Content uploaded by J. Jotautas Baronas
Author content
All content in this area was uploaded by J. Jotautas Baronas on Jan 04, 2021
Content may be subject to copyright.
EARTH, ATMOSPHERIC,
AND PLANETARY SCIENCES
Global silicate weathering flux overestimated because
of sediment–water cation exchange
Edward T. Tippera,1 , Emily I. Stevensona, Victoria Alcocka, Alasdair C. G. Knighta, J. Jotautas Baronasa,
Robert G. Hiltonb, Mike J. Bicklea, Christina S. Larkina, Linshu Fenga, Katy E. Relpha, and Genevieve Hughesa
aDepartment of Earth Sciences, University of Cambridge, Cambridge CB2 3EQ, United Kingdom; and bDepartment of Geography, Durham University,
Durham DH1 3LE, United Kingdom
Edited by Andrea Rinaldo, ´
Ecole Polytechnique F ´
ed ´
erale de Lausanne, Lausanne, Switzerland, and approved November 18, 2020 (received for review
August 4, 2020)
Rivers carry the dissolved and solid products of silicate mineral
weathering, a process that removes CO2from the atmosphere
and provides a key negative climate feedback over geological
timescales. Here we show that, in some river systems, a reac-
tive exchange pool on river suspended particulate matter, bonded
weakly to mineral surfaces, increases the mobile cation flux by
50%. The chemistry of both river waters and the exchange pool
demonstrates exchange equilibrium, confirmed by Sr isotopes.
Global silicate weathering fluxes are calculated based on riverine
dissolved sodium (Na+) from silicate minerals. The large exchange
pool supplies Na+of nonsilicate origin to the dissolved load,
especially in catchments with widespread marine sediments, or
where rocks have equilibrated with saline basement fluids. We
quantify this by comparing the riverine sediment exchange pool
and river water chemistry. In some basins, cation exchange could
account for the majority of sodium in the river water, significantly
reducing estimates of silicate weathering. At a global scale, we
demonstrate that silicate weathering fluxes are overestimated
by 12 to 28%. This overestimation is greatest in regions of high
erosion and high sediment loads where the negative climate
feedback has a maximum sensitivity to chemical weathering reac-
tions. In the context of other recent findings that reduce the net
CO2consumption through chemical weathering, the magnitude
of the continental silicate weathering fluxes and its implica-
tions for solid Earth CO2degassing fluxes need to be further
investigated.
cation exchange |global biogeochemical cycles |suspended
particulate matter |silicate weathering
For decades, silicate weathering has been postulated to pro-
vide the negative climate feedback on Earth that prevents
a runaway greenhouse climate like on Venus (1). Silicate min-
eral dissolution with carbonic acid converts atmospheric CO2
into carbonate, and releases essential nutrients to the terrestrial
and marine biosphere (2). There have been many attempts to
quantify the silicate weathering flux (3), mostly assuming that
riverine dissolved sodium (Na+) is derived only from silicate
minerals and rock salt. Here we show that there is a major
addition of nonsilicate Na+to the critical zone from ancient
seawater, weakly bonded to sedimentary rocks and supplied
to waters via the cation exchange process. The implication is
not only that the silicate weathering flux is overestimated at a
global scale, but that this nonsilicate Na+is most important in
regions previously thought to have the highest silicate weathering
fluxes (so called weathering-limited regions) and greatest climate
sensitivity.
Cation exchange is a rapid chemical reaction between cations
in the dissolved phase and mineral surfaces, particularly clays
(4). Major and trace cations such as calcium (Ca2+), magne-
sium (Mg2+), sodium (Na+), potassium (K+), and strontium
(Sr2+) form the cation exchange pool, which balances negative
charges on river-borne clay particle surfaces. This exchange takes
place on interlayer sites, between the tetrahedral and octahe-
dral layers, or on exposed surfaces (4). The importance of the
cation exchange pool is well recognized in soils and aquifers
(4, 5), has significant implications for enhanced weathering (6),
and has been proposed as an important mechanism for buffer-
ing the composition of river waters (7–9). However, data on the
riverine exchange pool are only available for two large river sys-
tems [Amazon and Ganges-Brahmaputra (10, 11)], despite its
significance in providing a source of elements that are imme-
diately bioavailable (12), and their potential for biasing the
quantification of silicate weathering (9).
It is increasingly recognized that rapidly reactive phases have
a strong influence on the chemistry of river waters (13, 14).
Cation exchange is a rapid reaction occurring continuously in
soils, as riverine freshwaters evolve downstream interacting with
particulate matter, and when they mix with seawater (15, 16).
Important examples of cation exchange are the “swapping” of
divalent cations Ca2+ and Mg2+ with Na+, in particular when
there is a major change in water composition such as when fluvial
clays reach the ocean,
Ca2+
clay + 2Na+
water 2Na+
clay + Ca2+
water.[1]
As a result, marine sediments have an exchange pool that is
dominated by Na+(17). Subsequently, these marine sediments
are uplifted and emplaced on the continents where Na+in
the exchange pool is released by cation exchange with Ca-
rich fresh waters (9). This has major implications for estimates
of silicate weathering fluxes and associated CO2consumption,
Significance
Large rivers transport water and sediment to floodplains and
oceans, supplying the nutrients that sustain life. They also
transport carbon, removed from the atmosphere during min-
eral dissolution reactions, which is thought to provide a key
negative climate feedback on long timescales. We demon-
strate that the (million-year) carbon flux associated with
mineral dissolution has been overestimated by up to 28%
because of a reactive pool of elements transported with river-
borne suspended sediment. This is most acute in regions of
high erosion, where silicate weathering is thought to be most
intense.
