Fungi have recently been found to comprise a significant part of the deep biosphere in oceanic sediments and crustal rocks. Fossils occupying fractures and pores in Phanerozoic volcanics indicate that this habitat is at least 400 million years old, but its origin may be considerably older. A 2.4-billion-year-old basalt from the Palaeoproterozoic Ongeluk Formation in South Africa contains filamentous fossils in vesicles and fractures. The filaments form mycelium-like structures growing from a basal film attached to the internal rock surfaces. Filaments branch and anastomose, touch and entangle each other. They are indistinguishable from mycelial fossils found in similar deep-biosphere habitats in the Phanerozoic, where they are attributed to fungi on the basis of chemical and morphological similarities to living fungi. The Ongeluk fossils, however, are two to three times older than current age estimates of the fungal clade. Unless they represent an unknown branch of fungus-like organisms, the fossils imply that the fungal clade is considerably older than previously thought, and that fungal origin and early evolution may lie in the oceanic deep biosphere rather than on land. The Ongeluk discovery suggests that life has inhabited submarine volcanics for more than 2.4 billion years.
The deep biosphere, hidden beneath land and sea, represents a major portion of life’s habitats and biomass on Earth1. In spite of significant discoveries from scientific ocean drilling and metagenomics, the deep biosphere remains largely uncharted and its geological history almost entirely unknown. The deep habitats are protected from most of the hazards of surface life, and the deep environments would have been potentially available to life from the early stages of Earth’s history. Here, we report filamentous structures preserved in carbonate- and chlorite-filled amygdales and fractures in basaltic lavas of the 2.4-Gyr-old Ongeluk Formation, South Africa. Their morphology, dimensions and striking similarity to fungi in Phanerozoic volcanics2,
The Ongeluk Formation is a 900-m-thick succession of basalts in the Griquatown West Basin, South Africa. The lavas are regionally extensive and comprise massive flows, pillow lavas and hyaloclastites that extruded onto the seafloor around 2.4 Ga; the basalts have undergone only very low-grade metamorphism (Supplementary Discussion). The fossiliferous sample (AG4) is a 25-cm-long ¼ core derived from drill depth 21.79–22.04 m of the Agouron drill hole GTF01, which penetrated the lower part of the Ongeluk Formation (Fig. 1). About 70 of the ~100 observed amygdales contain filaments.
The sample is a chlorite-altered basalt with a relict igneous texture consisting of pseudomorphs of pyroxene and plagioclase (Supplementary Discussion). The groundmass consists of intergrown chlorite, K-feldspar, quartz and calcite, with accessory apatite and Fe-Ti oxides. Amygdales and veins are present, characterized by chlorite and calcite representing mineral infills of original vesicles and fractures in the lavas. The spherical to subspherical amygdales are up to 1.5 mm in diameter (Figs 2 and 3). Most have rims composed of masses of very fine-grained, brownish green chlorite, chlorite 1. Thermometry of chlorite 1 yields metamorphic temperatures in the range of 179–260 °C (Supplementary Fig. 1; Supplementary Discussion). Where filaments are present, they are defined by chlorite 1. No carbonaceous material has been detected within the filaments (Supplementary Fig. 2a). Calcite typically forms a cylindrical layer of constant thickness around the filaments; the blocky arrangement of crystals in the calcite, without clear relation to filament morphology (Fig. 4e), suggests that the calcite has been recrystallized. Fine-grained chlorite 1 fills the space between the calcite cylinders (Figs 2b,d,e,g and 4; Supplementary Figs 3a,b, 4 and 5). A second generation of chlorite, chlorite 2, coarser-grained and apple green, commonly intergrown with quartz and chalcopyrite, is present in some amygdales and fractures. Chlorite 2 is not pervasive but overprints chlorite 1, including filaments defined by chlorite 1 (Fig. 4d,e; Supplementary Fig. 4d). Thermometry of chlorite 2 gives metamorphic temperatures of 319–411 °C (Supplementary Figs 1 and 5; Supplementary Discussion). Its association with chalcopyrite, occurrence in veins and otherwise non-pervasive distribution suggest that the growth of chlorite 2 was linked to hydrothermal fluids.
