Reassessing the biogenicity of Earth’s oldest trace fossil with implications for biosignatures in the search for early life

Significance

It has been argued that Archean subseafloor pillow lava sequences provide an environment in which to seek evidence for the earliest traces of life. Candidate titanite biosignatures of microbial activity have been reported in ∼3.45-Ga metavolcanic glass from the Barberton greenstone belt of South Africa. In this paper we present new in situ U–Pb age data, metamorphic constraints, and morphological observations on these titanite microtextures. Our data challenges a biological origin for these oldest purported trace fossils, with implications for the ecological niches where life may have first emerged. We therefore suggest alternative biosignatures and approaches should be considered in the search for subsurface life on early Earth and in extraterrestrial mafic–ultramafic rocks, for example, in martian basalts.

Abstract

Microtextures in metavolcanic pillow lavas from the Barberton greenstone belt of South Africa have been argued to represent Earth’s oldest trace fossil, preserving evidence for microbial life in the Paleoarchean subseafloor. In this study we present new in situ U–Pb age, metamorphic, and morphological data on these titanite microtextures from fresh drill cores intercepting the type locality. A filamentous microtexture representing a candidate biosignature yields a U–Pb titanite age of 2.819 ± 0.2 Ga. In the same drill core hornfelsic-textured titanite discovered adjacent to a local mafic sill records an indistinguishable U–Pb age of 2.913 ± 0.31 Ga, overlapping with the estimated age of intrusion. Quantitative microscale compositional mapping, combined with chlorite thermodynamic modeling, reveals that the titanite filaments are best developed in relatively low-temperature microdomains of the chlorite matrix. We find that the microtextures exhibit a morphological continuum that bears no similarity to candidate biotextures found in the modern oceanic crust. These new findings indicate that the titanite formed during late Archean ca. 2.9 Ga thermal contact metamorphism and not in an early ca. 3.45 Ga subseafloor environment. We therefore question the syngenicity and biogenicity of these purported trace fossils. It is argued herein that the titanite microtextures are more likely abiotic porphyroblasts of thermal contact metamorphic origin that record late-stage retrograde cooling in the pillow lava country rock. A full characterization of low-temperature metamorphic events and alternative biosignatures in greenstone belt pillow lavas is thus required before candidate traces of life can be confirmed in Archean subseafloor environments.

Filamentous titanite microtextures in ca. 3.472–3.432 Ga metavolcanic pillow lavas of the Barberton greenstone belt (BGB), South Africa, have been argued to represent Earth’s oldest trace fossils (1–4). Subsequent work in other Archean terrains, such as the Pilbara Craton of Western Australia and the Abitibi greenstone belt of Canada, have made similar claims for trace fossils in early metabasaltic pillow lavas (5, 6). The current paradigm for these titanite microtextures involves a complex bioalteration model in which endolithic microbes form hollow microtunnels by etching fresh volcanic glass in submarine environments on the early Earth (1–7). Subseafloor hydrothermal alteration is envisioned not only to provide a suitable environment for microbial activity, but also to result in mineralization of the hollow tubes by titanite (CaTiSiO₅) preserving the proposed trace fossils (1–3). Thus, it has been suggested that volcanic habitats represent a previously unexplored geological setting in which life may have thrived and possibly originated on the early Earth (1–3). It has also been argued that similar microtextures in altered extraterrestrial basalts may provide a useful biosignature in the search for life on Mars and beyond (8).

The principal argument for a biogenic origin of the purported Archean trace fossils is based on their apparent similarity in size, shape, and distribution to partially mineralized microtubules of argued biogenic origin from pillow lavas of the in situ oceanic crust and ophiolites (1, 2, 9). In young, in situ pillow lavas, abundant microbial DNA and geochemical signatures support the presence of a deep subseafloor biosphere (10, 11). Given therefore the major implications that the Archean titanite microtextures may hold for the earliest evidence of life on Earth, we have conducted syngenicity and biogenicity tests to evaluate a proposed subseafloor bioalteration model for their origin. We report in situ U–Pb dating of the titanite, quantitative microscale mapping of metamorphic conditions, and morphological evidence to evaluate the biogenicity of the titanite microtextures.

