Content uploaded by Keyron Hickman-Lewis
Author content
All content in this area was uploaded by Keyron Hickman-Lewis on Sep 13, 2020
Content may be subject to copyright.
DIVERSE COMMUNITIES OF BACTERIA AND
ARCHAEA FLOURISHED IN PALAEOARCHAEAN
(3.5–3.3 GA) MICROBIAL MATS
by KEYRON HICKMAN-LEWIS
1,2
, FRANCES WESTALL
2
and
BARBARA CAVALAZZI
1,3
1
Dipartimento Biologiche, Geologiche e Ambientali, Universit!
a di Bologna, via Zamboni 67, I-40126, Bologna, Italy; keyron.hickman-lewis@ cnrs-orleans.fr
2
CNRS Centre de Biophysique Mol"
eculaire, Rue Charles Sadron, 45071, Orl"
eans, France
3
Department of Geology, University of Johannesburg, PO Box 524, Auckland Park, 2006, Johannesburg, South Africa
Typescript received 9 April 2020; accepted in revised form 15 July 2020
Abstract: Limited taxonomic classification is possible for
Archaean microbial mats and this is a fundamental limitation
in constraining early ecosystems. Applying Fourier transform
infrared spectroscopy (FTIR), a powerful tool for identifying
vibrational motions attributable to specific functional groups,
we characterized fossilized biopolymers in 3.5–3.3 Ga micro-
bial mats from the Barberton greenstone belt (South Africa).
Microbial mats from four Palaeoarchaean horizons exhibit
significant differences in taxonomically informative aliphatic
contents, despite high aromaticity. This reflects precursor
biological heterogeneity since all horizons show equally
exceptional preservation and underwent similar grades of
metamorphism. Low methylene to end-methyl (CH
2
/CH
3
)
absorbance ratios in mats from the 3.472 Ga Middle Marker
horizon signify short, highly branched n-alkanes interpreted
as isoprenoid chains forming archaeal membranes. Mats from
the 3.45 Ga Hooggenoeg Chert H5c, 3.334 Ga Footbridge
Chert, and 3.33 Ga Josefsdal Chert exhibit higher CH
2
/CH
3
ratios suggesting mostly longer, unbranched fatty acids from
bacterial lipid precursors. Absorbance ratios of end-methyl to
methylene (CH
3
/CH
2
) in Hooggenoeg, Josefsdal and Foot-
bridge mats yield a range of values (0.20–0.80) suggesting
mixed bacterial and archaeal architect communities based on
comparison with modern examples. Higher (0.78–1.25) CH
3
/
CH
2
ratios in the Middle Marker mats identify Archaea. This
exceptional preservation reflects early, rapid silicification pre-
venting the alteration of biogeochemical signals inherited
from biomass. Since silicification commenced during the life-
time of the microbial mat, FTIR signals estimate the affinities
of the architect community and may be used in the recon-
struction of Archaean ecosystems. Together, these results
show that Bacteria and Archaea flourished together in Earth’s
earliest ecosystems.
Key words: Precambrian, Barberton greenstone belt, in-
frared spectroscopy, microbial mat, geobiology, biosignature.
PALAEOARCHAEAN microbial mats preserved in chert are
the oldest and most robust evidence for life on Earth.
The biological affinities and metabolisms of their architect
organisms are, however, subject to enduring debate (Nis-
bet & Fowler 1999; Tice & Lowe 2006; Bosak et al. 2007,
2010). They are assumed to have been anoxygenic and
photosynthetic (Tice & Lowe 2006; Tice 2009; Noffke
2010; Noffke et al. 2013; Westall et al. 2011; Homann
et al. 2015, 2018; Trower & Lowe 2016; Hickman-Lewis
et al. 2018a) but their phylogenetic placement and most
aspects of their palaeobiology remain unconstrained
because the Archaean fossil record is altered by processes
of maturation and metamorphism, at times susceptible
to ambiguity in interpretation, and lacks modern
environmental analogues. This lack of biological speci-
ficity is a fundamental limitation for constraining the nat-
ure of the earliest ecosystems. In this contribution, we
study taxonomically indicative signals using transmission
Fourier transform infrared spectroscopy (FTIR) analyses
of multiple microbial mat horizons (c. 3.47–3.33 Ga;
Figs 1, 2) from the Barberton greenstone belt, southern
Africa. We evaluate statistically significant differences in
the compositions of these carbonaceous microbial hori-
zons to distinguish aliphatic (and aromatic) groups
indicative of precursor biomolecules that provide infor-
mation about the biological affinities of the architect
microbial communities. The objective of this study is, for
the first time, to place unambiguous taxonomic and
©The Palaeontological Association doi: 10.1111/pala.12504 1
[Palaeontology, 2020, pp. 1–27]
phylogenetic constraints on the organisms inhabiting
Earth’s earliest ecosystems. Our finding that microbial
mats in Palaeoarchaean rocks preserve spectral vestiges of
their diverse architect communities represents a major
step toward reconstructing the ecologies of the earliest
ecosystems and the metabolic networks that regulated
Archaean biogeochemical cycling.
FTIR as a tool for palaeobiological characterization
FTIR is a vibrational spectroscopy technique that irradi-
ates samples with polychromatic IR light and quantifies
changes in dipole moment during molecular vibrational
motion (e.g. stretching, bending and rotating). The
absorption of IR light during these motions allows the
FIG. 1. A, generalized stratigraphic column of the ~3.5–3.3 Ga Onverwacht Group, Barberton greenstone belt, indicating the position
of the four studied horizons: the Middle Marker horizon (MM), Hooggenoeg Chert H5c (HC), Footbridge Chert (FC) and Josefsdal
Chert (JC); inset shows sampling localities within the context of the Barberton greenstone belt and Kaapvaal craton; adapted from
Byerly et al. (2018) and Homann et al. (2015). B, Middle Marker horizon outcrop (samples MM1 and MM2). C, Hooggenoeg Forma-
tion H5c chert outcrop (samples HC1 and HC2). D, Footbridge Chert outcrop (sample FC1). E, Josefsdal Chert outcrop (samples JC1
and JC2). Scale bars in cm; lens cap approx. 50 mm wide. Colour online.
2PALAEONTOLOGY
identification of atomic bonding in such chemical species
as molecules, functional groups and cation–anion pairs
(Benning et al. 2004; Chen et al. 2015; Olcott Marshall &
Marshall 2015). Among the most powerful uses of FTIR
is its ability to characterize cell compounds and struc-
tures, thereby taxonomically differentiating organisms
according to their dominant biomolecules (Naumann
et al. 1991, 1996; Helm & Naumann 1995; Choo-Smith
et al. 2001). Bacteria and Archaea, the two domains of life
relevant to deep time palaeobiology, have strikingly differ-
ent cell envelope compositions: bacterial membrane lipids
are composed of ester lipids (e.g. fatty acids), whereas
archaeal lipids are composed of ether lipids (e.g. iso-
prenoids) (Atlas 1989; Albers & Meyer 2011). These
chemical differences are identifiable by FTIR, as are other
features characteristic of either of these domains, for
instance the proteinaceous S-layer and glycosylated extra-
cellular proteins that are associated with Archaea (Helm
& Naumann 1995; Igisu et al. 2012).
In light of these potentials, FTIR has been demon-
strated as an effective means for non-destructive chemical
characterization of Phanerozoic organic materials (Gupta
et al. 2007; Steemans et al. 2010; Fraser et al. 2012; Hack-
ley et al. 2017) and Precambrian organic microfossils
(Marshall et al. 2005; Igisu et al. 2006, 2009, 2018; Javaux
& Marshall 2006; Qu et al. 2015). Reviews of the applica-
tion of FTIR to geological and palaeontological specimens
are given in Chen et al. (2015) and Olcott Marshall &
Marshall (2015). Transmission FTIR on thin sections (e.g.
Igisu et al. 2009, 2018; this study) is an in situ IR
FIG. 2. Petrographic characterization of microbial mat horizons in Palaeoarchaean biolaminites. A–B, Middle Marker horizon;
A, MM2; B, MM1. C–F, Hooggenoeg Formation H5c chert: C, E, HC1; D, F, HC2. G, Footbridge Chert; FC1. H, Josefsdal Chert; JC1.
Mats exhibit deformation fabrics resulting from the in vivo plasticity of EPS-bound microbial mats and are preserved in three dimen-
sions, a testament to the rapidity of their silicification. Scale bars represent: 4 mm (A); 1 mm (B); 300 lm (C, D, G, H); 50 lm (E,
F). Colour online.
HICKMAN-LEWIS ET AL.: DIVERSITY IN ARCHAEAN MICROBIAL MATS 3
approach that mitigates the potential contamination by
volatile hydrocarbons in laboratory air inherent to studies
of powdered samples (e.g. Derenne et al. 2008). In
micropalaeontological specimens, comparison with pub-
lished spectral libraries may be used to identify, for exam-
ple, functional groups attached to the molecular
framework of carbonaceous materials (Bellamy 1954;
Socrates 1980; Painter et al. 1985; Lin & Ritz 1993; Mar-
shall et al. 2005; Igisu et al. 2009; Steemans et al. 2010).
IR spectroscopy is generally applied to minimally altered
materials of low thermal maturity (Arouri et al. 1999;
Marshall et al. 2005; Steemans et al. 2010; Olcott Mar-
shall & Marshall 2015; Hackley et al. 2017), however,
studies on older palaeontological materials of higher
metamorphic grade (Igisu et al. 2006, 2009, 2018) have
also yielded informative palaeobiological datasets unavail-
able through other vibrational spectroscopy approaches.
Most FTIR vibrations of relevance to biology are the
organic spectral features occurring within the mid-IR
region between wavenumbers 700 and 4000 cm
!1
(Bellamy
1954; Socrates 1980; Painter et al. 1985; Helm & Naumann
1995; Naumann et al. 1996). This range includes aromatic,
aliphatic, hydroxyl, carboxyl and carbonyl functional
groups, as well as amides, polysaccharides and nucleic
acids. Regions of particular interest include the aliphatic
stretching region between 2800 and 3050 cm
!1
, and the
aromatic-alkenic-carbonyl region between 700 and
1800 cm
!1
, which contains information on both biological
compounds and their interactions with mineral substrates,
including silicates and carbonates (Benning et al. 2004; Yee
et al. 2004; Igisu et al. 2009, 2018; Guido et al. 2013).
Membrane lipid composition distinguishes biological
domains
The CH
2
/CH
3
ratio (Eqn 1; Lin & Ritz 1993) provides
both an estimation of the length of aliphatic chains and a
branching index such that an increase in the CH
2
/CH
3
ratio indicates longer or less branched aliphatic chains.
Marshall et al. (2005) used this ratio to approximate dif-
ferent n-alkane chain lengths in Meso-Neoproterozoic
acritarchs, thus quantifying structural variation in their
membrane biopolymers. Steemans et al. (2010) later used
the ratio to make taxonomic distinctions between Silurian
miospores. Herein, we use the CH
2
/CH
3
ratio to deduce
the n-alkane chain length of preserved biopolymers in
microbial mat material. It has been demonstrated that
differences in the absorbance (Abs) ratio of aliphatic CH
3
and CH
2
moieties after linear baseline correction, i.e. the
CH
2
/CH
3
and CH
3
/CH
2
ratios (Eqns 1, 2) highlighted in
Lin & Ritz (1993), Marshall et al. (2005) and Igisu et al.
(2006, 2009, 2012, 2018), can distinguish the nature of
precursor membrane lipids. As noted above, the
composition of membrane lipids is considered the most
fundamental distinction between extant Archaea and Bac-
teria (Albers & Meyer 2011).
CH2
CH3
¼AbsCH2
AbsCH3
¼Abs #2925 cm!1
Abs #2960 cm!1
ð1Þ
CH3
CH2
¼R3=2¼AbsCH3
AbsCH2
¼Abs#2960 cm!1
Abs#2925 cm!1
ð2Þ
In terms of FTIR spectral distinctions of CH
2
and CH
3
,
archaeal lipids are composed of mostly highly branched
isoprenoid chains with higher R
3/2
absorbance ratios,
whereas bacterial lipids consist of longer, straighter, fatty
acid-like hydrocarbons with lower aliphatic R
3/2
ratios
(Atlas 1989; Igisu et al. 2009). Igisu et al. (2009, 2012)
validated the logic of this approach. Igisu et al. (2009)
demonstrated a similarity in R
3/2
ratios between fossil and
extant cyanobacteria (Synechocystis sp. PCC6803) and sug-
gested that the CH
3
/CH
2
ratio of membrane lipids usually
controls the CH
3
/CH
2
ratio of the whole cell. Igisu et al.