Filament structure and morphology
The filaments extend from rims of chlorite 1 attached to amygdale and fracture walls, and form a tangled network inside vesicles and fractures in the rock (Figs 2, 3, 4; Supplementary Fig. 3). The density of the filamentous network typically decreases towards the centre of the cavities (Fig. 2b,d,j, 3a,e and 4b; Supplementary Fig. 3a,b). The chlorite rim represents an uneven basal film consisting of a jumbled mass with little space remaining between filaments (Figs 2j,k and 3e; Supplementary Fig. 3). Scanning electron microscopy (SEM)/back-scattered electron (BSE)/wavelength-dispersive X-ray spectroscopy (WDS) images confirm that the structure and composition are identical between filaments and basal film (Fig. 2j,k; Supplementary Figs 4 and 5).
Filaments are 2–12 μm wide; the width is usually constant within a filament. No internal septa have been identified, but original internal structure is not preserved (Fig. 2k,l). The filaments typically form straight or curved sections, rarely with irregular wiggly parts. Filaments frequently form loops of different diameter, from about 10 μm (Fig. 3h) to 80 μm or more (Fig. 3i).
Branchings at acute angles, Y-junctions, are common among the free filaments (Figs 2c and 3f,g). T-junctions also occur (Fig. 3g), although considerably less frequently. Filaments with different orientation commonly touch and entangle each other (Fig. 3f,i), and crossing filaments sometimes seem to merge seamlessly. Where none of the filaments change direction, the crossing is interpreted as coincidental (Fig. 2l). This phenomenon of taphonomic/diagenetic filament merging makes it sometimes difficult to identify true branching, where a single filament is split into two. When Y-junctions on the same apparently branching filament point in opposite directions (Fig. 3c), one or both junctions may represent false branching; this can also be indicated by the filament being thicker, or even appearing doubled, below a Y-junction. There are, however, a number of cases where the morphology of the junction leaves little doubt of true branching (Figs 2c and 3f,g). In particular, where successive Y-branching takes place from a stem of constant diameter, the branching is real and not due to bundling of separate filaments (Fig. 3f,g).
Anastomoses, where a branched-off filament meets and merges with another, occur with some frequency (Figs 2c,f and 3b,c). As with branching, it may be difficult to distinguish coincidental coming-together of independent filaments from true anastomoses, but the frequency of apparent anastomoses with consistent morphology (for example, Fig. 3b,c) indicates that the phenomenon is real. Nonetheless, anastomoses do not dominate the filament tangles to the extent that they form interlocking networks.
The filaments sometimes carry bulbous protrusions, 5–10 μm in diameter. These tend to congregate on the basal parts of filaments and on basal films, and be more rare on distal parts of filaments (Fig. 3j).
A recurring feature is a bundle of filaments giving off diverging branches to form a broom-like structure, here termed ‘broom’, that extends from the basal film or from the substrate (Figs 2e,g,h and 3d,e). In some vesicles, there are brooms consisting of tens of diverging filaments, some with their bases apparently attached to the vesicle wall and some produced by branching (Supplementary Fig. 3c,d).
The basalt is permeated by veins that are frequently seen to connect to the spherical/subspherical vesicles (Fig. 2a,i; Supplementary Video). The veins, down to 5 μm in width, are filled with chlorite and calcite similar to that which fills the vesicles. One large vein, >2.2 mm long and >0.2 mm wide, comprises a zone of densely intertwined filaments that occurs between the basalt wall rock and the centre of the vein. Filaments adjacent to the margin of the vein are commonly truncated by chlorite sheets and veinlets (Fig. 4), representing a later stage of chlorite growth (chlorite 2; Supplementary Discussion).
Biogenicity and syngenicity
Crucial to the interpretation of the filaments are the issues of biogenicity and syngenicity: do the filaments represent biological organisms and when did they form relative to the age of the rock? Filamentous fabrics are not uncommon in basaltic rocks, although most reported cases refer to tunnelling in volcanic glass and its alteration products8. Both biogenic and abiogenic mechanisms may be responsible for such tunnels, and distinguishing between the two causes is difficult and controversial9,
Although tunnels may take on a variety of shapes, including branching and dendritic ones13, several features of the Ongeluk structures are incompatible with tunnels. The frequent fusing of adjacent filaments (Fig. 3f,i), resulting in false branching (Fig. 3c), implies that they are physical entities often touching and entangling each other. This is consistent with flexible filaments in a void, but not with tunnelling in rock. Similarly, the recurring cases of anastomosis (Figs 2 and 3) are difficult to reconcile with tunnels.
