Articles

https://doi.org/10.1038/s41561-018-0190-9

Microbial life and biogeochemical cycling on land

3,220 million years ago
Martin Homann1*, Pierre Sansjofre1, Mark Van Zuilen2, Christoph Heubeck3, Jian Gong2,
Bryan Killingsworth 1, Ian S. Foster 1, Alessandro Airo4, Martin J. Van Kranendonk 5,
Magali Ader3 and Stefan V. Lalonde 1

The colonization of emergent continental landmass by microbial life was an evolutionary step of paramount importance in Earth
history. Here we report direct fossil evidence for life on land 3,220 million years ago (Ma) in the form of terrestrial microbial
mats draping fluvial conglomerates and gravelly sandstones of the Moodies Group, South Africa. Combined field, petrographic,
carbon isotope and Raman spectroscopic analyses confirm the synsedimentary origin and biogenicity of these unique fossil
mats as well as their fluvial habitat. The carbon isotope compositions of organic matter (δ13Corg) from these mats define a n­ arrow
range centred on −​21‰, in contrast to fossil mats of marine origin from nearby tidal deposits that show δ13Corg values as low
as −​34‰. Bulk nitrogen isotope compositions (2 <​ δ15N <​ 5‰) are also significantly different from their marine counterparts
(0 <​ δ15N <​ 3‰), which we interpret as reflecting denitrification in the terrestrial habitat, possibly of an atmospheric source of
nitrate. Our results support the antiquity of a thriving terrestrial biosphere during the Palaeoarchaean and suggest that a com-
plex and microbially driven redox landscape existed during the deposition of the Moodies Group, with distinct b ­ iogeochemical
cycling occurring on land by 3,220 Ma.

A
lthough there is abundant evidence that microbial life units indicates that deposition began about (3,223 ±​ 1) Ma and had
thrived in the oceans as far back as there is a sedimentary ended by about (3,219 ±​ 9) Ma22,23. The southwestward-plunging
record1–5, much less is known about microbial colonization Dycedale Syncline, about 2 km east of Barberton, hosts a steeply
of the land surface. Before 3,000 Ma, much of the Earth may have dipping succession, >​350 m thick, of Moodies Group conglomer-
been submerged6, and, accordingly, direct fossil evidence for terres- ates and cross-bedded sandstones. A large variety of sedimentary
trial life before the Mesoarchaean era is extremely rare7,8. (The term structures indicates that this succession records a transition from
‘terrestrial’ has multiple definitions; here we follow convention from alluvial-fluvial (terrestrial) to tide-influenced marine sedimenta-
literature on the Precambrian biosphere by considering any life on tion20,21 (Supplementary Fig. 2).
the emerged continental surface, aquatic or subaerial, as terrestrial
(see supplementary text).) The evidence is also inferential, largely Terrestrial microbial mats in fluvial sandstones
derived from the study of palaeosols as much as 3,200 Myr old9–13. Our study is focused on fossilized microbial mats discovered in this
A suite of suggestive biosignatures in hot spring deposits indicates unique terrestrial-to-marine transition in the Dycedale Syncline.
that life may have already been occupying terrestrial niches by The base of the section begins with a sedimentary unit, ~75 m thick,
3,480 Ma14. Here we present the discovery of a new locality in the including a ~40-m-thick, polymict, mostly clast-supported con-
Palaeoarchaean Moodies Group, Barberton Greenstone Belt (BGB), glomerate in the central part (unit B of ref. 22; Supplementary Fig. 2).
South Africa, where exceptionally preserved microbial mats are The poor sorting and angularity of clasts, poorly developed inter-
exposed in sediments of an ancient fluvial system. These terrestrial nal fabrics, clast imbrication, thin intercalated sandstone beds with
fossils represent a considerable expansion of the known diversity upper-plane bed horizontal lamination, and immature composition
of microbial life in the Moodies Group, which until now has been of this conglomerate indicate a proximal sediment source associated
restricted to marine settings15–19. with episodic, short-lived, high-energy unidirectional transport.
The Moodies Group is the uppermost of the three stratigraphic These fabrics and associations are typical for sheet-flow-dominated
units constituting the BGB (Supplementary Fig. 1) and represents the alluvial fans and/or proximal braided streams with highly variable
world’s oldest well-preserved alluvial to shallow-marine deposit20,21. discharge. The conglomerate is under- and overlain by 10- and
It consists of a thick (up to 3.5 km) succession of alluvial to shallow- 25-m-thick (respectively) gravelly sandstones with carbonaceous
marine quartz-rich sandstones with subordinate conglomerates, laminations and minor interbedded conglomerate beds. These lens-
mudstones, thin tuffs, banded iron formations and a single basaltic or wedge-shaped beds are 0.2 to 2 m thick, commonly vertically
lava21. The age of the Moodies Group is tightly constrained by sev- stacked, and show minor channel incision from erosional down-
eral dacitic tuffs and rare felsic dykes radiating from the Kaap Valley cutting, characteristic of fluvial deposition and transport (Fig. 1a).
tonalite that crosscut the Moodies Group along the northern mar- Pebble- to boulder-sized clasts (up to ~40 ×​ 40 cm) are subrounded,
gin of the BGB. Uranium–lead dating of single-zircons from these poorly sorted and embedded in a quartz-rich coarse-sandy matrix

European Institute for Marine Studies, CNRS-UMR6538 Laboratoire Géosciences Océan, Technopôle Brest-Iroise, Plouzané​, France. 2Institut de Physique
1

du Globe de Paris, CNRS-UMR7154, Paris, France. 3Department of Geosciences, Friedrich-Schiller-Universität, Jena, Germany. 4Center of Astronomy
and Astrophysics, Technische Universität Berlin, Berlin, Germany. 5Australian Centre for Astrobiology, and School of Biological, Earth and Environmental
Sciences, University of New South Wales, Sydney, New South Wales, Australia. *e-mail: [email protected]