Author contributions: E.T.T. designed research; E.T.T., E.I.S., V.A., A.C.G.K., and J.J.B. per-
formed research; E.T.T., E.I.S., V.A., A.C.G.K., J.J.B., C.S.L., L.F., K.E.R., and G.H. analyzed
data; and E.T.T., E.I.S., V.A., A.C.G.K., J.J.B., R.G.H., M.J.B., C.S.L., and K.E.R. wrote the
paper.y
The authors declare no competing interest.y
This article is a PNAS Direct Submission.y
Published under the PNAS license.y
1To whom correspondence may be addressed. Email: ett20@cam.ac.uk.y
This article contains supporting information online at https://www.pnas.org/lookup/suppl/
doi:10.1073/pnas.2016430118/-/DCSupplemental.y
Published December 21, 2020.
PNAS 2021 Vol. 118 No. 1 e2016430118 https://doi.org/10.1073/pnas.2016430118 |1 of 6
Downloaded at Cambridge University on December 22, 2020
because they are calculated using the Na+content of rivers (3).
Cerling et al. (9) proposed that the Na+-rich exchange pool
exerts an important control on natural waters, based on charge
balance arguments from river water chemistry, but this hypothe-
sis has never been rigorously tested (18) by determining the flux
and composition of the exchange pool of rivers around the world.
In this contribution, we present a large dataset of fluvial sedi-
ment cation exchange capacity (CEC) and composition in several
of the world’s largest river basins. By comparing with the con-
comitant dissolved load chemistry, we demonstrate that 1) the
exchange pool in river sediments is in equilibrium with the river
water; 2) the fraction of mobile elements in the exchange pool
relative to the dissolved pool can be significant, particularly in
rapidly eroding, weathering-limited catchments; and 3) given
reasonable inferences on the composition of old marine sedi-
mentary rocks, modern-day silicate weathering has been overes-
timated and carbonate weathering has been underestimated. The
results reduce the estimated magnitude of the silicate weathering
flux, but increase the supply of base cations (e.g., Ca2+, which
can be a limiting nutrient) to the biosphere, suggesting a greater
role of organic carbon burial compared with silicate weathering
for the long-term atmospheric CO2sink.
Samples, Their Global Reach, and Outline Methods
Paired river water, suspended sediment, and bed or bank sedi-
ment were collected from several of the world’s largest rivers and
their tributaries between 2013 and 2019, including the Murray
and Darling in Australia, the Hong He (Red River), Irrawaddy,
Salween, and Mekong in Southeast Asia, major tributaries of
the Ganges (Karnali, Narayani, Koshi, Rapti, and Bagmati), and
the Mackenzie and Yukon Rivers in North America. In addi-
tion, a number of small streams from Svalbard and the Alps
were analyzed. The data are compared to the Amazon and the
Ganges-Brahmaputra systems (10, 11). This sample collection
is from a very wide range of catchments, with diverse lithologi-
cal, climatic, geomorphological, and weathering characteristics.
The Mackenzie is a shale-rich Arctic basin [where the major-
ity of the clay is marine in origin (19)], whereas the Irrawaddy
is a tropical basin, draining mixed sedimentary and magmatic
terranes (20). The Himalayan tributaries of the Ganges are
weathering limited, whereas the Murray–Darling system is trans-
port limited. The Mackenzie is almost free from anthropogenic
influence, whereas the Hong He and Mekong are extensively
dammed (21).
Exchangeable ions on the suspended particulate matter (SPM)
were determined either by reacting the sediment with calcite-
saturated cobalt(III) hexammine chloride (CoHex; SI Appendix)
(11) or by using ammonium chloride (NH4Cl). NH4Cl is known
to induce the dissolution of carbonate minerals if present (5),
but, after filtering the data for exchange equilibrium (dis-
cussed below), the CEC and chemical compositions are within
uncertainty for NH4Cl and CoHex extractions (SI Appendix,
Figs. S1–S3).
Results
Exchange Equilibrium between the Dissolved Load and Exchange
Pool. The measured exchange pool chemistry is in equilibrium
with the river water chemistry, determined by comparing mea-
surements of the exchange pool composition with modeled equi-
librium values (SI Appendix). When βCa(measured ), the fraction
of Ca to other major cations in the exchange pool, deviated
from the equilibrium value by more than the uncertainty of the
CoHex data (SI Appendix, Fig. S2), samples were defined as not
in equilibrium and excluded from calculations. The majority of
these samples are NH4Cl extractions of carbonate-rich samples,
and the offset is consistent with carbonate dissolution. Stron-
tium isotopes (87Sr/86 Sr) provide a robust tracer of the origin
of the exchange pool (22). The very wide range in 87Sr/86 Sr in
the samples analyzed affords a unique opportunity to assess the
chemical equilibrium between the exchange pool and coeval river
water. There is a striking 1:1 relationship between 87Sr/86 Sr in
the exchange pool of the SPM and 87Sr/86 Sr in the river water
(Fig. 1), with a tighter relationship for the CoHex extractions
(R2= 0.98) compared to the NH4Cl extractions (R2= 0.92).
NH4Cl extractions which deviate significantly from the 1:1 line
(Narayani and Trisuli Rivers in Nepal, the Peel in Canada, and
the Salween) were offset because of small amounts of carbonate
dissolution.
