The age of Homo naledi and associated sediments in the Rising Star Cave, South Africa


previous item | next item

Citation
The age of Homo naledi and associated sediments in the Rising Star Cave, South Africa

Material Information

Title:
The age of Homo naledi and associated sediments in the Rising Star Cave, South Africa
Series Title:
Evolutionary Biology
Creator:
Dirks, Paul H. G. M.
Roberts, Eric M.
Hilbert-Wolf, Hannah
Kramers, Jan D.
Hawks, John
Dosseto, Anthony
Duval, Mathieu
Elliot, Marina
Evans, Mary
Publisher:
eLife Sciences
Publication Date:
Language:
English

Subjects

Subjects / Keywords:
Research Article ( local )
Evolutionary Biology ( local )
Homo Naledi ( local )
Hominin ( local )
Dinaledi Chamber ( local )
Dating ( local )
Paleoanthropology ( local )
Pleistocene ( local )
Genre:
serial ( sobekcm )

Notes

Abstract:
New ages for flowstone, sediments and fossil bones from the Dinaledi Chamber are presented. We combined optically stimulated luminescence dating of sediments with U-Th and palaeomagnetic analyses of flowstones to establish that all sediments containing Homo naledi fossils can be allocated to a single stratigraphic entity (sub-unit 3b), interpreted to be deposited between 236 ka and 414 ka. This result has been confirmed independently by dating three H. naledi teeth with combined U-series and electron spin resonance (US-ESR) dating. Two dating scenarios for the fossils were tested by varying the assumed levels of 222Rn loss in the encasing sediments: a maximum age scenario provides an average age for the two least altered fossil teeth of 253 +82/–70 ka, whilst a minimum age scenario yields an average age of 200 +70/–61 ka. We consider the maximum age scenario to more closely reflect conditions in the cave, and therefore, the true age of the fossils. By combining the US-ESR maximum age estimate obtained from the teeth, with the U-Th age for the oldest flowstone overlying Homo naledi fossils, we have constrained the depositional age of Homo naledi to a period between 236 ka and 335 ka. These age results demonstrate that a morphologically primitive hominin, Homo naledi, survived into the later parts of the Pleistocene in Africa, and indicate a much younger age for the Homo naledi fossils than have previously been hypothesized based on their morphology.

Record Information

Source Institution:
University of South Florida Library
Holding Location:
University of South Florida
Rights Management:
This item is licensed with the Creative Commons Attribution License. This license lets others distribute, remix, tweak, and build upon this work, even commercially, as long as they credit the author for the original creation.
Resource Identifier:
K26-05126 ( USFLDC: LOCAL DOI )
k26.5126 ( USFLDC: LOCAL Handle )

USFLDC Membership

Aggregations:
University of South Florida
Karst Information Portal

Postcard Information

Format:
serial

Downloads

This item is only available as the following downloads:


Full Text

PAGE 1

*Forcorrespondence: paul.dirks@ jcu.edu.auCompetinginterests: The authorsdeclarethatnocompetinginterestsexist.Funding: Seepage54 Received: 13December2016 Accepted: 25April2017 Published: 09May2017 Reviewingeditor: GeorgeH Perry,PennsylvaniaStateUniversity,UnitedStates CopyrightDirksetal.This articleisdistributedunderthetermsofthe CreativeCommons AttributionLicense, which permitsunrestricteduseandredistributionprovidedthattheoriginalauthorandsourcearecredited. Theageof Homonaledi andassociated sedimentsintheRisingStarCave,SouthAfrica PaulHGMDirks1,2*, EricMRoberts1,2,HannahHilbert-Wolf1,JanDKramers3, JohnHawks2,4,AnthonyDosseto5,MathieuDuval6,7,MarinaElliott2,MaryEvans8, RainerGru È n6,9,JohnHellstrom10,AndyIRHerries11,RenaudJoannes-Boyau12, TebogoVMakhubela3,ChristaJPlaczek1,JessieRobbins1,CarlSpandler1, JelleWiersma1,JonWoodhead10,LeeRBerger21DepartmentofGeoscience,JamesCookUniversity,Townsville,Australia;2EvolutionaryStudiesInstituteandtheNationalCentreforExcellencein PalaeoSciences,UniversityoftheWitwatersrand,Wits,SouthAfrica;3Department ofGeology,UniversityofJohannesburg,Johannesburg,SouthAfrica;4Department ofAnthropology,UniversityofWisconsin-Madison,Madison,UnitedStates;5School ofEarthandEnvironmentalSciences,UniversityofWollongong,Wollongong,Australia;6AustralianResearchCentreforHumanEvolution,EnvironmentalFutures ResearchInstitute,GriffithUniversity,Nathan,Australia;7Geochronology,Centro NacionaldeInvestigacioÂnsobrelaEvolucioÂnHumana(CENIEH),Burgos,Spain;8SchoolofGeosciences,UniversityoftheWitwatersrand,Wits,SouthAfrica;9ResearchSchoolofEarthSciences,TheAustralianNationalUniversity,Canberra, Australia;10SchoolofEarthSciences,TheUniversityofMelbourne,Parkville, Australia;11TheAustralianArchaeomagnetismLaboratory,Departmentof ArchaeologyandHistory,LaTrobeUniversity,Melbourne,Australia;12GeoarchaeologyandArchaeometryResearchGroup,DepartmentofGeoScience, SouthernCrossUniversity,Lismore,Australia Abstract Newagesforflowstone,sedimentsandfossilbonesfromtheDinalediChamberare presented.WecombinedopticallystimulatedluminescencedatingofsedimentswithU-Thandpalaeomagneticanalysesofflowstonestoestablishthatallsedimentscontaining Homonaledi fossilscanbeallocatedtoasinglestratigraphicentity(sub-unit3b),interpretedtobedepositedbetween236kaand414ka.Thisresulthasbeenconfirmedindependentlybydatingthree H. naledi teethwithcombinedU-seriesandelectronspinresonance(US-ESR)dating.Twodating scenariosforthefossilsweretestedbyvaryingtheassumedlevelsof222Rnlossintheencasing sediments:amaximumagescenarioprovidesanaverageageforthetwoleastalteredfossilteethof253+82/±70ka,whilstaminimumagescenarioyieldsanaverageageof200+70/±61ka.Weconsiderthemaximumagescenariotomorecloselyreflectconditionsinthecave,andtherefore,thetrueageofthefossils.BycombiningtheUS-ESRmaximumageestimateobtainedfromtheteeth,withtheU-Thagefortheoldestflowstoneoverlying Homonaledi fossils,wehave constrainedthedepositionalageof Homonaledi toaperiodbetween236kaand335ka.These ageresultsdemonstratethatamorphologicallyprimitivehominin, Homonaledi, survivedintothe laterpartsofthePleistoceneinAfrica,andindicateamuchyoungerageforthe Homonaledi fossils thanhavepreviouslybeenhypothesizedbasedontheirmorphology. DOI:10.7554/eLife.24231.001 Dirks etal .eLife2017;6:e24231. DOI:10.7554/eLife.24231 1of59 RESEARCHARTICLE

PAGE 2

Introduction Thefossilassemblageattributedto Homonaledi fromtheRisingStarCaveintheCradleofHumankind,UNESCOWorldHeritageArea,SouthAfrica(CoH)( Bergeretal.,2015 ),representsoneofthe richestandmostunusualtaphonomicassemblagesyetdiscoveredinthehomininfossilrecord( Dirksetal.,2015 ).TheremainsareexceptionallywellpreservedandrepresentthelargestcollectionoffossilsfromasingleprimitivehomininspecieseverdiscoveredinAfrica.The H.naledi fossils occurwithoutadirectassociationwithnon-homininmacrofossilremains,andarefounddeepinsidethedifficulttoaccessU.W.101-DinalediChamber( Dirksetal.,2015 ).TheDinalediChamberischaracterisedbyasedimentaryenvironmentthatisgeochemicallyandsedimentologicallydistinctfromtherestoftheRisingStarCave( Dirksetal.,2015 ),andthefossiliferousdeposititcontainsisprofoundlydifferentfromotherknownhominin-bearingcaveassemblagesintheCoH(e.g., ReynoldsandKibii,2011 ; Dirksetal.,2010 ; Pickeringetal.,2011a ; DirksandBerger,2013 ; Bruxellesetal.,2014 ).Thefossilsoccurasadenseboneaccumulationinmostlyunconsolidated muddysedimentthatlargelyoriginatedfromwithinthecavethroughweatheringofthedolomitehostrock( Dirksetal.,2015 ).Thefossilshavenotbeendateduntilnow. Inthispaperwepresentresultsofuranium-thorium(U-Th)disequilibrium,electronspinresonance (ESR),radiocarbon,andopticallystimulatedluminescence(OSL)datingincombinationwithpalaeo-magneticanalyses,toprovideagesforthefossilsandsurroundingdepositsintheDinalediChamber,andbuilduponthegeologicalcontextdescribedin Dirksetal.(2015) .DatesacquiredviaU-Thand ESRtechniqueswereobtainedusingadoubleblindapproachforeachtechniquetoensurerobust, eLifedigest Speciesofancienthumansandtheextinctrelativesofourancestorsaretypically describedfromalimitednumberoffossils.However,thiswasnotthecasewith Homonaledi .More than1500fossilsrepresentingatleast15individualsofthisspecieswereunearthedfromtheRisingStarcavesysteminSouthAfricabetween2013and2014.FounddeepundergroundintheDinalediChamber,the H.naledi fossilsarethelargestcollectionofasinglespeciesofanancienthumanrelativediscoveredinAfrica. Afterthediscoverywasreported,anumberofquestionsstillremained.Notleastamongthese questionswas:howoldwerethefossils?Thematerialwasundated,andpredictionsrangedfromanywherebetween2millionyearsoldand100,000yearsold. H.naledi sharedseveraltraitswiththe mostprimitiveofourancientrelatives,includingitssmallbrain.Asaresult,manyscientistsguessedthat H.naledi wasanoldspeciesinourfamilytree,andpossiblyoneoftheearliestspeciestoevolve inthegenus Homo . Now,Dirksetal.±whoincludemanyoftheresearcherswhowereinvolvedinthediscoveryof H. naledi ±reportthatthefossilsaremostlikelybetween236,000and335,000yearsold.Thesedates arebasedonmeasuringtheconcentrationofradioactiveelements,andthedamagecausedbytheseelements(whichaccumulatesovertime),inthreefossilizedteeth,plussurroundingrockandsedimentsfromthecavechamber.Importantly,themostcrucialtestswerecarriedoutatindependentlaboratoriesaroundtheworld,andthescientistsconductedthetestswithoutknowingtheresultsoftheotherlaboratories.Dirksetal.tooktheseextrastepstomakesurethattheresultsobtainedwerereproducibleandunbiased. Theestimateddatesaremuchmorerecentthanmanyhadpredicted,andmeanthat H.naledi wasaliveatthesametimeastheearliestmembersofourownspecies±whichmostlikelyevolvedbetween300,000and200,000yearsago.Thesenewfindingsdemonstratewhyitcanbeunwisetotrytopredicttheageofafossilbasedonlyonitsappearance,andemphasizetheimportanceofdatingspecimensviaindependenttests.Finallyintworelatedreports,Bergeretal.suggesthowaprimitive-lookingspecieslike H.naledi survivedmorerecentlythanmanywouldhavepredicted, whileHawksetal.describethediscoveryofmore H.naledi fossilsfromaseparatechamberinthe samecavesystem. DOI:10.7554/eLife.24231.002 Dirks etal .eLife2017;6:e24231. DOI:10.7554/eLife.24231 2of59 Researcharticle GenomicsandEvolutionaryBiology

PAGE 3

reproducibleresults,witheachlaboratoryusingtheirownanalyticalandcomputationalapproach.Approachestakenbyeachlaboratorythatcontributedtothispaperaredescribedindetailinthemethodologysection. TheageofthehomininsintheDinalediChamberhasimplicationsforourunderstandingofthe modeandtempoofthemorphologicalevolutionofhominins( HawksandBerger,2016 ),raising questionsaboutevolutionarystasisandtheroleofrefugia.Theresultschallengeourabilitytoassoci-ategivenhomininspeciestospecificculturesandbehavioursinthepast.Theseissuesarediscussedingreaterdetailinanaccompanyingpaper( Bergeretal.,2017 ). GeologicalsettingThecavesintheCradleofHumankind(CoH),SouthAfricahaveyieldedrichfossilassemblagesoflatePliocenetoearlyPleistoceneage,whichincludearangeofhomininspecies( A.africanus , A.prometheus , A.sediba , P.robustus , H.ergaster,H.naledi andearly Homo )andassociatedmammals, reptiles,andbirds(e.g., Vrba,1975 , 1995 ; Brain,1993 ; Tobias,2000 ; Bergeretal.,2010 , 2015 ).Forthepast3millionyears,hominin-bearingdepositsincavesformedinbroadlysimilarsettings,involvingdebrisconeaccumulationsnearcaveopenings( Partridge,1973 ; Wilkinson,1985 ; Brain,1993 ; Pickeringetal.,2007 ; deRuiteretal.,2009 ; DirksandBerger,2013 ; Herriesand Adams,2013 ; Dirksetal.,2010 , 2016b ; Bruxellesetal.,2014 ; Stratfordetal.,2014 ),with depositscementedbycarbonate-richwatersdrippingfromcaveceilings(e.g., Wilkinson,1985 ; Pickeringetal.,2011b ).IncontrasttoallotherhominindepositsintheCoH,thedepositsthathost H.naledi inRisingStarCavearecomposedoflargelyunconsolidated,mud-clastbrecciainamud matrixwithnoevidenceofcoarseclasticsedimentbeingcarriedinbywaterflow.Thissuggestsadifferentdepositionalregimeandtimingforthesedimentsandthefossils( Dirksetal.,2015 , Dirksetal.,2016a ). RisingStarCaveissituatedintheBloubankRivervalley,2.2kmWofSterkfonteinCave.Thecave systemcomprisesseveralkilometresofmappedpassageways( Figure1a )thatarestratigraphically boundtoa20±30m-thick,chert-poordolomitehorizoncappedbya1±1.3m-thickchertunitthatformstherooftothecavesystem( Dirksetal.,2015 ).Geologicalmappingandlaser-theodolitesurveysindicatethatthisroofisintactandnotpenetratedbysignificantshaftsthatopentosurface( Dirksetal.,2015 ; Krugeretal.,2016 ).Thebroadergeologicalsettingofthecaveisdiscussedin Dirksetal.(2015) ,( Dirksetal.,2016a ). TheDinalediChamber,whichcontainsmostofthefossilsof H.naledi, is ~ 30mbelowsurface and ~ 80minastraightlinefromthenearestpresent-dayopeningtothesurface( Figure1a ).The maincavityformingtheDinalediChamberis ~ 15mlongwithvariablewidthsnotexceeding2.5 meters( Figure1b ),andexpandsneartheintersectionwithacrosscuttingpassage,whichisthelocationofthemainexcavationsitetodate( Figure1b ).Thereisnoevidencethatthepresententrance intotheDinalediChamberhassignificantlychangedsincethedepositionofthefossilhominins,withsedimentaccumulatingmostlynearthecurrentaccesspoint( Dirksetal.,2015 , Dirksetal.,2016a ; Figure2 ).Samplesfordatingwerecollectedfromthevariousflowstonehorizonsandstratigraphic unitsexposedintheDinalediChamber( Figures1b , 2 , 3 , 4 and 5 )aswellasfromfossilmaterial itself( Figures4 , 6 and 7 ). LithologicandstratigraphiccontextfordatingTheDinalediChambercontainsdepositsoffine-grained,muddysedimentsintercalatedwithflow-stonedrapes.Thesedimentsincludevarioustypesoforange,laminatedmudstoneandmudclastbrecciadistributedacrossthreebroadlithostratigraphicunits(Units1,2and3; Dirksetal.,2015 ) thatfilledpartsofthechamberovertime.Basedonvariationsinsedimentcomposition,fossilcon-tentand/orstratigraphicpositionofeachunit,wehavedividedUnit1intosub-units1a,1band1c,andUnit3intosub-units3aand3b,tomorepreciselydefinethestratigraphicpackagestar-getedfordating( Figure2 ).Theunitsareseparatedbyerosionalunconformitiesorflowstone intercalations,butdonotallnecessarilyoccurindirectcontactwithoneanotherduetothecom-plexnatureofcavesasdepositionalsystems(e.g., Brain,1993 ; Martinietal.,2003 ).Inaddition, apartfromsedimentsaccumulatingalongthefloorofthecavechamber,sedimentintheformoforangemuddepositsalsoaccumulatedinsidefracturesandalongledgeshigherupintheDinalediChamber( Figure2b ),whereitformedasaresultofthecombinedeffectofinsituweatheringand Dirks etal .eLife2017;6:e24231. DOI:10.7554/eLife.24231 3of59 Researcharticle GenomicsandEvolutionaryBiology

PAGE 4

Figure1. LocationofRisingStarCaveandtheDinalediChamber.( a )Simplifiedgeologicalmapshowingthe positionoftheRisingStarCave(ingrey);( b )close-upmapoftheDinalediChambershowingthedistributionof thedatingsamples,including:U-Thflowstonesamples(yellowdots,blacktext);ESRsamples(purpledots,orangetext);andOSLsamples(reddots,bluetext).Ageestimatesforthedifferentsamplesareshown,withcrossreferenceto Tables1 , 7 and 8 . DOI:10.7554/eLife.24231.003 Dirks etal .eLife2017;6:e24231. DOI:10.7554/eLife.24231 4of59 Researcharticle GenomicsandEvolutionaryBiology

PAGE 5

Figure2. Geologicalfacemapandcross-sectionsthroughthesedimentpileatdifferentlocationsintheDinaledi Chamber,illustratingtherelationshipsbetweentheflowstonegroupsandsedimentaryunits.Thepositionsofthesectionlinesareshownin( a );afacemapoftheentryzoneoftheDinalediChamber(lookingNE)isshownin( b ); geologicalcross-sectionsthroughthecentralpartoftheDinalediChamberneartheexcavationpitareshownin( c ) and( d ). DOI:10.7554/eLife.24231.004 Dirks etal .eLife2017;6:e24231. DOI:10.7554/eLife.24231 5of59 Researcharticle GenomicsandEvolutionaryBiology

PAGE 6

depositionfromwaterflowingdownfracturesandsidewalls.Allunitsandsub-unitsaretime-trans-gressive,meaningthattheyarelithostratigraphicunitsandnotchronostratigraphicunitsthatoccurinstricttemporalorder.Periodsofsedimentationalternatedwithperiodsoferosion,duringwhichsedimentswereeitherredepositedorremovedfromthechamberviafloordrains,resultinginero-sionalremnantsofallunitsoccurringinavarietyofstratigraphicpositions( Dirksetal.,2015 ). Stalactiteshaveformedatdrippointsalongtheroofandassociatedstalagmitesformedbelow thesepoints.Inoneareabelowtheentrancetothechamber,thesedrippointsrepeatedlyformedflowstoneapronsovercavesedimentsthatdiptowardsthedeeperpartofthechamber.Flowstonealsoformedascascadesandcurtainsthatdevelopedwherewaterseepeddownfracturesandranalongthewallstolocallyspreadout,horizontally,acrossthesedimentscomprisingthecavefloor( Dirksetal.,2015 ).Theflowstoneshavepreliminarilybeensub-dividedintothreegroupsdemarcatingsemi-contemporaneousgenerationsofformation,whichwenamedFlowstoneGroups1,2and3basedontheirappearanceandrelationshipswitheachother,andwiththefloorsedimentsandotherlitho-stratigraphicunitsinthechamber.Inmakingthissubdivisionitwasrealisedthateachgroupofflowstoneswillprobablycomprisearangeofagesrepresentingseparateflowstoneformingevents( Dirksetal.,2015 ),afactborneoutbytheagespresentedbelow( Table1 ). FlowstoneGroup1 (FS1in Table1 ; Figures1b , 2 and 3 )includesremnantsofwhatareinterpretedtobegenerallyolderflowstoneunitsthatwerepartlydissolvedandresorbedtoleavebehindrimsorapronsalongthesidewallsofthecavechamber,somewithsedimentattachedbelowthem.FlowstoneremnantsinterpretedasFlowstoneGroup1aremostlyrestrictedtofivestaggeredrem-nants(Flowstones1a-e),oneabovetheotherinreversestratigraphicorder(oldestontop,youngestatthebottom),neartheentryshaftintotheDinalediChamber( Figure2b ). FlowstoneGroup2 ,the mostextensivegroupofflowstonesinthechamber(FS2in Table1 ,and Figures1b , 2 and 3 ),compriseswallapronsandsheetsthathavespreadoutacrosstheflooroftheDinalediChambertogetherwithdrippools,cascades,curtains,stalactitesandstalagmitesthatconnecttothesesheets,and,therefore,formedinconjunctionwiththem. FlowstoneGroup3 (FS3in Table1 and Figures1b , 2 and 3 )comprisestheflowstonedepositsthatareactivelyformingbelowexistingdrip points,andincludefreshgrowthofdelicatecrystalsofaragoniteandcalciteinfloorsedimentsandalongcavewalls. SedimentarydepositswithintheDinalediChambercanbeorganizedintothreeprimarystratigraphicunits( Dirksetal.,2015 ). Unit1 consistsofdepositsofnon-lithified,laminated,orangemud interpretedassuspensiondepositsinstandingwater(Facies1aof Dirksetal.,2015 ),andlaminated mudwithfinesandcontainingsmall-scaleripplecrosslaminationsandrodentremains(Facies1bof Dirksetal.,2015 ),reflectingdepositionbyshallow,flowingwateralongthecavefloor,withadditionalsandymaterialaccumulatingnearlocalentrypoints,wherefractureshigherinthechamberactassedimentconduits( Dirksetal.,2015 , Dirksetal.,2016a ). WithintheDinalediChamber Unit1 depositscanbedividedintothreesub-unitsprovisionally calledsub-units1a,1band1c.ItisassumedthatUnit1istime-transgressiveandfutureworkmayrevealadditionalsub-units.Sub-unit1aiscomposedoflaminatedorangemudstonewithisolatedlensesofsandymaterial,occursaserosionremnantsalongthecavefloor,andispossiblymoreextensivebeneathyoungerdepositsinthechamber.Sub-unit1bisdominatedbysandyorangemuddepositsthatarerichinmicro-faunalremain,stratigraphicallyoverliesdepositsofsub-unit1a( Figure2candd ),andmayhaveformedthroughthepartialerosionandre-depositionofsub-unit 1a.Depositsofsub-unit1caresimilarinappearanceandcompositiontothelaminated,muddysedi-mentsofsub-unit1a,buttheyoccuralongchertledges,solutionpocketsandfracturesinthecham-berwallsandalongtheentryshaft,higherupinthecavechamber( Figure2 ).Theorangemudis mostlytheproductofthecaveformationprocess,representingtheinsolubleresidueleftoverwhencavitiesdevelopviadissolutionofdolomite( Dirksetal.,2015 ).Someofthemud-bearingwaters seepingoutofthefractureswouldhaveflowedaswaterfilmsalongthecavewallstodepositmudonledgesandinfracturestoformsub-unit1c,whilstelsewherethiswaterwouldhavedrippedtothefloortocontributetothedepositionofsub-unit1aand1b. Unit2 iscomposedoflargelylithifiedmudclastbrecciaconsistingofangulartosub-angular clastsoflaminatedorangemudstone(similartothatfoundinUnit1),embeddedinabrownmudmatrix(Facies2of Dirksetal.,2015 ).Themudclastsareinterpretedtobederivedlocallydueto wettinganddryingoforangemuddeposits,whichledtoauto-brecciation,andsubsequenterosionandre-depositionofangularmudclasts( Dirksetal.,2015 ).Wehypothesizethatthemudclasts Dirks etal .eLife2017;6:e24231. DOI:10.7554/eLife.24231 6of59 Researcharticle GenomicsandEvolutionaryBiology