Nature Geoscience | www.nature.com/naturegeoscience

Articles Nature Geoscience

a b c
e

mm

cgl

15 cm 3 cm
mm d e

mm

cgl cgl

2 cm 3 cm

Fig. 1 | Field photographs of fluvial sandstones and conglomerates hosting fossil terrestrial mats from the 3,220-Myr-old Moodies Group. a, Overview
photograph showing interbedded fossil microbial mats (mm) and conglomerates (cgl). b, Mat-associated fluvial conglomerate, composed of subrounded
pebbles and cobbles. c,d, Microbial mats draping and onlapping interbedded clasts within the sandstones and on top of conglomerate beds (close-up view
of the framed area in a). e, Fluid-escape structure with well-defined central channel (arrow) that vertically disrupts the densely mat-laminated sandstone
(close-up view of the framed area in a).

(Fig. 1b). The transition to overlying gravelly, medium- to coarse- features is thought to be due to a combination of very early silici-
grained quartzofeldspathic sandstones is gradational. These hori- fication, low tectonic strain and the low temperature of post-dep-
zontal to low-angle planar cross-bedded sandstones are locally ositional hydrothermal overprinting (<​150 °C) in the interior part
interbedded with discontinuous mudstones with desiccation cracks, of the BGB18,24. Raman microspectroscopy demonstrates that the
indicating periods of subaerial exposure (Supplementary Fig. 3). carbonaceous laminae of both the terrestrial and marine mats18 are
Overlying strata gradually deepen upward through deltaic, and composed of organic carbon that has experienced similar peak tem-
medium-energy tidal, into subtidal siliciclastic deposits. The posi- peratures of ~365 °C (see Methods), consistent with the metamor-
tion of the conglomerate-bearing deposits at the base of this trans- phic grade of lower greenschist facies established by mineralogical
gressive, fining- and deepening-upward sequence, together with the indicators25 and previous Raman-based estimates of regional peak
absence of any sedimentary structure indicative of tidal or marine metamorphic temperatures for the central part of the BGB26 (Fig. 3,
conditions, further suggests that the gravelly sandstones and con- Supplementary Fig. 6, Supplementary Table 1). This confirms that
glomerates represent a terrestrial depositional environment, prob- the laminae are of syngenetic origin with the sandstone. Based on
ably a fluvial coastal braidplain that was updip of an estuary21. the combined evidence of carbonaceous composition, syngenicity,
The wavy and crinkly carbonaceous laminations within these cohesiveness, sediment trapping behaviour and the presence of fila-
gravelly sandstones and on top of the conglomerate beds show mentous microstructures, the laminae are confidently identified as
various features consistent with a biogenic origin. They are densely the fossilized remnants of microbial mats.
spaced at the millimetre scale, and both onlap and drape protruding The presence of microbial mats on land during the Palaeoarchaean
clasts (Fig. 1c,d). Laminae are bent upwards and plastically deformed provides insights into the timing of certain evolutionary innovations
by 10- to 50-cm-high subvertical fluid-escape structures, indicat- required for terrestrialization. The Archaean land surface was prob-
ing their cohesive water-impermeable nature and synsedimentary ably a harsh environment subject to repeated desiccation, fluvial
origin (Fig. 1e, Supplementary Fig. 4). High strength and cohesive- and/or aeolian abrasion and, presumably, intense ultraviolet radia-
ness of the laminae are further supported by their association with tion. Its colonization suggests that the terrestrial mats possessed a
coarse-grained sandstone and conglomerate beds that were repeat- variety of adaptations, including tolerance to high shear stresses by
edly emplaced on top of the laminae without severely damaging or formation of cohesive and resistive mats, production of hygroscopic
eroding them (Fig. 2a). However, petrographic analysis also reveals extracellular polymeric substance to maintain wetting during sub-
that during periods of increased current velocity, and therefore aerial exposure, synthesis of ultraviolet-screening pigments and an
higher shear stress, laminae were partially eroded, ripped up and enhanced capacity for DNA-repair to cope with cellular damage
reworked as fragments up to several centimetres in length (Fig. 2a–f). induced by desiccation and/or exposure to high-incidence ultravio-
In thin section, the laminae (0.5–4 mm thick) are composed of a let radiation. It seems that terrestrial mats of the Moodies Group
dense meshwork of interwoven filament-like microstructures that already possessed such coping mechanisms at 3,220 Ma.
drape horizontally laminated and rippled sandstones, onlap indi- Our new report of terrestrial mat fossils adds to the known
vidual clasts and envelop ‘floating’ detrital grains of fine-grained diversity of the Moodies ecosystem, which includes large sphe-
sand whose long axes are preferentially aligned parallel to bedding roidal microfossils15, widespread shallow-marine tufted microbial
(Fig. 2a,b). Individual carbonaceous filamentous structures are 1–3 µ​m mats with trapped gas bubbles16–18 and remnants of cavity-dwell-
in diameter and may exceed 100 µ​m in length, commonly bundled ing microbes thriving beneath the mats19. These coeval marine
and twisted around each other (Fig. 2c; Supplementary Fig. 5), microbial communities are preserved in sandstones in the nearby
consistent with, but not exclusive to, modern filamentous micro- Saddleback Syncline that show clear bidirectional palaeocurrent
organisms forming biofilms. The excellent preservation of these patterns characteristic of deposition under tidal influence18.