Chemistry of the Riverine Exchange Pool. The exchange pool chem-
istry is dominated by Ca2+ (βCa , the fraction of Ca in the
exchange pool, is typically >0.75, mean = 0.81). For the remain-
ing major cations, βMg = 0.16,βNa = 0.017, and βK= 0.018
(Fig 2A), on average. Although the river water and suspended
sediment exchange pool are in equilibrium (based on equilib-
rium calculations and Sr isotopes), they are distinct because of
exchange selectivity coefficients, such that Ca2+ and Mg2+ have
stronger affinities for the exchange pool compared to K+and
Na+(5). For example, the riverine exchange pool is enriched in
Ca2+ and depleted in Na+, K+, and Mg2+ relative to the river
water (Fig. 2). This has been observed previously in both the
Ganges-Brahmaputra and Amazon River basins as well as soil
pore waters (10, 11, 23).
Comparison with the Marine Exchange Pool. There is a clear differ-
ence between the composition of the exchange pool for riverine
sediments in equilibrium with river waters (present study), and
marine sediments. Modern marine sediments have an exchange
pool dominated by Na+(βNa = 0.6; refs. 15 and 17). When rivers
enter the oceans, the exchange pool of riverine particulate mat-
ter rapidly reacts with seawater, exchanging Ca2+ for Na+(Eq.
1and ref. 15). There are few measurements of unweathered
1:1 line
linear ts
through
simulated data
Nepal
Mackenzie
Mekong
Murray-Darling
Red
Salween
Yuk on
Irrawaddy
Fig. 1. Sr isotope ratio in the exchange pool as a function of Sr isotopes
in the river water. Open and closed (black outline) symbols used the NH4Cl
and CoHex methods, respectively. Gray symbols indicate samples where the
water and exchange pool are not in equilibrium. Uncertainties (500 parts
per million [ppm]) synthetically distributed about the mean of the data are
illustrated by the small points (Inset). Red lines are 100 examples of linear
fits through this synthetic CoHex data.
2 of 6 |PNAS
https://doi.org/10.1073/pnas.2016430118
Tipper et al.
Global silicate weathering flux overestimated because of sediment–water cation exchange
Downloaded at Cambridge University on December 22, 2020
EARTH, ATMOSPHERIC,
AND PLANETARY SCIENCES
Amaz
on (
Sayl
es & Man
gles
dor
f
197
9)
Mac
k
enzi
e
Damma Glacier
Gan
g
es-Brahmaputra (Lupker et al.
)
Irrawa
dd
y
Murra
y
Darli
ng
Mekong
Ne
p
a
l
Re
d
Salw
een
Svalbard
Y
u
k
o
n
Groundwater equilibrium (n=29782)
} 2S.D. } 2S.D.
AB
Fig. 2. Composition of the riverine exchange pool (A) and dissolved pool (B) as the percentage of the elements Ca2+, Mg2+, K+, and Na+. Amazon and
Ganges-Brahmaputra data are from refs. 10 and 11, respectively. Open and closed (black outline) symbols used the NH4Cl and CoHex methods, respectively.
Gray symbols indicate samples where water and exchange pool are not in equilibrium. Small red points are exchange pool compositions calculated in
equilibrium with groundwaters. Cluster of blue data points to indicate uncertainties are synthetic data distributed using the maximum uncertainties of the
data determined using a Monte Carlo simulation.
(pristine) exhumed marine rocks. Some have βNa in equilibrium
with seawater (24–26), but some have lower βNa of <0.18 (27,
28), suggesting a resetting by diagenetic or weathering processes.
There are many more measurements of continental groundwa-
ters, many of which are enriched in Na+, likely linked to halite
dissolution (nonsilicate Na+). We calculated the exchangeable
cation compositions in equilibrium with such groundwaters from
a compilation of almost 30,000 continental groundwaters (SI
Appendix, Fig. 2A, and ref. 29). The modal βNa was 0.80 (mean
of 0.56, interquartile range of 0.37 to 0.78; SI Appendix, Fig. S4),
demonstrating that the unweathered continental exchange pool
can retain a high βNa .
CEC. The CEC of SPM was calculated as the sum of exchange-
able Ca2+, Mg2+ , K+, and Na+in milliequivalent units per 100
g (meq/100 g) of dry sediment, following convention. The range
is comparable to that observed in other large river systems (<1
meq/100 g to 40 meq/100 g; SI Appendix, Fig. S5) (10, 11) and also
in soils (23). This range is lower than the CEC estimates for clay
minerals smectite (57 meq/100 g to 106 to meq/100 g) and kaoli-
nite (17 meq/100 g to 35 to meq/100 g) (30), since fluvial SPM
is always a mixture of multiple mineral phases. The sample from
the Damma Glacier in the Swiss Alps, draining exclusively gran-
ite, has the lowest CEC, whereas the rivers with high CEC drain
predominantly sedimentary terranes such as in the Mackenzie
Basin (31).
The CEC correlates with the aluminum/silicon (Al/Si) ratio
of the bulk sediment (11) (SI Appendix, Fig. S6; R2= 0.7, 0.8,
and 0.7, for the Mekong, Sun Koshi, and Koshi River datasets,
respectively). The Al/Si ratio is well documented to be a function
of the grain size (32), which is strongly related to the proportion
of clay mineral in the bulk sediment. Given that the fine frac-
tion of rivers is dominated by clay minerals (33), this covariation
provides a persuasive argument that the riverine CEC of SPM is
dominated by clays. In weathering-limited systems, the majority
of these clays are old marine sedimentary clays (19).
