Geobiology. 2017;1–19. wileyonlinelibrary.com/journal/gbi  | 1© 2017 John Wiley & Sons Ltd

Received:21July2016  |  Accepted:21June2017DOI: 10.1111/gbi.12248

O R I G I N A L A R T I C L E

Oncoidal granular iron formation in the Mesoarchaean Pongola Supergroup, southern Africa: Textural and geochemical evidence for biological activity during iron deposition

A. J. B. Smith1,2,3 | N. J. Beukes1,2,3 | J. Gutzmer1,4,5 | A. D. Czaja6,7,8 |  C. M. Johnson7,8 | N. Nhleko1,9

1PaleoproterozoicMineralizationResearchGroup,DepartmentofGeology,UniversityofJohannesburg,Johannesburg,SouthAfrica2DepartmentofScienceandTechnology–NationalResearchFoundationCentreofExcellenceforIntegratedMineralandEnergyResourceAnalysis,UniversityofJohannesburg,Johannesburg,SouthAfrica3DepartmentofGeology,UniversityofJohannesburg,Johannesburg,SouthAfrica4HelmholtzZentrumDresden-Rossendorf,HelmholtzInstituteFreibergforResourceTechnology,Freiberg,Germany5DepartmentofMineralogy,TUBergakademieFreiberg,Freiberg,Germany6DepartmentofGeology,UniversityofCincinnati,Cincinnati,OH,USA7DepartmentofGeoscience,UniversityofWisconsin,Madison,WI,USA8NASAAstrobiologyInstitute,UniversityofWisconsin,Madison,WI,USA9GeologicalSurveyandMinesDepartment,Mbabane,Swaziland

CorrespondenceA.J.B.Smith,PaleoproterozoicMineralizationResearchGroup,DepartmentofGeology,UniversityofJohannesburg,Johannesburg,SouthAfrica.Email:bertuss@uj.ac.za

Funding informationDepartmentofGeologyattheUniversityofJohannesburg;PalaeoproterozoicMineralizationResearchGroup(PPM);DepartmentofScienceandTechnology(DST);NationalResearchFoundation(NRF);CentreofExcellenceforIntegratedMineralandEnergyResourceAnalysis(CIMERA);NASAAstrobiologyInstitute

AbstractWedocumentthediscoveryofthefirstgranularironformation(GIF)ofArchaeanageand present textural and geochemical results that suggest these formed throughmicrobialironoxidation.TheGIFoccursintheNcongaFormationoftheca.3.0–2.8GaPongolaSupergroupinSouthAfricaandSwaziland.Itisinterbeddedwithoxideandsilicatefaciesmicriticironformation(MIF).ThereisastrongtexturalcontrolonironmineralizationintheGIFnotobservedintheassociatedMIF.TheGIFismarkedbyoncoidswithchertcoressurroundedbymagnetiteandcalciterims.Theserimsshowlaminated domal textures, similar in appearance to microstromatolites. The GIF isenrichedinsilicaanddepletedinFerelativetotheinterbeddedMIF.VerylowAlandtrace element contents in the GIF indicate that chemically precipitated chert wasreworked above wave base into granules in an environment devoid of siliciclasticinput.Microbiallymediated ironprecipitation resulted in the formationof irregular,domalrimsaroundthechertgranules.Duringstormsurges,oncoidsweretransportedanddeposited indeeperwaterenvironments.Texturalfeatures,alongwithpositiveδ56Fevaluesinmagnetite,suggestthatironprecipitationoccurredthroughincompleteoxidationofhydrothermalFe2+byiron-oxidizingbacteria.TheinitialFe3+-oxyhydroxideprecipitateswerethenpost-depositionallytransformedtomagnetite.Comparisonofthe Fe isotope compositions of the oncoidal GIF with those reported for theinterbeddeddeeperwater ironformation(IF) illustratesthattheFe2+pathwaysandsources for theseunitsweredistinct. It is suggested that thedeeperwater IFwasdepositedfromtheevolvedmarginofabuoyantFe2+

aq-richhydrothermalplumedistaltoitssource.Incontrast,oncoliticmagnetiterimsofchertgranulesweresourcedfromambientFe2+

aq-depletedshallowoceanwaterbeyondtheplume.

2  |     SMITH eT al.

1  | INTRODUCTION

TheimportanceofbiologicalactivityinthedepositionofPrecambrianironformations(IFs)hasbeenthesubjectofprotracteddebate(e.g.,Bekker etal., 2010; Beukes & Gutzmer, 2008; Kappler, Pasquero,Konhauser,&Newman, 2005;Klein, 2005;Konhauser etal., 2002).CommonlythreemodelsareproposedforbiologicalactivityrelevanttoironoxidationandprecipitationinproducingIFs:indirectoxidationviafreeO2producedbyoxygen-producingphotosyntheticmicrobes(e.g.,Cloud,1973;Klein&Beukes,1993);directoxidationviachem-olithoautotrophiciron-oxidizingbacteria(e.g.,Konhauseretal.,2002);or direct oxidation via anoxygenic photoautotrophic iron-oxidizingbacteria(e.g.,Kappleretal.,2005).OxidationofFe(II)mayalsooccurusingotherelectronacceptorssuchasnitrate(e.g.,Kappler,Johnson,Crosby,Beard,&Newman,2010).Despite thesepossibilities,all ar-gumentsinfavourofbiologicalactivityduringIFgenesissufferfromthegenerallackofdirect(i.e.,textural)evidenceforbiologicalactivity.SuchtexturalevidenceisprovidedbytheoccurrenceofoncoidsintheSokomanandAnimikiebasinIFs(Knoll&Simonson,1981;Planavskyetal., 2009) and microfossils and stromatolites in the Gunflint andBiwabikIFs(Barghoorn&Tyler,1965;Planavskyetal.,2009),allap-proximately1.9Gainage,ontheSuperiorCraton.

Oncoidsarea subtypeof coatedgrainsandaredefinedasbio-genically encrusted grains (Peryt, 1983a,b). Except for the olderexamplesnotedabove,modernexamplesofoncoidshavebeen re-cordedintheshallowmarineenvironment(e.g.,Alshuaibi,Duane,&Mahmoud,2012;Nguetchoua&Giresse,2010).Theterm“oncoidal”or “globoidal” is also applied byHofmann (2000) to a type of lam-inashapeforstromatolitesthatoccuraroundgrains,withtheauthoralsoincludingoncoidsintheclassificationofArchaeanstromatolites.TheoccurrenceofoncoidscontainingironmineralsinitsbiogenicallyformedcoatingsmaythusbetakenasevidenceforthesignificanceofmicrobialprocessesforthedepositionofIFs inancientsedimentarysuccessions.

Older coated grains in ironstones have been reported in thePalaeoproterozoicTimeballHillFormationoftheTransvaalSupergroupofSouthAfrica(Dorland,1999).However,itisimportanttonotethatthesegrainsareconcentricandlackirregularrimsandcanthereforebeclassifiedasooidssimilartowhatiscommonlyseeninPhanerozoicironstones (e.g., Garzanti, Haas, & Jadoul, 1989; Gehring, 1989;Kearsley, 1989),which do not necessarily require biological activityfor its formation (Richter,1983). In addition, suchoolitic ironstonesforminproximalmarinetodeltaicenvironmentswhichisincontrasttomoredistalmarinesettingsinwhichmostmicriticironformations(MIFs)formedduringthePalaeoproterozoic.

This study documents the discovery of a granular iron formation(GIF)fromtheMesoarchaeanMozaanGroupofthePongolaSupergroupofsouthernAfrica,whichwasfirstreportedbrieflybySmith,Beukes,Gutzmer,Johnson,andCzaja(2012)attheGoldschmidtConferenceinMontreal,Canada,duringJune2012.Theirregulartopologyofthegran-ules’coatings issimilartodomalmicrostromatolitic texturesandsug-gestsabiogenicorigin,definingthegranulesasoncoids.Thissuggeststhatmicrobialactivityplayedaroleduringirondepositioninashallow

marine environment that took place penecontemporaneously to thedepositionofMIFatgreaterwaterdepth.Acomparisonoftheironiso-topesoftheGIFandassociateddeeperwaterIFisalsopresentedandindicatesdistinctsourcesandpathwaysforFe2+intheshallowanddeepwaterdepositionalenvironmentsintheMozaanGroup.

2  | GEOLOGICAL AND STRATIGRAPHIC SETTING

TheMesoarchaeanPongola Supergroupon theKaapvaalCratonofsouthernAfrica(Figure1a)isoneoftheoldestwell-preservedlaterallyextensivesupracratonicsuccessionsintheworld(Beukes&Cairncross,1991;Wronkiewicz&Condie,1989).Itconsistsoftheessentiallyvol-caniclower(~3.0–2.97Ga)NsuzeGroupandthedominantlysiliciclas-ticupper(~2.96–2.84Ga)MozaanGroup(Gutzmer,Beukes,Pickard,&Barley, 1999;Hegner,Kröner,&Hunt, 1994;Mukasa,Wilson,&Young,2013;Wilson,Groenewald,&Palmer,2013).TheGIFthatisthefocusofthiscontributionishostedbytheNcongaFormationoftheOdwaleniSubgroupoftheMozaanGroup(Figure1b).Itiscloselyassociatedwithmassiveandbanded(fortexturalclassificationdetailsseeBeukes&Gutzmer,2008)MIF(Figure1c).

TheNcongaFormation isa laterallyextensiveargillaceousunitofsome100mthickintheMozaanGroupwedgedbetweenquartziteandshaleof theDelfkomFormationbelowandmafic lavaof theTobolskFormation above (Figure1a). It is mainly composed of siltstone andshale,cappedbyathinsharp-basedquartzite,andcontainstwodistinctmagneticmudstoneunits,onetowardsthebaseandanothernearthetop.TheseunitsaresimilarincharactertoseveralotherspresentintheMozaanGroupandthecorrelativeWitwatersrandSupergroupthat insomecasescontaininterbedsofMIF(Beukes&Cairncross,1991;Smith,2007;Smith,Beukes,&Gutzmer,2013).TheuppermagneticmudstoneoftheNcongaformation,however,differsfromallothersinthesensethatithoststheratheruniqueGIFdescribedinthiscontribution.

CurrentlytheGIFunitisknownfromtwolocalities.Mostdetailsandsamplesforthiscontributionwereobtainedfromanexplorationdrill core (PNG-4), drilled by AngloGold-Ashanti 20–25km northnortheastofNongomaintheKwaZulu-NatalProvinceofSouthAfricainwhichthemagneticmudstonewithtwointerbedsofGIFwasinter-sectedatdepthofaround1,295m(Figure1a,c).ThesecondlocalityisfromoutcroponaridgeintheKubuta-MooihoekareaofSwaziland(Nhleko, 2003), located approximately 90kmnorth northwest fromthelocationofdrillcorePNG-4(Figures1aand2).ThestratigraphicpositionoftheGIFintheoutcropisexactlysimilartothatinthedrillcore indicatingthat it isdevelopedoveranareaofat leasthundredormore km2. However, in anotherwell known outcrop area of theNcongaFormationintheHartlandarea(Figure1a),theGIFisnotde-veloped,although thehostmagneticmudstoneunit ispresent.Thissuggests that theGIFwasonlydeposited ina rathersmall localizedareaoftheoriginalWitwatersrand-Mozaandepository(Figure1a).

IntheKubuta-Mooihoekarea,theGIFoutcropsastwocontinuous,well-definedbedsseparatedbyathinmudstonelayeroveranexposedstrikelengthofsome300malongaridge(Figure2a).TheGIFbedsand

     |  3SMITH eT al.

mudstoneinterlayerhaveanaveragecombinedthicknessof0.8–1.0m.TopsurfacesofboththeGIFbedsdisplayhummockywaveripplemarks.TheseareespeciallywelldevelopedalongthetopsurfaceofthelowerGIFbed(Figure2a,b).ThelowerGIFbedalsocontainsgoodexamplesofcombinedhummockyandlinearwaveripplemarks(Figure2c).TheGIFbedstypicallydisplaypoorlydefinedflatbeddingbutsomecross-bedding is present (Figure2d).The topof theupperGIFbed showsreworkedgranulesurfaces inplacesthataresharplyoverlainbyMIFwithabundantangular flatmudshardsorchips inplaces (Figure2e).AlongstriketheMIFalsocontaininterlaminatedgranulelenticlescom-posedofgranulessimilartothose inthe immediatelyunderlyingGIF(Figure2f).TheGIFcontainsmud laminae,which in someplacesarehighlysilicifiedandinothersFe-rich.Inplacesthemudlaminaehaveabrokenupappearancewithrip-upsoccurringabovethem(Figure2g).TheGIFbedsoccurasstackedupwardfininggradedgranulelayers(i.e.,gradedbeds;Figures2hand3a–c).The layersthatare intheorderof5–10cmthickdisplaypoorlydefined flat towavybeddingwithsomelowanglediscontinuitiesthatcouldrepresentpartofhummockycross-stratification(Figure3a,c).Thebasesoftheupwardfininggranulelayersaremarkedbyloadcastsandthetopbymuddrapes(Figures2hand3a).Subangularchertintraclastsoccurthroughout(Figures2hand3a).