Nature Geoscience | www.nature.com/naturegeoscience

Nature Geoscience Articles
a
Mat fragment

Ripple

1 cm

Fig. 3a
250 µm

c d e

Fig. 3d

200 µm 200 µm 250 µm

f
e

1 mm

Fig. 2 | Reflected and transmitted light photomicrographs of the terrestrial microbial mats of the Moodies Group. a, Dark carbonaceous laminae of the
fossil mats, draping horizontally laminated and rippled sand and onlapping pebbles. Chips of eroded mat fragments are preserved in cross-laminated,
granular sandstone. b, Dense meshwork of interwoven filamentous microstructures with trapped detrital grains. c, Bundled filamentous structures in upper
part of the mat. d, Close-up view of eroded mat fragment. e, Partially eroded microbial mat laminae due to abrasion by impacting sand grains (arrow).
f, Erosional truncation of the mat by small channel (arrow).

To better characterize and distinguish the palaeobiological con- Carbon fixation in Moodies Group microbial mats
text of these unique terrestrial mats from their marine counter- The carbon isotope composition of preserved organic matter pro-
parts, we subsampled mat-rich horizons from both and analysed vides a more direct link to metabolic activity during mat growth.
them for the carbon isotope composition of organic matter and for In the terrestrial mats, δ13Corg values range between −​23.6‰ and
bulk nitrogen isotope composition. We also examined dolomite −​17.9‰, with a mean of −​21.2‰ (n =​ 36; Supplementary Table 3).
remnants that occur as bladed to blocky cement in millimetre- to These values contrast with isotopically lighter δ13Corg values of
centimetre-sized bedding-parallel cavities beneath the mats18,19 microbial communities from the coeval marine deposits, ranging
in both environments. In places, these carbonates are plasti- between −​33.9‰ and −​21.3‰, with a mean of −​27.4‰ (n =​  30;
cally deforming and rupture the mats, which further indicates Fig. 4). The observed difference between the two datasets is statis-
their early diagenetic formation, before sandstone lithification tically significant (two-tailed Welsh’s t-test, P <​ 0.0001). The ~6‰
(Supplementary Fig. 7). All carbonates yielded homogeneous shift to heavier δ13Corg values in the terrestrial realm thus suggests
mean δ13Ccarb values of (+​0.2 ±​  0.2)‰ and δ18Ocarb values of considerable environmental and metabolic diversity across this
(−​15.4 ±​  0.2)‰ (n =​ 16, Supplementary Table 2), common values Palaeoarchaean ecosystem landscape.
for dolomitic carbonates of Archaean age that rule out significant There are several non-mutually exclusive explanations for this
secondary exchange between carbon pools after burial. shift. First, cell size, growth rate and species-specific differences

Nature Geoscience | www.nature.com/naturegeoscience

Articles Nature Geoscience

a b c
TiO2 rutile

TiO2 anatase

Albite
Quartz

Carbonate

Kerogen

0 500 1,000 1,500 2,000

100 µm Wavenumber (cm–1)

d e f g
TiO2 rutile

TiO2 anatase

Pyrite

Mica

Tourmaline

Quartz

Kerogen

0 500 1,000 1,500 2,000

100 µm Wavenumber (cm–1)

Fig. 3 | Images of the kerogenous laminae and filamentous microstructures. a–f, Transmitted light photomicrographs of preserved kerogenous laminae
(a) and filamentous microstructures (d) of the terrestrial mats, with corresponding Raman component maps for mineral phases and G-peak intensity for
kerogenous phases (b,e,f), and representative Raman component spectra (c,g). Note that the analysed areas are close-up views of the samples shown in
Fig. 2b and c, respectively.

a b
Marine (n = 30) 6
0.35
Terrestrial (n = 36)
5
0.3

0.25 4
Density

0.2 3 δ15N (‰)

0.15 2

0.1 1

0.05 0
1σ
–1
0 –34 –32 –30 –28 –26 –24 –22 –20 –18 –34 –32 –30 –28 –26 –24 –22 –20 –18
δ13Corg (‰) δ13Corg (‰)

Fig. 4 | Carbon isotope composition of organic matter and bulk nitrogen isotope composition from terrestrial and marine microbial mats of the
Palaeoarchaean Moodies Group. a, Histogram of organic carbon δ13Corg. b, δ13Corg versus δ15N values for both environments.

in CO2 diffusion rates all influence εp, the carbon isotopic frac- cycle31,32, whether by oxygenic or anoxygenic phototrophs. Marine
tionation factor associated with phototrophic carbon fixation27,28. samples reach values that are isotopically as heavy, yet cover a larger
However, the influence of these factors on εp tends toward zero28 spread extending to lighter δ13Corg values, some as low as −​34‰.
as the partial pressure of CO2, pCO , approaches the elevated val- These features suggest that in the marine realm, mixing occurred
2
ues inferred for the Archaean29, and some bacterial species exhibit between material with the same isotopic composition as terrestrial
little variation even at low CO2 concentrations30. A more likely samples (−​21‰) and material with carbon that was isotopically
explanation for this shift is a mixing of carbon sources with differ- lighter than −​34‰. Under high pCO , carbon fixed by the Calvin–
2
ent isotopic compositions. The terrestrial samples exhibit a narrow Benson cycle is unlikely to reach such low values33, which are best
distribution in δ13Corg values, suggesting a relatively homogeneous explained instead by biomass derived from other carbon fixation
source centred around −​21‰. This δ13Corg composition is consis- pathways, notably the reductive acetyl-coenzyme A (CoA) pathway
tent with autotrophic carbon fixation through the Calvin–Benson (Wood–Ljungdahl pathway)33–35. This includes acetogenic bacteria,