Implications
Increased Total Reactive Pool of Elements. Although the CEC is
conventionally quoted in units of meq/100 g, it is convenient to
express the CEC of riverine SPM in microequivalents per liter
of water (µeq/L), taking into account the concentration of sus-
pended sediment in the water column. Expressing CEC in these
units allows direct comparison with the sum of the solute cations
in the river water. The ratio of exchangeable Ca to solute Ca
(Fig. 3) ranges from close to zero in some river systems to >0.5.
Half of our SPM samples carry more than ∼5% of total Ca in
the exchange pool, including major river systems such as the
Salween, Mekong, Yukon, and rivers from Nepal.
Given that the exchange pool is in equilibrium with the river
water, this implies that the total cationic load from chemical
weathering (particularly of carbonates) has been underestimated
at a global scale. The principal control on the ratio of exchange-
able to dissolved cations is the concentration of sediment in
the water column (Fig. 3). While the CEC varies by a factor of
∼40, the SPM content varies by three orders of magnitude (from
0.01 g/L to 10 g/L) in our dataset, and, in extreme cases, such as
glacial lake outburst floods, SPM concentrations can exceed 100
g/L (34). This is of particular importance to weathering fluxes,
because regions of the world with high erosion rates (and thus
} Max 2S.D.
]
M
a
ckenzi
e
Damma Glacier
Irrawadd
y
Murra
y
Dar
l
in
g
Mekong
Ne
p
a
l
Re
d
S
al
wee
n
Svalbard
Y
uko
n
Fig. 3. Ratio of Ca2+in the exchange pool relative to the dissolved pool
(marginal plot shows frequency distribution showing 50th and 75th per-
centiles) versus the concentration of SPM in the river. Red dashed line is
the best fit through the data. The black dotted lines bracket the data cal-
culated using the 90th and 10th percentiles of Ca2+in the exchange pool
and the water to delineate upper and lower bounds. Open and closed sym-
bols used the NH4Cl and CoHex methods, respectively. Gray symbols indicate
water and exchange pool are not in equilibrium. Cluster of blue data points
to indicate uncertainties are synthetic data distributed using the maximum
uncertainties of the data determined using a Monte Carlo simulation.
Tipper et al.
Global silicate weathering flux overestimated because of sediment–water cation exchange
PNAS |3 of 6
https://doi.org/10.1073/pnas.2016430118
Downloaded at Cambridge University on December 22, 2020
typically elevated SPM contents) deliver the greatest weather-
ing fluxes, consume the largest amounts of atmospheric carbon
dioxide, and have the greatest sensitivity for climate feedbacks
(35–37). It is these rivers from weathering-limited environments
that have the greatest capacity for interaction between the river
sediment and water, because they have the highest SPM con-
tents. Several of the rivers here have a monsoonal hydrograph
(Mekong, Irrawaddy, Salween, Nepal, Red) where SPM con-
centrations are highest during the monsoon (38), while solute
concentrations are at their lowest (39). Therefore, maximum
CaExchange/CaWater coincides with the time of greatest solute flux.
Transport-limited river basins with low SPM may also be sig-
nificantly influenced by cation exchange processes in the soil
environment, where the exchange pool may control the soil
pore waters for many hundreds of years after the complete dis-
solution of primary mineral phases (22, 23). Cation exchange
has long been proposed as a mechanism for buffering river
water chemistry (7, 40), but a quantification of the compo-
nent derived from ancient sedimentary rocks has proved elusive
(9). The new data are used to quantify the supply of cations
from the ancient marine exchange pool to modern weather-
ing systems, as a function of the SPM content, the CEC, and
chemical makeup of the lithogenic exchange pool. Below, we
investigate the significance of the exchange pool for the silicate
weathering fluxes.
A Reduced Silicate Weathering Flux. The substantial riverine
exchange pool has important consequences for the calculation of
silicate weathering fluxes, particularly where rivers erode large
volumes of marine sedimentary rocks or continental rocks that
have equilibrated with saline basement waters. In general, areas
of greatest dissolved load flux are spatially correlated to easily
eroded sedimentary rocks, and tectonically active regions con-
tain uplifted sedimentary sequences characteristic of continental
margins (41). Since the modern riverine exchange pool is domi-
nated by Ca2+ (Fig. 2A), and marine sedimentary rocks have an
exchange pool dominated by Na+, the primary exchange reaction
during uplift and exhumation is 2Na+
clay Ca2+
water. The Na+-
dominated ancient marine exchange pool is released to river
waters with an equivalent charge of Ca2+ removed from the river
water (Eq. 1; the reverse of the reaction that occurs when riverine
SPM is discharged into the sea). This is of importance, since most
riverine estimates of modern silicate weathering fluxes are based
on the assumption that, after NaCl salt correction, all remaining
Na+(denoted Na∗) is derived exclusively from the dissolution
of silicate minerals. Atmospheric CO2is converted to bicar-
bonate in solution via silicate mineral dissolution in carbonic
acid (42). In contrast, the release of Na+from the exchange
pool does not consume atmospheric CO2. Since the silicate
weathering flux and attendant CO2consumption are calculated
directly from the Na∗flux (42), any reduction in Na∗to account
for ancient marine Na+supplied via cation exchange (Na+
ex )
is equivalent to the reduction in long-term carbon drawdown
via silicate weathering. Denoting Nasil as the silicate contri-
bution of Na+, corrected for both halite and cation exchange
inputs, we quantified the percentage reduction in silicate
weathering as
100 ·
(Na∗−Nasil)
Na∗=100 ·Na+
ex
(Na+
river −Cl−
river),[2]
where a maximum estimate of Na+
ex (µeq/L) is given by the
product of βNa in equilibrium with seawater and the CEC
(milliequivalents per kilogram) of the riverine SPM (grams
per liter),
Na+
ex =CEC ·SPM ·βNa .[3]
Since some marine sediments are likely influenced by diagen-
esis lowering βNa , in addition to some fraction of the riverine
CEC resulting from neoformed clays in equilibrium with soil
pore waters (19) (SI Appendix), we considered a generalized case
where the reduction in the silicate weathering flux was parame-
terized as a function of βNa between 0 and 0.6, the equilibrium
value with seawater (SI Appendix, Fig. S7).