The morphology of the filaments and the mineral paragenetic sequence in the fractures (Fig. 4) are identical to those of the adjacent vesicles, implying that the vesicles, like the fractures, started out as voids and underwent the same history of colonization and paragenesis.
A number of different spherical or globular structures are found in volcanic and subvolcanic rocks14. They may be formed as gas bubbles in the magma (vesicles), as radial growth of crystals (spherulites), or as the result of immiscibility of component magmatic fluids (varioles). Vesicles usually become filled by secondary minerals formed at low temperatures, forming amygdales. The Ongeluk spherical structures are filled with minerals (mainly calcite and chlorite) characteristic of amygdales; they show neither spherulitic structure nor magmatic composition, and so may confidently be interpreted as having begun as gas bubbles (Supplementary Discussion).
The Ongeluk filaments fulfil established criteria15 distinguishing cryptoendoliths (cavity-dwellers) and chasmoendoliths (fracture-dwellers) from euendoliths (rock-borers) and abiotic processes forming microtunnels in rock (Supplementary Discussion). They show pre-metamorphic growth into fluid-filled cavities, curvilinear and branching forms with circular cross-section and non-uniform diameter, and preservation in clays with or without organic matter in carbonate-filled vesicles; all listed as characters typical of crypto- and chasmoendoliths15.
The authors of a previous study16 reported a variety of structures interpreted as ambient inclusion trails in an Archaean pyroclastic tuff. Their ‘type 1 microtubes’ show a compositional similarity with the Ongeluk structures: both have chloritic cores surrounded by calcite. They differ from the latter, however, in being straight and very regular.
A commonly stated criterion for biogenicity of microfossils is the presence of original organic carbon in the structures; this has even been cited as a necessary criterion12,17. However, organic carbon is seldom preserved in environments of highly oxidized minerals, such as calcite or hematite; organically preserved microfossils are predominantly found under preservational conditions of low permeability and reactivity, as in cherts18. As the absence of organic carbon in a fossil is seldom reported in the literature, the lack of such carbon is frequently overlooked. It is, however, a common condition18. For example, we have investigated well-preserved iron-oxidizing bacteria in a Quaternary microbialite where filaments are encrusted with hematite, and Raman spectroscopy failed to reveal any organic carbon signal19. The lack of detectable carbonaceous matter in the Ongeluk filaments is thus not a valid argument against their biogenicity.
With regard to syngenicity, the organisms must have invaded the Ongeluk lavas while the vesicles and fracture-controlled porosity were still open to the water column, a window that probably closed after ca. 10 Myr following the eruption of the lavas20. In any case, they should not be younger than ca. 2.06 Ga, at which time chloritization would have taken place (Supplementary Discussion). Supplementary Fig. 6 depicts the proposed formation sequence from invasion of the organism through diagenesis and metamorphism.
Raman spectroscopy indicates the presence of carbonaceous material in the Ongeluk host basalt that has not been subjected to temperatures higher than about 200 ± 30 °C; carbonaceous material in the Ongeluk basal sandstone and the underlying Makganyene diamictite yields temperatures of around 370 °C (Supplementary Fig. 7; Supplementary Discussion). The origin of the carbon is unknown; it shows no affinity to the filaments (Supplementary Fig. 2).
Biology of the filaments
Filamentous growth is a recurring characteristic in many multicellular prokaryotes and algae, but mycelial networks consisting of branching filaments are known mainly from three modern groups of organisms: actinobacteria, fungi and the fungus-like eukaryotic oomycetes. Mycelium-forming actinobacteria produce radiating networks of branching filaments, 0.15–1.5 μm in diameter. Anastomoses are generally absent21; occasional reports of anastomoses in Streptomyces have not been confirmed22. Many actinobacteria form spores, about 1 μm in diameter, on the mycelium, sometimes in sporangia 5–20 μm in size23. Actinobacteria have a wide distribution in aquatic and terrestrial habitats, including various extreme environments24.