Nature Geoscience | www.nature.com/naturegeoscience

Nature Geoscience Articles
methanogens and sulfate reducers that, with the exception of some of different diagenetic or metamorphic histories (see, for example,
examples of the latter36, are obligate anaerobes. The terrestrial mat ref. 34). Moreover, δ15N values are resistant to modification dur-
samples are rather remarkable in that the light carbon isotope sig- ing low-grade metamorphism42, in the case of kerogen varying no
nature that should be associated with alternative fixation pathways more than 1‰ for sediments reaching greenschist facies43. The low-
such as the reductive acetyl-CoA pathway is not observed. est C/N ratios observed are probably linked to the presence of clay
We suggest that this disparity reflects differences in fermentative minerals that retain nitrogen produced during diagenesis. Total
or respiratory processes occurring in the mats along a palaeoen- nitrogen contents (12–64 ppm, Supplementary Table 4) show no
vironmental transect. One possibility is that higher sedimenta- significant differences between marine and terrigenous sediments,
tion rates in the terrestrial realm promoted rapid burial of carbon and show no relation to clay content (~17–32% illite and muscovite
formed at the mat surface by the Calvin–Benson pathway, whereas in samples for which X-ray diffraction was performed), suggest-
lower sedimentation rates in marine settings permitted greater ing that variable contributions of nitrogen bound to clay, including
expression of the anaerobic reductive acetyl-CoA pathway at depth allochthonous clay, cannot explain the isotopic contrast between the
in the mat. A related possibility is that distinct microbial commu- two datasets, which are instead most likely to be recording different
nities inhabited these different environments. Indeed, in modern primary compositions of mat biomass.
microbial mats, production of CH4 is independent of diel cycling Three different biological mechanisms are known to produce
between ambient oxic and anoxic water conditions, yet it appears biomass with δ15N compositions greater than +​2‰ (see ref. 45 for a
strongly suppressed in intertidal mats and enhanced in subtidal mats detailed discussion). The first mechanism, and the only one possible
as the result of differences in their anaerobic community structure in the absence of oxidative nitrogen cycling, is partial assimilation of
at depth37. These include differences in the activity of sulfate reduc- NH4+, whereby preferential uptake of 14NH4+ can drive the residual
ers, which outcompete methanogens for organic substrates even at NH4+ toward isotopically heavier values (see, for example, ref. 46).
sulfate concentrations as low as 60 μ​M (ref. 38). The local presence However, only after most of the NH4+ pool has been assimilated
of sulfate in supratidal to braided-fluvial environments during would residual NH4+ achieve values as heavy as +​5‰, and we see no
Moodies Group deposition is indicated by common pseudomor- evidence in our dataset for light isotope enrichments that would indi-
phic relics of ­gypsum39 and isotopic evidence for sulfate reduction in cate this process. The two remaining hypotheses are partial nitrifica-
Moodies Group palaeosols13. However, other possibilities exist that tion and/or partial denitrification, both of which require oxidative
may have resulted in a greater contribution of carbon from pathways nitrogen cycling in the terrestrial mats. Partial nitrification requires a
other than Calvin–Benson, such as the Wood–Ljungdahl pathway, local source of O2 (with or without Mn oxide intermediates) and has
in the marine realm. It has been suggested that hydrogen gas was only been observed to generate such positive values in stratified water
the principal electron donor for photosynthetic mat growth in the bodies where O2 concentrations are highly dynamic as the result of
3,416-Myr-old Buck Reef Chert (Onverwacht Group, also in the seasonal overturning44, which does not apply to the fluvial setting of
BGB)3,40. Anoxygenic phototrophs metabolizing hydrogen using the the terrestrial mats. Finally, partial denitrification of a stable nitrate
reverse tricarboxylic acid or 3-hydroxypropionate CO2 fixation path- pool is the process that is most commonly evoked for the genera-
ways are characterized by carbon isotope compositions that tend to tion of isotopically heavy δ15N compositions in organic matter42,44 and
be heavier than −​14‰ (see review in ref. 36 and references therein), would also appear to be the most parsimonious explanation for the
for which no evidence is observed in our dataset. However, if locally isotopically heavy δ15N compositions of the terrestrial mats.
abundant, hydrogen might have been important in stimulating The source of nitrate to the terrestrial mat ecosystem may
anaerobic metabolism through the Wood–Ljungdahl pathway. Both have been atmospheric. Prebiotic generation of fixed nitro-
oxygenic and anoxygenic phototrophs may themselves produce con- gen (NO−, NO2− and NO3−) in the atmosphere by lightning dis-
siderable quantities of hydrogen gas via bidirectional hydrogenases, charge at pO <​  10−5 present atmospheric level is estimated to be
2
and under conditions of nitrogen limitation, this may also occur (2–4) ×​  1011 g N per year45,46, which translates to a global surface
through a nitrogenase-catalysed side reaction41. In the terrestrial flux of 0.1–0.2 μ​g N m−2 d−1. Although this flux may have been too
realm, rapid burial, the increased availability of sulfate and/or fixed diffuse to be a significant source of fixed nitrogen to the marine
nitrogen, and a depressed role for hydrogen are all plausible explana- biosphere47, fluvial mats would have had access to fixed nitrogen
tions for the contrasting carbon pools preserved in the mats, but it is that is integrated over a larger area by surface runoff. We calcu-
difficult to draw further inferences from carbon isotope data alone. late that the drainage area required to supply an atmospheric fixed
nitrogen flux to mats at a rate comparable to that of nitrogen fixa-
Isotopic insight into nitrogen cycling at 3,220 Myr tion by modern intertidal microbial mats (6–79 mg N m−2 d−1)48 is
Bulk nitrogen isotopic compositions of mat samples also record a only 0.02–0.62 km2. In other words, rainout of fixed nitrogen on the
contrast between the two palaeoenvironments that points to differ- early land surface should have had a profound influence on nitrogen
ences in mat community structure and respiratory processes. The cycling in early terrestrial ecosystems, one that appears expressed in
δ15N values of marine mats range between −​0.7‰ and +​3.1‰, the contrasting nitrogen isotope compositions of microbial mat bio-
with an average of +​1.8‰ (n =​ 10), in contrast to terrestrial δ15N mass between terrestrial and marine settings in the Moodies Group.
values, which are generally more positive, ranging between +​1.9‰ Our observations join those from the slightly younger
and +​5.6‰ with a mean of +​4.3‰ (n =​  10; Fig. 4, Supplementary (~3,000 Myr) Mesoarchaean fluvio-lacustrine Lalla Rookh sandstone
Table 4). The δ15N values of the two sample sets are statistically dif- (Western Australia) where a contrast in the carbon isotope compo-
ferent, even if the two marine data points that are lowest in δ15N sition of organic matter has also been observed33, albeit relative to
are considered as outliers (two-tailed Welsh’s t-test, P <​  0.002). marine sediments of similar age from other localities, whereas our
Whereas the marine samples show near-zero values consistent with comparison is made on approximately coeval sediments from the
atmospheric nitrogen fixation, values of up to +​5‰ in the terres- same basin. In the Lalla Rookh sandstone, the carbon isotopic con-
trial samples are outside the range of typical fractionations asso- trast occurred in the reverse sense (with more important δ13C deple-
ciated with growth on atmospheric N2. Ratios of C/N range from tion in lacustrine sediments) and without any evidence for oxidative
8 to 60 and show no covariation with δ15N values; similar to peak nitrogen cycling. Nonetheless, the ensemble of emerging evidence
metamorphic temperatures determined by Raman spectroscopy indicates that microbial communities already inhabited terrestrial
(Supplementary Table 1), C/N ratios show no significant differences surface environments, and fundamentally differed from their marine
between terrestrial and marine mat samples (Supplementary Fig. 8), counterparts in their biogeochemical cycling of carbon and nitrogen,
suggesting that the ­differences observed in δ15N are not the ­expression at the dawn of continental emergence around 3,220 Ma.

