1
The Nitrogen Cycle in the Late Archean Ocean
Linda V. Godfrey1 and Paul G. Falkowski1,2
1
Institute of Marine and Coastal Sciences, 2Department of Earth and Planetary Sciences,
Rutgers University, New Brunswick, NJ, 08901, USA
We measured the isotopic composition of nitrogen in kerogen (15Nkerogen)
extracted from minimally altered, organic-rich shales from the Campbellrand-Malmani
platform in South Africa formed in the Late Archean (2,700 to 2,400 Ma). Average
15Nkerogen values are 2.9‰ ± 1.0‰, which suggests that loss of 14N may have occurred
through coupled nitrification-denitrification or anammox pathways, implying the
presence of molecular oxygen in the upper ocean. There appears to be relatively little
change in 15Nkerogen through most of the core, suggesting that during the Late Archean,
low levels of oxygen were present in the upper ocean. There is a moderate increase in
15Nkerogen in sediments deposited between ~2520 and ~ 2460 Ma within the Klein Naute
Formation as water depth increased prior to the deposition of the Kuruman banded iron
formation, indicating an increase in the availability of oxygen 100 myr prior to the
disappearance of the mass-independent isotopic fractionation signal of sulfur. The
increase in 15Nkerogen suggests that denitrification during the late Archean could have led
to severe N-limitation of primary production and carbon burial, and hence the nitrogen
cycle in the Late Archean Ocean was potentially a major bottleneck that prevented
oxygen from accumulating in Earth’s atmosphere.
2
INTRODUCTION
N2 and O2 are, by far, the two most abundant gases in Earth’s atmosphere and the
fluxes of both elements are mediated almost entirely by microbial reactions. N2 is a
relatively inert molecule and has an atmospheric lifetime on the order of ~ 1 billion
years1. In contrast, O2, which has come to comprise 10 to 30% of the volume of the
atmosphere over the past ~500 million years 1; 2, is highly reactive and must be produced
continuously by the photobiological oxidation of water. It is unlikely that the gas was
present above trace levels in the atmosphere of Earth during the first two billion years of
the planet’s history, but when oxygenic photosynthesis first arose on Earth is not known
with certainty3 4 5 6 7. The disappearance of mass independent fractionation of sulfur
isotopes from the geological record (e.g., 8 9) and retention of Fe in paleosol weathering
profiles 10 suggests that by 2,300 - 2,200 Ma the accumulation of O2 in the atmosphere
was >10-5 PAL (present atmospheric level). Molecular fossil evidence suggests that
oxygenic photosynthetic organisms (i.e., cyanobacteria) might have been present and
producing O2 at 2,700 Ma, several hundred million years prior to the so-called “Great
Oxidation Event” (GOE)3,6 circa 2,300 Ma. While it is possible that the molecular
biomarkers for cyanobacteria have more diverse origins than originally thought4, it is
also possible that any O2 produced was consumed within the ocean or troposphere before
it escaped to the stratosphere, or initiated a feedback that limited the productivity of
photosynthetic organisms. In this paper, we explore the interaction between the N and O
cycles around 2,500 Ma at the close of the Archean.
3
Assuming the early Archean oceans were devoid of O2, the nitrogen cycle would
have been relatively simple; N2 would have been biologically reduced to NH4+, which, in
turn, would have become incorporated into organic matter. NH4+ would have either been
recycled in the water column as cells died (there were no grazers) or deposited in
sediments and buried as Norg or trapped in silicates as NH4+ (Figure 1). During
subduction, sedimentary N may be trapped in melts or returned to the atmosphere
predominantly as N2 since thermal N fixation requires atmospheric O2 to produce NOx 11.
This cycle is ultimately driven by tectonics and would have taken millions of years to
complete.
The evolution of oxygenic photosynthesis allowed the N-cycle to completely
close within the ocean on much shorter time scales (Figure 1). Once O2 became
available, chemoautotrophic organisms utilizing NH4+ could sequentially oxidize NH4+ to
NO2- and NO3-. Under suboxic conditions these two oxidized forms can be used by
facultative anaerobic heterotrophs (denitrifiers) to oxidize Corg leading to the production
of N2 (with trace fluxes of N2O). Alternatively, anammox bacteria consume NO2- to
anaerobically oxidize NH4+ to N2 (without the production of N2O). Regardless, N2
produced by heterotrophic denitrification or the anammox reaction is ultimately returned
to the atmosphere, short-circuiting tectonic processing. In the contemporary ocean the
oxidation of NH4+ is strictly biologically catalyzed by microbes that have an absolute
requirement for O2.