Since both Na∗and CEC ·SPM are determined for our sam-
ple set (Fig. 4, with the distributions indicated by the marginal
plots), the reduction in the silicate weathering flux can be directly
assessed as a function of βNa . Assuming βNa = 0.6(Fig. 4), the
equilibrium value with seawater, many samples show a major
reduction in Na∗(percentage reduction indicated by the con-
tours) and thus the total silicate weathering flux. Some rivers
plot above the 100% contour, indicating the entirety of river-
ine Na∗is derived from the exchange pool and that βNa is
<0.6in these catchments. For the Yukon, Irrawaddy, Salween,
Mackenzie, Mekong, and Nepalese Rivers, the reduction in Na∗
is greater than 25% (basin averages). In contrast, rivers such
as the Amazon and the Murray–Darling, show less than a 10%
reduction in Na∗. For βNa = 0.6, the average reduction in Na∗
is 53% (mean of the entire dataset). Assuming a more conserva-
tive βNa = 0.2reduces Na∗by 32% once cation exchange is taken
into account.
The contrasting behavior between different river basins
reflects the interplay of the variables Na∗, CEC, and SPM. For
example, although Nepalese rivers have among the lowest CEC
(mean of 8.5 meq/100 g), they also have the lowest Na∗values
(mean of 62 µmol/L) but highest suspended sediment concentra-
tions (3.8 g/L). In contrast, the Amazon dataset (10) has similar
Na∗(90 µmol/L), and much higher CEC (24 meq/100 g), but
very low SPM (0.08 g/L). As noted above, high erosion basins
10
20
30
40
50
60
70
80
90
100
40
0
40
40
50
50
50
0
0
0
50
50
0
0
50
0
50
5
50
50
0
0
0
50
50
50
0
50
50
50
50
50
0
50
0
50
50
50
50
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
50
50
0
0
50
0
5
5
50
0
0
50
50
0
5
5
5
5
5
5
0
0
5
0
0
5
5
5
0
0
5
5
5
5
5
5
5
50
0
0
5
5
50
0
5
5
5
5
5
5
5
5
5
}
2S.D.
Fig. 4. CEC ·SPM vs. Na* contoured for the percentage reduction in sili-
cate weathering flux 100 ·(Na* −Nasil )/Na* (0 = no change) calculated for
βNa =0.6. Open and closed symbols used the NH4Cl and CoHex methods,
respectively. Gray symbols indicate water and exchange pool are not in equi-
librium. Symbol legend is the same as for Fig. 2. Cluster of blue data points to
indicate uncertainties are synthetic data distributed using the mean uncer-
tainties of CEC ·SPM and Na* determined using a Monte Carlo simulation.
SI Appendix, Fig. S8 shows contours calculated for βNa =0.2.
4 of 6 |PNAS
https://doi.org/10.1073/pnas.2016430118
Tipper et al.
Global silicate weathering flux overestimated because of sediment–water cation exchange
Downloaded at Cambridge University on December 22, 2020
EARTH, ATMOSPHERIC,
AND PLANETARY SCIENCES
(weathering limited) are the most susceptible to a substantial
influence from the exchange pool because of their high sedi-
ment load (Fig. 3). While weathering-limited basins are thought
to be the most important for the silicate weathering feedback
(36, 37, 43), they also have the largest reduction in Na∗, even for
low βNa .
Extrapolation to Other Large Rivers. To estimate CEC for river
basins where data are not available, we exploited the lin-
ear dependency between CEC and the Al/Si of the SPM (SI
Appendix, Fig. S6). Uncertainty was determined using the covari-
ance matrix for Al/Si and CEC, which propagates the uncertainty
of Al/Si to the CEC using a Monte Carlo simulation. For sev-
eral of the world’s largest rivers, we used published values of
the Al/Si ratio (44–47) to determine the CEC and its associ-
ated uncertainty. Using Na∗and SPM concentration values (3),
we determined the maximum percentage reduction in silicate
weathering flux (100 ·(Na∗−Nasil)/Na∗; Eq. 3and Fig. 5).
The maximum discharge-weighted global average reduction in
Na∗was determined as 28% (βNa = 0.6). A lower bound was esti-
mated using βNa = 0.2with a 12% reduction in Na∗(SI Appendix,
Fig. S9). However, the global reduction in Na∗is uneven, with
basins with high SPM load and/or low Na∗having the largest
reductions in Na∗, at up to 100% for some of the world’s
largest rivers.
