Earth and Planetary Science Letters 299 (2010) 290–298

Contents lists available at ScienceDirect

Earth and Planetary Science Letters

j ourna l homepage: www.e lsev ie r.com/ locate /eps l

In-situ dating of the Earth's oldest trace fossil at 3.34 Ga

D. Fliegel a,⁎, J. Kosler a, N. McLoughlin a, A. Simonetti b, M.J. de Wit c, R. Wirth d, H. Furnes a

a Department of Earth Science & Center for Geobiology, University of Bergen, N-5007 Bergen, Norwayb Department of Civil Engineering & Geological Sciences, University of Notre Dame, Notre Dame, IN 46556, USAc AEON and Department of Geological Sciences, University of Cape Town, Rondebosch 7701, South Africad Helmholtz Centre Potsdam, GFZ German Research Centre for Geosciences, Potsdam, D 14473 Germany

⁎ Corresponding author.E-mail address: d.j.fliegel@gmail.com (D. Fliegel).

0012-821X/$ – see front matter © 2010 Elsevier B.V. Adoi:10.1016/j.epsl.2010.09.008

a b s t r a c t

a r t i c l e i n f o

Article history:Received 11 May 2010Received in revised form 3 September 2010Accepted 7 September 2010Available online 6 October 2010

Editor: M.L. Delaney

Keywords:trace fossilsPaleoArchean bioalterationBarberton Greenstone BeltU–Pb titanite datingearly lifesub-oceanic biosphere

Microbial activity in volcanic glass within the oceanic crust can produce micron sized pits and tunnels. Suchbiogenic textures have been described from the recent oceanic crust and mineralized equivalents in pillowlavas as old as 3.47–3.45 Ga from the Barberton Greenstone Belt (BGB) of South Africa. In meta-volcanicglasses these microbial traces are preserved by titanite mineralization (CaTiSiO5) and on the basis ofmorphological, textural and geochemical evidence have been argued to represent Earth's oldest trace fossils.Here we report the results of in-situ U–Pb dating of titanite that infills trace fossils from the HooggenoegComplex of the BGB using laser ablation MC-ICP-MS (multi-collector inductively coupled, plasma massspectrometry). This yields a titanite age of 3.342±0.068 Ga demonstrating the antiquity of the BGB tracefossils. This radiometric age confirms that a sub-seafloor biosphere was already established in thePaleoArchean and that it likely represented an important habitat for the emergence and evolution of earlymicrobial life on the Earth.

ll rights reserved.

© 2010 Elsevier B.V. All rights reserved.

1. Introduction

Sea floor pillow basalts contain tubular and granular alterationtextures in their glassy chilled margins (e.g. Fisk et al., 1998; Furneset al., 2008; McLoughlin et al., 2008; Staudigel et al., 2008; Thorsethet al., 1995). The biogenicity of such alteration textures has beenargued for on the basis of several lines of evidence including: themorphology of the alteration textures (e.g. McLoughlin et al., 2009;Staudigel et al., 2008 and refs therein); the δ13C values of disseminatecarbonates found in pillow rims and cores (e.g. Furnes et al., 2001a);microbiological staining and sequencing studies (e.g. Santelli et al.,2008 and refs therein); and chemical imaging of trace amounts oforganic material lining the textures (e.g. Furnes and Muehlenbachs,2003). The body of evidence used to advance a biogenic origin forthese glass alteration textures is extensively reviewed elsewhere (e.g.Furnes et al., 2001b; Staudigel et al., 2008). It is hypothesized thatmicroorganisms living in the sub-seafloor colonize fractures andvesicles in the glass in order to obtain electron donors, nutrients,shelter and perhaps to escape predation (Cockell and Herrera, 2008;Furnes et al., 2008). In Phanerozoic volcanic glass the granularbioalteration textures are by far the most abundant and the tubulartextures are less common (Fig. 34 in Furnes et al., 2008) but exhibit

striking morphological variations including annulated, twisted, spiralshape and branched forms (McLoughlin et al., 2009). A conceptualmodel of how such microbial communities are thought to etch thevolcanic glass has been developed in a series of papers (e.g. Furneset al., 2008; Staudigel et al., 2008; Thorseth et al., 1992; Thorseth et al.,1995). These bioalteration textures can also be regarded as tracefossils because they record the activities and behaviours of micro-organisms and recently an ichnotaxonomy has been proposed toelucidate themorphological diversity of these traces and enable globaland stratigraphic comparisons (McLoughlin et al., 2009).

The fossil record of life in volcanic glass extends far beyond theoldest in-situ oceanic crust. The preservation of bioalterationtextures in glass now transformed to metamorphic mineralassemblages is made possible by mineralization of the micro-textures on the sub-seafloor prior to deformation and metamor-phism. Titanite is a common mineralizing phase and micro-textureswith comparable morphologies to those found in the in-situ oceaniccrust have now been reported from Phanerozoic ophiolites andArchean greenstone belts (reviewed by Furnes et al., 2007). Thesequence of events is: microbial alteration, authigenic mineraliza-tion and subsequent metamorphism; this is summarized in Fig. 7 ofMcLoughlin et al. (2010b). The mineralization process appears toobscure some of the fine morphological detail seen in recentbioalteration textures, such as spirals and annulations, and onaverage results in filaments with larger diameters than that of theoriginal hollow tubes (Fig. 2 and Furnes et al., 2007). But in all other

291D. Fliegel et al. / Earth and Planetary Science Letters 299 (2010) 290–298

respects the mineralized tubes are comparable in morphology anddistribution to those found in fresh volcanic glass. In contrast,granular bioalteration textures, the most abundant bioalterationtexture in glassy pillow lava rims of the in-situ oceanic crust, that aretypically b0.5 μm in diameter are lacking from meta-volcanic glass,probably because these are obliterated by recrystallisation of thehost glass during metamorphism.

