53
© The Ecological Society of America www.frontiersinecology.org
R
ivers and other inland water systems represent
distinct ecosystems, geomorphological agents, sites
of biogeochemical storage and transformation, and con-
duits for material transport across continents and to the
oceans, and are linked with human health and economic
activity (Figures 1 and 2). Although the importance of
rivers in modulating the transfer of nutrients from the
land to coastal areas has long been recognized (Smith
and Hollibaugh 1993), rivers are typically considered as
passive “pipes” in regional and global budgets of carbon
(C) and weathering products (ie dissolved ions and sedi-
ment). Recent studies and syntheses have overturned
this view of rivers as passive transporters (Richey et al.
2002; Cole et al. 2007; Battin et al. 2009; Tranvik et al.
2009) and of weathering exports as unresponsive to
human impacts (Raymond et al. 2008). A key finding is
that the amount of C that rivers deliver to the oceans is
only a fraction of that entering rivers from terrestrial
ecosystems (Figure 3). Most of this C is returned to the
atmosphere in the form of carbon dioxide (CO
2
) before
reaching the oceans or is stored within river corridors as
sedimentary organic carbon (OC) after erosion and
transport from distant sites. Understanding the inter-
play between reactivity, transport, and coupling among
landscape components (ie uplands, riparian zones,
streams and rivers, wetlands, lakes, floodplains) is a fun-
damental challenge for characterizing the susceptibility
of riverine biogeochemical function to climate and land-
use change.
Here we concentrate on three important aspects of
water, C, and mineral cycles – linked together – to illus-
trate the role of rivers and other inland waters as biogeo-
chemical couplers of landscapes and Earth-system com-
ponents (ie continents, oceans, atmosphere):
(1) The lateral exports of C from terrestrial ecosystems by
streams and subsequent CO
2
outgassing to the atmos-
phere, highlighting their increasingly recognized
importance for closing terrestrial net ecosystem C
balance (see Panel 1);
(2) The erosion, mixing, and burial of minerals and asso-
ciated OC, including the coupling of C sequestration
with geomorphological drivers;
(3) The influence of river exports of alkalinity and major
ions on coastal responses to ocean acidification,
including the effects of human perturbation within
watersheds on mineral weathering.
COUPLED BIOGEOCHEMICAL CYCLES
Riverine coupling of biogeochemical cycles
between land, oceans, and atmosphere
Anthony K Aufdenkampe
1*
, Emilio Mayorga
2
, Peter A Raymond
3
, John M Melack
4
, Scott C Doney
5
,
Simone R Alin
6
, Rolf E Aalto
7
, and Kyungsoo Yoo
8
Streams, rivers, lakes, and other inland waters are important agents in the coupling of biogeochemical cycles
between continents, atmosphere, and oceans. The depiction of these roles in global-scale assessments of
carbon (C) and other bioactive elements remains limited, yet recent findings suggest that C discharged to the
oceans is only a fraction of that entering rivers from terrestrial ecosystems via soil respiration, leaching, chemi-
cal weathering, and physical erosion. Most of this C influx is returned to the atmosphere from inland waters as
carbon dioxide (CO
2
) or buried in sedimentary deposits within impoundments, lakes, floodplains, and other
wetlands. Carbon and mineral cycles are coupled by both erosion–deposition processes and chemical weather-
ing, with the latter producing dissolved inorganic C and carbonate buffering capacity that strongly modulate
downstream pH, biological production of calcium-carbonate shells, and CO
2
outgassing in rivers, estuaries, and
coastal zones. Human activities substantially affect all of these processes.