Results

The titanite microtextures were originally described from pillow lavas of the ca. 3.472-Ga Hooggenoeg Formation of the BGB that stratigraphically overlies ca. 3.482 Ga ultramafic komatiites of the Komati Formation (Fig. S1). The current study includes samples from drill core KD2b of the Barberton Scientific Drilling Project (BSDP) that intercepted the pillow lava type section for the proposed biotextures (12, 13) (Fig. S1). Petrographic investigation of the drill core samples reveals that the original titanite bioalteration textures (1) form part of a morphological continuum that was previously only partially illustrated. The size, morphology, and distribution of these titanite microtextures are evaluated against biogenicity criteria proposed for endolithic microborings (14, 15). The simplest morphological endmember is hornfelsic titanite spheres (Fig. 1 A–E) that may coalesce to form irregular bands. The spheres and bands develop filamentous projections (Fig. 1 A, C, and D), showing a growth continuum from spheres with one or two short projections to larger clusters of titanite microtextures with more numerous and longer projections (Fig. 1 B and D–F; Fig. S2). The latter morphotype with well-developed titanite filaments has been the focus of all previous studies but forms only a less abundant endmember of the morphological continuum. The individual filaments are highly variable in length (e.g., Fig. 1F, from 14 to 460 µm) and width (from 2 to 39 µm) and are either constant in diameter, tapered (Fig. 1E), or club shaped (Fig. 1G). The filaments do not uniquely radiate at high angles from titanite “root zones” as in modern oceanic basalts (Fig. 1 G and H), but rather at varying angles from irregular titanite bands and spheres disseminated throughout the chlorite matrix (Fig. 1 E and F). The new drill core samples reveal that the titanite filaments have a much wider range of diameters and are typically an order of magnitude thicker in comparison with the modern tubular microtextures (Fig. 1 H and I). In addition, they lack complex morphological features such as annulations and spiral forms (16) seen in younger volcanic glass (Fig. 1H). These new findings indicate no morphological similarity between the Archean titanite microtextures and the modern tubular bioalteration textures.

Fig. 1.

Titanite microtextures in Archean metavolcanic pillow lavas now comprising a chlorite dominated matrix, compared with partially mineralized microtunnels from younger oceanic crust. (A–F) A continuum of titanite microtextures from spheres with or without filamentous projections, to well-developed clusters with radiating filaments, LA tracks for U–Pb titanite dating are shown (white bands in C). (G) Partially mineralized microtubules radiating at high angles from a fracture in volcanic glass from Ocean Drilling Program hole 418A in the West Atlantic. (H) Curvilinear and spiral-shaped (Inset) microtunnels in volcanic glass of the Troodos ophiolite Cyprus (drill core CY-1A). (I) Histogram of measured Archean titanite filament widths (n = 303) in 12 samples from the Barberton drill core showing that they are much larger in diameter (dark green line = mean of 12 µm and light green band = SD) and span a wider range compared with microtubules in younger volcanic glass (dark purple line = mean of 1.3 µm and light purple band = SD, replotted from ref. 2). (Scale bars: 50 µm A–H except G and H, Insets, which are 10 µm.)

As a syngenicity test, the U–Pb age of titanite growth was determined by in situ measurements using single- and multicollector laser ablation, inductively coupled plasma mass spectrometry (LA-ICP-MS). Further details on the methods and analytical standards are presented in SI Materials and Methods. The ²⁰⁷Pb/²⁰⁶Pb and ²³⁸U/²⁰⁶Pb isotopic ratios measured by single-collector LA-ICP-MS on spherical hornfelsic titanite (sample B77) are plotted on a Tera–Wasserburg Concordia diagram (Fig. 2; Tables S1 and S2). The measured U–Pb and Pb–Pb isotope ratios are directly plotted requiring no additional assumptions for age determination. Most of the U–Pb ratios plot near the upper intercept, indicating significant incorporation of common Pb in the titanite (Fig. 2A). However, the measured isotopic ratios do show a significant spread enabling a lower intercept value to be calculated yielding an age of 2.913 ± 0.31 Ga for titanite growth (Fig. 2 A and B). Using a Monte Carlo statistical approach, the lower intercept values indicate a mean peak age of 2.871 ± 0.33 Ga for sample B77 (Fig. 2C).

Fig. 2.

U–Pb isotopic data for hornfelsic titanite in pillow lava metabasalt sample B77 of the Hooggenoeg Formation, South Africa, determined by single-collector LA-ICP-MS. (A) Measured ²⁰⁷Pb:²⁰⁶Pb and ²³⁸U:²⁰⁶Pb isotopic ratios plotted on a Tera–Wasserburg Concordia diagram yielding a lower intercept age of ca. 2.913 ± 0.31 Ga. (B and C) Monte Carlo statistical calculation plotted as a histogram gives an average lower intercept age of 2.871 ± 0.31 Ga. These titanite ages overlap with late Archean mafic dyke intrusion ages and not with early seafloor hydrothermal alteration.