Trace fossils found in meta-volcanic glass of the BarbertonGreenstone Belt (BGB) in South Africa are the focus of this study(Fig. 1). Their geological setting and key textural characteristics aresummarized below. Briefly, they are filamentous micro-textures thatare b15 μm in diameter and extend up to 200 μm from a “root zone”that represents the initial crack from which the microbial alterationcommenced (Banerjee et al., 2006; Furnes et al., 2004). Based ontextural and geochemical evidence, these tubular titanite structures(Fig. 2) have been interpreted to represent the oldest physical tracesof life on Earth (Furnes et al., 2004). The goal of this study was toobtain an in-situ U–Pb date for the titanite that infills the BGB tracefossils in order to constrain the timing of their mineralization and toobtain a minimum age for the sub-seafloor microbial activity thatproduced these trace fossils.

2. Geological setting and sample locality

The PaleoArchean BGB of South Africa contains some of theworld'soldest and best-preserved pillow lavas (de Wit et al., 1987). Thetectonostratigraphy of the Onverwacht Suite comprises the Thee-spruit, Komati, Hooggenoeg, Noisy, Kromberg and Mendon Com-plexes that total 15 km in thickness (Fig. 1 and de Wit et al. in press).These complexes are separated by major shear zones or unconfor-mities described in detail in de Wit et al (in press). The extrusivecomponents include pillow lavas, minor hyaloclastite breccias and

Fig. 1. (A) Map of South-Africa showing the location of the Barberton Greenstone Belt (BGB).the Komati river in upper Hooggenoeg Complex (star). The eruptive age of the sample localitand 3.455 Ga (see text for explanation).

sheet flows with several interlayered chert horizons that are betweenca. 0.5 to 20 m thick and together this sequence is overlain by thesedimentary rocks of the Fig Tree and Moodies Groups (Fig. 1). Pillowlavas from the Hooggenoeg Complex show no or little deformation(Fig. 2). The Onverwacht Suite has been interpreted to representfragments of Archean oceanic crust, termed the Jamestown OphioliteComplex (de Wit et al., 1987), that developed in association withsubduction and island arc activity approximately 3.55 to 3.22 Ga (deRonde and de Wit, 1994). The magmatic sequence of the OnverwachtSuite is exceptionally well-preserved, relatively undeformed awayfrom its margins with the surrounding granitoids and has experiencedgreenschist facies or lower degrees of metamorphism (e.g., Groschet al., 2009a; Tice et al., 2004). This relatively well preserved windowonto the early Archean Earth has thus provided the focus of manyeffort to seek traces of life in both the sediments (e.g. Javaux et al.,2010; Tice and Lowe, 2004) and volcanics (e.g. Furnes et al., 2004;Grosch et al., 2009b).

The ~2700 m thick lava pile of the Hooggenoeg Complexcomprises several eruptive pulses described in Furnes et al (inpress). The age of the Hooggenoeg Complex is bracketed by two U–Pbzircon ages of ~3.47 and ~3.45 Ga the locations of which are shown onFig. 1 (modified from Furnes et al., in press). A maximum age foreruption of the Hooggenoeg Complex is provided by detrital zirconsfrom the basal Middle Marker chert that is exposed along thesoutheastern limb of the Onverwacht Fold where it has been dated at3.472±0.005 Ga (Armstrong et al., 1990). Above the HooggenoegComplex and within the Noisy Complex appears a large daciticintrusion in the eastern part of the northern limb of the OnverwachtFold which has yielded a concordant U/Pb SHRIMP date of 3.445±0.004 Ga (de Wit et al., 1987). The overlying Noisy Complex asrecently defined by Furnes et al (in press) has also yielded severalminimum age constraints upon eruption of the Hooggenoeg lavas

(B) Sketch map of the BGB. (C) Geological map of BGB indicating the sample location ony is bracketed between two U–Pb zircon ages (locations indicated by solid dots) of 3.472

Fig. 2. (A) Komati river with the sampling location of the pillow lavas. Pillow lavas in the photograph are well preserved and little deformed. Petrographic thin-sections of titanitemineralized trace fossils, in a greenschist facies meta-volcanic glass from Barberton Greenstone Belt South Africa. (B) Clusters of titanite mineralized filaments radiating from central“root zones” in a inter pillow hyaloclastite (IPH) sample now composed of chlorite (green) and quartz (white). (C) Combined transmitted-reflected light image showing a laserablation track in the “root zone” of a titanite mineralized trace fossil (dotted line). (D) and (E) Examples of trace fossils showing a central titanite mineralized “root zone” (dashedline) with a medial fracture along which fluids once circulated, and from which titanite mineralized filaments radiate and are segmented (arrows) by the growth of cross-cuttingchlorite. Sample number 29-BG-03. (F) Size distribution of tubular textures in the Barberton pillow lavas, redrawn from Furnes et al. (2007).

292 D. Fliegel et al. / Earth and Planetary Science Letters 299 (2010) 290–298

including, igneous zircons from a flow banded porphyritic lava nearthe top of the volcaniclastic sequence that gave a concordant SHRIMPdate of 3.451±0.005 Ga (de Vries et al., 2006). In addition, along theKomati River downstream from our sample site, zircons from a daciticcrystal-tuff near the top of the Noisy Complex (location shown onFig. 1) yielded a magmatic U–Pb zircon age of 3455.2±7.5 Ma (Bigginet al., submitted).

The Hooggenoeg samples investigated here were collected from anon-vesicular and variolitic pillow lava outcrop on the banks of theKomati River in the south eastern limb of the Onverwacht anticline(Fig. 1). This trace fossil bearing horizon was also the target of ascientific drill hole KD2b described in Grosch et al., (2009b). Thepillow lavas are very little deformed and display well-developedchilled margins ~10 mm thick of formerly glassy material that gradeinwards into a variolitic zone, consisting of a mixture of altered glassand microcrystalline material (Fig. 2a and d of Banerjee et al., 2006).The pillow rims now comprise a greenschist facies assemblage ofchlorite, quartz, epidote, titanite and calcite. Pockets of interpillowhyaloclastites are well developed at this locality and this outcrop alsoprovided thematerial illustrated in Figs. 1 and 3 of Furnes et al. (2004)and is referred to as locality 6 in Banerjee et al. (2006).