PAGE 7

formingUnit2arepartlyderivedfromerosionofdepositsofsub-unit1c,andpartlyfromayetunidentifiedunitthatwaslikelydepositedinfractureswithinandabovethechamberentryzone.Twomacro-fossils(partialshaftsoflongbones)thatarenon-specific,butnothominin,havebeenfoundinUnit2. Unit2sedimentsareonlyexposedashangingremnantsattachedbelowtheremainsofacompositeflowstonesheet(Flowstone1a)neartheentranceshaftintothechamber( Figure2b ; Dirksetal.,2015 ).Notethatin Dirksetal.(2015) Unit2wasoriginallydefinedtoalsoincludesedimentsbelowFlowstones1b-e;however,basedonournewdatingresults,thereviseddefinitionofUnit2hasbeennarrowedtoonlyincludethemoreinduratedanddistinctlydarkercolouredero-sionalremnantsofmudclastbrecciaunderFlowstone1a,whicharenotablefortheirabsenceofhomininfossils.Unit2sedimentsaccumulatedasaslopingdebrisconeofmudclastbrecciabelowaverticalfracturesystembeforebeingcoveredbyflowstone(Flowstone1a).ThedebrisconeofmudclastbrecciawassubsequentlyerodedleavingbehindhangingerosionremnantsofUnit2belowaflowstoneapron( Figures2b and 3l ).TheprocessesthatcausederosionoftheUnit2debriscone ledtothedepositionofUnit3sedimentalongtheflooroftheDinalediChamberasshownin Figure8 . Unit3 iscomposedoflargelyunlithified,clast-supported,mudclastbreccia(Facies2of Dirksetal.,2015 ),dominatedbyreworkedangulartosub-angularmudclasts,whichareinterpretedasbeinglocallyderivedfromthereworkingofUnits1and2.Unit3accumulatedbelowthehangingremnantsoftheUnit2debrisconeneartheentryshaft,andalsoextendsalongthecurrent,slopingcavefloortotheSWendofthechamber( Figures2c and 8 ).Unit3sedimentsaredynamic inthesensethattheyarepoorlylithifiedinmostplacesandactivelyslumptowards,anderodeinto,floordrainsthatoccurinpartsofthechamberwheresedimentisbeingwasheddowntodeeperlev-elsinthecave(likelyasaresultoffluctuationsinthegroundwaterlevel).RemainsofUnit3sedimentareattachedtoapron-likeerosionalremnantsofFlowstones1b-eneartheentranceshaft( Figure2a and 3l ).ErosionalremnantsofUnit3underFlowstone1ccontaininsitulongbonesconsistentwith H.naledi ,whichareactivelyerodingoutandaccumulatingalongthepresentcavefloor.Notethat Dirksetal.(2015) originallyincludedtheseerosionalremnantsaspartofUnit2.Everywhereelse, Unit3depositsarespreadacrossthecaveflooraslooselypacked,semi-moist,orangemudclastsofvaryingsizesinwhichbonematerialof H.naledi isdistributed.Unit3ispartlycoveredbysheetsof FlowstoneGroups1,2and3. Unit3hasbeendividedintoalowerandanuppersub-unit,termedsub-unit3aand3b( Figure2 ),basedontherespectiveabsenceorpresenceofhomininfossils.Sedimentsbelongingto sub-unit3aarenotdirectlyexposedinthechamber,buttheirpresencehasbeenconfirmedinthedeepestpartoftheexcavationarea( Figure2d ).Incontrastsub-unit3bisexposedwithinthetalus coneneartheentryshaftandalongthecavefloor,andcontainsalloftheknown H.naledi fossilsin thechamber( Figure2candd ).Thethicknessofsub-unit3bisthoughttobenomorethan20±30 cm(seebelow).Thedistributionoffossils,units,andflowstonesAllhomininbonesidentifiedintheDinalediChamberarecontainedindepositsofsub-unit3b.Bonesattributedto H.naledi havebeenrecoveredas:(a)isolatedelementsthatweatheredoutfromerosionremnantsofsub-unit3bbelowFlowstones1b-e;(b)asfragmentedremainsscatteredacrossthecavefloor;and(c)aspartlyarticulatedremainsfromasingleexcavationpitdowntoadepthof ~ 20 cmbelowthecurrentfloorlevel( Dirksetal.,2015 ). Preliminarygroundpenetratingradarwork( Naidoo,2016 )suggeststhatUnit3depositsalong theflooroftheDinalediChambercouldbeupto1.5mthick.A50cm-deepsondagewasduginthecentreoftheexcavationpit,whichitselfis20cmdeep,toindicateaminimumdepthof70cmforthemudclastbrecciapileofUnit3.Thetop20cmofthissedimentcontains H.naledi remains andispartofsub-unit3b( Figure2d ).Adiscretecontactoccursat15±20cmdepth,belowwhichno morefossilswereencounteredwiththeexceptionofasinglejuvenilebaboontooth(sample1841; Figure7 )thatwasrecoveredfromadepthof55±60cmbelowtheoriginalcavefloorsurfaceinsedimentofsub-unit3a( Figures2 and 8 ). Stainingpatternsonbonefragments,skeletalelementrepresentation,andthefactthatbones canbeseentoweatheroutfromerosionalremnantsofsub-unit3b,indicatethatpartofthefossil Dirks etal .eLife2017;6:e24231. DOI:10.7554/eLife.24231 7of59 Researcharticle GenomicsandEvolutionaryBiology

PAGE 8

assemblagehasbeenreworked( Dirksetal.,2015 ).Thepresenceofwell-articulatedremainsinthe excavationpitawayfromthechamberentranceindicatesthatsomeoftheremainsenteredthecaveintact.Themixedtaphonomicsignaturesuggeststhatfossilsenteredthecaveoveraperiodoftime,whichisminimallyassumedtobeduringdepositionofsub-unit3b,andbeforedepositionofFlow-stone1c.Fossilentrymayhavecontinuedassedimentaccumulationsofsub-unit3bneartheentryshaftwerereworkedandredistributedalongthecavefloor( Figure8 ). ThestratigraphicrelationshipsintheDinalediChambersuggestthatUnit1sedimentswere depositedoveralongperiod,whichbothpredatesandspansthemorelimiteddepositionaltime-framesofUnits2and3.Hence,Unit1istime-transgressive,meaningthatthesesedimentswere(andare)constantlyformingindifferentpartsofthechamberduetoweatheringofthedolomiticcavewalls(i.e.,wadformation sensu Martinietal.,2003 ),andthattheirageisdependentupon whereinthecavethematerialislocated.Atpresent,wecanonlydivideUnit1intothreesub-units,butwehypothesizethatanoldersub-unitconsistingoflaminatedorangemudstoneexists(orexisted)higherupinthechamberaswell(possiblyonlyonledgesandinfractures),whichwaserodedtoprovidesomeofthesedimentthatformedUnit2andpartsofUnit3,neartheentryshaft. Flowstone1a,whichoverliesremnantsofUnit2,istheoldestflowstoneunitinthechamber,and displaysevidenceofmultiplephasesofflowstoneformationfollowedbypartialdissolution( Figure2b ).Flowstonedissolutionoccurredduringtimeperiodswhenthewatertablewaselevated andthechamberwasfilledwithstandingwater.TheerosionremnantsofFlowstone1adiptowardsthedeeperpartofthechamber,indicatingthatatthetimeofitsformation,aslopingdebrisconeofUnit2sedimentwaspresent.ErosionofUnit2sedimentsfromunderneathFlowstone1aonlyoccurredaftertheflowstonehadformedandlithifiedthetopofUnit2.FollowingerosionofUnit2,depositionofUnit3began,assedimentandmudclastsspreadoutoverthecavefloorandalsofilledmuchofthespaceunderneathFlowstone1a.Thishasledtoaninvertedstratigraphynearthecaveentrance,althoughanormalstratigraphyisdocumentedatthebottomofthechamber,wherethecavefloorisflatlyingandsedimentofUnit3progressivelybuiltup( Figures2b and 8 ).Atsome pointduringtheseprocessesremainsof H.naledi enteredthecavechamber,markingthestartof depositionofsub-unit3b.Followingdepositionofsub-unit3bandthehomininremains,Flowstones1b-eweredepositedoversub-unit3bintheentryzone.TheseflowstoneshavebeeninterpretedasyoungerthanFlowstone1a,butolderthantheFlowstoneGroup2sheetsalongthecavefloor.Inotherwords,afterdepositionofUnit3commencedtoformthetalusconeneartheentranceofthechamber,partsoftheconeslumpedanderodeddowntowardsdeeperpartsofthechamberafterFlowstones1b-eweredeposited,butbeforeFlowstoneGroup2wasdeposited.Thisslumpingmotionwasprobablydrivenbysedimentbeingremovedfromthebaseofthestratigraphicpilethroughfloordrains. FlowstoneGroup2coverserosionremnantsofFlowstones1a-eascoatingsandstalactitesalong driprims.Inplaces,FlowstoneGroup2alsocoverserosionremnantsofUnit1andUnit3alongtheflooranddisplaysvariablerelationshipswithUnit3( Figure2 ).WherepartsofUnit3havebeen erodedviafloordrains,hangingremnantsofFlowstoneGroup2canbefoundattachedtothewallsasfringingaprons,upto10cmabovethecurrentfloorlevel,establishingthefactthatpartsofthefloorarecurrentlyinastateoferosion.Inotherplaces,FlowstoneGroup2sheetsdirectlyoverlieUnit3andthe H.naledi fossilsitcontains.ThesevaryingrelationshipsindicatethatFlowstoneGroup 2sheetsweredepositedoveranextendedperiodoftime,post-datingdepositionandpartialreworkingofsub-unit3b. Insummary,thestratigraphiccontextindicatesthatthe H.naledi fossilsenteredthecaveduring deposition(andpossiblyduringpartialreworking)ofsub-unit3b,afterdepositionoftheoldersedimentsofUnit1(sub-unit1a)andUnit2.Severalisolated,non-homininbonefragmentsinhangingerosionremnantsofUnit2andasinglebaboontoothinfloorsedimentsinsub-unit3aweredepositedpriortotheentryofthehomininremains.TheaccumulationofUnit3alongthecavefloorinvolvedadynamicinterplaybetweentheaccumulationofmudclastbrecciabelowsedi-mententrypointsorinsitusedimentsources(Unit1andUnit2)inthechamber,anderosionthroughfloordrainsresultingincontrastingstratigraphicrelationshipsacrossthechamber( Figures2 and 8 ). Dirks etal .eLife2017;6:e24231. DOI:10.7554/eLife.24231 8of59 Researcharticle GenomicsandEvolutionaryBiology

PAGE 9

Datingthe H.naledi fossils MostfossildepositsintheCradleofHumankindthathavebeendatedarebetween0.5and3.7Maoldandconsistofbonematerialencasedinwell-cementedhardclasticrockscommonlyreferredtoascavebreccia(e.g., Wilkinson,1985 ; O'ReganandReynolds,2009 ; Herriesetal.,2009 ; Pickeringetal.,2011b ; Grangeretal.,2015 ).Intheabsenceofvolcanicdeposits,itisgenerally difficulttoobtainaccurateagesforthefossils,notjustbecausereliabletechniquesarefew,butmostlybecausethestratigraphicsequencesinthecavesarecomplex,discontinuousandfrequentlyreworked(e.g., Brain,1993 ; Pickeringetal.,2011a ; Bruxellesetal.,2014 ; Stratfordetal.,2014 ). Workershavereliedonacombinationofbiochronologyoffaunalremains,palaeomagneticworkandarangeofradiometricmethods,includingU-Pb,U-ThandESRdatingtargetingflowstonesandfossilteeth(e.g., Vrba,1975 ; Partridgeetal.,1999 ; Bergeretal.,2002 ; Walkeretal.,2006 ; Herriesetal.,2006 , 2013 , 2014 ; HerriesandShaw,2011 ; Dirksetal.,2010 ; Pickeringand Kramers,2010 ; Pickeringetal.,2011a ; HerriesandAdams,2013 ),aswellaslimitedcosmogenic (10Be,16Al)dating(e.g., Partridgeetal.,2003 ; Grangeretal.,2015 ; Dirksetal.,2016b ).Whilst someofthesetechniquesarewellestablished,otherssuchastheapplicationofcosmogeniciso-chrons(e.g., Grangeretal.,2015 )arerelativelynewandnotwithoutsignificantanalytical(andinterpretative)challenges( KramersandDirks,2017 ),andalleffortsarestronglydependentonthe stratigraphicinterpretationofthefossilsorunitsthatarebeingdated. UnlikeotherfossildepositsintheCradleofHumankind,theremainsintheDinalediChamber arelargelyrestrictedtohominins.Thismakesitimpossibletousebiochronologyasapreliminarytechniquetoassesstheageofthefossils.Inaddition,thefossilsarecontainedinmostlyunconsoli-datedmuddysedimentwithclearevidenceofamixedtaphonomicsignatureindicativeofrepeatedcyclesofreworkingandmorethanoneepisodeofprimarydeposition( Dirksetal.,2015 ).This indicatesthatcautionisrequiredwheninterpretingthestratigraphyandtheageofthefossilstheycontain. Inpreparationforthisstudy,trialdatingofthedepositsintheDinalediChamberwasundertaken toobtainanindicationoftheageofthedepositandthebesttechniquestoapply.PreliminaryworkwasfocussedonassessingtheviabilityofU-seriestechniquesforflowstonedating,using14Cfordatingbonefragments,andusingOSLtotestsamplesofquartz-bearingUnit1( Dirksetal.,2015 ).InitialtestswerecarriedoutattheUniversityofJohannesburg(UJ)toassesssuitabilityforU-Pbdating,whichallowsforthedatingofolder(>500ka)flowstonematerial(e.g., Walkeretal.,2006 ; Pickeringetal.,2010 ; PickeringandKramers,2010 ),ontheassumptionthatthe H.naledi material couldbeolderthan1Mabasedonitsprimitivemorphology( Bergeretal.,2015 ; Demboetal., 2016 ; HawksandBerger,2016 ; Thackeray,2016 ; Hawksetal.,2017 ).Itwasfoundthattheolder flowstonesintheDinalediChambercontainedexcessivecommonPbcausedbytheinclusionofdetritalmaterial(mainlyclays)makingthemunsuitableforU-Pbdating( Dirksetal.,2015 ).Incontrast,preliminarytestswithU-ThdisequilibriumdatingatJamesCookUniversity(JCU)returnedpromisingresults.U-Thdatingismorepreciseinthe<500karangethanU-Pbdating,andismuchlesscriticallyaffectedbydetritalmaterial.TheinitialtestswithU-Thdisequilibriumdatingrevealedthatthefossilsmaybemuchyoungerthanoriginallyanticipated(e.g., Demboetal.,2016 ; Thackeray,2016 ),andmostlywellwithintherangeoftheU-Thtechnique.Therefore,U-Pbdatingwas notpursuedfurther. PreliminarytestswithOSLwereconductedattheUniversityoftheWitwatersrand(Wits)onsamplesfromUnit1,whichwereassumedtobeolderthanthefossilsof H.naledi .Thesepreliminary studies,andtheresultscontainedinthispaper,arethefirstOSLresultsforcavesedimentsfromtheCoH,andagainindicatedthatthe H.naledi fossilswereprobablyrelativelyyoung(i.e.,<500 ka). Testswithradiocarbon(14C)datingwereundertakenthroughacommercialfacility(BetaAnalytic Inc.inFlorida),toensureafastturn-aroundtimeforresults.Atthetimethesedatingtestsweredone,itwasalreadyknownfromU-ThandOSLteststhatthe H.naledi fossilswouldbetoooldto bedatedby14C.Nevertheless,analyseswerecarriedoutaspartoftheduediligenceprocess,and theresultsofthesetestsarepresentedhere.Followingthisinitialwork,nofurtherradiocarbonstud-ieswerecarriedout. Thepreliminaryresultshaveguidedthesubsequentdatingstrategyandsamplingapproach reportedhere.Thedatingstrategywasdesignedtoachievethreeobjectives:(i)establishadetailed Dirks etal .eLife2017;6:e24231. DOI:10.7554/eLife.24231 9of59 Researcharticle GenomicsandEvolutionaryBiology

PAGE 10

stratigraphyforthecavesedimentsintheDinalediChamber;(ii)datesedimentaryunitsthatpoten-tiallybracketthefossil-bearingdeposits;and(iii)datethefossilsdirectly. Toobtainanupperagelimitforthefossil-bearingdepositsofUnit3(i.e.sub-unit3b),weconductedU-ThdatingofflowstonesthatdirectlyoverlieUnit3.Alargenumberofsuchflowstonesweresampledwiththeaimoffindingtheoldestflowstonedirectlyoverlying H.naledi fossils.To obtainaloweragelimitforsub-unit3b,erosionalremnantsofUnit1sedimentsthatwereatleastpartiallycoveredbyfossil-bearingsub-unit3bsediments,weresampledforOSLdatingontheassumptionthatsub-units1aand1bintheseareasareolderthansub-unit3b( Dirksetal.,2015 ). ThiswasdoneinthefullknowledgethatOSLdatingofcavesedimentsiscomplexanddifficulttointerpret(e.g., Robertsetal.,2009 ),andprobablyimprecise.Asaninternalcontrol,wealsosampledflowstonesthatcovertheoutcropsofsub-units1aand1bfromwhichOSLsamplesweretaken.TheseflowstonesweredatedwithU-Thwiththeexpectationthattheyareyoungerthantheunderly-ingUnit1sediments.InadditiontoOSL,Flowstone1a,whichoverliesUnit2sediments,wassam-pledforpalaeomagneticanalyses.Thisflowstonewastargeted,becauseitwasexpectedtobetheoldestflowstoneinthechamberandpossiblyolderthan780ka,andhencecouldpotentiallyrecordreversemagneticpolarity(e.g., Singer,2014 ).Inthiscase,thiswouldconstraintheminimumageof Unit2. Thebestageestimatesfor H.naledi canbeobtainedbydirectlydatingfossilmaterial.Itwasclear frompreliminaryteststhatthiscouldnotbeachievedwith14C,andinsteadcombinedESRandU-Th disequilibriumdatingtechniques(US-ESR; Gru È netal.,1988 )wereappliedtothree H.naledi teeth thatwerefreshlycollectedfromnearthesiteoftheoriginalexcavation( Figures1 , 2 , 4 and 6 ),as wellasasinglebaboontooth(cf. Papio )thathadbeenrecoveredfromsub-unit3abelowthehominin-bearinghorizon( Figures2d and 7 ). OnceresultswereobtainedforESRandU-Thdating,itbecameapparentthatOSLdatingwould onlyprovidegeneralageconstraintsthatconfirmedtheESRresults,butintheirownrightdidnotreturnadditionalageconstraintsforthefossils.OSLresultswerealsodifficulttointerpretinthecomplexcaveenvironmentthatwasstronglyaffectedbyRnloss(seeDiscussion).Itwas,therefore,decidednottopursuemoredetailedOSLstudiesatthisstage,eventhoughwedidcarryoutprelim-inarytestsforsinglegrainandfeldsparanalysesattheUniversityofWollongong,toassessthesuit-abilityofthesetechniques.Pilotresultsareencouraging,andsuggestthatfuture,detailedOSLstudiesareworthpursuing. Results U-ThdatingofflowstonesU-Thdatingof17flowstonesamples( Figure3 )hasyieldedminimumdepositionalageestimates forthesedimentaryunitstheyoverlie,andhasprovidedinsightsintothetimingofflowstoneforma-tionevents( Tables1 , 2 and 3 ).ThreeseparatecheckswerebuiltintotheU-Thdatingstrategyto ensurerobustresultswouldbeobtained.IndependentdatesforthesamesampleswereobtainedbylaboratoriesatJCUandattheUniversityofMelbourne(UoM),withresultsdisplayingahighdegreeofconcordance.Ininstanceswheresampleswereobtainedfromthesameflowstonelayer,butatdifferentstratigraphiclevels(e.g.,samplepairsRS13andRS18,RS22andRS23,RS16andRS17,andRS11andRS21)allagesareconsistentwithstratigraphicorder,andblindduplicatesofthesamesample(RS1andRS15)returnedidenticalresultswithinerror,indicatingthatresultsarebothaccurateandprecise.DoubleblindU-ThresultsfromJCUandUoMareshownin Tables2 and 3 ,respectively.ThedistributionofflowstoneagesacrosstheDinalediChamberisshownin Figures1b and 2 . TheoldestdatedflowstoneintheassemblageisFlowstone1aoverlyingUnit2,whichyieldsage estimatesof502+181/±53ka(RS23)and478+107/±41ka(RS22)( Table1 ).Thenextoldestage comesfromaflowstoneinterpretedasFlowstoneGroup1overlyingsedimentofsub-unit1btotheWoftheexcavationpit( Figure1b )withanageof290 ± 6ka(RS5; Table1 ).Thisageisyounger thantheOSLagederivedfromsub-unit1a(OSL5),butisslightlyolderthanOSLagesderivedfromsub-unit1binthislocation(OSL4; Figure2c ).Thissuggeststhatsub-unit1awasdepositedpriorto precipitationofthisflowstoneandthatsub-unit1bformedoutofstratigraphicorderduetoerosionandredepositionofthetopofsub-unit1abeneaththisflowstone.However,theU-ThdateforRS5 Dirks etal .eLife2017;6:e24231. DOI:10.7554/eLife.24231 10of59 Researcharticle GenomicsandEvolutionaryBiology

PAGE 11

Dirks etal .eLife2017;6:e24231. DOI:10.7554/eLife.24231 11of59 Researcharticle GenomicsandEvolutionaryBiology

PAGE 12

Dirks etal .eLife2017;6:e24231. DOI:10.7554/eLife.24231 12of59 Researcharticle GenomicsandEvolutionaryBiology

PAGE 13

shouldbeinterpretedwithcautionastheflowstonehasaporoustexture( Figure3a ),whichprobablyindicatessomedegreeofdissolutionand/orrecrystallizationoftheprimarycalcite,andmayhaveaffectedtheU-Thsystematics(seeDiscussion). Flowstonesamplesthatoverliesub-unit3b,whichcontainsthe H.naledi fossils,yieldageestimatesthatfallwithinfourdistincttimeperiods: ~ 242ka(RS18=242 ± 5ka[JCU]and242.9 ± 6.6ka Figure3. Fieldandclose-upphotographsofallflowstonesamplescollectedforU-Thdating.Theflowstone groups(i.e.,FlowstoneGroups1,2or3),samplenumbers,andages(2 s uncertainty),aslistedin Table1 ,are shownbeloweachsample.AgesreportedherearefromJCU,unlessotherwisestated. DOI:10.7554/eLife.24231.005 Dirks etal .eLife2017;6:e24231. DOI:10.7554/eLife.24231 13of59 Researcharticle GenomicsandEvolutionaryBiology