Anthropogenic Influences. It is worth noting that, in recent
decades, many large river systems have suffered major reductions
in their sediment loads because of hydropower dam construction
(21). For example, the Red River samples are heavily affected
by reduced SPM contents because of dams. Firstly, for rivers
with a significant anthropogenic influence, such as the Red
and the Mekong Rivers, the reduction in Na∗determined here
underestimates the long-term reduction, because the suspended
particulate load is lower than in its natural state. Secondly, the
total flux of labile cations being delivered to the world’s flood-
plains and oceans is being reduced by sediment trapping in dams
because of the significant flux of elements carried in the exchange
pool. This is a temporary effect from a geological perspective,
but it can skew estimates of chemical weathering determined
downstream of major dams.
Conclusions
We measured the chemistry and magnitude of the exchange
pool in eight of the largest river systems on Earth. Strontium
isotopes and comparison of the dissolved and exchange pool
chemistry indicate chemical equilibrium between the dissolved
and exchange pools. In some river systems, the flux of mobile ele-
ments in the exchange pool bonded weakly to mineral surfaces on
SPM is comparable with that in the dissolved pool. This exchange
pool delivers an important flux of base cations to the world’s
floodplains and oceans, currently in a state of perturbation by the
rapid proliferation of dams. We demonstrate that, if at least part
of the riverine exchange pool was marine in origin, this Na+-rich
marine exchange pool contributes to the modern river chemistry
by exchanging with calcium. By comparing the exchange pool
chemistry to that of the river water, we demonstrate that the
global silicate weathering flux is 12 to 28% lower than previously
thought, and up to 100% lower in some river systems. The reduc-
tion in the calculated silicate weathering flux is most marked in
regions of the world with high erosion and high sediment loads,
where chemical weathering reactions that control the negative
climate feedback are most sensitive. This finding, that the mag-
nitude of the silicate weathering flux needs to be reevaluated,
adds to a series of recent studies (2, 48) that emphasize CO2
release during chemical weathering, and raise questions for the
canonical view of the silicate weathering feedback.
Materials and Methods
Materials and methods are summarized here; further details are provided
in SI Appendix. Water and suspended sediment samples were mostly col-
lected midchannel from boats, or, for smaller rivers, from bridges. Some
recent bank deposits, from just above the water line, were collected and
analyzed. Sediment was reacted with calcite-saturated CoHex (SI Appendix)
(11), where the Co(NH3)3+
6ion displaces the exchangeable cations, but is
buffered to calcite saturation, preventing the dissolution of calcite present
80°S
70°S
60°S
50°S
40°S
30°S
20°S
10°S
0°
10°N
20°N
30°N
40°N
50°N
60°N
70°N
80°N
30°W 10°W 10°E 30°E 50°E 70°E 90°E 110°E 130°E 150°E 170°E 170°W 150°W 130°W 110°W 90°W 70°W 50°W 30°W
80°S
70°S
60°S
50°S
40°S
30°S
20°S
10°S
0°
10°N
20°N
30°N
40°N
50°N
60°N
70°N
80°N
30°W 10°W 10°E 30°E 50°E 70°E 90°E 110°E 130°E 150°E 170°E 170°W 150°W 130°W 110°W 90°W 70°W 50°W 30°W
Amazon (10%)
Brahmaputra (24%)
Congo (8%)
Danube (NA%)
Don (0%)
Fraser (35%)
Ganges (26%)
Indus (49%)
Irrawaddy (53%)
Karnali (100%) Koshi (87%)
Lena (NA%)
Loire (4%)
Murray − Darling (1%)
Mackenzie (71%)
Mekong (34%)
Mississippi (43%)
Narayani (100%)
Niger (94%)
Nile (12%)
Ob (8%)
Orinoco (21%)
Rapti (48%)
Hong (Red River) (6%)
Salween (100%)
Seine (NA%)
St Lawrence (4%)
Wisla (NA%)
Volga (NA%)
Xun Jiang (85%)
Yangtze (86%)
Huang He (100%)
Yenisey (NA%)
Yukon (100%)
25
50
75
100
100
Na
*
Na
sil
Na
*
Na
0.6
Fig. 5. Global map of large river basins colored for percentage reduction in the silicate weathering flux (100 ·(Na* −Nasil )/Na*) calculated for βNa =0.6
(SI Appendix, Fig. S8 is equivalent calculated for βNa =0.2). Basins in gray have chloride in excess of sodium (Na* <0).
Tipper et al.
Global silicate weathering flux overestimated because of sediment–water cation exchange
PNAS |5 of 6
https://doi.org/10.1073/pnas.2016430118
Downloaded at Cambridge University on December 22, 2020
in some of the samples. Additional samples were reacted with ammonium
chloride (NH4Cl) where the NH+
4ion displaces exchangeable cations. Cation
concentrations were determined by inductively coupled plasma optical
emission spectroscopy or ion chromatography, and anions were determined
by IC. The exchange pool extractions were measured either by ICP-OES
or spectrometric ultraviolet absorbance, using matrix matched calibration
lines. After filtering the data for exchange equilibrium, the CEC and chem-
ical compositions are within uncertainty for NH4Cl and CoHex extractions
(SI Appendix, Figs S1–S3). Radiogenic strontium isotopic compositions were
measured on a Neptune Plus multicollector–ICP–mass spectrometer (Thermo
Scientific, University of Cambridge).
Data Availability. All study data are included in the article and SI Appendix.
