishing human effects
from naturalvariabilitywill inevitably abate
in the face of increased understanding of
climate and biogeochemical cycles. Our
present state of uncertainty arises largely
from lack of integrationof information.Nevertheless, scientists' abilities to predict the
future will always have a component of uncertainty.This uncertaintyshould not be confused with lack of knowledge nor should it be
used as an excuse to postpone prudentpolicy
decisions based on the best informationavailable at the time.
COMPASS
References and Notes
1. M. I. Hoffert et al., Nature 395, 881 (1998).
2. J. T. Houghton, G. J. Jenkins, J. J. Ephraums,Eds.,
Climate Change: The IPCC Scientific Assessment
(CambridgeUniv. Press, Cambridge,1996).
3. The International Geosphere-Biosphere Programme
(IGBP)is an umbrella organization that coordinates
(but does not fund) large multinationalresearchprograms. Several of these programs have focused on
the carbon cycle in specific reservoirs:The flagship
programfor terrestrialecosystems is Global Change
and TerrestrialEcosystems (GCTE);for the oceans, it
is the Joint GlobalOcean FluxStudy (JGOFS);for the
atmosphere, it is InternationalGlobal Atmospheric
Chemistry(IGAC);and for paleochemistryand paleoclimate, it is Past Global Changes (PAGES).
4. S. Arrhenius,Philos. Mag. J. Sci. (London,Edinburgh,
Dublin) 41, 237 (1896).
5. M. I. Budyko, The Earth'sClimate: Past and Future
(Academic Press, New York,1982).
6. J. F. B. Mitchell,T. C.Johns,J. M. Gregory,S. F. B.Tett,
Science 376, 501 (1995).
7. H. J. Smith, H. Fischer,M. Wahlen, D. Mastroianni,B.
Deck, Nature 400, 248 (1999).
8. J. R. Petit et al., Nature 399, 429 (1999).
9. The reconstruction of paleotemperatures from the
ice cores is based on the H/D and 18/160 fractionation in the precipitation. Necessarily, the inferred
temperature is that of the upper troposphere, not
that at ground level. Moreover,the inferredtemperature is not a global mean but rathera value for the
region, in this case, Vostok in Antarctica. Despite
these caveats, the glacial-interglacial temperature
trends and anomalies are highly consistent on time
scales of 100,000 years for the four cycles obtained
from the ice core.
10. T. Crowley,Science 289, 270 (2000).
11. On time scales of millions of years, vulcanism and
weathering reactions are critical determinants of atmospheric CO2. The weathering reactions can be
summarizedby the following:
weathering
COz+ CaSiO3? CaCO3+ SiOz
12.
13.
14.
15.
16.
17.
18.
metamorphosis
where Mg can substitute for Ca. CO2 is resuppliedto
the atmosphere by vulcanism [J. F. Kasting, O. B.
Toon, J. B. Pollack,Sci. Am. 258, 90 (1988)]. On time
scales of decades, these processes are relatively insignificant in determining atmospheric CO2 as compared with the exchanges of CO2 between oceanic
and terrestrialecosystems.
The preindustrialconcentration of CO2 was -280
ppmv throughout the Holocene. The lower mean
value is due to the fact that the glacial periods (with
low CO2concentrations)were about four times longer than the interglacialperiods. In fact, most of the
recent history of Earth is dominated by a glacial
climate, with relatively short punctuations of interglacial phases. If the low CO2levels of glacial periods
had persisted through the Holocene, plants that provided food and fiber for early human civilizationmay
not have been cultivatable [R.F. Sage, Global Change
Biology 1, 93 (1995)].
Given projectedrates of increase in atmosphericCO2,
the pH of the surface waters will decrease by -0.2
units within this century. This decrease will significantly hinder the biological precipitationof calcium
carbonates [J. Klyepas et al., Science 284, 118
(1999); C. Langdonetal., GlobalGeochem. Cycles 14,
639 (2000)].
T. Volk and M. I. Hoffert, in The Carbon Cycle and
Atmospheric COz: Natural Variations Archean to
Present, E.T. Sunquistand W. S. Broeker,Eds.(American Geophysical Union, Washington, DC, 1985), pp.
99-110.
J. L. Sarmiento and M. Bender,Photosynth. Res. 39,
209 (1994).
W. S. Broecker and T.-H. Peng, Nature 356, 587
(1992).
T. Stocker and A. Schmittner, Nature 388, 862
(1998).