(2012) used cultured Bacteria and Archaea to prove that
the R
3/2
ratio varies between the two domains, and
between cellular components. This approach has been
applied to distinguishing fossilized organisms, for instance
by demonstrating that the R
3/2
values of microfossils from
the 1.88 Ga Gunflint Formation (including Gunflintia
spp. and Huroniospora) and the 850 Ma Bitter Springs
Formation (including Cephalophytarion and Glenobotry-
oidion) are consistent with a bacterial affinity (Igisu et al.
2009), corroborating their palaeobiological determination
as cyanobacteria (Barghoorn & Tyler 1965; Schopf &
Blacic 1971; Awramik & Barghoorn 1977).
FTIR spectra change little with maturation and
metamorphism
Experimental approaches have demonstrated that FTIR
spectra evolve during maturation and metamorphism, pri-
marily through the addition of heat (Fraser et al. 2014; Jar-
dine et al. 2015); however, well-preserved biomolecules
have been identified in identical extant and extinct species
(Fraser et al. 2012; Hackley et al. 2017). Generally, the peak
intensities of aliphatic bands are observed to reduce in
magnitude, whereas aromatic bands show limited relative
change (Fraser et al. 2014; Jardine et al. 2015). Conse-
quently, the ratio of the intensity of the aliphatic C–H
stretching region (2800–3050 cm
!1
) to that of the aromatic
ring stretch (~1600 cm
!1
) reflects the degree of aromaticity
of the carbonaceous material (Eqn 3), which generally
increases with metamorphic grade. Thus, Archaean micro-
fossils subjected to lower greenschist metamorphism are
largely composed of aromatic carbonaceous materials. This
4PALAEONTOLOGY
can be compared with Raman spectral characteristics, par-
ticularly the relative intensities of the D and G-bands,
which denote the thermal maturity (i.e. syngenicity) of car-
bonaceous materials and the nature of their non-polar aro-
matic frameworks. Raman spectroscopy of Palaeoarchaean
carbonaceous materials suggests that they are composed of
heterogeneous, but generally poorly ordered, weakly
graphitized, mixed sp
2
and sp
3
carbon (Marshall et al.
2007, 2010). Relics of other carbon-bearing groups, for
example amides, carbonyl (C=O) and carboxyl (R–COOH)
groups, may be also present in fossilized carbonaceous
materials (Steemans et al. 2010; Jardine et al. 2015). The
relative proportions of carbon-bearing groups are assessed
using the following three equations:
Aliphatic
Aromatic ¼Absal
Absar
¼Abs3000!2800 cm!1
Abs1600 cm!1
ð3Þ
Aliphatic
Carbonyl ¼Abs2925 cm!1
Abs1710 cm!1
ð4Þ
Carbonyl
Aromatic ¼Abs1710 cm!1
Abs1600 cm!1
ð5Þ
Since organic matter has a tendency to remain stable dur-
ing diagenesis due to the insolubility of the lipid fraction in
water (Peters et al. 2005), low-maturity organic materials
have the potential to preserve IR spectral characteristics
diagnostic of the precursor material (Igisu et al. 2009;
Guido et al. 2012; Olcott Marshall & Marshall 2015) even
through low-grade metamorphism (Igisu et al. 2018). Natu-
rally matured kerogens have been shown to undergo only a
small initial change in aliphatic compositions due to scis-
sion of the C–C bond adjacent to aromatic moieties (Lin &
Ritz 1993; Lis et al. 2005; Igisu et al. 2009). This chain
shortening can, nonetheless, be compensated by simultane-
ous breakage of a C–C bond next to a tertiary carbon atom
(Lis et al. 2005). Thus, the aliphatic chemistry accessible
through FTIR analyses may undergo no measurable change
during diagenesis. Increasing metamorphic alteration (sub-
greenschist facies) nonetheless leads to muting (i.e.
decreased intensity) of the FTIR peaks associated with bio-
logical materials (Guido et al. 2013; Igisu et al. 2018).
MATERIAL AND METHOD
Sampling and fundamental characterization
Samples were collected during field campaigns to the Bar-
berton greenstone belt between 2003 and 2012 and all
sampling locations may be re-located by means of their
global positioning system (GPS) co-ordinates. All samples
are stored at the Lithoth!
eque of the CNRS Centre de Bio-
physique Mol"
eculaire, Orl"
eans, France. Samples of
banded, carbonaceous, microbial mat-rich cherts were
obtained from the 3.472 Ga Middle Marker horizon,
3.45 Ga Hooggenoeg Formation H5c chert, 3.334 Ga
Footbridge Chert and ~3.33 Ga Josefsdal Chert (Figs 1,
2). The petrography and microbial fossils in these cherts
are described in detail elsewhere (Walsh & Lowe 1999;
Westall et al. 2011, 2015; Hickman-Lewis et al. 2018a,b,
2020a).
The palaeoenvironmental settings of these four hori-
zons were addressed in detail by Hickman-Lewis et al.
(2020a), who used a combination of bulk ICP-MS and
in situ laser ablation ICP-MS to determine that these
microbial mats flourished under complex aqueous
regimes resulting from the confluence of anoxic marine,
hydrothermal and terrigenous (riverine) contributions. It
was found that the Hooggenoeg and Josefsdal mats grew
in conditions relatively more influenced by marine and
hydrothermal activity than the Middle Marker and Foot-
bridge mats. Trace and rare earth element geochemistry
showed that the mats flourished in variably restricted
coastal basins with strong continental influences from
mixed ultramafic, mafic and felsic sources, which presum-
ably provided essential nutrients, particularly transition
metals, to sustain mat growth (Hickman-Lewis et al.
2020a). The microbial mats themselves have been inter-
preted as originating from anoxygenic photosynthetic
microbes due to their occurrence in anoxic shallow-water
sedimentary successions (Walsh & Lowe 1999; Westall
et al. 2011, 2015; Hickman-Lewis et al. 2018a,b, 2020a).
Previously reported Raman spectra of these microbial
mats show bands at ~1350 and ~1600 cm
!1
, correspond-
ing to disordered and graphitic carbonaceous material;
i.e. poorly ordered, randomly layered, polyaromatic
domains that have undergone metamorphic temperatures
of 285–400°C (e.g. Xie et al. 1997; Tice et al. 2004; Wes-
tall et al. 2011, 2015; Hickman-Lewis et al. 2018a, 2020b).
We will not focus further on these Raman spectra in this
contribution. It suffices to say that this characterization
demonstrates that, while this carbonaceous material has
undergone some degree of maturation, it has not been
completely graphitized; thus, based upon well-constrained
experimental work, it is plausible to identify C–H and
other bonds within its structure (Fraser et al. 2014; Jar-
dine et al. 2015).
Transmission Fourier transform infrared spectroscopy
Transmission FTIR was performed on microbial mats
from the above-mentioned Middle Marker horizon,
HICKMAN-LEWIS ET AL.: DIVERSITY IN ARCHAEAN MICROBIAL MATS 5
Hooggenoeg Chert H5c, Footbridge Chert and Josefsdal
Chert. Representative microbial fabrics are detailed in
the photomicrographs shown in Figure 2. Optical micro-
scopy was used both to identify mat fabrics in thin sec-
tion and to confirm that all regions of interest were free
from oxidative weathering, percolative chemical alter-
ation or post-diagenetic remineralization that may influ-
ence FTIR spectra (see Jardine et al. 2015). Samples
were prepared as detached doubly polished thin sections,
which were removed from their glass slides by dissolu-
tion of cyanoacrylate adhesive. High-magnification opti-
cal microscopy was again used after detachment to
confirm that all adhesive had been removed from the
regions of analysis.
FTIR measurements were conducted at the Centro
Interdipartimentale Grandi Strumenti (Universit!
a degli
studi di Modena e Reggio Emilia, Modena, Italy) using a
Bruker VERTEX 70 instrument equipped with a GLOBAR
(silicon carbide) source and a deuterated triglycine sul-
phate (DTGS) detector with a KBr window. Several
microbial mats were analysed within each detached sec-
tion and multiple regions of interest (between 4 and 20)
were measured within each mat. Representative complete
spectra are shown in Figure 3. Each reported FTIR spec-
trum is an average of 16 spectra accumulated within the
studied area (<1 mm
2
; i.e. completely within the micro-
bial mat horizon) within a spectral range of 7500–
370 cm
!1
with a 4 cm
!1
spectral resolution. Therefore,
between 64 and 320 spectra were obtained for each sam-
ple. A reference background spectrum in air (T
bkg
) was
taken before each analysis and subtracted from the sample
spectrum to remove environmental noise from the signal.
FTIR spectra for samples (T
sample
) are reported as absor-
bance (Abs =log
10
[1/transmittance]) relative to the refer-
ence (absorbance =log
10
[T
sample
/T
bkg
]). All spectra
presented are reported as absorbance versus wavenumber
(cm
!1
). To derive characteristic frequencies and compare
relative intensities, the studied regions of each back-
ground-corrected spectra were baseline-corrected. The
baseline correction was applied as a two-point straight
line between two end-points of the desired frequency
region according to the methods of Benning et al. (2004),
Marshall et al. (2005) and Igisu et al. (2006, 2009, 2018).
First-order derivative spectra were calculated to identify
any spectral features occluded by adjacent or overlapping
peaks (Fraser et al. 2012).
Systematic band identifications were made in spectral
regions crucial for determining the composition of car-
bonaceous material (Figs 4–7). Peak identification took
into consideration the variable wavenumbers associated
with similar or identical features in previous publications,
e.g. wavenumbers between 2920 and 2930 cm
!1
have been
attributed to the antisymmetric methylene CH
2
stretch
(Bellamy 1954). Peak identifications are indicated in Fig-
ure 4 for the aliphatic region and Figure 5A for the aro-
matic-alkenic region. Increasing maturity can lead to
changes in peak position, for example, aromatization of
lipids leads to the movement of their diagnostic peaks to
lower frequencies (Benning et al. 2004; Fraser et al. 2014;
Jardine et al. 2015). Given the often highly controversial
nature of suggesting biomolecular complexity in traces of
ancient life, we have identified IR bands as corresponding
to specific molecular features only where no plausible alter-
natives (such as matrix mineralogy) exist, following listings
in Bellamy (1954), Socrates (1980), Painter et al. (1985),
Benning et al. (2004), Yee et al. (2004), Marshall et al.
(2005), Chen et al. (2015) and Igisu et al. (2009, 2018).
Numerous peaks within the 1400–2000 cm
!1
range are
interpreted as Si–O (Igisu et al. 2009; Fig. 6B) therefore,
although first-order derivative spectra enable the distinc-
tion of certain peaks in this region (Fig. 6), identifications
were made in a highly conservative fashion.
Spectra were also obtained for the cyanoacrylate adhe-
sive, finding major absorption bands at 2988, 2944, 2908,
and 2876 cm
!1
within the aliphatic C–H stretching
region and weak bands at 1450 and 1375 cm
!1
(Fig. 7H),
consistent with cautions raised by Igisu et al. (2009).
All FTIR data for this study are available in Hickman-
Lewis et al. (2020c).
RESULTS
3.472 Ga Middle Marker horizon
Crinkly microbial laminations exhibiting local micro-
tufted morphologies were analysed in two samples, MM1
and MM2 (07SA21 and 07SA23 in the Orl"
eans
Lithoth!
eque classification), representative photomicro-
graphs of which are shown in Figure 2A–B. In the alipha-
tic C–H stretching region (2800–3050 cm
!1
; Fig. 4A),
both samples exhibit bands at 2960 cm
!1
(asymmetrical
FIG. 3. Representative selections of transmission Fourier transform infrared (FTIR) absorbance spectra in the range 500–4000 cm
!1
.
A, Middle Marker horizon sample MM2. B, Hooggeoneg Formation H5c sample HC1. C, Footbridge Chert sample FC1. D, Josefsdal
Chert sample JC1; highlighted zones are relevant to the composition of the aromatic and aliphatic moieties in carbonaceous materials
(after Socrates, 1980; Painter et al. 1985; Marshall et al. 2005; Igisu et al. 2009; Guido et al. 2013; Chen et al. 2015). The region
around 2300 cm
!1
is discounted from discussion, since it results from ambient CO
2
.
6PALAEONTOLOGY
HICKMAN-LEWIS ET AL.: DIVERSITY IN ARCHAEAN MICROBIAL MATS 7
end-methyl CH
3
stretching), and weak peaks at 2920–
2925 cm
!1
(asymmetrical methylene CH
2
stretching) and
2850 cm
!1
(symmetrical CH
2
stretching). Sample MM2
shows a further broad, high-intensity band at 2870–
2875 cm
!1
, i.e. symmetrical CH
3
stretching. Although this
band may be interpreted to result from the adhesive
(Fig. 7H), the absence of bands at 2944 and 2908 cm
!1
,
in addition to the ubiquitous presence of the band at
2960 cm
!1
, lead us to attribute the band at 2870–
2875 cm
!1
to an indigenous CH
3
component.