3. Description of the BGB trace fossils

The BGB trace fossils occur in pillow rims and interpillowhyaloclastites. They are mineralized by micro-crystalline titanite andappear dark brown in thin-section. They occur as dense clusters andbands containing a central “root zone” from which they radiate(Fig. 2). Sometimes a paler-coloured band, a few microns wide can beseen at the center of the “root zone”, recording the former location of afracture in the volcanic glass where the bioalteration initiated. Themineralized tubular structures are curvi-linear and may taper slightlytowards their ends, occurring in clusters of up to tens of filaments.They have an average width of 4 μm and are up to 200 μm in lengthwith an average length of 50 μm(Fig. 7 in Furnes et al., 2007). They arecommonly segmented by cross-cutting chlorite grains (Fig. 2 D and E).

4. Archean trace fossil— a summary of the evidence for biogenicityand antiquity

There are three reported occurrences of mineralized microbialtraces from Archean pillow lavas. The first and also the oldestexamples yet described come from the BGB of South Africa (Furnes

293D. Fliegel et al. / Earth and Planetary Science Letters 299 (2010) 290–298

et al., 2004; Banerjee et al., 2006; Furnes et al., 2007). Morphologicallycomparable textures, have also been described from the 3.35–3.31 Gapillow lavas of the Euro Basalt in Western Australia (Banerjee et al.,2007) and 2.52 Ga pillow lavas of the Wutai Group of North WestChina (McLoughlin et al., 2010b). The evidence for the biogenicity ofthese trace fossils has been extensively explained elsewhere (seeBanerjee et al., 2006; Furnes et al., 2004; Furnes et al., 2008) andrather here, we summarise the major features that support thebiogenicity and especially antiquity of the Barberton trace fossils.

A biogenic origin for the titanite mineralized textures from theBarberton has been advanced on the basis of 3 lines of evidencesummarized below:

a. The BGB titanite mineralized tubular structures have comparablemorphologies, size and distribution to trace fossils found in recentvolcanic glass

The micro-textures are restricted to pillow rim and interpillowbreccia samples and extend away from healed fractures along whichseawater once flowed (Fig. 2). These tubular structures are similar insize, shape, and distribution to features documented in glassy pillowrims from ophiolites such as the Troodos ophiolite and in-situ oceaniccrust (Furnes et al., 2008). Thus the BGB structures are interpreted asbeing the mineralized remains of microbial trace fossils.

b. The δ13C values of carbonate in the BGB pillows rims records microbialfractionation

Disseminated carbonate from bulk rock samples of pillow rimsfrom the Onverwacht Suite have δ13C values of +3.9 to−16.4 per mil(‰), which are different from those of the crystalline interiors of+0.7to −6.9‰. Amygdules containing later, secondary carbonate haveδ13C values that cluster around zero. The δ13C values from crystallineinterior samples are bracketed between those of primary mantle CO2

(−5 to−7‰) and of Archean marine carbonate (0‰), whereas theglassy samples extend to lower δ13C values. Such isotopic contrastsare also seen in pillow lava rims from ophiolites and oceanic crust,where the generally low δ13C values of disseminated carbonate areattributed tometabolic byproducts formed duringmicrobial oxidationof dissolved organic matter in the pore waters (Fig. 2 in Furnes et al.,2008). The isotopically low δ13C values of carbonate in the formerlyglassy rims of the BGB pillows are thus interpreted to have alsoformed by microbial fractionation (Banerjee et al., 2006; Furnes et al.,2004).

c. X-ray element mapping of carbon on the walls of the BGB tubularstructures

Furnes et al. (2004) and Banerjee et al. (2006) report X-rayelemental maps of thin, b1 μm thick carbon linings on the walls ofsome of the BGB micro-textures. This carbon does not correlate withthe elemental maps of calcium, iron, and magnesium indicating thatthe carbon is not bound in carbonate (e.g. Fig 7 in Banerjee et al., 2006,analyticalmethods explained in Banerjee andMuehlenbachs 2003). Inrecent volcanic glass, biofilms and organic remains containing nucleicacids are commonly observed along the interior surface of microbiallygenerated pits and tunnels (e.g. Furnes et al., 1999). Thus the BGBcarbon has been interpreted to represent decayed organic materialleft on the interior surfaces of microbially generated trace fossils thatwas subsequently preserved during later titanite mineralization(Furnes et al. 2004).

No plausible abiotic mechanism has yet been advanced to explainthe BGB micro-textures. Abiotic processes that have been tested andrejected include: purely chemical dissolution, and the migration ofambient inclusions. These abiotic mechanisms are discussed in detailin Banerjee et al. (2006); Staudigel et al. (2008); and most recently in

McLoughlin et al. (2010a) and cannot explain the morphology,distribution and geochemical characteristics of the BGB trace fossils.

The antiquity of the Barberton trace fossils has been argued forfrom textural observations. The micro-tunnels are segmented bymetamorphic chlorite (e.g. Fig. 2), implying that the microbial traceswere mineralized by titanite before growth of the cross-cuttingmetamorphic chlorite. A low-grade, sea-floor metamorphic event isargued to have occurred shortly after extrusion of the Hooggenoegpillow lava succession (Brandl and deWit, 1997), andwas followed byvery low grade dynamothermal metamorphism between 3.23 and3.29 Ga (Schoene et al., 2008). The last regional deformation in thevicinity of the BGB occurred between 3.1 and 3.2 Ga, and the rockshave not been thermally disturbed since. Maximum temperature forthis regional metamorphism have been estimated to be between 200–400 °C on the basis of a geothermometer involving graphitecrystallinity measured by laser-Raman spectroscopy on carbonaceouscherts (Tice et al., 2004).

Direct dating of Archean trace fossils in volcanic glass was firstundertaken on samples from the Pilbara Craton of Western Australia(Banerjee et al., 2007). This study reported a U–Pb age of 2.9±0.1 Gafor titanite mineralized trace fossils that is ~400 Ma younger than theaccepted eruptive age for the host pillow lavas of 3.35 Ga given by aninterbedded tuff (Nelson, 2005). The titanite date corresponds to theage of a regional metamorphic event related to the last phase ofdeformation andwidespread granite intrusion in the Northern Pilbaraat 2.93 Ga (van Kranendonk et al., 2002). This U–Pb titanite agetherefore represents a minimum estimate for the timing of titaniteformation, especially given that metamorphic chlorite cross-cuts thetitanite tubules and so must post-date the titanite mineralization. Thisimplies ab400 Ma post eruptive period during which the bioaltera-tion textures from the Pilbara were mineralized; but does not excludethe possibility that the textures formed soon after eruption, remainedhollow and were mineralized somewhat later. Our goal with theBarberton samples is thus to investigate the duration of this Archeanbioalterationwindow and to see if we couldmore tightly constrain theantiquity of another example of Archean trace fossils.