S. Manabeand R.J. Stouffer,Nature 364, 215 (1993).
www.sciencemag.org SCIENCE VOL290
19. J. L. Sarmiento, T. M. C. Hughes, R. J. Stouffer, S.
Manabe,Nature 393, 245 (1998).
20. F. Joos, G.-K. Plattner, T. Stocker, O. Marchal, A.
Schmittner,Science 284, 464 (1999).
21. P. G. Falkowski,R. T. Backer,V. Smetacek, Science
281, 200 (1998).
22. E. Laws, P. Falkowski,W. O. Smith, H. Ducklow, J.
McCarthy,Global Biogeochem Cycles, in press.
23. This is the apparent steady-state value, not the
change in the net exchange with the atmosphere.
24. This so-called "Strangelove"ocean scenario is uncertain within about a factor of 2 because we do not
have better estimates of export fluxes of carbon in
the oceans (22).
25. This (perhaps counterintuitive)effect is a direct outcome of the calcificationreaction,which can be summarizedas 2HCO3+ Ca <->CaCO3+ COz + H2O.The
mean residencetime of Ca in the oceans is 8.5 x 105
years (it is even longer for Mg2+). On time scales of
millions of years, as the Ca supply from weathering
keeps pace with the precipitationof carbonates,carbonate formationand burialbecome significantsinks
for COz.In fact, carbonatesare the largestreservoirof
carbonon Earth(Table 1). As COzdissolvesin seawater,
it forms carbonicacid, which lowers pH (13). Thus,as
atmosphericCOz increases,the pH of the upperocean
decreases until the bufferingcapacity of the ocean is
restoredby the dissolutionof carbonates.
26. J. L. Sarmiento, T. M. C. Hughes, R. J. Stouffer, S.
Manabe,Nature 393, 245 (1998).
27. J. L Sarmiento and U. Siegenthaler,PrimaryProductivity and Biogeochemical Cycles in the Sea, P. G.
Falkowski,Ed. (Plenum, New York, 1992), pp. 316317.
28. Followingup on the work of the JGOFSprogram,the
responses of marine ecosystems to changes in climate and CO2 in the coming century have been
identified by biological oceanographersas a key research topic for the next decade in funding agency
planning.
29. The storage of carbon in terrestrial ecosystems is
primarilyrelated to the relative resistance of lignin
and other polymeric carbon complexes to degradation or herbivory.Terrestrialplants contain, on average, substantially more organic carbon per unit of
nitrogen or phosphorus than their marine counterparts, the phytoplankton,which are primarilycomposed of protein. Although lignins and other carbonrich polymers may accumulate in terrestrialecosystems on time scales of decades, on longer time scales
most of these molecules are oxidized, so that the
accumulationof organic carbon in soils is a miniscule
fraction of the total carbon fixed by the ecosystem.
Lakesmay also store substantial amounts of organic
matter in sediments [W. E. Dean and E. Gorham,
Geology 26, 535 (1998)].
30. R. Myneni,C. Keeling,C. Tucker,G. Asrar,R. Nemani,
Nature 386, 698 (1997).
31. P. P, Tans, I. Y. Fung,T. Takahashi,Science 247, 1431
(1990).
32.
33.
34.
35.
36.
37.
38.
39.
40.
R.F. Keelingand S. R.Shertz,Nature 358, 723 (1992).
M. Battle et al., Science 287, 2467 (2000).
S. Fan et al., Science 282, 442 (1998).
An invaluablefurtherconstraint on both oceanic and
terrestrial carbon uptake is provided by following
annual and interannual changes in atmospheric Oz
concentrations. With this approach, it is assumed
that intraannualdecreases in 0O and increases in CO2
are due to the combined effects of fossil fuel burning
and deforestation. Increases in 02 with decreased
COz are a consequence of terrestrial uptake, and
decreases in CO2with minimal changes in 02 represent oceanic uptake of the former gas.
NPPis the difference between gross primaryproduction and plant respiration.
D. S. Schimel, Global Change Biol. 1, 77 (1995).
S. P. Long and B. G. Drake, Crop Photosynthesis:
Spatial and TemporalDeterminants,N. R. Bakerand
H. Thomas, Eds. (Elsevier, Amsterdam, 1992), pp.
69-95.
C. Kornerand F. Bazzaz,CarbonDioxide, Populations,
and Communities(AcademicPress,New York,1996).