In the studieddrill core (Figure1c), theNcongaFormation con-sistsof40mofupwardcoarseningshale-siltstonefollowedby50mof upward fining siltstone-shale that becomes slightlymagnetic to-wards the top before it is overlain with sharp gradational contactbya20-cm-thicksilicatefaciesfelutite (terminologyafterBeukes&Gutzmer,2008).Thefelutiteisoverlainwithsharpcontactby50cmofGIF.TheGIFunit comprises three stacked gradedbedsof gran-ularmagnetite in chert (Figure3b).A brown,massive silicate facies

IF, also termed silicate facies felutite, sharply overlies theGIF.Thefelutite,whichis2.6mthick, isagainoverlainwithsharpbasalcon-tactby50cmofthreestackedbedsofGIF(Figure3c).LensesofGIF,representingstarvedripples,occurinthefelutiteimmediatelybelowtheGIF(Figure1c).TheGIFisoverlainby1mofmixedfaciesmicro-banded IF, also termedmixed facies ferhythmite (terminology afterBeukes&Gutzmer,2008)thatcontainsgreenchertbands.Thefer-hythmitegradesupwardsintobrownfelutite(typicalmagneticmud-stone reportedbySmithetal.,2013) that isoverlainbyaquartzitewithasharperosivecontact.

ThedepositionalageoftheNcongaIFsisconstrainedbythedepo-sitionofthePongolaSupergroupthathasbeenplacedatamaximumof2,985±1Ma(U-Pbzirconage)bythefelsicvolcanicrocksoftheNsuzeGroup (Hegner etal., 1994) and aminimumof 2,954±9Ma (U-Pb zircon age) by theTobolskvolcanics in the upper part of theMozaanGroup(Mukasaetal.,2013).

3  | METHODOLOGY

Five samples of IF and two samples of GIF (Table1) were takenfrom the drill core (PNG-4) for this study, as the drill core is fairlypristine and has only been affected by lower greenschist faciesregionalmetamorphismwhichwould have had aminimal effect onthe depositional and diagenetic minerals, which are dominated bynon-hydrousphases(i.e.,chert,oxidesandcarbonates)thathavenotbeenrecrystallized.Surfacesampleswerenotusedforthisstudy,astheyhavebeenaffectedbysupergenealteration,where,forexample,oxidationofallmagnetitetohaematitemayoccur.

F IGURE  1  (a)LocationofthePongolaandWitwatersrandbasinsofsouthernAfricawiththelocationofthestudiedborehole.(b)SimplifiedstratigraphyoftheMozaanGroup.(c)StratigraphicsectionoftheNcongaFormationfromboreholePNG4usedinthisstudywithsamplelocationsindicated

4  |     SMITH eT al.

3.1 | Petrography and mineralogy

X-raypowderdiffraction (XRD)wasconductedonmilledpowdersofthe studied units to identify major minerals present. Measurementswere carried out using a Panalytical X’Pert diffractometer with anX’CeleratordetectorhousedatSpectrum,thecentralanalyticalfacilityoftheScienceFacultyattheUniversityofJohannesburg.AnalysesofthemeasuredXRDpatternswerecarriedoutonX’PertHighScorePlussoftware.FurtherdiffractometersettingsarereportedinSmith(2007).ThiswascomplementedbypetrographicstudiesonsevenpolishedthinsectionsofGIFandMIFusinglightandscanningelectronmicroscopy(SEM).SEMstudieswereconductedonasetofcarbon-coatedpolished

thin sections with a Jeol 5600 SEM equipped with a Noran Si(Li)energydispersiveX-rayspectrometry(EDS)detector,withanultra-thinberylliumwindow,atSpectrum.Thesampleswereexaminedunderhighvacuumwitha15kV,15mAelectronbeambymeansofsecondary(SEI)andbackscatteredelectron(BSE)imaging.Mineralswereidentifiedbysemi-quantitative EDS-spot analyses. Additional BSE imaging andEDSanalyseswereperformedonpolishedblockspreparedfromboththe lower and upper GIF beds in the drill core at the University ofWisconsin,MadisonusingaHitachiS-3400variablepressurescanningelectronmicroscopeequippedwithaSi(Li)EDSdetector.Analyseswereperformedathighvacuumandwith15kVacceleratingvoltage.

F IGURE  2 FieldphotographsoftheNcongaFormationGIFandMIFoutcropintheKubuta-MooihoekareaofSwaziland.(a)ThegeneraloutcropillustratingtwosteeplydippingGIFbedsseparatedbyanFe-richmudbeddrapingthehummockytopsurfaceofthelowerGIFbed.(b)CloseupphotographofthehummockywaveripplemarksonthelowerGIFbed(planview).(c)ExampleofahummockyandlinearwaverippledsurfaceinthelowerGIFbed(planview).(d)Cross-beddingintheGIF.(e)ReworkedgranulesurfaceattopcontactofupperGIFbedwithoverlyingMIF.MIFcontainsreworkedmudshards.(f)BeddingparallelgranulelenticlesinMIFoverlyingtheupperGIFbed.(g)Silicifiedmudrip-upsaboveapartiallybrokenupsilicifiedmudlaminainGIF.(h)SedimentaryfeaturesintheGIFwhichincludegradedbedding,chertintraclasts,mudintraclastsandmuddrapes

     |  5SMITH eT al.

3.2 | Geochemistry

AsetofsevensamplesoftheIFandGIFwasanalysedformajorele-mentcomposition(Table1)usingX-rayfluorescencespectrometryatSpectrumusingfusionbeads.Inaddition,asetoffoursamplesoftheIFswassenttoACMElaboratoriesinVancouver,Canadaforrareearthelement(REE)analysisbyICP-MS(Table1).ForthesamplesanalysedatACME laboratories, concentrations above10 times the detectionlimithaveaprecisionof15%.Analyticalqualitywascheckedusingcer-tifiedstandardreferencematerials.TwosamplesofGIFwereanalysedforREEattheGeochemistryLaboratoryatJacobsUniversity,Bremenusing ICP-MS (Table1).Alldataproducedat JacobsUniversityhavebeen interference-corrected (e.g., BaO on Eu) and analytical qualitywas frequentlycheckedbyanalysing internalandexternal referencematerialssuchasIFreferencestandardsIF-G(IsuaBIF,Greenland;e.g.,Viehmann, Bau, Hoffmann, &Münker, 2015) and FeR-3 and FeR-4(TemagamiBIF,Canada;e.g.,Bau&Alexander,2009).

Calcite-bearingwhole rock samples of the GIF and ferhythmitewereanalysed for their carbonandoxygen stable isotopecomposi-tion(Table1)attheStableIsotopeLaboratory intheDepartmentofEarthSciences,UniversityofCapeTown.Analyseswerecarriedoutbyselectivelydissolvingthecarbonatemineralsinphosphoricacidat50°C,extractingtheproducedcarbondioxidegasandanalysingitonaDeltaXPmass spectrometer indual inletmode.The internal stan-dardusedforthecarbonatemeasurementwastheNM(NamaqualandMarble)standard.Measuredvalueswerecorrectedbyapplyingfrac-tionationvaluesforcalcite(α=1.009;Al-Aasm,Taylor,&South,1990).

TheFeisotopecompositionsofonespecimenofeachofthethreedistinct lithologies (felutite, ferhythmite and GIF) were measured.Multiple ~3mm by 3mm regionswithin each specimenwere sam-pled using a tungsten-carbide hand scribe. These samples includedtwofromthefelutitethatcontainedmagnetiteandstilpnomelane,six

fromtheferhythmite(twoeachofthesilicatefacies,magnetitefaciesandmixedfacies)andfourfromtheGIF(threefromgranule-richareasandonefromaminnesotaite-bearingchertdevoidofgranules).Theminnesotaite-bearing chert sample from the GIF contained a smallamountofmagnetite.InordertomeasuretheFeisotopecompositionof theminnesotaite alone, a strongmagnetwasused to remove asmuchoftheresidualmagnetiteaspossible.

Theresultingpowders fromallof thesamplesweredissolved inheatedSavillexbeakersusingultrapure,concentratedHFandHNO3 for24hr.Thesamplesweredriedandre-suspendedinultrapure8m HCluntilcompletedissolutionwasachieved.Theironwasseparatedfromtherestofthematrixbyanionexchangechromatography(Beard&Johnson,1999;Beard,Johnson,VonDamm&Poulson,2003).AllsamplesweremeasuredfortotalFecontentsbytheFerrozinemethod(Stookey,1970)beforeandafterchromatographicseparationtoen-surethattherewasnolossofFe.Yieldsthroughthecolumnswereall>90%andmostwere≥98%.IsotopicanalysesofthepurifiedFewereperformed using a Micromass Isoprobe MC-ICP-MS and an Aridusdesolvatingnebulizer following themethodsofAlbaréde andBeard(2004)andBeardetal.(2003).

Ironisotopedataarereportedhereinstandarddeltanotationasbothδ56Feandδ57Fe,inunitsofpermil(‰):

The reference ratio for these values is the average of igneousrocks(Beardetal.,2003).TheIRMM-014standardhasaδ56Fevalue

(1)δ56Fe=

[(

56Fe∕54Fe)

sample(

56Fe∕54Fe)

standard

−1

]

×103

(2)δ57Fe=

[(

57Fe∕54Fe)

sample(

57Fe∕54Fe)

standard

−1

]

×103

F IGURE  3 Surfaceoutcrophandsample(a)andstudieddrillcorethinsectionpictures(b,c)oftheNcongaFormationGIFillustratingthefollowingsedimentaryfeaturesoftheunit:stackedbeds(a);upwardfininggradedbeds(a–c);lowanglecross-lamination(a,c);subangularchertintraclasts(a);muddrapes(a);andloadcasts(a).Theinsetinpanelbshowsdetailsofthegranuleswithchertcoresandmagnetite–calcitecoatings

6  |     SMITH eT al.

TABLE  1 Wholerockmajor,rareearthelementandstableisotopegeochemistryofthestudiedsamplesfromtheNcongaFormation.Theδ13Candδ18Odatapresentedareforbulkcarbonate