5. Analytical techniques

A multi-collector ICP-MS (Thermo Finnigan Neptune) coupled toa 213 nm Nd:YAG laser (New Wave Research UP-213) was used tomeasure Pb/U and Pb isotopic ratios in titanite at Bergen University.The sample introduction system was modified to enable simulta-neous nebulisation of a tracer solution (detector cross calibrationand mass bias correction) and laser ablation of the solid sample.Natural Tl (205Tl/203Tl=2.3871; Dunstan et al., 1980), 209Bi andenriched 237 Np (N99%) were used in the tracer solution which wasaspirated to the plasma using a 50 μl/min PFA nebuliser, dualcyclonic–double pass quartz glass spray chamber and a T-piece tubeattached to the back end of the plasma torch. Helium was used asablation/carrier gas. The aerosol was mixed prior to the torch via aT-piece.

The laser was set up to produce energy density of 3 J/cm2 at arepetition rate of 10 Hz. The laser spot size varied between 8 to 60 μm indiameter according to the sample volume available for laser ablation.The laser beam was focused on the sample that was mounted in ca.3 cm3 ablation cell. The “root zones” from 11 different trace fossil foundin a polished thin section, sample numbers 101-BGB-08 were analyzed.During ablation the stagewasmoved beneath the stationary laser beamat a speed of 10 μm/s. Typical signal intensities measured for the 204Pb,206Pb–207Pb and 238Uwere ca. 2000, 30000 and 20000 cps, respectively.The acquisition consistedof a 25 smeasurementof all analytes in the gasblank and aspirated tracer solution, followed bymeasurement of U andPb isotope signals from titanite, along with the continuous signal fromthe tracer solution for another 25 seconds. The data were acquired instatic mode in a mixed channeltron (IC)–Faraday (FAR) detector array

294 D. Fliegel et al. / Earth and Planetary Science Letters 299 (2010) 290–298

295D. Fliegel et al. / Earth and Planetary Science Letters 299 (2010) 290–298

for isotopes 202Hg (IC1), 203Tl (IC2), 204Hg+Pb (IC3), 205Tl (IC4), 206Pb(IC5), 207Pb (IC6), 208Pb (IC7), 209Bi (FAR L4), 237 Np (FAR H4) and 238U(IC8). Solution signal of 203Tl was used to cross-calibrate thechanneltrons before each analysis.

The data were corrected for channeltron dead time and processedoff line in a spreadsheet-based program (LamTool; Kosler et al., 2008)and plotted on concordia diagrams using IsoplotEx (Ludwig, 2003).Data reduction included correction for channeltron yield, gas blankandmass discrimination using the solution signals of Tl, Bi and Np andrepeat analyses of the Khan titanite (Heaman, 2009). Correction forcommon Pb was based on measurement of the non-radiogenic 204Pband utilized the Pb isotope composition of Archean galena from theFrench Bob's mine in the Barberton Greenstone Belt; it is one of thegalena samples used to define the two-stage model of lead isotopicevolution (Stacey and Kramers, 1975). Titanite sample from theLillebukt Alkaline Complex in northern Norway (0.520±0.005 Ga;Pedersen et al., 1989) was periodically measured for data qualitycontrol and it gave a pooled concordia age of 0.524±0.024 Ga (6measurements, 2 sigma).

Energy dispersive x-ray (SEM-EDX) images were acquired with aZeiss Supra 55VP scanning electron microscope equipped with aThermo Noran EDS detector at the University of Bergen. The EDXspectra were collected at 15 mm working distance using 15 keVacceleration voltage and 60 μm aperture size. The most abundantfluorescence for each element line was used for imaging.

Focused ion beam milling (FIB) and transmission electronmicroscopy were done at the GeoForschungsZentrum in Potsdam(GFZ) using a FEI FIB2000 TEM and a FEI TecnaiG2 F20 X-TWINinstruments. The field emission gun of the TEM was operated at200 keV accelerating voltage. Total of four different FIB foils wereobtained from different parts of sample number 29-BGB-03 andanalyzed by TEM. Datasets from all FIB foils were found to be similar.Further details of the FIB sample preparation and analysis aredescribed in Wirth (2004).

6. Results

Scanning and transmission electron microscope investigations ofthe BGB sample, together with previous textural and mineralogicalobservations confirm the similarities between the titanite that infillthe tubular structures and the titanite in the “root zones” (Figs. 2 and3). Elemental distribution maps obtained by SEM-based EDX analysisindicate that there is no appreciable difference in compositionbetween the tubular structures and their respective “root zones”(Fig. 3). In addition, the TEM data shows no variation at the micron orsub-micron scale in porosity, size and shape of the crystals betweenthe polycrystalline titanite in the micro-textures and that in the “rootzones” (Fig. 3G and H). Moreover, the TEM images reveal thehomogeneity of the phase composition of the trace fossils, suggestingthat mineralization of the tubular structures and the “root zones”occurred simultaneously or within a very short time span. This meansthat a U–Pb age obtained from the central “root zones” will berepresentative of the whole trace fossil. Unfortunately, our LAMC ICP-MS requires a 40 μm spot size which is too large to date an individualtube. But as explained above, this is not a concern becausepetrographic arguments suggest that mineralization of the “rootzone” and tubes was synchronous.