Like actinobacteria, fungi are widely distributed in terrestrial and aquatic habitats, and they have recently been shown to be common inhabitants of deep marine sediments and crustal rocks5,25,
Fungi have recently been found to play a leading role in the Phanerozoic subsurface biota through the discoveries of fossilized fungal mycelia in vesicles in Devonian, Eocene and Quaternary submarine volcanics2,
The oomycetes were previously thought to be fungi, but molecular systematics now places them close to the photosynthetic stramenopiles34. Anastomoses between hyphae occasionally occur, but as a form of conjugation, not a mechanism to form interlocking networks35.
When compared with modern mycelial organisms, the Ongeluk fossils in hyphal dimensions, network architecture and mode of life seem most consistent with fungi. If the 5–10 μm bulbous protrusions are spores, those too agree with fungal but not actinobacterial dimensions. Ongeluk anastomoses closely mimic those in modern fungi (compare our Figs 2c,f and 3b,c with anastomoses in ref. 36, Fig. 1). Other features of the Ongeluk fossils, such as the basal film and the tendency of filaments to protrude from it as brooms, are consistent with fungal mycelial morphology (for example, the mycelial cords developed by many fungi under conditions of starvation37). The growth habit of the Ongeluk filaments in basaltic vesicles is morphologically almost identical to that seen in fungi in Phanerozoic volcanics (Supplementary Fig. 8)2,
Although on the basis of morphology we cannot exclude the possibility that the Ongeluk fossils represent a separate branch of fungus-like organisms, the similarities with fungi in the corresponding Phanerozoic settings are striking. The presence of fungi in early Palaeoproterozoic submarine volcanic rocks would, however, overturn current concepts on the timing and circumstances of fungal origin and evolution. There is a strong consensus that fungi and nucleariids comprise the sister group of holozoans within the clade Opisthokonta38,
Estimates of node ages from molecular clocks rely on calibration against the fossil record. Whereas the Phanerozoic fossil record is sufficiently reliable to yield useful calibration points49, the Proterozoic record is notoriously spotty, and interpretations of Proterozoic fossils are frequently controversial (for example, alleged Proterozoic fungi46). Ages of Proterozoic nodes are therefore typically based on extrapolations from Phanerozoic calibration points. Irrespective of the formidable molecular clock problems, the existence of fungi near the beginning of the Proterozoic, before or at the very early stage of the Great Oxidation Event, would raise issues about the existence of other major eukaryote branches at the time.
Whether or not the Palaeoproterozoic Ongeluk fossils represent fungi, the occurrence of remarkably similar fossils in Phanerozoic vesicular basalts (Supplementary Fig. 8)2,
The sample was cut to produce 14 petrographic thin sections and 7 pillars, 2 mm wide. The sections were studied using optical microscopy and SEM/environmental scanning electron microscopy (ESEM), as well as synchrotron-radiation X-ray tomographic microscopy (SRXTM).
The images in Fig. 2j–l were obtained with a Philips XL30 ESEM with a field emission gun (XL30 ESEM-FEG) and a high-contrast BSE detector. The acceleration voltage was 20 kV. The samples were not coated.
Chlorite and carbonate (calcite) within the amygdales were analysed with an Oxford Instruments X-Max50 energy-dispersive X-ray detector (EDS) mounted on a TESCAN VEGA3 SEM. Aztec software was used to collect and process the X-ray spectra from carbon-coated thin sections. Fully quantitative data were collected at 15 kV accelerating voltage and 1–2 nA beam current. The system count rate was calibrated using pure copper, and standard spectra were collected on the VEGA3 from jadeite (Na), periclase (Mg), corundum (Al), wollastonite (Si and Ca), orthoclase (K), rutile (Ti), chromite (Cr), rhodonite (Mn) and pyrite (Fe). Analytical precision is ±1–2% relative for major elements (>10 wt%) and ±5–10% relative for minor elements (<10 wt%). Twenty-five point analyses from chlorite 1 and 25 point analyses from chlorite 2 were collected at 15 kV and 20 nA using a JEOL 8530F electron microprobe fitted with five WDS. Quantitative analyses were derived using Probe for EPMA from Probe Software. Standards used were jadeite (Na), periclase (Mg), corundum (Al), wollastonite (Si and Ca), orthoclase (K), rutile (Ti), Cr2O3 (Cr), spessartine (Mn) and magnetite (Fe). Analytical precision is ±1% relative for major elements (>10 wt%) and ±5–10% relative for minor elements (<10 wt%).