Front Ecol Environ 2011; 9(1): 53–60, doi:10.1890/100014
In a nutshell:
• Rivers and other inland waters receive, transport, process, and
return to the atmosphere amounts of carbon (C) of similar
magnitude to the net ecosystem C balance of the terrestrial
ecosystems in their watersheds
• Burial of C on continents – within sedimentary deposits of
inland waters – is an order of magnitude greater than burial of
C in the oceans
• Human-accelerated chemical weathering of minerals in water-
sheds affects coastal-zone acidification
1
Stroud Water Research Center, Avondale, PA
*
(aufdenkampe@
stroudcenter.org);
2
Applied Physics Laboratory, University of
Washington, Seattle, WA;
3
Yale School of Forestry and Environmental
Studies, New Haven, CT;
4
Bren School of Environmental Science and
Management, University of California, Santa Barbara, CA;
5
Woods
Hole Oceanographic Institution, Woods Hole, MA;
6
Pacific Marine
Environmental Laboratory, NOAA, Seattle, WA;
7
School of
Geography, College of Life and Environmental Sciences, University of
Exeter, Devon, UK;
8
Department of Soil, Water and Climate,
University of Minnesota, St Paul, MN
N
Nitrogen
14.007
Rivers and the coupling of biogeochemical cycles AK Aufdenkampe et al.
54
www.frontiersinecology.org © The Ecological Society of America
Our consideration of coupling focuses primarily on that
between continents, oceans, and atmosphere by rivers
and other inland waters. We mainly consider biogeo-
chemical coupling between C and inorganic species, such
as alumino-silicate and carbonate minerals, in the con-
text of geochemical and geomorphological processes,
rather than coupling between different bioactive ele-
ments (ie C, nitrogen [N], phosphorus [P], oxygen [O],
and sulfur [S]), which we assume to be implicit in our dis-
cussion of C processing. Also, we consider most inland
waters – such as streams, lakes, impoundments, and the
many types of wetlands – as being components of river
systems. Although each of these inland water types repre-
sents a distinct ecosystem with associated processes,
nearly all receive and/or discharge materials from/to the
river networks to which they are coupled.
n
From land to river to atmosphere: C under the radar
In the early 1990s, a decade’s worth of oceanographic and
global C modeling identified a significant “missing carbon
sink” on continents that was equivalent to about one-
third of global fossil-fuel emissions, launching numerous
research efforts to quantify the net storage of C in the
major terrestrial ecosystems. Nearly two decades later, a
consensus has yet to emerge regarding the spatial distribu-
tion of terrestrial C sinks. One roadblock is that net
ecosystem production (NEP) measured at local scales does
not often extrapolate well to larger scales (Ometto et al.
2005; Stephens et al. 2007). These discrepancies may be
due to the assumption that all NEP is converted entirely
to storage, without consideration of lateral export (Panel
1). Recent findings suggest that, in many cases, lateral
exports – through rivers, airborne transport of reduced C
compounds, and commercial trade in food and forest prod-
ucts – can be of equal magnitude to storage fluxes (Richey
et al. 2002; Billett et al. 2004; Ciais et al. 2008). This obser-
vation may help reconcile disparate estimates of C seques-
tration and provide improved estimates of terrestrial NEP
(Billett et al. 2004; Cole et al. 2007).
Aquatic biogeochemists have established that nearly all
fresh waters contain CO
2
in concentra-
tions that are supersaturated with respect
to that of the atmosphere. The partial
pressure of dissolved CO
2
(pCO
2
; Panel
2) in water in equilibrium with the
atmosphere is equivalent to the concen-
tration of CO
2
in the atmosphere, which
is currently about 390 parts per million
(ppm) by volume. Measured pCO
2
values
typically range from 1000 to more than
12 000 ppm in rivers (Cole and Caraco
2001; Richey et al. 2002; Johnson et al.
2008; Humborg et al. 2010) and from 350
Figure 1. The delta of Russia’s Lena River. Terrestrial and
aquatic ecosystems closely intermingle in many continental areas
across the world, such as boreal ecosystems.
Figure 2. The Negro River in the central
Amazon, just upstream of the confluence with
the Branco River. River floodplains and inland
waters are dynamic systems that link Earth
systems across large spatial and temporal scales.