A typical candidate bioalteration texture was also analyzed (sample B74) to determine the age of titanite growth using both the Tera–Wasserburg and Wetherill Concordia methods (Fig. 3 A and B; Tables S1–S3). A first estimate of the titanite age was conducted using the single-collector LA-ICP-MS, yielding a lower intercept age of 2.828 ± 0.44 Ga (Fig. 3A; Tables S1 and S2). To improve the precision of this age estimate, further isotopic measurements were conducted using the more sensitive multicollector LA-ICP-MS (Table S3). This method requires a common Pb correction (4, 17). In this study, correction for common Pb was based on measurement of the nonradiogenic ²⁰⁴Pb in the titanite and used the Pb isotope composition of an Archean galena from the Old Star Mine. This galena sulphide has the least radiogenic Pb isotope composition so far reported from the Archean Kaapvaal Craton, giving the best estimate of common Pb at the time of titanite growth (18). This gave data error ellipses with a significant spread beneath the Wetherill Concordia curve, and yielded an upper intercept age of 2.819 ± 0.20 Ga (Fig. 3B). This titanite age estimate is identical to that determined by the Tera–Wasserburg method, but with dramatically improved precision.

Fig. 3.

U–Pb isotopic data for Archean titanite microtextures in sample B74, previously interpreted as trace fossils, determined by LA-ICP-MS. (A) Measured ²⁰⁷Pb:²⁰⁶Pb and ²³⁸U:²⁰⁶Pb isotopic ratios determined by single-collector LA-ICP-MS and plotted on a Tera–Wasserburg Concordia diagram, yielding a lower intercept age of 2.828 ± 0.440 Ga. (B) A more precise age of 2.819 ± 0.200 Ga determined using a multicollector LA-ICP-MS, and the Wetherill Concordia method.

Using an electron microprobe mapping approach combined with thermodynamic modeling of chlorite crystallization conditions (19, 20), microscale metamorphic temperature conditions surrounding a candidate biosignature (in B80) were calculated (Fig. 4 A–D). The resulting map covers an area of 450 × 450 μm, with a 3-μm pixel size (image pixel resolution of 150 × 150). The filamentous titanite microtextures of interest are highlighted by white arrows in Fig. 4A in a TiO₂ (wt%) map. A temperature map of the metamorphic conditions in the chlorite matrix surrounding the titanite indicates conditions of between 420 °C and 150 °C, with a mean value around 350 °C (Fig. 4 B–D). A high proportion of pixels recording conditions above the mean of T = 350 °C are plotted in the lower half of the image (Fig. 4 B–D). In contrast, lower temperature matrix conditions (below T = 350 °C) occur surrounding the titanite microfilaments (Fig. 4C).

Fig. 4.

Quantitative microscale maps of candidate titanite biosignatures and surrounding metamorphic conditions in Archean metabasaltic pillow lavas from the BGB. (A) Map of TiO₂ (wt%) showing the image of titanite filaments of interest (arrows) highlighted in the dashed-line box. (B) Calculated metamorphic conditions in the chlorite matrix using a chlorite thermodynamic modeling approach (SI Materials and Methods). The dashed-line box highlights a typical area with well-developed titanite microfilaments surrounded by chlorite pixels recording relatively low temperature conditions of less than T = 350 °C (pixel group 2). (C and D) Image pixel grouping above and below the mean value of T = 350 °C, indicating a higher proportion of pixels recording conditions below T = 350 °C surrounding the titanite microfilaments (dashed-line box in C). In D the y axis variable Al(iv) = atoms per formula unit of Al⁴⁺ in tetrahedral coordination modeled in the chlorite crystal structure.