4
Nitrification is divided into two sequential steps, separating at NO2-. Members of
the Archaea and Bacteria can oxidize NH4+, but the bacteria responsible for NH4+
oxidation are not closely related to the NO2- oxidizers12 13. This phylogenetic separation
suggests that a long period of time could have elapsed between oxidation of NH4+ to
NO2- and oxidation of NO2- to NO3-. If correct, we propose that the closure of the Ncycle might have first occurred with the formation of NO2- via the anammox reaction, and
later, with the formation of NO3- when more oxygen became available14, via “classical”
heterotrophic denitrification 13. If the net production of O2 is low, an imbalance between
nitrification and denitrification can lead to rapid loss of fixed inorganic N, which, in turn
can potentially lead to N limitation of primary production15. Once pO2 reaches ~15 M
or more15, the rate of nitrification exceeds that of denitrification, and both NO3- and O2
can accumulate.
The organisms, whose organic molecular remnants form the kerogen we analyzed,
acquired N in the upper ocean. There are two isotopic end-members for the N sources. If
the primary source of fixed N was NH4+ , supplied either by local fixation or from
isopycnal mixing from the anoxic interior, the isotopic composition would be similar to
N2. If, however, O2 was present in the upper ocean, subsequent nitrificationdenitrification reactions would have led to preferential return of 14N to the atmosphere16.
Furthermore a redoxcline would have formed where a slightly more oxidized surface
ocean overlay an anoxic interior. Under such conditions the organic matter would
become isotopically enriched in 15N. Hence, to first order, the sedimentary record of N
5
isotope compositions of organic matter reflects the redox state of in the upper ocean, and
changes in 15N reflect the evolution of redox state in the upper ocean 17 18.
We analyzed samples for isotopes of N in the bulk rock, in extracted kerogen, and
in silicate interlayers and structural sites from a core (denoted GKP01) drilled close to the
distal edge of the Campbellrand-Malmani platform. The core contains siliclastic material,
carbonates and banded iron formations deposited before, during and following the
drowning of the Campbellrand-Malmani platform; the youngest samples from the core
just predate the Great Oxidation Event (GOE). The core is described in detail by
Schroeder et al. 19, and dated sections are detailed in Anbar and Beukes et al.(in press).
The 15N composition of bulk sediments increase from ~+2 ‰ to +10 ‰ towards the top
of the core, which at first consideration suggests increasing interaction between the
production of O and the N-cycle, or a “whiff of O2” 20, causing loss of 14N through
coupled nitrification-denitrification reactions. However, 15N values in kerogen and
fixed sites of siliclastics indicate that there almost certainly was a post-depositional
addition of N enriched in 15N which is only weakly bound to surface sites. This process
has also been described in Greenland from metasediments in the Isua formation21. The
isotopic system was effectively closed, and hence, we infer that the isotopic composition
of kerogen bound N is a reliable indication of the redox state of the upper ocean during
the late Archean.
The values of 15Nkerogen range from +0.5 to +7.3‰ within the 250 million year
window recorded by the core, with an average value of 3.2‰ ± 1.5‰ (Figure 2). Nearly
6
all samples that are more than 1 standard deviation from the mean either show evidence
of thermal degradation of organic matter, based on elevated 13Corg, or are associated
with volcanic tuffs or tuff zones. The importance of tuff zones are two fold: (1) they
have increased porosity and therefore are susceptible to post-depositional alteration via
fluid flow, and (2) because the ionic radii of NH4+ and K+ are similar, the two ions can
exchange, and potentially affect 15N 22. If these samples are excluded, the 15Nkerogen
average and standard deviation remain almost the same, 2.9‰ ± 1.0‰.
