What are the contributions in this paper?

Although the coastal contribution to the anthropogenic carbondioxide budget was neglected in past assessments reported by the Intergovernmental Panel on Climate Change ( IPCC ) and others, it has been recognized recently. In this Review, the authors discuss their current understanding of the sources, fates and exchanges of organic and inorganic carbon in the coastal ocean, with an emphasis on the factors that contribute to net carbon fluxes within and between coastal subsystems.

What is the likely explanation for the increase in DIC fluxes?

It is also increasingly likely that global DIC fluxes have increased as a result of liming and anthropogenic acid additions to watersheds74,79, and that POC fluxes have decreased owing to changing precipitation regimes and land and river management.

Why do the authors assign an overall uncertainty for shelves?

because of less-constrained fluxes in enclosed seas and low latitude open shelves (up to 75% uncertainty)22 (Fig. 3c), the authors assign an overall uncertainty for shelves of 50–75%.

What are the main drivers of the change in the coastal carbon cycle?

Activities such as land-use modification, waterway impoundment, nutrient inputs, wetland degradation and climate change add even greater complexity and uncertainty, making it difficult to differentiate the natural and anthropogenic drivers affecting changes in the coastal carbon cycle.

What is the main driver of the change in riverine carbon exports to the coastal ocean?

Human perturbations to rivers and estuaries Land management is now considered to be a primary driver for changing riverine carbon exports to the coastal ocean (Fig. 1a).

What is the likely cause of the increase in river carbon fluxes?

Although future climate change is predicted to lead to an increase in river carbon fluxes (Fig. 1a), it is also likely to lead to increased uncertainties in predicting these fluxes.

What is the importance of reducing the uncertainty in the air–surface CO2 flux estimates?

The uncertainty in present-day air–surface CO2 flux estimates in coastal systems must also be reduced before meaningful predictions of the effects of climate change on future fluxes can be made.

What is the important factor in the estimation of the ocean’s CO2 flux?

The very large uncertainty associated with estimates of air–water CO2 fluxes in coastal waters and wetlands (about 50% or ± 0.2 Pg C yr−1; Fig. 2) further impedes satisfactory assessment of the overall CO2 exchange from land and ocean to the atmosphere.

What is the estimate of carbon fluxes from estuaries to continental shelves?

Lateral carbon fluxes from estuaries to continental shelves are estimated from the estuarine mass balance of river and wetland inputs, estuarine CO2 degassing, and marsh and estuarine net ecosystem productivities (NEP), and therefore have moderate confidence.

(Open Access) The changing carbon cycle of the coastal ocean (2013) | James E. Bauer

Q: What is the effect of rapid river transport on the carbon budgets of the coastal ocean?

rapid river transport times associated with large hydrological events will result in the bypassing of terrestrial carbon processing in rivers and concomitant episodic disturbance to coastal carbon budgets21, and lead to a shift in the timing of terrestrial carbon delivery to the coastal ocean under future climate change scenarios.

Aquatic Biogeochemistry Laboratory, Department of Evolution, Ecology and Organismal Biology, The Ohio State University, Columbus, Ohio 43210, USA.

School of Marine Science and Policy,

University of Delaware, Newark, Delaware 19716, USA.

School of Forestry and Environmental Studies, Yale University, New Haven, Connecticut 06511, USA.

Department of Geological Sciences,

University of Florida, Gainesville, Florida 32611, USA.

Department of Marine Science, University of Georgia, Athens, Georgia 30602, USA.

Department of Earth & Environmental Sciences, Université

Libre de Bruxelles, Brussels 1050, Belgium.

he coastal ocean consists of several distinct but tightly connected

ecosystems that include rivers, estuaries, tidal wetlands and the

continental shelf. Carbon cycling in the coastal waters that connect

terrestrial with oceanic systems is acknowledged to be a major component

of global carbon cycles and budgets

1–3

. Carbon fluxes within and between

coastal subsystems, and their alteration by climate and anthropogenic

changes, are substantial. It is therefore essential to understand, and accu-

rately account for, the factors regulating these fluxes and how they affect

the ocean and global carbon budgets.

Although the coastal contribution to the anthropogenic carbon-

dioxide budget was neglected in past assessments reported by the

Intergovernmental Panel on Climate Change (IPCC) and others, it has

been recognized recently

3–5

. Constraining the exchanges and fates of

different forms of carbon in coastal settings has been challenging and

is so far incomplete, due to the difficulty in scaling up relatively few

observational studies. Rapid expansion of organic and inorganic carbon

data collection (especially on the partial pressure of CO

, p

) in coastal

waters over the past decade, as well as new biogeochemical contexts

for coastal systems dynamics, make this an exciting time for the field.