PAGE 14

[UoM])forFlowstone1c;88±106ka(RS1,RS8,RS13-17);50ka(RS6);and24±32ka(RS10,RS19-21)forFlowstoneGroup2deposits.TheresultsforRS18provideaminimumageforthe H.naledi fossils inthispartofthecave. Anactivelyforming(i.e.,dripping)stalagmiteofFlowstoneGroup3(RS11)ontopofanolder baseofFlowstoneGroup2returnedayoungerageof9±10ka(RS11=9.05 ± 0.06[JCU]and 9.95 ± 0.06ka[UoM]).Thefinalflowstonesamplethatwasdatedinthisstudywascollectedonthe landsurfaceabovethecavesystem.Itwassampledfroma ~ 14cmwideverticalflowstone-filledfractureexposedonthesurfaceabovethesouthernendoftheDinalediChamberitself.Thisistheonlypossiblealternativeentry-wayintotheDinalediChamberthatwehaveobserved.Thereplicatesam-ples(RS9)analysedatJCUandUoMbothindicatedsecularequilibrium,whichconfirmsthattheflowstonesealedthisfracturesometimebefore ~ 600ka,eliminatingthisthinflowstone-filledfracture asapossiblealternativeentrancefor H.naledi intotheDinalediChamber.Itshouldalsobenoted thatnoevidenceofatalusconeoranyotherevidenceofsedimententryintothechamberbelowthispointhasbeenobserved. Thespatialdistributionoftheflowstonesbelongingtothedifferentagegroups( Figure1b )indicatesthattheoldestflowstones(FlowstoneGroup1)occurneartheentryzoneintothechamberandasanerosionalremnant(RS5)nearthebackofaWNW-trendingfractureWoftheexcavationpit.The88±106kaflowstonesformedinthreeseparatepartsofthechamber( Figure1b ):(a)ontop ofolderflowstonesneartheentry;(b)aswalldrapesaboveadrain6mSWoftheentry;and(c) Table1. SummarytableofU-ThdisequilibriumagesobtainedforsamplesfromtheDinalediChamberbyJamesCookUniversity (JCU-1)andtheUniversityofMelbourne(UoM-2).Thedetailedanalyticalresultsareshownin Tables2 and 3 .Samplelocationsare shownin Figure1b .Thedataarerankedbyincreasingageoftheoldestflowstonehorizonwithinthesample,basedontheJCUages. Thegreyshadinghighlightsthedifferentagegroupingsobservedwithintheflowstones:24±32ka, ~ 50ka,88±105ka, ~ 242ka, ~ 290ka and>440ka.Agesarereportedrelativeto1950.SampleIDFlowstonegroupUnderlyingunitAge1(ka)2 s1(ka)Age2(ka)2 s2(ka) RS19FS2sub-unit3b 24.7 0.2 24.53 0.43 RS11FS3(toptoRS21)FS2 9.05 0.06 9.946 0.063 RS21FS2(basetoRS11)sub-unit3b 28.4 0.4 28.62 0.29 RS10FS2sub-unit3b(andbone) 30.1 0.3±± RS20FS2sub-unit1a(Facies1a;OSL5) 30.4 0.2 32.12 0.38 RS6FS2sub-unit1a,sub-unit3b 49.8 0.3 50.82 0.43 RS15FS2(blindduplicateofRS1)sub-unit3b 92.6 1.0 91.40 0.65 RS1FS2(blindduplicateofRS15)sub-unit3b±± 91.04 0.72 RS8FS2(belowFS1a-e)sub-unit3b 95.0 1.0 96.29 0.69 RS14FS2indrain,alongdolostonewall 100.1 1.2 96.20 0.36 RS17FS2(toptoRS16)indrain,alongdolostonewall 102.6 0.8 98.6 1.4 RS16FS2(basetoRS17)indrain,alongdolostonewall 104.0 1.9 99.1 1.4 RS13FS2(rimtoRS18)sub-unit3b±± 88.46 0.67 RS18FS1c(coretoRS13)sub-unit3b 242 5 242.9 6.6 RS5FS1sub-unit1b(Facies1b;OSL4) 290 6±± RS22FS1a(toptoRS23)Unit2 equilibrium ± 478 +107/ 41 RS23FS1a(basetoRS22)Unit2 equilibrium ± 502 +181/ 53 RS9n/a(surfaceoutcrop)n/a equilibrium ± equilibrium ± 1JamesCookUniversity(JCU),AdvancedAnalyticalCentre.2UniversityofMelbourne(UoM),paleochronologylaboratory. DOI:10.7554/eLife.24231.006 Dirks etal .eLife2017;6:e24231. DOI:10.7554/eLife.24231 14of59 Researcharticle GenomicsandEvolutionaryBiology

PAGE 15

Table2. U-ThdatatableforJamesCookUniversity.Uncertaintiesinclude:analyticalerror,decayconstantuncertainty,anduncertainty oninitial230Th/232Th.Agesarereportedrelativeto1950andassumeaninitial230Th/232Thactivityof0.83 ± 0.5,andtheequationgiven in Placzeketal.(2006) .Decayconstantsfor234Uand230Tharefrom Chengetal.(2013) . SampleIDU(ppm)234U/238U2 s230Th/238U2 s232Th/238U2 s Age(kaBP)2 s (ka)234U/238Uinitial2 s (activity)(activity) (corrected)(activity) RS112.3141.7720.0500.1440.0010.00010880.00000059.050.061.81840.0003RS190.6521.8550.0010.3870.0020.0021760.00000824.70.21.9890.001RS210.4211.9460.0010.4600.0040.0019200.00001528.40.42.1090.002RS100.8461.8850.0010.4660.0030.0007920.00000330.10.32.0530.001RS200.7951.8550.0010.4630.0030.0013630.00000530.40.22.0220.001RS60.5601.9660.0010.7470.0030.0009740.00000249.80.32.2630.002RS150.4001.9120.0011.1640.0080.004720.0000192.61.02.4840.007RS80.3281.8130.0031.1200.0080.003160.0000295.01.02.3730.007RS140.7341.6390.0951.0390.0080.002980.00002100.11.22.1750.008RS170.6801.6090.0011.0320.0050.0006790.000001102.60.82.1500.005RS160.9731.5830.0001.0240.0110.0004030.000006104.01.92.120.01RS180.1521.8480.0011.8560.0130.011750.0000524253.660.05RS50.0901.7280.0011.8180.0090.017320.0000529063.920.07RS230.3141.1870.0021.3150.0110.003460.00002>400±±RS220.3671.2090.0011.3220.0080.0001250.000001>400±±RS90.7371.0070.0021.0290.0040.0004620.000001>400±± DOI:10.7554/eLife.24231.007 Table3. U-ThdatatablefortheUniversityofMelbourne.Activityratiosaredeterminedafter Hellstrom(2003) and Drysdaleetal. (2012) .Agesarecorrectedforinitial230Thusing Equation1 of Hellstrom(2006) ,thedecayconstantsof Chengetal.(2013) ,andan initial230Th/232Thactivityof1.5 ± 1.5.Theinitial234U/238Uratiosarecalculatedusingcorrectedages,whicharereportedrelativeto 1950.SampleIDU(ppm)234U/238U2 s230Th/238U2 s232Th/238U2 s Age(kaBP)2 s (ka)234U/238Uinitial2 s (activity)(activity) (corrected)(activity) RS111.5181.8080.0030.15970.00090.00008750.00000049.9460.0631.8310.004RS190.5011.8840.0110.39160.00260.0043220.00001024.530.431.9470.011RS210.3611.9680.0110.46540.00300.00113420.000001928.620.292.0490.011RS200.6261.8780.0110.49250.00320.00238370.000004032.120.381.9610.011RS60.2762.0230.0040.78560.00210.004960.0001050.820.432.1810.004RS130.0762.0060.0041.18370.00470.0047860.00005888.460.672.2910.005RS150.3811.9340.0041.16610.00290.006390.0001291.370.652.2090.005RS140.6651.6260.0031.00100.00150.0012620.00001496.240.361.8220.003RS80.2571.8310.0041.13970.00340.0057460.00006096.290.692.0910.004RS170.5171.6370.0091.02480.00660.00239630.000003798.61.41.8410.010RS160.9051.5900.0100.99630.00670.00170990.000003799.11.41.7800.011RS180.1042.0010.0112.03200.01400.0205570.000041242.96.62.9870.027RS220.3241.2280.0071.30170.00830.00012010.0000008478+107/±41±±RS230.2061.2250.0071.30160.00930.0078180.000016502+181/±53±±RS90.8961.0100.0021.02040.00180.0009160.000012±±±± DOI:10.7554/eLife.24231.008 Dirks etal .eLife2017;6:e24231. DOI:10.7554/eLife.24231 15of59 Researcharticle GenomicsandEvolutionaryBiology

PAGE 16

Table4. SummarytableofU-Thdisequilibriumagesobtainedforthethree H.naledi teeth(samples1767,1788and1810)andthe baboontooth(sample1841)fromtheDinalediChamberobtainedbySCU-UoW.NoagecalculationswerecarriedoutforUconcentra-tionsof 0.5ppmorU/Th 250(indicatedinredandunderlined).Meanvaluesinthistableonlyincorporatevaluesfromwhichmeaningfulagescouldbecalculated(indicatedinblack),howeverallvalues(i.e.,redandblack)wereaveragedtoobtaintherelevantmeanvaluesreportedin Table4 .Alluncertaintiesaregivenas2 s . Sample1767 U(ppm)U/Th230Th/238U2 s234U/238U2 s Age(ka) 2s(ka)(234U/238U) i2 s 1767-1D7.226852.1670.0246.2590.00943.51.16.9490.0261767-2D7.759962.2610.0236.2820.01045.51.17.0090.0301767-3D8.03 196 2.2250.8256.2760.012±±±± 1767-4D8.559512.2090.0316.3010.00944.11.47.0070.0301767-5*E3.6912382.2590.0316.1970.05546.21.86.9240.1261767-6*E1.76 108 2.2391.1336.1650.038±±±± 1767-7*E2.15 109 2.3370.9476.2310.024±±±± 1767-8*E2.465182.2760.0216.2530.01946.11.16.9860.048Mean:1767D7.848772.2120.0266.2810.00944.51.26.9880.0291767E3.088782.2680.0266.2250.03746.21.46.9550.087Sample1788U(ppm)U/Th230Th/238U2 s234U/238U2 s Age(ka)2s(ka)(234U/238U)i2 s 1788-1D6.673902.9670.0266.4230.01161.41.57.4530.0541788-2D7.08 176 3.3700.8336.4410.010±±±± 1788-3D7.17 60 3.2063.1266.3940.049±±±± 1788-4D7.4513913.3130.0236.4450.01070.31.47.6450.0561788-5D5.5244233.2690.0236.3490.01070.41.47.5310.0521788-6D5.0740903.4160.0146.3780.01474.11.17.6340.0541788-7D5.3947293.3850.0206.4000.01472.91.47.6400.0541788-8D5.9332093.4270.0156.3930.01374.21.17.6540.0541788-9D5.2443293.4490.0146.4130.01474.51.07.6850.0521788-10D4.8931613.3900.0106.4030.01173.00.97.6450.0521788-11D4.825563.3940.0146.4160.01472.91.07.6590.0521788-12D5.4816063.3560.0176.3840.01472.31.17.6090.0521788-13D5.048383.3170.0256.4200.01470.71.57.6230.0581788-14D5.69 93 3.2812.4266.4080.013±±±± 1788-15D5.03 72 3.3153.7316.4270.014±±±± 1788-16E 0.13 3 1.78618.1493.8340.267±±±± 1788-17E0.68 25 0.7529.1496.2480.273±±±± 1788-18E 0.4 16 0.80113.0536.2360.050±±±± 1788-19E 0.08 3 1.78336.2314.3010.288±±±± 1788-20E1.023062.9900.1175.5410.15475.19.36.6170.3941788-21*E 0.33 50 2.04127.1355.7930.141±±±± 1788-22*E 0.12 30 1.51324.8015.9750.098±±±± 1788-23*E 0.25 34 1.36817.0715.9880.079±±±± 1788-24*E 0.36 90 1.23713.5556.1670.055±±±± 1788-25*E 0.41 107 1.0848.6726.2060.033±±±± 1788-26*E 0.48 102 1.30211.3336.3840.081±±±± 1788-27*E 0.49 165 0.6867.7336.3670.037±±±± 1788-28*E 0.31 167 1.6156.9755.6020.246±±±± Table4continuedonnextpage Dirks etal .eLife2017;6:e24231. DOI:10.7554/eLife.24231 16of59 Researcharticle GenomicsandEvolutionaryBiology

PAGE 17

Table4continued Sample1767 U(ppm)U/Th230Th/238U2 s234U/238U2 s Age(ka) 2s(ka)(234U/238U) i2 s 1788-29*E 0.44 6 2.31111.8985.5760.306±±±± 1788-30E 0.44 62 0.9885.3106.0890.075±±±± 1788-31E 0.29 8 1.06619.2566.1510.056±±±± 1788-32E 0.23 95 0.99417.4516.3520.064±±±± 1788-33E 0.41 6 1.10321.6516.3440.049±±±± 1788-34E 0.28 51 1.34011.4506.3820.061±±±± 1788-35E 0.35 4 1.28621.0886.3210.062±±±± 1788-36E 0.4 115 1.21612.8966.3720.041±±±± 1788-37E 0.3 61 1.10617.0596.3130.073±±±± 1788-38E0.542792.8100.2376.3000.06458.912.27.2620.270Mean:1788D5.5927933.3350.0186.4020.01371.51.27.6160.0541788E0.782932.9000.1775.9200.10967.010.86.9360.332Sample1810U(ppm)U/Th230Th/238U2 s234U/238U2 s Age(ka)2s(ka)(234U/238U)i2 s 1810-1D7.073483.2310.0215.8140.01777.91.67.0030.0561810-2D8.294113.1120.0305.8630.01073.42.16.9860.0621810-3D8.889793.1060.0275.9290.01072.11.87.0460.0601810-4D9.198333.0490.0445.9930.01169.42.67.0790.0661810-5D9.175082.9370.0475.9900.00766.22.87.0200.0661810-6D9.12 55 3.1436.9195.9810.012±±±± 1810-7D7.954323.0990.0185.9770.01371.11.37.0890.0541810-8D8.844892.9860.0746.0350.060674.17.0880.0841810-9D9.39159053.1220.0135.8700.00673.61.16.9990.0521810-10D9.7878393.1650.0175.8730.01174.81.37.0240.0541810-11D9.0372423.1740.0305.8880.01574.82.07.0430.0581810-12D9.5396263.1570.0195.8890.00974.31.47.0360.0541810-13D10.19102403.0940.0185.9040.00872.21.37.0160.0521810-14D10.64144633.1550.0305.9580.01073.11.97.0990.0581810-15E 0.005 1 0.384146.0361.9650.186±±±± 1810-16E 0.002 2 1.06048.1681.0140.108±±±± 1810-17E 0.004 1 5.40317.5732.3570.194±±±± 1810-18E 0.24 55 3.1959.1094.0540.062±±±± 1810-19E0.544264.0090.1865.0540.109130.821.86.8720.4661810-20E0.853283.6250.1194.2870.137146.822.75.9840.5241810-21E 0.41 48 5.0094.4744.4940.186±±±± 1810-22E 0.15 7 7.6908.2254.3490.351±±±± 1810-23E 0.03 2 9.9121.8435.1530.599±±±± 1810-24E 0.01 0 0.661166.0561.8770.144±±±± 1810-25E 0.02 2 7.4089.3864.9640.557±±±± 1810-26*E0.73 3 3.7624.1004.7350.100±±±± 1810-27*E 0.14 3 5.2714.5104.2550.112±±±± 1810-28*E 0.18 5 3.5418.0794.5620.081±±±± 1810-29*E 0.25 9 3.5117.0954.5620.058±±±± Table4continuedonnextpage Dirks etal .eLife2017;6:e24231. DOI:10.7554/eLife.24231 17of59 Researcharticle GenomicsandEvolutionaryBiology

PAGE 18

deepwithinaN-trendingfracture,8mNoftheexcavationpit.The ~ 50kaflowstonesoriginate fromaW-trendingfracture,6mWoftheentryshaftwhereitoverliessub-unit1aandsub-unit3bsediments.The24±32kaflowstonesalloriginatefromaroundtheareawheretheexcavationpitislocatedattheintersectionpointofthreefracturesets( Figure1b ).Theyoungestflowstonesample comesfrombelowanactivedrippoint,1.5mEoftheexcavationpit,andsimilaractivelyformingflowstonescanbeseeninotherpartsofthechamber.TheflowstoneagegroupingsindicatethatepisodicwetperiodsintheDinalediChamberalternatedwithperiodsduringwhichnoflowstonewasdeposited.U-ThanalysesonteethU-ThdisequilibriumanalysesoffourtoothsampleswereconductedtoconstrainUuptakemodelsintodentaltissuesusedinESRdating.TheanalyseswerealsousedtoprovideapparentU-Thageestimates( Tables4 and 5 ).Analysesofallfourteeth(samples1767,1788and1810from H.naledi , andsample1841from Papiosp.) wereperformedattheUniversityofWollongong(UoW),incollaborationwithSouthernCrossUniversity(SCU).Duplicateanalysesoftwoofthetoothsamples(samples1788and1810)wereconductedatGriffithUniversity(GU)incollaborationwiththeAustralianNationalUniversity(ANU). Table4continued Sample1767 U(ppm)U/Th230Th/238U2 s234U/238U2 s Age(ka) 2s(ka)(234U/238U) i2 s 1810-30*E 0.21 3 4.0275.2914.0730.149±±±± 1810-31*E 0.09 1 3.87539.1364.0290.072±±±± 1810-32*E 0.05 2 2.4699.0994.1870.072±±±± 1810-33*E 0.06 1 2.60224.1694.4260.141±±±± 1810-34*E0.9120103.1310.0684.5610.032105.27.05.7980.1281810-35E 1.01 3 4.29115.5144.0850.025±±±± 1810-36E 0.04 99 6.2978.8904.0600.242±±±± 1810-37E 0.14 5285.7533.9324.3850.234±±±± 1810-38E 0.02 55 5.68728.4904.2110.437±±±± 1810-39E 0.01 17 4.20331.0484.4740.314±±±± 1810-40E2.0915863.9930.0494.9930.037132.56.26.8140.146Mean:1810D9.0753323.1070.0305.9220.014572.31.97.0400.0601810E1.1010883.6900.1054.7240.0788128.814.46.5950.316Sample1841U(ppm)U/Th230Th/238U2 s234U/238U2 s Age(ka)2s(ka)(234U/238U)i2 s 1841-1E2.51 78 4.4153.2525.8510.035±±±± 1841-2E1.96 51 4.2687.6315.8420.044±±±± 1841-3E2.372184.3190.0415.8710.021115.53.57.7580.0901841-4E1.883504.2610.0465.8910.016112.63.67.7300.0821841-5E2.52144.3780.0455.8460.032118.74.37.7840.1241841-6E2.5 12 4.4282.7445.8810.044±±±± 1841-7E2.4 63 4.4841.7445.9460.044±±±± 1841-8E2.14 47 4.4992.4675.9620.037±±±± Mean:1841E2.252614.3190.0445.8690.023115.63.87.7570.099 DOI:10.7554/eLife.24231.009 Dirks etal .eLife2017;6:e24231. DOI:10.7554/eLife.24231 18of59 Researcharticle GenomicsandEvolutionaryBiology

PAGE 19

Table5. SummarytableofU-Thdisequilibriumagesobtainedfortwo H.naledi teeth(samples1788and1810)fromtheDinaledi ChamberobtainedbyGU-ANU.NoagecalculationswerecarriedoutforUconcentrationsof 0.5ppmorU/Th 250(indicatedin redandunderlined).NegativeU/ThvaluesareduetotheThbackgroundbeinghigherthanthemeasuredvalues.Meanvaluesinthistableonlyincorporatevaluesfromwhichmeaningfulagescouldbecalculated(indicatedinblack).Alluncertaintiesaregivenas2 s .CS =ClosedSystem;Diff=diffusion(i.e.,calculatedagesarebasedontheassumptionofcontinuousdiffusionafter Sambridgeetal. (2012) . Sample1810aU(ppm)U/Th230Th/238U2 s234U/238U2 s Age±CS(ka) 2 s (ka)Age±Diff(ka)2 s (ka) (234U/238U) i*2 s 1E 0.03 273.71131.25084.43020.8881n/a±±±±± 2E 0.02 193.16480.93904.20830.4703n/a±±±±± 3E 0.04 333.02571.05315.12200.4988n/a±±±±± 4E 0.05 353.63521.38974.92240.4912n/a±±±±± 5E 0.19 2583.45040.19654.81060.1376n/a±±±±± 6D6.07 29723.29090.06665.98010.055977.22.387.32.77.190.11 7D6.10 53543.26180.08245.97680.031276.32.686.23.37.170.08 8D6.39114363.31690.08005.98270.051477.92.688.33.37.210.119D6.4761933.33180.08995.94700.087379.03.289.73.87.180.1710D6.65 50553.49850.10486.04620.040382.53.494.44.57.370.11 11D6.9551493.54650.09106.05310.040683.83.096.34.07.400.1112D7.1532443.52380.09976.05010.042383.23.395.44.37.390.11 Mean:1±5E0.07 ± 0.063.43210.30034.79620.1504112.015.7137.125.5 ±± 6±12D6.54 ± 0.313.40180.07496.00700.042880.12.591.13.27.270.11 Sample1810bU(ppm)U/Th230Th/238U2 s234U/238U2 s Age±CS(ka) 2 s (ka)Age±Diff(ka)2 s (ka) (234U/238U) i*2 s 1E 0.01 176.844214.04350.02285.5626n/a±±±±± 2E 0.00 39.233016.2333 2.28384.6085n/a±±±±± 3E 0.00 ±216.168827.6564 0.10337.6336n/a±±±±± 4E 0.00 ±314.9980967.0421 0.7695259.2590n/a±±±±± 5E 0.02 1827.1338296.11363.625048.0742n/a±±±±± 6E0.86 24934.51760.17864.65880.0795189.116.7381.3137.97.240.44 7E0.98 6034.87970.14164.87370.0681201.414.00.00.07.840.40 8D4.49204233.57780.06705.93270.077387.12.7100.83.17.310.159D5.35 101283.30460.06595.91420.067778.72.489.22.87.140.13 10D5.67 41973.40770.07775.94800.045981.32.692.83.37.230.10 Mean:1±5E 0.01 ± 0.01 8.7750204.69881.654793.9630n/a----6±7E0.92 ± 0.124.71010.14654.77300.0602195.713.8471.0269.47.540.42 8±10D5.17 ± 0.703.42140.07905.93190.052382.12.793.73.67.230.13 Sample1788aU(ppm)U/Th230th/238U2 s234U/238U2 s Age±CS(ka) 2 s (ka)Age±Diff(ka)2 s (ka) (234U/238U) i*2 s 1E 0.03 48772.10953.30583.57401.6107n/a±±±±± 2E 0.01 213.78455.52711.35252.7713n/a±±±±± 3E 0.00 ±110.503027.8940 2.490910.1171n/a±±±±± 4E 0.00 ±19.0249113.8912 0.712032.3636n/a±±±±± 5E 0.00 ±26.679566.27500.776918.7506n/a±±±±± 6E 0.01 ±63.02310.98442.19040.4875n/a±±±±± 7E 0.24 1052.81390.20766.37910.1624n/a±±±±± Table5continuedonnextpage Dirks etal .eLife2017;6:e24231. DOI:10.7554/eLife.24231 19of59 Researcharticle GenomicsandEvolutionaryBiology

PAGE 20

SummariesoftheU-Thanalyticaldataandagesarereportedin Table4 (SCU-UoW)and Table5 (GU-ANU).In Table4 onlyclosedsystemdatesarereported,while Table5 alsolistsdatesbasedon thecontinuousdiffusionmodelof Sambridgeetal.(2012) .Inbothdatasets,theUcontentin enamelismuchlowerthanindentine.NotethatapparentU-Thagesfortheteetharelikelyto Table5continued Sample1810aU(ppm)U/Th230Th/238U2 s234U/238U2 s Age±CS(ka) 2 s (ka)Age±Diff(ka)2 s (ka) (234U/238U) i*2 s 8E 0.24 2041.64951.58116.01230.2961n/a±±±±± 9E 0.19 5792.40754.26206.33411.3187n/a±±±±± 10E 0.48 1893.17172.28626.10060.1341n/a±±±±± 11E1.34138333.87920.28646.35210.102488.59.2102.912.67.870.3212E2.577554.17700.06096.32750.097298.63.0117.73.18.040.19 Mean:1±6E 0.01 ± 0.01 3.51884.45312.34717.1463n/a--±± 7±10E0.29 ± 0.132.64840.24696.18240.097856.26.561.07.7 ±± 11±12E1.96 ± 1.234.07460.09416.33610.048295.13.2112.34.77.960.26 Sample1788bU(ppm)U/Th230Th/238U2 s234U/238U2 s Age±CS(ka) 2 s (ka)Age±Diff(ka)2 s (ka) (234U/238U) i*2 s 1E 0.02 42.39451.87743.50401.4368n/a±±±±± 2E 0.02 141.96561.32993.30990.9022n/a±±±±± 3E 0.01 ±82.81562.10342.53590.8082n/a±±±±± 4E 0.02 1602.102458.18543.23427.4009n/a±±±±± 5E 0.03 312.38591.40844.12851.5222n/a±±±±± 6E 0.03 202.895111.29114.10463.9747n/a±±±±± 7E 0.02 102.84863.53834.93622.3343n/a±±±±± 8E 0.03 182.83251.61135.70520.7139n/a±±±±± Mean:1±8E 0.02 ± 0.01 2.55976.76184.13081.2209n/a ±±±±± Sample1788cU(ppm)U/Th230th/238U2 s234U/238U2 s Age±CS(ka) 2 s (ka)Age±Diff(ka)2 s (ka) (234U/238U) i*2 s 1D5.44215783.92810.07076.42600.074088.62.6103.03.17.970.152D5.391550373.89080.05656.44160.052487.22.0101.02.47.960.113D4.9517083.89010.08286.40850.079287.83.0102.03.67.930.164D3.8716533.80330.08596.37860.106885.83.399.03.67.850.205D4.2511683.95690.08006.40510.095790.03.1105.03.67.970.196D5.1214933.94330.05796.49510.096187.82.5102.02.58.040.177D5.3426593.80200.05816.47130.058184.02.096.72.47.940.118D5.0610933.99480.06726.44790.063090.32.4105.53.08.030.139D4.7810184.04810.07186.44680.058692.02.5108.03.38.060.1310D5.228173.90110.05826.51870.081386.12.399.62.48.040.1511D5.254253.88720.08506.44150.062487.12.8101.03.67.960.1412D5.463453.95610.05846.46580.073388.82.3103.32.58.030.14 Mean:1±12D5.01 ± 0.283.91750.07966.44790.046187.92.6102.03.67.980.06 DOI:10.7554/eLife.24231.010 Dirks etal .eLife2017;6:e24231. DOI:10.7554/eLife.24231 20of59 Researcharticle GenomicsandEvolutionaryBiology