ACKNOWLEDGMENTS. This work was funded by National Environmen-
tal Research Council Grants NE/K000705/1, NE/M001865/1, NE/N007441/1,
and NE/P011659/1 to E.T.T., and NERC Arctic Bursary Award & European
Research Council Starting Grant ROC-CO2 678779 to R.G.H. Mackenzie River
samples were collected under research licenses 15288 and 16106. Many peo-
ple assisted in the collection of samples in the field. C. Parish built our
field equipment. H. Chapman conducted some of the Sr isotope chemistry.
S. Souanef-Ureta conducted some of the NH4Cl extractions. M.-L. Bagard
ensured the flawless running of the Cambridge Plasma Labs. The manuscript
was improved by two thoughtful anonymous reviewers.
1. R. A. Berner, Z. Kothavala, Geocarb III: A revised model of atmospheric CO2over
phanerozoic time. Am. J. Sci. 301, 182–204 (2001).
2. R. G. Hilton, A. J. West, Mountains, erosion and the carbon cycle. Nat. Rev. Earth
Environ. 1, 284–299 (2020).
3. J. Gaillardet, B. Dupr ´
e, P. Louvat, C. J. All`
egre, Global silicate weathering and CO2
consumption rates deduced from the chemistry of large rivers. Chem. Geol. 159, 3–30
(1999).
4. G. Sposito et al., Surface geochemistry of the clay minerals. Proc. Natl. Acad. Sci. U.S.A.
96, 3358–3364 (1999).
5. C. A. J. Appelo, Multicomponent ion exchange and chromatography in natural
systems. Rev. Mineral. Geochem. 34, 193–227 (1996).
6. J. A. Andrews, W. H. Schlesinger, Soil CO2dynamics, acidification, and chemi-
cal weathering in a temperate forest with experimental CO2enrichment. Global
Biogeochem. Cycles 15, 149–162 (2001).
7. D. W. Clow, M. A. Mast, Mechanisms for chemostatic behavior in catchments: Impli-
cations for CO2consumption by mineral weathering. Chem. Geol. 269, 40–51
(2010).
8. J. S. Scheingross et al., Preservation of organic carbon during active fluvial transport
and particle abrasion. Geology 47, 958–962 (2019).
9. T. E. Cerling, B. L. Pederson, K. L. Von Damm, Sodium-calcium ion exchange in the
weathering of shales: Implications for global weathering budgets. Geology 17, 552–
554 (1989).
10. F. L. Sayles, P. C. Mangelsdorf Jr, Cation-exchange characteristics of Amazon River sus-
pended sediment and its reaction with seawater. Geochem. Cosmochim. Acta 43,
767–779 (1979).
11. M. Lupker, C. France-Lanord, B. Lartiges, Impact of sediment–seawater cation
exchange on Himalayan chemical weathering fluxes. Earth Surf. Dyn. 4, 675–684
(2016).
12. G. E. Likens, C. T. Driscoll, D. C. Buso, Long-term effects of acid rain: Response and
recovery of a forest ecosystem. Science 272, 244–246 (1996).
13. R. S. Hindshaw, S. M. Aciego, A. M. Piotrowski, E. T. Tipper, Decoupling of dissolved
and bedrock neodymium isotopes during sedimentary cycling. Geochem. Pers. Lett.
8, 43–46 (2018).
14. S. Samanta, T. K. Dalai, Dissolved and particulate Barium in the Ganga (Hooghly)
River estuary, India: Solute-particle interactions and the enhanced dissolved flux to
the oceans. Geochem. Cosmochim. Acta 195, 1–28 (2016).
15. F. L. Sayles, P. C. Mangelsdorf Jr, The equilibration of clay minerals with sea water:
Exchange reactions. Geochem. Cosmochim. Acta 41, 951–960 (1977).
16. P. Michalopoulos, R. C. Aller, Rapid clay mineral formation in Amazon delta sed-
iments: Reverse weathering and oceanic elemental cycles. Science 270, 614–617
(1995).
17. P. Y. Henry, “Relationship between porosity, electrical conductivity, and cation
exchange capacity in Barbados Wedge sediments” in Proceedings of the Ocean
Drilling Program, Scientific Results, T. H. Shipley et al., Eds. (Texas A&M University,
1997). vol. 156, pp. 137–149.
18. R. A. Berner et al., Comment and reply on “Sodium-calcium ion exchange in the
weathering of shales: Implications for global weathering budgets”. Geology 18,
190–191 (1990).
19. M. Dellinger et al., Lithium isotopes in large rivers reveal the cannibalistic nature
of modern continental weathering and erosion. Earth Planet Sci. Lett. 401, 359–372
(2014).
20. E. Garzanti, J. G. Wang, G. Vezzoli, M. Limonta, Tracing provenance and sediment
fluxes in the Irrawaddy River basin (Myanmar). Chem. Geol. 440, 73–90 (2016).
21. X. Li, J. P. Liu, Y. Saito, V. L. Nguyen, Recent evolution of the Mekong Delta and the
impacts of dams. Earth Sci. Rev. 175, 1–17 (2017).
22. T. Bullen, A. White, A. Blum, J. Harden, M. Schulz, Chemical weathering of a soil
chronosequence on granitoid alluvium: II. Mineralogic and isotopic constraints on
the behavior of strontium. Geochem. Cosmochim. Acta 61, 291–306 (1997).
23. A. F. White et al., Chemical weathering of a marine terrace chronosequence, Santa
Cruz, California II: Solute profiles, gradients and linear approximations of con-
temporary and long-term weathering rates. Geochem. Cosmochim. Acta 72, 36–68
(2009).