Sample name

Unit

PNG4 1288 PNG4 1294 PNG4 1297.15 PNG4 1297.18 PNG4 1298 PNG4 1299.4 PNG4 1304

Rock type Felutite Felutite Ferhythmite Ferhythmite GIF Felutite GIF

SiO2 wt% 43.42 45.18 74.08 41.53 55.55 39.54 79.37

TiO2 wt% 0.15 0.07 0.01 0.04 0.00 0.13 0.00

Al2O3 wt% 4.12 1.59 0.16 0.64 0.04 3.33 0.04

Fe2O3 wt% 44.78 48.36 17.44 45.61 24.03 49.05 17.26

MnO wt% 0.08 0.15 0.09 0.13 0.20 0.20 0.07

MgO wt% 3.87 2.36 1.24 1.38 0.53 4.10 0.16

CaO wt% 0.28 0.92 4.34 7.16 13.05 0.20 2.34

Na2O wt% 0.56 0.19 −0.02 0.06 −0.01 0.52 −0.01

K2O wt% 1.46 0.71 0.11 0.31 0.03 1.37 0.02

P2O5 wt% 0.08 0.06 0.02 0.03 0.01 0.08 0.02

Ni ppm 23.5 4.1 7.6 3.6 40.2 7.1

Cu ppm 7.3 3.0 1.9 1.5 1.1 2.5

Pb ppm 2.2 1.6 4.2 3.2 1.1 2.2

Sc ppm 2.0 1.0 1.0 1.0 5.0 1.0

Ba ppm 16.9 0.6 9.1 <0.5 105.4 11.7

Co ppm 4.3 1.3 2.0 <0.5 6.1 0.7

Cs ppm 18.7 0.3 5.7 <0.1 45.6 0.8

Ga ppm 2.3 <0.5 0.9 <0.5 3.8 <0.5

Hf ppm <0.5 <0.5 <0.5 <0.5 <0.5 <0.5

Nb ppm 0.9 <0.5 0.5 <0.5 1.6 <0.5

Rb ppm 50.4 1.5 17.4 <0.5 116.1 1.9

Sr ppm 25.1 67.8 117.2 143.5 44.7 71.0

Th ppm 1.4 0.2 0.7 0.2 1.7 <0.1

U ppm 0.2 <0.1 0.1 <0.1 0.3 <0.1

V ppm 19.0 7.0 10.0 6.0 32.0 9.0

W ppm 0.4 0.4 0.6 0.4 0.5 1.0

Zr ppm 10.8 1.2 5.9 0.6 18.1 1.0

Y ppm 7.8 2.1 4.1 0.73 8.6 1.38

La ppm 6.20 1.30 1.90 0.63 7.40 0.75

Ce ppm 11.10 4.30 4.20 0.72 15.20 1.37

Pr ppm 1.41 0.28 0.47 0.08 1.79 0.15

Nd ppm 5.40 1.20 1.90 0.29 7.40 0.62

Sm ppm 1.10 0.20 0.40 0.06 1.60 0.15

Eu ppm 0.22 0.10 0.16 0.06 0.31 0.04

Gd ppm 1.02 0.17 0.60 0.07 1.29 0.19

Tb ppm 0.16 0.03 0.08 0.01 0.25 0.03

Dy ppm 1.10 0.25 0.46 0.06 1.21 0.18

Ho ppm 0.22 <0.05 0.10 0.01 0.27 0.04

Er ppm 0.71 0.13 0.29 0.04 0.76 0.10

Tm ppm 0.11 <0.05 <0.05 0.01 0.10 0.01

Yb ppm 0.61 0.11 0.21 0.04 0.71 0.09

Lu ppm 0.08 0.01 0.05 0.01 0.12 0.01

LOI wt% 1.9 3.5 5.4 9.7 3.9 1.7

(Continues)

     |  7SMITH eT al.

of−0.09‰on this scale (Beard&Johnson,2004).Analytical preci-sionandaccuracyweredeterminedbymultipleanalysesofsamples,includingmultipleanalysesofthesamesolution,aswellastestsolu-tionsmadeoflaboratorystandardsofknownFeisotopiccomposition,namely IRMM-014, J-M andHPS Fe. Repeat analyses of individualsamplesolutionsyielda2-σ error in δ56Fevaluesof±0.07‰(n=14).

4  | RESULTS

4.1 | Petrography and mineralogy

TheGIFcanbedescribedasamassivegreychertrockthatcontainsabundant black round and oval-shaped magnetite-coated gran-ules (Figures3b,c and 4a) of 0.2–2mm in size. The coated grains

are irregular in outline and do not have a concentric internal tex-ture (Figure4a,c,d).The coatedgranules are thusverydistinct fromchemicallyprecipitatedfree-rollingooids(Richter,1983).Manyofthegranulerimsdisplayinternaldomalstructuresaswellasirregularlami-nationsofvarioussizesthataresimilarinappearancetostromatoliticmicrotextures (Figure4c–f).The stromatolite-shapeddomesarevis-ibleincrosssection(Figure4c,d)andplanview(Figure4e),dependingwhetherthethinsection intersectsthegranulesthroughtheircoresor rims, respectively.Thecoresofgranulesarecomposedofmicro-crystalline quartz (chert) that in turn is overgrown by rather poorlydefined zones of calcite intergrown with microcrystalline quartz.Therimsofthegranulesarethencoatedbymagnetitewithsubordi-natecalcite.Thesecoatingshavevariablethicknesses(<10–100μm; Figure4d,f).Wherewelldeveloped, themagnetite±calcitecoatings

F IGURE  4 Transmittedlightopticalphotomicrographs(a,b)andSEMimages(c–f)illustratingtheimportantpetrographicfeaturesoftheGIFoccurringintheNcongaFormation.(a)Generaltextureillustratingmagnetite(Mag)andcalcite(Cal)coatedchert(Qz)granulesinachertmatrix.Redblocksindicatemagnetiteandcalciterimfragmentsoccurringinthechertmatrix.(b)Minnesotaite(Mns)needlesinthechertmatrix.(c,d)Crosssectionviewofdomalfeatures,similarinappearancetomicrostromatolites,inmagnetiterimsofgranules.(e)Planviewofdomalfeaturesinmagnetiterimsofgranuleswithminorminnesotaiteinthechertmatrix.(f)Irregularlaminationindomalrimsofgranules

Sample name

Unit

PNG4 1288 PNG4 1294 PNG4 1297.15 PNG4 1297.18 PNG4 1298 PNG4 1299.4 PNG4 1304

Rock type Felutite Felutite Ferhythmite Ferhythmite GIF Felutite GIF

TotalC wt% 0.23 1.04 1.66 2.93 0.04 0.56

δ13CPDB ‰ −14.61 −14.17 −11.62 −15.70

δ18OSMOW ‰ 8.75 8.54 8.28 8.64

δ18OPDB ‰ −21.49 −21.70 −21.95 −21.60

TABLE  1  (Continued)

8  |     SMITH eT al.

display internal, irregulartoveryfinewrinkly laminations (Figure4f).Fragmentsofthecoatingsarealsovisibleinthechertmatrixbetweenthe granules (Figure4a). Fe-bearing carbonates are notably absent(Figure5).Themagnetitecrystalsinthecoatingsareeuhedralandveryfine,2–20μminsize.Insurfaceoutcropsamples,themagnetitehas

beenreplacedbyhaematiteduetosupergenealteration.However,nohaematite ispresent inanyofthedrillcoresamplesoftheGIF.Theovalshapeandbedding-parallelorientationofmanyofthegrainssug-gesttheywereaffectedbycompaction,althoughitcouldalsobeduetoasortingeffect.

Thegranulesdefinegradedbedsthatare4–5cmthick,withthenumberandsizeofthecoatedgrainsdecreasingfromthebottomtothetopoftheindividualbeds(Figure3b,c).Granulesareenclosedin amicrocrystalline quartz (chert) cement or matrix,with tracesofminnesotaiteoccurring (Figure4b).Crosscuttingveinsarefilledwithcalcite.

ThefelutitethatoccursbetweentheGIFunitsandabovethefe-rhythmite isbestdescribedasamassivelytexturedsilicatefacies IFwithstilpnomelaneasthedominantmineralphase(Figure6a,b).Thestilpnomelaneoccursasfine(10–30μm)needle-shapedcrystalssur-roundedbymicrocrystallinequartz.Magnetiteispresentasisolated,sub-toanhedralcrystalsinawiderangeofsizes(<10–100μm)thatovergrowandreplacestilpnomelane(Figure6b).Tracesoffinecrystal-linepyriteoccur.

The ferhythmite that immediately overlies the upper GIF unitis amixed oxide and silicate faciesmicrobanded IF (Figure6c).Themicrobands comprise magnetite, quartz, Fe-silicates (stilpnomelaneand minnesotaite) and calcite in variable proportions (Figure6c,d).Themagnetite-bearingmicrobandscontainsmall(~10μm),subhedralmagnetiteaggregatesassociatedwithquartzandminorstilpnomelaneandcalcite.Thecalcite,wherepresent,appearstocrosscutmagnetitemicrobands (Figure6d). Large (~200μm), slightly compacted spher-oidsof stilpnomelane and calciteoccur locally in themagnetitemi-crobands.The silicatemicrobands consistof fine stilpnomelaneandminnesotaite needles enclosed in quartz. The mixed silicate-calcitemicrobandsaresimilar tothesilicatemicrobandsexceptthatcalciteoccurs as anhedral, isolated crystals. Contacts betweenmicrobandsare generally sharp (Figure6c), but can occasionally be gradational.Justasinthefelutite,tracesoffinecrystallinepyritealsooccurintheferhythmite(Figure6d).

4.2 | Geochemistry

The SiO2 content of the GIF varies between 55.6 and 79.4wt %(Figure7a; Table1). The other two major constituents are Fe2O3

T (17.3–24.0wt %) and CaO (2.34–13.1wt %). These compositionsreflectthemajormineralconstituents,whicharequartz,magnetiteandcalcite.Allothermajorelementsshowconcentrationsof<0.5wt%.Traceelementsshowextremelylowcontentsthataregenerallyclosetoorbelowdetectionlimits(Table1).OnlyTiandSrconcentrationsexceed50ppm.

Compared to the GIF, the felutite has significantly higher con-centrations of Fe2O3 (44.8–49.0wt %) and lower contents of SiO2 (39.5–45.2wt%).ConcentrationsofAl2O3(1.59–3.33wt%)andMgO(2.36–4.10wt%) are also higher in the felutite (Figure7a;Table1),as are concentrations of trace elements. The ferhythmite, in turn,containssimilarSiO2concentrations(41.5–74.1wt%)totheGIFandrathervariableFe2O3(17.4–45.6wt%)contents,withAl2O3andMgO

F IGURE  5 BackscatterSEMimageofagranulerimfromtheNcongaFormationGIFindicatingthelocationofthreeEDSspotanalysesalongwiththeEDSspectraacquiredforthefollowingmajorminerals:(1)magnetite;(2)quartz;and(3)calcite

     |  9SMITH eT al.

contentsintermediatetothatoffelutiteandGIF(Figure7a).CaOcon-tents (4.34–7.16wt%) in the ferhythmiteare in the same rangeasintheGIF.AswiththeGIF,onlyTiandSrshowcontentsinexcessof50ppmintheferhythmite.

RareearthelementconcentrationsintheGIFareexceptionallylowwith a total (ΣREE) of only 2.1–3.7ppm.ΣREE is somewhat higherintheferhythmite(8.1–10.8ppm)andhighestinthefelutitesamples(29.4–38.4ppm).Post-ArchaeanAustralianShale (PAAS)normalizedplots of theREEs andY (inserted betweenDy andHo afterBau&Dulski, 1996) for the studied samples (Figure7b) showslightheavyREE(HREE)enrichmentover lightREE(LREE) intheGIF.Thestrati-graphicallyupperGIFunitshowslowerREEcontentsandprominentpositiveEuandYanomalieswhencompared to the lowerGIFunit.TheferhythmitesamplesshowpositiveEuandYanomalies,andonesampleshowsatruepositiveCeanomaly(Bau&Dulski,1996).HREEenrichmentisalsoobserved.ThefelutitesamplesshowflatnormalizedREYpatternswithnoprominentanomalies.

Stable isotope analyses of calcite in the GIF and ferhythmiteshowδ18OPDBrangesbetween−22.0and−21.5‰relativetoPeedeeBelemnite (PDB). The δ13CPDB ranges between −15.7 and −11.6‰(Figure7c;Table1).

The δ56Fe values of magnetite in felutite (sample PNG4-1294)rangefrom−0.18to−0.02‰(average=−0.10±0.23‰2-SD,n = 2; Table2 and Figure8). In the ferrhythmite (sample PNG4-1297.15),magnetite-richmicrobandshaveδ56Fevaluesthatrangefrom−0.01to +0.04‰ (average=+0.01±0.07‰ 2-SD, n=2), silicate faciesmicrobands have δ56Fe values that range from −0.56 to −0.47‰(average=−0.52±0.13‰2-SD, n=2) and themixed faciesmicro-bands have δ56Fe values that range from −0.31 to −0.12‰ (aver-age=−0.21±0.27‰2-SD,n=2)(Table2andFigure8).FortheGIF(samplePNG4-1298),magnetiteshaveδ56Fevaluesthatrangefrom+0.39to+0.48‰(average=0.44±0.09‰2-SD,n=3).Minnesotaiteinchertcementofcoatedgranuleshasaδ56Fevalueof+0.04‰(n = 1; Table2andFigure8).

5  | DISCUSSION

Thepetrography and isotopegeochemistryprovide insight into themineral paragenesis of the GIF and the associated iron-rich units,whereas the field observations, lithostratigraphy and geochemistryprovideinsightintothedepositionalenvironmentofalliron-richunits.Thesecharacteristicsarediscussedwiththeaimofprovidingadepo-sitionalmodelforthisiron-richsuccessionaswellasamodelfortheoriginofthestructuresinthegranulecoatingsoftheGIF.