A new generation of coupled hydrodynamic biogeochemical models

can now mechanistically incorporate the factors that control carbon

dynamics, such as elemental stoichiometry and biological turnover of

both internally and externally supplied organic matter and nutrients,

and their inputs and residence times

6–8

. These new tools will provide

a more predictive understanding of how coastal systems respond to

human impacts and climate perturbations.

In this Review, we discuss our current understanding of the sources,

fates and exchanges of organic and inorganic carbon in the coastal ocean,

with an emphasis on the factors that contribute to net carbon fluxes

within and between coastal subsystems. Carbon inputs and transfor-

mations are considered in the contexts of net air–water exchanges of

, carbon burial in coastal subsystems, and exports to the open ocean.

We explicitly address the growing recognition of how the coastal car-

bon cycle has fundamentally shifted in recent years owing to a variety

of human activities. This synthesis shows that the present-day coastal

ocean is a net sink for atmospheric CO

and a burial site for organic and

inorganic carbon, and represents an important global zone of carbon

transformation and sequestration. The purported shift of the coastal

ocean from a CO

source to a CO

sink over the past 50 to 100years also

has ramifications for its future role in the ocean and global carbon cycles.

Riverine carbon inputs to coastal systems

Riverine supply of many elements, including carbon of largely terrestrial

origin, is important to the steady-state chemistry of the oceans (Fig.1a).

Although estimates of riverine fluxes of both organic carbon

9–11

and inor-

ganic carbon

3,12

continue to be improved by new geospatial tools and by

scaling and modelling approaches, these fluxes are not greatly different

from earlier estimates

(Fig.2 and Box1). Average annual carbon fluxes

to all major ocean basins and seas are now available

10,12

. Fluxes are gener-

ally well correlated with river discharge, except in certain regions where

factors such as high peat and carbonate coverage, and high erosion rates

in watersheds also control carbon inputs

10–14

Factors regulating riverine carbon fluxes

Climate has long been recognized as an important driver of river carbon

supply to the coastal ocean (Fig.1a). Watersheds with high precipita-

tion have higher riverine discharge rates, and studies have long docu-

mented a primary regulation of carbon fluxes by discharge

, owing to the

importance of transport limitation. Temperature also regulates important

abiotic and biotic processes that can alter water throughput, flow paths,

dissolution rates and watershed carbon stocks. The net effect of tempera-

ture on carbon fluxes can therefore vary between regions and among the

different organic and inorganic forms of carbon (for example, see refs15

and 16) (Fig.1a).

In addition to annual precipitation and temperature, it is now clear

that hydrologic ‘events’, such as extreme rainfall from tropical storms, are

disproportionately important to riverine organic carbon transport. The

erosive power of these storms is responsible for most particulate organic

carbon (POC) export from watersheds to the coastal ocean, especially in

mountainous regions

. Increases in riverine dissolved organic carbon

(DOC) concentrations — and, hence, in annual riverine DOC export

to coastal systems — can also result from these events. For example, a

single tropical storm can be responsible for more than 40% of average

annual riverine DOC export

. On decadal time scales, single large flood

events can export 80–90% of POC from mountainous regions

. Climate

models suggest that although the change in storm frequency is difficult

to predict, the most intense storms will probably become more frequent

and this will consequently affect riverine DOC and POC transport to

coastal waters.

We can currently estimate with moderate to high certainty the riv-

erine transport of terrestrial carbon to the coastal ocean (Fig.2 and

The carbon cycle of the coastal ocean is a dynamic component of the global carbon budget. But the diverse sources and sinks of

carbon and their complex interactions in these waters remain poorly understood. Here we discuss the sources, exchanges and

fates of carbon in the coastal ocean and how anthropogenic activities have altered the carbon cycle. Recent evidence suggests that

the coastal ocean may have become a net sink for atmospheric carbon dioxide during post-industrial times. Continued human

pressures in coastal zones will probably have an important impact on the future evolution of the coastal ocean’s carbon budget.

The changing carbon cycle

of the coastal ocean

James E. Bauer

, Wei-Jun Cai

, Peter A. Raymond

, Thomas S. Bianchi

, Charles S. Hopkinson

& Pierre A. G. Regnier

5 DECEMBER 2013 | VOL 504 | NATURE | 61

REVIEW

doi:10.1038/nature12857

Box1). However, changing climate is expected to have major impacts

on future river carbon fluxes and their uncertainties. Numerous stud-

ies now support precipitation as a dominant effect (compared with,

for example, temperature) on these fluxes in the coming decades

19,20

Furthermore, rapid river transport times associated with large hydro-

logical events will result in the bypassing of terrestrial carbon process-

ing in rivers and concomitant episodic disturbance to coastal carbon

budgets

, and lead to a shift in the timing of terrestrial carbon delivery

to the coastal ocean under future climate change scenarios. The mag-

nitude of these changes remains difficult to assess because the current

generation of Earth system models does not simulate riverine carbon

fluxes (Box2).