24. K. L. Von Damm, “Geochemistry of shale groundwaters: Survey of available data and
postulated mineralogic controls on composition” (Tech. Rep. ORNL/TM-10488, US
Department of Energy, 1987).
25. S. Zhang, J. J. Sheng, Z. Qiu, Maintaining shale stability using polyether amine while
preventing polyether amine intercalation. Appl. Clay Sci. 132-133, 635–640 (2016).
26. A. M. Fern ´
andez et al., Applying the squeezing technique to highly consolidated clay-
rocks for pore water characterisation: Lessons learned from experiments at the Mont
Terri rock laboratory. Appl. Geochem. 49, 2–21 (2014).
27. E. Gaucher et al., ANDRA underground research laboratory: Interpretation of the
mineralogical and geochemical data acquired in the Callovian–Oxfordian formation
by investigative drilling. Phys. Chem. Earth 29, 55–77 (2004).
28. L. Jin et al., Mineral weathering and elemental transport during hillslope evolution
at the Susquehanna/Shale Hills Critical Zone Observatory. Geochem. Cosmochim. Acta
74, 3669–3691 (2010).
29. M. Blondes et al., U.S. Geological Survey National Produced Waters Geochemical
Database v2.3. https://doi.org/10.5066/F7J964W8. Accessed 11 December 2020.
30. C. Ma, R. A. Eggleton, Cation exchange capacity of kaolinite. Clay Clay Miner. 47,
174–180 (1999).
31. R. Millot, J. Gaillardet, B. Dupr ´
e, C. J. All `
egre, Northern latitude chemical weathering
rates: Clues from the Mackenzie river basin, Canada. Geochem. Cosmochim. Acta 67,
1305–1329 (2003).
32. J. Bouchez, J. Gaillardet, C. France-Lanord, L. Maurice, P. Dutra-Maia, Grain size
control of river suspended sediment geochemistry: Clues from Amazon River depth
profiles. Geochem. Geophys. Geosyst. 12, 12 (2011).
33. J. L. Guyot et al., Clay mineral composition of river sediments in the Amazon Basin.
Catena 71, 340–356 (2007).
34. K. L. Cook, C. Andermann, F. Gimbert, B. R. Adhikari, N. Hovius, Glacial lake outburst
floods as drivers of fluvial erosion in the Himalaya. Science 362, 53–57 (2018).
35. A. J. West, A. Galy, M. Bickle, Tectonic and climatic controls on silicate weathering.
Earth Planet Sci. Lett. 235, 211–228 (2005).
36. I. J. Larsen, D. R. Montgomery, H. M. Greenberg, The contribution of mountains to
global denudation. Geology 42, 527–530 (2014).
37. K. Maher, C. P. Chamberlain, Hydrologic regulation of chemical weathering and the
geologic carbon cycle. Science 343, 1502–1504 (2014).
38. C. Andermann, A. Crave, R. Gloaguen, P. Davy, S. Bonnet, Connecting source and
transport: Suspended sediments in the Nepal Himalayas. Earth Planet. Sci. Lett. 351-
352, 158–170 (2012).
39. E. T. Tipper et al., The short term climatic sensitivity of carbonate and silicate weather-
ing fluxes: Insight from seasonal variations in river chemistry. Geochem. Cosmochim.
Acta 70, 2737–2754 (2006).
40. R. A. Berner, J. L. Rao, S. Chang, R. O’Brien, C. K. Keller, Seasonal variability of
adsorption and exchange equilibria in soil waters. Aquat. Geochem. 4, 273–290
(1998).
41. M. E. Raymo, W. F. Ruddiman, P. N. Froelich, Influence of late Cenozoic mountain
building on ocean geochemical cycles. Geology 16, 649–653 (1988).
42. A. Galy, C. France-Lanord, Weathering processes in the Ganges-Brahmaputra basin
and the riverine alkalinity budget. Chem. Geol. 159, 31–60 (1999).
43. A. J. West, Thickness of the chemical weathering zone and implications for erosional
and climatic drivers of weathering and for carbon-cycle feedbacks. Geology 40, 811–
814 (2012).
44. G. Bayon et al., Rare earth elements and neodymium isotopes in world river sediments
revisited. Geochem. Cosmochim. Acta 170, 17–38 (2015).
45. J. Gaillardet, B. Dupr ´
e, C. J. All `
egre, Geochemistry of large river suspended sediments:
Silicate weathering or recycling tracer? Geochem. Cosmochim. Acta 63, 4037–4051
(1999).
46. T. Ahmad, P. P. Khanna, G. J. Chakrapani, S. Balakrishnan, Geochemical characteris-
tics of water and sediment of the Indus River, Trans-Himalaya, India: Constraints on
weathering and erosion. J. Asian Earth Sci. 16, 333–346 (1998).
47. V. V. Gordeev, B. Beeskow, V.Rachold, “Geochemistry of the Ob and Yenisey estuaries:
A comparative study” (Tech. Rep. 557, Alfred Wegener Institute, 2007).
48. M. A. Torres, A. J. West, G. Li, Sulphide oxidation and carbonate dissolution as a
source of CO2over geological timescales. Nature 507, 346–349 (2014).
6 of 6 |PNAS
https://doi.org/10.1073/pnas.2016430118
Tipper et al.
Global silicate weathering flux overestimated because of sediment–water cation exchange
Downloaded at Cambridge University on December 22, 2020