In the aromatic/alkenic region (~1300–1800 cm
!1
;
Fig. 5A) sample MM1 exhibits weak bands at 1338 and
1360 cm
!1
that are also attributable to CH
3
and an intense
band at 1456 cm
!1
assigned to asymmetrical CH
2
stretching.
Bands at 1444 and 1464 cm
!1
are attributed to C–Hstretch-
ing with a contribution from aromatic ring stretching. The
bands at 1394, 1415 and 1728–1743 cm
!1
are assigned to the
symmetrical stretch of C–O, COO and the >C=Oester
stretch, respectively, of carboxyl groups. A distinct feature at
1473 cm
!1
and a shoulder at 1770–1778 cm
!1
are both
attributed to ester C=O, while bands at 1633–1651 cm
!1
indicate highly conjugated C=O. A weak feature around
1537 cm
!1
may reflect aliphatic COOH. Olefinic and aro-
matic groups, together with contributions from carboxyl
COOH account for the peaks at 1554, 1568 and 1574 cm
!1
.
Furthermore, carbonyl C=O and carboxyl COOH (1689 and
1703 cm
!1
) are superposed onto the Si–O peak 1682 cm
!1
.
Carbonyl and carboxyl groups also account for well-defined
bands on the shoulder of this peak at 1710 and 1720 cm
!1
.
Further weak shoulders to this Si–O peak at 1666 and
1672 cm
!1
probably result from alkyl and/or phenyl groups.
Bands at 1487 and 1493 cm
!1
correspond to aromatic ring
stretches with a contribution from Si–O, and the bands at
1504 and 1600 cm
!1
, the latter particularly prominent, are
FIG. 4. Representative transmission Fourier transform infrared (FTIR) absorbance spectra in the range 2800–3050 cm
!1
. A, Middle Mar-
ker horizon sample MM1. B, Hooggenoeg Formation H5c sample HC2. C, Footbridge Chert sample Fc1. D, Josefsdal Chert sample JC1.
8PALAEONTOLOGY
attributed to aromatic ring stretching (C=C). Finally, a weak
shoulder on the 1600 cm
!1
C=C band at 1618 cm
!1
may be
dNH. Other peaks within the 1400–2000 cm
!1
range at
1514–1527 and 1790–1792 cm
!1
are interpreted as Si–Oafter
Igisu et al. (2009). It should be noted that the FTIR absorp-
tion intensity of samples from the Middle Marker is one
order of magnitude lower than in the other samples analysed.
Sample MM2 exhibits additional peaks at 1373 cm
!1
,
attributed to CH
3
, at 1430 cm
!1
, also attributed to ali-
phatic C–H stretching, and at 1732 cm
!1
, attributed to
the >C=O ester stretch. These spectra do not exhibit the
bands at 1568 cm
!1
and 1770 cm
!1
indicating lower con-
tributions from C=O. The ester C=O moves to a lower
wavenumber (1469 cm
!1
), whereas the weak dNH feature
on the 1600 cm
!1
C=C band moves to a higher
wavenumber (1622 cm
!1
). All other peak identifications
remain the same as in MM1.
3.45 Ga Hooggenoeg Chert H5c
Two sequences of microbial laminations from the same
outcrop of Hooggenoeg Formation unit H5c were stud-
ied: HC1, containing flat-laminated biofilm sequences
several millimetres in thickness; and HC2, in which the
laminations are undulatory (representative examples in
Fig. 2C–F) (03SA15 and 03SA15B in the Orl"
eans
Lithoth!
eque). Both samples show absorption bands in the
aliphatic C–H stretching region at 2850, 2920 and
2960 cm
!1
(spectra of HC2 given in Fig. 4B) attributed
to symmetrical and asymmetrical CH
2
and CH
3
stretching
as detailed above. Some spectra in HC2 also show a very
weak feature at 2870–2875 cm
!1
corresponding to the
symmetrical CH
3
stretch. In spectra from HC1, the broad
feature centred on 2925 cm
!1
probably incorporates a
minor contribution from the band at 2890 cm
!1
(C–H
FIG. 4. (Continued)
HICKMAN-LEWIS ET AL.: DIVERSITY IN ARCHAEAN MICROBIAL MATS 9
stretch), which forms a shoulder. Additionally, the aro-
matic C–H feature at 3010 cm
!1
is visible in a limited
number of spectra from both samples, although is close
to being within the noise of the spectra.
In the aromatic/alkenic region, sample HC1 exhibits weak
bands at 1340, 1348, 1356–1360 and 1433 cm
!1
are attribu-
ted to CH
3
and a deviation at 1456 cm
!1
assigned to asym-
metrical CH
2
stretching. Bands at 1433 and 1456 cm
!1
are
attributed to C–H stretching from polysaccharides. Bands at
1392, 1415 and 1691 cm
!1
are assigned to the symmetrical
stretch of C–O, COO and the >C=O ester stretch, respec-
tively, of carboxyl groups. The band at 1547 cm
!1
is attrib-
uted to amide II, comprising C–N and N–H. Distinct bands
at 1635–1649 cm
!1
indicate highly conjugated C=O. Olefi-
nic and aromatic groups, together with contributions from
carboxyl COOH may account for the weak band at
1560 cm
!1
. Carbonyl and carboxyl groups also account for
well-defined bands at 1710 and 1720–1722 cm
!1
, and weak
features in some spectra at 1730 cm
!1
, on the shoulder of
the 1684 cm
!1
Si–O peak. A band at 1481 cm
!1
reflects an
FIG. 5. Representative transmission Fourier transform infrared (FTIR) absorbance spectra in the range 1300–1800 cm
!1
. A, Middle Mar-
ker horizon sample MM1. B, Hooggenoeg Formation H5c sample HC2. C, Footbridge Chert sample FC1. D, Josefsdal Chert sample JC1.
10 PALAEONTOLOGY
aromatic ring stretch with a contribution from Si–O, and
the bands at 1502 and 1604 cm
!1
are attributed to aromatic
ring stretching (C=C). A weak shoulder on the 1600 cm
!1
C=C band at 1622 cm
!1
may be dNH. Other peaks within
the 1400–2000 cm
!1
range at 1514–1527 and 1788–
1792 cm
!1
are interpreted as Si–O.
Sample HC2 (Fig. 5B) does not exhibit the CH
3
peak
at 1348 cm
!1
, but rather exhibits two peaks at 1543 and
1550 cm
!1
attributed to C–H and N–H in amide II. HC2
exhibits additional features at 1489–1491 cm
!1
, assigned
to aromatic ring stretching, a wider array of bands reflect-
ing highly conjugated C=O at 1628, 1643 and 1651 cm
!1
,
and two additional bands at 1714 and 1739 cm
!1
, which
are attributed to the >C=O ester stretch of carboxyl or
carbonyl groups.
3.334 Ga Footbridge chert
Sample FC1 (03SA09 in the Orl"
eans Lithoth!
eque), com-
prising black-grey-white cherts with exceptionally well-
preserved biofilm-like microbial horizons, exhibits weak,
but consistently present, absorption bands in the aliphatic
C–H stretching region at 2850, 2920–2925 and
2960 cm
!1
(Fig. 4C) attributed to symmetrical and asym-
metrical CH
2
and CH
3
stretching.
In the aromatic/alkenic region (Fig. 5C), sample FC1
exhibits weak bands at 1338, 1360 and weak features at
1435 cm
!1
attributed to CH
3
. Bands at 1435 and
1454 cm
!1
are attributed to C–H stretching from
polysaccharides. Bands at 1390–1394, 1415 and 1684–
1689 cm
!1
are assigned to the symmetrical stretch of C–
O, COO and the >C=O ester stretch, respectively, of car-
boxyl groups. An additional band around 1739 cm
!1
is
also assigned to the >C=O ester stretch. Distinct bands at
1630–1649 cm
!1
indicate highly conjugated C=O. Olefi-
nic and aromatic groups may account for the weak band
at 1554 cm
!1
. Carbonyl and carboxyl groups are candi-
date identifications for weak bands at 1724–1732 cm
!1
,
on the shoulder of the 1684 cm
!1
Si–O peak. A band at
1488–1491 cm
!1
reflects an aromatic ring stretch with a
contribution from Si–O. The bands at 1502 and
FIG. 5. (Continued)
HICKMAN-LEWIS ET AL.: DIVERSITY IN ARCHAEAN MICROBIAL MATS 11
FIG. 6. Mathematical first-order derivative spectra averaged from each sample in the ranges: A, 1300–1800 cm
!1
; B, 2800–
3050 cm
!1
. Spectral characteristics indicate similarity in biopolymer composition but differences in biopolymer concentrations, i.e. the
absorbance ratio differs depending upon the sample. Deviations around 2850, 2880, 2925 and 2960 cm
!1
indicate between-sample
variation in aliphatic biochemistry.
12 PALAEONTOLOGY
1600 cm
!1
are also attributed to aromatic ring stretching
(C=C). Other peaks within the 1400–2000 cm
!1
range at
1516–1523 and 1788 cm
!1
are interpreted as Si–O.
3.33 Ga Josefsdal chert
FTIR spectra of sample JC1 (12SA18), a flat-laminated
microbial mat with well-preserved individual laminae,
show absorption bands of moieties in the aliphatic C–H
stretching region (Fig. 4D) centred on 2850 (symmetrical
CH
2
stretching), 2920 (asymmetrical methylene CH
2
stretching) and 2950–2960 cm
!1
(asymmetrical CH
3
stretching), with a small number of spectra also showing a
minor feature at 2870–2875 cm
!1
corresponding to the
symmetrical CH
3
stretch. A very minor band at 2890 cm
!1
(C–H stretch) may exist as a shoulder to the broad feature
centred on 2925 cm
!1
. Two spectra from JC1 also exhibit
the aromatic C–H bands at 3010 and 3030 cm
!1
. JC2
(12SA36) shows only minor peaks at 2850 and 2925 cm
!1
and weak deviations around 2960 cm
!1
, suggesting a more
minor aliphatic component.
In the aromatic/alkenic region (Fig. 5D), sample JC1
features bands at 1340, 1360 and 1433 cm
!1
attributed to
CH
3
. Features at 1450, 1466 and 1470 cm
!1
are also
attributed to C–H stretching from polysaccharides. Bands
at 1394 and 1714 cm
!1
are assigned to the C–O and sym-
metrical >C=O ester stretch of carboxyl or carbonyl
groups. The strong band at 1415 cm
!1
is attributed to
COO from carboxylate, while bands at 1572, 1682 and
1705 cm
!1
reflect C=O and COOH from carboxyl, aug-
mented by an Si–O band. Bands at 1730–1734 and
1747 cm
!1
(on the shoulder of the 1684 cm
!1
Si–O
peak) are further evidence for a strong carbonyl–carboxyl
contribution to JC1. The band at 1540 cm
!1
is attributed
to amide II, comprising C–N and N–H. Bands at 1635–
1653 cm
!1
denote highly conjugated C=O bonds. Fur-
thermore, a small feature at 1670 cm
!1
is attributed to
alkyl or phenyl groups. Olefinic and aromatic groups,
together with contributions from carboxyl COOH may
account for the weak band at 1554–1558 cm
!1
. A band at
1490 cm
!1
reflects an aromatic ring stretch with a contri-
bution from Si–O, and the bands at 1502 and 1600 cm
!1
provide evidence of aromatic ring stretching (C=C).
Other peaks within the 1400–2000 cm
!1
range at 1518–
1522 and 1788 cm
!1
are interpreted as Si–O.
In general, the intensity and diversity of peaks, particu-
larly within the aliphatic stretching region, are consider-
ably higher in JC1 than JC2, indicating superior
preservation of structural variation in the former. A smal-
ler complement of bands in the aromatic/alkenic region
of JC2 also supports this. JC2 exhibits only a single weak
feature at 1454 cm
!1
attributable to C–H stretching in
polysaccharides and limited weak features at 1684–1687
and 1734 cm
!1
, suggesting lower concentrations of pre-
served carboxyl and carbonyl groups. The 1710 cm
!1
fea-
ture attributed to the >C=O ester stretch of carboxyl or
carbonyl groups is very minor and the features at 1540,
1554–1558 and 1572 cm
!1
are absent.