Estuaries as modulators of carbon fluxes

Estuaries are transitional regions that range from predominantly river water

to predominantly sea water. As a result, estuaries typically display strong

gradients in biogeochemical parameters (for example, salinity, organic and

inorganic carbon) during river and seawater mixing (Fig.1b). Although

most estuaries are geographically confined, estuaries of high-discharge riv-

ers (for example, the Amazon) can extend onto, and even across, the conti-

nental shelf

. Numerous factors, such as the geomorphology of the estuary

and the magnitude and stoichiometry of nutrient inputs (Box3), control the

fluxes and cycling of carbon in estuaries (Fig.2). These important physical–

biogeochemical reactors greatly modify the amounts and characteristics of

organic and inorganic carbon transported between land and the ocean

2,23,24

Organic carbon in estuaries

DOC and POC in estuaries are derived from terrestrial, marine and

estuarine primary production (Fig.1b). Owing to their unique bio-

chemical and isotopic characteristics, specific sources of organic

carbon have generally been easier to quantify than their fates within

estuaries

23,24

. In situ production of organic carbon in some estuarine

waters can be significant to the coastal carbon budget because it can

equal or exceed the river or marine supply

25,26

Mineral sorption and desorption and photochemical dissolution

can lead to an interchange between DOC and POC in estuaries

27,28

Figure 1 | Processes that affect organic and inorganic carbon cycling

and fluxes in the major coastal ocean subsystems. a, Natural and

anthropogenic processes altering riverine carbon inputs to the coastal ocean.

Inputs can be altered through changes in the water balance (precipitation and

evapotranspiration) and carbon stocks and flows in watersheds. Hydrological

alterations include the effects of climate change on the amount and frequency

of rainfall events and temperature regulation of evapotranspiration. Land

management practices such as irrigation and clearance of vegetation that alter

rates of evapotranspiration can also be important. Recent studies have indicated

that inputs of sulphuric acid, agricultural practices, peatland disturbance,

permafrost thaw, wetland removal and reservoir construction can alter carbon

stocks (biogeochemical response) and flows through varied mechanisms at

the drainage-network level. Carbon flux is equal to carbon concentration

(the hydrological and biogeochemical response) multiplied by discharge

(precipitation minus evapotranspiration). b, Major processes affecting carbon

sources and fluxes in estuaries. Estuaries contain a mixture of organic and

inorganic carbon sources derived from terrestrial materials carried by fresh river

water (in which the salinity is zero), marine sources carried in shelf sea water

(with a salinity of more than or equal to 30), and uniquely estuarine materials.

Organic carbon is lost owing to salinity-induced flocculation, sedimentation,

microbial respiration and photooxidation. Estuaries can modulate the export

of carbon to the shelf depending on whether the estuary is a net carbon source

or carbon sink. c, A representative continental shelf at its interface with a low-

salinity river or estuarine plume. Physical and biogeochemical processes control

the source, transport and fate of organic carbon. Carbon is exchanged at the

interface between plume and shelf waters through sorption and desorption.

Organic carbon transport to the open ocean is supplemented by physical

resuspension, bioturbation and mobile and fluidized mud layers. The benthic

nepheloid layer contains significant amounts of suspended sediment, which

may be deposited to and resuspended from depocentres. Primary production

in inner shelf waters may be limited by high sediment loads in plumes,

whereas regions of upwelling in outer shelf waters can lead to elevated primary

production. DOC, dissolved organic carbon; POC, particulate organic carbon;

DIC, dissolved inorganic carbon.

