Oxidation of petrogenic organic carbon in the Amazon floodplain as a source of atmospheric CO2
ABSTRACT The two long-term sources of atmospheric carbon are CO2 degassing from metamorphic and volcanic activity, and oxidation of organic carbon (OC) contained in sedimentary rocks, or petrogenic organic carbon (OCpetro). The latter flux is still poorly constrained. In this study, we report particulate organic carbon content and 14C activity measurements in Amazon River sediments, which allow for estimates of the OCpetro content of these sediments. A large decrease of OCpetro content in riverine sediments is observed from the outlet of the Andes to the mouth of the large tributaries. This loss reveals oxidation of OCpetro during transfer of sediments in the floodplain, and results in an escape of ~0.25 Mt C/yr to the atmosphere, which is on the same order of magnitude as the CO2 consumption by silicate weathering in the same area. Raman microspectroscopy investigations show that graphite is the most stable phase with respect to this oxidation process. These results emphasize the significance of OCpetro oxidation in large river floodplains in the global carbon cycle.
GEOLOGY, March 2010
Geology, March 2010; v. 38; no. 3; p. 255–258; doi: 10.1130/G30608.1; 3 fi gures; 1 table.
The two long-term sources of atmospheric carbon are CO2
degassing from metamorphic and volcanic activity, and oxidation of
organic carbon (OC) contained in sedimentary rocks, or petrogenic
organic carbon (OCpetro). The latter fl ux is still poorly constrained.
In this study, we report particulate organic carbon content and 14C
activity measurements in Amazon River sediments, which allow for
estimates of the OCpetro content of these sediments. A large decrease
of OCpetro content in riverine sediments is observed from the outlet
of the Andes to the mouth of the large tributaries. This loss reveals
oxidation of OCpetro during transfer of sediments in the fl oodplain,
and results in an escape of ~0.25 Mt C/yr to the atmosphere, which is
on the same order of magnitude as the CO2 consumption by silicate
weathering in the same area. Raman microspectroscopy investiga-
tions show that graphite is the most stable phase with respect to this
oxidation process. These results emphasize the signifi cance of OCpetro
oxidation in large river fl oodplains in the global carbon cycle.
CO2 degassed from Earth’s interior is partly scavenged by chemical
reactions occurring during weathering of silicate rocks and subsequent
carbonate precipitation in the ocean (Garrels et al., 1976). It is also con-
sumed by photosynthesis followed by burial of organic matter in marine
sediments (Hayes and Waldbauer, 2006). These two mechanisms have
respectively built up the two major carbon reservoirs of Earth’s surface:
limestones (50 × 106 Gt C), and 14C-free organic matter disseminated in
sedimentary rocks, or petrogenic organic carbon (OCpetro, 12.5 × 106 Gt
C; Berner, 1990). The oxidation of petrogenic OC is a source of CO2 to
the atmosphere (Berner, 2004). However, quantifying the modern rates of
OCpetro oxidation remains a challenge for understanding and modeling the
geological carbon and oxygen cycles. Although a few studies based on soil
profi les have attempted to determine rates of OCpetro oxidation (e.g., Keller
and Bacon, 1998, Petsch et al., 2000), budgets of fossil organic carbon
oxidation at river-catchment scale have not received much attention (e.g.,
Galy et al., 2008b; Hilton et al., 2008).
The dissolved and particulate loads transported by rivers derive from
chemical weathering of rocks and physical erosion of soils and rocks.
This includes organic material, which consists of a mixture of recent bio-
spheric carbon (OCrecent), and OCpetro (Blair et al., 2004; Komada et al.,
2004; Leithold et al., 2006). The oxidation of OC transported in rivers is
thought to mostly affect OCrecent and to have no effect on the geological
budget of atmospheric CO2. In their study of the Amazon River, Hedges
et al. (1986) showed that the organic material transported by the Ama-
zon River mostly consists of OCrecent derived from the highly productive
lowland ecosystems. During fl uvial transport, the oxidation of this dis-
solved and particulate organic matter results in the escape of ~500 Mt C/
yr to the atmosphere (Richey et al., 2002). Most of this oxidation derives
from OCrecent (Mayorga et al., 2005), and therefore has no impact on the
long-term regulation of atmospheric CO2. However, a signifi cant fraction
of particulate organic matter can be of petrogenic origin. Distinguishing
between OCpetro and OCrecent in rivers is thus of prime importance because
only the oxidation of OCpetro represents an input of C to the active reser-
voirs at Earth’s surface.