Fig. 3. Scanning electron microscope images of the trace fossils taken in backscatter mode (transmission electronmicroscope, high-angle annular darkfield images (TEMHAADF)of focusein the elemental composition or mineralogy of the “root zones” and individual tubes. (A) Segmcentral “root zone”. (B) Titanium elemental map. (C) Silica elemental map. (D)Magnesium elemtubular trace fossil that intersects the surface of the platinum coated sample. The tubes are filledin orientation of the individual titanite grains. The semi-circular cross-section of the tube is clea“root zone”. No difference in the size, shape or porosity of the titanite crystals can be seen by com(H). Variation in grey level of (G) and (H) is attributed to thedifferent thickness of both foils, whhole in the amorphous carbon foil used on the TEM grid). Images A–F are from sample 29-BGB

Titanite U–Pb isotopic data obtained by LA MC ICP-MS for tracefossils are given in Table 1 and plotted on a concordia diagram inFig. 4. The pooled concordia age for 11 trace fossil “root zones” is3.342±0.068 Ga. The mean isotopic composition of the studiedtitanites plots slightly above the concordia curve. We interpret thisas resulting from a small inaccuracy in the common lead correctionthat was applied to the raw isotopic data. Accurate common Pbcorrection is crucial for obtaining correct U–Pb titanite ages butproblems in determining the initial lead isotopic composition forPrecambrian lavas in general, and for rocks from the BGB in particular,are well known and have been discussed in length by Brevart et al.(1986). Briefly, the Pb isotopic composition of sulfides (mostlygalena) from the BGB shows a considerable spread in the 206/204Pb,207/208Pb and 208/204Pb isotopic ratios (Brevart et al., 1986; Saager andKoppel, 1976). In addition, none of the published galena Pb isotopiccompositions can produce entirely concordant titanite ages and thegalena samples come from a different part of the BGB stratigraphythan the material studied herein. Therefore we propose that forcommon Pb correction of the studied titanite samples, it is moreappropriate to use a general model for common Pb evolution, forexample, Stacey and Kramers (1975) or part of it. We therefore choseto use the common lead composition of a galena from the French Bob'smine in the Barberton Onverwacht Suite that defines the Archean partof the 2-stage lead evolution model of Stacey and Kramers (1975).This is considered by us to be the most appropriate for the commonlead correction of the titanite infilling the studied trace fossils. Itshould also be noted that choice of a different initial Pb compositionfor correction of titanite data presented in this study affects theconcordance of the pooled concordia ages but has no significant effecton the reported age of the trace fossils.

7. Discussion

The 3.342±0.068 Ga titanite agemeasured here records the age ofmetamorphic titanite growth and equilibration within these tracefossils. This demonstrates that the trace fossils are Paleoarchean in ageand provides a minimum age estimate for their formation. In otherwords, this titanite age underestimates the timing of microbialalteration of the BGB volcanic glasses but nonetheless, confirmstheir antiquity. The Hooggenoeg lavas have never experiencedmetamorphic temperature that exceed the closure temperature ofthe U–Pb isotopic system in titanite (Tc=550–650 °C), thus this agerepresents the main phase of titanite growth within the textures (c.f.Schoene and Bowring, 2007). Furthermore, petrographic, SEM andFIB-TEM observations confirm that mineralization was synchronousin all parts of the trace fossils i.e. in both the “root zones” and tubularcavities (Figs. 2 and 3).

The difference in age between the infilling titanite (3.342±0.068 Ga) and the eruptive age of the Hooggenoeg Complex (3.47–3.45 Ga) is attributed to either: a long-lived bioalteration windowduring which microbial alteration persisted; or a time gap after whichmicrobial alteration ceased and before the titanite mineralizationcommenced. Both of these possibilities are supported by analogy tothemodern oceanic crust and ophiolites. An early onset to bioalterationon the Archean seafloor is supported by comparison to modern oceanicspreading centres, where microbes are observed to rapidly colonizejuvenile glassy surfaces (Furnes et al., 2007). These bioalteration traces

A) and elemental distribution maps obtained using energy dispersive X-rays (B–F). Alsod ionbeam(FIB) foils cut fromthe trace fossils (G)–(H). Together these shownodifferencesented, filamentous trace fossils that intersect the surface of the sample and radiate from aental map. (E) Calcium elemental map. (F) Iron elemental map. (G) FIB foil from a single,with polycrystalline, porous titanite and the variation in grey contrast is due to differencesrly seen and the interface with the chlorite matrix. (H) FIB-foil from a titanite mineralizedparison to the polycrystalline titanite in the individual trace fossil tubes (G) or “root zones”ich is also reflected in theHAADF image (note the faint dark circle visible inH is due to the a-03 and G–H from sample 101-BGB-08.

Fig. 4. U–Pb concordia diagram showing the LA-MC-ICP-MS titanite data measured on“root zones” of microbial trace fossils from the Hooggenoeg Complex of the BarbertonGreenstone Belt, South Africa. The data-point ellipses represent one sigma uncertainty;the pooled concordia age (grey ellipse) is 3.342±0.068 Ga and is shown at the 2 sigmalevel.

Table1

Laserab

lation

MC-ICP-MSU–Pb

data

fortitanite

from

11“roo

tzo

nes”

oftracefossils

from

sample10

1-BG

B-08

oftheBa

rbertonGreen

ston

eBe

lt.

Ana

lysis

Isotop

icratios

Calculated

ages

Ma

206Pb

/204Pb

1)

207Pb

/235U

±1σ

206Pb

/238U

±1σ

207Pb

/206Pb

±1σ

Rho2)

207Pb

/235U

±1σ

206Pb

/238U

±1σ

207Pb

/206Pb

±1σ

#1

19.342

30.563

84.37

060.71

610.03

860.30

960.01

450.38

3505

501

3481

188

3518

72#2

20.921

23.176

93.16

210.66

010.03

350.25

460.01

010.37

3234

441

3268

166

3214

63#3

19.157

23.829

33.15

440.67

030.03

360.25

780.01

050.38

3261

432

3307

166

3233

64#4

20.000

24.219

34.91

020.72

020.04

460.24

390.01

350.31

3277

664

3497

216

3145

88#5

18.248

29.092

24.46

870.74

970.04

200.28

150.01

520.36

3456

531

3606

202

3371

84#6

20.040

24.310

03.26

470.73

300.03

480.24

050.00

960.35

3281

441

3544

168

3123

64#7

21.097

21.244

52.73

280.67

500.03

190.22

830.00

870.37

3150

405

3325

157

3040

61#8

20.040

30.213

45.84

470.74

220.04

890.29

520.02

190.34

3494

676

3579

236

3445

115

#9

18.519

22.543

13.41

790.70

860.03

520.23

070.01

000.33

3207

486

3453

172

3057

69#10

19.268

23.223

12.95

190.68

190.03

330.24

700.00

920.38

3236

411

3351

164

3165

59#11

18.692

24.164

13.22

210.69

960.03

500.25

050.01

020.38

3275

437

3419

171

3188

65

Isotop

icratios

correctedforinitialc

ommon

Pbus

ingco

mpo

sition

ofga

lena

from

theFren

chBo

b'sMine,

Onw

erwacht

Group

(Staceyan

dKramers,19

75).