DISCUSSION
FTIR spectra may be evaluated both qualitatively, to iden-
tify preserved molecular complexity, and quantitatively,
since the intensity of an absorption band increases pro-
portionally with its frequency in the corresponding mole-
cule or functional group (Marshall et al. 2005; Chen et al.
2015). FTIR spectra from the studied mats show similar
overall spectral characteristics (Fig. 3), but fine-scale dif-
ferences, particularly the variable concentrations and
compositions of aliphatic moieties (CH
2
and CH
3
; Figs 4,
5). Averaging the mathematical first-order derivative spec-
tra of each horizon reveals strong similarities amongst the
samples, indicating that the same or similar range of
chemical components are present in all carbonaceous
materials (Table 1). The major source of variability
occurs in the aromatic and especially aliphatic regions
(Fig. 6), best explained as differential peak intensity
reduction due to the metamorphic maturation of car-
bonaceous materials (Fraser et al. 2014; Jardine et al.
2015). The many spectral features identified indicate that
relics of diverse molecules and functional groups are pre-
sent in this carbonaceous material.
The following discussion initially addresses the origin
of the FTIR signals and evaluates whether diagenetic over-
printing can account for the spectral variation observed.
Secondly, we describe the (bio)polymer compositions pre-
served, and use the aliphatic chain length and branching
model of Lin & Ritz (1993) to determine the lengths and
degree of branching of aliphatic moieties. Using this
information, we follow the approach of Igisu et al. (2009,
2018), complemented by a detailed set of statistical tests,
to distinguish the contributions of the domains Bacteria
and Archaea to these microbial mats and demonstrate
measurable differences between individual mat-bearing
horizons. We conclude with a brief discussion of the
implications of this exceptional biopolymer preservation
for microbial silicification rates.
Origin of FTIR signals: biology vs diagenesis
FTIR spectra, although preserving a unique biochemical
record, are not immune to the effects of organic material
maturation through diagenesis; even the FTIR spectra of
recalcitrant molecules begin to alter at around 250–300°C
(Fraser et al. 2014). Since the samples studied herein have
HICKMAN-LEWIS ET AL.: DIVERSITY IN ARCHAEAN MICROBIAL MATS 13
14 PALAEONTOLOGY
undergone metamorphic temperatures of 285–400°C (Xie
et al. 1997; Tice et al. 2004; Hickman-Lewis et al. 2020b)
it is necessary to evaluate whether original biochemistry
or post-depositional maturation dominates the FTIR
signal.
Igisu et al. (2009) and Fraser et al. (2014) conducted
experimental maturation of carbonaceous materials and
noted three trends in spectral evolution: (1) aliphatic/aro-
matic ratios reduced due to decreased peak intensities
particularly within the aliphatic region (consistent with
low absorption intensities in Fig. 4); (2) spectral detail
becomes increasingly simplified with increasing maturity
due to the loss of various water-soluble groups (consis-
tent with overall spectral form in Fig. 3); and (3) relative
changes in spectral characteristics (e.g. relative concentra-
tions of aliphatic vs aromatic molecules) change pre-
dictably with increasing maturation. Jardine et al. (2015)
also found that chemical alteration under heat led
towards a general homogenization (simplification) of
spectral characteristics, and concomitant weakening of the
taxonomically indicative aliphatic peaks. Critically, how-
ever, it remains possible to recover taxonomic signals
even after extensive alteration and time (e.g. Palaeozoic
sporopollenin; Steemans et al. 2010; Fraser et al. 2012)
and chemically distinguish taxa even after alteration (Jar-
dine et al. 2015). Although some studies have suggested
that aliphatic molecules may be produced during diagene-
sis (e.g. Kodner et al. 2009), this is not sustained by
either experimental work or observation (Gupta et al.
2007; Igisu et al. 2009; Fraser et al. 2014), and is thus
more logically explained as the selective preservation of
non-hydrolysable aliphatic macromolecules derived from
the original biomass (de Leeuw et al. 2006). Oxidative
polymerization of non-fossilized organic materials, advo-
cated as a method of diagenetic aliphatic increase by Ver-
steegh & Blokker (2004), could not have occurred in the
studied silicified sediments since microquartz (chert) is
impermeable; there is therefore no possibility for sec-
ondary organic moieties to form on or within this already
fossilized biomass. As highlighted in the Introduction,
membrane lipid compositions generally dominate the tax-
onomically indicative aliphatic window of FTIR spectra
even after fossilization (Marshall et al. 2005; Igisu et al.
2006, 2009, 2018).
Overall, alterations in FTIR spectra primarily occur
within the oil window or zeolite-grade conditions of
metamorphism (Igisu et al. 2009; Fraser et al. 2014;
Hackley et al. 2017) after which longer-term maturation
produces only minor changes (Igisu et al. 2009; Jardine
et al. 2015). Selective preservation of lipids relative to
other cellular components may occur during post-mor-
tem diagenesis, but the relative preservation of the alipha-
tic CH
2
and CH
3
bands that enable membrane lipid
identification is similar (Igisu et al. 2009). Therefore,
despite the muting of spectral characteristics during dia-
genesis, variation in spectra from silicified microbial mats
is overwhelmingly explained by diversity in original bio-
mass composition. The fact that aliphatic, aromatic and
other carbon-bearing groups are identified (Figs 4, 5) in
this unambiguously microbial material evidences partial
degradation and kerogenization consistent with the ther-
mal history of the host rock (Bonneville et al. 2020). All
of the above evidence is singularly supportive of our FTIR
spectra being dominated by biology, not diagenesis, and
justifies the use of these spectra for taxonomic distinction.
This is consistent with recent findings using advanced
geochemistry demonstrating that diagenesis does not
obliterate all biochemical specificity in even the most
ancient fossils (Marshall et al. 2005; Igisu et al. 2006,
2009, 2012, 2018; Westall et al. 2011, 2015; Alleon et al.
2016a; Hickman-Lewis et al. 2017, 2018a, 2020b; Wacey
et al. 2017).
Statistically significant differences in biopolymer composition
between microbial mats
In order to evaluate variation in biopolymer compositions
between and within samples and between horizons, we
conducted a number of statistical analyses to demonstrate
the variation of spectral parameters (Figs 8–10; Table 2).
Even simple statistical treatments (box plots and scatter
graphs) evince a clear variation in aliphatic molecules and
functional groups in spite of the highly aromatized
FIG. 7. Details and interpretation of IR spectra from the studied samples. A, summarized band identification in the aliphatic stretch-
ing region (2800–3050 cm
!1
), with aliphatic CH
2
(methylene) and CH
3
(end-methyl) bonds identified. B, summarized band identifica-
tion in the aromatic/alkenic region (1300–1800 cm
!1
); lines indicate band identification; see main text and Figure 5A for details. C–
G, examples of the aliphatic stretching region for mats from: C, the Middle Marker horizon (sample MM1); D–E, the Hooggenoeg
Formation H5c (samples HC1 and HC2, respectively); F, the Footbridge Chert (sample FC1); G, the Josefsdal Chert (sample JC1).
H, FTIR characterization of the carbon-based cyanoacrylate adhesive used within the same region (2800–3050 cm
!1
) demonstrating
that all bands identified in Figure 3 and parts A–G of this figure are not contaminated by adhesive, which is characterized by high-
intensity absorption bands at 2988, 2944, 2908, and 2876 cm
!1
within the aliphatic C–H stretching region (shown) and weaker bands
at 1450 and 1375 cm
!1
in the aromatic-alkenic region (not shown). All horizontal axes represent wavenumber and vertical axes repre-
sent absorbance.
HICKMAN-LEWIS ET AL.: DIVERSITY IN ARCHAEAN MICROBIAL MATS 15
structure of the carbonaceous materials. This is evident
from the large ranges in CH
3
/CH
2
, CH
2
/CH
3
, Abs
1710
/
Abs
1600
and Abs
2925
/Abs
1710
ratios, whereas the aliphatic/
aromatic ratio remains uniformly low in all samples
(Fig. 8). Comparison of the R
3/2
ratio with measured val-
ues from extant biology (bacterial and archaeal cell com-
ponents), microfossils from Proterozoic horizons and
organic materials from the contemporaneous Dresser For-
mation (Fig. 8A; Igisu et al. 2009, 2012, 2018, 2019; Qu
et al. 2015; Bonneville et al. 2020) indicates that most of
the microbial mats studied exhibit a wide range in this
ratio and significant overlaps both with extant and extinct
biology, probably reflecting a community signal. The pre-
cise meaning of this signature for microbial diversity in
the Palaeoarchaean is addressed below. Whereas the
Josefsdal and Footbridge cherts (samples JC1, JC2 and
FC1) show considerable overlap in R
3/2
and CH
2
/CH
3
values (Fig. 8A, B; considering the range between the
25th and 75th percentiles), the ratios in the Middle Mar-
ker and Hooggenoeg Chert mats (especially HC2) show
significantly different values. Taken together, CH
3
/CH
2
and CH
2
/CH
3
ratios suggest that the Hooggenoeg and
Middle Marker mats were dominated by Bacteria (charac-
terized by long, unbranched membrane lipids) and
TABLE 1. Band and peak identifications in FTIR spectra.
Wavenumber (cm
!1
) Designation Occurrence (sample number)
1338, 1340, 1348, 1356–
1360
CH
3
MM1, MM2, HC1, HC2, FC1, JC1
1433–1435, 1454–1456,
1470
CH
2
,C–H stretch from polysacharides MM1, MM2, HC1, HC2, FC1, JC1
1444, 1464 C–H stretching with aromatic ring stretching MM1, MM2, HC1, HC2, FC1
1489–1491 Aromatic ring stretching HC2
1373 CH
3
MM2
1390–1394 C–O MM1, MM2, HC1, HC2, FC1, JC1
1415 COO MM1, MM2, HC1, HC2, FC1, JC1
1430, 1433 Aliphatic C–H stretching MM2, HC2
1469–1473 Ester C=O MM1, MM2
1481, 1487–1493 Aromatic ring stretches with a contribution from Si–
O
MM1, MM2, FC1, JC1
1502–1504 C=C All
1514–1527 Si–O All
1537 Aliphatic COOH MM1(?), MM2(?)
1540–1543, 1547, 1550 Amide II (C–N and N–H) HC1, HC2, JC1
1554–1560, 1568, 1572–
1574
Olefinic and aromatic groups, with carboxyl COOH MM1, MM2, HC1(?), HC2(?), FC1(?), JC1,
JC2
1600–1604 C=C All
1618–1622 dNH MM1(?), MM2(?), HC1(?), HC2(?)
1628–1653 Highly conjugated C=O MM1, MM2, HC1, HC2, FC1
1666, 1672 Si–O All
1670 Alkyl and/or phenyl groups MM1(?), MM2(?), JC1
1682 Si–O All
1684–1689, 1691, 1703–
1705
Carbonyl C=O and carboxyl COOH MM1, MM2, HC1, HC2, JC1, JC2
1710, 1714, 1720–1722 Carbonyl and carboxyl groups; >C=O ester stretch MM1, MM2, HC1, HC2, FC1, JC1, JC2(?)
1728–1747 >C=O ester stretch MM1, MM2, HC1(?), HC2, FC1, JC1, JC2
1770–1778 Ester C=O MM1, MM2
1788–1792 Si–O All
2850 Symmetrical CH
2
stretch All
2870–2875 CH
3
MM2, HC2(?)
2890 C–H stretch HC1(?), JC1(?)
2920–2925 Asymmetrical methylene CH
2
stretch All
2960 Asymmetrical end-methyl CH
3
stretch MM1, MM2, HC1, HC2, FC1, JC1, JC2(?)
3010, 3030 Aromatic C–H JC1
Identifications were made after Socrates (1980), Painter et al. (1985), Benning et al. (2004), Marshall et al. (2005), Igisu et al. (2009),
Guido et al. (2013) and Chen et al. (2015).
16 PALAEONTOLOGY
Archaea (characterized by shorter, highly branched mem-
brane lipids), respectively, whereas the other mats were
inhabited by mixed consortia. Concentrations of aliphatic
moieties are relatively similar throughout all samples and
lower than in reported datasets for Proterozoic material;
all aliphatic/aromatic ratios measured were below 0.07
(Fig. 8C). Abs
1710
/Abs
1600
ratios vary widely between and
within samples (Fig. 8D), whereas Abs
2925
/Abs
1710
ratios
vary widely between, but to a lesser extent within, sam-
ples (Fig. 8E). In summary, box plots indicate diverse ali-
phatic and carbonyl/carboxylic compositions in the
studied carbonaceous materials. Aliphatic ratios resemble
the biopolymer compositions detected in contemporane-
ous (Igisu et al. 2018) and younger (Marshall et al. 2005;
Igisu et al. 2006, 2009; Javaux & Marshall 2006; Qu et al.