Element distribution maps
Qualitative element distribution maps of amygdales from thin section ZBF061 were generated using an FEI Verios XHR SEM, operating with 15 kV accelerating voltage and approximately 1–2 nA beam current. X-ray maps were collected and processed using an Oxford Instruments X-Max80 EDS and Aztec software. The section was coated with a 20-nm-thick carbon film before analysis. Element distribution maps for temperature mapping were collected from part of an amygdale from thin section ZBF061 with the JEOL 8530F at 15 kV accelerating voltage and 50 nA beam current. The maps were collected and processed (deadtime, background and overlap corrections) using Probe for EPMA software from Probe Software. The corrected data were read into XMapTools50, where they were calibrated using the WDS point analyses and converted into quantitative maps of the element oxides. Pixel by pixel (1 × 1 μm) temperature estimates, based on the calibration of ref. 51, were made in XMapTools and plotted as a temperature map.
Raman spectra were collected using a confocal laser Raman microspectrometer (Horiba instrument LabRAM HR 800; Horiba Jobin Yvon, Villeneuve d’Ascq, France), equipped with a multichannel air-cooled (–70 °C) 1,024 × 256 pixel charge-coupled device detector at the Department of Geological Sciences, Stockholm University. Acquisitions were obtained with a 1,800 lines mm−1 grating. Excitation was provided by an Ar-ion laser (λ = 514 nm) source. Spectra were recorded using a low laser power of 0.1–1 mW at the sample surface to avoid laser-induced degradation of the samples. Sampling was carried out using an Olympus BX41 microscope coupled to the instrument, and the laser beam was focused through a ×100 objective to obtain a spot size of about 1 μm. The spectral resolution was ~0.3 cm−1 per pixel. The typical exposure time was 10 s with 10 accumulations. The accuracy of the instrument was controlled by repeated use of a silicon wafer calibration standard with a characteristic Raman line at 520.7 cm−1. Instrument control and data acquisition were made with LabSpec 5 software.
SRXTM was carried out at the TOMCAT beamline of the Swiss Light Source at the Paul Scherrer Institute, Villigen, Switzerland. X-ray energy was set to 15 keV for petrographic thin sections and 28 keV for sawn-out 2 mm pillars. Objectives ×4, ×10 and ×20 were used, for a voxel size of 1.625 μm, 0.65 μm and 0.325 μm, respectively. Pillars were first scanned in total at low magnification to identify fossiliferous vesicles later to be scanned at higher magnifications. For the results presented here, 1,501 projections were acquired equiangularly over 180°, online post-processed and rearranged into flat- and darkfield-corrected sinograms. Reconstruction was performed on a Linux PC farm using highly optimized routines based on the Fourier transform method52. Slice data derived from the scans were analysed and rendered using Avizo software.
The illustrated material is deposited at the Swedish Museum of Natural History, Stockholm. The datasets generated and/or analysed during the current study are available from the corresponding author (S.B.) on request.
How to cite this article: Bengtson, S. et al. Fungus-like mycelial fossils in 2.4-billion-year-old vesicular basalt. Nat. Ecol. Evol. 1, 0141 (2017).
Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Our work has been supported by the Agouron Institute, Swedish Research Council (2012-4364, 2013-4290), Danish National Research Foundation (DNRF53), Australian Research Council (DP110103660, DP140100512), Paul Scherrer Institute (20130185, 20141047), Australian Microscopy & Microanalysis Research Facility, National Science Foundation (EAR-05-45484), NASA Astrobiology Institute (NNA04CC09A), Natural Sciences and Engineering Research Council, and the European Commission CALIPSO programme (312284). We thank V. Belivanova for technical assistance, P. von Knorring for drafting Supplementary Fig. 6, J. Peckmann for supplying images for Supplementary Fig. 8c,d and A. Tehler for discussions.