296 D. Fliegel et al. / Earth and Planetary Science Letters 299 (2010) 290–298

can remain hollow or becomepartially infilledwith authigenicmineralsprincipally clays and iron-oxyhydroxides over time and these may beenriched in titanium, a possible precursor to titanite (e.g. Fig. 12b inStaudigel et al., 2008). Theoretically, as long as there is still fresh glasspresent and seawater circulation continues, then the microbialbioalteration may go on for millions of years. The major phase oftitanite mineralization is argued to occur much later during greenschistfacies metamorphism, the timing of which is determined by theregionally geological history. Bioalteration textures can remain hollowfor long periods of time, as for example, micro-textures from the littledeformed 0.092 Ga Troodos ophiolite (Furnes et al., 2001c) and the0.122 Ga Ontong Java Plateau (Banerjee and Muehlenbachs, 2003) arelargelyunmineralized.A summaryof this sequenceof events is shown inFig. 7 of McLoughlin et al. (2010b).

The nature and timing of metamorphism of the Hooggenoegvolcanic glasses is only sparsely characterized. The 3.342±0.068 Gametamorphic titanite age reported here is comparable to an ageobtained by step-heating 40Ar/39Ar analysis of 3.326±0.005 Ga fromkomatitic basalts of the underlying Komati Complex (Fig. 5 in López-Martínez et al., 1992). Alternatively, the titanite age obtained heremay correspond to heating associated with the intrusion of gabbroicmagma into the Onverwacht Suite that is recorded by a U/Pbbaddelleyite age of 3.351±0.006 Ga obtained from a meta-gabbroin the underlying Komati Complex (Kamo and Davis, 1994).Subsequent to the metamorphic event discussed here the Hooggen-oeg Complex experienced low grade dynamothermal metamorphismduring a major deformation event between 3.23 and 3.29 (de Rondeand de Wit, 1994; Schoene et al., 2008). Localised deformation alongthemargins of the BGB occurred between 3.1 and 3.2 Ga and the rockshave not been thermally disturbed since. Whether the new U–Pbtitanite age reported here represents a sub-seafloor burial metamor-phism event, a regional low-grade metamorphic overprint, or acontact metamorphic effect arising from gabbroic intrusives, it doesnot change the fact that the trace fossils are Paleoarchean in age.Moreover, this corroborates and provides an absolute age constrainton previous textural observations of metamorphic chlorite cross-cutting the titanite infilled trace fossils (Fig. 4).

This is only the second study to use direct in-situ dating techniquesto obtain an absolute age for a candidate biosignature from thePaleoarchean. Other traces of Archean life such as organic

Fig. 5. Schematic overview of processes leading to mineralization of tubular alteration textures by titanite. After eruption of the lava pile at 3.47–3.45 Ga, the tubular textures wereetched into the host glass. The alteration progressed until mineralization by titanite ceased at 3.34±0.07 Ga. The titanite mineralization in the non chloritized cavities was mostlikely related to exclusion of titanium from the chlorite during metamorphism in the presence of calcium and silicon. The two photographs show well preserved pillows fromHooggenoeg Complex (left) and tubular textures mineralized by titanite.

297D. Fliegel et al. / Earth and Planetary Science Letters 299 (2010) 290–298

microfossils, stromatolites and geochemical signatures are notpossible to date directly. The age obtained here is approximately 0.4billion years older than the previously published age of 2.9±0.1 Gafrom comparable trace fossils in the 3.35 Ga Euro Basalt of the PilbaraCraton in Western Australia (Banerjee et al., 2007). Furthermore, thetime gap between eruption of the host lavas and the mineralization ofthe textures, or the co called “window for bioalteration”, is alsosignificantly smaller, being b0.13 Ga in the case of the Barberton tracefossils in contrast to ~0.44 Ga for the Pilbara trace fossils (Banerjeeet al., 2007) and ~0.71 Ga for the Wutai bioalteration textures(McLoughlin et al., 2010b). This time window is controlled by theregional geological history and is much shorter in the case of theBarberton trace fossils, because the greenschist facies metamorphicevent responsible for their mineralization and preservation occurredsooner after eruption. A schematic overview of the process is shownon Fig. 5. It is in these types of settings where the titanite growthwithin the trace fossils was relatively early that this U–Pb datingtechnique is most powerful in constraining the antiquity of the tracefossils.

8. Conclusion

We report a minimum age of 3.342±0.068 Ga for titanitemineralization of trace fossils from pillow lavas of the BarbertonGreenstone belt. These trace fossils therefore represent the oldestdirectly dated microbial trace of life on Earth and confirm previoustextural arguments for the antiquity of these trace fossils (Furneset al., 2004). It follows that the sub-seafloor biosphere was alreadyestablished in the Paleoarchean and likely provided an importantcradle for the emergence and evolution of microbial life on the Earth.This study further illustrates the importance of U–Pb dating toassessing the antiquity and syngenicity of microbial traces in volcanicglass, as may one day be applied to seeking traces of life in olderterrestrial pillow lava sequences.

Acknowledgments

Funding for this study was received from the Norwegian ResearchCouncil. A. Schreiber is thanked for preparing the FIB foils, E.G. Grosch

and H. Staudigel for constructive discussions. The authors wish tothank J. Ellingsen for help with the artwork in Fig. 1 and L. M. Heamanfor providing the Khan titanite standard. This is AEON contributionnumber 85.

References

Armstrong, R.A., Compston, W., deWit, M.J., Williams, I.S., 1990. The stratigraphy of the3.5–3.2 Ga Barberton Greenstone Belt revisited: a single zircon ion microprobestudy. Earth Planet. Sci. Lett. 101, 90–106.