2015) biogenic organic materials.
Analysis of variance (ANOVA) found significant differ-
ences in CH
2
/CH
3
and CH
3
/CH
2
with high coefficients of
determination (0.767 and 0.820, respectively) and
p<0.01 (Table 2). Samples were grouped using Tukey’s
and REGWQ (Ryan–Einot–Gabriel–Welsh Studentized
Range Q) tests; in all but two cases (the CH
3
/CH
2
ratios
of HC2 and JC2) these groupings were identical for the
two tests (Table 2). A confidence interval of p <0.05 was
not attained in comparisons of the aliphatic/aromatic,
Abs
1710
/Abs
1600
and Abs
2925
/Abs
1710
ratios, indicating that
the means of these ratios are not statistically distinct
across the seven studied samples; this concurs with the
lack of sample-specific grouping in box plots (Fig. 8D, E).
Two-sample Student’s t-tests were conducted on all
spectral parameters from all pairs of samples. These tests,
determining whether mean values between samples are
significantly different, invariably support the distinction
of samples by the other tests. Three illustrative examples
follow. Comparing samples HC2 and MM2, which are
the most distinct based on box plots and PCA (see
below), and joint most distinct based on ANOVA, the Stu-
dent’s t-test finds that the null hypothesis can be rejected
at a confidence level of p <0.01 for CH
2
/CH
3
, CH
3
/CH
2
,
Abs
1710
/Abs
1600
and Abs
2925
/Abs
1710
ratios, but cannot be
rejected for the aliphatic/aromatic ratio; i.e. the two sam-
ples are significantly distinct in four out of five tested
spectral parameters. Comparing MM1 and MM2, which
directly overlap in all other tests (box plots, ANOVA, PCA),
no spectral parameters, except for the Abs
2925
/Abs
1710
ratio, enable rejection of the null hypothesis; i.e. the two
samples are similar in four out of five parameters. Com-
paring HC1 and JC1, which partly overlap in box plots,
scatter plots, ANOVA and PCA, the Student’s t-test finds
significant differences in the CH
3
/CH
2
and Abs
2925
/
Abs
1710
ratios but is unable to reject the null hypothesis
for the CH
2
/CH
3
, aliphatic/aromatic and Abs
1710
/Abs
1600
ratios; i.e. the samples are similar in three out of five
parameters.
Scatter plots further sustain these distinctions: the
CH
3
/CH
2
ratio (and by extension its reciprocal CH
2
/
CH
3
) shows sample-specific grouping independent of ali-
phatic/aromatic ratios (Fig. 9A), whereas overlapping
groupings of ratios incorporating aliphatic and carbonyl/
carboxylic groups (Fig. 9B–F) suggests that most
between-sample and within-sample variation (even in the
same chert) results from the relative concentrations of ali-
phatic and, to a lesser extent, carbonyl/carboxylic groups.
Since, in fossilized microbial mats, carbonyl/carboxylic
groups are more likely to be derived from extracellular
polymeric substances whereas aliphatic compounds derive
from cell membranes, we add statistical support to the
findings of Igisu et al. (2009) that variations in the taxo-
nomic indices of FTIR spectra (and by extension, car-
bonaceous materials) are dominated by membrane lipids.
Finally, all spectral data were subjected to principal
component analysis (PCA), an ordination technique
based on an orthogonal transformation to convert a set
of observations into a minimal number of possibly corre-
lated dimensions (the eponymous principal components).
Considering the CH
3
/CH
2
, Abs
1710
/Abs
1600
and Abs
2925
/
Abs
1710
ratios as the key spectral data from which taxo-
nomic conclusions can be made, the first two PCA axes
account for 89.65% of total variance in the dataset
(Fig. 10A). Since the aliphatic/aromatic ratio is relatively
constant (Fig. 8C), this indicates very clear differences
between the aliphatic fractions of each sample. PCA axis
1, accounting for 68.03% of the variance, clearly separates
samples with high (MM1, MM2) and low (HC2, JC1 and
JC2) CH
3
/CH
2
ratios, whereas some overlap exists
between samples FC1, JC1, JC2 (to a lesser extent) and
HC1. FC1, exhibiting the widest 25th–75th percentile
range in CH
3
/CH
2
, predictably overlaps with both groups.
A second PCA analysis, as above but excluding the CH
3
/
CH
2
ratio, explained only 78.89% of variance (Fig. 10B),
confirming that most variance is captured within the ali-
phatic fractions and that higher intra-sample variability
occurs in the carbonyl/carboxylic groups (Figs 8D–E, 9D–
F). This is expected since these biomass components are
localized within specific portions of microbial mats and
would thus be heterogeneously distributed in the aliphatic
residues. Since an overwhelming percentage of total vari-
ance (almost 90%) can be explained through the PCA
analyses that reflect the CH
3
/CH
2
ratios, we advocate
these as the crucial parameters for explaining sample dif-
ferences related to composition and taxonomy.
In conclusion, a suite of statistical techniques shows
similarities and differences between the spectral character-
istics of the studied samples. Particularly strong distinc-
tions emerge in the CH
2
/CH
3
and CH
3
/CH
2
ratios which,
as outlined above, are derived directly from the Palaeoar-
chaean organisms that constructed these microbial mats
and thus have taxonomic significance for their
HICKMAN-LEWIS ET AL.: DIVERSITY IN ARCHAEAN MICROBIAL MATS 17
classification (see Lin & Ritz 1993; Marshall et al. 2005;
Igisu et al. 2009, 2012).
Biopolymer composition, aliphatic chain length and
branching
Peak identifications of the following molecules and func-
tional groups were made: (1) CH
3
, CH
2
and other C–H
stretching attributed to polysaccharides; (2) C–O, COO
and the >C=O ester stretch of carboxyl groups; (3) ali-
phatic carboxyl COOH; (4) ester C=O and highly conju-
gated carbonyl C=O; (5) olefinic and aromatic groups;
(6) diverse carbonyl and carboxyl groups; (7) alkyl and/or
phenyl groups; (8) aromatic ring stretches (C=C); and (9)
dNH. Other peaks were interpreted as Si–O. Although
absorption intensity varies between samples, the consis-
tency of wavenumbers attributable to specific moieties
and the large number of band identifications that support
other band identifications, for example numerous car-
bonyl–carboxyl and aliphatic groups, confirm the preser-
vation of biomolecular diversity. No other reasonable
explanation exists for the recurrent observation of spectral
diversity in microbial structures. This demonstrates that,
despite their great age and more than 3 Ga of geological
processing, even the most ancient microbial mats preserve
a rich complement of biomolecules testament to a wealth
of biological diversity in their architect communities.
The carbonaceous materials constituting these mats
therefore comprise predominantly polyaromatic hydrocar-
bon frameworks with small amounts of aliphatic C–H
groups, among others. All horizons show features in the
aliphatic stretching region between 2800 and 3050 cm
!1
,
namely bands at 2850 cm
!1
(symmetrical CH
2
stretch-
ing), 2920–2925 cm
!1
(asymmetrical methylene CH
2
stretching) and 2950–2960 cm
!1
(asymmetrical CH
3
stretching), with many spectra also showing a minor fea-
ture at 2870–2875 cm
!1
corresponding to the symmetri-
cal CH
3
stretch (Fig. 7A). Other minor peaks, e.g.
2988 cm
!1
, which are represented in only a small number
of spectra (e.g. MM1, Fig. 7C), are more challenging to
identify. The peak at 2988 cm
!1
is also exhibited by the
spectra of the carbon-based cyanoacrylate adhesive used
in thin section preparation (Fig. 7H), and could be con-
tamination. We do not believe, however, that this is the
case, primarily because the significant differences in
absorption intensity (an order of magnitude) between the
spectra of the adhesive and those of the carbonaceous
materials suggest that the 2988 cm
!1
has different origins
in each instance. Secondly, spectra exhibiting the
2988 cm
!1
band exhibit a second band at 3010 cm
!1
,
which is also attributable to the C–H aromatic stretch.
Thirdly, no other bands characterizing the spectrum of
the adhesive (2944, 2908, and 2876 cm
!1
) exist in these
spectra. We consider that to find only one adhesive band
at such low intensity is inconsistent with contamination
and interpret the 2988 cm
!1
band as indigenous to bio-
genic carbonaceous matter.
The CH
2
/CH
3
ratio described by Lin & Ritz (1993)
provides an estimate of the chain length of the aliphatic
hydrocarbon moiety bridging the aromatic structure,
together with its degree of branching. Naturally matured
kerogens, when compared with artificially matured kero-
gens, indicate that the CH
2
/CH
3
ratio undergoes only a
minor initial decrease during maturation (Lin & Ritz
1993; Lis et al. 2005), suggesting that aliphatic chains
become only slightly shorter and/or more branched dur-
ing low-grade metamorphism. This is due to the slight
preferential degradation of methylene CH
2
over end-
methyl CH
3
under similar thermal alteration conditions
(Igisu et al. 2009). The CH
2
/CH
3
ratios observed in the
studied mats fall within the range of other micropalaeon-
tological samples (Arouri et al. 1999; Marshall et al. 2005;
Igisu et al. 2009) and, as with R
3/2
ratios, show statisti-
cally significant variation between samples (Fig. 8B).
Using the simplified n-alkane model of Lin & Ritz (1993),
we calculate that the Middle Marker mats are character-
ized by aliphatic biopolymers with short, highly branched
n-alkanes with chain lengths of between 5 and 6; the
Hooggenoeg Chert H5c mats are characterized by longer,
less branched chains of 7–13 units; and the Footbridge
and Josefsdal cherts by broadly similar intermediate val-
ues of 5–9 and 6–9, respectively (Fig. 8B; Table 3). An
outlier point in Josefsdal Chert sample JC1 (Fig. 8B) may
represent a region of increased concentration of long
(chain length 11–12), unbranched aliphatic biopolymers.
Microbial diversity within Palaeoarchaean mats
Understanding microbial diversity in mat-building com-
munities from the Archaean is a critical outstanding ques-
tion in Precambrian palaeobiology. The aliphatic fraction
(between 2850 and 2960 cm
!1
) is a taxonomically indica-
tive signal, and our statistical analyses above have demon-
strated that this fraction exhibits clear differences between
the studied samples and horizons (Figs 4, 6B, 7). The ali-
phatic composition of membrane proteins varies between
the three domains of life: Bacteria, Archaea and Eukarya
(Atlas 1989; Igisu et al. 2012; Albers & Meyer 2011). This
differential signal can be applied to the fossil record, evi-
denced, for example, by FTIR signals of cyanobacterial
microfossils that are consistent with bacterial affinities
(Igisu et al. 2006, 2009, 2018) but inconsistent with
archaeal or eukaryotic affinities (Marshall et al. 2005;
Bonneville et al. 2020). FTIR analyses of Palaeoarchaean
18 PALAEONTOLOGY
FIG. 8. Box plots showing distributions of spectral characteristics. A, CH
3
/CH
2
(R
3/2
) with measured values from extant biology
(bacterial and archaeal cell components), microfossils from Proterozoic horizons and organic materials from the contemporaneous
Dresser Formation (compiled after Igisu et al. 2009, 2012, 2018, 2019; Qu et al. 2015; Bonneville et al. 2020). B, aliphatic ratio
CH
2
/CH
3
shows considerable overlap for Josefsdal (JC) and Footbridge (FC) samples and significantly different values in the Middle
Marker (MM) and Hooggenoeg chert (HC) mats (especially sample HC2). C, the aliphatic/aromatic ratio (Abs
2800–3050
/Abs
1600
) is
similar throughout all samples. D, Abs
1710
/Abs
1600
ratios vary widely between and within samples. E, Abs
2925
/Abs
1710
ratios vary widely
between, but to a lesser extent within, samples. In all box plots, the cross indicates the mean value, the horizontal line indicates the
median value, and circles outside the box denote outliers.