In this study we report particulate organic carbon (POC) and 14C
activity measurements in river sediments collected throughout the Ama-
zon River system. Sediments were collected along river depth profi les
in order to capture the entire range of the granulometric spectrum of
erosion products. Concentrations of OCpetro were measured in these sedi-
ments and coupled with structural characterization of OCpetro by Raman
microspectroscopy. This allows us to estimate the fi rst order of carbon
input to the atmosphere by OCpetro oxidation during transfer of sediments
in the fl oodplains of the Amazon Basin, and gives a lower bound on the
estimate of CO2 release to the atmosphere by the oxidation of OCpetro in
the Amazon Basin.
SETTING, SAMPLING, AND ANALYTICAL METHODS
The Amazon is the world largest river in terms of drainage area and
water discharge to the ocean (Meybeck and Ragu, 1997). Isotopic studies
(Allègre et al., 1996) have clearly shown that most of the Amazon River
sediments are derived from the Andes. There, Amazon tributaries drain
extensive outcrops of easily erodible sedimentary and metasedimentary
rocks, such as black shales in the Bolivian Andes.
We sampled the two main tributaries of the Amazon, the Solimões
and the Madeira Rivers, at their mouth, as well as the Amazon mainstream
at Obidos, in June 2005 and March 2006 (Fig. 1). The Beni River, which
supplies most of the sediments to the Madeira River, was sampled at the
outlet of the Andes, near Rurrenabaque, where it enters the Madeira fl ood-
plain, in February 2001. At each location, river water was sampled at vari-
ous river depths along vertical profi les, from channel surface to bottom,
and fi ltered at 0.22 µm porosity; bed sediments were also dredged. Within
the channel of large rivers, granulometric sorting induces important varia-
tions of chemical composition of river sediments from the surface to the
bottom (Galy et al., 2008a). The sampling technique used here allows us
to characterize the entire range of erosion products in terms of grain size
distribution and mineralogy.
POC content was determined using a modifi ed Eurovector
EuroEA3028-HT elemental analyzer coupled to a GV Instruments
IsoPrime continuous-fl ow isotope mass spectrometer at the Centre de
Recherches Pétrographiques et Géochimiques, Vandoeuvre-lès-Nancy,
France (Galy et al., 2007). 14C activity was determined by accelerator
© 2010 Geological Society of America. For permission to copy, contact Copyright Permissions, GSA, or firstname.lastname@example.org.
Oxidation of petrogenic organic carbon in the Amazon fl oodplain as
a source of atmospheric CO2
Julien Bouchez1,2, Olivier Beyssac3, Valier Galy4, Jérôme Gaillardet1,2, Christian France-Lanord5, Laurence Maurice6,
and Patricia Moreira-Turcq7
1Institut de Physique du Globe de Paris, CNRS-UMR 7154, 4, place Jussieu 75252 Paris cedex 05, France
2Université Paris Diderot, 75205 Paris cedex 13, France
3Laboratoire de Géologie, Ecole Normale Supérieure, CNRS-UMR 8538, 24 rue Lhomond, 75231 Paris cedex 05, France
4Woods Hole Oceanographic Institution, 360 Woods Hole Road, Woods Hole, Massachusetts 02543, USA
5Centre de Recherches Pétrographiques et Géochimiques, CNRS-UPR 2300, BP 20, 54501 Vandoeuvre-lès-Nancy, France
6 Laboratoire des Mécanismes de Tranfert en Géologie, Institut de Recherche pour le Développement, 14 avenue Edouard Belin,
31400 Toulouse, France
7Institut de Recherche pour le Développement, 32 avenue Henri Varagnat, 93140 Bondy, France
256 GEOLOGY, March 2010
mass spectrometry at LMC14 National Facility, Saclay, France, after
off-line organic matter combustion and CO2 cryogenic purifi cation.