Banerjee, N.R., Muehlenbachs, K., 2003. Tuff life: bioalteration in volcaniclastic rocksfrom the Ontong Java Plateau. Gechem. Geophys. Geosyst. 4 (4). doi:10.1029/2002GC000470.

Banerjee, N.R., Furnes, H., Muehlenbachs, K., Staudigel, H., de Wit, M., 2006.Preservation of similar to 3.4–3.5 Ga microbial biomarkers in pillow lavas andhyaloclastites from the Barberton Greenstone Belt, South Africa. Earth Planet. Sci.Lett. 241, 707–722.

Banerjee, N.R., Simonetti, A., Furnes, H., Muehlenbachs, K., Staudigel, H., Heaman, L., VanKranendonk, M.J., 2007. Direct dating of Archean microbial ichnofossils. Geology35, 487–490.

Biggin, A., de Wit, M.J., Langerijs, C., Zegers, T.E., Voûte, S., Dekkers, M.J. and Drost, K. inreview. Paleomagnetism of Paleoarchean rocks of the Onverwach Suite, BarbertonGreenstone Belt (Southern Africa): potential geomagnetic reversals and evidencefor Vaalbara Supercraton at c. 3.5 Ga.. Earth Planet Sci. Lett., submitted.

Brandl, G., deWit, M.J., 1997. The Kaapvaal Craton, South Africa. In: deWit, M.J., Ashwal,L. (Eds.), Greenstone Belts. Oxford University Press, Oxford, pp. 581–607.

Brevart, O., Dupre, B., Allegre, C.J., 1986. Lead-lead age of Komatiitic lavas andlimitations on the structure and evolution of the Precambrian mantle. Earth Planet.Sci. Lett. 77, 293–302.

Cockell, C.S., Herrera, A., 2008. Why are some microorganisms boring? TrendsMicrobiol. 16, 101–106.

de Ronde, C.E.J., de Wit, M.J., 1994. Tectonic history of the Barberton Greenstone BeltSouth Africa: 490 million years of Archean crustal evolution. Tectonics 13,983–1005.

de Vries, S.T., Nijman, W., Armstrong, R.A., 2006. Growth-fault structure and strati-graphic architecture of the Buck Ridge volcano-sedimentary complex, upperHooggenoeg Formation, Barberton Greenstone Belt, South Africa. Precambrian Res.149, 77–98.

de Wit, M.J., Furnes, H., Robins B. (in press). Geology and Tectonostratigraphy of theOnverwacht Suite, Barberton Greenstone Belt, South Africa. Precambrian Res.

de Wit, M.J., Hart, R.A., Hart, R., 1987. The Jamestown Ophiolite Complex, Barbertonmountain belt: a section through 3.5 Ga oceanic crust. J. Afr. Earth Sci. 6 (no. 5),681–730.

Dunstan, L.P., Gramlich, J.W., Barnes, I.L., Purdy, W.C., 1980. Absolute isotopicabundance and the atomic weight of a reference sample of thallium. J. Res. NatlBur. Stand. 85, 1–10.

Fisk, M.R., Giovannoni, S.J., Thorseth, I.H., 1998. The extent of microbial life in thevolcanic crust of the ocean basins. Science 281, 978–979.

298 D. Fliegel et al. / Earth and Planetary Science Letters 299 (2010) 290–298

Furnes, H., de Wit, M.J, Robins, B., Sandstå, N.R., (in press) Volcanic evolution of theUpper Onverwacht Suite, Barberton Greenstone Belt, South Africa. PrecambrianRes.

Furnes, H., Muehlenbachs, K., 2003. Bioalteration Recorded in Ophiolitic Pillow Lavas.In: Robinson, P.T. (Ed.), Dilek, Y. Special Publications, London, Ophiolites in Earth'sHistory. J Geol Soc London, pp. 415–426.

Furnes, H., Muehlenbachs, K., Tumyr, O., Torsvik, T., Thorseth, I.H., 1999. Depth of activebio-alteration in the ocean crust: Costa Rica Rift (Hole 504B). Terra Nova 11,228–233.

Furnes, H., Muehlenbachs, K., Torsvik, T., Thorseth, I.H., Tumyr, O., 2001a. Microbialfractionation of carbon isotopes in altered basaltic glass from the Atlantic Ocean,Lau Basin and Costa Rica Rift. Chem. Geol. 173, 313–330.

Furnes, H., Staudigel, H., Thorseth, I.H., Torsvik, T., Muehlenbachs, K., Tumyr, O., 2001b.Bioalteration of basaltic glass in the oceanic crust. Geochem. Geophys. Geosyst. 2(8). doi:10.1029/2000GC000150.

Furnes, H., Muehlenbachs, K., Tumyr, O., Torsvik, T., Xenophontos, C., 2001c. Biogenicalteration of volcanic glass from the Troodos ophiolite, Cyprus. J. Geol. Soc. Lond.158, 75–84.

Furnes, H., Banerjee, N.R., Muehlenbachs, K., Staudigel, H., de Wit, M., 2004. Early liferecorded in archean pillow lavas. Science 304, 578–581.

Furnes, H., Banerjee, N.R., Staudigel, H., Muehlenbachs, K., McLoughlin, N., de Wit, M.,Van Kranendonk, M., 2007. Comparing petrographic signatures of bioalteration inrecent to Mesoarchean pillow lavas: tracing subsurface life in oceanic igneousrocks. Precambrian Res. 158, 156–176.

Furnes, H., McLoughlin, N., Muehlenbachs, K., Banerjee, N., Staudigel, H., Dilek, Y., deWit, M., Van Kranendonk, M., Schiffman, P., 2008. Oceanic Pillow Lavas andHyaloclastites as Habitats for Microbial Life Through Time—A Review. In: Dilek, Y.,Furnes, H., Muehlenbachs, K. (Eds.), Links Between Geological Processes, MicrobialActivities and Evolution of Life. Springer, pp. 1–68.

Grosch, E.G., de Wit, M.J., Furnes, H., McLoughlin, N., 2009a. Metamorphic constraintsacross a mafic-ultramafic tectonite zone between the Mendon and Krombergcomplexes, Barberton greenstone belt, South Africa. Geol. Soc. Am. v.41 (No.7), 529Abstract with Programs.