Agricultural

soil disturbance

Soil sulphate

input

Tropical peatland

disturbance

Reservoirs

Wetlands

removal

Permafrost

thaw

DOC

POC

Precipitation

• Frequency of storms

• Amount of rainfall

Evapotranspiration

• Vegetation cover

• Temperature regulation

• Irrigation

Discharge =

Carbon ux

Sunlight

Saltmarsh

DOC, DIC/CO

additions from submarine

groundwaters, sediment porewaters

Lateral DOC, POC,

DIC/CO

exchanges

from saltmarshes

DIC ux if

estuary is net

source

DIC ux if

estuary is net

sink

Terrestrial

DOC, POC,

DIC/CO

Fresh river

water

uptake

Shelf

seawater

Benthic

resuspension

Flocculation,

sedimentation

Marine

DOC, POC,

DIC/CO

a b

DIC

Microbial respiration:

OC+O

→ CO

Photooxidation:

OC+hν → CO

Primary

production

Net CO

ux

to atmosphere

Solar UV

irradiation

Inner plume

sedimentation

Shelf sediments

Disturbance

Shelf break

Upwelling

Benthic

nepheloid layer

Depocentre

resuspension

Biological

resuspension

Physical

resuspension

Mobile and uidized

mud layers

Sorption/

desorption processes

Mid-plume

sedimentation

Outer-plume

exchange

Cross-shelf exchange

Low salinity

waters

Carbon concentration

62 | NATURE | VOL 504 | 5 DECEMBER 2013

REVIEW

INSIGHT

Estuaries also experience significant losses of organic carbon owing to

the combined influences of microbial degradation and photochemical

oxidation

29–31

, scavenging, sedimentation, and salinity-induced floc-

culation of DOC and POC

(Fig.1b). Individual sources of organic

carbon have unique reactivity and residence times that affect their

degradation

to climatically important gases such as CO

, methane

and volatile organic carbon in estuarine waters and sediments

, and

their export to continental shelves. Estuaries can therefore modulate

organic carbon exports to shelves relative to riverine organic carbon

fluxes alone

Most estuaries have tremendous internal spatial and temporal

heterogeneity in carbon processing and fluxes, making it difficult

to quantify even a single estuary’s net carbon balance. As a result,

we lack the measurements from a representative number of systems

for accurate global or even regional estimates of the direction and

magnitude of net organic carbon fluxes that occur within estuar-

ies. In addition, the complex interplay between organic carbon and

both inorganic and organic nutrient inputs (Box3) from land and

ocean have an important, but poorly quantified, role in regulating

the balance between net organic carbon production (autotrophy) and

consumption (heterotrophy) in estuaries

35,36

. Process-based models of

coupled estuarine hydrodynamics and biogeochemistry have recently

addressed interactions between the organic and inorganic carbon

cycles at the scale of individual estuaries (for example, see refs8, 21,

35, 37), but none are currently suitable for regional or global applica-

tions (Box2).

Estuarine carbon dioxide and inorganic carbon exchange

emissions from European estuaries were recognized to be a signifi-

cant component of the regional CO

budget

about 15years ago. Subse-

quent studies estimated global estuarine CO

emissions to be on the order

of 0.2–0.4 Pg C yr

−1

(refs2, 33, 39, 40). Estuaries occupy a small portion

of global ocean area (about 0.2%), and, therefore, their CO

emissions

are a disproportionately large flux when compared with CO

exchanges

between the open ocean and atmosphere

(Fig.3a). However, the uncer-

tainty in the global estuarine CO

emission flux is high (Fig.2,3a,b and

Box1) due to very limited spatial and temporal coverage during field

observations, large physical and biogeochemical variability and insuf-

ficient use of generalized hydrodynamic–biogeochemical models in

estuaries

2,22

In addition to very high p

and correspondingly high rates of CO

degassing, dissolved inorganic carbon (DIC) is also usually enriched in

estuarine waters and exported to continental shelf waters (Fig.2). The

elevated p

and DIC result from in situ net respiration of internally and

externally supplied organic carbon and lateral transport of DIC from riv-

ers and coastal wetlands

2,33,40,42,43

and CO

-rich groundwaters

. Addition-

ally, low-DIC estuaries

generally experience higher CO

degassing fluxes

than high-DIC estuaries, as waters of the latter retain CO

longer owing

to their greater buffering capacity

45,46

Exchanges with tidal wetlands

Highly productive tidal wetlands flank many estuaries and laterally export

both dissolved and particulate carbon to estuaries and coastal systems

2,47

Regional and global estimates of wetland fluxes are hampered by a scar-

city of reliable estimates of wetland surface area and studies of carbon

export. The limited studies that are available suggest that wetlands act as

a net source of carbon to estuaries that could be comparable with riverine

carbon supply

2,3

(Fig.2). However, although some of the carbon exported

from wetlands is recycled in estuaries, significant amounts are buried in

estuarine sediments and exported to continental shelves (Fig. 2).

Globally, estuaries are net heterotrophic, meaning that respired organic

carbon exceeds that supplied by rivers and wetlands, and produced in

Figure 2 | Organic and inorganic carbon fluxes in the estuarine, tidal

wetland and continental shelf subsystems of the coastal ocean. Fluxes

between adjacent subsystems and other components of the earth system

are regulated by a number of processes (the major ones are shown here).