PAGE 21

provideapparentageestimates,whichwillapproachtheageforUuptakeeventsthataffectedtheteethduringwetperiodsinthechamber,typicallyafterdeposition.Theseagesshould,therefore,beregardedasminimumageestimatesfortheteeth,anddonotrepresentdepositionalagesforthefossils. Sample1767: Thisextremelywornupperpremolarcrown( Figure6a )isheavilyweathered,and onlyasmallfragmentofenamelwasleftattachedtothedentine.Itcould,therefore,onlybedatedonce(atSCU-UoW).BothdentineandenamelyieldconsistentresultswithapparentU-Thagesof44.5 ± 0.6kafordentineand46.1 ± 0.7kaforenamel,andinitial234U/238Uactivityratiosat 6.99 ± 0.01and6.99 ± 0.04respectively.Theseresultssuggestthatasingleuptakeeventisdated. ThetoothischaracterisedbyanextremelyhighUcontentintheenamelwhencomparedtotheotherteeth( Table4 ).Uraniumconcentrationgradientsshowtheeffectsofdiffusionintothe enamelfromallexternalsurfaces,withenrichmentattheEnamelDentineJunction(EDJ). Sample1788: Thislowerrightsecondmolarwascoveredbyathinlayerofsedimentandiswellpreserved( Figure6b ).Uraniumconcentrationsinenamelanddentinevaryacrossthesurface,with minorhotspotsandleachingzonesnearenamelcracksandalongthedentinecanal.TheEDJisenrichedinU,showingadiffusiongradientintotheenamel,andresultinginelevatedUconcentra-tionsinspotslocatedclosetotheEDJ( Figure9 ).TheUuptakehistoryappearscomplexandheterogeneous,andprobablyinvolvedseveralepisodes.MostoftheUconcentrationsintheenamelaretoolowtoprovideameaningfulage.However,partsoftheenamelandthedentineyielded Figure4. Locationofthethree H.naledi toothsamples(samples1767,1788and1810)andonebaboon(cf. Papio ) toothsample(sample1841)usedforcombinedU-seriesandESRdating.( a )MapoftheDinalediChamber showingthepositionoftheexcavationpitandthepositionoffigures( b )and( c );( b )close-upoftheSEcornerof theexcavationpitshowingthesamplesiteforsample1810andthe50cmdeepsondagefromwhichsample1841wasrecovered;( c )theareatotheWoftheexcavationpitfromwhichsamples1767and1788werecollected. DOI:10.7554/eLife.24231.011 Dirks etal .eLife2017;6:e24231. DOI:10.7554/eLife.24231 21of59 Researcharticle GenomicsandEvolutionaryBiology

PAGE 22

consistentmeasurementsforwhichSCU-UoWprovideageswithmeanvaluesof71.5 ± 0.6kafor dentineand67.0 ± 5.4kaforenamelwithinitial234U/238Uactivityratiosof7.62 ± 0.03and 6.94 ± 0.17respectively.GU-ANUobtainedacombinedageof95.1 ± 3.2kafortwoenamelspots withhigherUconcentrations,withpartsoftheenamelwithlowerU-enrichmentyieldingacombinedageof56.2 ± 6.5ka.GU-ANUalsoprovideaconsistentmeanapparentageof87.9 ± 2.6ka,associatedwithinitial234U/238Uactivityratiosof7.98 ± 0.06(individualspotsagreeingwithinerror)for dentinewhichisinterpretedastheageofanUuptakeevent. Sample1810: Thislowerleftmolarorpremolarfromtheexcavationpit( Figures4 and 6c ),is nearcompleteandonlymoderatelyweathered.UraniumdiffusionpatternsshowUaccumulatingattheEDJwithslowdiffusionintotheenameltissue.TheUconcentrationsinmostoftheenamelaretoolowtocalculateameaningfulage.AreasofenamelwithhigherUconcentrationsreturnolderages( Tables4 and 5 ).SCU-UoWprovideameanageforhigh-Uspotsinenamelof128.8 ± 7.2ka andanassociatedinitial234U/238Uactivityratioof7.60 ± 0.16.GU-ANUreportameanapparent U-Thageof195.7 ± 13.8ka,whichismuchhigherthantheadjacentdentinespots(at81.1 ± 2.7ka), butiscoupledwithrealisticinitial234U/238Uactivityratiosof7.24and7.84withoverlappingerror limits.ThisindicatesthatasecondaryoverprintofthedentinetookplaceforwhichtheUsourcehadasimilar234U/238UcompositionasthesourceoftheinitialUuptakeevent.Dentinemeasurements areconsistentalongthemeasuredsectionswithsmallregionsaffectedbyleachingandenrichmentnearcracksandthepulpcavity.ThedentineanalysesdonebySCU-UoW( Table4 )yieldsimilarages withconsistentinitial234U/238Uactivityratiosofaround7.04 ± 0.03,andameanapparentageof 72.3 ± 1.0ka.ThecombinedanalyticaldatafordentinefromGU-ANUinsamples1810Aand1810B yieldapparentU-Thagesof80.1 ± 2.5kaand82.1 ± 2.7karespectively,coupledwithconsistentinitial234U/238Uactivityratios( Table5 ). Sample1841 Thebaboontoothconsistsofanenamelcrownthatisstructurallyintact,butthe enamelisfriableandweathered( Figure7 ).TheUdistributionwithintheenamelappearshomogenous,however,Thconcentrationsarelowandtheresolutionoftheelementaldistributionispoor,whichimpairsthequalityoftheU-Thageestimates.ArecentUuptakeeventmayhaveoccurredaffectingenamelincontactwithsediment,resultinginameanapparentU-Thageestimateof115.6 ± 1.9kawithameaninitial234U/238Uactivityratioof7.76 ± 0.05( Table4 ). GU-ANUalsoprovideageestimatesinwhichthecontinuousdiffusionassumptionsof Sambridgeetal.(2012) havebeenapplied.Theresultsobtainedforsamples1788and1810are Figure5. SamplesoforangelaminatedmudstoneofUnit1forOSLdating.( a )sampleOSL3withanestimated MAMageof231 ± 41katakenfromsub-unit1b;( b )sampleOSL4withanestimatedMAMageof241 ± 37ka, takenfromsub-unit1bandcoveredbyaflowstonesheetdatedat290 ± 6ka(RS5);( c )sampleOSL5withan estimatedMAMageof353 ± 61ka,takenfromsub-unit1aandcoveredbyaflowstonesheetdatedat32.1 ± 0.4 ka(RS20).Thescalebarineachofthephotographsis10cm. DOI:10.7554/eLife.24231.012 Dirks etal .eLife2017;6:e24231. DOI:10.7554/eLife.24231 22of59 Researcharticle GenomicsandEvolutionaryBiology

PAGE 23

about20%olderthantheclosedsystemages( Table5 ),butshowmuchlessconsistencyandarenot furtherconsidered. Collectively,theresultsshowthattheteethareolderthan70ka,andconsideringsample1810 withaminimumageofaround200ka,thatthe H.naledi fossilsareprobablyolderthan200ka(see Discussion).US-ESRdatingCombinedU-seriesandESRdating(US-ESR; Gru È netal.,1988 )ofthreehomininteeth(samples 1767,1788and1810)andababoontooth(sample1841)wasperformedatSCU.Blindduplicateanalysesoftwoofthehomininsamples(samples1788and1810)wereperformedatthe`CentroNacionaldeInvestigacioÂnsobrelaEvolucioÂnHumana'(CENIEH),SpainincollaborationwithGU(CENIEH-GU).Inobtainingtheages,eachlaboratorycarriedoutindependentsamplepreparation,andESRandU-seriesanalysesofthefossilteeth.Estimatesfortheenvironmentaldoseratesusedintheagecalculationswerestandardizedforbothlaboratories( Table6 )inordertoproducecomparableresults(seediscussionandmethodologysectionsfordetails). Figure6. Photographsof H.naledi teethusedforESRdating.( a )U.W.101±1767;( b )U.W.101±1788;( c )U.W.101± 1810.Theorderofimagesforeachpanelis:buccal,distal,lingual,mesial,andocclusalviews.Thescalebarineachpanelis1cm. DOI:10.7554/eLife.24231.013 Dirks etal .eLife2017;6:e24231. DOI:10.7554/eLife.24231 23of59 Researcharticle GenomicsandEvolutionaryBiology

PAGE 24

Analysesandresultsfrombothlabsarepresentedin Table7 and Figures10 , 11 and 12 .Results arepresentedfortwoscenarios:scenario1inwhichtheteetharefullycoveredinsedimentthatcon-tains25 ± 10%waterandexperienced80%Rnloss;andscenario2inwhichtheteetharefullycoveredinsedimentthatcontains25 ± 10%waterandexperienced noRnloss.Scenario1reflectsthe measuredpresent-daysituationandisinterpretedasamaximumageestimate.Scenario2providesaminimumageestimate( Table7 ).Togetherthesescenariosprovideourbestestimatefortheage rangeofthefossilteeth. CombinedUS-ESRagesdeterminedbySCUforsamples1810,1788and1767underscenario1 conditions(i.e.,themaximumagescenario)are284 ± 51ka,247 ± 41kaand104 ± 29ka(2 s uncertainty),respectively( Table7 ).CombinedUS-ESRagesdeterminedbyCENIEH-GUforsamples1810 and1788underscenario1conditionsare267 ± 68kaand211 ± 28ka(2 s uncertainty),respectively ( Table7 ). CombinedUS-ESRagesdeterminedbySCUforsamples1810,1788and1767underscenario2 conditions(i.e.,theminimumagescenario)are230 ± 40ka,194 ± 34kaand87 ± 22ka(2 s Figure7. Photographsofthebaboon(cf. Papio )tooth(sample1841),recoveredfromthesondageinthe excavationpit,andusedforESRdating.Viewsare:( a )buccal,( b )occlusal,( c )lingual,and( d )internal. DOI:10.7554/eLife.24231.014 Dirks etal .eLife2017;6:e24231. DOI:10.7554/eLife.24231 24of59 Researcharticle GenomicsandEvolutionaryBiology

PAGE 25

uncertainty),respectively( Table7 ).CombinedUS-ESRagesdeterminedbyCENIEH-GUforsamples 1810and1788underthisscenarioare210 ± 50kaand163 ± 24ka(2 s uncertainty),respectively ( Table7 ). Resultsforsample1767arebasedonanomalouslyhigh( ~ 20times)Uconcentrationsinenamel, andprobablyyieldanomalouslylowageestimates( Duvaletal.,2012 ).Thisresultis,therefore,consideredtobeunreliableandhasbeenexcludedfromfinalageestimates(seeDiscussion). TheobserveddifferenceinageestimatesobtainedbySCUandCENIEH-GUforsamples1788 and1810,aremostlikelyexplainedbynaturaldosevariationswithinthetestedenamellayers(seeDiscussionandmethodologysections),andwehavenoreasontopreferoneageresultoveranother.Theoptimalageestimateforthe H.naledi fossils,therefore,combinestheresultsfrombothlaboratorieswithaveragemaximum(i.e.,scenario1)ageestimatesforsamples1788and1810of229+60/±46kaand276+59/±77ka(2 s uncertainty)respectively,andaverageminimum(i.e.,scenario2)ageestimatesof179+49/±40kaand220+50/±60ka(2 s uncertainty)respectively. Togethertheseresultsprovideanagerangeof139±335kaforthe H.naledi remains,althoughdatingofflowstoneencasing H.naledi boneshelpstobetterconstrainthisrange(seeDiscussion). CombinedUS-ESRagesforthebaboontooth(sample1841)determinedbySCUusingscenario1 and2conditionsare723 ± 181kaand635 ± 148karespectively.Thetoothcontainednoinnerdentine( Figure7 ),andwasfilledwithsediment.Incalculatingtheageitwas,therefore,assumedthat sedimentoccurredonbothsidesoftheenamellayer.Sample1841wasrecoveredfromsub-unit3adirectlybelowtheoccurrenceofarticulated H.naledi remainsintheexcavationpit( Dirksetal., 2015 ).Theageresultsprovideanupperagelimitforthedepositionofthe H.naledi bearinglayer, andmarkanearlierstageofdepositionofmudclastbrecciainthecaveassignedtosub-unit3a,whichpredatestheentryof H.naledi fossilsintothecave. OSLdatingOpticallystimulatedluminescence(OSL)datingofthreesamplesofsedimentfromUnit1intheDinalediChamber(samplesOSL3andOSL4fromsub-unit1b,andsampleOSL5fromsub-unit1a; Figure5 )wasperformedattheUniversityoftheWitwatersrand(Wits).Themeasurementswerecarriedoutonsmallaliquotscontaining ~ 30grains.SummariesoftheOSLanalyticaldataandagesare reportedin Table8 . Thereporteddoseratesforthesamplesrangefrom0.7to0.9Gyka 1( Table8 ),withsignificant within-samplescatter,resultinginuncertaintiesonageestimatesof15±18%.OverdispersioninDerangesfrom50±70%,whichismuchhigherthanwouldbeexpectedforawell-bleachedsample,andindicatesthatitismostappropriatetoapplyaMinimumAgeModel(MAM)tothedataset,inwhichtheMAMageislikelytorepresentamaximumestimatefortheageofthesediments(seeDiscus-sion).AswithESR,significantRnlosswasdetectedinthesamplesofUnit1,andcorrectionstothemeasuredUconcentrationswereapplied( Table8 ).TheMAMcalculationsforthethreesamples yieldmaximumageestimatesforthesedimentsof231 ± 41ka(OSL3),241 ± 37ka(OSL4),and 353 ± 61ka(OSL5).TheMAMapparentagesforOSL3andOSL4wereobtainedfromthesandy faciesofsub-unit1bsedimentsandyieldagesthatareyoungerthantheageofaFlowstone1sheet(sampleRS5at290 ± 4ka; Table1 )thatcoverstheoutcropofsub-unit1afromwhichsampleOSL4 wastaken( Figure2c ).Thisdiscrepancycanbeattributedtoinvertedstratigraphyassociatedwith erosionofthetopoftheoldersub-unit1aafterthedepositionofFlowstone1(RS5)byrunningwaterandsubsequentdepositionofsub-unit1bbetweensub-unit1aandtheflowstone( Figure2c ). SampleOSL5wasobtainedfrommuddysedimentofsub-unit1a,andyieldsanolderagethanthecoveringflowstones(RS5andRS20).NotethatifaCentralAgeModel(CAM)isappliedtotheOSLdata,resultsaresignificantlyolderat560 ± 102ka(OSL3),546 ± 79ka(OSL4),and849 ± 132ka (OSL5),however,thismodelisconsideredunrealisticwithinacaveenvironment( Galbraithetal., 1999 ). PalaeomagneticanalysisofflowstonePalaeomagneticanalysisofonecompositesampleofFlowstone1a( Figure13 ),coveringerosional remnantsofUnit2neartheentryshaftintotheDinalediChamber,wasperformedatLatrobeUni-versity,Melbourne(LTU).ThepalaeomagneticresultsforFlowstone1aarepresentedin Figure13 and Table9 .ThepalaeomagneticsamplefromFlowstone1acomprisesthreedistinctphasesof Dirks etal .eLife2017;6:e24231. DOI:10.7554/eLife.24231 25of59 Researcharticle GenomicsandEvolutionaryBiology

PAGE 26

flowstoneformation(frombasetotop:A,BandC; Figure13a,b )thathavebeenanalysedfortheir palaeomagneticorientation( Figure13c ).NotethatthepalaeomagneticsampleofFlowstone1a wastakenup-dipfromthepositionwheretheU-ThsampleofFlowstone1awastaken(i.e.,RS22andRS23).Thebasal,phaseAflowstoneobservedinthepalaeomagneticsampletapersoutdown-dip( Figure13a )andisnotpresentintheU-Thsample;thus,RS23atthebaseoftheU-Thsample correspondstophaseBcarbonateinthepalaeomagneticsample,andRS22tophaseCcarbonate( Figure2b ). PhaseAflowstoneatthebaseofthesamplerecordsaconsistentreversedmagneticpolarity, withaclearoverprintthatisanti-paralleltothecharacteristic,reversedpolarityremanence(ChRM)andisremovedby ~ 10mT.ThereversedpolarityChRMisthenstablebetween10and40mT, althoughwithinthisrangetherearetworeversedpolaritycomponentsthatcanbeidentifiedinsomesamples(between10and19mTandthenbetween20and40mT; Figure13c ).Bothcomponentshavealmostidenticaldeclinationvalues,butthelowerfieldcomponenthasashallowerinclina-tion.Thisisnotuncommonwithinspeleothems( HerriesandShaw,2011 )becauseasingle subsampleof2.5cmdepthismeasuringtheremanencerecordedinmultiplelayersofspeleothemaswellasmultiplelayersofdetritalcontamination.PhaseBflowstonerecordsaweak,normalpolar-ityChRMthatisconsistentlyisolatedbetween7and36mT( Figure13c ).Theresultsfromthisphase havethehighestmeanmaximumangulardeviation(MAD)values( Table9 )duetothesmalland oddlyshapednatureofthesamplesthatprovidelessconsistentmeasurementsbetweeneachspininthemagnetometer.PhaseCflowstonealsorecordsanormalChRMthatissimilartothatseenwithinphaseB( Figure13c ),althoughwithslightlysteeperinclinations( Figure13c ,StereoPlot). ThesamplesdonothavestrongsecondaryoverprintsasseeninmanyPlio-Pleistocenepalaeocavedepositsfromtheregion( Dirksetal.,2010 ),whichmayindicatethattheyareyoungerorhave adifferentsedimentsourceand,thus,mineralogyholdingtheremanence.ThecoercivityofphaseBandCflowstonesissimilar,anddistinctfromphaseAflowstone,althoughallthedemagnetisationspectrasuggestthedominantmineralholdingtheremanenceisferrimagnetic(magnetite/maghae-mite)asatmanyotherCoHsites( Herriesetal.,2006 ; HerriesandShaw,2011 ; Herriesetal., 2014 ). RadiocarbondatingThreeweatheredbonefragmentsof H.naledi wereanalysedviaradiocarbondatingatBetaAnalytic Inc.(Florida,USA).Analysesindicatedthatnocollagenwaspresentinanyofthesamplesandthattheboneappearedpossiblycremated.Thiswasinvestigatedwithabonecarbonateextractiontech-nique.TestsdidnotsupportcremationandindicatedthatextensivesecondaryCaCO3replacement hadoccurred,providingagesof33.0 ± 0.2kaand35.50 ± 0.16kafortwoofthefragments.We interprettheseagestorelatetolatecalciteprecipitationinthebonesthatmayreflectawetperiodinthecave. Discussion Thestrategyindatingthefossilshasbeenbuiltonthreecomponents:(i)constructadetailedstrati-graphicmodelforthecavesedimentsintheDinalediChamberinwhichthepositionofthehominin-bearingdepositscanbesecurelyplaced;(ii)datethesedimentaryunitsthatpotentiallybracketthehominin-bearingdeposits;and(iii)datethe H.naledi fossilsdirectly.Wewillstartthisdiscussionby commentingonthereliabilityofthevariousdatingtechniques,andhowtheyshouldbeviewedrela-tivetoeachother.Wewillthendiscusstheoutcomesofeachofthethreestrategiccomponents,andtheirimplicationsfortheageofthe H.naledi material.Thebroaderimplicationsoftheagefor themorphologicalevolutionofhomininsinsouthernAfricaisdiscussedindetailin Bergeretal., 2017 ,andwillbebrieflysummarizedattheendofthisdiscussion. ReliabilityoftheageestimatesThedatingtechniquesappliedinthisstudydonotallworkinthesameway,andhencetheresultsmustbevieweddifferently.U-seriesanalysisoncarbonates,14Canalysisofboneandpalaeomagneticanalysesofflowstones(e.g., HerriesandShaw,2011 ; TaylorandBar-Yosef,2014 ; HellstromandPickering,2015 )arewellestablisheddatingtechniquesrequiringfew,ifany,aÁpriori Dirks etal .eLife2017;6:e24231. DOI:10.7554/eLife.24231 26of59 Researcharticle GenomicsandEvolutionaryBiology

PAGE 27

assumptions.Incontrast,ESRandOSLresultsarestronglydependentonmodelassumptionsfortheenvironmentalconditionsthataffectedthelocationsfromwhichthesamplesweretaken.U-ThresultsSamplepreparationandanalyticalproceduresforU-Thdisequilibriumdatingarewellestablished(seemethodologysection),andarelikelytoreturnhighlyreproducible(i.e.,highprecision)results.Thisiscertainlythecasehere,wheresamplesanalysedinduplicateinindependentlaboratories(JCUandUoM)returnedidenticalresultswithinanalyticaluncertainties( Tables1 , 2 and 3 ).Theonly exceptionisthatUoMwasabletoprovideagesforolderflowstonesthatdateclosetothedetectionlimitforthetechnique,whereasJCUdidnotprovideagesover400ka. Althoughtheseresultsdemonstratehighlevelsofprecision,furtherevaluationofthegeological meaningoftheagesisrequired.Inparticular,theeffectsofpossiblepost-depositionalUuptake,whichwouldresultinapparentagesthatareyoungerthanthetrueage,canbeassessedwithtex-turalanalyses(e.g., Pickeringetal.,2010 ),andtheuseofinitial234U/238Uratios( Tables2 and 3 ; Kronfeldetal.,1994 ).Mostoftheanalysedflowstonesdisplaysimilarinitial234U/238Uratios(1.8 to2.4),withtwonotableexceptionsinsamplesRS18(243 ± 7ka)andRS5(290 ± 6ka)thatrecord234U/238Uratiosof3to4.Thesearetwooftheoldestsamplesanalysed,andbothdisplayevidence forrecrystallizationandsecondarydissolution.ConsideringthatthegroundwaterreservoirintheCoHhashigh(i.e.,>7)initial234U/238Uratios( Kronfeldetal.,1994 )itislikelythattheelevated234U/238UratiosinsamplesRS5andRS18areduetoUuptakeduringinteractionwithgroundwater afterflowstonedeposition.Inthiscase,thesetwoageresultsshouldbeviewedasminimumageestimatesfortheflowstones. Aswithflowstone,analyticalproceduresforU-Thdisequilibriumdatingofteetharewellestablished(e.g., Gru È netal.,2014 ).However,inthisstudytherearesignificantdifferencesintheU-Th disequilibriumagesofteethreportedbythetwolaboratories( Tables4 and 5 ).Thisdifferencein measuredagesismostlikelyduetothefactthatthedifferentlaboratoriesdateddifferentfragmentsoftheteeth.TheU-ThanalysesshowthattheUdistributionwithintheteeth,andespeciallywithinenamel,ishighlyvariable,asreflectedinthewiderangeofU-Thages( Tables4 and 5 ).Wherea toothdisplaysconsistentagesandinitial234U/238Uratiosoveralargedomain,orwherewehave observedclosecoincidenceinU-Thagesandinitial234U/238Uratiosfordentineandadjacentenamel samples,weinterprettheresultstosuggestthataUuptakeeventhasbeendated.ConsideringthevariabledistributionofUwithinthetoothsamples,itispossiblethatthesametoothrecordsmorethanoneUuptakeeventindifferentdomainswithinthetooth(mostnotablywhencomparingenamelvsdentinedomains; Gru È netal.,2014 ).SinceThisimmobile,eachUuptakeeventthat affectsthetoothwillshiftpre-existingU-Thsystematicstoportrayayoungerage( Gru È netal., 2014 ).ThevariableU-ThdisequilibriumagesareaclearindicationthatUuptakeeventstookplace, andprovideaminimumageestimateforthetrueageoftheteeth.Whereapparentagesareconsis-tentacrossmuchofthetooth(e.g.,sample1767)theymayapproachthetrueageoftheUuptakeevents.Theresultspresentedinthisstudyshowthatthehomininteethareclearlyolderthan70ka,asevidencedfromtheconsistentresultsonthedentineofsamples1788and1810.Animportantresultfromthismethodistheolderclosedsystemage(ca.200ka)fromenamelobtainedforsample1810BbyGU-ANU( Table5 ).ThisresultindicatesthattherewasanearlyphaseofUuptakeinthe tooththatwaslateroverprintedintheassociateddentine.Inthiscase,sample1810hasaminimumageofaround200ka,suggestingthatthe H.naledi fossilsareolderthan200ka. US-ESRandOSLresultsESRandOSLdatingtechniquesaredosimetricdatingtechniques( AdamiecandAitken,1998 ; MurrayandWintle,2000 ),anddeliverresultsthataregenerallymuchmorevariablethanU-series datingduetotherelativelackofcontroloftheconditionsunderwhichthesamplesaccumulatedradiationdamage.Inaddition,analyticalprotocolsbetweenlabsmayvarysignificantly(e.g.,SCUusedanX-raygunwhilstCENIEH-GUusedgamma-raysastheirirradiationsource),andfinalresultsareheavilydependentonarangeofaÁpriorimodelassumptions(suchaswatercontent,Rnlossorburialhistorythroughtime).ForOSL,similaraÁpriorimodelassumptionsmustbemade,andtheinterpretationofresultsisfurtherhinderedbytheuseofcompositesamplesinwhichindividualgrainsmayhaveexperienceddifferentburialhistories( Galbraithetal.,1999 , 2005 ).Singlegrain Dirks etal .eLife2017;6:e24231. DOI:10.7554/eLife.24231 27of59 Researcharticle GenomicsandEvolutionaryBiology