HICKMAN-LEWIS ET AL.: DIVERSITY IN ARCHAEAN MICROBIAL MATS 19
0.04
0.06
FIG. 9. Bi-dimensional scatter plots showing variation in spectral characteristics between samples. A, R
3/2
ratio vs aliphatic/aromatic
(Abs
3050–2800
/Abs
1600
) ratio showing that while the former is highly variable between samples, the latter is essentially constant. B, R
3/2
ratio vs aliphatic/carbonyl–carboxyl (Abs
2925
/Abs
1710
) ratio, indicating a weak positive correlation. C, R
3/2
ratio vs carbonyl–carboxyl/
aromatic (Abs
1710
/Abs
1600
) ratio, indicating a weak positive correlation. D, carbonyl–carboxyl/aromatic (Abs
1710
/Abs
1600
) ratio vs ali-
phatic/carbonyl–carboxyl (Abs
2925
/Abs
1710
) ratio, showing no meaningful correlation. E, aliphatic/carbonyl–carboxyl (Abs
2925
/Abs
1710
)
ratio vs aliphatic/aromatic (Abs
3050–2800
/Abs
1600
) ratio, showing no meaningful correlation. F, carbonyl–carboxyl/aromatic
(Abs
1710
/Abs
1600
) ratio vs aliphatic/carbonyl–carboxyl (Abs
2925
/Abs
1710
) ratio, showing no meaningful correlation. Taking all scatter
plots into consideration, we find that only the R
3/2
ratio exhibits significant and meaningful variation between samples.
20 PALAEONTOLOGY
FIG. 10. Principal components analysis (PCA) of spectral parameters, considering: A, the CH
3
/CH
2
, Abs
1710
/Abs
1600
and Abs
2925
/
Abs
1710
ratios, explaining 89.65% of variance in the dataset; PCA axes 1 and 2 were sufficient to explain most variance in the data; per-
centages in parentheses indicate the variance explained by that axis; B, the Abs
3050–2800
/Abs
1600
(aliphatic/aromatic), Abs
1710
/Abs
1600
and Abs
2925
/Abs
1710
ratios, explaining only 78.89% of variance in the data. Insets show loadings plots; blue diamonds denote the fac-
tors (Abs
3050–2800
in A and R
3/2
in B). Coloured regions were drawn to conservatively include all data from a single sample, discount-
ing outliers. This indicates both similarities (overlap) and differences between the spectral characteristics of the samples approximately
following the PCA axes. Coloured points indicate individual analyses and grey/white symbols (triangles, squares, diamonds, circles)
indicate the centroid of each point cloud.
HICKMAN-LEWIS ET AL.: DIVERSITY IN ARCHAEAN MICROBIAL MATS 21
mats presented herein thus provide a means of linking
taxonomy and phylogeny to mat fabrics through in situ
chemistry. Since FTIR spectral features diagnostic of pre-
cursor biology are resilient to diagenetic degradation (Lin
& Ritz 1993; Peters et al. 2005; Lis et al. 2005; Steemans
et al. 2010; Fraser et al. 2014), the aliphatic moieties pre-
served reflect compositional variation in precursor (archi-
tect) biological materials at the moment of fossilization of
the microbial mat.
Calculated end-methyl to methylene CH
3
/CH
2
(R
3/2
)
ratios suggest significant diversity in mat-builders from
different horizons, however, all ratios fall within the range
of extant prokaryotes (Fig. 8A; Table 3). In the Josefsdal
Chert, R
3/2
ratios varying between 0.195 and 0.769, with
25th and 75th percentiles of 0.337 and 0.568, respectively,
are consistent with both bacterial and archaeal origins
(Igisu et al. 2009, 2018), however, the many low values
below 0.4 necessitate a strong input from bacterial mem-
brane lipids. Although higher values between 0.45 and
0.75 are consistent with archaeal lipid origins (Atlas 1989;
Igisu et al. 2012) and thus suggest a mixed ecosystem, we
nonetheless find that the Josefsdal Chert mats were bacte-
rially dominated. Mats of the Footbridge Chert tell a sim-
ilar story: R
3/2
ratios between 0.337 and 0.796 (with one
outlier at 1.270), with 25th and 75th percentiles at 0.426
and 0.754 respectively, are consistent with mixed bacterial
and archaeal origins. The wider distribution of points and
lack of low R
3/2
ratios when compared to the Josefsdal
Chert suggest that at least some carbonaceous material is
derived from archaeal membrane lipids; however, a con-
centration of points around 0.6–0.7 also indicates a
bacterial cellular signal. In the Hooggenoeg Chert sam-
ples, the limited range of low R
3/2
ratios of between 0.184
and 0.506, is consistent with a signal dominated by bacte-
rial membrane lipids. Finally, the Middle Marker horizon
shows a distribution unlike the other three formations,
with higher R
3/2
ratios between 0.779 and 1.249, and 25th
and 75th percentiles at 0.824 and 1.023 respectively,
potentially diagnostic of a near-pure archaeal consortium.
Although the R
3/2
ratio may not necessarily be applicable
to archaeal material (due to the fact that the highly sim-
plified n-alkane model of Lin & Ritz (1993) upon which
it is based has no direct link to isoprenoids) the R
3/2
val-
ues for the Middle Marker chert mats are inconsistent
with bacterial membrane lipid origins. In the absence of
other logical solutions, we follow the extrapolation of the
n-alkane model by Igisu et al. (2009), who suggested that
the R
3/2
ratio of archaeal lipids would average 1.16, not
dissimilar to our own measurements in mats from the
Middle Marker horizon. Nonetheless, measurements of
archaeal lipids by Igisu et al. (2012) found that values
could be lower (R
3/2
=0.50–0.75), while archaeal cells
had much higher values (R
3/2
=0.75–1.10). We there-
fore suggest a strong contribution from archaeal cell
material in the Middle Marker mats. The outlier point
(R
3/2
=1.270) in the Footbridge Chert mats may also
reflect an archaeal cell contribution, perhaps evidencing
within-mat heterogeneity. This is supported by the con-
centration of elevated R
3/2
ratios in Footbridge Chert
material relative to the conspicuously bacterially domi-
nated Hooggenoeg and Josefsdal cherts. PCA (Fig. 10),
scatter plots (Fig. 9A–C) and ANOVA of coupled to the
TABLE 2. ANOVA results.
Unit Sample Grouping based
on Tukey’s test
Grouping based on
REGWQ test
Coefficient of
determination
p-value
CH
2
/CH
3
Middle Marker horizon MM1 D D 0.767 5.12 910
!23
MM2 D D
Hooggenoeg Chert H5c HC1 B B
HC2 A A
Footbridge Chert FC1 C–DC–D
Josefsdal Chert JC1 B B
JC2 B–CB–C
CH
3
/CH
2
(R
3/2
value)
Middle Marker horizon MM1 A A 0.820 2.13 910
!27
MM2 A A
Hooggenoeg Chert H5c HC1 C–DC–D
HC2 C D
Footbridge Chert FC1 B B
Josefsdal Chert JC1 C C
JC2 C C–D
Groupings of samples are based on Tukey’s and Ryan–Einot–Gabriel–Welsh Studentized Range Q (REGWQ) tests. The coefficient of
determination is high in each case, and all sample differences are deemed statistically significant with a p-value much lower than 0.01.
22 PALAEONTOLOGY
Tukey’s and REGWQ tests (Table 2) all note a higher
degree of similarity between R
3/2
in the Footbridge Chert
and Middle Marker mats than between any other samples
and the Middle Marker mats. We thus suggest that the
Footbridge Chert mats were home to a well-mixed bacte-
rial–archaeal consortium.
Palaeoarchaean mats thus reflect diverse microbial con-
sortia in which Bacteria and Archaea flourished alongside
one another in shallow-water anoxygenic environments
(Fig. 11; Table 3). The Middle Marker mats may reflect an
unusual, uniquely archaeal system. Convincing fossil evi-
dence for Archaea is extremely rare and primarily consists
of either very negative carbon isotopes consistent with the
archaeal clade conducting anaerobic oxidation of methane
(Ueno et al. 2006) or Fe and Ni-enriched carbonaceous
materials that may preserve co-factors from the palaeo-
metallome of archaeal biomass (Hickman-Lewis et al.
2020b). Alternatively, and in accordance with the lower
intensity of FTIR spectral characteristics in the Middle
Marker horizon, the difference in Bacteria-dominated and
Archaea-dominated systems may indicate that while Bacte-
ria were the most common primary producers, Archaea
were the most common heterotrophic degraders, and the
Middle Marker mats have simply undergone further degra-
dation prior to silicification; i.e. their silicification was less
rapid. This alternative, however, rests on the assumption
that the relative microbial compositions of producer and
degrader niches have been constant through time.
We have thus demonstrated, for the first time, unequivo-
cal distinctions between the bacterial and archaeal commu-
nities constructing microbial mats in Earth’s early
ecosystems and have shown that taxonomic specificity may
be preserved in FTIR spectra for more than 3 Gyr.
Rapid silicification allows exceptional preservation of FTIR
signatures
FTIR spectral features diagnostic of the precursor materials,
for example the CH
3
/CH
2
ratio, may be preserved through
diagenesis and catagenesis (Peters et al. 2005; Igisu et al.
2006, 2018; Guido et al. 2013; Hackley et al. 2017). Post-
mortem biological reworking combined with deep burial at
depths greater than 3 km and temperatures higher than
250°C, however, appear to cause diagenetic modification of
IR bio-indicators (Peters et al. 2005; Lis et al. 2005; Fraser
et al. 2014). The samples studied herein have undergone
lower greenschist metamorphism (~3–4 kbar, 285–400°C)
i.e. alteration higher than that presupposing the preserva-
tion of IR spectral heterogeneity. In spite of this, we have
unambiguously demonstrated that variance in FTIR spectra
is due to biomolecular diversity in the original biomass.
Furthermore, there is no evidence for post-diagenetic bio-
logical recombination of the carbonaceous materials within
samples, which were obtained from clean, unweathered
cherts free of textural or chemical alteration. This inconsis-
tency requires explanation. Why, contrary to expectations,
do the carbonaceous materials within these cherts preserve
primary bio-indicative signals?
We suggest that silicification, which was suggested to
be early and rapid in Palaeoarchaean cherts due to the
elevated concentrations of Si in the oceans of the time
TABLE 3. Summary of measured parameters from FTIR spectra with inferences made regarding the composition of relic biopolymers
and the architect microbial community, reflecting the community composition of the mats at the time of fossilization.
Unit Sample CH
3
/CH
2
(R
3/2
) Biological
community
CH
2
/CH
3
n-alkane
chain
length
(approx.)
Aliphatic/
aromatic
Middle
Marker
07SA21 0.785–0.994 Archaea-
dominated
1.033–1.284 5–6 0.014–0.047
07SA23 0.811–1.249 Archaea-
dominated
0.813–1.233 5–6 0.008–0.053
Hooggenoeg
Chert H5c
03SA15 0.287–0.506 Bacteria-
dominated
1.975–3.593 7–10 0.032–0.066
03SA15B 0.184–0.370 Bacteria-
dominated
2.706–5.447 9–13 0.024–0.046
Footbridge
Chert
03SA09 0.377–0.796 (outlier discounted) Mixed
community
0.787–2.652 5–9 0.024–0.048
Josefsdal
Chert
12SA18 0.195–0.769 Mixed
community
1.300–3.044 (outlier discounted) 6–9 0.019–0.042
12SA36 0.364–0.586 Mixed
community
1.706–2.750 7–9 0.024–0.037
HICKMAN-LEWIS ET AL.: DIVERSITY IN ARCHAEAN MICROBIAL MATS 23
(Maliva et al. 2005; Westall et al. 2015; Alleon et al.
2016a; Hickman-Lewis et al. 2020a) inhibits the alteration
of even fine-scale biomolecular heterogeneity in organic
material. Microbes can also mediate their own silicifica-
tion within extracellular polymeric substances (Moore
et al. 2020) which, being a major component of microbial
mats, may account for the exceptional diversity of
biopolymers observed in these mats relative to individual
microfossils in previous studies. This corroborates previ-
ous experimental determination that early diagenetic sili-
cification preserves relics of original biomolecular
composition at up to zeolite facies (150–170°C; Alleon
et al. 2016a) and even as far as sub-greenschist facies
(250°C and 250 bars; Alleon et al. 2016b). We here pro-
pose that, when the silicification of biomass is exception-
ally rapid, as was the case for these rocks, biochemical
moieties reflecting palaeobiology may endure through
even lower greenschist metamorphism. Furthermore, since
silicification began during the life cycle of the organisms
building these microbial mats, hence the volumetric
preservation of biofilms, (Walsh & Lowe 1999; Walsh &
Westall 2003; Hickman-Lewis et al. 2018a) and three-
dimensional preservation of micromorphologies within
(Hickman-Lewis et al. 2017, 2019), the IR signal pre-
served is that of the architect mat-building community at
the moment of silicification. Differentiating between the
IR signals preserved in microbial mats of various mor-
phologies and within various palaeoenvironmental set-
tings could aid both in constraining their original
biological communities and in the eventual reconstruction
of the biomes of Archaean life of which they were a part.
CONCLUSIONS
Palaeoarchaean surface productivity was distributed
throughout systems of basins teeming with microbial life,
the fragmentary records of which are preserved in green-
stone belts. Constraining the microbial communities
within is one of the principal outstanding challenges fac-
ing Archaean palaeontology.