5.1 | Mineral paragenesis

Microcrystalline quartz (chert) is the most common mineral in thefelutite, ferhythmite and GIF. All the iron-bearing minerals occurclosely associated and intergrown with chert, except in the GIFwhere the magnetite and calcite only occur as rims around chertnuclei (Figure4) that are suspended in a chert matrix (Figure3).Theserelations,aswellastheconsistentlyhighSiO2contentsofallsamples,would suggest continuous precipitation and deposition ofsilicainthebackgroundinallIFunitsoftheNcongaFormation.Themicrocrystallinenatureofthequartzsuggeststhatitwasderivedfromrecrystallizationofanamorphousprecursor(Clout&Simonson,2005;Maliva,Knoll,&Simonson,2005).RecentstudiesofmixedFe-Sigelsindicate that thiswas themost likelyprimaryprecipitate forSi andFe in theArchaean oceans, and later burial diagenesis, dewateringand recrystallization of suchmaterialswould produce amixture ofFe minerals and quartz (Rasmussen, Krapež, Muhling, & Suvorova,2015;Wu,Percak-Dennett,Beard,Roden,&Johnson,2012;Zheng,Beard,Reddy,Roden,&Johnson,2016).ThebulkFe:SiratiosofthesamplesanalysedinthisstudyspantherangeofFe-Sigelsstudiedinexperiment.

Stilpnomelaneandminnesotaitearecommoninboththeferhyth-miteandfelutiteandarecommonsilicatesindiagenetictolow-gradeIFmineral assemblages (Klein, 2005).AmorphousFe2+-rich silica gel

F IGURE  6 Transmittedlightopticalphotomicrograph(a)andSEMimages(b-d)ofthefelutite(a,b)andferhythmite(c,d)intheNcongaFormation.(a,b)Massivelytexturedstilpnomelane(Stp)withmagnetite(Mag)inquartz(Qz).(c,d)Contactbetweenoxidefaciesandsilicatefaciesbandsintheferhythmite.Oxidefaciesbandscomprisefine-grainedeuhedralmagnetitewithminorcalcite(Cal)inquartzwhereassilicatefaciesbandcomprisestilpnomelanewithminorcalciteandmagnetiteinquartz.TheacceleratingvoltageforallBSEMimageswas15kV.Py:pyrite

(a) (b)

(c) (d)

10  |     SMITH eT al.

iscommonlyregardedasprecursortosuchsilicates(Klein,2005)andcould have been precipitated in a non-redox process. Alternatively,Fe2+- silicatesmay also form as the product of diagenetic reactions

whereFe(III)-Sigelswerereducedinsitu(Percak-Dennettetal.,2011),or interacted with porewater aqueous Fe(II) (Zheng etal., 2016).AnotherpossibilityisthattheFe2+-silicatesformedfromthediagenesis

F IGURE  7 Selectedmajorelementgeochemistrybarcharts(a),PAAS-normalizedREEdiagrams(b)andbinaryplotofcarbonandoxygenstableisotopes(c)oftheGIF,ferhythmiteandfelutiteoftheNcongaFormation.WitwatersrandSupergroupdatafromSmithetal.(2013)andAsbesheuwelsIFdatafromKaufman(1996)

     |  11SMITH eT al.

TABLE 2 IronisotopedatafromsmallbulksamplesofvariousFe-richlithologiesstudied

Sam

ple

Dom

inan

t Fe

min

eral

ogya

Fe (w

t %)

Repl

.b

Indi

vidu

al a

naly

ses

Ave

rage

s for

repe

ats

Gra

nd a

vera

gesc

δ56Fe

(‰)

2 SE

δ57Fe

(‰)

2 SE

δ56Fe

(‰)

2 σ

δ57Fe

(‰)

2 σ

δ56Fe

(‰)

2 σ

δ57Fe

(‰)

2 σ

CoresamplePNG4-1294,Felutite

1Magnetite

29.2

a−0.18

0.02

−0.25

0.02

−0.10

0.23

−0.14

0.32

2Magnetite

30.2

b0.

030.

010.

040.

02−0.02

0.13

−0.03

0.19

b−0.07

0.01

−0.09

0.01

CoresamplePNG4-1297.15,Ferhythmite

1Fe-silicate

10.4

a−0.47

0.03

−0.72

0.04

−0.52

0.13

−0.77

0.15

2Fe-silicate

15.4

b−0.57

0.02

−0.84

0.02

−0.56

0.03

−0.83

0.05

b−0.55

0.02

−0.81

0.03

3Magnetite

56.8

c0.

080.

010.

150.

020.

040.

120.

080.

200.

010.

070.

010.

18

c0.

000.

010.

010.

02

4Magnetite

59.0

d−0.01

0.02

−0.05

0.02

5Mixed

21.6

e−0.31

0.02

−0.40

0.02

−0.21

0.27

−0.30

0.29

6Mixed

33.0

f−0.12

0.03

−0.19

0.03

CoresamplePNG4-1298,Oncoidalgranularironformation

1Magnetite

25.7

a0.

350.

010.

540.

020.

450.

310.

670.

380.

440.

090.

660.

12

a0.

560.

020.

810.

02

2Magnetite

21.6

b0.

390.

020.

600.

030.

480.

260.

720.

35

b0.

570.

020.

840.

03

3Magnetite

16.3

c0.

410.

020.

580.

020.

390.

070.

600.

07

c0.

370.

010.

630.

02

4Fe-silicate

5.6

d0.

020.

010.

040.

010.

040.

050.

070.

09

0.06

0.02

0.10

0.02

a Repl.=replicates.Repeatedlettersrefertomultipleanalysesofthesamesamplealiquot.Differentlettersrefertodistinctsamplealiquotsprocessedseparately.

b Basedonpetrographicanalyses.

c Averagesofmeanvaluesformultiplesamplesofthesamemineraltypewithinacore.

12  |     SMITH eT al.

ofprecipitatedgreenrust,aferrous-ferrichydroxylsalt(Halevy,Alesker,Schuster,Popvitz-Biro,&Feldman,2017).Indeed,theonlymildeffectofcompactionontheshapeofstilpnomelanespheressuggestsadia-geneticratherthansynsedimentaryoriginofstilpnomelane.

Two texturally distinct types of magnetite are recognized.Largereuhedraland isolatedmagnetitecrystalsoccur in thefelutite(Figure6b),whereasminute(~2–20μminsize)euhedraltosubhedralmagnetitecrystalsformaggregatesormassivemonomineralicmicro-bandsintheferhythmite(Figure6c,d)andrimsaroundgranulesintheGIF(Figure4f).Basedontexturalrelations,thelargereuhedralmagne-titecrystalsinthefelutitemaywellreplacestilpnomelane,indicatingthattheyarelateintheparageneticsequence.Thiseuhedralmagne-titeisthusinterpretedasformingbylow-grademetamorphicdecom-positionofstilpnomelanetomagnetiteandquartz.

In contrast,minutemagnetite crystals appear tobe finely inter-grownwithmicroquartzandcalciteandappeartobeveryearlyinthemineral paragenesis. This magnetite is thus considered to be earlydiagenetic inorigin.This isconsistentwith insituO isotopestudiesofmagnetiteinIFsthatshowfine-grainedpuremagnetitecommonlyhas relatively low δ18Ovalues, indicating an early, low-temperatureformation (Lietal.,2013).Theminutemagnetite in the ferhythmiteandGIFareinferredtohaveformedattheexpenseofaninitialFe3+-oxyhydroxide-silicaprecipitate.Possibleoriginsforthe initialprecip-itateincludeoxidationofFe2+

aqbyfreemolecularoxygengeneratedby photosynthesis (Klein & Beukes, 1993), chemolithoautotrophs(Konhauser,2007;Konhauseretal.,2002)oranoxygenicphototrophiciron-oxidizingbacteria(Kappleretal.,2005;Posth,Hegler,Konhauser&Kappler,2008).ThestableFeisotopedataplaceconstraintsonthese

F IGURE  8 Theironisotopecompositionsoftheiron-richmineralsintheNcongaFormationrelativetotheirstratigraphicpositioninboreholePNG4(seeTable2).Theareassampledbyhandscribeareindicatedbywhiteshapesontheimagesofthepolishedsections.NotethatthelowironcontentoftheFe-silicatesinsamplePNG4-1298necessitatedcombiningthematerialfromthetwoareassampledintoonesample.Thelegendinthemiddlerightpanelappliestoallisotopepanels.TheverticalgreybarineachisotopepanelindicatestheaverageofigneouscrustandtheapproximateestimatedironisotopevalueoftheArchaeanocean(Johnsonetal.,2008a).GIF,granularironformation;Fe-sil,ironsilicates

     |  13SMITH eT al.

possiblemodelsasdiscussedbelowinthesectiontitled,Constraints on Iron Source and Deposition.Itisunlikelythatthefine-grainedmag-netitewas produced by metamorphic breakdown of Fe carbonates(e.g.,McCollom,2003),whichwouldproduceabundanthydrocarbonsandlikelyretainresidualsiderite,neitherofwhichisobservedinoursamples.

Thereareseveralpossiblepathwaysforthediageneticformationofmagnetite.Diagenetic reduction inFe3+ through theoxidationoforganicmatter is a possible origin ofmagnetite in theGIF and theferhythmite, and magnetite is a common product of microbial dis-similatoryironreduction(e.g.,Roden&Lovley,1993).CarbonatesinboththeGIFandferhythmiteareCa-richrich,however,andcontainverylittletonoFe(Figure5),asmightbeexpectedifassociatedwithmicrobial iron reduction. Iron-rich carbonates are common in otherMesoarchaeanBIFs in theWitwatersrandandPongolaSupergroupsaswellasinNeoarchaeanBIFsoftheTransvaalSupergroup,consistingofpredominantlyankeriteandsiderite thathaveverynegativeδ13Cvalues that indicate inheritance throughoxidationoforganicmatter(Heimannetal.,2010;Smith,2007;Smithetal.,2013).TheFe-poornatureofthecarbonatesintheGIFandferhythmiteunitsstudiedheresuggestsanotherpathwayforcarbonateformationthatisnotrelatedtomagnetite formation.Alternatively, the fine-grainedmagnetite intheGIFandferhythmitecouldberelatedtoanon-redoxreactionofFe3+oxyhydroxideswithFe2+-rich fluids in thewatercolumnand inthesediments(Ohmoto,2003)andisapossibleformationpathwayforfine-grainedmagnetiteintheNcongaFormation.RecentexperimentalstudiesofFe-Sigels,however,indicatethatonlylimitedconversionofFe3+hydroxidesoccursinthepresenceofaqueousFe2+(Zhengetal.,2016),insufficienttoproducemagnetitestoichiometryforFe2+,evenusingaqueousFe2+ contentsof~1mm,whichgreatlyexceedsmostestimates for Archaean seawater (e.g., Czaja etal., 2012; Holland,1984).ItisthereforepossiblethatiftheFe3+precursorsedimentwas

an Fe3+-Si gel and not Fe3+ hydroxide, conversion to amixed Fe2+-Fe3+-Si gel that, upon dewatering, produced magnetite and quartz,may in fact requiremicrobial iron reduction despite the absenceofFe-bearingcarbonates.

The lightδ13Csignatureof the calcite (Figure5c) in the rimsofthe granules in theGIF indicates that carbonatewas likely derivedthroughoxidationoforganiccarbon.Itisunclear,however,whattheelectron acceptor may have been. Sulphate (SO2−

4) was unlikely to

havebeentheoxidant,asthiswouldhaveformedsulphide,andpyriteisallbutabsentfromthestudiedsamples(onlytraceamountsoccurintheIFandnoneintheGIF).Inaddition,ferriciron(Fe3+)wouldnotbethoughttohavebeentheoxidant,asFe-bearingcarbonates(an-keriteorsiderite)areabsentinthestudiedsamples,asnotedabove.IntheirdetailedstudyoftheKurumanIF,Heimannetal.(2010)notedthatclearevidenceformicrobialFe3+reductionwassuppliedbytheoccurrenceofhaematiteinclusionsinFe-bearingcarbonates(ankeriteandsiderite)thatsimultaneouslyhadnegativeδ13Candpositiveδ56Fevalues.TheabsenceinoursamplesofFe-bearingcarbonates,aswellashaematite,makesitdifficulttoargueforFe3+astheelectrondonorthatwascoupledtoorganiccarbonoxidationtoproducethelowδ13Ccalcite.Ametamorphic origin for the calcite is also considered un-likely,asithasnoexclusivemineralassociationandthereforeshowsno petrographic evidence for being part of ametamorphicmineralassemblage,forexampleanexclusiveassociationwithmagnetiteifitwouldhavebeenderivedfromprecursorsiderite.Thecalcitetexturereportedhereisalsoverydifferentfromcontactmetamorphiccalciteassociatedwith grunerite formed from the breakdown of iron-richcarbonatesdocumented in theScottsHillMember IFwhichoccursstratigraphicallybelowtheNcongaFormation in theMozaanGroup(Smith,2007).