PAGE 28

analysesarepossible( Duller,2008 ),butfarmorelabourintensiveand,forthepurposeofthisstudy, werenotrequired.Moreover,thisisthefirsttimethatOSLhasbeenappliedtocavedepositsintheCoH,andthereisnocomparativedataavailableforthisarea.However,theresultsareconsistentwiththeinterpretedstratigraphyandotherdatinginthechamber,andhighlightthepotentialforadditionalwork. ToassurethatallUS-ESRresultsobtainedbythedifferentlaboratoriescanbecomparedinan objectivemanner,westandardizedtheaÁpriorimodelassumptions( Table6 ),thatis,thesameanalyticalresultsforsedimentchemistry,backgroundradiation,watercontent,radon-lossandcosmo-genicradiation,neededtocalculatetheenvironmentaldoserate,wereusedbybothlaboratories.Therefore,anydifferencesinageestimatesrelatetothemeasuredequivalentdose(DE)andlaboratorymethodologies,andarenottheresultofmodelassumptions.Bothlaboratoriesalsoappliedthesamecombinationsofmodelparametersthatwouldmostlikelyresultineithermaximumormini-mumageestimates.Thesearepresentedastwoscenariosinwhichthedeterminedagerangewilloverlapwiththetrueageofthefossils( Table7 ). WhenexaminingtheUS-ESRresultsfortheteeth( Table7 ),twotrendsareapparent:(i)The resultsforsamples1810and1788fromSCUandCENIEH-GUagreewellwithinthelisteduncertain-ties;and(ii)Thecalculatedagesforsample1767aremuchyoungerthantheotherteeth.Regardingthesecondissue,samples1767and1788werecollectedfromthetopofsub-unit3bwithin1.3mfromoneanother,withthedifferencethatsample1767laydirectlyonsurfaceinthecentreofthefloor,whereassample1788laytothesideunderacoverof2cmofloosesediment( Figure4 ).Given theirpositioninthecavetheteethareexpectedtobeofsimilarage.However,theiragesasobtainedbySCUvarybyafactorof>2(scenario1:104 ± 29vs247 ± 42ka;scenario2:87 ± 22ka vs194 ± 34ka; Table7 ).Themaindifferencebetweenthetwosamplesisthattheinternaldoserate ofsample1767is ~ 20timeshigher( Table7 ),duetothehighUconcentrationintheenamel( ~ 2.5 ppm).Suchhighconcentrationscanleadtodoserateoverestimationsduetoinappropriatealphaefficiencyvalues( Duvaletal.,2012 ).ThehighUvaluesinsample1767areprobablytheresultofits greaterexposuretowater,confirmedbyitsextremelyweatherednature( Figure6a ).Giventhese factors,wedonottrustthereliabilityoftheageresultsforsample1767andhaveexcludedthistoothfromthefinalageestimatesforthefossils. TheageestimatesfromthemaximumandminimumagescenariosascalculatedherearedependentontheamountofestimatedRnlossovertime,thatis,inassessingthebestageestimateforthe H.naledi fossilsamajorissueiswhether222Rndegassingwasaprocessthatoperatedcontinuouslyoverthepast300kaornot.222Rnisanoblegas,radio-isotopethatformsaspartofthe238U decaychain,anditsescapehasamajoreffectonthetotalamountofgammaradiationgeneratedbythesediments(e.g., Gue  rinetal.,2011 ),andconsequently,thecalculatedUS-ESRage( Table7 ). Thereasonwhy222Rndegassingoccurssoreadilyprobablyrelatestothecrystallographicposition occupiedbyUatoms.UraniumandTharechieflyhostedinverythinFe-Mnoxy-hydroxidecoatingsongrainswithinthesediments( Dirksetal.,2015 ; Makhubelaetal.,2017 ),andgiven(alpha)-recoil inthe226Radecay,escapeof222Rn(half-life3.82days)islikely.Thisprocesswillhaveoccurred throughoutthehistoryofthecavegiventhattheformationofFe-Mnoxy-hydroxidesisanintegralpartofthechemicalprocessthatoccurswithintheunconsolidatedcavesedimentsasaresultofoxi-dation,auto-brecciationandweatheringreactions( Dirksetal.,2015 ).Itis,therefore,assumedthat effectiveRndegassingwillhavebeenanimportantprocessinthecavethroughouttheaccumulationhistoryoftheUnit3sediments,althoughitseemsunlikelythatanextremedegassingenvironmentsimilartothe80%measuredtodaywasmaintainedduringtheentirehistoryofburial.Therefore,weinterpretthemaximumUS-ESRageestimatescalculatedforscenario1conditionstobetheclosesttothetrueageofthe H.naledi fossils.IfRndegassingwassomewhatlessinthepast,theolderage bracketwouldmoveintothedirectionof270ka(i.e.,themaximumagelimitcalculatedundersce-nario2conditions). TheOSLresultsaremoredifficulttointerpretthantheUS-ESRresults,notonlybecausewehave toassumemodelparameterstoestimatetheenvironmentaldoserate,butalsobecausewehavefewconstraintsontheoriginandprovenanceofthequartzgrainsthatweresampled.OSLanalyseswerecarriedoutforaliquotsof ~ 30grainsandeachanalysis,therefore,isanaverageofthesignals containedinthegrains.Thereissignificantwithin-samplescatterinmeasuredtotaldose(De)values betweendifferentaliquots,meaningthatageestimatesareimprecise(uncertaintiesof15±18%).OverdispersioninDerangesfrom50±70%,whichismuchhigherthanwouldbeexpectedforawellDirks etal .eLife2017;6:e24231. DOI:10.7554/eLife.24231 28of59 Researcharticle GenomicsandEvolutionaryBiology

PAGE 29

bleachedsample( Galbraithetal.,1999 ),andisareflectionofthecaveenvironmentfromwhichthe samplesweretaken( Duller,2008 ).Weinterpretthistomeanthatthequartzgrainsineachaliquot consistofamixtureofgrainswithsomederivedfromoutsidethecave(andcarriedintothecaveoveraperiodoftime),andothersderivedfromquartzveinsandchertbedswithinthecavethathaveonlybeenpartlybleachedorhaveneverbeenbleachedatsurfaceatall.Addtothisthenaturalsignalvariabilitythatcanoccurduetocrystallographicorientationandlatticedefects( Olleyetal., 2004 ; Galbraithetal.,1999 , 2005 )andalargeoverdispersionresults.TheMinimumAgeModel (MAM)waschosenasthepreferredstatisticalmodeltocalculateageestimatesforthesediments.Thisisastatisticalaveragingtechniqueforaliquotsthatdisplayahighdegreeofwithin-samplescat-ter( Olleyetal.,2004 ; Galbraithetal.,2005 ),butstillcombinessomegrainsthatarewell-bleached withothersthatareonlypartlybleached.Thus,theMAMageislikelytoprovideamaximumageestimatefortheUnit1sediments,andmustbeviewedwithcaution.Amorecomprehensivesam-plingcampaignwillberequiredinfuturetofullyestablishtherangeofagesforUnit1acrossthechamber.Nevertheless,thepreliminaryresultsareconsistentwiththerangeofagessuggestedforUnit3byESRandU-Thanalysesoncappingflowstones( Figure14 ). AnupdatedstratigraphyfortheDinalediChamberTheimportanceofadeepunderstandingofthestratigraphicpositionofthefossilsandthegeologi-calprocessesthatledtotheirdepositioncannotbeoverstatedconsideringtheextremelycomplex Table6. SummarytableofmodelparametersusedinESRdatingseparatedbysamplenumberandlaboratory.Seetextfordetailed discussion.Sample:1767178818101841Laboratory:SCUSCUCenieh-guSCUCenieh-guSCU Enamel: De(Gy) 194 ± 4231 ± 8159 ± 10296 ± 14232 ± 301676 ± 127 U(ppm) 2.52 ± 0.530.38 ± 0.170.07 ± 0.070.32 ± 0.120.16 ± 0.162.28 ± 0.48234U/238U 6.21 ± 0.035.95 ± 0.326.258 ± 0.3494.04 ± 0.184.773 ± 0.0605.87 ± 0.03230Th/234U 0.37 ± 0.050.55 ± 0.520.598 ± 0.0380.92 ± 0.050.950 ± 0.0340.785 ± 0.038 Alphaefficiency* 0.13 ± 0.020.13 ± 0.020.13 ± 0.020.13 ± 0.020.13 ± 0.020.13 ± 0.02 Initialthickness( m m) 1027 ± 2101049 ± 2771486 ± 2481150 ± 2501527 ± 257650 ± 145 Water(%) 000000Dentine: U(ppm) 7.88 ± 0.665.76 ± 0.864.71 ± 0.279.08 ± 0.445.81 ± 0.37±234U/238U 6.28 ± 0.096.40 ± 0.036.448 ± 0.0465.93 ± 0.035.969 ± 0.035±230Th/234U 0.35 ± 0.110.62 ± 0.020.608 ± 0.0120.52 ± 0.090.572 ± 0.010± Water(%) 10 ± 510 ± 510 ± 510 ± 510 ± 5± Sediment: U(ppm) 3.0 ± 0.32.9 ± 0.12.9 ± 0.13.2 ± 0.33.2 ± 0.30.64 ± 0.06²Th(ppm) 7.9 ± 0.48.3 ± 0.68.3 ± 0.68.6 ± 0.48.6 ± 0.44.72 ± 0.47²K(%) 1.17 ± 0.141.21 ± 0.141.21 ± 0.141.23 ± 0.141.23 ± 0.141.47 ± 0.15²Water(%) 25 ± 1025 ± 1025 ± 1025 ± 1025 ± 1025 ± 10 Depthbelowgroundsurface(cm) 0225555 GammaDoserate( m Gya 1) 25 10%Water,80%Rndegassing 25 10%Water,noRndegassing 534 ± 69 724 ± 116 534 ± 69 724 ± 116 534 ± 69 724 ± 116 534 ± 69 724 ± 116 534 ± 69 724 ± 116 534 ± 69 724 ± 116 Cosmicdoserate( m Gya 1) 15 ± 115 ± 115 ± 115 ± 115 ± 115 ± 1 *After Woodroffeetal.(1991) ;²Arelativeerrorof ± 10%wasassumed. DOI:10.7554/eLife.24231.015 Dirks etal .eLife2017;6:e24231. DOI:10.7554/eLife.24231 29of59 Researcharticle GenomicsandEvolutionaryBiology

PAGE 30

natureofsedimentarycavefillinmanycavesystems(e.g., Wilkinson,1985 ; Brain,1993 ; Sasowsky,1998 ; Stocketal.,2005 ; Stratfordetal.,2014 ; Sutiknaetal.,2016 )involvingrepeatedcycles ofdeposition,erosionandreworking,leadingtocomplexandsometimescontradictoryageresults(e.g., Grangeretal.,2015 ; KramersandDirks,2017 ).ThisproblemiswellillustratedwiththeongoingdebateontheageofStw573(`littlefoot')inthenearbySterkfonteinCave,whereafter20yearsofdatingeffortsnodefinitiveageisyetestablished(see Partridgeetal.,1999 , 2003 ; Bergeretal.,2002 ; Walkeretal.,2006 ; HerriesandShaw,2011 ; PickeringandKramers, 2010 ; Grangeretal.,2015 ; KramersandDirks,2017 ).Anothergoodexampleillustratingthedifficultiesoflinkingcavestratigraphytoadefinitiveageforthehomininfossilstheycontainispresentedbythe H.floresiensis remainsintheLiangBuacave,Indonesia( Morwoodetal.,2004 ; Robertsetal.,2009 ; Sutiknaetal.,2016 ). ThestratigraphywithintheDinalediChamberhasbeenpreviouslydescribedby Dirksetal. (2015) .Theagespresentedherehelptoresolveoutstandingquestionsaboutthestratigraphyinthe DinalediChamber,andallowustomorecloselydefinethedistributionofcorrelativestratigraphicunits( Figure14 ),andthusconstraintheageofthe H.naledi fossils. Table7. SummaryofESRdatingresults(2 s uncertainties)fortwoend-memberscenarios:(i)completeburialofthesamples,80%Rn lossinthesedimentandpostTh-230equilibriumindentaltissue(i.e.,maximumagescenario);(ii)completeburialofthesamplesandpost-Rnequilibriuminsediment(i.e.,minimumagescenario).Seetextfordetaileddiscussion.Sample: 1767178818101841 Laboratory:SCUSCUCenieh-guSCUCenieh-guSCU Scenario1:25 ± 10%Water,completeburialand80%222Rndegassing(maximumagescenario) internaldoserate( m Gya 1)1142 ± 515190 ± 12947 ± 47323 ± 175176 ± 1761411 ± 596 alpha( m Gya 1)*008 ± 208 ± 20²betadoserate,dentine( m Gya 1)73 ± 3391 ± 6264 ± 1675 ± 4151 ± 14±³betadoserate,sediment( m Gya 1)101 ± 24105 ± 3186 ± 1795 ± 2486 ± 18358 ± 74 gammaandcosmic( m Gya 1)549 ± 69549 ± 69549 ± 69549 ± 69549 ± 69549 ± 69 totaldoserate( m Gya 1)1865 ± 521935 ± 162754 ± 871042 ± 194870 ± 1902318 ± 606 penamel 0.030.49 0.02 0.70 0.770.91 pdentine0.080.13 0.061.020.54± Age(ka) 104 ± 29247 ± 42211 ± 28284 ± 51267 ± 68723 ± 181 CombinedSCU/CENIEH-GUage(ka)229+60/±46276+59/±77Averageagefor1788&1810(ka)253+82/±70 Scenario2:25 ± 10%Water,completeburialandno222Rndegassing(minimumagescenario) internaldoserate( m Gya 1)1277 ± 552216 ± 16551 ± 51335 ± 193184 ± 1841520 ± 630 alpha( m Gya 1)*008 ± 208 ± 20 betadoserate,dentine( m Gya 1)82 ± 35102 ± 7869 ± 1887 ± 5059 ± 16 ± betadoserate,sediment( m Gya 1)132 ± 26134 ± 33111 ± 19126 ± 26112 ± 19380 ± 81 gammaandcosmic( m Gya 1)739 ± 116739 ± 116739 ± 116739 ± 116739 ± 116739 ± 116 totaldoserate( m Gya 1)2230 ± 5861191 ± 219978 ± 1291287 ± 2321102 ± 2192639 ± 647 penamel 0.310.06 0.37 0.83 0.910.67 pdentine 0.22 0.22 0.400.540.10± Age(ka) 87 ± 22194 ± 34163 ± 24230 ± 40210 ± 50635 ± 148 CombinedSCU/CENIEH-GUage(ka)179+49/±40220+50/±60Averageagefor1788&1810(ka)200+70/±61 *usingalphaattenuationvaluesof Gru È n(1987) .²consideredasnegligiblegiventhelowradioelementconcentrationsinthesedimentandthehightotaldoseratevalue.³for1841,thebetadoserateonbothsidesoftheenamellayerisderivedfromthesediment. DOI:10.7554/eLife.24231.016 Dirks etal .eLife2017;6:e24231. DOI:10.7554/eLife.24231 30of59 Researcharticle GenomicsandEvolutionaryBiology

PAGE 31

TheoldestagesreturnedfromtheDinalediChamberarefromthebaboontooththatisembeddedinsedimentattributedtosub-unit3a,followedbyU-ThagesforFlowstone1a,whichdirectlycoverserosionalremnantsofUnit2( Figure14 ).Flowstone1aconsistsofatleastthreegenerations offlowstonegrowth,named,fromoldesttoyoungest,phaseA,BandC( Figures2b , 13a,b and 14 ).WedatedphasesBandCviaU-Th,at502ka(RS23; Table1 )and478ka(RS22; Table1 ) respectively,buttheuncertaintiesarelargebecausetheagesareclosetotheupperdatinglimitoftheU-Thtechnique.Theoldest,phaseAlayeratthebaseofFlowstone1arecordsreversedmag-neticpolarityindicatingthatitformedbefore780ka( Singer,2014 ).Thethreeageestimatesare consistentwiththestratigraphicpositionofthethreephasesinFlowstone1a,andtheyindicatethattheerosionremnantofUnit2encrustedbyFlowstone1aisalsoolderthan780ka.ThefactthattheerosionremnantsofFlowstone1adipintothechambersuggeststhatatthetimeofformationofallthreephasesofFlowstone1a,thedebrisconeofUnit2sedimentwasstillinplace;thatis,erosionofUnit2sedimentfrombelowFlowstone1awouldhaveoccurredsometimeafter585ka(i.e.,theolderagelimitforphaseCinFlowstone1a)andpossiblyaslateas437ka(i.e.,theyoungeragelimitofphaseCinFlowstone1a).Unit2containsrarefossilsofmacrofauna,includingalongboneintheerosionremnant,whichmustbeolderthan780kaaswell.Weinterpretthefloorsedimentsofsub-unit3athatcontainthebaboontooth(sample1841)torepresent,atlastinpart,thereworkedremainsofthedebrisconethatonceexistedbelowFlowstone1a( Figure8 )asmaterialwas removedfromthechamberviafloordrains,therebyundercuttingthedebriscone,whichrespondedbyslowlyslumpingintothechamber±aprocessongoingtoday.Thebaboontoothcouldbepartoftheoriginal,Unit2debrisconeand,therefore,olderthan780ka,althoughtheUS-ESRageindicatesthatitcouldalsobeyounger(withamidpointageof ~ 679kabetweenamaximumageof 723 ± 181kaandtheminimumageof635 ± 148ka,withapossibleagerangeof487kato904ka). Ifthetoothisyoungerthan780ka,itwouldsuggestthatitwasderivedfromdifferentsedimentarydepositsnottestedinthisstudy,orthatitmayhaveenteredthechamberseparatelyduringerosionoftheUnit2debriscone,anditspresencemayreflectmoredirectentrypointsfromsurfaceintothechambernowsealedbyflowstone.ThispossibilitywastestedwithsampleRS9,aflowstonesamplefillingathin(<14cmwide)fractureinthedolomiteonthesurfacethatoccursabovetheDinalediChamber.ThisflowstoneyieldedequilibriumU-Thresultsmeaningthatitformedbefore ~ 600ka ( Table1 ),whichisconsistentwiththeinterpretationthattheDinalediChamberwasclosedtodirect entryofcoarser-grainedsedimentfromthesurfacepriortotheentryof H.naledi intothecavesystemandremainedcloseduntilthepresent( Dirksetal.,2016a ). BelowFlowstone1aarefiveotherflowstones(Flowstones1b-eandFlowstoneGroup2; Figure2b ),whicheachcovererosionalremnantsofsedimentsthatweoriginallygroupedasUnit2 ( Dirksetal.,2015 ).Thegeochronologyresultspresentedhere( Table1 )nowpermitabetterevaluationoftheflowstonestratigraphyinthechamber,anditisevidentthatUnit2representsasignifi-cantlyolderstratigraphicunitthatisrestrictedtodepositsdirectlybelowFlowstone1a,butnottothesedimentdepositsbelowFlowstones1b-e.Flowstone1creturnsanageof ~ 243ka( Table1 ), suggestingthatthesedimentsbelowFlowstones1b-earesignificantlyyoungerthanUnit2sedi-mentsbelowFlowstone1a( Figure14 ).ItwasnotedbeforethatthesedimentsbelowFlowstones 1b-earelessinduratedandlessweatheredthanthesedimentsbelowFlowstone1a,andthattheycontain H.naledi material( Dirksetal.,2015 ).Theirdistinctappearanceandfossilcontentisnow Table8. SummaryofOSLresultsobtainedbytheUniversityoftheWitwatersrandforsamplesofUnit1fromtheDinalediChamber (samplesOSL3,4and5).TheageswerecalculatedusingeffectiveUconcentrationvalues(takingdisequilibriumintoaccount;seetextfordetails).CAM=CentralAgeModel;MAM=MinimumAgeModel.SampleIDH2O(%)Th(ppm) U(ppm)pre-Rn U(ppm)post-RnK(%) Totaldr(Gy/ka)2 s Totalde(Gy)CAM2 s Totalde(Gy)MAM2 s CAMAge(ka) CAM2 s MAMage(ka) MAM2 s Overdispersion(%) OSL318.9 ± 53.71 ± 1.600.75 ± 0.1770.193 ± 0.0440.45 ± 0.120.760.07428.5968.92176.427.75601032314163 OSL425.8 ± 53.38 ± 1.600.485 ± 0.1770.097 ± 0.0440.47 ± 0.120.700.06379.8943.58168.020.7546792413755 OSL522.7 ± 55.11 ± 1.600.692 ± 0.1770.138 ± 0.0440.56 ± 0.120.900.07759.54102.33315.6748.688491323536168 DOI:10.7554/eLife.24231.017 Dirks etal .eLife2017;6:e24231. DOI:10.7554/eLife.24231 31of59 Researcharticle GenomicsandEvolutionaryBiology