Applying FTIR spectroscopy to multiple microbial mat
horizons from the Palaeoarchaean of the Barberton green-
stone belt, we have demonstrated that spectral character-
istics can delineate relics of the biopolymer compositions
of their architect anoxygenic photosynthetic communities
and, more critically, the relative proportions of the
domains of life in these mats. The range of biomolecular
relics preserved in these mats is considerably greater than
in individual microfossils and reflects the organic compo-
sition of original biomass, i.e. cells and extracellular prod-
ucts. Using the CH
3
/CH
2
and CH
2
/CH
3
ratios, together
with qualitative assessments of absorption band intensity,
we have quantitatively compared the architect communi-
ties of four horizons (the 3.472 Ga Middle Marker,
3.45 Ga Hooggenoeg Chert H5c, 3.334 Ga Footbridge
Chert and the 3.33 Ga Josefsdal Chert) and shown statis-
tically significant differences between them. Whereas mats
of the Middle Marker horizon appear to reflect domi-
nantly archaeal communities, the Hooggenoeg mats result
from dominantly bacterial sources. The Josefsdal mats,
some Hooggenoeg mats and, especially, the Footbridge
mats were built by mixed consortia, indicated by wider
ranges of CH
3
/CH
2
and CH
2
/CH
3
ratios, which show evi-
dence for bacterial and archaeal precursor material.
Such exceptional preservation of biomolecular relics is
largely inconsistent with the lower greenschist metamor-
phism experienced by these rocks. The retention of spec-
tral characteristics demands almost immediate
syndepositional preservation, which is consistent with
early, rapid silicification of these horizons. Chertified fos-
siliferous horizons provide an unparalleled window on
the primitive biosphere and serve as palaeobiological tes-
tament to the diversity of Archaean microbial ecosystems.
We suggest that Bacteria and Archaea were already flour-
ishing together in these ecosystems by 3.5 Ga.
Acknowledgements. This work was funded by the INACMa
(Inorganic Nanoparticles in Archaean Carbonaceous Matter –a
key to early life and palaeoenvironmental reconstructions) pro-
ject (EU-FP7 Grant no. 618657) to BC. FTIR analyses were con-
ducted with Fabio Bergamini (Centro Interdipartimentale
Grandi Strumenti, Modena, Italy). We thank Assimo Maris
FIG. 11. Venn diagram qualitatively summarizing the results
of all measurements and statistical analyses. Samples can clearly
be categorized along a shifting scale from bacterial to archaeal
ecosystem dominance. Most microbial mats comprised a variably
mixed community.
24 PALAEONTOLOGY
(Universit!
a di Bologna), for valuable comments on the statistical
analyses and Sylvain Janiec (Institut des Sciences de la Terre
d’Orl"
eans, France) for producing all thin sections. Four review-
ers provided many helpful comments and suggestions that led to
significant improvement of the paper.
Author contributions. KHL and BC conceived the research. FW
collected the samples. KHL performed the experiments and anal-
ysed the data. All authors contributed to discussion of the data.
KHL wrote the paper, to which all authors contributed.
DATA ARCHIVING STATEMENT
FTIR Data for this study are archived at the Centre National de la
Recherche Scientifique, France, and are available from the Dryad
Digital Repository: https://doi.org/10.5061/dryad.v41ns1rsn
Editor. Barry Lomax
REFERENCES
AL B E R S , S. V. and M E YE R , B. H. 2011. The archaeal cell
envelope. Nature Reviews Microbiology,9, 414–426.
AL L E O N , J., B E R N A R D , S., L E G U I L L OU , C., MA R I N -
CA R B O N NE , J., PO N T , S., BE Y S S A C , O., MCKEE-
G A N , K. D. and R O B E R T , F. 2016a. Molecular preserva-
tion of 1.88 Ga Gunflint organic microfossils as a function of
temperature and mineralogy. Nature Communications,7, 11977.
---
DA V A L , D., SK OU R I - P A N E T , F., PO N T ,
S., D E L B E S , L. and R O BE R T , F. 2016b. Early entombment
within silica minimizes the molecular degradation of microor-
ganisms during advanced diagenesis. Chemical Geology,437,
98–108.
AR O U R I , K., G R E E N W O O D , P. F. and WAL T E R , M. R.
1999. A possible chlorophycean affinity of some Neoprotero-
zoic acritarchs. Organic Geochemistry,30, 1323–1337.
A T L A S , R. M. (ed.) 1989. Microbiology: Fundamentals and
applications. Macmillan.
AW R A M I K , S. M. and B AR G H O O RN , E. S. 1977. The
Gunflint microbiota. Precambrian Research,5, 121–142.
BA R G H O O R N , E. S. and T Y L E R, S. A. 1965. Microorgan-
isms from the Gunflint Chert. Science,147, 563–575.
BE L L A M Y, L. J. 1954. The infra-red spectra of complex mole-
cules. John Wiley & Sons.
BE N N I N G, L. G., PH OE N I X , V. R., Y E E , N. and T O B I N ,
M. J. 2004. Molecular characterization of cyanobacterial silici-
fication using synchrotron infrared micro-spectroscopy.
Geochimica et Cosmochimica Acta,68, 729–741.
BO N N E V IL L E , S., DE L PO M D O R , F., P R E
´AT , A., CHE -
VA L I E R , C., A R A K I , T., KA Z E M I A N , M., ST E E L E ,
A., SC H R E I BE R , A., W I R TH , R. and B E N N I NG , L. G.
2020. Molecular identification of fungi microfossils in a Neo-
proterozoic shale rock. Science Advances,22, eaax7599.
BO S A K , T., GR E E N E , S. and N E W M A N , D. K. 2007. A
likely role for anoxygenic photosynthetic microbes in the for-
mation of ancient stromatolites. Geobiology,5, 119–126.
-
B U S H , J. W. M., F L Y N N , M. R., L I A N G , B., ON O ,
S., P E T R OF F , A. P. and SI M, M. S. 2010. Formation and
stability of oxygen-rich bubbles that shape photosynthetic
mats. Geobiology,8, 45–55.
BY E R L Y , G. R., LO W E , D. R. and HE U B E C K , C. 2018.
Geologic evolution of the Barberton greenstone belt –a
unique record of crustal development, surface processes, and
early life. 3.55–3.20 Ga. 569–613. In VAN KRANEN-
DO N K , M. J., BE N N E T T, V. C. and HOF M A N N , J. E.
(eds) Earth’s oldest rocks. 2nd edn. Elsevier.
CH E N , Y., ZO U , C., M A S T AL E R Z , M., HU , S., G A S -
AW A Y , C. and T A O , X. 2015. Applications of micro-Four-
ier Transform Infrared Spectroscopy (FTIR) in the geological
sciences—a review. International Journal of Molecular Sciences,
16, 30223–30250.
CH O O - S MI T H , L. Q., M A Q U E L I N , K., V A N VREE S -
WI J K , T., BR U I N I N G , H. A., PU P P E L S , G. J., TH I , N.,
KI R S C H NE R , C., NA U M A N N , D., AM I , D., VI L L A ,
A. M., OR SI N I , F., DO G LI A , S. M., L A M F A RR A J , H.
and S O C K A L IN G U M , G. D. 2001. Investigating microbial
(micro)colony heterogeneity by vibrational spectroscopy.
Applied & Environmental Microbiology,67, 1461–1469.
DE LE E U W , J. W., V E R S T E E GH , G. J. M. and V A N
BE R G E N , P. F. 2006. Biomacromolecules of algae and plants
and their fossil analogues. Plant Ecology,182, 209–233.
D E R E N NE , S., RO B E R T , F. and S K R Z Y P C Z A K - B O N -
DU E L L E , A. 2008. Molecular evidence for life in the 3.5 bil-
lion year old Warrawoona chert. Earth & Planetary Science
Letters,272, 476–480.
FR A S E R , W. T., SC O T T , A. C., F O RB E S , A. E. S.,
GL A S S P OO L , I. J., PL O T N I CK , R. E., KEN I G , F. and
LO M A X , B. H. 2012. Evolutionary stasis of sporopollenin
biochemistry revealed by unaltered Pennsylvanian spores. New
Phytologist,196, 397–401.
-
WA T S O N , J. S., SE P H T O N , M. A., L O M A X , B. H.,
HA R R I N G T O N , G., GO S LI N G , W. D. and S E LF , S.
2014. Changes in spore chemistry and appearance with
increasing maturity. Review of Palaeobotany & Palynology,201,
41–46.
GU I D O , A., MA S T A N D R E A , A., TO S T I, F., D E M A S I ,
F., B L A N C O , A., D ’ E LI A , M., OR O F IN O , V., FO N T I,
S. and RU S S O , F. 2012. Characterization of fossil organic
matter with Fourier-Transform Infrared (FTIR) Spectroscopy:
an attempt to record extraterrestrial life. Memorie Della Societ!
a
Astronomica Italiana Supplementi,20, 64.
--
and RU S S O F. 2013. Biotic vs abiotic carbonates:
characterisation of the fossil organic matter with Fourier-
Transform Infrared (FT-IR) spectroscopy. Bollettino Della
Societa
`Paleontologica Italiana,52, 63–70.
GU P T A , N. S., B R I GG S , D. E. G., CO L L I N S O N , M. E.,
EV E R S H ED , R. P., M I C H E L S , R. and P AN C O S T , R.
D. 2007. Molecular preservation of plant and insect cuticles
from the Oligocene Enspel Formation, Germany: evidence
against derivation of aliphatic polymer from sediment. Organic
Geochemistry,38, 404–418.
HA C K L E Y, P. C., WA L T E R S , C. C., K EL E M E N , S. R.,
MA S T A L ER Z , M. and LO W E R S , H. A. 2017. Organic
petrology and micro-spectroscopy of Tasmanites microfossils:
HICKMAN-LEWIS ET AL.: DIVERSITY IN ARCHAEAN MICROBIAL MATS 25
applications to kerogen transformations in the early oil win-
dow. Organic Geochemistry,114, 23–44.
HE L M , D. and N A U M A N N , D. 1995. Identification of some
bacterial cell components by FT-IR spectroscopy. FEMS
Microbiology Letters,126, 75–79.
HI C K M A N - L E W I S , K., G A R W O O D , R.J., W I T H E RS ,
P.J. and W A C E Y , D. 2017. X-ray microtomography as a tool
for investigating the petrological context of Precambrian cellu-
lar remains. 33–56. In B R A S I ER , A. T., MCI LR O Y , D. and
MCLO U G H L IN , N. (eds) Earth system evolution and early
life: A celebration of the work of Martin Brasier. Geological
Society of London Special Publications, 448.
-
CA V A L A ZZ I , B., FO U C HE R , F. and W E S T A L L , F.
2018a. Most ancient evidence for life in the Barberton Green-
stone Belt: microbial mats and biofabrics of the ~3.47 Ga
middle marker horizon. Precambrian Research,312, 45–67.
-
WE S T A L L, F. and CA V AL A Z Z I , B. 2018b. Traces of
early life from the Barberton greenstone belt. 1029–1058. In
VA N K R A N E N D O N K , M. J., B E N N E T T , V. C. and
HO F M AN N , J. E. (eds) Earth’s oldest rocks. 2nd edn. Elsevier.
-
GA U T R E T, P., AR BA R E T , L., SO R I E U L , S., DE
WI T , R., FO U C H E R , F., CA V AL A Z Z I , B. and W E S -
TA L L , F. 2019. Mechanistic morphogenesis of organo-sedi-
mentary structures growing under geochemically stressed
conditions: keystone to the interpretation of some Archaean
stromatolites? Geosciences,9, 359.
-
GO U R C E R O L , B., WE S T A L L , F., MA N ZI N I , D.
and CA V A L A Z Z I , B. 2020a. Reconstructing Palaeoarchaean
microbial biomes flourishing in the presence of emergent
landmasses using trace and rare earth element systematic. Pre-
cambrian Research,342, 105689.
-
CA V A L A ZZ I , B., SO R IE U L , S., GA UT R E T , P.,
FO U C H E R, F., W H I T E H O U S E , M. J., JE O N , H.,
CO C K E L L, C. S., GE O R G E L I N, T. and WE S T A L L , F.
2020b. Metallomics in deep time and the influence of ocean
chemistry on the metabolic landscapes of Earth’s earliest
ecosystems. Scientific Reports,10, 4965.