Wespeculatethat,intheabsenceofaclearmineralogicalrecordforsulphateorFe3+reduction,thatoxidationoforganiccarbonmight

F IGURE  9 DepositionalmodelfortheoncoidalGIFrelativetothedeeperwaterferhythmiteandfelutiteoftheNcongaFormation.Theverticalelevationisexaggeratedforillustrativepurposes.BIFdepositionalmodeladaptedfromSmithetal.(2013)

14  |     SMITH eT al.

haveoccurredviaanoxidantthatdidnotleaveamineralogicalbyprod-uct,andhencemayhavebeenamobileorvolatilephase.Onepossibil-ityisthattheoxidantwasnitrate(NO−

3),whichcouldhaveparticipated

viathemicrobiallymediatedreaction(Castanier,LeMétayer-Levrel,&Perthuisor,2000;Konhauser,2007):

Wesuggest thatnitratemayhaveexisted in thewatercolumnat 2.9Ga, based on the increasing evidence for incipient oxygencontents,includinginshallowmarinesettings,asfarbackas3.2GabasedonCrandMostableisotopes,aswellasthecombinationofFeandU-Th-Pbisotopes(Croweetal.,2013;Planavskyetal.,2014;Satkoski,Beukes,Li,Beard,&Johnson,2015).Althoughequilibriumthermodynamic relations suggest thatnitrate isunlikely tocoexistwithaqueousFe2+,significantquantitiesofaqueousFe2+andnitrateisfoundinporefluids inmodernmarinesettings,dependentupondiageneticandauthigenicmicrobialcycling(e.g.,Lauferetal.,2016).DirectevidencefornitrateintheenvironmentinthelateArchaeancomesfromNisotopes,wherepositiveδ15NvaluesinNeoarchaeanrocks have been interpreted to reflect nitrate cycling (Godfrey &Falkowski,2009;Thomazo&Papineau,2013),althoughinterpreta-tionofNisotopedatainArchaeanrockscanbedifficult(e.g.,Aderetal.,2016;Stüeken,Kipp,Koehler,&Buick,2016).Toourknowl-edge, however, there are no N isotope data from Mesoarchaeanrockssimilartothosestudiedherethatprovidedirectevidencefornitratecycling.

5.2 | Depositional setting

TheNcongaFormationismarkedbyaverycloseandcogeneticassoci-ationofmicrobandedIF(ferhythmite),massivelytexturedIF(felutite)andGIF.ThetexturaldifferencebetweenGIFandtheIFcanatleastpartlybeattributedtodifferenthydrodynamicconditionsinthedep-ositional environment (Simonson&Goode, 1989). Ferhythmite andfelutitearegenerallyassociatedwithdeepwater,belowwavebaseenvironments(Figure9).TheGIF,incontrast,ismarkedbytheoccur-renceofcoarse-coatedgrainsthatrequireformationonashallowsub-mergedshelfwithconstantwaveactionthatwaslocallyabletoreworkchertprecipitates(Figure9).AsthininterbedsofGIFaresandwichedbetweendeeperwaterIFdeposits(Figure1c),itmaybeassumedthatgranuleswerewashedinfromtheshallowsubmergedshelfbystormwavesurgesintodeeperwaterenvironments(Figure9).Thisinterpre-tationissupportedbytherecognitionofthefollowingfeatures:cross-bedding(Figure2d);internallowanglecross-lamination(Figure3a,c);gradedbedding(Figures2hand3a–c);loadcastsandmuddrapesatthe base and top of graded beds, respectively (Figures2h and 3a);subangular chert intraclasts (Figures2h and 3a); reworked granulesurfaces at the topofGIFbeds (Figure2e); granule lenticleswithinMIFoverlyingtheGIF(Figure2f);partiallybrokenmudlaminaewithoverlyingrip-upsofthemudlaminae(Figure2g);hummockyandlin-earwave ripplemarks (Figure2b,c); and fragments of themagnet-ite–calcitecoatingsinthechertmatrix(Figure4a).Thefelutitebelow

theupperGIFunitcontainssomestarvedripplesofGIF(Figure1c),illustratingthatsomeofthecoatedgrainsweretransportedastrac-tionloadintothedepositionalenvironmentoftheMIF(Figure9).Thequartziteunitthatcapsthedeepwateriron-richunitdoessowithanerosive contact (Figure1c),most likelymarking a forced regression(Posamentier,Allen,James,andTesson(1992)thatoccurredafterthedepositionoftheiron-richunits.Insequencestratigraphicterms,theIFs in theMozaanGroupareconsideredtohavebeendeposited instarvedshelfsettingsdevelopedduringperiodsofmaximumratesofrelativesea-levelriseinthebasin,thatisalongmaximumfloodingsur-faces(Beukes&Cairncross,1991).

As is typical for IFs, felutite, ferhythmiteandGIFof theNcongaFormationareallmarkedbyverylowconcentrationsofconstituentsreflectingdetrital influx (Al2O3,TiO2, Zr;Table1).Concentrations intheGIF,however,areexceptionally lowandsignificantly lower thaninfelutiteandferhythmite.Thismaysuggestthatthedepositionalen-vironmentoftheGIFwaswellshieldedfromdetrital inputfromthecontinent.Here,wesuggestthisshieldedshallowenvironmenttohavebeen a submerged palaeo-high in theMozaanBasin in the formofashelfbank isolatedfromland (Figure9).Suchasettingwouldalsoexplain the rather restricted distribution of the GIF, surrounded bymagneticmudstone,intheNgoncaFormationandthelargeroriginalWitwatersrand-Mozaanbasin(Figure1a).

ThedeepwaterIFsintheNcongaFormationcompriseonlymixedoxideandsilicatefaciesferhythmiteandsilicatefaciesfelutite.OtherIFsintheMozaanGroupandthecorrelativeWestRandGroupoftheWitwatersrand Supergroup show more complete stratigraphic se-quences (Smith etal., 2013). From a distal to proximal depositionalsetting, oxide facies is found to grade intomixed oxide-carbonate-faciesgrading intosilicate faciesand finally into iron-richmudstoneandiron-poorshaleandquartziteinthemostproximalsettings(Smithetal., 2013).The abundance of silicate facies IF, lack of pure oxidefaciesIFandthetoperosivecontactwithquartziteinthestudiedse-quenceoftheNcongaFormationsuggestthatonlythemoreproximalenvironmentsarepreservedinthestudieddrillcore.Theoriginallat-eralfaciesdistributioninthedepositionalenvironmentoftheNcongaFormationcanonlybeassumedtohavebeenacompletemineralogicalfaciessequence(Figure9).

5.3 | Origin of structures in granule rims

ThepresenceofstratiformandconicalstromatolitesinpartlysilicifieddolostonehasbeenestablishedinthelowerNsuzeGroupbyBeukesandLowe(1989)andevidenceforsediment-stabilizingbacterialmatshasbeendocumentedinthequartzitesoftheNtombe(Noffke,Hazen,& Nhleko, 2003) and Sinqeni Formations (Noffke, Beukes, Bower,Hazen, & Swift, 2008) of the Mozaan Group. These occurrencesillustrate that biological activity in the formof filamentousbacteriawaslikelypresentatearliertimesduringthedepositionofthePongolaSupergroupthantheNcongaFormation.

Theminutedomaltexturesandirregularlayeringobservedaroundthe granules that resemble stromatolites (Figure4c–f) along withthe carbon isotope compositionof the calcite in the rim (Figure7c)

(3)2(CH2O)+NO−

3+Ca

2+→CaCO3+CO2+NH

+

4

     |  15SMITH eT al.

stronglysuggestabiologicalinfluenceontheprecipitationoftheiron-richrims.Theserimsarethereforeinterpretedtobemicrostromato-litesthatgrewaroundthegranules.Theroundandoval-shapedchertcoresofthegranuleswouldsuggestthatprecipitationofaniron-richphaseoccurredontherimsofchertpeloidsthatwerebeingshapedbyepisodicreworking,mostlikelybywaveaction.Becausestromat-olitesgrowupwardstowardssunlight,therollingofthesepeloidsontheseafloorduetowavereworkingalsoexplainsthemulti-directionalnature observed in the microstromatolites coating the granules(Figure4c,d).Furtherevidence foradirect rolebybiologicalactivityisprovidedbymarkednegativevaluesforδ13CPDBincalciteintimatelyassociatedwithmagnetite (Figure7c).Suchnegativeδ13Cvalues in-dicatecarbonthatwassourcedfromtheoxidationoforganiccarbonandstronglysuggeststhatorganicmattercontributedasthesourceofcarbonforcarbonatemineralization (DesMarais,2001;Schidlowski,1995).Furthermore,thelikelybiologicaloriginofthegranulecoatingsindicatethattheycanbeclassifiedasoncoids,whicharedefinedasbiogenicallyencrustedgrains(Peryt,1983a)andarealsoclassifiedasastromatolitesubtype(Hofmann,2000).TheNcongaGIFcanthereforebeclassifiedasanoncoidalGIF.

Iron precipitates responsible for oncoidal textures could havebeen deposited by either chemolithoautotrophic (Konhauser, 2007;Konhauser etal., 2002) or anoxygenic phototrophic iron-oxidizingbacteria (Kappleretal.,2005;Posthetal.,2008) formingstrongmi-crobialfilmsormats(Emerson&Revsbech,1994)aroundchertpeloidnucleitosurviveinthewave-dominatedshallowmarineenvironment.Suchmicrobialfilmsarewell-knowntotrapdetritalparticles(Kearsley,1989),butintheabsenceofsuchdetritusincaseoftheNcongaon-coidalGIFthefilmsformedaroundconsolidatednucleiofchertandactedaschemicaltrapsforiron(Fortin,Ferris,&Beveridge,1997).Thebacterial films then grew in the shapeof domalmicrostromatolites.Successivemicrobialfilmsaroundthegrainscreatedtheirregularlay-ering(Figure4f)andcausedchertandorganicmaterialtobetrapped.

It hasbeen suggested that bacterial surfaces couldhave servedas nucleation sites for iron mineralization in IFs (Konhauser, 1998;Konhauser&Ferris,1996;Warren&Ferris,1998).Inmicrobialmatsfrom Icelandichot springsmanycontemporaryexampleshavebeenstudiedofbacterialcellsbeingcompletelyencrustedbyiron-richpre-cipitates, potentially enhancing their preservation potential (Ferris,2000;Konhauser& Ferris, 1996).These encrusted cells showmor-phologyandlayeringrathersimilartothatofthemicrostructuresintheNcongaoncoids,eventhoughthecontemporaryexamplesaresmallerinsize.

ItshouldbenotedthatDoddetal.(2017),whodocumentedhae-matitefilaments,carbonaterosettesandmagnetite-rimmedgranulesin thePalaeoarchaeanNuvvuagittuq supracrustal belt (NSB) jaspersof Quebec, Canada, state that the latter granules are comparableto the ones in theNcongaGIF as briefly presented by Smith etal.(2012). Furthermore they interpret the granules in theNSB jaspersto be diagenetic and that biomass oxidation played amajor part intheirgrowth.TheimplicationisthatadiageneticoriginforthegranulesintheNcongaGIFcouldbeconsidered.However,althoughtherearesomesimilaritiessuchasgranulesizesandthepresenceofmagnetite

andchert,thereareseveralmarkeddifferencesbetweentheNcongagranulesandtheNSBjaspergranules:themagnetiterimmorphologiesaredrasticallydifferent,withthecoarser,angularandeuhedralmag-netite in theNSB jasper granule rims not completely encapsulatingthegranulesandlackinganydomalstructures;wherecalciteisamajorconstituentintheNcongagranules,itismostlyabsentfromtheNSBjasper granules; apatite and haematite are common constituents oftheNSBjaspergranulesandarecompletelyabsentfromtheNcongagranules;andthecomplexinternalmineralogyandwhatcouldbein-terpretedasinternalgrowthstructuresintheNSBjaspergranules,arecompletelyabsentintheNcongagranules.Therefore,itisnotaccurateto call theNconga granules and theNSB jasper granules compara-ble.AlthoughdiagenesisandorganiccarbonoxidationdidplayaroleinthemineralogycurrentlyobservedintheNcongaGIF,adiageneticgrowthoriginforthegranulesasawholeisnotsupportedbythedataforthefollowingreasons:abundantevidenceofwavetransportandreworking(Figures2and3);noobservedintergrowthofgranulesandtheirrims,butratherastackedandgradednaturetothegranulesbeds(Figures3and4);andacompletelackofinternalgrowthstructuresinthegranules(Figure4a,c,d).Thesedimentologicalevidencedoesshowthatthegranuleswereformedatadifferentsettingfromwheretheywere finally deposited.However, if theywere initially growndiage-neticallyandonlythenreworkedbystormsurges,theGIFwouldhavehadamoreclasticandbrokenuptextureratherthanthemostlyintactcoatedgranulesobserved (Figure3).Reworked rock fragmentswithaninternalgranulartextureshouldalsobepresentandthatisnotthecase.Thisisbecauseifthegranulesinitiallygrewdiageneticallyinthesedimentatconditions that requiresignificantburial,asopposedtogranulesformingandbeingcoatedinasilicagel-richenvironmentasproposedhere,itwouldbealithifiedrockunitbeingreworked.