Carbon can flux both within (values in black) and across (values in red) the

boundaries of the coastal ocean. All organic carbon (OC) and inorganic

carbon (IC) fluxes are presented as positive values, arrows indicate direction

of flux. Particulate and dissolved OC fluxes are presented as total OC values.

The balance between gross primary production (GPP) and total system

respiration (both autotrophic, A, and heterotrophic, H; R

) is net ecosystem

production (NEP), with negative values indicating conversion of OC to IC.

The IC burial flux takes into consideration calcification. The methods used to

estimate flux values and their associated uncertainties are described in Box1.

Typical uncertainties for carbon fluxes: *95% certainty that the estimate is

within 50% of the reported value; †95% certainty that the estimate is within

100% of the reported value; ‡uncertainty greater than 100%. Units are

Pg C yr

−1

(1 Pg = 10

g) rounded to ± 0.05 Pg C yr

−1

. Within-river fluxes and

transformation of carbon are excluded from this analysis.

Atmosphere

Exchange with

open ocean

Riverine input

from land

Coastal

ocean

River ow,

tides

Sediment OC/IC

accumulation, dissolution

OC 0.2†

IC 0.15†

Tides

uptake

Coastal sediments

Tidal

wetlands

xation CO

emissions 0.45 net CO

uptake†

Cont.

shelves

GPP

0.55†

0.25†

OC 0.3†

IC 0.1†

OC 0.5†

IC 0.45†

OC 0.45*

IC 0.4*

0.85 net C input*

OC 0.3†

IC 0.15†

0.45 net C burial†

OC 0.15 to 0.35‡

IC 0.5 to 0.7‡

0.85 net C export†

OC 0.05‡ OC 0.05‡

0.25*†

0.1†

–0.15 to

0.05‡ R

Estuaries

GPP

–0.2† R

GPP

0.35† R

5 DECEMBER 2013 | VOL 504 | NATURE | 63

REVIEW

INSIGHT

situ, by 0.2 Pg C yr

−1

(ref. 33). The outgassing of estuarine CO

, derived

from organic carbon respiration and DIC inputs from rivers and wetlands,

releases around 0.25 Pg C yr

−1

of CO

into the atmosphere (Figs2,3a,b).

Although highly uncertain, carbon burial in wetlands and estuaries is

probably around 0.1 Pg C yr

−1

(refs3, 48, 49) (Fig.2 and Box1). Our mass

balance analysis suggests that estuaries export about 10% more carbon to

continental shelves than they import from rivers (Fig.2).

Tremendous geomorphological differences between estuaries and a

lack of synthetic modelling lead to considerable variability and uncer-

tainty in estimates of estuarine carbon dynamics and export to continen-

tal shelves. Estuaries are typically net sources of CO

to the atmosphere

and augment the organic and inorganic carbon supply to shelves (Fig.2).

Climate and land-use changes (specifically sea-level rise and declining

river sediment export) are likely to decrease net carbon burial in estuaries

and wetlands

; however, with anticipated increased inputs from rivers,

estuarine export of carbon to continental shelves is likely to increase.

Carbon on continental shelves

Continental shelves are dynamic interfaces where terrestrial, estuarine

and marine organic carbon is recycled (Fig.1c). Continental shelves

occupy only 7–10% of global ocean area

22,50

(Fig.3a). However, shelves

contribute 10–30% of global marine primary production, 30–50% of inor-

ganic carbon and around 80% of organic carbon burial in sediments

and could contribute up to about 50% of the organic carbon supplied to

the deep open ocean

. Thus, continental shelves are disproportionately

important to ocean carbon cycles and budgets.