PAGE 32

confirmedwiththedating.Therefore,wereinterpretthe H.naledi -bearingsedimentsbelowFlowstones1b-easpartofUnit3(sub-unit3b),whichmeansthatall H.naledi -bearingsedimentsinthe chamberarenowpartofsub-unit3b( Figure8 ). ApartfromthepoorlyconsolidatederosionalremnantsbelowFlowstones1b-e,hominin-bearing Unit3sedimentsalsocovermostoftheflooroftheDinalediChamber.Thelayerofsub-unit3bsedi-menthasbeeninterpretedasarelativelythinsheet( ~ 20cm)ofrubblymudclastbrecciamaterial mixedwith H.naledi fossilsbasedonoutcropsintheexcavationpitandpreliminarygroundpenetratingradarresults( Naidoo,2016 ; Figures2 and 8 ).Agebracketsforsub-unit3bwereobtained bydatingunderlyingoutcropsofsub-units1aand1bviaOSL,andbydatingoverlyingflowstoneunits.InthiscontextsampleOSL5isthemostrelevantforobtainingamaximumageestimateforsub-unit3b,becauseitwastakenfromanoutcropofUnit1thatisoverlainbyUnit3.Here,amaxi-mumagelimitof414ka(theuppererrorlimitofOSL5)canbeassignedtosub-unit3biftheOSLagesaretakenatfacevalue.Notehowever,thatthemaximumagelimitsforsub-unit1basdeter-minedfromsamplesOSL3andOSL4aresignificantlylessat272kaand278ka(uppererrorlimits),respectively( Table8 ),whichsupportstheinterpretationthatsub-unit1bformedduetoerosionand redepositionontopofsub-unit1a.Theminimumagelimitofsub-unit3bcanbemoreconfidentlyconstrainedasitisoverlainbyflowstoneswithagerangesbetween97kaand24ka( Table1 ).More importantly,ahangingremnantofsub-unit3bwithhomininmaterialiscoveredbyFlowstone1cwithaloweragelimitof236kaobtainedfromthecoreofastalactiteoverlyingtherimoftheflow-stoneremnant.Thisageprovidesthebestminimumageestimateforsub-unit3b,andbyextensionaminimumageforthe H.naledi fossils. U-ThdatingofanerosionalremnantofFlowstone1thatoccursdirectlyaboveanoutcropofsubunit1bfromwhichsampleOSL4wastaken,andappearstocoverit,providesanageof290 ± 6ka, suggestingthatUnit1inthislocationmustbeolderthan284ka.TheOSLresults(usingtheMAMmodel),however,suggestthattheUnit1sedimentsinthislocationmustbeyoungerthan278ka( Table8 ;usingMAMagemodels).ThisapparentparadoxmayindicatethattheOSLagesareunreliable,butcouldmeanthatsub-unit1bwasdepositedinaninversestratigraphicorderinrelationtotheflowstone.Thisissupportedbyphysicalevidenceinthechamberinwhichagapbetweentheflowstonedrapeandunderlyingsedimentwidenstowardsthebackoftheoutcrop( Figure2c ), implyingthatsub-unit1bwasdepositedbelowanerosionalremnantofFlowstone1,andis,there-fore,youngerthanthisflowstone.Recrystallization-dissolutiontextures,andanomalouslyhigh234U/238Uratios(>3)suggestthattheagereportedforthisflowstonesample(RS5)shouldbe treatedasaminimumage. Althoughtheageconstraintsforsub-units1aand1bareimprecisetheydosuggestthatUnit1in thispartofthechamberisyoungerthanUnit2,andthattheredmudclastsformingUnit2sedimentwerederivedfromsourcematerialmatchingourdescriptionofsub-unit1a,butpositionedhigherupinthecave.Thissourcematerialwaspossiblypartofsub-unit1c,oritcouldrepresentpartofanolderandasyetundefinedsub-unitofUnit1.Noageassessmentforsub-unit1cdepositsweredone,becauseaccumulationsaretoosmalltobetestedwithOSLortoodifficulttoaccess. TheU-Thagesfromflowstonesandteethplaceconstraintsonthechangingphysicalenvironment experiencedinthecavechamberovertime.FlowstonedepositionintheDinalediChamberoccurredduringdiscreteperiodsincluding24±32ka,50ka,88±105ka,andduringoldereventsaround242ka,290ka,between ~ 437kaand683ka,and>780ka.Theflowstonesareassociatedwithrelatively lowinitial234U/238Uratiosof1.8±2.4( Tables2 and 3 ).Flowstonesformedindifferentpartsofthe chamberasdrippointsshifted,butnoclearpatterninagedistributionisapparent( Figure1b )other Table9. FinalmeanpalaeomagneticdataforallsubsamplesanalysedfromeachphaseofFlowstone1aasshownin Figure13 .MAD =meanmaximumangulardeviationforindividualsamples;K=precision/sampledispersalparameter;Plat=palaeolatitude).Flowstone1a Declination(O) Inclination(O)MADKPlat.Polarity PhaseC15.5 39.7370.875.4N PhaseB26 28.17.4156.263.3N PhaseA156.415.95.730.2 60.0R DOI:10.7554/eLife.24231.018 Dirks etal .eLife2017;6:e24231. DOI:10.7554/eLife.24231 32of59 Researcharticle GenomicsandEvolutionaryBiology

PAGE 33

Figure8. Cartoonillustratingthesedimentaryhistoryresultinginthedepositionandredistributionofsedimentof Units2and3,andFlowstoneGroups1to3intheDinalediChamber.Notethatallhomininfossilsarecontainedinsub-unit3b,butthatthissub-unithasbeenrepeatedlyreworkedafteritsinitialdeposition.FossilentryoccurredduringtheinitialstagesofdepositionofUnit3belowtheentryshaftandpredateddepositionofFlowstone1c. H. naledi fossilsmayhavecontinuedtoentertheDinalediChamberasolderpartsofUnit3wereerodedfrombelow Flowstone1c,andasremnantsofallolderunitswerereworkedtobeincorporatedintoUnit3sedimentsthataccumulatedalongtheflooroftheDinalediChamber. DOI:10.7554/eLife.24231.019 Dirks etal .eLife2017;6:e24231. DOI:10.7554/eLife.24231 33of59 Researcharticle GenomicsandEvolutionaryBiology

PAGE 34

thanthattheflowstonesdeeperinthecavegenerallyappeartobeyounger.TheU-ThagesfortheteethalsodefineanumberofapparentUuptakeeventsataround43±48ka,70±75ka,80±90ka,100±120kaand ~ 200ka( Tables4 and 5 ).Theinitial234U/238Uratiosassociatedwitheachofthese eventsissimilarandanomalouslyhigh(6.9±8.1). Theperiodsoftimeduringwhichflowstonesformedinthecave,byandlarge,donotoverlap withtheperiodsoftimeduringwhichU-uptakeappearstohaveoccurredintheteeth,althoughbothtypesofeventswereprobablyassociatedwithwetperiodsinthechamber.Thesystematicdif-ferenceintheinitial234U/238Uratiosobtainedfromflowstones,ascomparedtoteeth,indicatesthat anisotopicallydistinctwatersourceledtoU-uptakeintheteeth,suchasgroundwaterinthedolo-miticaquiferoftheMalmaniGroupwithreportedanomalouslyhighinitial234U/238Uratios(e.g., Kronfeldetal.,1994 ).Workingontheassumptionthatthehighinitial234U/238Uratiosarederived fromthegroundwaterreservoir,theobservedagegroupingssuggestthatUuptakeeventsinteethrepresent(partial)inundationeventsoftheDinalediChamber,whilsttheflowstoneformationeventsreflectperiodsduringwhichthegroundwatertablehaddroppedbelowfloorlevel,butextensivedripstilloccurredwithintheDinalediChambercausedbywaterderivedfromthesurfacewithgener-allylower234Uexcess.TheagegroupingsalsoindicatethatepisodicwetperiodsintheDinaledi Chamberalternatedwithperiodsduringwhichnoflowstonewasdeposited.Ageestimatesfor H.naledi andimplicationsforhomininevolution Figures1b and 14 summarizetheresultsofalldatingmethodsappliedtotheDinalediChamber duringthecourseofthisstudy.Itisclearfromtheseresultsthatthe H.naledi assemblageinUnit3 isofmid-tolate-MiddlePleistoceneage.Thebestageestimatesforthe H.naledi fossilscomefrom theaveragedUS-ESRagesforsamples1788and1810:229+60/±46ka(maximum)and179+49/±40ka(minimum)forsample1788,and276+59/±77ka(maximum)and220+50/±60ka(minimum)forsample1810,withanagerangeof139kato335ka.Themaximumagescenarioprovidesanaver-ageageforbothteethof253+82/±70ka,andtheminimumagescenarioprovidesanaverageageof200+70/±61ka.ConsideringtheobservedRnlossinthecavesediments,themaximumageesti-mateisconsideredtobeclosertothetrueage. Theloweragelimitof139kamustbedisregardedandshiftedto ~ 200kaconsideringaU-Th minimumageestimateforpartsofenamelontoothsample1810( Table5 ).Thisminimumagelimit canbeconstrainedfurtherto236kabasedonU-ThageestimatesforFlowstone1c(242.0 ± 5.0ka; 242.9 ± 6.6ka; Table1 ),whichdirectlycoversfossilmaterialof H.naledi, notingthattheseages mayrepresentminimumageestimatesfortheflowstoneasaresultofpossibleUuptakeduringaperiodofraisedgroundwaterlevels( Figure14 ). Themaximumagelimitof335kaforthe H.naledi fossilsrelieson80%Rnlossthroughoutthe burialhistoryofthefossilsandisprobablyanover-estimate.Aseparatemaximumageestimateforsub-unit3bcanbeobtainedfromthebaboontooth(sample1841)insub-unit3asedimentsdevoidofhomininfossils,underlyingthepartlyarticulatedremainsof H.naledi ,withUS-ESRageestimates varyingfrom635 ± 148ka(minimum)to723 ± 181ka(maximum)withanagerangeof487kato904 kaandamid-pointageof679ka.Theolderageestimateisconsideredtobemorelikely(consider-ingthemeasuredRnloss).Thesemaximumageestimatesareconsistentwiththedirectageesti-matesforthe H.naledi fossils,butdonotfurtherconstraintheupperagelimit( Figure14 ). Afurthermaximumageestimateforthe H.naledi fossilscanbeobtainedfromageestimatesfor sub-unit1ausingsampleOSL5.Thissampleprovidesamaximumagelimitforsub-unit3b,of ~ 414 ka( Table8 ;usingMAM).AlthoughOSLagesarepoorlyconstrained,anddonotprovideverypreciseagelimits,thisestimateisbroadlyconsistentwiththeestimatedageforreworkingofUnit2sedimentsfrombelowFlowstone1a(437±585ka)andtheUS-ESRageestimatesfortheteeth. Consideringallageresultspresentedherethemostparsimoniousageestimateforthe H.naledi fossilsissometimebetween236kaand335ka.Moreworkwillbeneededinfuturetoconstraintheseagesfurther( Figure14 ). Untilnow,ithasbeengenerallyassumedthatmorphologicallyprimitivehomininslike H.naledi ( Bergeretal.,2015 )didnotsurviveintothelaterpartsofthePleistoceneinAfrica.Thisgeneral assumptionhascommonlyguidedtheinterpretationoffossildiscoverieswithpoorgeologicalorstratigraphiccontext,includinginitialestimatesfortheageofthe H.naledi fossils( Thackeray,2016 ; Demboetal.,2016 ).Thenewageestimatesfor H.naledi showthatanapproximateageforthe homininfossilfragmentscannotbesimplydeducedfromtheirmorphology( Thackeray,2016 ; Dirks etal .eLife2017;6:e24231. DOI:10.7554/eLife.24231 34of59 Researcharticle GenomicsandEvolutionaryBiology

PAGE 35

Demboetal.,2016 ).Detailedgeologicalinvestigationsarecriticalbeforeanyattempttoascribean agetothefossilsismade,andeventhengreatcaremustbetakenininterpretingresults,whichmaynotalwaysbeconclusive(e.g.seefamousexamplessuchasSterkfonteinorLiangBua; Partridgeetal.,2003 ; Wilkinson,1985 ; Walkeretal.,2006 ; Robertsetal.,2009 ; Pickeringand Kramers,2010 ; Grangeretal.2015 ; Sutiknaetal.,2016 ; KramersandDirks,2017 ). ItisgenerallyassumedthatallAfricanfossilhomininsproducingMiddleStoneAgearchaeological industriesinthepast300kawerepartofasinglevariablespeciesofearly H.sapiens oranimmediateprecursor(e.g., McbreartyandBrooks,2000 ; LahrandFoley,2001 ; Stringer,2002 ).Thenew agesnowshowthat H.naledi existedatthesametimeasthefirstMiddleStoneAgetoolswereproducedinsouthernandeasternAfrica,whilstskeletalevidenceshowsthat H.naledi wasprobably capableoftooluse( Bergeretal.,2015 ; Hawksetal.,2017 ).Thisraisesthepossibilityof H.naledi beingresponsibleforsomeoftheMSAtraditions.Theimplicationsofthenewagesfor H.naledi are discussedindetailin Bergeretal.,2017 . Materialandmethods FlowstonesamplesforU-ThdatingAtotalofseventeenflowstonesamples(RS1,RS5-6,RS8,RS10-11,andRS13±23)fromtheDinalediChamberweredatedviaU-Thgeochronology,includingonesetofblindduplicates(RS1andRS15; Table1 , Figures1b and 3 ).Inaddition,oneflowstonesample(RS9)wastakenonthesurface (WGS84571240±7121866)fromashallowpitabout11mSWoftheprojectedsurfacepositionoftheexcavationpitintheDinalediChamber( Figure3d ).Foreachsampleapowderwasprepared andthensplit,withonehalfbeingdatedatJCUandtheotherhalfbeingdatedatUoM;thatis,foreachsamplebothJCUandUoMdatedthesamematerial. IntheDinalediChambersamplesofFlowstoneGroups1,2and3werecollectedfromavarietyof stratigraphicpositions( Figures1b , 2 and 3 ).Allthesampledflowstonesformedassheets,crusts,or drapesoverlyingoldersedimentunits,withtheexceptionofsamplesRS13andRS18,whichweretakenfromasmallstalactitethatformedalongthelipofanerosionalremnantofFlowstone1c.Additionally,samplesRS14,RS16andRS17weretakenfromaflowstonedrapealongadolostoneside-wallofthechamber.Inallinstancestheflowstoneshaveafreeuppersurface,thatis,theyarenotoverlainorcoveredbysediment,andallflowstonesareinterpretedtobeyoungerthanthesedi-mentunitstheycover.TheoneexceptiontothisrulecouldbeRS5,whichcomesfromapartlyresorbederosionremnantofFlowstone1thatappearstooverlieUnit1,butisseparatedfromthetopsurfaceofUnit1byasmallopeningthatwidenswithdepth( Figure2c ).Thisleavesthepossibilitythatthisflowstoneisanerosionalremnant,andthatUnit1(sub-unit1a)sedimentsbuiltupbelowitafterithadbeendeposited.ThelocationofeachsamplewithintheDinalediChamberisshownin Figure1b ,andoutcropandclose-upphotosofthesampledflowstonesareshownin Figure3 . RS1andRS15 areblindduplicatesamplestakenfromahangingremnantofFlowstoneGroup2 thatoccursinaN-trendingfracture, ~ 10mNoftheexcavationpitand ~ 3mNofamajoroutcropof sub-unit1asedimentfromwhichsampleOSL3wascollected.Theflowstoneoverlieslargelyuncon-solidatedfloorsedimentsofUnit3,whichinthislocalityhaveerodedfromunderneaththeflowstonetoleaveahangingremnant ~ 8cmabovethecurrentfloorlevel( Figure3i ).Theflowstoneconsistsof a15±18mm-thicklayerofcalciteoverlyinganirregularsurfaceofmudclastbreccia,locallyincorpo-ratingandgrowingaroundlargemudclaststhatwerelyingonthepalaeo-surface.Theflowstoneisgrey-whiteincolourandpreserves3±6mmscalelaminationsvisibleduetosubtlecolourvariations.Theflowstoneisrecrystallizedwithelongated,acicularcrystalsofcalcitegrowingfromthebasetothetopofthelayeracrossallinternallaminations.Theuppersurfaceoftheflowstonehasarough,pittedappearanceasaresultofpartialresorptionordissolutionofcalcitealongthegrainbound-ariesoftheneedle-likecrystals.SamplesRS1andRS15weretakenfroma3mm-thickzone,3mmabovethebasalcontactoftheflowstonelayer( Figure3i ). RS5 wassampledfromathinsheetofFlowstoneGroup1inaWNW-trendingfracture, ~ 4mW oftheexcavationpit( Figure1b ).Theflowstoneoverliesorangesandymudstonebelongingtosubunit1bfromwhichsampleOSL4wastaken( Figure3a ),butappearstobeseparatedfromthisunit byanarrowopeningthatwidenstothebackoftheoutcrop( Figure2c ).Theflowstoneconsistsof an8±22mmthick,creamwhitelayerofcarbonatewithasponge-like,porous,sugarytexturethat Dirks etal .eLife2017;6:e24231. DOI:10.7554/eLife.24231 35of59 Researcharticle GenomicsandEvolutionaryBiology

PAGE 36

largelymasks(duetorecrystallizationoftheprimarycalcite)underlyingmm-scalelaminations.Theflowstonelayerappearstobepartlydissolvedalongitsstratigraphictopwiththeexternalsurfaceofthelayertruncatinginternallaminations.SampleRS5wastakenfroma5mm-thickhorizon,3mmbelowthetopoftheflowstonelayer( Figure3a ). RS6 wassampledfromathinsheetofFlowstoneGroup2inaWNW-trendingfracture, ~ 6mW oftheentryzone( Figure1b ).TheflowstoneoverlieserosionremnantsofUnit1andUnit3.Atthe samplesite,sedimentsofUnit3havebeenpartlyerodedfromunderneaththeflowstoneleavingahangingremnant,5±10cmabovethecurrentfloorlevel( Figure3b ).Theflowstoneconsistsofan8± 12mmthickcrustoverlyinganirregularsedimentsurface,incorporatingfine(<3mm)mudclastsinthebaseofthelayer.Itiswhite-greyincolourandsemi-transparent,withlimitedevidenceofinternallayeringexceptforaslightlylightercolouredbasallayerthatisseveralmmthick.Theentirelayerisrecrystallizedwithfine,radiatingacicularcrystalsofcalcitegrowingupwardfromthebasalcontact.Theuppersurfacehasaroughpittedappearanceasaresultofpartialresorption/dissolutionalonggrainboundariesoftheacicularcrystals.SampleRS6wastakenfroma3mm-thickhorizonatthebottomoftheflowstone( Figure3b ). RS8 wassampledfromathinsheetofFlowstone2ontheflooroftheentryzonebelowthestack ofhangingremnantsofFlowstone1a-e.TheflowstoneoverliessedimentsofUnit3thatarepartlyerodedfromunderneaththeflowstoneleavinga5±10cmgapbetweentheflowstonesheetandthecurrentfloor( Figure3c ).Theflowstoneconsistsofa5±22mmthicklayeroverlyinganirregularsurfaceofmudclastbreccia.Thegrey-whiteflowstonepreserves3±5mmthicklayeringvisibleassubtlecolourvariations,despitepartialrecrystallizationandreplacementbyelongatedacicularcrystalsofcalcite.Thebasal10mmiscomposedofameshoffinearagoniteneedles.Abovethiszoneacicularcalcitereplacesthearagonite.Theuppersurfaceoftheflowstonehasarough,pittedappearanceasaresultofpartialresorption/dissolutionalonggrainboundariesoftheacicularcrystals.SampleRS8wastakenfromthebottom3±5mmoftheflowstonelayer( Figure3c ). RS9 wassampledfromasurfaceoutcropinashallowminepitthatoccursabout11mSWofthe projectedsurfacepositionoftheexcavationpitintheDinalediChamber( Figure3d ).Thesample siteoccursalongstrikeofthesamefracturealongwhichtheDinalediChamberwasformedatdepth,anditisthereforepossiblethatthissurfaceoutcropinthepastcouldhavelinkedtothecavesystembelow.Theflowstoneconsistsofa12±25mmthicklayerofcarbonateoverlyinganirregularsurfaceofconsolidatedmudclastbrecciasimilartosedimentsofUnit2intheDinalediChamber.Thegrey-whiteflowstonepreserves1±5mmscalelaminationsvisibleduetocolourvariationsandthepres-enceofseveralthinbrownmarkersurfaces.Thelaminationsarelocallyrecrystallizedandovergrownbyradiating,elongatedcrystalsofcalcitegrowingfromthebasetothetopoftheflowstonelayer.Additionally,fibrousaragoniteneedlesarewidelydistributedinsheaf-likepatterns.Theuppersur-faceoftheflowstonehasarough,pittedappearanceasaresultofpartialresorption/dissolutionalongthegrainboundariesoftheacicularcrystals.SampleRS9wastakenfroma4mm-thickwhitehorizon,2mmabovethebasalcontactoftheflowstonelayer( Figure3d ). RS10 wassampledfromathincrustofFlowstoneGroup2,about2mWoftheexcavationpit and0.2mWofthelocationwheretoothsample1810wasfound( Figure3e ).Theflowstonedirectly overliesmudclastbrecciaofUnit3andhomininbonefragments.Itconsistsofa4±8mm-thicklami-natedflowstonecrustoverlyinganirregularsedimentsurfaceandincorporatesfine(<3mm)mud-stoneclastswithinitsbasallaminae.Thisflowstoneiswhitetogreyincolourandisfinelylaminatedandpartlyrecrystallized,withrecrystallizationvisibleaswhite,fine,radiatingneedlesofaragonitegrowingupwardfromthebasalcontact,alonganirregularalterationfrontintogrey-white,laminatedcalcitenearthetopofthelayer.RS10wastakenfromthelaminatedbottom3mmoftheflowstonelayer( Figure3e ). RS11 and RS21 representtwosamplestakenfromthetopandbottom,respectively,ofthesame flowstoneunitdevelopedalongthefloorbelowanactivedrippoint,1.5mEoftheexcavationpit.ThisflowstoneconsistsofabasallayerofgreyflowstoneinterpretedasFlowstoneGroup2overlyingUnit3sediments,coveredbyawhiteflowstonelayercontainingsmallstalagmitesinterpretedasFlowstoneGroup3( Figure3f ).Thesamplecollectedfordatingconsistsofa20±28mmthick,finely laminatedflowstoneincludingthetwodistinctlayersdescribedabove( Figure3f ).Thebasallayer (RS21; Figure3f )is3±12mmthick,andconsistsofbrown-greycalciterestingonUnit3sediment. Thislayerisstronglyrecrystallizedwithradial,acicularcrystalsofcalcitegrowingupwardfromthelowercontact,whichlocallyappearstohavereplacedanearliergenerationofaciculararagonite. Dirks etal .eLife2017;6:e24231. DOI:10.7554/eLife.24231 36of59 Researcharticle GenomicsandEvolutionaryBiology