5.4 | Constraints on iron pathways, sources and deposition

Rare Earth Elements and Y abundances (“REY”) are an importantgeochemical tool in understanding the origin of IFs (e.g., Bau &Dulski, 1996; Bekker etal., 2010; Klein & Beukes, 1992). ThePAAS-normalized REY patterns of the ferhythmite show a typicalPrecambrianIFsignaturewithHREEenrichmentandpositiveEuandYanomalies(Klein,2005;Planavskyetal.,2010).LREEdepletionandHREE enrichment is typical for oceanwater precipitates (Brookins,1989; Fryer, 1983). The positive Eu anomaly is consistent withhydrothermalinputintothedepositionalsetting(Bau&Dulski,1996;Dymek&Klein,1988;Klein&Beukes,1993)andsuggeststhattheiron in thedeeperwater IFwas introducedbyhydrothermal fluids,whichwouldbeexpectedtobereduced(e.g.,Slack,Grenne,Bekker,Rouxel,&Lindberg,2007;Bekkeretal.,2010;Figure9).

TheoncoidalGIFunitsshowsomebutnotalloftheREYcharac-teristicsseen intheferhythmite,withthe lowerunitshowingHREEenrichment, but no Eu orY anomalies, and the upper unit showingpositiveEuandYanomaliesbutlessHREEenrichment(Figure7b).TheflatPAAS-normalizedpatternsobservedinthefelutitearemostlikelyduetoREEsfromdetritalmaterialthatwouldhavedominatedtheREE

16  |     SMITH eT al.

signatureofthechemicalprecipitates.Thisconclusionissupportedbyelevatedconcentrationsofconstituentsreflectingdetritalinflux(e.g.,Al2O3,TiO2andZr)inthislithology(Table1).

Ironisotopegeochemistryprovidesfurtherinsightintoironcyclingand therefore themechanismof iron deposition (Czaja etal., 2012,2013;Johnson,Beard,Beukes,Klein,&O’Leary,2003;Johnson,Beard,Klein, Beukes, & Roden, 2008a; Planavsky etal., 2012).The largestiron isotope fractionations occur with redox processes (Johnson,Beard,&Roden,2008b),whereinpartialoxidationofferrousironpro-ducesaferricironprecipitatethathasrelativelyhighδ56Fevalues,andaresidualferrousironpoolhasrelativelylowδ56Fevaluesrelativetotheoriginalsource(Bullen,White,Childs,Vivit,&Schulz,2001;Croal,Johnson,Beard,&Newman,2004;Planavskyetal., 2012).Thisob-servationappliestopureFe3+precipitates,aswellasFe3+-Sigels(Wuetal.,2012).Ingeneral,itisnotpossibletousestableFeisotopecom-positionstodistinguishbetweenbiologicalandabiologicaloxidationpaths(Bullenetal.,2001;Croaletal.,2004),butthemagnitudeoftheincreaseinδ56FevaluesfortheFe3+precipitatesdoconstraintheex-tentofoxidationandhencecanbeusedtoestimateambientoxygenlevels(e.g.,Czajaetal.,2013).Thiswillbefurtherexploredbelow.

AhydrothermalsourceofironlocatedindeepwaterenvironmentsoftheArchaeanOceanwouldmost likelyhavehadaδ56Fevalueofabout 0‰ (Johnson etal., 2008a) (Figure9).Oxidation andprecipi-tationofaportionofthisferrousironindistal(pelagicoroutershelf)settingsasoxidefaciesIFwouldhaverenderedtheprecipitatedironenrichedwithpositiveδ56Fevalues,andtheremainingdissolvedfer-rous ironwouldhavesomewhatnegativeδ56Fevalues.Thisprocesswouldhavebeenultimatelyexpressedinmoreelevatedδ56FevaluesindistalIFsandlowerδ56Fevaluesinmoreproximallithotypes.ThisprocessmightexplainthedistributionofFeisotopedataintheIFsoftheNcongaFormation(Table2andFigures8and9).

ThedominanceofFe-silicates intheferhythmiteandthefelutitesuggeststhattheyweredepositedinamoreproximalclosertoshoreenvironmentwithhigherinputofterrestrialdetritus.ThehydrothermalFe2+

aq,ifsourcedfromtheopenocean,wouldbeexpectedtohaveneg-ativeδ56FevaluesintheseproximalFe-silicatesifitrepresentstheresi-duefromextensivedepositionofoxidefaciesIFinamoredistalsetting(Figure9).Precipitationof ferricoxyhydroxides inthemoreproximalsettings,ifformedfromlow-δ56FeFe2+

aq,mighthavenear-zeroδ56Fe

values,giventhe56Fe/54FefractionationsbetweenFe3+-bearingmin-eralsandFe2+

aq,asdiscussedabove(Table2andFigure8).Theferroussilicateprecursortominnesotaiteandstilpnomelane,mostlikelypre-cipitatedthroughanon-redoxprocess,wouldbeexpectedtoinheritthenegativeδ56Fevaluesof thehydrothermalFe2+

aq,asseen in thesilicate-richbandsandzonesintheferhythmiteandfelutite(Figure8);thisisbasedontheanticipationthatFe2+silicateswillhavelittleisoto-picfractionationrelativetoFe2+

aq(Polyakov&Mineev,2000).AnimportantfindinginthisstudyisthedistinctFeisotopecompo-

sitionsoftheoncoidalGIF(Table2andFigure8),whichwasformedintheshallowestsettingoftheunitsstudied.Thepositiveδ56FevaluesoftheoncoidssuggestthattheFewhichprecipitatedinshallow,waveagitatedconditions,hadadifferentsourceorpathwaythanthefelutiteandferhythmite.IftheironintheoncoidalGIFwassourcedfromthe

samehydrothermalplumeasthedeeperwaterferhythmite,continueddistillationofFe2+

aqintheplumeasitspreadfromthedeeptoshallowshelfshouldhaveresulted inδ56Fevalues intheGIFthatweremorenegativethanintheferhythmite.Incontrasttotheferhythmiteandfe-lutite,however,themagnetiteintheoncoidalGIFhassignificantlyposi-tiveδ56Fevaluesandthecoexistingminnesotaiteδ56Fevaluesareclosetozero.ThissuggeststhattheFe2+

aqsourcefromwhichtheironintheoncoidrimsprecipitatedhadδ56Fevaluesclosetozero(Figure9).TheironintheoncoidalGIFisthus interpretedtobemost likelysourcedfromashallowoceanpoolthathadalowerconcentrationofFe2+

aq and notfromthedeepwaterhydrothermalplume.Wefavourthismodeloverthepossibilitythatintenseupwelling,resultinginpartialoxidationofFe2+,andoffshoreflowoftheevolvedwatermassproducedtheFeisotopevariationsmeasured.ThelattermodelwouldmostlikelyresultintheoppositetrendforFe-silicateFeisotopevaluesmeasured.

Itisimportanttoalsonotethatthemagnetiteδ56FevaluesforGIFarelikelytobeminimumvaluesbecausetheinitialprecipitatewouldhavebeenan ironoxyhydroxide.Later interactionwithFe2+

aq, likelyhavingaδ56Fevalueofaround0‰,wouldhavetendedtodecreasetheδ56Fevaluefromtheinitial ironoxyhydroxide,andasimplemix-ingcalculationindicatesthattheoriginalFeoxyhydroxideprecipitateslikely had an average initial δ56Fe value of about 0.66‰ (c.f. Czajaetal., 2013).Thenear-zeroδ56Fevalueof theminnesotaite-bearingchertbetweentheoncoids,however,suggeststhatthepuremagne-titecomponentintheoncoidrimsprobablyhasslightlyhigherδ56Fevaluesthananalysedforthesmallbulksamples.

A lackofabundantFe2+aq intheshallowpartofthisdepositional

basin, interpreted to have been an isolated submerged shelf bank(Figure9), is supportedby the lowerFe content in theoncoidalGIFwhencomparedtothedeeperwaterIF(Table1),aswellasthestron-gertexturalcontrolonmagnetiteintheGIF,whichonlyoccursinmi-crostromatoliticrimsofoncoids(Figure4).Consistentwiththetexturalandothergeochemicaldata,partialoxidationofa lowconcentrationof Fe2+

aq on the isolated bank, as indicated by significantly positiveδ56FevaluesintheoncoidalGIFFeoxides,alsosuggestsoxidationina lowoxygenenvironment, possiblyby chemolithoautotrophsormi-croaerobicanoxygenicphotosyntheticironoxidizers(APIO)ratherthanbyreactionwithfreeoxygen.Czajaetal. (2013)modelledFeisotopefractionationduringoxidationbyfreeoxygenandbyAPIOandarguedthatlowamountsofironthathassignificantlypositiveδ56FevaluesarebestexplainedbyAPIOratherthanreactionwithmolecularoxygen.Itisimportanttonote,however,thatthepossiblepresenceofnitrate,asinferredfortheformationofcalciteinthemineralparagenesissectionabove,could indicatethattheremayhavebeensomefreeoxygenintheenvironment (i.e.,microaerobicconditions). Inaddition, theenvi-ronmentinwhichthegranulesformedmusthavebeenhighlydepletedinironasessentiallyonlychertwasprecipitatedandtheFe-richmin-eralsareessentiallyconfinedtotheoncoidrims(Figure4),whichisinstarkcontrasttopervasiveFemineralizationinthedeeperwaterMIF(Figure6).IntheIFsoftheWitwatersrandSupergroup,whichiscorrela-tivetothoseoftheMozaanGroup,thereisalsoanotableabsenceofFeinunitsdepositedabovewavebaseandtheinferredbaseofthephoticzone(Smithetal.,2013).ThisimpliestheremovalofhydrothermalFe2+

     |  17SMITH eT al.

indeeperwaterfaciesbychemolithoautotrophsundermicro-oxiccon-ditions(Figure9).ChemolithoautotrophyinsteadofAPIOcouldthere-forehavebeenresponsiblefortheprecipitationofFeintheGIF.

6  | CONCLUSION

We describe the oldest knownwell-preserved oncoidal GIF, whichtextural and compositional data indicate likely formed throughbiologicalactivityintheshallowwaterenvironmentsofthePongolaSupergroupasearlyas3.0–2.8Ga.Initialironoxidesareproposedtohave formed by iron-oxidizing bacteria as part of layeredmicrobialfilms that were developed in a shallow water wave-dominatedenvironment.Evolutionof these filmscontinued intoshapessimilartodomalmicrostromatolites.Thelowerironcontent,strongtexturalcontrolonironmineralizationandthestableFeisotopesystematicsoftheoncoidalGIF,whencomparedtothedeeperwaterIFs,illustratethatthereweredifferentpathwayssourcesfortheironinthedeeperandshallowerocean.OxidationofFe2+

aqinthedeepwaterevidentlytook place in the presence of abundant iron, as suggested by thewealthofiron-bearingmineralthroughouttheIFs.Microbeslivingintheshallowwaterhadamuchmore limitedsourceof iron,and theironwasonlyprecipitatedinthebacterialfilamentsthatgrewaroundchert granules thatwere formed bywave reworking of chert bedson an isolated shallow submerged shelf bank, developed locally onapalaeo-high, intheoriginalveryextensiveWitwatersrand-Mozaandepository.