Grosch, E.G., McLoughlin, N., deWit, M., Furnes, H., 2009b. Drilling for the Archean rootsof life and tectonic Earth in the Barberton Mountains. Sci. Drilling 8, 24–28.

Heaman, L.M., 2009. The application of U–Pb geochronology to mafic, ultramaficand alkaline rocks: an evaluation of three mineral standards. Chem. Geol. 261,42–51.

Javaux, E.J., Marshall, C.P., Bekker, A., 2010. Organic-walled microfossils in 3.2-billion-year-old shallow-marine siliclastic deposits. Nature 463, 934–938.

Kamo, S.L., Davis, D.W., 1994. Reassessment of Archean crustal development in theBarberton Mountain Land, South-Africa, based on U–Pb dating. Tectonics 13,167–192.

Kosler, J., Forst, L., Slama, J., 2008. Lamdate and Lamtool: Spreadsheet-Based DataReduction for Laser Ablation ICP-MS. In: Sylvester, P. (Ed.), Laser ablation ICP-MS inthe Earth sciences: current practices and outstanding issues. Mineral. Assoc. Can.short course series, pp. 315–317.

López-Martínez, M., York, D., Hanes, J.A., 1992. A Ar-40/Ar-39 geochronological study ofKomatiites and Komatiitic basalts from the Lower Onverwacht Volcanics —

Barberton Mountain Land, South-Africa. Precambrian Res. 57, 91–119.Ludwig, K., 2003. User's Manual for Isoplot 3.00, A Geochronological Toolkit for

Microsoft Excel. 4.

McLoughlin, N., Furnes, H., Banerjee, N., Staudigel, H., Muehlenbachs, K., deWit, M., VanKranendonk, M., 2008. Micro-Bioerosion in Volcanic Glass: Extending theIchnofossil Record to Archean Basaltic Crust. In “Current Developments inBioersion. In: Wisshak, M., Tapinla, L. (Eds.), Current Developments in Bioerosion.Erlangen Earth Conference, Springer, Berlin, pp. 372–396.

McLoughlin, N., Furnes, H., Banerjee, N.R., Muehlenbachs, K., Staudigel, H., 2009.Ichnotaxonomy of microbial trace fossils in volcanic glass. J. Geol. Soc. Lond. 166,159–169.

McLoughlin, N., Staudigel, H., Furnes, H., Eickmann, B., Ivarsson, M., 2010a. Mechanismsof microtunneling in rock substrates — distinguishing endolithic biosignaturesfrom abiotic microtunnels. Geobiol 8, 245–255.

McLoughlin, N., Fliegel, D.J., Furnes, H., Staudigel, H., Simonetti, A., Zhao, G.C., Robinson,P.T., 2010b. Assessing the biogenicity and syngenicity of candidate bioalterationtextures in pillow lavas of the 2.52 GaWutai Greenstone Terrane of China. Chin. Sci.Bull. 55, 188–199.

Nelson, D.R., 2005. Sample GSWA 178042, in Compilation of geochronology data, June2005 update: Western Australia Geological Survey, CD-ROM.

Pedersen, R.B., Dunning, G.R., Robins, B., 1989. U–Pb ages of Nepheline SyenitePegmatites from the Seiland Magmatic Province. In: Gayer, R.A. (Ed.), TheCaledonide Geology of Scandinavia. Graham & Trotman, London, pp. 3–8.

Saager, R., Koppel, V., 1976. Lead isotopes and trace-elements from sulfides of ArcheanGreenstone Belts in South-Africa — contribution to knowledge of oldest knownmineralizations. Econ. Geol. 71, 44–57.

Santelli, C.M., Orcutt, B.N., Banning, E., Bach, W., Moyer, C.L., Sogin, M.L., Staudigel, H.,Edwards, K.J., 2008. Abundance and diversity of microbial life in ocean crust. Nature453, 653–657.

Schoene, B., Bowring, S.A., 2007. Determining accurate temperature-time paths fromU–Pb thermochronology: an example from the Kaapvaal craton, southern Africa.Geochim. Cosmochim. Acta 71, 165–185.

Schoene, B., de Wit, M.J., Bowring, S.A., 2008. Mesoarchean assembly and stabilizationof the eastern Kaapvaal craton: a structural-thermochronological perspective.Tectonics 27, TC5010. doi:10.1029/2008TC002267 2008.

Stacey, J.S., Kramers, J.D., 1975. Approximation of terrestrial lead isotope evolution by a2-stage model. Earth Planet. Sci. Lett. 26, 207–221.

Staudigel, H., Furnes, H., McLoughlin, N., Banerjee, N., Connell, L.B., Templeton, A., 2008.3.5 billion years of glass bioalteration: volcanic rocks as a basis for microbial life.Earth Sci. Rev. 89, 156–176.

Thorseth, I.H., Furnes, H., Heldal, M., 1992. The importance of microbiological activity inthe alteration of natural basaltic glass. Geochim. Cosmochim. Acta 56, 845–850.

Thorseth, I.H., Furnes, H., Tumyr, O., 1995. Textural and chemical effects of bacterial-activity on basaltic glass — an experimental approach. Chem. Geol. 119, 139–160.

Tice, M.M., Lowe, D.R., 2004. Photosynthetic microbial mats in the 3416-Myr-old ocean.Nature 431, 549–552.

Tice, M.M., Bostick, B.C., Lowe, D.R., 2004. Thermal history of the 3.5–3.2 Ga Onverwachtand Fig Tree Groups, Barberton greenstone belt, South Africa, inferred by Ramanmicrospectroscopy of carbonaceous material. Geology 32, 37–40.

Van Kranendonk, M.J., Hickman, A.H., Smithies, R.H., Nelson, D.R., Pike, G., 2002.Geology and tectonic evolution of the Archean North Pilbara terrain, Pilbara Craton,Western Australia. Econ. Geol. Bull. Soc. Econ. Geol. v. 97, 695–732.

Wirth, R., 2004. Focused Ion Beam (FIB): a novel technology for advanced application ofmicro- and nanoanalysis in geosciences and applied mineralogy. Eur. J. Mineralog.16, 863–876.