Shelf organic carbon sources and sinks

Continental shelf primary production is often related to shelf width and

the magnitude of river discharge

(Fig.1c). On broad river-dominated

shelves (for example, Mississippi and Yangtze), primary production on the

inner shelf may be limited by high particulate loads, and inputs of river-

and estuary-derived organic carbon may dominate the water column and

sediments. On broad shelves with lower river discharge (for example,

South Atlantic Bight), sunlight may reach the sea floor and support a

significant benthic contribution to shelf primary production and organic

carbon

. On narrow shelves, pelagic and/or benthic primary production

is supported by oceanic inputs of nutrients and is recycled on the shelf

Organic carbon burial rates on continental shelves are controlled

by multiple mechanisms (Fig.1c). When river- and estuarine-derived

organic carbon enters the coastal zone, about 90% of it is associated with

mineral matrices in organo-clay aggregates

. In many highly productive

upwelling regions along shelves, organic carbon particles may aggregate

by glue-like exopolymers from phytoplankton

. In addition to burial,

organic carbon in shelf environments is transported to the open ocean

through physical resuspension, bioturbation, and mobile and fluid muds

along the sea floor

(Fig.1c). Flocculation processes similar to those in

estuaries (Fig.1b) can also be important in transporting selected forms of

DOC and POC from shelf waters to sediments and altering their chemical

composition

Organic carbon reactivity on shelves

Although terrestrial DOC supplied by rivers can theoretically account for

the 4,000–6,000year residence time of DOC in the global ocean

, little

terrestrial material is actually detected in the oceans. How this large flux

of terrestrial DOC is processed in the coastal ocean is a major remain-

ing question in coastal carbon research. Recent findings show that much

(90% or more) of the reactive aromatic fraction of terrestrial DOC is

altered in inner shelf waters by sunlight-driven photoreactions to pro-

duce highly stable, ubiquitous and presumably long-lived components of

oceanic DOC

56,57

. Bacterial alteration of terrestrial plant-derived lignin

can account for additional DOC losses (up to 30% of the photochemical

losses) in river-dominated shelf waters

. There is, thus, a growing consen-

sus that much of the terrestrial DOC becomes chemically altered, rather

than completely oxidized to CO

, on shelves. This may help to reconcile

the disagreement between riverine inputs and ocean DOC residence

times. In contrast to terrestrial organic carbon, 50–90% of the organic

carbon that is derived from marine primary production is rapidly recycled

Carbon fluxes and their associated uncertainty estimates presented in

this Review are not based on statistical treatment of multiple observed

data, as data coverage is poor and often skewed. Rather, most fluxes

are based on ranges presented in the literature and the quality of the

individual values. Because of the large degree of heterogeneity in the

main coastal subsystems and concomitant lack of data, most carbon

fluxes in these subsystems have relatively high uncertainties. Riverine

carbon fluxes have been estimated over the past three decades and

are known with the highest degree of confidence (95% certainty

that the estimate is within 50% of the reported value

2,3,9–14

) (Fig.2).

Wetland net primary production rate, carbon burial, CO

degassing and

dissolved inorganic carbon (DIC) export rates are based on values in the

literature

2,49,86

. The estuarine CO

degassing flux is based on the best

known syntheses of field data

2,33,40,66

with moderate confidence (95%

certainty that the estimate is within 100% of the reported value) (Figs2,

3). Organic carbon burial of 0.03–0.2 Pg C yr

−1

in wetland and estuarine

sediments is the most poorly constrained flux (95% certainty that the

estimate is greater than 100% of the reported value), and we adopt

a conservative value

2,3

(Fig.2). Lateral carbon fluxes from estuaries to

continental shelves are estimated from the estuarine mass balance

of river and wetland inputs, estuarine CO

degassing, and marsh and

estuarine net ecosystem productivities (NEP), and therefore have

moderate confidence.

Owing to recent marked improvements in the quantity and quality

of data, present-day continental shelf CO

gas exchange (0.25 Pg C yr

−1

)

is among the best estimated of all coastal carbon fluxes (within 50%

uncertainty)

2, 39, 40, 61, 66

. However, because of less-constrained fluxes in

enclosed seas and low latitude open shelves (up to 75% uncertainty)

(Fig. 3c), we assign an overall uncertainty for shelves of 50–75%. On

the basis of the CO

uptake flux, the known global shelf surface area

(26 × 10

), average gas transport parameter (9.3 cm h

−1

), and

present-day atmospheric p

(Fig.4b), an average surface water p

of 350 ± 18 p.p.m. is estimated for the post-industrial shelf. Present-

day carbon burial in shelf sediments is also estimated with moderate

confidence

2, 3, 50, 89

For the pre-industrial shelf (Fig.4a) a net CO

degassing flux of

around 0.15 Pg C yr

−1

was estimated as the mean of the upper

2,50

and

lower

bounds of reported values. To be compatible with this CO

degassing flux and a pre-industrial atmospheric p

of 280p.p.m., an

average surface water p

of 298±18 p.p.m. is estimated (Fig. 4a). For

pre-industrial time, we estimate a shelf NEP of −0.15 Pg C yr

−1

as the

mean between an upper limit, assuming that 60% of terrestrial organic

carbon is respired on shelves and a low bound of zero from ref.68.