ACKNOWLEDGMENTS

TheauthorswishtothankTheDepartmentofGeologyattheUniversityof Johannesburg, the Palaeoproterozoic Mineralization ResearchGroup(PPM),andtheDepartmentofScienceandTechnology(DST)andNationalResearchFoundation(NRF)fundedCentreofExcellencefor IntegratedMineral and Energy Resource Analysis (CIMERA) fortheirfundingandsupport.TheNASAAstrobiologyInstitutearealsothanked for funding. Michael Bau at Jacobs University Bremen isthankedfor theuseofhisgeochemical laboratoryforREEanalyses.ChrisHarrisat theUniversityofCapeTown isalso thanked for theuseofhisstable isotope laboratory.AngloGold-Ashantiare thankedfor access to their exploration drill core on which this study wasconducted. Dominic Papineau, an anonymous reviewer and KurtKonhauserarethankedfortheircommentsthataddedgreatvaluetothiscontribution.

REFERENCES

Ader,M.,Thomazo,C.,Sansjofre,P.,Busigny,V.,Papineau,D.,Laffont,R.,…Halverson,G.P.(2016).Interpretationofthenitrogenisotopiccompo-sitionsofPrecambriansedimentaryrocks:Assumptionsandperspec-tives.Chemical Geology,429,93–110.

Al-Aasm,I.S.,Taylor,B.E.,&South,B. (1990).Stable isotopeanalysisofmultiple carbonate samples using selective acid extraction.Chemical Geology (Isotope Geoscience Section),80,119–125.

Albarède,F.,&Beard,B.(2004).Analyticalmethodsfornon-traditionaliso-topes.Reviews in Mineralogy and Geochemistry,55,113–152.

Alshuaibi,A.,Duane,M.J.,&Mahmoud,H.(2012).Microbial-activatedsed-imenttrapsassociatedwithoncoliteformationalongaperitidalbeach,northernArabian (Persian)Gulf, Kuwait.Geomicrobiology Journal,29,679–696.

Barghoorn,E.S.,&Tyler,S.A. (1965).MicroorganismsfromtheGunflintChert:ThesestructurallypreservedPrecambrianfossilsfromOntarioarethemostancientorganismsknown.Science,147,563–575.

Bau, M., & Alexander, B.W. (2009). Distribution of high field strengthelements (Y, Zr, REE,Hf,Ta,Th,U) in adjacentmagnetite and chertbandsandinreferencestandardsFeR-3 and FeR-4fromtheTemagamiiron-formation,Canada,andtheredoxleveloftheNeoarcheanocean.Precambrian Research,174,337–346.

Bau,M.,&Dulski,P.(1996).Distributionofyttriumandrare-earthelementsin the Penge and Kuruman iron-formations, Transvaal Supergroup,SouthAfrica.Precambrian Research,79,37–55.

Beard,B.L.,&Johnson,C.M.(1999).Highprecisionisotopemeasurementsofterrestrialandlunarmaterials.Geochimica et Cosmochimica Acta,63,1653–1660.

Beard,B.L.,&Johnson,C.M.(2004).FeisotopevariationsinthemodernandancientEarthandotherplanetarybodies.Reviews in Mineralogy & Geochemistry,55,319–357.

Beard,B.L.,Johnson,C.M.,VonDamm,K.L.,&Poulson,R.L.(2003).Ironisotope constraints on Fe cycling and mass balance in oxygenatedEarthoceans.Geology,31,629–632.

Bekker,A.,Slack,J.F.,Planavsky,N.,Krapež,B.,Hofmann,A.,Konhauser,K.O.,&Rouxel,O.J.(2010).Ironformation:Thesedimentaryproductofacomplexinterplayamongmantle,tectonic,oceanic,andbiosphericprocesses.Economic Geology,105,467–508.

Beukes,N.J.,&Cairncross,B.(1991).Alithostratigraphic-sedimentologicalreference profile for the Late Archaean Mozaan Group, PongolaSequence:Application to sequence stratigraphyandcorrelationwiththeWitwatersrand Supergroup. South African Journal of Geology,94,44–69.

Beukes, N. J., &Gutzmer, J. (2008).Origin and paleoenvironmental sig-nificance of major iron formations at the Archean-Paleoproterozoicboundary. Reviews in Economic Geology,15,5–47.

Beukes,N.J.,&Lowe,D.R.(1989).Environmentalcontrolondiversestro-matolite morphologies in the 3000 Ma Pongola Supergroup, SouthAfrica.Sedimentology,36,383–397.

Brookins,D.G.(1989).Aqueousgeochemistryofrare-earthelements.InB.R.Lipin&G.A.McKay(Eds.),Reviews in mineralogy vol. 21: Geochemistry and mineralogy of rare-earth elements (201–225). Washington, DC:MineralogicalSocietyofAmerica.

Bullen,T.D.,White,A.F.,Childs,C.W.,Vivit,D.V.,&Schulz,M.S.(2001).Demonstrationof significantabiotic iron isotope fractionation inna-ture.Geology,29,699–702.

Castanier, S., LeMétayer-Levrel, G., & Perthuisor, J.-P. (2000). Bacterialrolesintheprecipitationofcarbonateminerals.InR.E.Riding&S.M.Awramik(Eds.),Microbial sediments(pp.32–39).Berlin:Springer-Verlag.

Cloud,P.(1973).Paleoecologicalsignificanceofthebandediron-formation.Economic Geology,68,1135–1143.

Clout, J.M. F., & Simonson, B.M. (2005). Precambrian iron formationsand ironformation-hosted ironoredeposits.Economic geology 100th anniversary volume (pp. 643–679). Littleton: Society of EconomicGeologists.

Croal, L. R., Johnson,C.M., Beard, B. L., &Newman,D. K. (2004). Ironisotope fractionation by Fe(II)-oxidizing photoautotrophic bacteria.Geochimica et Cosmochimica Acta,68,1227–1242.

Crowe,S.A.,Døssing,L.N.,Beukes,N.J.,Bau,M.,Kruger,S.J.,Frei,R.,&Canfield,D.E.(2013).Atmosphericoxygenationthreebillionyearsago.Nature,501,535–538.

Czaja,A.D.,Johnson,C.M.,Beard,B.L.,Roden,E.E.,Li,W.,&Moorbath,S.(2013).BiologicalFeoxidationcontrolleddepositionofbandediron

18  |     SMITH eT al.

formationintheca.3,770MaIsuaSupracrustalBelt(WestGreenland).Earth and Planetary Science Letters,363,192–203.

Czaja,A.D.,Johnson,C.M.,Roden,E.E.,Beard,B.L.,Voegelin,A.R.,Nagler,T.F.,…Wille,M.(2012).EvidenceforfreeoxygenintheNeoarcheanocean based on coupled iron–molybdenum isotope fractionation.Geochimica et Cosmochimica Acta,86,118–137.

DesMarais, D. J. (2001). Isotopic evolution of the biogeochemical car-boncycleduringthePrecambrian.InJ.W.Valley,&D.R.Cole(Eds.),Reviews in mineralogy and geochemistry: Stable isotope geochemistry(pp.555–578). Washington, DC: Mineralogical Society of America andGeochemicalSociety.

Dodd,M.S.,Papineau,D.,Grenne,T.,Slack,J.F.,Rittner,M.,Pirajno,F.,…Little,C.T.S.(2017).EvidenceforearlylifeinEarth’soldesthydrother-malventprecipitates.Nature,543,60–64.

Dorland,H.C.(1999).Paleoproterozoic laterites, red beds and ironstones of the Pretoria Group with reference to the history of atmospheric oxygen. MScdissertation,DepartmentofGeology,RandAfrikaansUniversity,Johannesburg,147p.

Dymek,R.F.,&Klein,C.(1988).Chemistry,petrologyandoriginofbandediron-formation lithologies from the 3800Ma Isua supracrustal belt,WestGreenland.Precambrian Research,39,247–302.

Emerson,D.,&Revsbech,N.P. (1994). Investigationofan iron-oxidizingmicrobialmatcommunitylocatednearAarhus,Denmark:Fieldstudies.Applied and Environmental Microbiology,60,4022–4031.

Ferris,F.G.(2000).Microbe-metalinteractionsinsediments.InR.E.Riding,& S. M. Awramik (Eds.),Microbial sediments (pp. 121–126). Berlin:Springer-Verlag.

Fortin,D., Ferris, F.G.,&Beveridge,T.J. (1997). Surface-mediatedmin-eral development by bacteria. In J. F. Banfield, & K. F. Nealson(Eds.), Geomicrobiology: Interactions between microbes and minerals: Mineralogical Society of America Reviews in Mineralogy(Vol.35,pp.161–180).Washington,DC:MineralogicalSocietyofAmerica.

Fryer,B.J.(1983).Rareearthelementsiniron-formation.InA.F.Trendall,&R.C.Morris(Eds.),Iron-formation: Facts and problems(pp.345–358).Amsterdam:Elsevier.

Garzanti,E.,Haas,R.,&Jadoul,F.(1989).IronstonesintheMesozoicpas-sivemarginsequenceoftheTethysHimalaya(Zanskar,NorthernIndia):Sedimentologyandmetamorphim.InT.P.Young,&W.E.G.Taylor(Eds.),Phanerozoic ironstones: Geological Society [London] Special Publication (Vol.46,pp.229–245).London:GeologicalSocietyofLondon.

Gehring, A. U. (1989). The formation of goethitic ooids in condensedJurassicdepositsofnorthernSwitzerland. InT.P.Young,&W.E.G.Taylor(Eds.),Phanerozoic ironstones: Geological Society [London] Special Publication (Vol. 46, pp. 133–139). London: Geological Society ofLondon.

Godfrey, L.V.,&Falkowski, P.G. (2009).The cycling and redox stateofnitrogenintheArchaeanocean.Nature Geoscience,2,725–729.

Gutzmer,J.,Beukes,N.J.,Pickard,N.J.,&Barley,M.E. (1999).SHRIMPageofaquartzporphyrysill in theMozaangroup:GeochronologicalimplicationforthePongolaandtheWitwatersrandSupergroups.South African Journal of Geology,102,139–146.

Halevy, I., Alesker, M., Schuster, E. M., Popvitz-Biro, R., & Feldman, Y.(2017).Akey role forgreen rust in thePrecambrianoceansand thegenesisofironformations.Nature Geoscience,10,135–139.

Hegner,E.,Kröner,A.,&Hunt,P.(1994).ApreciseU-PbzirconagefortheArcheanPongolaSupergroupvolcanicsinSwaziland.Journal of African Earth Science,18,139–141.

Heimann, A., Johnson, C. M., Beard, B. L., Valley, J. W., Roden, E. E.,Spicuzza,M.J.,&Beukes,N.J.(2010).Fe,C,andOisotopecomposi-tionsofbandedironformationcarbonatesdemonstrateamajorrolefordissimilatoryironreductionin~2.5Gamarineenvironments.Earth and Planetary Science Letters,294,8–18.

Hofmann,H.J. (2000).Archeanstromatolitesasmicrobialarchives. InR.E.Riding,&S.M.Awramik (Eds.),Microbial sediments (pp.315–327).BerlinHeidelberg:Springer-Verlag.

Holland,H.D.(1984).The chemical evolution of the atmosphere and oceans. 598p.Princeton,NJ:PrincetonUniversityPress.

Johnson, C. M., Beard, B. L., Beukes, N. J., Klein, C., & O’Leary, J. M.(2003).Ancientgeochemicalcycling in theEarthas inferred fromFeisotopestudiesofbandedironformationsfromtheTransvaalCraton.Contributions to Mineralogy and Petrology,144,523–547.

Johnson, C. M., Beard, B. L., Klein, C., Beukes, N. J., & Roden, E. E.(2008a). Iron isotope constrain biologic and abiologic processes inbandedironformationgenesis.Geochimica et Cosmochimica Acta,72,151–169.

Johnson,C.M.,Beard,B.L.,&Roden,E.E. (2008b).The iron isotopefingerprints of redox and biogeochemical cycling in modern andancient Earth. Annual Review of Earth and Planetary Sciences, 36,457–493.

Kappler,A.,Johnson,C.M.,Crosby,H.A.,Beard,B.L.,&Newman,D.K.(2010).Evidenceforequilibriumironisotopefractionationbynitrate-reducing iron(II)-oxidizingbacteria.Geochimica et Cosmochimica Acta,74,2826–2842.