USGS EROS Center/NASA Landsat Project Science Office
USGS EROS Center/NASA Landsat Project Science Office
C
Carbon
12.011
AK Aufdenkampe et al. Rivers and the coupling of biogeochemical cycles
55
© The Ecological Society of America www.frontiersinecology.org
to more than 10 000 ppm in lakes and reser-
voirs (Sobek et al. 2005; Marotta et al. 2009)
– with tropical waters typically exhibiting
higher concentrations of CO
2
than temper-
ate waters, and rivers and wetlands typically
having higher concentrations of CO
2
than
lakes (Table 1). These values show large net
freshwater-to-atmosphere CO
2
fluxes, which
must be balanced by substantial inputs from
terrestrial systems, as either OC or CO
2
-rich
groundwater (Mayorga et al. 2005; Johnson
et al. 2008). The translation of water–air
CO
2
gradients to areal fluxes requires multi-
plication by gas exchange velocities (k),
which vary as a function of turbulence in
the surface water and, to a lesser degree,
with temperature (Table 1 and WebPanel
1). The CO
2
fluxes from one square meter of
surface water can therefore be much higher
than CO
2
fluxes into one square meter of
the adjacent terrestrial ecosystem (Table 1).
Regional and global estimates of CO
2
out-
gassing fluxes rely as much on estimates of
inland water area as on the CO
2
fluxes per
unit area of surface water. Over the past few
years, our estimation of inland water area has
substantially increased as a result of (1)
improvements in remote-sensing approaches,
often combining data captured from multiple
airborne- and satellite-borne sensors to assess
flooding beneath vegetation (Hess et al. 2003; Prigent et al.
2007), and (2) compilations and algorithms to estimate
areal coverage of water bodies (Lehner and Döll 2004;
Downing 2009). Recent studies have more than doubled
the global land area known to be seasonally to perma-
nently inundated to over 20 million km
2
(Lehner and Döll
2004; Downing 2009), an area roughly equivalent to 15%
of global land surfaces, excluding Antarctica and
Greenland (Table 1). These improved estimates of inun-
dated area translate to proportional increases in the magni-
tudes of outgassing fluxes.
Published estimates of the global flux of CO
2
outgassing
from inland surface waters range from 0.75 to 1.4 peta-
grams of C per year (Pg C yr
−1
; Figure 3), which is globally
important when compared with estimates of net C accu-
mulation on continents (2.2 Pg C yr
−1
) or in the oceans
(2.2 Pg C yr
−1
; Cole and Caraco 2001; Cole et al. 2007;
Battin et al. 2009; Takahashi et al. 2009; Tranvik et al.
2009). Yet these CO
2
fluxes from inland waters directly to
the atmosphere are rarely considered in global or regional
Figure 3. The coupling of land, oceans, and atmosphere by rivers, lakes, and
wetlands. All numbers are fluxes in units of Pg C yr
−1
, with values based on an
analysis by Battin et al. (2009); accumulation fluxes within both land and
ocean each equal 2.2 Pg C yr
–1
. The CO
2
outgassing and continental burial
fluxes from Battin et al. (2009) are substantially larger than those published by
Cole et al. (2007), primarily on account of more complete consideration of
high-latitude lakes. A more balanced inclusion of tropical waters and wetlands,
and temperature dependencies on pCO
2
and k, as we consider in Table 1,
would require a further increase in outgassing fluxes to the atmosphere. These
flux values have direct consequences to net C balances on land because of the
need to balance the global C budget.
Atmosphere
1.2
Rivers, lakes, wetlands
0.6
Geosphere
Panel 1. Terms of terrestrial ecosystem carbon budgeting
Net ecosystem production (NEP) is the difference between gross primary production and the community respiration of an ecosystem,
typically reported in units of C per unit time per unit area. NEP represents the amount of C available for storage or export over a given
time interval, with a positive value for net autotrophic ecosystems and a negative value for net heterotrophic ecosystems. Although a
fundamental property of an ecosystem, NEP is challenging to measure over annual and decadal time scales. The historical approach for
terrestrial ecosystems has been to measure changes in C stocks in biomass and soils over many years, with the implicit assumption that
lateral C export is negligible. A more recent approach has been to deploy sensors for CO
2
and wind on a tower – emerging from the
forest canopy – to take measurements of the turbulent vertical fluxes of CO
2
every few seconds for a period of years. Data from these
“eddy covariance flux” towers are integrated to quantify the net amount of CO
2
transported into the forest from the atmosphere, which
is known as net ecosystem exchange (NEE, where NEE < 0 represents a land sink for CO
2
). The assumption has been that lateral export
is negligible and that NEE equals NEP, but several new studies have called into question these assumptions and assert that lateral export
of C from terrestrial ecosystems – via rivers and other processes – may be substantial relative to NEP or NEE (Richey et al. 2002; Billett
et al. 2004; Cole et al. 2007; Ciais et al. 2008). As a result, an effort has been made to remind the terrestrial ecosystem research commu-
nity of the original definition of NEP, which includes export (Lovett et al. 2006), and to introduce a less ambiguous term, net ecosystem
carbon balance (NECB), to describe only the flux of C available for local storage (Chapin et al. 2006).