Nature Geoscience | www.nature.com/naturegeoscience

Articles Nature Geoscience

Methods 26. Tice, M. M., Bostick, B. C. & Lowe, D. R. Thermal history of the 3.5–3.2 Ga
Methods, including statements of data availability and any asso- Onverwacht and Fig Tree Groups, Barberton greenstone belt, South Africa,
inferred by Raman microspectroscopy of carbonaceous material. Geology 32,
ciated accession codes and references, are available at https://doi. 37 (2004).
org/10.1038/s41561-018-0190-9. 27. Popp, B. N. et al. Effect of phytoplankton cell geometry on carbon isotopic
fractionation. Geochim. Cosmochim. Acta 62, 69–77 (1998).
Received: 29 March 2018; Accepted: 25 June 2018; 28. Hayes, J. M., Strauss, H. & Kaufman, A. J. The abundance of in marine
Published: xx xx xxxx organic matter and isotopic fractionation in the global biogeochemical cycle
of carbon during the past 800 Ma. Chem. Geol. 161, 103–125 (1999).
29. Driese, S. G. et al. Neoarchean paleoweathering of tonalite and metabasalt:
References Implications for reconstructions of 2.69 Ga early terrestrial ecosystems and
1. Nutman, A. P., Bennett, V. C., Friend, C. R. L., Van Kranendonk, M. J. & paleoatmospheric chemistry. Precambrian Res. 189, 1–17 (2011).
Chivas, A. R. Rapid emergence of life shown by discovery of 3,700-million- 30. Laws, E. A., Popp, B. N., Cassas, N. & Tanimoto, J. 13C discrimination
year-old microbial structures. Nature 537, 535–538 (2016). patterns in oceanic phytoplankton: likely influence of CO2 concentrating
2. Allwood, A. C., Walter, M. R., Kamber, B. S., Marshall, C. P. & Burch, I. W. mechanisms, and implications for palaeoreconstructions. Funct. Plant Biol.
Stromatolite reef from the Early Archaean era of Australia. Nature 441, 29, 323–333 (2002).
714–718 (2006). 31. Schidlowski, M. A 3,800-million-year isotopic record of life from carbon in
3. Tice, M. M. & Lowe, D. R. Photosynthetic microbial mats in the 3,416-Myr-old sedimentary rocks. Nature 333, 313–318 (1988).
ocean. Nature 431, 549–552 (2004). 32. Eigenbrode, J. L. & Freeman, K. H. Late Archean rise of aerobic microbial
4. Wacey, D., Kilburn, M. R., Saunders, M., Cliff, J. & Brasier, M. D. Microfossils ecosystems. Proc. Natl Acad. Sci. USA 103, 15759–15764 (2006).
of sulphur-metabolizing cells in 3.4-billion-year-old rocks of Western 33. Stüeken, E. E. & Buick, R. Environmental control on microbial diversification
Australia. Nat. Geosci. 4, 698–702 (2011). and methane production in the Mesoarchean. Precambrian Res. 304,
5. Dodd, M. S. et al. Evidence for early life in Earth’s oldest hydrothermal vent 64–72 (2018).
precipitates. Nature 543, 60–64 (2017). 34. Slotznick, S. P. & Fischer, W. W. Examining Archean methanotrophy.
6. Flament, N., Coltice, N. & Rey, P. F. The evolution of the 87Sr/86Sr of marine Earth Planet. Sci. Lett. 441, 52–59 (2016).
carbonates does not constrain continental growth. Precambrian Res. 229, 35. Havig, J. R., Hamilton, T. L., Bachan, A. & Kump, L. R. Sulfur and carbon
177–188 (2013). isotopic evidence for metabolic pathway evolution and a four-stepped Earth
7. Beraldi-Campesi, H. Early life on land and the first terrestrial ecosystems. system progression across the Archean and Paleoproterozoic. Earth Sci. Rev.
Ecol. Process. 2, 1 (2013). 174, 1–21 (2017).
8. Wellman, C. H. & Strother, P. K. The terrestrial biota prior to the origin of 36. Baumgartner, L. K. et al. Sulfate reducing bacteria in microbial
land plants (embryophytes): a review of the evidence. Palaeontology 58, mats: changing paradigms, new discoveries. Sediment. Geol. 185,
601–627 (2015). 131–145 (2006).
9. Watanabe, Y., Martini, J. E. & Ohmoto, H. Geochemical evidence for 37. Hoehler, T. M., Bebout, B. M. & Des Marais, D. J. The role of microbial mats
terrestrial ecosystems 2.6 billion years ago. Nature 408, 574–578 (2000). in the production of reduced gases on the early Earth. Nature 412,
10. Rye, R. & Holland, H. D. Life associated with a 2.76 Ga ephemeral pond? 324–327 (2001).
Evidence from Mount Roe #2 paleosol. Geology 28, 483–486 (2000). 38. Lovley, D. R. & Klug, M. J. Sulfate reducers can outcompete methanogens
11. Crowe, S. A. et al. Atmospheric oxygenation three billion years ago. Nature at freshwater sulfate concentrations. Appl. Environ. Microbiol. 45,
501, 535–538 (2013). 187–192 (1983).
12. Mukhopadhyay, J. et al. Oxygenation of the Archean atmosphere: new 39. Nabhan, S., Luber, T., Scheffler, F. & Heubeck, C. Climatic and geochemical
paleosol constraints from eastern India. Geology 42, 923–926 (2014). implications of Archean pedogenic gypsum in the Moodies Group (<​3.2 Ga),
13. Nabhan, S., Wiedenbeck, M., Milke, R. & Heubeck, C. Biogenic Barberton Greenstone Belt, South Africa. Precambrian Res. 275,