Discussion

The new titanite ages reported here are ca. 650 My younger than the eruptive age of the Hooggenoeg pillow lavas (3.472–3.432 Ga) (21), rejecting the antiquity of the titanite microtextures. The previous bioalteration model (1–7) argues for seafloor alteration that was penecontemporaneous with pillow lava extrusion, and was based on a reported overlap between metamorphic Ar–Ar ages in komatiites (22), and U–Pb ages (23, 24) from a metagabbro in the Komati Formation at ca. 3.488 Ga. However, these age constraints are derived from the Komati Formation, which is much older and unrelated to the eruption of the overlying Hooggenoeg Formation pillow lavas that hosts the titanite microtextures. Thus, there are currently no direct age constraints on early subseafloor alteration preserved in the Hooggenoeg Formation pillow lavas. Rather, we highlight that the late Archean U–Pb ages of ca. 2.8–2.9 Ga reported here for the hornfelsic titanite and the candidate bioalteration textures, overlaps with the age of local, late-stage mafic dykes and diabase intrusions in the southwestern BGB (Fig. S1 and refs. 12, 13, 25, and 26). Such mafic diabase intrusions, like the one intercepted by drill core KD2b (Fig. S1), have been dated at ca. 2.96 Ga (25, 26). The U–Pb age for titanite growth determined here is markedly different from that previously reported in a multicollector LA-ICP-MS study of ref. 4, which estimated titanite growth at 3.34 ± 0.068 Ga. This apparently precise titanite age does not however, correspond to any known regional or local metamorphic event in the BGB and falls midway between estimated seafloor alteration at ca. 3.488 Ga (22) and regional tectono-thermal metamorphism at ca. 3.23 Ga (25). The previous titanite age of ca. 3.34 Ga was estimated using the common Pb composition of the French Bob mine galena in the common lead correction (4). In the current study, we used two independent analytical techniques (single- and multicollector LA-ICP-MS) and two U–Pb dating methods (Tera–Wasserburg and Wetherill Concordia) for interpreting the data. The single-collector LA-ICP-MS technique required no assumptions on common lead and plots measured isotope ratios directly on the Tera–Wasserburg diagram. In the multicollector LA-ICP-MS technique, we used an appropriate common Pb correction and the least radiogenic Old Star Mine galena composition, considered to give the best estimate of common Pb composition on the Archean Kaapvaal Craton at the time of titanite growth (18). Both analytical techniques used here, yield robust age estimates of titanite growth that were in agreement at ca. 2.9 Ga and correspond to a known local thermal event in the pillow lava sequence. As such, the new U–Pb titanite ages presented here do not support an early ca. 3.488 Ga hydrothermal subseafloor bioalteration model. More likely, we argue that the titanite microtextures are abiogenic metamorphic porphyroblasts, which formed as a result of local thermal contact metamorphism of the Hooggenoeg pillow lava country rock.

The new in situ metamorphic information indicates a textural connection between the titanite microfilaments and low-temperature metamorphic microdomains. Variable metamorphic temperature conditions (T = 420 °C to 150 °C) are recorded on a microscale in the chlorite matrix (Fig. 4A–D). However, a high proportion of the low-temperature pixels recording conditions of less than T = 350 °C are concentrated around the well-developed titanite filaments. This suggests that the titanite microfilaments most likely grew progressively under retrograde metamorphic conditions. We argue that the metamorphic data supports an abiogenic origin for titanite growth as a result of thermal contact metamorphism, and that the titanite microtextures record variable retrograde cooling conditions and high fluid-rock ratios.

The metamorphic model for the origin of the titanite filaments presented herein is in direct contrast to the complex bioalteration model presented previously (Fig. S3 and refs. 1–7). The bioalteration model infers early Archean microbial etching to have taken place in the glassy pillow lava rims (refs. 1–7; Fig. S3A). A poorly described process of titanite infilling is then argued to preserve the hollow microbial tubes (refs. 1–7; Fig. S3B). In many cases, an inferred “window for bioalteration” can be as long as 150–200 My and this time difference between volcanic eruption and titanite infilling is argued to allow development of the titanite microtextures (e.g., refs. 4 and 27; Fig. S3B). We point out that although microbial etching has been hypothesized to explain tubular microtunnels in recent basaltic glass, this process has not been confirmed in either natural settings or in laboratory experiments (16, 28). In the current study we propose an alternative, and more likely, abiotic model for the origin of the titanite microtextures (Fig. S3B). In this model the titanite and associated greenschist assemblage forms due to a combination of metamorphic net transfer reactions and exchange hydration reactions during decreasing metamorphic temperature (retrograde cooling) and alteration of the pillow basalt. Fresh basaltic glass and anhydrous phenocrysts become unstable resulting in the appearance of actinolite + chlorite + epidote + quartz + titanite during thermal contact metamorphism. The titanite filaments develop during microscale retrograde cooling with the continuous growth of titanite on preexisting filament nucleation sites depending on the degree of microscale fluid–rock interaction and the decreasing temperature gradient. A detailed model diagram with reaction history and titanite porphyroblast development is provided in Fig. S3, depicting this abiotic mechanism of titanite growth.