However, for present-day shelf NEP, we assumed two scenarios (see

‘Pre- and post-industrial shelf carbon budget’). For the pre-industrial

and present-day comparison of shelf CO

fluxes, we held constant

river input, estuarine CO

exchange flux with the atmosphere, NEP

and sediment burial constant and then calculated the various organic

carbon and inorganic carbon export fluxes by mass balance. Our level

of confidence for these estimates is low (Fig.4a, b).

BOX 1

Coastal carbon flux estimates and their uncertainties

64 | NATURE | VOL 504 | 5 DECEMBER 2013

REVIEW

INSIGHT

on continental shelves. Reactive marine material may also enhance the

metabolism of less reactive terrestrial organic carbon in shelf waters and

sediments

Carbon-dioxide

exchange in shelf waters

Tsunogaietal.

first pointed out, in 1999, the importance of con-

tinental shelf CO

uptake to the carbon cycling and climate change

communities, proposing a value as high as around 1 Pg C yr

−1

, equal

to 50% of the open ocean CO

uptake known at the time. Most recent

syntheses are based on up-scaling methods whereby different shelf

systems are classified by dividing them into a few provinces

or typol-

ogies

. These estimates suggest a lower, but, relative to the global

ocean and land

, still significant net atmospheric CO

uptake flux of

0.25 Pg C yr

−1

(Fig.3a).

Inner continental shelf waters close to land tend to be sources of CO

mostly due to their high rates of respiration of terrestrial and estuarine

organic carbon and lateral transport of high CO

waters from adjacent

inshore systems. By contrast, mid- to outer-shelf waters are a sink of

(ref.62). This general pattern results from decreased terrestrial

organic carbon supply, increased primary production as light condi-

tions improve offshore, and increased accessibility to nutrients supplied

by upwelling and mixing across the shelf break

. This pattern and the

inshore to offshore shift from CO

release to CO

uptake across shelves

can be altered greatly in larger river plumes

or in upwelling dominated

shelves

(Fig.1c) and depends to a significant extent on physical con-

ditions such as wind stress and river discharge. Furthermore, a strik-

ing latitudinal contrast in shelf-water–atmosphere CO

fluxes emerges

from a global synthesis of shelf systems. Present-day shelves located

between 30° and 90° latitude are, in general, sinks for atmospheric CO

whereas shelves located between the equator and 30° tend to be sources

of CO

(or nearly neutral) to the atmosphere

61,66

(Fig.3c). This latitudi-

nal pattern could, in part, be explained by the fact that around 60% of

river organic carbon is exported to lower latitude shelves and respired

under the higher mean temperatures in these systems

2,10

. Of increasing

importance is that the western Arctic Ocean margin, in particular the

nutrient-rich Chukchi Sea, has become a rapidly increasing global shelf

sink for atmospheric CO

over the past decade

due to greater annual

retreat of sea ice — a major barrier to gas exchange — as a result of

climate warming.

Autotrophic–heterotrophic balance

Before the extensive alteration of terrestrial landscapes and industrial

fertilizer production, estuarine and continental shelf waters were on the

whole thought to be net heterotrophic (they released more CO

to the

atmosphere than was fixed by primary production, as a result of their

respiration of terrestrial and tidal wetland organic carbon inputs

68,69

Indeed, almost every river and estuary globally, for which data are avail-

able, is today a strong source of CO

(refs1, 40, 66). These systems were

probably an equally strong — if not stronger — source of CO

in the past

when atmospheric p

was lower.

There is currently no consensus as to whether present-day continental

shelves are net autotrophic or heterotrophic, owing, in part, to a lack of

concurrent respiration measurements to go along with the abundant pri-

mary productivity measurements that have been made in shelf waters

The complexity of coastal systems further hampers proper upscaling for

modelling (Box2). A school of thought suggests that continental shelves

as a whole are now net autotrophic because of increased anthropogenic

nutrient supply

50,63,69

, which in some systems exceeds deep ocean nutrient

inputs by upwelling or mixing across the shelf break

. This view is consist-

ent with the observation that most shelves are a net sink for atmospheric

(refs40, 61, 66). Thus, it has been postulated that the coastal ocean

as a whole has shifted from a net heterotrophic to an increasingly net

autotrophic state, which has in turn favoured a reversal from shelves being

a CO

source to a CO

sink in recent decades

. This hypothesis is sub-

stantiated by results from coarse-grained box models (Box2). Although

it is important to establish whether such a shift has occurred, if it has, its

exact magnitude and timing remain highly uncertain.

