LETTERS PUBLISHED ONLINE: 21 AUGUST 2011 | DOI: 10.1038/NGEO1238 Microfossils of sulphur-metabolizing cells in 3.4-billion-year-old rocks of Western Australia David Wacey 1,2 * , Matt R. Kilburn 1 * , Martin Saunders 1 , John Cliff 1 and Martin D. Brasier 3 Sulphur isotope data from early Archaean rocks suggest that microbes with metabolisms based on sulphur existed almost 3.5 billion years ago, leading to suggestions that the earliest microbial ecosystems were sulphur-based 1–5 . However, mor- phological evidence for these sulphur-metabolizing bacteria has been elusive. Here we report the presence of microstruc- tures from the 3.4-billion-year-old Strelley Pool Formation in Western Australia that are associated with micrometre-sized pyrite crystals. The microstructures we identify exhibit indica- tors of biological affinity, including hollow cell lumens, carbona- ceous cell walls enriched in nitrogen, taphonomic degradation, organization into chains and clusters, and δ 13 C values of -33 to -46 Vienna PeeDee Belemnite (VPDB). We therefore identify them as microfossils of spheroidal and ellipsoidal cells and tubular sheaths demonstrating the organization of multiple cells. The associated pyrite crystals have  33 S values between -1.65 and +1.43 and δ 34 S values ranging from -12 to +6 Vienna Canyon Diablo Troilite (VCDT) 5 . We interpret the pyrite crystals as the metabolic by-products of these cells, which would have employed sulphate-reduction and sulphur-disproportionation pathways. These microfossils are about 200 million years older than previously described 6 microfossils from Palaeoarchaean siliciclastic environments. Evidence of cellular organization would represent one of the strongest lines of evidence for a Palaeoarchean biosphere, but this has been beset with controversy 7,8 . At present, microbial mats 9–11 together with sulphur isotope analysis 1–5 provide the best insights into Palaeoarchean microbial metabolisms and ecosystems, with evidence reported for phototrophs 9–11 plus hydrogen-based 10 and sulphur-based 1–5 metabolisms. However, these reports lack evidence for accompanying cellular morphology. Here, we provide such evidence in the form of well-preserved cells closely associated with pyrite in the basal sandstone member of the ∼3,400 Myr-old Strelley Pool Formation (SPF), Western Australia. The SPF crops out across eleven greenstone belts within the East Pilbara Terrane, spanning a ∼75 Myr hiatus in volcanism between the 3,520–3,427 Myr-old Warrawoona Group and the 3,350–3,315 Myr-old Kelly Group 12 . Our microfossils come from the East Strelley greenstone belt (Supplementary Fig. S1), where the SPF lies above an unconformity on top of eroded ∼3,515 Myr- old volcanics 13 . Here, the SPF records a marine transgression across one of Earth’s earliest preserved shorelines, with the basal sandstone deposited in a shallow-water beach or estuarine setting 14 , and the overlying carbonates deposited in a marine carbonate platform setting 11,15 . Early silica cements in the sandstones include isopachous phreatic cements (Supplementary Fig. S2), 1 Centre for Microscopy, Characterization and Analysis, The University of Western Australia, 35 Stirling Highway, Crawley, Western Australia 6009, Australia, 2 School of Earth and Environment, The University of Western Australia, 35 Stirling Highway, Crawley, Western Australia 6009, Australia, 3 Department of Earth Sciences, University of Oxford, South Parks Road, Oxford OX1 3AN, UK. *e-mail:David.Wacey@uwa.edu.au; martinb@earth.ox.ac.uk. accompanied near the base by dripstone and meniscus fabrics (Supplementary Fig. S3) formed in the vadose zone, indicating partially gas-filled pore spaces (see ref. 16), probably in the photic zone. Associated density concentrates of rounded detrital pyrite with mass-independently fractionated sulphur isotope signatures show this gas was low in oxygen 5 . The microfossils reported here are slightly older than both the stromatolites 11 and microfossils of unknown metabolic affinity from stratiform chert higher up in the SPF (ref. 17; see Supplementary Table S1 for comparison). Candidate microfossils (Figs 1, 2) are restricted to stratiform black sandstone at the base of the member, or to rounded, reworked clasts of this black sandstone found ∼0.5–2 m above (Supplementary Fig. S1). Their syngenicity is constrained by field, petrographic and geochemical data. First-order Raman spectra from the microfossils possess disordered ‘D1’ and ordered ‘G’ carbon peaks (Supplementary Fig. S4) with D1/G peak heights and areas consistent with thermally mature disordered carbonaceous material that has experienced approximately lower greenschist facies metamorphism 18 . High-resolution transmission electron microscopy (HRTEM) reinforces the Raman data, revealing mostly disordered carbon plus small domains where ordered lattice fringes have the 0.34 nm interplane spacing of graphite (see ref. 19). Together, these data rule out a post-metamorphic origin for the candidate microfossils. However, as the SPF experienced greenschist facies metamorphism over an extended time period 20 , these data cannot prove a syn-depositional age for the microfossils. Instead, syngenicity is confirmed by the spatial occurrence of the microfossils and the nature of silica cementation. Microfossils are restricted either to beds of carbonaceous/pyritic sandstone or to reworked intraclasts eroded from that lithology; they are absent from surrounding pale sandstone that lacks carbon and pyrite (Supplementary Fig. S5). Both the bedded black sandstone and the black clasts have similar successions of quartz cements and similar distributions of microfossils within those cements. At least two generations of microfossils are present; an earlier generation, found within the earliest isopachous silica cements, coating sand grains directly (Fig. 1c,f,i,j); and a later generation found within remaining pore spaces, in places coexistent with meniscus and dripstone silica cements (Supplementary Fig. S3). The presence of cemented intraclasts requires those cements to have formed within reach of erosion close to the sediment–water interface, and the high ‘minus-cement’ porosities indicate all inter-granular cements were in place before significant compaction took place 21 . Given that the sandstone was buried by ∼20 m of carbonate sediment by 3,350 Myr ago and ∼2 km of basalt by 3,325 Myr ago (ref. 20), the microfossils and their enclosing cements must be >3,350 Myr in age. No NATURE GEOSCIENCE | ADVANCE ONLINE PUBLICATION | www.nature.com/naturegeoscience 1 © 2011 Macmillan Publishers Limited. All rights reserved.