Samples were decarbonated before combustion (Galy et al., 2007). 14C
values are given after correction for 13C fractionation (normalization to
a δ13C of −25‰), and expressed as percentage of modern carbon (pMC)
comparative to 95% of the 14C activity of the oxalic acid standard OXI.
Petrogenic carbon was characterized by Raman microspectroscopy
using a Renishaw InVia Raman microspectrometer at the Laboratoire
de Géologie, Ecole Normale Supérieure, Paris, France (Bernard et al.,
2008). Raman spectra were obtained directly on raw sediments, and on
thin sections for bedrocks.
RESULTS: 14C AGE OF OCrecent AND OCpetro CONTENT
In depth profi le sediments, 14C content shows a wide range of varia-
tion, between 37.5 and 86.2 pMC (Table 1). A fi rst-order positive relation
between pMC and POC is observed, the coarser bed sediments being the
most depleted in both OC and 14C.
In river sediments, OC can be interpreted as a binary mixture of
OCpetro and OCrecent with distinct 14C content (e.g., Blair et al., 2004). Fol-
lowing the approach of Galy et al. (2008b), we plot our results in a dia-
gram of POC × pMC as a function of POC (Fig. 2). Depth profi les from
different sampling locations defi ne linear trends, at 95% confi dence level,
regardless of the sampling period. These correlations indicate that samples
from a given depth profi le have fairly constant absolute OCpetro concentra-
tion and 14C activity of the OCrecent component (see Appendix for details).
The values of OCpetro content in the samples and 14C age of the OCrecent pool
can be both determined from the slope and intercept of each line (Table 1).
Similar observations were made on the Ganga-Brahmaputra system (Galy
et al., 2008b).
The highest OCpetro content, 0.26% (±0.11%, 2σ uncertainty), is
obtained for the Beni River at Rurrenabaque. There, OCpetro composes as
much as 50% of the POC, and likely originates from the large outcrops of
black shales drained by this river. Lowland sampling locations (Solimões,
Madeira, and Amazon) all display lower OCpetro content, between 0.02%
and 0.06% (or even lower, regarding the uncertainties reported in Table 1).
Thus, there is a large apparent decrease in OCpetro concentration between
the entry and the outlet of the Madeira fl oodplain.
Because the Beni contributes as much as ~40% of the Madeira sedi-
mentary budget (Guyot et al., 1996), potential addition of supposedly
OCpetro-free sedimentary material by other tributaries of the Madeira River
could only lead to an OCpetro content decrease by a factor of slightly more
than two. Hence, the apparent tenfold decrease in OCpetro content could
mainly be due either to a preferential burial of OCpetro-rich material in the
fl oodplain, or to a loss by oxidation.
Burial of sedimentary material occurs in the Amazon basin between
the Andean source of sediments and the Amazon mouth, in particular
between Rurrenabaque and the mouth of the Madeira (Guyot et al., 1996).
If sediment storage is the cause of the observed decrease of OCpetro con-
centration reported here, it would imply the preferential sedimentation of
an OCpetro-enriched component. As stated above, sampling along depth
profi les allows us to take into account the entire range of riverine particu-
late matter size distribution and mineralogy. Our results (Fig. 2) show that
the absolute OCpetro content is the same along all depth profi les, despite
expected variations in particle size distribution with depth (Curtis et al.,
1979). Selective burial of a given size fraction should therefore not affect
Solimões and Madeira
Modern C (%)
Figure 1. Map of Amazon basin and sampling sites.
Figure 2. Modern carbon content [particulate organic carbon (POC)
× percentage of modern carbon (pMC), expressed in wt% of entire
sample] versus POC for sediments collected along depth profi le in
different rivers of Amazon basin. Linear regression solution for each
sampling location is also shown. Open symbols—bedload sedi-
ments, closed symbols—suspended load sediments. Samples are
plotted regardless of their position in the hydrological cycle.
TABLE 1. SAMPLE LIST AND RESULTS
(r2 = 0.995)
(r2 = 0.998)
(r2 = 0.999)
(r2 = 0.986)
Note: Analytical absolute uncertainties (2σ) are 0.5 m for sampling depth, 0.3%
for percentage of modern carbon (pMC), and 0.02% for particulate organic carbon
GEOLOGY, March 2010 257
OCpetro concentration of suspended sediments. This strongly suggests that
the decrease of OCpetro concentration along the course of the Madeira is
due to OCpetro oxidation.