Similar to estuaries, continental shelves are highly heterogeneous

coastal subsystems. Carbon dynamics in some shelves are controlled

entirely by ocean circulation, whereas in others they are controlled

largely by riverine inputs. Uncertainties in shelf carbon fluxes

are significant (Fig.2) but cannot yet be adequately constrained

Earth system models (ESMs) are climate models that can include

physical processes, as well as biogeochemical cycles, and which allow a

representation of anthropogenic processes

. They describe processes

within and between the atmosphere, ocean, cryosphere, and terrestrial

and marine biosphere. ESMs include coarse-grained box models,

models of intermediate complexity and comprehensive tridimensional

global climate models that incorporate biogeochemical processes,

such as carbon cycle and atmospheric chemistry. However, with the

exception of a few box models, ESMs are at present limited by their lack

of coupling between atmosphere, land and ocean components through

lateral flows of carbon (and nutrients; Box3) along the land–ocean

continuum

. The delivery of riverine carbon is included as a forcing

condition in large-scale ocean component models

, but spatially

resolved ESMs do not simulate the riverine carbon fluxes dynamically.

Attempts to estimate the historical evolution of the aquatic fluxes

from land to ocean and their effects on estuarine and continental shelf

carbon dynamics have so far relied solely on globally averaged box

models

. Although these models are extremely valuable for testing

further conceptual ideas, they rely on highly parameterized process

formulations. In addition, global box models do not account for the wide

diversity of estuarine and shelf systems, nor do they mechanistically

represent the effect of land-use changes on terrestrial biogeochemistry

and riverine fluxes.

State-of-the-art ESMs contain routing schemes for riverine water

flows at a spatial resolution of 0.5°. At this resolution, about 500rivers

that contribute around 80% of the total organic carbon delivery to

the oceans may now be resolved. Furthermore, with the development

of eddy-permitting runs at a resolution of about 0.25°, shelf carbon

dynamics are increasingly better captured. Including these dynamics

will allow better resolution of important processes such as the air–sea

exchange that results from large-scale coastal eddies or the

physical transport divergence of carbon across coastal boundaries

and the shelf break. The largest rivers of the world produce coastal

plumes which can also be reasonably well captured

, whereas narrow

coastal systems such as eastern boundary currents will still require

higher resolution

. For estuaries and tidal wetlands a resolution of

0.25–0.5° is too coarse, and specific modelling approaches that rest

on mechanistically rooted upscaling strategies need to be designed

to better constrain their roles in ocean and global carbon cycles and

assess their sensitivity to anthropogenic disturbances. Improved

mechanistic descriptions of ecosystem and biogeochemical processes

are also needed as the current formulations may not be able to

sufficiently capture the complexity of coastal ocean dynamics and their

response to enhanced terrestrial inputs

BOX 2

Modelling the coastal carbon cycle

5 DECEMBER 2013 | VOL 504 | NATURE | 65

REVIEW

INSIGHT

The changing carbon cycle of the coastal ocean

Figures

Citations

Global Carbon Budget 2020

Global Carbon Budget 2018

Global Carbon Budget 2016

Global Carbon Budget 2019

Global Carbon Budget 2017

References

The Physical Science Basis : Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change

The Physical Science Basis

Supporting Online Material for Spreading Dead Zones and Consequences for Marine Ecosystems

Spreading Dead Zones and Consequences for Marine Ecosystems

Biogeochemistry : An Analysis of Global Change

Related Papers (5)

Anthropogenic perturbation of the carbon fluxes from land to ocean

Carbon dioxide in water and seawater: the solubility of a non-ideal gas

Plumbing the Global Carbon Cycle: Integrating Inland Waters into the Terrestrial Carbon Budget

Relationship between wind speed and gas exchange over the ocean

Sedimentary organic matter preservation: an assessment and speculative synthesis

Frequently Asked Questions (14)

Q1. What are the contributions in this paper?

Q2. Why do fluxes in the main coastal subsystems have relatively high uncertainties?

Q3. What is the likely explanation for the increase in DIC fluxes?

Q4. What is the effect of rapid river transport on the carbon budgets of the coastal ocean?

Q5. How can the authors better constrain their roles in ocean and global carbon cycles?

Q6. Why do the authors assign an overall uncertainty for shelves?

Q7. What are the main drivers of the change in the coastal carbon cycle?

Q8. What is the main driver of the change in riverine carbon exports to the coastal ocean?

Q9. What is the likely cause of the increase in river carbon fluxes?

Q10. What are the main factors that hinder the estimates of wetland fluxes?

Q11. What is the importance of reducing the uncertainty in the air–surface CO2 flux estimates?

Q12. What is the important factor in the estimation of the ocean’s CO2 flux?

Q13. How is the current carbon burial estimated for the post-industrial shelf?

Q14. What is the estimate of carbon fluxes from estuaries to continental shelves?