Land Ocean
2.7
0.9
P
Phosphorus
30.974
Rivers and the coupling of biogeochemical cycles AK Aufdenkampe et al.
56
www.frontiersinecology.org © The Ecological Society of America
C balances. Furthermore, recent re-evaluations of the
factors used to generate regional CO
2
outgassing esti-
mates – pCO
2
concentrations, k, and areal extent of
inundation – along with a more balanced consideration
of the tropics and wetlands all support an increase to pub-
lished values for regional and global CO
2
outgassing
fluxes (Table 1 and WebPanel 1). Such estimates would
have direct consequences on net C balances on land
because of the need to balance global, regional, and even
local C budgets.
The effort to quantify the net C sink in mature rain-
forests in the Amazon Basin is an excellent case study,
demonstrating the importance of rivers in coupling ter-
restrial and atmospheric C cycles (Richey et al. 2002;
Ometto et al. 2005). Net ecosystem exchange (NEE;
Panel 1) of C – measured by eddy covariance flux towers
situated throughout the Amazon Basin – has ranged from
high forest sinks to modest sources (–6 to +1 megagrams
[Mg] C ha
–1
yr
–1
; Ometto et al. 2005), yet biomass and soil
surveys generally yield lower sink estimates (Ometto et al.
2005; Chave et al. 2008). Richey et al. (2002) suggested
that lateral C exports to streams could play a potential
role in balancing this discrepancy. In response, two stud-
ies examined whether lateral stream exports might be
important relative to terrestrial fluxes (Waterloo et al.
2006; Johnson et al. 2008). For streams in the seasonally
dry southern Amazon, a net export of 0.40 Mg C ha
–1
yr
–1
was calculated (Johnson et al. 2008) for dissolved CO
2
,
which is largely outgassed within the few hundred meters
downstream of groundwater seeps or springs, representing
one-half of total deep-soil respiration; Johnson et al.
(2008) also estimated that an additional 0.10 Mg C ha
–1
yr
–1
was exported as dissolved organic
carbon (DOC). Because no eddy covari-
ance tower was present at the sites stud-
ied by Johnson et al. (2008), comparisons
with published Amazon NEE values from
other sites led us to estimate that stream
export of groundwater CO
2
and DOC
could account for 10–100% of NEE at
that site (Ometto et al. 2005; Johnson et
al. 2008). Waterloo et al. (2006) mea-
sured DOC and particulate organic car-
bon (POC) exports in a 2nd- to 3rd-
order blackwater stream within an eddy
covariance tower site, and estimated 0.19
Mg C ha
–1
yr
–1
of OC export to represent
5–6% of NEE. This estimate does not
include dissolved CO
2
, yet our few mea-
surements of CO
2
degassing from these
highly acidic streams immediately
upstream of Waterloo et al.’s (2006) OC
monitoring site are as high as, or even
higher than, those of Johnson et al.
(2008), suggesting that >15% of NEE
may be exported as dissolved CO
2
and OC (Aufdenkampe unpublished).