ESTIMATE OF THE MAGNITUDE OF THE CO2 SOURCE
Given the important amount of sediments transported in the Madeira
fl oodplain (Guyot et al., 1996), the oxidation fl ux resulting from the large
decrease in OCpetro content during the transfer of sediments in the fl ood-
plain should be signifi cant. A fi rst-order oxidation fl ux of OCpetro can be
estimated using previous works on sedimentary budgets in the Madeira
River basin. Among the 212 Mt/yr of sediments delivered by the Beni
River to the plain, approximately one-half is buried in the foreland basin
(Guyot et al., 1996). The amount of Beni sediments actually transiting
through the plain is thus ~100 Mt/yr. Hence, given the OCpetro concentra-
tion reported in this study, 100 Mt/yr of sediments represents an OCpetro
fl ux of 0.26 Mt C/yr supplied to the plain and not buried. At the outlet, 100
Mt/yr of sediments represents an OCpetro fl ux of 0.02 Mt C/yr that exits the
plain. The difference of ~0.25 Mt C/yr is thus the oxidation fl ux of OCpetro
in the Madeira fl oodplain. This is a fi rst-order estimate, but also a lower
bound of the OCpetro oxidation fl ux of the Madeira basin, as we assume
that no OCpetro is delivered to the Madeira fl oodplain by its two other main
tributaries. In addition, this estimate does not take into account the oxida-
tion of OCpetro upstream in Rurrenabaque and in Andean soils that we are
not able to address here. Moreover, we assume that no oxidation affects
the sediments buried in the foreland basin. The fl ux of 0.25 Mt C/yr is
thus a minimum bound of the OCpetro-derived CO2 outgassing fl ux. This
number is in the same order of magnitude as the net CO2 sequestration fl ux
in this basin associated with silicate weathering (0.8 Mt C/yr; Gaillardet
et al., 1997).
OCpetro STRUCTURAL CHARACTERIZATION
OCpetro is derived from organic carbon initially trapped in sediments
and has been structurally and chemically transformed during diagen-
esis and metamorphism. Structural characterization of OCpetro by Raman
microspectroscopy has been performed on both riverine sediments and
bedrock samples. Because volcanic rocks of the high cordillera may not
contain any signifi cant amount of solid OC, the main sources of OCpetro
are most likely the sediments, mainly black shales, drained by the Beni
River. Three samples representative of the main bedrock lithologies from
the Tipuani, Mapiri, and Coroico basins have been investigated (Fig. 1).
They contain two main OCpetro fractions (Fig. 3): one is rather disordered,
exhibiting Raman spectra typical of greenschist facies (Beyssac et al.,
2002), in agreement with the thermal history of these rocks. The sec-
ond is highly graphitic, and supposedly represents a detrital pool. Both
fractions are found in all riverine sediments either as isolated particles
or as inclusions or aggregates within minerals (mostly quartz, phyllosili-
cates, or plagioclases; Fig. 3). As shown in Figure 3, the graphitic phases
become dominant in samples of downstream sediment (Beni River and
then Madeira River), while the disordered fraction progressively disap-
pears. Graphite thus appears to be the most stable phase with respect to
the oxidation process.
DISCUSSION AND CONCLUSION
This study thus shows that the oxidation of OCpetro during fl uvial
transport is a signifi cant fl ux for the long-term atmospheric CO2 budget.
Fluvial oxidation of OCpetro may counteract the consumption fl ux of CO2
by silicate weathering, which is conventionally thought to be the only sig-
nifi cant process, with organic carbon sequestration, to control atmospheric
CO2 at geological time scales (Berner, 2004; Wallmann, 2001). The deg-
radation of physically mobilized ancient organic matter in large fl uvial
systems is probably dependent on a number of factors, such as residence
time of particles in fl oodplains (Blair et al., 2003), or on climatic con-
ditions. The important oxidation fl ux found here is probably favored by
the warm and oxidative conditions that prevail in the soils of Amazonian
fl oodplains. Whether this oxidation of OCpetro occurs via biotic (Petsch et
al., 2001) or abiotic (Chang and Berner, 1999) pathways is beyond the
scope of the paper, but would need further investigations.