Secular trends in 15N
A trend of increasing kerogen 15N from the early Archean to the mid-Proterozoic
is apparent despite a fairly wide range of up to ±5‰ within a single location and
stratigraphic age which could reflect outcrop heterogeneities arising from postdepositional events (Figure 3). Between 3,500 and 3,250 Ma in the Paleoarchean the
isotopic values are approximately 0 ‰23-25. At the base of GKP01, dated at 2,670 Ma,
15Nkerogen averages +2.9 ‰. The near 0 ‰ values of early Archean organic matter are
consistent with an N-cycle operating within a reducing environment and an isotopic
composition of fixed inorganic N reflecting atmospheric N2. In order to explain the
modest increase in 15Nkerogen ~ 2,670 Ma, there must have been some loss of 14N from
the upper ocean. Assuming that the isotopic signal of kerogen-bound N reflects that of
the organic matter produced at the time, the only known mechanism to explain such the
isotopic enrichment is denitrification, a process which requires the production of free
molecular O214. Although exactly when oxygenic photosynthesis first evolved is not
known with certainty3-6,26, our results suggest that it probably arose before 2,670 Ma. It
7
is unlikely that the extremely high 15N values for bulk N between 2,700 and 2,750 Ma,
reported and explained by Jia and Kerrich27, reflect the nitrogen isotope composition of
fixed N available for uptake by marine organisms. Those studies were limited to areas
affected by mineralization events and it is more probable that high values of 15N resulted
from devolatilization of 14N during the invasion of hydrothermal fluids 28 rather than
surface fixed nitrogen that was incorporated into organic matter 28.
Short-term changes in 15N and their relation to environment and oxygen.
The consistently elevated 15N values found in kerogen through most of the core
suggest that the ocean inventory of fixed N was stable for long periods of time during the
last few hundred million years of the Archean. There are a small number of samples in
GKP01 that are more than 1 standard deviation from the average 15Nkerogen which are not
associated with tuffs and do not show chemical signs of thermal alteration from elevated
values of 13Corg. If we assume that consumption of available fixed inorganic N is
complete, changes in the 15N value of organic matter formed in the surface of an ocean
reflects the proportion and 15N value of fixed N assimilated. Consequently, changes in
15N can record when oxygenic photosynthesis influences the N-cycle and coupled
nitrification-denitrification leads to loss of 14N. However, until the N isotope signature of
organic matter is transferred to the deep ocean by sinking and reminineralization of
organic matter, there would be a difference between the N isotope composition of fixed
inorganic N between the upper ocean and the interior, with the greatest change across a
redox boundary at the base of the upper mixed layer. If this redox boundary occurred
within the euphotic zone, N could be assimilated from the more oxidized surface layer
8
where dissolved N was lower in concentration but had higher 15N values, or from the
reduced deep ocean where NH4+ was stable, at higher concentration, and but with lower
15N values. We first address values of 15N more than 1 standard deviation below the
average.
Two samples with 15Nkerogen 2‰ lower than the core-average value of 2.9‰
occur in the Monteville Formation, deposited between 2650 and 2588 Ma 2930, at 1029
and 1008 m either side of an ash layer. Similarly low 15Nkerogen values of 0.9 ± 1.0‰ in
these samples have been obtained from the Oaktree Formation in the Transvaal sub-basin
25
, which corresponds to the lowermost to middle Reivilo Formation30,31. The Monteville
and Oaktree Fms represent transgressive events across the Kaapvaal craton30,31 and it is
possible that this geological process allowed increased mixing of deep water, containing
isotopically lighter sources of N, with the upper ocean. We cannot constrain this process
through time because thermal degradation, indicated by higher d13C values of organic
matter in this section of the core, appears to have altered the values of 15Nkerogen.
Samples with relatively high values 15Nkerogen of +4.5 and +5‰ are found at 292
and 267 m in the Klein Naute Formation, corresponding to deposits formed shortly after
2,521±3 to 2,465±4 Ma 32-34. Viewed as a separate unit, there is an increase of about 2.5
‰ during its deposition. The Klein Naute Formation consists of mudstones, and the
absence of detrital carbonate places it in an environment of deeper water than earlier
formations represented in GKP01, well below storm wave base. During deposition of the
Klein Naute Fm and into the Kuruman Fm chert content increases, reflecting increased
water depth and lower sedimentation rates occur while 15Nkerogen first increases towards
the top of the Klein Naute Formation, and then decreases. These trends suggest a
9
temporal rather than lithologic control on the N-cycle during this period, unless there was
a marked increase in ventilation of isotopically distinct N from the ocean interior to the
surface layers.
The increase in 15N indicates that the N-cycle shifted from the steady state
conditions indicated from the average value of +2.9‰ through most of the core. The
initial, small, positive value indicates that coupled nitrification/denitrification reactions
were superimposed on a background of reduced N. The latter N sources were fixation in
the upper ocean and upwelling from the anoxic interior. That the isotopic composition
of organic N remained relatively stable for 150 to 200 million years strongly suggests that
the N cycle was relatively balanced, with little change in the inventory of dissolved
inorganic N species. The increase in 15Nkerogen in the Klein Naute Formation implies an
increase in denitrification in the surface ocean. This, in turn, requires an increase in O2.