Neither study was able to compare local NEE measure-
ments with the full suite of stream C exports, including
DOC, POC, and total dissolved inorganic carbon (DIC,
which includes CO
2
; Panel 2). Nevertheless, in a peat-
land system in Scotland, Billett et al. (2004) were able to
account for all lateral C exports via streams within the
footprint of an eddy covariance tower and demonstrated
that NEE changed from a net sink of 0.278 Mg C ha
–1
yr
–1
to a net source of 0.083 Mg C ha
–1
yr
–1
when lateral
stream fluxes were considered. One challenge, however,
of comparing stream C exports with NEE is whether the
evaded CO
2
(from groundwater or respired from organic
matter) is already captured within the footprint of the
eddy covariance tower. Regardless, these studies suggest
that lateral transport of C by streams should be consid-
ered when evaluating net ecosystem carbon balance
(Chapin et al. 2006).
n
Carbon sequestration from soils to sea:
erosion is key
Geologists and oceanographers have long hypothesized
that there is a connection between tectonic uplift, mineral
erosion, C burial, and atmospheric oxygen over geological
time scales (Berner 1989), with river systems implicitly
coupling the biogeochemical cycles between continents,
oceans, and atmosphere. Studies of OC turnover in soils
and sediments have begun to describe the mechanisms for
the coupling of mineral and biogeochemical cycles via
organo-mineral complexation, which is a critical factor in
stabilizing OC (Hedges and Keil 1995). Yet OC and min-
eral production are typically spatially separated. Biological
Panel 2. Dissolved carbon dioxide, explained
Dissolved carbon dioxide (CO
2
), carbonic acid (H
2
CO
3
), bicarbonate (HCO
3
–
), and
carbonate (CO
3
2–
) are collectively referred to as dissolved inorganic carbon (DIC).
These chemical species readily interconvert from one to another as a function of
their relative concentrations, pH, temperature, and the concentrations of other
buffering and complexing species. With a gas species (CO
2
) at one end of the reac-
tion chain and several mineral species at the other (eg calcium carbonate, CaCO
3
),
the carbonate buffering system – as this series of related reactions is known – can
only be understood by simultaneously solving a series of thermodynamic equations.
The carbonate buffering system is further coupled to mineral cycles because car-
bonate alkalinity (Carb
alk
= [HCO
3
–
] + 2×[CO
3
2–
]) is controlled by the thermody-
namic requirement to balance the electrostatic charge imbalance between strong
base cations (eg calcium [Ca
2+
], manganese [Mg
2+
], sodium [Na
+
], potassium [K
+
])
and strong acid anions (eg chlorine [Cl
–
], sulfate [SO
4
2–
]) that is created during the
chemical weathering of minerals. These equations and associated thermodynamic
constants have been thoroughly studied (Dickson et al. 2007).
Concentrations of dissolved “free” CO
2
are often reported in terms of partial
pressure, pCO
2
, which is the equivalent atmospheric CO
2
concentration that would
be in equilibrium with the water sample. Typical pCO
2
units are parts per million
(ppm) volume of CO
2
per volume of air, or microatmospheres (µatm). Thus, pCO
2
does not directly refer to a concentration of free CO
2
in water but rather the result
of the application of Henry’s Law of gas dissolution equilibrium, which simply states
that at a given temperature the concentration of a dissolved gas is proportional to
the partial pressure of that gas in equilibrium with the solution.
S
Sulfur
32.065
AK Aufdenkampe et al. Rivers and the coupling of biogeochemical cycles
57
© The Ecological Society of America www.frontiersinecology.org
primary production mainly occurs where there is light –
that is, above the soil surface or in the low-turbidity
euphotic zone of lakes and oceans. Mineral surfaces are
produced at the bedrock–soil interface and subsoil, where
primary minerals are chemically weathered to secondary
alumino-silicate clays and iron and aluminum hydroxides.
In most systems, the flux of OC mobilized into river corri-
dors and the coastal oceans is substantially greater than the
system’s capacity to stabilize it via organo-mineral com-
plexation, burial into anoxic environments, or other
mechanisms. Thus, most OC is rapidly metabolized or pho-
tolysed and returned to the atmosphere as CO
2
(Cole et al.
2007). Therefore, the rate of erosion, delivery, and mixing
of fresh mineral surfaces with fresh organic matter could
likely control watershed- to global-scale C sequestration
fluxes over both current and geological time scales (Hedges
2002; Kennedy et al. 2006).