Galy et al. (2008b) showed that 30%–50% of the OCpetro present
in the Himalayan source rocks was preserved and is still present in the
marine sediments of the Bengal fan. Our estimate of the OCpetro preserva-
tion in the Madeira fl oodplain, 15%, is an upper bound of the extent of
OCpetro preservation in the Madeira basin, as it does not take into account
the oxidation taking place in Andean soils, downstream of the sampling
locations, or even in the ocean, before or after deposition. The Amazon
basin is hence a better incinerator of OCpetro than the Himalayan system.
This is likely due to differences in the sources of OCpetro. Low-grade meta-
morphic rocks with disordered OC are common in the Andes, while high-
grade metamorphic rocks generating highly graphitic OC are widespread
in the Himalaya (Beyssac et al., 2004). Disordered OC is more prone to
oxidation than graphite because of its chemistry (aromatic skeleton with
radicalization) and structure, as microporosity and nanoporosity enhance
Over geological time scales, geodynamic (e.g., metamorphic grade,
erosion intensity) settings probably control the extent of preservation of
OCpetro during the erosion-transport-sedimentation cycle. Over shorter
Raman shift (cm–1)
Figure 3. Representative Raman spectra of riverine and bedrock ma-
terial, with location of main graphite G band, and D1, D2, and D3
defect bands. Minerals associated with carbon are also indicated.
Fossil organic matter was found as free particles, inclusions in min-
erals such as quartz or rutiles, or aggregates with phyllosilicates.
Free particles were as large as 20 µm in diameter.
258 GEOLOGY, March 2010
time scales (tens to hundreds of thousands of years), and for a given geo-
dynamic context, climate is likely to control the oxidation or preservation
of OCpetro, through erosion, temperature, and probably the nature of micro-
bial communities (and their metabolic activity) present in the fl oodplain.
We speculate that, in response to an atmospheric CO2 rise, increased
global temperature would probably enhance oxidation of petrogenic OC
in large river fl oodplains and associated CO2 outgassing. This newly
described mechanism possibly constitutes a positive feedback in the long-
term carbon cycle.
This study was funded by the Centre National de la Recherche Scientifi que
Institut National des Sciences de l’Univers (CNRS-INSU) program Reliefs de la
Terre, and realized in the frame of the HyBAm project (cooperation agreement with
CNPq 492685/2004-5). We sincerely thank the following Brazilian Institutions and
Universities: Agência Nacional de Águas, Universidade de Brasília, Universidade
Federal Fluminense, and the Serviço Geológico do Brasil. We thank C. Guilmette
for technical assistance in the stable isotopes laboratory and R. Hilton for improving
the quality of the text. We greatly acknowledge R.A. Berner and J.M. Hayes for their
thoughtful reviews.This is Institut de Physique du Globe de Paris contribution 2564.
APPENDIX: OCpetro AND 14C AGE OF OCrecent CALCULATION
We use a method described in Galy et al. (2008b). Briefl y, we describe
the OC pool as a binary mixture of OCpetro, derived from the rocks, and OCrecent
derived from the biosphere (vegetation, soils, and autotrophic production in the
river). These two components have distinct 14C activity, OCpetro being 14C free
(pMCpetro = 0). For each sample, the absolute content of Modern C (POC × pMC;
“modern” referring here to a present 14C standard) can thus be written as:
Modern C = POC × pMCrecent – OCpetro × pMCrecent,
where pMCrecent is the 14C activity of OCrecent and OCpetro is the absolute content
of petrogenic organic carbon. In a Modern C vs. POC plot, samples having the
same pMCrecent and the same %OCpetro defi ne a single straight line. The pMCrecent
is given by the slope of the line and allows the calculation of the age of the recent
component. Moreover, the absolute content of OCpetro is given by the opposite of
the intercept/slope ratio.
Despite the auto-correlated nature of the two plotted variables, and as shown
in Table 1, the relationships we obtain are more signifi cantly correlated than in
the case of randomly distributed POC and pMC (either assuming an uniform
or normal distribution, within boundaries defi ned by the ranges covered by the
values measured in our samples).