A net accumulation of O2 in the upper ocean does not appear to have been translated to
an atmospheric signal. It is likely that the concentration of O2 was extremely low, and if a
small portion of the gas evaded to the troposphere, it was almost certainly chemically
reduced. Whatever the specific mechanisms, our data suggest that oxygenic
photosynthesis arose much earlier than the loss of the mass independent isotopic
fractionation of sulfur 9, which is presumed to be a proxy for the presence of stratospheric
ozone 35
At the start of deposition of the Kuruman BIF, two samples have lower values of
15N, suggesting that conditions favoring denitrification and loss of N reversed. This
could have been due to N-limitation of oxygenic photosynthesis or an increase in the flux
of Fe(II) into the surface ocean scavenging O2. All values of 15N higher than +5‰
10
appear to be in samples that are post-depositionally altered, based on a large difference
between kerogen and bulk 15N or exhibiting high 13Corg indicative of thermal
devolatilization processes, and hence are not indicative of increased denitrification.
Since there is no reason why anaerobic Fe oxidation should cease, it seems likely, given
the available data, that enhanced volcanism may have led to a source of reductants
(including Fe(II)) that consumed free O2.
CONCLUSIONS
The isotopic record of N in kerogen in the Late Archean is consistent with the
presence of free O2 produced by oxygenic photosynthesis in the upper ocean. These
results, combined with records from the Phanerozoic, indicate that 15N of organic matter
can be used to track redox changes in the upper ocean through time ocean 17 18. Further,
this conclusion is consistent with other, independent, geochemical proxies from South
Africa and Australia20,26,36. Regardless, the net flux of O2 from the surface ocean to the
atmosphere must have been trivial. Indeed, the atmospheric concentration of the O2
appears to have been insufficient to produce ozone8,37. The record of 15N we obtained
from the Campbellrand-Malmani platform suggests that the N cycle approached quasi
steady-state conditions throughout the 250 myr interval recorded in the core. However, as
the platform drowned, 15N values suggest an increase in denitrification, the cause of
which is unknown. Increased denitrification implies an increase in the production of O2.
Indeed, the data from GKP01 fall on a trend of increasing 15N values that continued to
11
the middle of the Proterozoic. Whether there are sharp changes marking abrupt changes
in oxygen production, or whether N isotopes record oxygenation with multiple chemical
systems buffering the increase in pO2 waits to be revealed.
MATERIAL AND METHODS
Samples for nitrogen and organic carbon isotope analyses were obtained from
drill core GKP01, which recovered more than 1300 m of Transvaal Supergroup sediment
from the Griqualand West sub-basin in South Africa. Nitrogen isotope ratios were
measured on total (“bulk” N) sediments, refractory organic molecules extracted as
kerogen, and clays (“fixed” N) in order to address post-depositional alterations in the N
isotopic signatures which always present a problem in interpretation of the N cycle from
isotopic analyses. Low temperature regional heating in Griqualand West of 110 to 170º
C38 is unlikely to have caused devolatilization of rock-bound N, but a compression driven
fluid flow event around 2,140 Ma39 40 or later circulation of deep groundwater may have
altered 15N; measurement of 15N in different phases addresses these concerns.
Samples were ground in a ceramic ball mill and used for the bulk measurements
or as starting material for extractions of fixed N and kerogen. Fixed N, i.e., NH4+ or
small organic molecules trapped in layered silicates, was measured for concentration and
isotope composition after surface adsorbed N and organic matter was removed by
oxidation with KOBr in a 2N KCl solution following the method described by Silva and
Bremner41. Nitrogen bound in organic material (kerogen) was isolated from
exchangeable or fixed NH4+ by digesting 200 mg of decalcified sample in 50 mL
polypropylene centrifuge tubes in 20 mL of 3N HCl- 15N HF at 50ºC. The samples were
12
centrifuged, the acid poured off and the samples rinsed with 0.5N KCl three times to
remove any remaining acid or re-adsorption of NH4+. The kerogen was floated off in
chloroform on to precombusted GF/F filters which were dried and segments taken with a
clean razor blade for N-analysis. For analysis of 13Corg, samples were decarbonated in
silver capsules with 2N HCl. The majority of the isotopic analyses were performed using
a GVI Isoprime CF-IRMS system; a subset were also analyzed using the off-line Dumas
combustion method and a Finnigan-MAT 252 dual inlet mass spectrometry to ensure that
potential kinetic effects of combustion using a continuous flow method did not affect our
measurement of N isotopes.