The size, shape, and distribution of the Archean titanite microtextures do not compare well with microtunnels from the in situ oceanic crust or argued biogenic origin. In particular, the wide range in diameters of the titanite filaments (Fig. 1I) is not compatible with the restricted size distribution expected for a biogenic population (29). Likewise, the morphological continuum of titanite microtextures reported here (Fig. 1; Fig. S2), spanning titanite spheres and bands with a few short projections to more well-developed clusters with an increasing number and length of filaments, does not reflect the more restricted shape variation expected from a biogenic population (29). Rather, this morphological spectrum of increasing complexity recorded by the titanite microtextures is more compatible with an abiotic growth continuum. The distribution of the titanite microtextures is also not compatible with an early seafloor origin, as they are not uniquely rooted in early fractures and rather, they can be isolated from sites of early fluid infiltration, also they radiate at highly variable angles from the root zones contrary to observation from seafloor basalts (7, 9). As such, the titanite microtextures fail several criteria concerning their size, shape and distribution required for microbial trace fossils (14, 15), therefore questioning the biogenicity of the Archean microtextures.

Recent high-sensitivity nanoscale secondary ion microprobe mapping has found that previously reported carbon and nitrogen linings (1, 3) to the Barberton microtextures are absent (30). This lack of carbonaceous linings earlier interpreted as decayed organic remains, thereby removes a key geochemical line of evidence in support of a biogenic origin for the microtextures. The abiotic process of ambient inclusion trails (AITs) has been explored for microtextures in metavolcanic glass (3, 14, 15), and this involves tubular microtextures formed by crystal migration under elevated fluid pressures. In this instance however, AITs can be excluded given that the microtextures do not show polygonal cross-sections and longitudinal striations, which are characteristics of AITs (15). Also crucially, the sulfide crystals sometimes associated with the microtextures (30) do not occur as terminal crystals to the filaments, and do not have sizes and shapes that match the dimensions of the microtextures.

In summary, we illustrate that the candidate titanite biotextures fail an antiquity test and are not syngenetic with an early subseafloor bioalteration model, nor do they satisfy size and morphological criteria for biogenicity. Rather, they are late Archean, abiogenic metamorphic microtextures that formed 870–470 My after eruption and seafloor alteration of the Hooggenoeg pillow lavas. We argue that a reassessment of titanite microtextures in all variably overprinted greenstone belt pillow lavas (e.g., refs. 2, 5, and 6) is required to exclude such abiotic thermal processes before a case for microbial bioalteration can be made. We therefore contend that the oldest bona fide trace fossil now reverts to ca. 1.7 Ga microborings in silicified stromatolites from China (31).

Numerous lines of textural and geochemical evidence have supported the concept of microbial life in Archean sedimentary environments. In the BGB these include carbonaceous remains found in silicified platform facies (e.g., ref. 32), shallow marine siliclastics (e.g., refs. 33 and 34), and sulfur isotope evidence from tectono-sedimentary environments (e.g., ref. 35). In contrast, much further work is now required to confirm if microbial life existed in the Archean mafic to ultramafic subseafloor environment. We conclude that filamentous titanite is not a reliable biosignature for investigating potential subseafloor life, and that alternatives, such as sulfur isotope fractionations recorded by basalt-hosted sulfides, could be more promising (11, 30). It may yet be possible to identify robust evidence of life in mafic–ultramafic habitats on the early Earth and beyond. Seeking alternative biosignatures, combined with quantitative microscale mapping of low-temperature alteration and redox conditions (20), may help to constrain potential habitats for early subsurface microbial life and conditions for preservation of their biosignatures.

Materials and Methods

Our samples come from drill core KD2b of the BSDP that intercepted pillow lavas of the Hooggenoeg Formation, along the Komatii river in the BGB, South Africa. A geological map and lithological log of the drill core can be found in SI Materials and Methods. In situ U–Pb titanite dating by LA was conducted using both a single-collector and a multicollector ICP-MS. These included a Thermo Finnigan Element 2 single-collector sector field ICP-MS combined with a 193-ArF excimer laser (Resonetics RESOlution; M50-LR) and a Thermo Finnigan Neptune multicollector ICP-MS coupled to a Resonetics RESOlution M50 laser at the Department of Earth Science, University of Bergen. Full description on the analytical methods and Tables S1–S3 containing measured U–Pb isotope ratios by single- and multicollector ICP-MS are provided in SI Materials and Methods. The optical images shown here were obtained using a Nikon LV100Pol polarizing microscope and photographed using a DS-Fi1 color camera with 5.24-megapixel resolution combined with NIS-Elements BR 2.30 software. This software includes a microscale measuring tool that was used in the titanite morphological analysis. A Cameca SX100 electron microprobe with five wavelength-dispersive spectrometers at the Department of Geosciences, University of Oslo (Oslo, Norway), was used to acquire point analyses and X-ray compositional maps. The X-ray maps were treated using the new software XMapTools to derive quantitative compositional and metamorphic chlorite temperature maps (SI Materials and Methods).