A series of reports and letters responding to Van Oost et
al. (2007) serves to highlight an ongoing debate over
whether anthropogenic erosion might result in a globally
important C sink (Stallard 1998), an important net CO
2
source (Lal 2003), or a combination that yields a slight net
sink (Van Oost et al. 2007). Previous studies have explic-
itly considered one or more of three mechanisms linking
erosion–deposition processes to alterations of net CO
2
fluxes between the land and the atmosphere: (1) complete
to partial replacement of soil OC at eroding agricultural
sites via crop production (Stallard 1998); (2) burial and
inhibited decomposition of OC eroded from topsoils
(Stallard 1998); and (3) enhanced decomposition of soil
OC due to the breakdown of protective soil structures dur-
ing erosion and transport (Lal 2003). Nevertheless, these
studies did not consider a fourth mechanism that could
substantially augment estimates of global C sequestration
due to erosion: (4) the transport of OC-poor minerals
from deep soil horizons into environments where stable C-
mineral complexes are formed with otherwise fresh, reac-
tive C (Aufdenkampe et al. unpublished).
The conventional scientific perspective is that, glob-
ally, rivers discharge about 0.4 Pg C yr
–1
as OC to the
oceans, with about one-half in the form of DOC and one-
half as POC, and an additional 0.5 to 0.4 Pg C yr
–1
as DIC
(Schlünz and Schneider 2000; Figure 3). These estimates
have changed surprisingly little over time, despite revi-
sions to both sediment delivery fluxes and the concentra-
tions of POC associated with those sediments (Berner
1989; Schlünz and Schneider 2000). Geomorphologists
have known, however, that only 5–25% of eroded sedi-
ment actually reaches the oceans, with most being
deposited at the bottom of hillslopes (ie colluvium), in
floodplains and wetlands (ie alluvium), and in reservoirs,
lakes, and estuaries (Figure 4). This is apparent when
comparing recent global estimates of modern fluxes of
12.6 Pg yr
–1
of sediment delivered to the oceans (Syvitski
et al. 2005) with estimates of modern erosion fluxes of
50–150 Pg yr
–1
of soil (Stallard 1998; Wilkinson and
McElroy 2007). Conservatively assuming that the OC
content of these sediments averages 1%, then about
Table 1. Estimates of CO
2
outgassing from inland waters, for zones based on atmospheric circulation
Zone-class Area of inland waters pCO
2
Gas exchange velocity Areal outgassing Zonal outgassing
(1000s km
2
) (ppm) (k
600
, cm hr
–1
) (g C m
–2
yr
–1
) (Pg C yr
–1
)
min – max median median median median
Tropical (0˚– 25˚)
Lakes and reservoirs 1840–1840 1900 4.0 240 0.45
Rivers (>60–100 m wide) 146–146 3600 12.3 1600 0.23
Streams (<60–100 m wide) 60–60 4300 17.2 2720 0.16
Wetlands 3080–6170 2900 2.4 240 1.12
Temperate (25˚–50˚)
Lakes and reservoirs 880–1050 900 4.0 80 0.08
Rivers (>60–100 m wide) 70–84 3200 6.0 720 0.05
Streams (<60–100 m wide) 29–34 3500 20.2 2630 0.08
Wetlands 880–3530 2500 2.4 210 0.47
Boreal and Arctic (50˚–90˚)
Lakes and reservoirs 80–1650 1100 4.0 130 0.11
Rivers (>60–100 m wide) 7–131 1300 6.0 260 0.02
Streams (<60–100 m wide) 3–54 1300 13.1 560 0.02
Wetlands 280–5520 2000 2.4 170 0.49
Global Percent of global land area
Lakes and reservoirs 2800–4540 2.1%–3.4% 0.64
Rivers (>60–100 m wide) 220–360 0.2%–0.3% 0.30
Streams (<60–100 m wide) 90–150 0.1%–0.1% 0.26
Wetlands 4240–15 220 3.2%–11.4% 2.08
All inland waters 7350–20 260 5.5%–15.2% 3.28
Notes: see WebPanel 1 for associated references.
N
Nitrogen
14.007