Uncertainties on the determined slope and intercept (and thus on %OCpetro
and pMCrecent) are yielded by a full inversion method (Tarantola and Valette,
1982). Relatively low uncertainties on pMCrecent (i.e., on the slope) stem from the
good alignment of data points.
Allègre, C.J., Dupré, B., Négrel, P., and Gaillardet, J., 1996, Sr-Nd-Pb isotope
systematics in Amazon and Congo River systems: Constraints about ero-
sion processes: Chemical Geology, v. 131, p. 93–112, doi: 10.1016/0009-
Bernard, S., Beyssac, O., and Benzerara, K., 2008, Raman mapping using ad-
vanced line-scanning systems: Geological applications: Applied Spectros-
copy, v. 62, p. 1180–1188, doi: 10.1366/000370208786401581.
Berner, R.A., 1990, Atmospheric carbon dioxide levels over Phanerozoic time:
Science, v. 249, p. 1382–1386, doi: 10.1126/science.249.4975.1382.
Berner, R.A., 2004, The Phanerozoic carbon cycle: Oxford, Oxford University
Press, 150 p.
Beyssac, O., Goffé, B., Chopin, C., and Rouzaud, J.-N., 2002, Raman spec-
tra of carbonaceous material in metasediments: A new geothermometer:
Journal of Metamorphic Geology, v. 20, p. 859–871, doi: 10.1046/j.1525
Beyssac, O., Bollinger, L., Avouac, J.-P., and Goffé, B., 2004, Thermal meta-
morphism in the Lesser Himalaya of Nepal determined from Raman spec-
troscopy of carbonaceous material: Earth and Planetary Science Letters,
v. 225, p. 233–241, doi: 10.1016/j.epsl.2004.05.023.
Blair, N.E., Leithold, E.L., Ford, S.T., Peeler, K.A., Holmes, J.C., and Perkey,
D.W., 2003, The persistence of memory: the fate of ancient sedimentary
organic carbon in a modern sedimentary system: Geochimica et Cosmochi-
mica Acta, v. 67, p. 63–73, doi: 10.1016/S0016-7037(02)01043-8.
Blair, N.E., Leithold, E.L., and Aller, R.C., 2004, From bedrock to burial: The
evolution of particulate organic carbon across coupled watershed-continen-
tal margin systems: Marine Chemistry, v. 92, p. 141–156, doi: 10.1016/
Chang, S., and Berner, R.A., 1999, Coal weathering and the geochemical car-
bon cycle: Geochimica et Cosmochimica Acta, v. 63, p. 3301–3310, doi:
Curtis, W.F., Meade, R.H., Nordin, C.F., Price, N.B., and Sholkovitz, E.R., 1979,
Non-uniform vertical distribution of fi ne sediment in the Amazon River:
Nature, v. 280, p. 381–383, doi: 10.1038/280381a0.
Gaillardet, J., Dupré, B., Allègre, C.J., and Négrel, P., 1997, Chemical and physi-
cal denudation in the Amazon River Basin: Chemical Geology, v. 142,
p. 141–173, doi: 10.1016/S0009-2541(97)00074-0.
Galy, V., Bouchez, J., and France-Lanord, C., 2007, Determination of total or-
ganic carbon content and δ13C in carbonate-rich detrital sediments: Geostan-
dards and Geoanalytical Research, v. 31, p. 199–207, doi: 10.1111/j.1751
Galy, V., France-Lanord, C., and Lartiges, B., 2008a, Loading and fate of particu-
late organic carbon from the Himalaya to the Ganga-Brahmaputra delta:
Geochimica et Cosmochimica Acta, v. 72, p. 1767–1787, doi: 10.1016/
Galy, V., Beyssac, O., France-Lanord, C., and Eglinton, T.I., 2008b, Recycling of
graphite during Himalayan erosion: A geological stabilization of carbon in
the crust: Science, v. 322, p. 943–945, doi: 10.1126/science.1161408.
Garrels, R.M., Lerman, A., and Mackenzie, F.T., 1976, Controls of atmospheric O2
and CO2—Past, present and future: American Scientist, v. 63, p. 306–315.