PAGE 37

Thesecrystalsovergrowmm-scalelaminationsdefinedbywhitetobrowncolourvariations.RS21wassampledfromthebasal-to-centralsegmentofthislayer.Thebasallayerisoverlainbyan8±16mmthicklayeroffinelylaminated(sub-mmscale)whitespeleothem.Thebasal2±3mmofthislayerisrecrystallizedwithfinearagoniteneedlesradiatingoutfromthebase.Abovethattheflowstoneconsistsoffinelamellaedefinedbysubtlegreytowhitecolourvariations.Eachlaminaiscomposedofbotryoidalaggregatesofaciculararagonite.RS11wastakenfromthetop3mmofthiswhiteflow-stone( Figure3f ). RS13 and RS18 aretwosamplestakenfromthesamestalactitedevelopedalongthelipofan erosionalremnantofFlowstone1cneartheentranceintotheDinalediChamber( Figure1b ).Inthis location,Flowstone1ccoverstheerosionalremainsofamudclastbrecciacontainingalongboneconsistentwiththe H.naledi assemblage( Figure3g ),andinterpretedasUnit3.Thestalactiteconnectstopartofthespeleothemlayersthatcoverthebone. Thestalactitepreserveswell-developedinternallayering,withlayersasymmetricallydeveloped aroundacore,inwhichlayersthickentowardstheoutwardfacingsideofthestalactite.Fromcoretorimthestalactitereachesamaximumthicknessof53mmandincludesthreeseparatezonesthatcanbedistinguishedbasedoninternaltexture,layering,andcolour( Figure3g ).Theinnermostzone formsan8mm-thick,finelylaminated(sub-mmscale)corecentredonasmallmudclastandistermi-natedbyathinbrown,mud-richrim.Thiscoreissurroundedbya ~ 23mm-thickcentralzonethat consistsofmorecoarselylayered(3±10mm-thick),creamtogrey-whitecolouredcalcitewithlayeringpreservedassubtlecolourvariations,whichatitsbaseshowsreplacementbyradiatingsheavesofaragoniteneedles.Thiszoneismantledbya ~ 22mm-thickouterzoneofwhitecalcitepreserving onlyremnantsofinternallayering.Thisoutermostzoneischaracterizedbyextensivereplacementofabundant,olderaragoniteneedlesbycoarse-grainedcalcite,creatingapatchytexture.Thesamplehasbeenaffectedbyrecrystallizationresultingintheformationofradiatingacicularcrystalsofcalcitethatgrowfromthecoreoutward.Thewhiteouterlayeralsoshowsevidenceoffurtherrecrystalliza-tion,withtheformationofcoarse(2±4mm),equantcalcitegrains,manywithhighlyirregulargrainboundariesthatovergrowtheaciculargrains.Theoutersurfaceofthestalactitehasarough,pittedappearanceasaresultofpartialresorption/dissolutionalongthesurfacewithdissolutionalongthegrainboundariesoftheacicularcrystals.RS18wassampledfroma3mm-thickhorizonnearthebaseofthecentralzone.RS13wassampledfromtheoutermostpartoftheouterzone( Figure3g ). RS14 wassampledfromanirregularlyshaped,cascade-likecrustofFlowstoneGroup2,along theside-wallofthemainfloordrain,betweentheentryshaftandexcavationpit( Figures1b and 3h ).Thesampleofflowstonecrustconsistsofan8±11mmthicklayerofcreamcolouredcarbonate displayingmm-scalelaminationsandasugary,recrystallizedtexturewithnumerous,fineporespacesalonglaminarsurfaces.SampleRS14wastakenfroma3mm-thickzoneencompassingseveralfinelaminations, ~ 5mmabovethebaseoftheflowstonelayer( Figure3h ). RS16 and RS17 aretwoseparatesamplesfromacascade-likecrustofFlowstoneGroup2that formedwithinthemainfloordrainbetweentheentryshaftandexcavationpit( Figure1b ).Thisflowstonedevelopedontopofthedolomiteback-wallofthedrain( Figure3h ).Thesampleconsistsofa mostlycreamcolouredflowstonethatisupto73mmthick,withamassivesugary,recrystallizedtex-ture,preservingmm-scalelaminationsvisibleduetosubtlecolourvariations.Manyoftheequantcal-citegrainscontainremnantaragoniteneedles,reflectinganearlierphaseofaragonitegrowth.Laminationsalongthebasal3±4mmofthisflowstonearebrownincolour,whereasthetop10±12mmofthisflowstonelayerconsistsofwhatispossiblyaseparate,youngerunitoflaminatedgrey-browncarbonatewithsmallporespacesdevelopedalongsomeofthelaminarsurfaces.SampleRS16wastakenfroma3±4mm-thicklayerdirectlyabovethebasalzonewithbrownlaminations.SampleRS17wastakenfroma4mm-thickzoneatthetopofthecream-colouredlaminatedflow-stone,immediatelybelowthedarkercolouredtopunit( Figure3h ). RS19 wastakenfromawedge-shapedsampleofFlowstoneGroup2thatformedalongthelipof adrippooldirectlyoverlyingUnit3sediments,1.2mSoftheexcavationpit( Figure1b ).Theflowstoneiscream-colouredandupto30mmthick,andoverliesaflatsedimentsurface.Itpreservescomplexinternallayeringandavarietyoftextures( Figure3j ).Abasal2±4mm-thicklayercharacterisedbyfinecalciteneedlesgrowingatrightanglestolayeringisoverlainbyanirregularmassofdarkercreamcoloured,unstructuredcalcitewithasugarytextureandnumerousfinevoids.Towardsitstopthisunstructuredmassisinterlayeredwithandcoveredbyseveral1±4mm-thicklamellaecharacterisedbyfinecalciteneedlesgrowingatrightanglestolayeringsimilartothebasallayer, Dirks etal .eLife2017;6:e24231. DOI:10.7554/eLife.24231 37of59 Researcharticle GenomicsandEvolutionaryBiology

PAGE 38

butcontainingporespaces.Theseinturnareoverlainbyan8mm-thickzoneoffinelylaminatedflowstonewithafine-grained(<0.5mm)sugarytextureandabundantfineporespaces.RS19wastakenfromthebasalflowstonelayerwithradiatingcalcitecrystals( Figure3j ). RS20 wassampledfromathinsheetofFlowstoneGroup2inaWNW-trendingfracture, ~ 2mW oftheexcavationpit( Figure1b ).Theflowstoneformsacascade-likedrapethatdirectlyoverlies orangemudstoneofsub-unit1afromwhichsampleOSL5wastaken( Figure3k ).RS20occurs directlyabovesampleRS10,whichwastakenatthefootofthecascade( Figure1b ).Thesampled flowstonelayerisirregularinshapeasitcoverssedimentswithtopography.Itconsistsofafinelylaminated15±22mmthickcrustoverlyingsub-unit1asedimentandincludesa ~ 6mmthick,laminatedbasalunitofwhiteflowstonethatincorporatessmall(3±10mm)mudclasts,overlainbymorecoarselylaminated(mm-scale)greybrowncarbonatefreeofinclusions.Theflowstoneisinternallylocallyrecrystallizedwithaciculararagonitecrystalsgrowingwithin ~ 3mmthickzonesatrightangles tolayering.RS20wassampledfromthebasallayerofthesample( Figure3k ). RS22 and RS23 aretwosamplesfromanerodedrimofFlowstone1aneartheentryshaftinto theDinalediChamber( Figure1b ).Thisflowstoneoverlieserosionalremnantsofwell-induratedmud clastbrecciaassignedtoUnit2( Figure3l ),andmostlyconsistsofcoarselyrecrystallizedwhitecarbonate.TheflowstonesamplefromwhichRS22andRS23weretakenoccursdowndipfromthepalaeo-magneticsampletakenfromFlowstone1aasdescribedbelow( Figures2b and 13 ).The flowstonelayeris ~ 25mmthickandcomprisesa6mmthick,basalunitofgreytobrowncalcitethat isrecrystallizedintofine(sub-mmscale),equigranulargrainsofcalciteovergrowinganolderaciculartexture.ThebasallayercorrespondstophaseBcarbonatedescribedforthepalaeo-magneticsam-ple,andisoverlainbyacleanwhitecalciteunit(i.e.,phaseCinthepalaeo-magneticsample)thatiscoarselyrecrystallizedwith2±4mmequidimensionalcalcitegrainsovergrowing(andpartlydestroy-ing)anoldertextureformedbyaciculargrains.Smallvoidsoccurintheinterstitialspacesbetweenthecoarsecalcitegrains.RS23wastakenfromthegreybasalunit(i.e.,phaseB)oftheflowstonelayerandRS22wassampledfromthecentralpartoftheupperwhiterecrystallizedunit(i.e.,phaseC, Figures3l and 13b ).Inthissample,flowstonebelongingtotheolderphaseAcarbonateisno longerpresent,asthislayerpinchesoutalongdipbetweenthepointwherethepalaeomagneticsamplewastakenandthepointwheretheU-Pbsamplewastakenascanbeseenin Figures2b and 13 . ToothsamplesforESRdatingThree H.naledi teeth(samples1810,1767and1788)andonebaboontooth(sample1841)werecollectedfromUnit3,intheDinalediChamber( Figures1b , 4 , 6 and 7 )forESRdatingandU-Thanalysis. Sample1767 (fullcataloguenumberU.W.101±1767)isanextremelywornupperpremolarcrown ( Figure6a )obtainedfromapproximately1mSWoftheexcavationpit( Figure4 ),andoccurredon surfacesurroundedbymudclastfragmentsofUnit3.Thistoothisdeeplyweathered,andpreservedonlyasmallrimofenamelonthebuccalmarginwithamaximumheightof4.5mm.Otherwisethecrownisaconcavedentinesurfaceworntothecervix,withbrightwhite,highlybleacheddentine.Thetoothisbrittleandappearsstronglyaffectedbywateraction. Sample1788 (fullcataloguenumberU.W.101±1788)isalowerrightsecondmolar( Figure6b )obtainedfromapproximately2m WSWoftheexcavationpit( Figure4 ),embeddedwithinlooselypacked,mudclastbrecciaofUnit 3, ~ 2cmbelowthegroundsurfacelevel.Thistoothispartlybrokenattheroot,buthasanotherwisewellpreservedcrownwiththick,light-greyenamel.Thedistalrootofthetoothispresentandcomplete,butthelingualrootisbrokenoffjustbelowthecervix.Thetoothismorphologicallycon-sistentasanantimereofU.W.101±284.Thedentineishighlybleachedandbrittleandappearsaffectedbywateraction. Sample1810 (fullcataloguenumberU.W.101±1810)isalowerleftthird premolarorpossiblylowerleftfourthpremolar( Figure6c )obtainedfromtheSEcorneroftheexcavationpit( Figure4 ),embeddedwithinsedimentsofUnit3, ~ 5cmbelowtheoriginalgroundsurface level.Thistoothiswell-preservedwiththick,light-blue-greyenamel,andshowslittleevidenceofbleachingorweathering,withonlyaveryslightpolishingwearonthedistalcrestoftheprotoconid.Themorphologyofthecrownissimilartootherlowerthirdpremolarsinthecollection,however,itisslightlyshorterthanmanyofthose. Sample1841 (fullcataloguenumberU.W.101±1841)isawellpreservedtoothcrown,morphologicallyconsistentwithalowerleftsecondmolarof Papiosp .This isacompleteenamelcrownofanuneruptedtoothwithnowearfacetsorinterproximalfacetsinevi-dence,andtherootshadnotformed( Figure7 ).Thespecimenwasrecoveredfromasediment Dirks etal .eLife2017;6:e24231. DOI:10.7554/eLife.24231 38of59 Researcharticle GenomicsandEvolutionaryBiology

PAGE 39

sampletakenatadepthof55±60cmbelowtheoriginalgroundsurfaceofthecavefloornearthebaseofthesondageduginthecentreoftheexcavationpit( Figures1b , 2d and 4 ).Thistooth occursinsub-unit3a, ~ 40cmbelowthestratigraphicallylowestoccurrenceofpartlyarticulated remainsof H.naledi ,andrepresentstheonlynon-homininmacrofossilrecoveredfromUnit3. SedimentsamplesforOSLdatingThreeOSLsamples(OSL3,OSL4andOSL5)werecollectedintheDinalediChamberfromerosionremnantsofsub-unit1a,inwhichsandy,laminatedmudstonesareexposed( Figures1b and 5 ).Thesesandyintercalationsweretargetedbecausetheycontainfine-grainedquartzandfeldspar grainsthatcanbeextractedforanalysis.Foreachsamplea30cmlengthofaluminiumpipingwithadiameterof5cmwashammeredintothesedimentsinahorizontaldirectionorparalleltosedimen-tarylaminationsvisiblewithintheunits.AcoresamplewithinthepipewasextractedforOSLanaly-sestogetherwithasedimentsamplefromthesameunittodeterminebackgroundradiationfrommeasuredvaluesofU,ThandK. SampleOSL3comesfromanerosionremnantofsedimentsofsub-unit1bcollectednearthe intersectionpointoftwofracturestrendingNandErespectively, ~ 6mNoftheexcavationpit ( Figure1b ).SamplesOSL4andOSL5comefromanerosionremnantofUnit1sedimentalongan ENE-trendingfracture, ~ 3mWoftheexcavationpit.SampleOSL4isobtainedfromsub-unit1b directlybelowathin,partlyresorbedflowstonesheetattributedtoFlowstoneGroup1( Figure3a ). SampleOSL5occursasanerosionalremnantofsub-unit1a,1mE,andstratigraphically10±20cmbelowsampleOSL4.TheUnit1sedimentsinthislocationarepartlycoveredbyUnit3sedimentsandacascadeofFlowstoneGroup2( Figure3k ). FlowstonesampleforpalaeomagneticanalysisThespeleothemsampledforpalaeomagneticanalysiscomesfromFlowstone1aneartheentryzoneintotheDinalediChamber( Figure2b ).Thesampleislayeredandcomprisesthreedistinctphases (frombasetotop:A-C)separatedbythinclastichorizonsthatmarkdisconformities( Figure13a,b ). Thelowerphase(phaseA)isinterstratifiedwithvisibleclasticlaminations.TheinternallaminationsaretruncatedalongthelowersurfaceofthesampleindicatingthatphaseAflowstonewaspartlydissolvedduringaphreaticeventaftertheflowstonehadbeendeposited.Themiddlephase(phaseB)consistsofintercalatedflowstonespeleothemanddetritalsedimentlayers,withthedetri-tallayeringconcentratedtowardsthedownslopepartofthesample.DetritalmaterialisgenerallylessthaninphaseA,indicatingdecreasingamountsofclasticcontamination.LikephaseA,theinter-nallayeringofphaseBistruncatedalongthelowersurfaceofthesample,indicatingthatthephre-aticdissolutioneventoccurredafterdepositionofphaseB.ThereisnoapparenttruncationsurfacebetweenphaseAandBflowstone,indicatingthatphaseBformedontopofphaseAafteraperiodofnon-depositionduringwhichdetritalmaterialaccumulatedontopofphaseA.TheuppersurfaceofphaseBtruncatesinternallayeringreflectingdissolutionduringaphreaticevent.Thiswasfol-lowedbyaperiodofnon-depositionofflowstoneduringwhichathinlayerofdetritalclasticmaterialaccumulated,beforedepositionoftheupperlayer(phaseC)offlowstoneoccurred.PhaseCcom-prisesyounger,cleanerflowstonewhichdisplaysextensiverecrystallizationalongthebaseoftheunitwiththeformationofelongatedcalcitecrystals.ThiszoneofrecrystallizationcouldpotentiallyalsoindicateayoungerinfillofacavitythathadformedbetweenphaseBandCflowstone,andhasbeenavoidedduringsampling.PhaseCflowstonecontainslittletonodetritalinclusions,andsug-geststhatsedimentinfluxintothechamberwasnotoccurringduringitsdeposition. PhaseBandphaseCflowstonecorrelatewithsamplesRS23andRS22respectively,thatwerecollectedforU-Thdating.ThetopofthesamplewasorientedinthecavetomagneticN( 18.2odegreesfromtrueNatthislocationwitha 62.9oinclination).Theinclinationwasaccountedforby markingthesampleonacompletelyflatsurfaceoftheblock.BonesamplesforradiocarbondatingThreeweatheredbonefragmentsof H.naledi werecollectedforradiocarbondatingincluding:(i)a tibiashaftfragment,53mminlength(U.W.101±567);(ii)afemurshaftfragmentcomprisingthewholecircumferenceoftheshaft,79mminlength(U.W.101±857),and;(iii)ametatarsalor Dirks etal .eLife2017;6:e24231. DOI:10.7554/eLife.24231 39of59 Researcharticle GenomicsandEvolutionaryBiology

PAGE 40

metacarpalshaftwithnoarticularmorphology,48mminlength(U.W.101±065).AllsampleswerecollectedfromthesurfaceofUnit3asisolatedfragmentsneartheexcavationpit.AnalyticalandcomputationaldatingmethodsU-Thdating BackgroundtoU-Thdating Cavecarbonatesareprovenarchivesofsedimentchronologyandterrestrialclimatevariation(e.g., PickeringandKramers,2010 ; Pickeringetal.,2010 ).Cavecarbonateslessthan ~ 600kainage canbepreciselydatedusingtheU-Thdisequilibriumdatingmethod(e.g., HellstromandPickering, 2015 ).ThedisequilibriumtechniquediffersfromotherradiogenictechniqueslikeU-PborRb-Srin whichthedaughterproductisstableandaccumulatesindefinitely.Inthe238Udecaychain,234U (half-lifeof245.5ka)decaysto230Th(half-lifeof75.4ka),whichitselfdecaysto226Ra.Thismeans thatratherthanaccumulatingindefinitelywithtime,theconcentrationof230Thinasamplegradually movestoapointwherethenumberofdecaysof230Thatomsequalsthenumberof230Thatomsproducedviathedecayof234U(`secularequilibrium').BecauseThisnearlyinsolubleinsurfacewaters, flowstonedoesnotusuallycontaininitial230Th,incontrasttoU,whichissolubleinoxidizingwaters andisincorporatedintoflowstoneatlowconcentrations(generally<5ppm;e.g., Pickeringetal., 2010 ).InU-Thdating,theaccumulationof230Thtowardssecularequilibriumandthedecayofits immediateparent(234U)yieldstheage.IncalculatingaU-Thage,theratioof234Utoitsparent238U mustalsobedetermined,asthisratioisgenerallyelevatedabovesecularequilibriuminnaturalwaters.Theassumptionthatinitial230Thconcentrationsinflowstoneareclosetozeromaybeviolatedbyvaryingamountsof`detritalmatter',carryinginitial230Th.However,manycavecarbonates formwithessentiallynoinitialTh.WherecorrectionsforinitialTharesmall,potentialvariabilityintheisotopiccompositionofthisThisgenerallyincorporatedintheageerror.TheU-Thdatingtech-niquehasanupperagelimitof ~ 600ka,determinedbythehalf-lifeof230Th,andbytheprecision withwhichthevariousisotopescanbemeasuredinthelaboratory( Chengetal.,2013 ). AfterinitialtestsforU-ThdisequilibriumdatingatJCUprovedsuccessful,asuiteofflowstone sampleswascollectedfromtheDinalediChamberandpreparedasduplicatesamplesfordoubleblinddatinganalysesatJamesCookUniversity(JCU)andattheUniversityofMelbourne(UoM). Methodology:JamesCookUniversity(JCU) Samplepreparation ± Foreachsamplea100±200mgflowstonefragmentwasremovedwitha hand-heldDremeltool.Thepurestflowstonelayersweretargetedforsampling,anddarklayerscon-tainingdetritalmaterialwereavoided.Samplesweretakenparalleltolayering,typicallyacrossa3mmthickzoneandanyattachedsedimentorimpuritieswereremoved. Priortoitbeingdissolved,eachflowstonesamplewascleaned.Sampleswerefirstloadedinto centrifugetubesfilledwithethanolandagitatedinanultrasonicbathfor10min.Thisprocedurewasperformedtwiceforeachsample,withcleanethanolforeachbath.Thecentrifugetubescontainingthesampledfragmentswerethenfilledwithultrapurewater(18.2M W )andagitatedinanultrasonic bathforapproximately2min.Thisprocedurewasperformedthreetimesforeachsample,changingtheultrapurewatereachtime,afterwhichsamplesweredriedunderafumehood. Datingprotocol ± Chemicalprocedureslargelyfollow Horwitzetal.(1992) and (1993) .Dried samplesweretransferredtoTefloncontainersandspikedwithamixed229Th-233U-236Utracerprior todissolutionin3MHNO3.ChemicalseparationofUfromThwasperformedusingeitherachromographicextractionmethodusingEichromUTEVAresin( Horwitzetal.,1992 , 1993 )orEichromTRU resin( Louetal.,1997 ).TheUTEVAprocedureproducedhigherUyieldsandbetterseparationofU fromTh,andhassincebeenadoptedasthestandardprocedureatJCU.Analyticalblanksfrombothprocedureswereindistinguishableandlow(<10picogramsU).Priortoanalysissamplesweretreatedwith32%H2O2todestroyanyparticlesofresin. UraniumandThisotopemeasurementswereperformedwithaThermoScientificNeptuneMCICP-MSinstrumentattheAdvancedAnalyticalCentreofJCU.Theinstrumentisequippedwithasin-glecentralsecondaryelectronmultiplier(SEM)and9Faradaydetectorsandamplifierswith1011W resistors.MeasurementswereperformedusingaCetacAriduscombinedwitha100 m lPFAnebulizer forsampleintroduction.Anexponentialmassfractionationlawwasusedtocorrectisotopicratios.ForUisotopeswenormalizedratiostoa235U/238Uratioof137.794( Goldmannetal.,2015 ).ForTh Dirks etal .eLife2017;6:e24231. DOI:10.7554/eLife.24231 40of59 Researcharticle GenomicsandEvolutionaryBiology

PAGE 41

isotopes,sampleswerebracketedwithstandardsolutionsofCRM-145andcorrectedusingthemassbiasdeterminedfortheUstandard.ForisotopesmeasuredontheSEM,wecorrectedfortailsusingthelogmeanoftherelevanthalfmasses.ForisotopesmeasuredontheFaradaysdetectors,theshapeofthetailandtailingparametersweredefineddailyusinga>25VofasolutionofU005A;withtheseparametersacorrectionforthefamilyofUtailswasmade. Werepresentageuncertaintiesthatincludepropagated2 s envelopesonisotoperatiosand decayconstants,aswellasanerrorenvelopeontheassumedinitial230Th/232Th.Theassumedvalue forinitial230Th/232This0.83 ± 0.50(2 s ).Thesearethemainsourcesofuncertaintyfortheageestimates,andwetreatthemasthreesourcesofuncorrelateduncertainty.Uncertaintyonmeasurediso-toperatiosreflectscountingerrorsandvariabilitywithinamassspectrometricanalysis.Decayconstantuncertaintyenvelopesareafter Chengetal.(2013) ,andwepropagateuncertaintyon tracer(spike)isotopeconcentrations.Uncertaintiesarerepresentedassymmetricinboththeolderandyoungerdirectionandareexpressedinka. Methodology:TheUniversityofMelbourne(UoM) Samplepreparation ± SampleswerepreparedatJCUasdescribedabove,andsenttoUoMfor analysis. Datingprotocol ± AnalysesatUoMweredonebymulti-collectorinductivelycoupledplasma massspectrometry(MC-ICP-MS),usingtheanalyticalmethodsof Hellstrom(2003) .Anassumed valuefortheinitial230Th/232Thactivityratioof1.5 ± 1.5(2 s )wasusedtocalculatecorrectedagesin conjunctionwiththedecayconstantsof Chengetal.(2013) .Allagesareexpressedinkaandcorrectedforinitial230Thusing Equation1 of Hellstrom(2006) .234U/238Uactivityratiosweredeterminedafter Hellstrom(2003) and Drysdaleetal.(2012) .Alluncertaintiesarepresentedas2 s . ESRdating BackgroundtoESRdating Electronspinresonance(ESR)datingincombinationwithU-seriesdataofferalargelynon-destruc-tiveapproachfordirectdatingofhumanfossilremains(e.g., Gru È n,1989 , 1997 , 2009 ; JoannesBoyauetal.,2010 ; Joannes-BoyauandGru È n,2011 ; Joannes-Boyau,2013 ).Thebasicprinciplesof ESRdatingcanbefoundin Gru È n(1989) , (1997) ,andarebrieflysummarizedherewithreferenceto homininteeth. Whenthehighlycrystallinematerial(hydroxyl-apatite)thatformstoothenamelisexposedtoionizingradiation,resultingfromtheradioactivedecayofnaturallyoccurringradiogenicisotopes(mainlyU,ThandK)andcosmogenicrays,unpairedelectronsinthecrystallatticecanbemovedfromtheirnormalvalencebandtoahigherenergylevelorexcitationstate.SomeoftheseexcitedelectronsbecometrappedinchargedeficitsiteswithinthecrystalstructuretoformparamagneticcentresthatcanbemeasuredwithESRspectroscopy.Thenumberoftrappedelectronsbuildsupovertimeasafunctionofthestrengthofthegammarayintensity(ordoserate)inthesurroundingenvironment.Forteethburiedincavesedimentdeepunderground,radiationismainlyderivedfromradioisotopescontainedinthesurroundingsedimentsandthetoothitself.Thus,atoothactsasanaturaldosimeterinwhichthetotalaccumulateddoseanddoseratecanprovideanageestimate( Gru È n,1989 , 1997 ).TheESRsignalcontainedwithinatoothandthedoseratecanbemeasured directly.However,thedoseratecanvaryovertimeasUcanbehighlymobileinwetcaveenviron-ments(e.g., Gru È n,2009 ),andcanmoveinandoutofteeth( Gru È netal.,2008a ).Theenvironmental doseratecanbemeasuredusinginsitugammarayspectrometry,orcanbedeterminedfrommea-suredvaluesofnaturalisotopes(mainlyU,ThandK)andcalculationsofthecosmicraycontributionsfollowingmodels.Theinternaldoserate(insidethetooth)canbedeterminedbymeasuringthepresentday230Th/234U-ratiosfromwhichanuptakehistoryforthetoothcanbemodelled(as opposedtoapplyingassumeduptakemodels). UsingthecombinedresultsofU-ThanalysesandESRmeasurements,moreaccurateageestimatescanbeobtainedthanwouldbepossiblewithESRdatingalone.ToovercometheproblemofnotaccuratelyknowingthecomplexUuptakehistoryofatooth,theU-ThdataisusedtoestablishtherelationshipbetweentheESRequivalentdoseandapparentU-Thageasdefinedbytheequation: Ut …†ˆ U m t = T …† p ‡ 1 (1) Dirks etal .eLife2017;6:e24231. DOI:10.7554/eLife.24231 41of59 Researcharticle GenomicsandEvolutionaryBiology