Kappler, A., Pasquero, C., Konhauser, K. O., & Newman, D. K. (2005).Deposition of banded iron formations by anoxygenic phototrophicFe(II)-oxidizingbacteria.Geology,33,865–868.

Kaufman,A.J. (1996).Geochemical andmineralogical effectsof contactmetamorphism on banded iron-formations: An example from theTransvaalBasin,SouthAfrica.Precambrian Research,79,171–194.

Kearsley,A. T. (1989). Iron-rich ooids, theirmineralogy andmicrofabric:Cluestotheiroriginandevolution.InT.P.Young,&W.E.G.Taylor(Eds.),Phanerozoic ironstones: Geological Society [London] Special Publication (Vol.46,pp.141–164).London:GeologicalSocietyofLondon..

Klein, C. (2005). Some Precambrian banded iron-formations (BIFs) fromaround theworld:Their age, geologic setting,mineralogy,metamor-phism,geochemistry,andorigin.American Mineralogist,90,1473–1499.

Klein,C.,&Beukes,N.J.(1992).Time,distribution,stratigraphy,andsedi-mentologicsetting,andgeochemistryofPrecambrianironformations.InJ.W.Schopf,&C.Klein(Eds.),The proterozoic biosphere: A multidis-ciplinary study(pp.139–146).NewYork:CambridgeUniversityPress.

Klein,C.,&Beukes,N.J.(1993).Proterozoiciron-formations.InK.Condie(Ed.),Proterozoic crustal evolution(pp.383–418).Amsterdam:Elsevier.

Knoll, A. H., & Simonson, B. (1981). Early proterozoic microfossils andpenecontemporaneous quartz cementation in the sokoman iron for-mation,Canada.Science,211,478–480.

Konhauser, K.O. (1998).Diversity of bacterial ironmineralization.Earth Science Reviews,43,91–121.

Konhauser,K.O.(2007).Introduction to geomicrobiology.Malden:Blackwell.Konhauser,K.O.,&Ferris,F.G.(1996).Diversityofironandsilicaprecipi-

tationbymicrobialmatsinhydrothermalwaters,Iceland:ImplicationsforPrecambrianironformations.Geology,24,323–326.

Konhauser, K. O., Hamade, T., Raiswell, R., Morris, R. C., Ferris, F. G.,Southam,G.,&Canfield,D.E.(2002).CouldbacteriahaveformedthePrecambrianbandedironformations?Geology,30,1079–1082.

Laufer,K.,Nordhoff,M.,Røy,H.,Schmidt,C.,Behrens,S.,Jørgensen,B.B.,&Kappler,A.(2016).Coexistenceofmicroaerophilic,nitrate-reducing,andphototrophicFe(II)oxidizersandFe(III)reducersincoastalmarinesediment.Applied and Environmental Microbiology,82,1433–1447.

Li,W.,Huberty,J.M.,Beard,B.L.,Kita,N.T.,Valley,J.W.,&Johnson,C.M.(2013).Contrastingbehaviorofoxygenandironisotopesinbandedironformationsrevealedbyin situisotopicanalysis.Earth and Planetary Science Letters,384,132–143.

Maliva, R.G., Knoll,A.H.,& Simonson,B.M. (2005). Secular change inthePrecambriansilicacycle:Insightsfromchertpetrology.Geological Society of America Bulletin,117,835–845.

McCollom,T.M.(2003).Formationofmeteoritehydrocarbonsfromther-mal decomposition of siderite (FeCO3).Geochimica et Cosmochimica Acta,67,311–317.

Mukasa,S.B.,Wilson,A.H.,&Young,K.R.(2013).Geochronologicalcon-straints on themagmatic and tectonic development of the Pongola

     |  19SMITH eT al.

Supergroup(CentralRegion),SouthAfrica.Precambrian Research,224,268–286.

Nguetchoua, G., & Giresse, P. (2010). Sand bodies and incised valleyswithintheLateQuarternarySanaga-Nyongdeltacomplexonthemid-dlecontinentalshelfofCameroon.Marine and Petroleum Geology,27,2173–2188.

Nhleko, N. (2003). The Pongola Supergroup in Swaziland. PhD thesis,DepartmentofGeology,RandAfrikaansUniversity,Johannesburg,299p.

Noffke,N.,Beukes,N.,Bower,D.,Hazen,R.M.,&Swift,D.J.P. (2008).An actualistic perspective into Archean worlds – (cyano-)bacteriallyinduced sedimentary structures in the siliciclasticNhlazatse Section,2.9GaPongolaSupergroup,SouthAfrica.Geobiology,6,5–20.

Noffke,N.,Hazen,R.,&Nhleko,N.(2003).Earth’searliestmicrobialmatsinasiliciclasticmarineenvironment(2.9GaMozaanGroup,SouthAfrica).Geology,31,673–676.

Ohmoto,H. (2003).Nonredox transformations ofmagnetite-hematite inhydrothermalsystems.Economic Geology,98,157–161.

Percak-Dennett,E.M.,Beard,B.L.,Xu,H.,Konishi,H.,Johnson,C.M.,&Roden,E.E.(2011).Ironisotopefractionationduringmicrobialdissimi-latoryironoxidereductioninsimulatedArchaeanseawater.Geobiology,9,205–220.

Peryt,T.M. (1983a).Classificationofcoatedgrains. InT.M.Peryt (Ed.),Coated grains(pp.3–6).Berlin:Springer-Verlag.

Peryt, T. M. (1983b). Oncoids: Comment on recent develop-ments. InT.M. Peryt (Ed.),Coated grains (pp. 273–275). Berlin:Springer-Verlag.

Planavsky, N. J., Asael, D., Hofmann, A., Reinhard, C. T., Lalonde, S. V.,Knudsen,A.,…Rouxel,O.J.(2014).Evidenceforoxygenicphotosyn-thesis half a billion years before the Great Oxidation Event.Nature Geoscience,7,283–286.

Planavsky,N.,Bekker,A.,Rouxel,O.J.,Kamber,B.,Hofmann,A.,Knudsen,A.,&Lyons,T.W.(2010).RareEarthElementandyttriumcompositionsofArcheanandPaleoproterozoicFeformationsrevisited:Newperspec-tivesonthesignificanceandmechanismsofdeposition.Geochimica et Cosmochimica Acta,74,6387–6405.

Planavsky, N., Rouxel, O. J., Bekker, A., Hofmann,A., Little, C. T. S., &Lyons,T.W. (2012). Iron isotopecompositionof someArcheanandProterozoic iron formations. Geochimica et Cosmochimica Acta, 80,159–169.

Planavsky,N., Rouxel,O., Bekker,A., Shapiro, R., Fralick, P., &Knudsen,A. (2009). Iron-oxidizing microbial ecosystems thrived in latePaleoproterozoic redox stratified oceans.Earth and Planetary Science Letters,286,230–242.

Polyakov,V.B.,&Mineev,S.D. (2000).TheuseofMossbauerspectros-copyinstable isotopegeochemistry.Geochimica et Cosmochima Acta,64,849–865.

Posamentier, H. W., Allen, G. P., James, D. P., & Tesson, M. (1992).Forced regressions in a sequence stratigraphic framework: con-cepts, examples, and exploration significance. AAPG Bulletin, 76,1687–1709.

Posth,N.R.,Hegler,F.,Konhauser,K.O.,&Kappler,A.(2008).AlternatingSiandFedepositioncausedbytemperaturefluctuationsinPrecambrianoceans.Nature Geoscience,1,703–708.

Rasmussen,B.,Krapež,B.,Muhling,J.R.,&Suvorova,A.(2015).PrecipitationofironsilicatenanoparticlesinearlyPrecambrianoceansmarksEarth’sfirstironage.Geology,43,303–306.

Richter,D.K. (1983).Classificationof coatedgrains,discussion. InT.M.Peryt(Ed.),Coated grains(pp.7–8).Berlin:Springer-Verlag.

Roden,E.,&Lovley,D.(1993).DissimilatoryFe(II)reductionbythemarinemicroorganismDesulfuromonas acetoxidans. Applied and Environmental Microbiology,59,734–742.

Satkoski,A.M.,Beukes,N.J.,Li,W.,Beard,B.L.,&Johnson,C.M.(2015).Aredox-stratifiedocean3.2billionyearsago.Earth and Planetary Science Letters,430,43–53.

Schidlowski, M. (1995). Isotope fractionations in the terrestrial carboncycle:Abriefoverview.Advances in Space Research,15,441–449.

Simonson,B.M.,&Goode,A.D.T.(1989).FirstdiscoveryofferruginouschertarenitesintheearlyPrecambrianHamersleyGroupofWesternAustralia.Geology,17,269–272.

Slack,J.F.,Grenne,T.,Bekker,A.,Rouxel,O.J.,&Lindberg,P.A.(2007).SuboxicdeepseawaterinthelatePaleoproterozoic:Evidencefromhematiticchertandironformationrelatedtoseafloor-hydrothermalsulfidedeposits,cen-tralArizona,USA.Earth and Planetary Science Letter,255,243–256.

Smith,A.J.B. (2007).The paleo-environmental significance of the iron-for-mations and iron-rich mudstones of the Mesoarchean Witwatersrand-Mozaan Basin, South Africa.MScdissertation,DepartmentofGeology,UniversityofJohannesburg,Johannesburg,208p.

Smith,A.J.B.,Beukes,N.J.,&Gutzmer,J. (2013).Thecompositionanddepositional environments of Mesoarchean iron formations of theWest Rand Group of theWitwatersrand Supergroup, South Africa.Economic Geology,108,111–134.

Smith,A.J.B.,Beukes,N.J.,Gutzmer,J.,Johnson,C.M.,&Czaja,A.D.(2012). Iron isotope fractionation in stromatolitic oncoidal iron for-mation, Mesoarchean Witwatersrand-Mozaan Basin, South Africa.Mineralogical Magazine,76,2384.

Stookey, L. L. (1970). Ferrozine - a new spectroscopic reagent for iron.Analytical Chemistry,42,779–781.

Stüeken,E.E.,Kipp,M.A.,Koehler,M.C.,&Buick,R.(2016).TheevolutionofEarth’s biogeochemical nitrogen cycle.Earth- Science Reviews,160,220–239.

Thomazo,C.,&Papineau,D.(2013).BiogeochemicalcyclingofnitrogenontheearlyEarth.Elements,9,345–351.

Viehmann,S.,Bau,M.,Hoffmann,J.E.,&Münker,C.(2015).GeochemistryoftheKrivoyRogBandedIronFormation,Ukraine,andtheimpactofpeakepisodesof increasedglobalmagmaticactivityon the traceel-ement composition of Precambrian seawater. Precambrian Research,270,165–180.

Warren,L.A.,&Ferris,F.G.(1998).Continuumbetweensorptionandpre-cipitation of Fe(III) on microbial surfaces. Environmental Science and Technology,32,2331–2337.

Wilson,A.,Groenewald,B.,&Palmer,C.(2013).Volcanicandvolcaniclasticrocksof theMesoarchaeanPongolaSupergroup inSouthAfrica andSwaziland:Distribution,physicalcharacteristics,stratigraphyandcor-relations.South African Journal of Geology,116,119–168.

Wronkiewicz,D.J.,&Condie,K.C.(1989).GeochemistryandprovenanceofsedimentsfromthePongolaSupergroup,SouthAfrica:Evidencefora3.0-Ga-oldcontinentalcraton.Geochimica et Cosmochimica Acta,53,1537–1549.

Wu, L., Percak-Dennett, E.M., Beard, B. L., Roden, E. E., & Johnson, C.M. (2012). Stable iron isotope fractionation between aqueous Fe(II)and model Archaean ocean Fe–Si coprecipitates and implicationsfor iron isotope variations in the ancient rock record.Geochimica et Cosmochimica Acta,84,14–28.

Zheng, X.-Y., Beard, B. L., Reddy, T. R., Roden, E. E., & Johnson, C. M.(2016).AbiologicsiliconisotopefractionationbetweenaqueousSiandFe(III)-SigelinsimulatedArchaeanseawater:ImplicationsforSiisotoperecordsinPrecambriansedimentaryrocks.Geochimica et Cosmochimica Acta,187,102–122.

How to cite this article:SmithAJB,BeukesNJ,GutzmerJ,CzajaAD,JohnsonCM,NhlekoN.OncoidalgranularironformationintheMesoarchaeanPongolaSupergroup,southernAfrica:Texturalandgeochemicalevidenceforbiologicalactivityduringirondeposition.Geobiology. 2017;00:1–19. https://doi.org/10.1111/gbi.12248