overgrowth on detrital pyrite in ca. 3.2 Ga Archean paleosols. Geology 44, 119–134 (2016).
763–766 (2016). 40. Tice, M. M. & Lowe, D. R. Hydrogen-based carbon fixation in the earliest
14. Djokic, T., Van Kranendonk, M. J., Campbell, K. A., Walter, M. R. & Ward, C. R. known photosynthetic organisms. Geology 34, 37 (2006).
Earliest signs of life on land preserved in ca. 3.5 Ga hot spring deposits. 41. Bandyopadhyay, A., Stöckel, J., Min, H., Sherman, L. A. & Pakrasi, H. B. High
Nat. Commun. 8, 15263 (2017). rates of photobiological H2 production by a cyanobacterium under aerobic
15. Javaux, E. J., Marshall, C. P. & Bekker, A. Organic-walled microfossils in conditions. Nat. Commun. 1, 139 (2010).
3.2-billion-year-old shallow-marine siliciclastic deposits. Nature 463, 42. Ader, M. et al. Interpretation of the nitrogen isotopic composition of
934–938 (2010). Precambrian sedimentary rocks: assumptions and perspectives. Chem. Geol.
16. Noffke, N., Eriksson, K. A., Hazen, R. M. & Simpson, E. L. A new window 429, 93–110 (2016).
into Early Archean life: microbial mats in Earth’s oldest siliciclastic tidal 43. Ader, M. et al. Nitrogen isotopic evolution of carbonaceous matter during
deposits (3.2 Ga Moodies Group, South Africa). Geology 34, 253 (2006). metamorphism: methodology and preliminary results. Chem. Geol. 232,
17. Heubeck, C. An early ecosystem of Archean tidal microbial mats (Moodies 152–169 (2006).
Group, South Africa, ca. 3.2 Ga). Geology 37, 931–934 (2009). 44. Stüeken, E. E. A test of the nitrogen-limitation hypothesis for retarded
18. Homann, M., Heubeck, C., Airo, A. & Tice, M. M. Morphological adaptations eukaryote radiation: nitrogen isotopes across a Mesoproterozoic basinal
of 3.22 Ga-old tufted microbial mats to Archean coastal habitats (Moodies profile. Geochim. Cosmochim. Acta 120, 121–139 (2013).
Group, Barberton Greenstone Belt, South Africa). Precambrian Res. 266, 46. Kasting, J. F. & Walker, J. C. G. Limits on oxygen concentration in the
47–64 (2015). prebiological atmosphere and the rate of abiotic fixation of nitrogen.
19. Homann, M. et al. Evidence for cavity-dwelling microbial life in 3.22 Ga tidal J. Geophys. Res. 86, 1147 (1981).
deposits. Geology 44, 51–54 (2016). 47. Navarro-gonz, R., Molina, M. J. & Molina, L. T. Nitrogen fixation
20. Eriksson, K. A., Simpson, E. L. & Mueller, W. U. An unusual fluvial to tidal by volcanic lightning in the early Earth. Geophys. Res. Lett. 25,
transition in the mesoarchean Moodies Group, South Africa: a response to 3123–3126 (1998).
high tidal range and active tectonics. Sediment. Geol. 190, 13–24 (2006). 48. Stüeken, E. E., Kipp, M. A., Koehler, M. C. & Buick, R. The evolution
21. Heubeck, C. et al. Geological constraints on Archean (3.22 Ga) coastal-zone of Earth’s biogeochemical nitrogen cycle. Earth Sci. Rev. 160,
processes from the Dycedale Syncline, Barberton Greenstone Belt. South Afr. 220–239 (2016).
J. Geol. 119, 495–518 (2016). 49. Joye, S. B. & Paerl, H. W. Nitrogen cycling in microbial mats—rates and
22. De Ronde, C. E. J. & Kamo, S. L. An Archaean arc–arc collisional event: a patterns of denitrification and nitrogen-fixation. Mar. Biol. 119,
short-lived (ca 3 Myr) episode, Weltevreden area, Barberton greenstone belt, 285–295 (1994).
South Africa. J. Afr. Earth Sci. 30, 219–248 (2000).
23. Heubeck, C. et al. Timing of deposition and deformation of the Moodies
Group (Barberton Greenstone Belt, South Africa): very-high-resolution of Acknowledgements
Archaean surface processes. Precambrian Res. 231, 236–262 (2013). This work was greatly supported by LabexMER ANR-10-LABX-19 and Prestige
24. Farber, K., Dziggel, A., Trumbull, R. B., Meyer, F. M. & Wiedenbeck, M. COFUND-GA-2013-609102 to M.H., and Deutsche Forschungsgemeinschaft (DFG)
Tourmaline B-isotopes as tracers of fluid sources in silicified Palaeoarchaean grant He2418/13–1 to C.H. S.V.L. and M.V.Z. acknowledge support from the European
oceanic crust of the Mendon Formation, Barberton greenstone belt, South Research Council (ERC) under the European Union’s Horizon 2020 research and
Africa. Chem. Geol. 417, 134–147 (2015). innovation programme (grant agreement n° 716515 for S.V.L. and grant agreement
25. Xie, X., Byerly, G. R. & Ferrell, R. E. Jr. IIb trioctahedral chlorite from the n° 646894 for M.V.Z.). We thank N. and D. Oosthuizen for access to the private
Barberton greenstone belt: crystal structure and rock composition constraints Mountainlands nature reserve, S. Bläsing and M. Grund for assistance with sample
with implications to geothermometry. Contrib. Mineral. Petrol. 126, collection, J.-P. Oldra for thin section preparation, and O. Lebeau, C. Tanvet, C. Liorzou,
275–291 (1997). M.-L. Rouget and B. Gueguen for assistance with isotopic and elemental analysis.