Acknowledgements: This research was supported by a grant from the Agouron
Foundation and the NASA Exobiology program under grant NNX7AK14G. We thank
Gray Bebout and Long Li for their help in analyzing samples, and Woodward Fischer,
Andrew Knoll, Edward Stolper, Nic Beukes and Joseph Kirschvink for the discussions
and comments.
13
FIGURE CAPTIONS
Figure 1. Hypothesized nitrogen cycle for the late Archean. Upwelling of ocean interior
NH4+, N-fixation and efficient N recycling support surface productivity. Nitrogen fluxes
in a totally anoxic ocean are indicated by red arrows. The only sink of ocean-N is loss to
sediments. Once oxygenenic photosynthesis starts, N can exist in higher oxidation states
(fluxes indicated by blue arrows). Coupled nitrification-denitrification allows N to
escape from the ocean as N2 which causes an increase in remaining 15Nfixed as wells as a
decrease in the ocean N-inventory. If the rates of oxygen production are sufficiently
high, nitrification rates exceed those of denitrification, and partially denitrified nitrate
with high 15N can accumulate and be available for assimilation. The fluxes associated
with anammox (indicated by green arrows) are included. The isotope signature of surface
ocean processes is carried by sinking organic matter to sediments.
Figure 2. Changes in the isotopic composition of nitrogen in kerogen, N fixed in layered
silicate, bulk sediments and K/Na ratios over the last 200 million years of the Archean
recorded in core GKP01. Potassium and Na data are from Schröder et al. (2006).
Samples that have not been thermally altered or are associated with volcanic tuffs
average 2.9‰ ± 1.0‰. Samples from the Monteville Formation with low 15Nkerogen may
do so due to incomplete Nfixed assimilation when flooding occurred over the Kaapvaal
craton bringing a new supply of N to this location. There is an increase in 15Nkerogen
across the Klein Naute Formation, a time of platform drowning. The increase in
15Nkerogen at this time reflects an enrichment of 15N in the surface ocean due to coupled
14
nitrification-denitrification reaction and loss of 14N to the atmosphere. The decrease in
15Nkerogen into the Kuruman banded iron formation suggests that the availability of O2 for
interaction with the N-cycle could reflect rates of O consumption by Fe(II) oxidation, or
that N loss during the Klein Naute led to N-limitation of oxygenic photosynthetic
organisms. Dates for the units are: Ventersdorp Supergroup 2710 – 2700 Ma42; Vryburg
Fm. 2669 ± 529; Lockammona Fm. 2650 ± 829; lower Nauga Fm. 2588 ± 6 Ma30; upper
Nauga above Kamden Mb. 2552 ± 11 Ma and 2549 ± 7 Ma34,43; upper Nauga below
Klein Naute Fm. 2521 ± 3; 2516 ± 432,34; Kuruman Fm. 2460 ± 5 Ma33.
Figure 3. The long-term change in 15Nkerogen through the Precambrian showing the
gradual increase in 15Nkerogen that reflects the increased interaction between the O and Ncycles. Data from this study and Beaumont and Robert23, Hayes et al.24, Jia and
Kerrich27, Yamaguchi25.
15
REFERENCES
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
R.A. Berner, Geology 34, 413 (2006).
P.G. Falkowski, Katz, M.E., Milligan, A.J., Fennel, K., Cramer, B.S., Aubry,
M.P., Berner, R.A., Novacek, M.J., Zapol, W.M., Science 309, 2202 (2005).
J.J. Brocks, Buick., R., Logan, G.A., Summons, R.E., Geochimica et
Cosmochimica Acta 67, 4289 (2003).
SE. Rashby, Sessions, AL., Summons, RE., Newman, DK., Proceedings of the
National Academy of Science 104, 15099 (2007).
J.L. Kirschvink, Kopp, R.E., Philosophical Transactions of the Royal Society B
363, 2755 (2008).
P.G. Falkowski, Knoll A.H. ed., The Evolution of Aquatic Photoautotrophs.
(Academic Press, New York, 2007).
B. Rasmussen, I. R. Fletcher, J. J. Brocks et al., Nature 455 (7216), 1101 (2008).
J. Farquhar, Bao, H., Thiemens, M., Science 289, 756 (2000).
A. Bekker, H. D. Holland, P. L. Wang et al., Nature 427 (6970), 117 (2004).
R. Rye, Holland, HD., American Journal of Science 298, 621 (1998).
T. A. Mather, A. G. Allen, B. M. Davison et al., Earth and Planetary Science
Letters 218 (1-2), 17 (2004).