Guyot, J.-L., Filizola, N., Quintanilla, J., and Cortez, J., 1996, Dissolved solids
and suspended sediment yields in the Rio Madeira basin, from the Bolivian
Andes to the Amazon, in Walling, D.E., and Webb, B.W., eds., Erosion and
sediment yield: Global and regional perspectives: International Association
of Hydrological Sciences Publication 236, p. 55–63.
Hayes, J.M., and Waldbauer, J.R., 2006, The carbon cycle and associated redox
processes through time: Royal Society of London Philosophical Transac-
tions, v. 361, p. 931–950, doi: 10.1098/rstb.2006.1840.
Hedges, J.I., Quay, P.D., Grootes, P.M., Richey, J.E., Devol, A.H., Farwell, G.W.,
Schmidt, F.W., and Salati, E., 1986, Organic Carbon-14 in the Amazon River
system: Science, v. 231, p. 1129–1131, doi: 10.1126/science.231.4742.1129.
Hilton, R.H., Galy, A., Hovius, N., Chen, M.-C., Horng, M.-J., and Chen, H.,
2008, Tropical-cyclone-driven erosion of the terrestrial biosphere from
mountains: Nature Geoscience, v. 1, p. 759–762, doi: 10.1038/ngeo333.
Keller, C.K., and Bacon, D.H., 1998, Soil respiration and georespiration distin-
guished by transport analyses of vadose CO2, 13CO2 and 14CO2: Global Bio-
geochemical Cycles, v. 12, p. 361–372, doi: 10.1029/98GB00742.
Komada, T., Druffel, E.R.M., and Trumbore, S.E., 2004, Oceanic export of relict
carbon by small mountainous rivers: Geophysical Research Letters, v. 31,
L07504, doi: 10.1029/2004GL019512.
Leithold, E.L., Bair, N.E., and Perkey, D.W., 2006, Geomorphic controls on
the age of particulate organic carbon from small mountainous and up-
land rivers: Global Biogeochemical Cycles, v. 20, GB3022, doi: 10.1029/
Mayorga, E., Aufdenkampe, A.K., Masiello, C.A., Krusche, A.V., Hedges, J.I.,
Quay, P.D., Richey, J.E., and Brown, T.A., 2005, Young organic matter as a
source of carbon dioxide outgassing from Amazonian rivers: Nature, v. 436,
p. 538–541, doi: 10.1038/nature03880.
Meybeck, M., and Ragu, A., 1997, River discharges to the oceans: An assessment
of suspended solids, major ions and nutrients: Nairobi, Kenya, United Na-
tions Environment Programme, World Health Organization, Environmental
and Assessment Document, 245 p.
Petsch, S.T., Berner, R.A., and Eglinton, T.I., 2000, A fi eld study of chemical
weathering of ancient sedimentary organic matter: Organic Geochemistry,
v. 31, p. 475–487, doi: 10.1016/S0146-6380(00)00014-0.
Petsch, S.T., Eglinton, T.I., and Edwards, K.J., 2001, 14C-dead living biomass:
Evidence for microbial assimilation of ancient organic carbon during shale
weathering: Science, v. 292, p. 1127–1131, doi: 10.1126/science.1058332.
Richey, J.E., Melack, J.M., Aufdenkampe, A.K., Ballester, V.M., and Hess, L.L.,
2002, Outgassing from Amazonian rivers and wetlands as a large tropical
source of atmospheric CO2: Nature, v. 416, p. 617–620, doi: 10.1038/416617a.
Tarantola, A., and Valette, B., 1982, Generalized nonlinear inverse problems
solved using the least squares criterion: Reviews of Geophysics and Space
Physics, v. 20, p. 219–232, doi: 10.1029/RG020i002p00219.
Wallmann, K., 2001, Controls on the Cretaceous and Cenozoic evolution of sea-
water composition, atmospheric CO2 and climate: Geochimica et Cosmo-
chimica Acta, v. 65, p. 3005–3025, doi: 10.1016/S0016-7037(01)00638-X.
Manuscript received 11 August 2009
Revised manuscript received 6 October 2009
Manuscript accepted 7 October 2009
Printed in USA