Nature Geoscience | www.nature.com/naturegeoscience

Nature Geoscience Articles
Author contributions Additional information
M.H. and C.H. carried out fieldwork and collected samples in South Africa. P.S., M.A. Supplementary information is available for this paper at https://doi.org/10.1038/
and S.V.L. helped with the acquisition and interpretation of elemental and isotopic data. s41561-018-0190-9.
M.V.Z. and J.G. performed Raman analysis. B.K., I.S.F., A.A. and M.J.V.K. contributed to the Reprints and permissions information is available at www.nature.com/reprints.
discussion of the data. M.H. wrote the manuscript with contributions from all co-authors.
Correspondence and requests for materials should be addressed to M.H.
Competing interests Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in
The authors declare no competing interests. published maps and institutional affiliations.

Nature Geoscience | www.nature.com/naturegeoscience

Articles Nature Geoscience

Methods results are reported in delta notation against the V-PDB standard (Vienna Pee
Optical microscopy. Standard 30-μ​m-thick, polished thin sections, oriented Dee Belemnite) with an average analytical error of 0.12‰ (2σ; Supplementary
perpendicular to bedding, were analysed with an Olympus BX60 petrographic Table 3). For C and O isotope composition of the mat-associated carbonates, CO2
microscope and a Zeiss Axio Scope.A1 equipped with a ×​63 oil objective lens. was released from powdered samples by reaction with 100% H3PO4 at 72 °C in a
High-resolution scans of entire thin sections were performed with a Zeiss Axio Kiel IV automated carbonate preparation device. The CO2 was analysed using for
Zoom v16 motorized stereo microscope at the Institut de Physique du Globe de isotope compositions with a Finnigan MAT 253 mass spectrometer. Ratios 18O/16O
Paris (IPGP) in Paris. and 13C/12C were also expressed in delta notation relative to the V-PDB standard
(Supplementary Table 2). Precision for δ18O and δ13Ccarb was 0.2‰ (2σ) and 0.1‰
Raman spectroscopy. Raman analyses were performed using a Renishaw InVia (2σ), respectively.
Raman microscope coupled to an Olympus BX61 confocal microscope, within
the PARI analytical platform at the IPGP. Measurements were made with a 514- Nitrogen isotope analysis. Individual carbonaceous mat laminae were manually
nm excitation (Ar-ion laser) and adjusted to an on-sample intensity of 0.2 mW separated from fresh and unweathered sandstone samples (as described above),
with a spot size of <​2 µ​m (×​50 objective). Beam centring and Raman spectral ground in an automated agate mill grinder and sieved to ensure a grain size
calibration were performed on a pure silicon chip with a specific Raman band smaller than 140 µ​m. Bulk rock δ15N measurements were performed, as these
at 520.4 cm−1. All spectra were detected using a grating with 1,800 lines per have been suggested to be more likely to record the primary isotopic composition
millimetre and a detector configuration in streamspot mode, providing a spectral of the original biomass in kerogen-poor greenschist-facies metasediments than
range of 2,000 cm−1 in static mode. Individual spot analyses were obtained in measurements of kerogen isolates52. To concentrate nitrogen in the insoluble
both static mode (2 ×​ 20 s exposure, centred at 1,150 cm−1 with a spectral range residue, the samples were first decarbonated in HCl 6 M for 12 h overnight at room
of 100–2,000 cm−1) and extended mode (1 ×​ 20 s exposure, spectral range 100– temperature, followed by 2 h at 80 °C in a fume hood. Residual material was rinsed
4,000 cm−1). To determine Raman spectral indicators of the carbonaceous fractions, three times with Milli-Q water, then centrifuged and dried at 50 °C overnight.
the individual spectra were truncated to 900–1,900 cm−1, and a linear background Approximately 500–1,400 mg of powdered samples were analysed following the
subtraction was performed, using the program Wire 2.0. Peak decomposition method detailed in ref. 53. Conventional sealed-tube combustion with CuO2 and
was performed using two generally reported methods: a two-peak fit, assigning Cu rods (but in this study without CaO grains) was used to convert total nitrogen
a D-peak at approximately 1,350 cm−1 and a G-peak at approximately 1,600 cm−1 to N2 (Dumas combustion), which was then purified using a secondary vacuum
(following the procedure outlined in ref. 49); and a four-peak fit, assigning a D1-, extraction line as shown in figure 1 of ref. 54. Nitrogen isotope ratio measurements