A. Schramm, Geomicrobiology Journal 20, 313 (2002).
DJ Arp, PSG Chain, and MG Klotz, Ann. Rev. Microbiol. 61, 503 (2007).
P.G. Falkowski, Nature 387, 272 (1997).
K. Fennel, Follows, M., Falkowski, P.G., American Journal of Science 305, 526
(2005).
K.L. Casciotti, Sigman, D.M. Ward, B.B., Geomicrobiology Journal 20, 335
(2003); J.D. Cline, Kaplan, I.R., Marine Chemistry 3, 271 (1975); Y. Yoshida,
Nature 307, 528 (1988).
PG Falkowski and LV Godfrey, Proc. Roy. Soc. B 363, 2705 (2008).
T. M. Quan, B. van de Schootbrugge, M. P. Field et al., Global Biogeochemical
Cycles 22 (2) (2008); H. C. Jenkyns, A. Matthews, H. Tsikos et al.,
Paleoceanography 22 (3) (2007).
S. Schröder, Lacassie, J.P., Beukes, N.J., South African Journal of Geology 109,
23 (2006).
A.D. Anbar, Duan, Y., Lyons, T.W., Arnold, G.L., Kendall, B., Creaser, R.A.,
Kaufman, A.J., Gordon, G.W., Scott, C., Garvin, J., Buick, R., Science 317, 1903
(2007).
D. Papineau, S. J. Mojzsis, J. A. Karhu et al., Chemical Geology 216 (1-2), 37
(2005).
G. E. Bebout, Earth and Planetary Science Letters 151 (1-2), 77 (1997).
V. Beaumont, Robert, F., Precambrian Research 96, 63 (1999).
J.M. Hayes, Kaplan, I.R., Wedeking, K.M., in Earth's Earliest Biosphere: Its
Origin and Evolution, edited by J. W. Schopf ( Princeton University Press,
Princeton, 1983), pp. 93.
K. Yamaguchi, PhD, Pennsylvania State University, 2002.
J.R. Waldbauer, Sherman, L.S., Sumner, D.Y.,Summons, R.E., Precambrian
Research in press (2008).
16
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
Y. Jia, Kerrich, R., Geochemistry Geophysics Geosystems 5, doi:
10.1029/2004GC000716 (2004).
N. Dauphas, Marty, B., Earth and Planetary Science Letters 225, 441 (2004).
J. Gutzmer and NJ. Beukes, 1998.
D.M. Martin, Clendenin, C.W., Krapez, B., Mcnaughton, N.J., Journal of the
Geological Society 155, 311 (1998).
D.Y. Sumner, Beukes, N. J., South African Journal of Geology 109, 11 (2006).
W. Altermann and D. R. Nelson, Journal of Sedimentary Geology 120, 225
(1998).
A.L. Pickard, Precambrian Research 125, 275 (2003).
D.Y. Sumner, Bowring, S.A., Precambrian Research 79, 25 (1996).
J. Farquhar, H. M. Bao, and M. Thiemens, Science 289 (5480), 756 (2000).
A.J. Kaufman, Johnston, D.T., Farquar, J., Masterson, A.L., Lyons, T.W., Bates,
S., Anbar, A.D., Arnold, G.L., Garvin, J., Science 317, 1900 (2007).
S. Ono, Wing, B., Johnston, D., Farquhar, J., Rumble, D., Geochimica et
Cosmochimica Acta 70, 2238 (2006).
T. Miyano, and Beukes, N.J., Transactions of the Geological Society of South
Africa 87, 111 (1984).
D.H. Cornell, Schutte, S.S., Eglington, B.L., Precambrian Research 79, 101
(1996); J. Gutzmer, Banks, D.A., Lüders, V., Hoefs, J., Beukes, N.J., von
Benzing, K.L., Chemical Geology 201, 37 (2003).
A. Bekker, C. Holmden, N. J. Beukes et al., Earth and Planetary Science Letters
271 (1-4), 278 (2008).
J.A. Silva, Bremner, J.M., Proceedings of the Soil Science Society of America 30,
587 (1966).
R.A. Armstrong, Compston, W., Retief, E.A., Williams, I.S., and Welke, H.J.,
Precambrian Research 53, 243 (1991).
E.S. Barton, Altermann, W., Williams, I.S., Smith, C.B., Geology 22, 343 (1994).