-
WE S T A L L, F. and CAV A L A Z ZI , B. 2020c. Data from:
Diverse communities of bacteria and archaea flourished in
Palaeoarchaean (3.5–3.3 Ga) microbial mats. Dryad Digital
Repository. https://doi.org/10.5061/dryad.v41ns1rsn
HO M A N N , M., H E U BE C K , C., AI R O , A. and T IC E , M.
M. 2015. Morphological adaptations of 3.22 Ga-old tufted
microbial mats to Archean coastal habitats (Moodies Group,
Barberton Greenstone Belt, South Africa). Precambrian
Research,266, 47–64.
-
SA N S J O FR E , P., VA N Z U I L E N , M., HE UB E C K ,
C., G ON G , J., KI L L I N G S W O R T H , B., FO S TE R , I. S.,
AI R O , A., VA N K R A N E N DO N K , M. J., A D E R , M. and
LA L O N D E, S. V. 2018. Microbial life and biogeochemical
cycling on land 3,220 million years ago. Nature Geoscience,11,
665–671.
IG I S U , M., NAK A S H I MA , S., UE N O , Y., AWR A M I K , S.
M. and M A R U Y AM A , S. 2006. In situ infrared microspec-
troscopy of ~850 million-year-old prokaryotic fossils. Applied
Spectroscopy,60, 1111–1120.
-
UE N O , Y., SH I MO J I M A , M., NAK A S H I MA , S.,
AW R A M I K , S. M., OHT A , H. and M A R U Y AM A , S.
2009. Micro-FTIR signature of bacterial lipids in proterozoic
microfossils. Precambrian Research,173, 19–26.
-
TA K A I , K., UEN O , Y., NI S H IZ A W A , M.,
NU N O U R A , T., H I R A I , M., KA N E K O , M., NA R -
AO K A , H., SH I M O J I M A , M., H O R I , K., NA K A-
SH I M A , S., OH TA , H., MA R U Y A M A , S. and
I S O Z A K I, Y. 2012. Domain-level identification and quantifi-
cation of relative prokaryotic cell abundance in microbial
communities by micro-FTIR spectroscopy. Environmental
Microbiology Reports,4, 42–49.
-
UE N O , Y. and TAK A I , K. 2018. FTIR microspec-
troscopy of carbonaceous matter in ~3.5 Ga seafloor
hydrothermal deposits in the North Pole area, Western Aus-
tralia. Progress in Earth & Planetary Science,5, 85.
-
KO M I Y A , T., A W RA M I K , S. M., I K E M O T O , Y.,
GE N G , Y., UE H A R A , H. and T A K A I , K. 2019. Fourier
transform infrared microspectroscopic characterization of
Neoproterozoic organic microfossils from the Fifteenmile
Group in Yukon, Canada. Island Arc,28, e12310.
JA R D I N E, P. E., F R A S E R , W. T., L O M A X, B. H. and
GO S L I N G , W. D. 2015. The impact of oxidation on spore
and pollen chemistry. Journal of Micropalaeontology,34, 139–149.
JA V A U X , E. J. and M A R S H A L L , C. P. 2006. A new
approach in deciphering early protist paleobiology and evolu-
tion: combined microscopy and microchemistry of single
Proterozoic acritarchs. Review of Palaeobotany & Palynology,
139,1–15.
KO D N E R , R. B., SU M M O N S , R. E. and K N O L L, A. H.
2009. Phylogenetic investigation of the aliphatic, non-hydro-
lyzable biopolymer algaenan, with a focus on green algae.
Organic Geochemistry,40, 854–862.
LI N , R. and R I T Z , G. P. 1993. Studying individual macerals
using i.r. microspectroscopy, and implications on oil versus
gas/condensate proneness and low rank generation. Organic
Geochemistry,20, 695–706.
LI S , G. P., MA S T A L ER Z , M., SC H I M M E L M AN N , A.,
LE W A N , M. D. and ST A N K I E W I C Z, B. A. 2005.
FTIR absorption indices for thermal maturity in comparison
with vitrinite reflectance R
0
in type-II kerogens from
Devonian black shales. Organic Geochemistry,36, 1533–1552.
MA L I V A , R. G., KN OL L , A. H. and S IM O N S O N , B. M.
2005. Secular change in the Precambrian silica cycle: insights
from chert petrology. Geological Society of America Bulletin,
117, 835–845.
MA R S H A LL , C. P., J A VA U X , E. J., KN O L L , A. H. and
W A L T E R , M. R. 2005. Combined micro-Fourier transform
infrared (FTIR) spectroscopy and micro-Raman spectroscopy
of Proterozoic acritarchs: a new approach to palaeobiology.
Precambrian Research,138, 208–224.
-
LO V E , G. D., S N A PE , C. E., H I L L , A. C., A L L-
WO O D , A. C., W A L TE R , M. R., V A N K R A N E N -
DO N K , M. J., BOW D E N , S. A., SY L V A , S. P. and
SU M M O N S, R. E. 2007. Structural characterization of kero-
gen in 3.4 Ga Archaean cherts from the Pilbara Craton, Wes-
tern Australia. Precambrian Research,155,1–23.
-
E D W A RD S , H. G. M. and J E H L IC K A , J. 2010. Under-
standing the application of Raman spectroscopy to the detec-
tion of traces of life. Astrobiology,10, 229–243.
26 PALAEONTOLOGY
MO O R E , K. R., PAJ U S A L U, M., G O N G , J., SO J O , V.,
MA T R E U X , T., B R AU N , D. and B O S A K , T. 2020. Bio-
logically mediated silicification of marine cyanobacteria and
implications for the Proterozoic fossil record. Geology, pub-
lished online 29 May. https://doi.org/10.1130/g47394.1
NA U M A N N, D., H E L M , D. and LA BI S C H I N S K I , H.
1991. Microbiological characterizations by FT-IR spectroscopy.
Nature,351, 81–82.
-
SC H U L T Z , C. and H E L M , D. 1996. What can infrared
spectroscopy tell us about the structure and composition of
intact bacterial cells? 279–310. In M A N T S CH , H. H. and
CH A P M A N , D. (eds) Infrared spectroscopy of biomolecules.
Wiley-Liss.
NI S B E T , E. G. and F O WL E R , C. M. R. 1999. Archaean
metabolic evolution of microbial mats. Proceedings of the Royal
Society B,266, 2375–2382.
N O F F K E, N. 2010. Microbial mats in sandy deposits from the
Archean Era to today. Springer, 175 pp.
-
CH R I S T IA N , D., WA C EY , D. and H A Z EN , R. M.
2013. Microbially induced sedimentary structures recording an
ancient ecosystem in the ca. 3.48 billion-year-old Dresser For-
mation, Pilbara, Western Australia. Astrobiology,13, 1103–
1124.
OLCOTT MARSHALL, A. and MARSHALL, C. P.
2015. Vibrational spectroscopy of fossils. Palaeontology,58,
201–211.
PA I N T E R, P. C., ST A R S I NI C , M. and C O L E M A N , M.
M. 1985. Determination of functional groups in coal by Four-
ier transform interferometry. 169–240. In F E R R A R O , J. R.
and BA S I L E , L. J. (eds) Fourier transform infrared spec-
troscopy. Vol. 4. Academic Press.
PE T E R S , K. E., W A L T E R S , C. C. and M O L D O WA N , J.
M. 2005. The biomarker guide. Vol. 2. Cambridge University
Press, 1155 pp.
QU , Y., EN G D A H L , A., ZH U , S., VA J D A , V. and
MCLO U G H L IN , N. 2015. Ultrastructural heterogeneity of
carbonaceous material in ancient cherts: investigating
biosignature origin and preservation. Astrobiology,15, 825–
842.
SC H O P F , J. W. and B L A C I C , J. M. 1971. New microorgan-
isms form the Bitter Springs formation (late Precambrian) of
the north-central Amadeus Basin, Australia. Journal of Paleon-
tology,45, 925–960.
SO C R A T ES , G. 1980. Infrared and Raman characteristic group
frequencies. John Wiley & Sons, 366 pp.
ST E E M A N S , P., LE P O T , K., M A R S H A L L , C. P., L E
HE R I S S E, A. and J A V A U X , E. J. 2010. FTIR characterisa-
tion of the chemical composition of Silurian miospores (cryp-
tospore and trilete spores) from Gotland, Sweden. Review of
Palaeobotany & Palynology,162, 577–590.
TI C E , M. M. 2009. Environmental controls on photosynthetic
microbial mat distribution on and morphogenesis on a
3.42 Ga clastic-starved platform. Astrobiology,9, 989–1000.
-
and LO W E , D. R. 2006. The origin of carbonaceous mat-
ter in pre-3.0 Ga greenstone terrains: a review and new evi-
dence from the 3.42 Ga Buck Reef Chert. Earth-Science
Reviews,76, 259–300.
-
BO S T I C K, B. C. and L O W E, D. R. 2004. Thermal his-
tory of the 3.5–3.2 Ga Onverwacht and Fig Tree Groups, Barber-
ton greenstone belt, South Africa, inferred by Raman
microspectroscopy of carbonaceous material. Geology,32, 37–40.
TR O W E R , E. J. and LO WE , D. R. 2016. Sedimentology of
the <3.3 Ga upper Mendon Formation, Barberton greenstone
belt, South Africa. Precambrian Research,281, 473–494.
UE N O , Y., YA M A D A , K., YO S H I D A , N., MA R UY A M A ,
S. and I S O Z A KI , Y. 2006. Evidence from fluid inclusions
for microbial methanogenesis in the early Archaean era. Nat-
ure,440, 516–519.
V E R S T EE G H , G. J. M. and B L O K K E R, P. 2004. Resistant
macromolecules of extant and fossil microalgae. Phycological
Research,52, 325–339.
WA C E Y , D., BA T T I S O N , L., G A RW O O D , R. J., H I C K -
MA N - L E W I S , K. and B R A S IE R , M. D. 2017. Advanced
analytical techniques for studying the morphology and chem-
istry of Proterozoic microfossils. 81–104. In B R A S I E R , A. T.,
MCIL R O Y , D. and M CL O U G H L I N , N. (eds) Earth system
evolution and early life: A celebration of the work of Martin Bra-
sier. Geological Society of London Special Publications, 448.
WA L S H , M. M. and L O W E, D. R. 1999. Modes of accumula-
tion of carbonaceous matter in the early Archaean: a petrographic
and geochemical study of the carbonaceous cherts of the Swazi-
land Supergroup. 115–132. In LO W E, D. R. and B Y E RL Y , G.
R. (eds) Geologic evolution of the Barberton Greenstone Belt,
South Africa. Geological Society of America Special Paper, 329.
-
and WE S T A L L, F. 2003. Archean biofilms preserved in
the Swaziland supergroup, South Africa. 307–316. In KRUM-
BE I N , W. E., P A T E RS O N , D. M. and ZA VA R Z I N , G.
A. (eds) Fossil and recent biofilms: A natural history of life on
earth. Kluwer.
WE S T A L L, F., C A V A L A Z ZI , B., LE M E L LE , L., MA R -
RO C C H I , Y., R O U Z A U D , J. N., S I MI O N O V I C I , A.,
SALOME
´, M., MO S T E F AO U I , S., AN DR E A Z Z A, C.,
FO U C H E R, F., T O P O R S K I , J., JA U S S , A., TH I E L , V.,
SO U T H A M, G., M A C L E A N, L., W I R I C K , S., HO F-
MA N N , A., ME I B O M , A., RO B ER T , F. and D E
´FARGE,
C. 2011. Implications of in situ calcification for photosynthesis
in a <3.3 Ga-old microbial biofilm from the Barberton Green-
stone Belt, South Africa. Earth & Planetary Science Letters,
310, 468–479.
-
CA M P B E L L , K. A., B R E
´HE
´RE T , J. G., F O U C H E R,
F., GA U T R E T , P., H U B E R T , A., S O R I E U L , S.,
GR A S S I N E A U , N. and G U I D O , D. M. 2015. Archean
(3.33 Ga) microbe-sediment systems were diverse and flour-
ished in a hydrothermal context. Geology,43, 615–618.
XI E , X., BY E R LY , G. R. and FE RR E L L , R. E. Jr 1997. IIb
trioctahedral chlorite from the Barberton greenstone belt: crys-
tal structure and rock composition constraints with implica-
tions to geothermometry. Contributions to Mineralogy &
Petrology,126, 275–291.
YE E , N., BE N N IN G , L. G., P H OE N I X , V. R. and FE R-
RI S , F. G. 2004. Characterization of metal-cyanobacteria
sorption reactions: a combined macroscopic and infrared spec-
troscopic investigation. Environmental Science & Technology,
38, 775–782.
HICKMAN-LEWIS ET AL.: DIVERSITY IN ARCHAEAN MICROBIAL MATS 27