Because the saturation function of an enzyme is not
a linear function of substrate concentration, the relationship between the change in CO2 and rubisco
13 OCTOBER
2000
295
SCIENCE'S
41.
42.
43.
44.
45.
46.
47.
48.
49.
50.
51.
activity is not linear. These levels of CO2 can be
reached within the next 50 years.
M.U. F. Kirschbaum,
Soil Biol.Biochem.27, 753 (1995).
F. A. Bazzaz,Annu. Rev. Ecol. Syst. 21, 167 (1990).
P. Jarvisand S. Linder,Nature 405, 904 (2000).
P. Vitousek, H. Mooney, J. Lubchenco,J. Melillo,Science 277, 494 (1997).
J. N. Galloway, W. H. Schlesinger, H. Levy, A.
Michaels,J. L.Schnoor,GlobalBiogeochem. Cycles 9,
235 (1995).
R. Howarth,Ed.,Nitrogen Cyclingin the North Atlantic Ocean and Its Watersheds(KluwerAcademic,Boston, 1996).
C. Kroezeand S. Seitzinger,Nutr. Cycl. Agroecosyst.
52, 195 (1998).
J. J. Walsh, Nature 350, 53 (1991).
J. P. Christensen, J. W. Murray,A. H. Devol, L. A.
Codispoti,Global Biogeochem. Cycles 1, 97 (1987).
S. P. Seitzingerand A. E. Giblin,Biogeochemistry35,
235 (1996).
There are exceptions of course. Phosphoruslimitation
occurs in the Mediterraneanand parts of the western
52.
53.
54.
55.
56.
57.
58.
59.
60.
61.
COMPASS
North Atlantic [J. Wu, W. Sunda, E. Boyle, D. Kari,
Science 298, 759 (2000)].
K.Coale et al. Nature 383, 495 (1996).
P. Falkowski,Nature 387, 272 (1997).
I. Fung et al., Global Biogeochem. Cycles 14, 281
(2000).
A. Dai, I. Fung,A. Del Genio,J. Clim. 10, 2943 (1997).
The change in nitrogen inventories in the ocean may
makes a substantialcontributionto the drawdownof
COz on glacial-interglacialtime scales (53, 60).
S. Hall and P. Matson, Nature 400, 152 (1999).
K.J. Nadelhoffer et al., Nature 398, 145 (1999).
The estimated annual global eolian flux of fixed
inorganicnitrogen (NOy) to the ocean is 12.3 Tg, of
which 9 Tg is from anthropogenic sources (45). Assuming that all that nitrogenwas used to fix CO2,the
net stimulation amounts to -65 Tg of carbon per
year. This is less than 1% of the global annual anthropogenic emission of CO2.
W. S. Broeckerand G. M. Henderson,Paleoceanography 13, 352 (1998).
R. Keelingand B. Stephens, Nature 404, 171 (2000).
Mind the
NEW! Science
62. The transition into glacial periods is gradual, with
several phases, and occurs over several tens of thousands of years, whereas the transition to an interglacial state occurs within 10,000 years.
63. J. Sarmiento and C. Le Quere, Science 274, 1346
(1996).
64. P. Crutzen and E. Stoermer, IGBPNewsl. 41, 17
(2000).
65. This article is based on a workshop on the global
carbon cycle held at the Royal Swedish Academy of
Science in November 1999. The workshopwas organized by the InternationalBiosphere-GeosphereProgramme (IBGP)and the Royal Swedish Academy of
Science, in collaboration with Stockholm University
and the Swedish Universityof AgriculturalSciences.
Financialsupport was provided by the Swedish Millennium Committee and by MISTRA(the Swedish
Foundation for Strategic EnvironmentalResearch).
The workshopwas the first of five Stockholm workshops contributing to the IGBPsynthesis project.
Thanksto J. Raven for comments.
gap.
Online's Content Alert Service
With Science's Content Alert Service, Europeansubscribers
(and those aroundthe world)can eliminatethe information
gap
betweenwhenScience publishesand whenitarrivesinthe post.
This free enhancement to your Science Online subscription
delivers e-mail summaries of the latest news and research
articlespublishedeach Fridayin Science- instantly. Tosign
up for the ContentAlert service, go to Science Online and
eliminatethe gap.
Science
www.sciencemag.org
For more information about Content Alerts go to www.sciencemag.org.
Click on the Subscription button, then click on the Content Alert button.
296
2000
13 OCTOBER
VOL290 SCIENCEwww.sciencemag.org