D2-, D3- and G-peaks at about 1,350 cm−1, 1,620 cm−1, 1,500 cm−1 and 1,600 cm−1, for N2 were performed with a dual-inlet Thermo-Fisher Delta V Plus mass
respectively (following the procedure outlined in ref. 49; Supplementary Fig. 6). spectrometer at the IPGP in Paris (Supplementary Table 4). Each purified nitrogen
In all obtained spectra, a D4-peak at approximately 1,200 cm−1 was absent and gas sample was analysed twice. Nitrogen blanks were lower than 0.1 micromoles,
was therefore not assigned during the peak decomposition procedure. Spectral thus representing less than 10% of the measured nitrogen. External reproducibility
data of the decomposed peaks (position, width, height, area) were recorded and of the δ15Nsed measurements was ±​0.4‰ (1σ).
used for calculating the Raman indicators D-FWHM (D-peak, full width at half
maximum) and ID/IG (height-based D/G) for the two-peak fit, and D1-FWHM, Data availability. All data supporting the study are available within the article and
R1 (height-based D1/G) and R2 (area-based D1/D1 +​  D2 +​ G) for the four-peak fit. its Supplementary Information file.
Two geothermometers could then be calculated, that of ref. 50 using temperature
T =​  −​445R2 +​ 641, and that of ref. 51 using T =​  (−​2.15 ×​  D1-FWHM) +​  478 (see
Supplementary Table 1). Raman hyperspectral maps were obtained for selected References
areas within the thin sections 14-452-1B2 and 14-452-1B9, in streamspot mode 45. Papineau, D. et al. High primary productivity and nitrogen cycling after the
(point-by-point scanning) using 1 ×​ 6 s or 1 ×​ 10 s exposures per point. Raman Paleoproterozoic phosphogenic event in the Aravalli Supergroup, India.
maps were generated using the software Wire 2.0 by selecting representative Precambrian Res. 171, 37–56 (2009).
spectra for each component (minerals and kerogen) within the hyperspectral 50. Sforna, M. C., van Zuilen, M. A. & Philippot, P. Structural characterization by
dataset (Fig. 3), and subsequent component analysis for each spectral point was Raman hyperspectral mapping of organic carbon in the 3.46 billion-year-old
determined using a background subtraction with a second-order polynomial fit. Apex chert, Western Australia. Geochim. Cosmochim. Acta 124,
Maps of individual components were subsequently merged using the imaging 18–33 (2014).
program ImageJ. 51. Beyssac, O., Goffé, B., Chopin, C. & Rouzaud, J. N. Raman spectra of
carbonaceous material in metasediments: a new geothermometer. J.
Carbon isotope analysis. For isotope analysis of organic carbon, large rock Metamorph. Geol. 20, 859–871 (2002).
samples (up to 50 ×​ 20 cm) were collected in the field from the freshest and 52. Kouketsu, Y. et al. A new approach to develop the Raman carbonaceous
least weathered outcrops available. In the laboratory, all outer surfaces of the material geothermometer for low-grade metamorphism using peak width. Isl.
rocks were removed with a rock saw, and the rocks were cut into smaller blocks, Arc 23, 33–50 (2014).
devoid of any cracks or fractures. Mat horizons, 2–10 mm thick, were then cut 53. Stüeken, E. E., Zaloumis, J., Meixnerová, J. & Buick, R. Differential
out individually with a thin rock saw and broken into smaller pieces, cleaned metamorphic effects on nitrogen isotopes in kerogen extracts and bulk rocks.
in an ultrasonic bath and dried. The resultant material was crushed into a fine Geochim. Cosmochim. Acta 217, 80–94 (2017).
powder using an automated agate mill grinder, which was cleaned with pure 54. Ader, M., Boudou, J.-P., Javoy, M., Goffe, B. & Daniels, E. Isotope study on
quartz sand, distilled water and ethanol between each run. The powders were organic nitrogen of Westphalian anthracites from the Western Middle field of
decarbonated with 6 M HCl for 12 hours, then warmed at 80 °C for 2 hours in a Pennsylvania (U.S.A.) and from the Bramsche Massif (Germany). Org.
fume hood. Residues were rinsed with Milli-Q water, then centrifuged three to Geochem. 29, 315–323 (1998).
four times until they approached a neutral pH. Decarbonated samples (10–50 mg) 55. Li, L., Cartigny, P. & Ader, M. Kinetic nitrogen isotope fractionation
were loaded into tin capsules and analysed for their carbon isotopic composition associated with thermal decomposition of NH3: experimental results and
at the Pôle Spectométrie Océan (PSO, Brest) using a Thermo Scientific Delta potential applications to trace the origin of N2 in natural gas and
V Plus mass spectrometer coupled to a Flash 2000 elemental analyser. Isotopic hydrothermal systems. Geochim. Cosmochim. Acta 73, 6282–6297 (2009).

Nature Geoscience | www.nature.com/naturegeoscience