PAGE 42

InwhichU(t)istheUconcentrationatt,UmisthemeasuredUconcentration,Tistheageofthe sampleandptheuptakeparameterlinkedtotheuptakemodeldeterminedfromU-Thmeasure-mentsinthetooth(closedsystemorearlyuptake:p= 1;linearuptake:p=0,lateuptake:p>0 ( Gru È n,2009 ). ThemethodologiesusedtoobtainUuptakemodelsandESRdatesareexplainedseparatelyfor eachofthelaboratories.Forallreportedages,decayconstantsfor234Uand230Tharefrom Chengetal.(2000) . MethodologyforcombinedU-seriesandESRdating:SouthernCrossUniversity(SCU) Samplepreparation ± Smallfragmentsofenamelwereremovedfromeachofthefourteethwith thehelpofahand-helddiamondsawfollowingprotocolsinGru È netal.(2008b)and JoannesBoyau,2013 .Anydentineattachedtotheenamelfragmentwasremoved(1841hadnoobservable dentine),afterwhichalayerof100 m mwasremovedfromtheoutersurfaceofeachfragmentwitha rotarytool.Foreachtooth,theenamelfragmentsandasectionofdentinedirectlyunderlyingeachenamelfragmentwereanalysedforUandThconcentrations.U-seriesanalyseswerealsoperformedonenamelanddentinefromtheremainingtoothintheimmediatevicinityofthefragmenttoassessvariationsinU/ThisotopicratiosandcalculateESRages. Datingprotocols:ESRdoseevaluation ± ESRdatingwasperformedatroomtemperatureona FreibergMS5000ESRX-bandspectrometerata0.1mTmodulationamplitude,10scans,2mWpower,100Gsweep,and100KHzmodulationfrequency.AtSCU,thesampleswereirradiatedwithX-raysinaFreibergX-rayirradiationchamber,withaVarianVF50X-raygunatavoltageof40KVand0.5mAcurrent,withdoseratecalibrationsdependingontheoutputvalueoftheX-raygun.ESRintensitieswereextractedfromT1-B2peak-to-peakamplitudesoftheESRsignalofenamel( Gru È n,2000a ). Eachenamelfragmentwasmountedintoapara-filmmouldwithinaTeflonsampleholderto recordtheangulardependencyintheESRresponse.IrradiationwasperformedbyexposuretotheX-raygunwithnoshieldingofthesource.Toestimatetheequivalentdose(DE),eachfragmentwas irradiatedinstepswithexponentiallyincreasingirradiationtimes(forsamples1767,1788and1810,thesestepswere:90s,380s,1080s,1800s,3600s,7200s,14400s,2,8800sand7,9200s,andforsam-ple1841:90s,380s,1080s,1800s,3690s,7200s,1,4400s,2,8800s,4,3200sand9,9000s).Duringeachirradiationstep,theoutputoftheX-raygunwasrecordedtocalculatethedoseratereceivedbythesample(forsamples1767,1788,1810and1841thesevalueswere0.178Gy/s,0.169Gy/s,0.188Gy/sand0.240Gy/srespectively.Foreachirradiationstepthefragmentwasmeasuredover180oinx,yandz-configurations( Joannes-BoyauandGru È n,2011 ; Joannes-Boyau,2013 ).Isotropic andbaselinecorrectionswereapplieduniformlyacrossthemeasuredspectra.TheamountofNOCOR'sinthenaturalsignalwasestimatedusingtheprotocolof Joannes-Boyau(2013) .TheESR doseresponsecurveswereobtainedbyusingmeanESRintensitiesandassociatedstandarddevia-tionsfromtherepeatedmeasurements. FittingprocedureswerecarriedoutwithMCDOSE2.0softwareusingaMarkovChainMonte Carlo(MCMC)approachbasedontheMetropolis-Hastingsalgorithm.TheprogramusesaBayesianframework,wherethesolutionispresentedasafullprobabilitydistributionoftheequivalentdose( Metropolisetal.,1953 ).DEvalueswereobtainedbyfittingaSingleSaturatingExponential(SSE) attheappropriatemaximumirradiationdose(Dmax)following DuvalandGru È n(2016) (using1767 Dmax=1264Gy,1788Dmax=2465Gy,1810Dmax=2735Gyand1841Dmax=3526Gy). Datingprotocols:Useriesanalysis ± TheUandThconcentrationsineachoftheenamelfragmentsandsurroundingdentineweremeasuredbylaserablation,usinganESINW193ArFExcimerlasercoupledtoaMC-ICPMSNeptunePlusattheUoW.Sectionsofenamelanddentineweremappedusingsmallrasterstodocumentcompositionalvariabilityintheteethandconstraindiffu-sionprocesses( Figure9a ). Atotalof86separateU-seriesanalyticalrunswereperformedacrossthedentineandenamelsurfaceforthethree H.naledi teeth.Eachindividualrunconsistsofanaveragevalueobtainedacrossa rasterorablationtrackmeasuring200 m mx700 m minsize;thatis,asuccessionofshortmeasurementswastakenalongarasterandthenaveragedintoonevalue.Measurementswereperformedatanablationrateof20Hzandatranslationspeedof50 m m/s. Dirks etal .eLife2017;6:e24231. DOI:10.7554/eLife.24231 42of59 Researcharticle GenomicsandEvolutionaryBiology

PAGE 43

Therasterswerepositionedinaseriesoftransectsfollowingthegrowthaxisofthedentaltissue inlocationsimmediatelyadjacenttotheareafromwhichtheESRfragmentwastaken,andontheESRfragmentsthemselves( Figure9a ).ToevaluatetheESRinternaldosimetryofeachtooth,three zoneswereanalysed:(i)onealongtheenamelgrowthaxis;(ii)oneperpendiculartotheDentineEnamelJunction(DEJ),and;(iii)oneonthedentine( Figure9a ).Additionalmeasurementswere takenacrosseachtoothinareasawayfromtheESRfragmenttoassessUvariabilityanddiffusiongradients.TheU-seriesageestimates( Table4 )arebasedontheaverageU-seriesisotopevaluesfor eachablationrun.NoagecalculationsweredoneforareasofthetoothwheretheUconcentrationswerebelow0.5ppmorwheretheU/Thratiowasbelow250(markedinredin Table4 ).Ageswere calculatedwithIsoplot3.75( Ludwig,2012 ),anduncertaintiesarereportedas2 s . BaselineanddriftwerecorrectedusinganalysisoftheNIST612glassstandard,whiletwocoral standards(theMIS7FaviidandMIS5PoritescoralsfromtheSouthernCookIslands; Woodroffeetal.,1991 )wereusedtocorrect234U/238Uand230U/238Uratiosandassesstheaccuracyofmeasurements.EachcoralstandardwasanalysedbysolutionMC-ICPMSatUoWandusedforreference.Toaccountforpotentialmatrixeffects,abovidtoothfragmentfromSouthAfricawithknownisotopeconcentrations(U-seriesatequilibrium)wasusedtoverifymeasurements.Toaccountfortailingeffects,measurementswerecarriedoutathalf-massesof229.5and230.5for230Thand 233.5and234.5for234U. MethodologiesforcombinedU-seriesandESRdating:CENIEH,GriffithUniversityandAustralianNationalUniversity Samplepreparation ± Oneenamelfragmentwasextractedfromeachofthetwo H.naledi teeth 1788and1810,andanalyzed. Datingprotocols:ESRdoseevaluation ± Thetwoenamelfragments(samples1788and1810) weremeasuredbyESRfollowingtheanalyticalproceduredevelopedby Gru È netal.(2008b ).Dose evaluationswerecarriedoutattheCENIEHusingasub-exponentialdosestepdistribution( Gru È nandRhodes,1992 )forthefollowinggammadoses(6.705Gy/min):0,13.4,40.2,93.9,201.2, 415.7,844.8,1649,3058,5471,8690,12713,17943,25989,34861and61412Gy. ESRmeasurementswereperformedatroomtemperaturewithaEMXmicro6/1BrukerESRspectrometercoupledtoastandardrectangularER4102STcavity,withthefollowingacquisitionparam-eters:1±10scans,2mWmicrowavepower,1024pointsresolution,15mTsweepwidth,100kHzmodulationfrequency,0.1mTmodulationamplitude,20msconversiontimeand5mstimecon-stant.Toensurereproduciblemeasurements,eachfragmentwasmountedinapara-filmmouldwithinaTeflonholderinaZ-configurationonly( Gru È netal.,2008b ),whichcanbeinsertedintoa BrukerER218PG1programmablegoniometer.BecausetheESRsignalsinfragmentsshowverystrongangulardependences,theESRspectraofeachdosestepwererecordedevery10 Ê over360 Ê . ESRintensitieswereextractedfromT1-B2peak-to-peakamplitudesoftheESRsignalofenamel(after Gru È n,2000a ). FittingprocedureswerecarriedoutwithMicrocalOriginPro9.1.softwareusingaLevenberg-Marquardtalgorithmbychi-squareminimisation.DatawereweightedbytheinverseofthesquaredESRintensity(1/I2)( Gru È nandBrumby,1994a ).Thedoseresponsecurves(DRC's)wereobtainedby averagingtheT1-B2ESRintensitiesrecordedforalltheanglesatagivenirradiationdose.ThefinalDEvalueswereobtainedbyfittingaSingleSaturatingExponential(SSE)throughtheESRintensities andbyselectingtheappropriatemaximumirradiationdose(Dmax)inordertoavoiddoseoverestimation( DuvalandGru È n,2016 ; Table10 ). Datingprotocols:Useriesanalysis ± LaserablationU-seriesanalyseswerecarriedoutatthe AustralianNationalUniversity(ANU),usingacustom-builtlasersamplingsysteminterfacedbetweenanArFExcimerlaserandaFinniganNeptuneMC-ICP-MS( Egginsetal.,2003 , 2005 ),following principlesandproceduresdescribedin Gru È netal.(2014) .Toaccountfortailingof238Uintotheion countersfor234Uand230Th,measurementswerecarriedoutathalf-massesof229.5and230.5for230Thand233.5and234.5for234U.Thetailingcorrectionswere7cts/V(238U)and17cts/V(238U)for230Thand234U,respectively. U-seriesdatawereobtainedfromdistinctivespotanalysesonpolishedsurfaces.Firstly,allspot locationswerecleanedwiththelasersetto263 m mfor10s,followedbytheanalyticalrunwiththe lasersetto203 m mfor60s.Therimfromthecleaningrunisclearlyvisiblein Figure9 .Theenamel ofsample1788( Figure9b )wasanalysedalongtwotransectsatanobliqueangletothesurfaceto Dirks etal .eLife2017;6:e24231. DOI:10.7554/eLife.24231 43of59 Researcharticle GenomicsandEvolutionaryBiology

PAGE 44

increasethenumberofanalysesacrosstheenamellayer.Thedentinewasanalysedonapiecethatseparatedjustbelowthetwoenameltracks.TheresultsfromtheenamelspotsintransectAforsam-ple1788arebiasedtowardsthedomainclosetothedentine.Itwasfurthernotedthatthelasttwoanalysescontainedsomedentinematerial,andtheseanalyseswere,therefore,neglected( Figure9d ).Sample1810wasanalysedalongtwotracksperpendiculartotheenamelsurface ( Figure9c ).Forsample1810,analysesofenamelspotsonlyshowhigherUconcentrationsintransectB( Figure9h )wherethespotsoccurclosetothedentineboundary,whileintransectA ( Figure9g )allenamelspotshaveuniformlylowUconcentrations.Theaverageenamelconcentrationswerederivedfromtheaveragesofthetwotransects. Theanalyticaldataoftheenamelanddentinesections( Table5 )werecombinedtoprovidethe datainputfortheESRagecalculations.BecausetheenamellayeranalyzedbyESRwasnotcleaned Figure9. PhotographsillustratingthesamplingapproachestakenbySCU-UoWandGU-ANUinobtainingthe U-Thresultspresentedin Tables4 and 5 .( a )Comparisonofsamplinggridsacrosstheenamel-dentineboundary measuredbySCU-UoW(redlines)vs.GU-ANU(bluecircles).SCU-UoW(redlines)measuredaseriesofparallel,shallow(<5 m m)pitsalonggridlinesacrosstheteethandaveragedUconcentrationsacrosseachgrid.GU-ANU (bluecircles)measuredtheaveragecompositionofthetoothinsinglespotsthatwerelaser-boredalongprofilesacrosstheteeth,andreportresultsforeachspot.( b,c )LocationsofLA-ICP-MSspotanalysesforteethsamples 1788( b )and1810( c )conductedbyGU-ANU.Thedetailedtransectsareshowninpanels( d )to( h ). DOI:10.7554/eLife.24231.020 Dirks etal .eLife2017;6:e24231. DOI:10.7554/eLife.24231 44of59 Researcharticle GenomicsandEvolutionaryBiology

PAGE 45

onbothsides,anaverageexternalalphadoseratecomponentof8 ± 2 m Gya 1wasusedbasedon thealphaattenuationvaluesof Gru È n(1987) . ApparentU-ThageswerecalculatedwithIsoplot3.75( Ludwig,2012 ),anduncertaintiesare reportedas2 s .NoagecalculationsweredoneforareasofthetoothwheretheUconcentrations werebelow0.5ppmorwheretheU/Thratiowasbelow250(markedinredin Table5 ).Ageswere calculatedassumingclosedsystembehaviour(CS),andcomparedwithageresultsassumingcontinu-ousdiffusionmodels(after Sambridgeetal.,2012 ). Determiningtheenvironmentaldoserate ApartfromradiationderivedfromUandThcontainedwithinthedentaltissueoftheteeth,agecal-culationsmusttaketheenvironmentaldoserateintoaccount.Theenvironmentaldoseratemainlyresultsfromgammaandbetaradiationderivedfromtheimmediatesurroundingsoftheteethwithanadditionalcomponentfromcosmicradiation,whichinadeepcaveenvironmentisusuallyverysmall.Incalculatingtheageswehaveusedacosmicdoserateof15 ± 1 m Gya 1forallsamples assumingthatthefossilswereoverlainbya20mthickroofofdolomitewithadensityof2.80 ± 0.05 g/cm3( BarboutiandRastin,1983 ). GammaandbetaradiationfromtheimmediatesurroundingsoftheteethcomesfromU,Thand Kinthesedimentsinwhichtheteethareembedded,andispartlyattenuatedbythewatercontentofthesediments.Thus,theU,Th,Kandwatercontentofthesedimentsneedstobemeasured( Table6 ).Themeasuredwatercontentinsedimentsurroundingsample1841is27.7%.Thisvalueis expectedtovaryovertime,butnotbymuchgiventhattheDinalediChamberoccursdeepinsidethecave,closetothewatertable,wheresedimentsareexpectedtohavealwaysbeenclosetowatersaturated.Therefore,incalculatingtheageswehaveassumedthewatercontenttobe25 ± 10%. Uranium,Th,KconcentrationsweremeasuredbyICP-MSatSCUandattheUoWusingbothleach-ing(ina1:3mixtureofnitricandhydrochloricacid)andtotaldissolution(ina1:3mixtureofnitricandhydrochloricacidwithadditionalHFusinganAgilent7700solutionandaThermoIcap)meth-ods.LeachingresultsgavelowerUconcentrationsreflectingincompletedissolutionofallU-bearingphases.Onlythetotaldissolutionresults( Table6 )havebeenusedtocalculatetheenvironmental doserate. Iftheteethwerenotfullycoveredbysedimentuptoadepthofabout30cm(whichistheaverageattenuationlengthofgamma-raysinmaterialwithadensityof ~ 2.5g/cm3)anestimatehasto bemadefortheburialhistoryoftheteethaswellasthemoregeneralbackgroundradiationinthecave.TheenvironmentaldoserateintheDinalediChamberwasmeasuredinsituwithaportablegammarayspectrometeronthesurfaceofUnit3,nexttotheplaceswheresamples1767,1788and1810werecollected.WemeasureddoseratesforThandK,thatwereequaltoorsomewhathigherthanthosederivedfromthechemicalanalysisofthesediments,whichsuggeststhatthebackgroundgammaradiationinthecavemaybealittlehigherthaninthesediments.However,theinsitugammameasurementsvariedconsiderablyfromoneplacetothenext,whereasthemeasuredcon-centrationsofU,ThandKinsub-units3aand3bareconsistent( Table6 ).Samples1767,1788and 1810werecollectedfromnearsurfacewheretheymayhaveresidedforalongtime,and,therefore,mayhavereceivedbetween25%and50%oftheirgammadoseratefromthewidercaveenviron-ment.However,theresultinggammadoseratesforthisscenarioarenotsignificantlydifferentfromasituationinwhichtheteetharefullyburiedwhenconsideringthespectrometerreadings.Giventheuncertaintieswiththeinsitugammarayspectrometermeasurements(seebelow),weareconfi-dentthattheanalyticaldataofthesedimentprovidemorerobustconstraintsforcalculatingtheenvironmentaldoserates.Therefore,theteethhavebeenmodelledasiffullyencasedinsedimentassuminginfinitematrixdoseratesforsedimentandusingthemeasuredvaluesofU,ThandKlistedin Table6 . Thegammadoseratescalculatedforsedimentcollectedfromtheimmediatevicinityofthesampledteeth( Figure10 )werecomparedwithgammadoseratescalculatedfromU,ThandKconcentrationsforsedimentsamplescollectedatregularintervalsalonga53cmdeep,verticalprofileinthesondageatthecenteroftheexcavationpit( Figure10 ).Forthisprofilesamples1to4werecollectedfroma13cmthicklayerofhominin-bearingsedimentbelongingtosub-unit3b,andsamples5±8werecollectedfromunderlyingsedimentsbelongingtosub-unit3a,toatotaldepthof53cm.Thevariationinthegammadoserateshowsnocleartrendwithdepth( Figure10 ),andthesamples alongtheverticalprofilewereusedtoobtainanaveragegammadoserate.Sample2fromthetop Dirks etal .eLife2017;6:e24231. DOI:10.7554/eLife.24231 45of59 Researcharticle GenomicsandEvolutionaryBiology

PAGE 46

3cmofUnit3wasexcludedbecauseofanomalouslyhighThconcentrations(thissamplealsocon-tainsmanysmallbonefragments). Anadditionalprobleminestimatingthegammaradiationforthesedimentsmayariseif222Rn lossoccursasaresultofdegassing.Thisisofimportance,becausevirtuallyall(98%)ofthegammaradiationgeneratedinthe238Udecaychainisproducedby222Rnanditsdaughters(e.g., Gue  rinetal.,2011 ).If222Rnlossisdetected,thencorrectionshavetobemade.Insitugammaray spectrometermeasurementsyieldedKandThvaluesbroadlyinagreementwithanalysedconcentra-tions,butdidnotdetectU.SinceUisdeterminedviathe(gamma)emissioninthedecayof214Bi, whichisapost-222Rnnuclideinthe238Udecaychain,weconcludethattheabsenceofthissignalis dueto222Rnloss( Aitken,1985 ).The222Rnlosswasconfirmedbyanalyzingasedimentsamplethat wascollectednexttosample1841,forwhichhighresolution,Ge-gammaspectrometryshowed ~ 80%Rnloss. ItisdifficulttoestimatewhetherRndegassingoccurredduringtheentireburialhistoryofthe teeth.Therefore,agecalculationswereperformedfortwodifferentscenarios:(i)asituationwhereitisassumedthatthepresentdaysituationof80%Rnlosspersistedduringtheentireburialhistory,and(ii)asituationinwhichnoRnlossoccurred( Table7 ).Assumingwatercontentof25 ± 10%,the averagegammadoserateforthesedimentshasbeencalculatedat724 ± 116 m Gy.a 1assumingno radonloss,and534 ± 69 m Gy.a 1assuming80%Rnloss( Figure10 ). CombinedUS-ESRagecalculations Allagecalculationsareprovidedwith2 s uncertainties,andwerecarriedoutusingtheUS-ESRprogramof Shaoetal.(2014) ,whichutilizesthedoserateconversionfactorsof Gue  rinetal.(2011) .In doingthecalculations,theinputparametersandcriterialistedin Table6 wereused,andageresults aregivenin Table7 .Inallcalculationswehavemadethefollowingassumptions:(i)post-230Th daughterelementsareinequilibriumindentaltissues,whichisthestandardassumptioninESRdat-ing;and(ii)complete(effective)burialofthesamplesoccurred(i.e.,infinitematrixassumptionforthegammadoseratemeasuredinsediment). Resultshavebeencalculatedforthetwoscenarios( Table7 ):(1)80%Rnlossinthesediment;and (2)post-Rnequilibriumindentaltissueandsediment(i.e.,noradondegassing).Inbothscenariosthewatercontenthasbeentakenas25 ± 10%.Scenario1isbasedonmeasured,current-dayvalues andprobablyprovidesamaximumageestimate,becauseitisunlikelythatthehighdegreeofRn Figure10. Gammadoseratereconstructionsderivedfromanalyticaldataofsedimentsamplescollectedaround ESRsamples1767,1788and1810(closedcirclesanddiamonds),combinedwithsamplesfromaverticalprofileintheexcavationpitandsondage(opencirclesanddiamonds).Thedatashowlittlevariationindoseratewithdepth(seetextforexplanation). DOI:10.7554/eLife.24231.021 Dirks etal .eLife2017;6:e24231. DOI:10.7554/eLife.24231 46of59 Researcharticle GenomicsandEvolutionaryBiology

PAGE 47

lossobservedtodaywouldhaveoccurredthroughoutthehistoryofthecave.Scenario2willprovideaminimumageestimate,becauseRnlosshasbeenobserved.Thesetwoscenariosprovidethemostreasonablerangefortheageofthefossilteeth. AdditionalmodellingwascarriedouttoassesstheeffectsofearlyUuptakeontheESRage,ina closedsystemsetting(CSUS-ESR, Gru È n,2000b ).ThisshowedthatthedifferencesinUS-ESRand CSUS-ESRmodelagesarelessthan4%,thatsis,whichiswellwithintheuncertaintyrangesofthelistedUS-ESRages( Table7 ).Therefore,CSUS-ESRagesarenotfurtherconsidered. ESRdoseevaluation:inter-laboratorycomparison Eachlaboratory(SCUandCENIEH)independentlymeasuredtheDEvaluesfollowingthemethods outlinedabove.ToensurereliableDEvalues,themaximumirradiationdose(Dmax)wasselectedin accordancewiththerecommendationsmadeby DuvalandGru È n(2016) ,inordertoavoidDEoverestimation(i.e.,forDEvaluesbetween100±500Gy,theDmax/DEratiohastobekeptbetween5±10; forDevaluesbetween1000±2000GytheDmax/DEratiohastobekeptbetween1±2; Table10 ).Normaliseddoseresponsecurves(DRC's)areshownin Figure11 . Samples1767,1788and1810haveDEvaluesthatvarywithinanarrowrangebetween200and 300Gy,whiletheDEvalueofsample1841issignificantlyhigher(>1500Gy).Forthetwosamples thatweremeasuredbybothlaboratories,theSCUproceduresystematicallyprovideshigherDEresults.Thesedifferencespartlyresultfromdifferentfittingprocedures:whenplottingtheSCUexperimentaldatapointsusingtheCENIEH-GUfittingprocedure,thecalculatedDEvaluesshift by<2%forsamples1767,1788and1841,and 6%forsample1810( Table10 ).Bothlaboratories alsouseddifferentirradiationsources(gamma-raysatCENIEH,andX-raysatSCU),whichmayhavehadsomeinfluenceonresults.Athirdfactorthatmayhavemostfundamentallyinfluencedthediffer-enceinmeasuredDEvaluesistheradiationsensitivityobservedforeachfragment( Figure11 );for sample1810,theDRC'sfromCENIEHandSCUaresimilar,butforsample1788theyaredifferent,indicatingthattheresponseofthetwofragmentsfromsample1788totheradiationsourcewasdis-tinctlydifferent.Eachlaboratoryhadsampleditsownfragmentfromdifferentdomainsofthetooth,withdifferentradicalconcentrationsandcrystallinity,andtheU-concentrationsinthefragmentusedbySCUweresomewhathigherthanthoseusedbyGU-CENIEH( Tables4 and 5 ).However,inspite ofthesedifferences,theagesagreewithinerror( Table7 ). Theuseofenamelfragmentsinmeasuringthedoseallowsustodifferentiatebetweentherelativecontributionsofnon-orientedCO2 -radicals(NOCOR's)versusanisotropicradicals(AICOR's)(e. g., Gru È netal.,2008b ). Joannes-BoyauandGru È n(2011) recentlyshowedthatgammairradiationin thelaboratorymayproduceadditionalunstableNOCOR'swhencomparedtogammairradiationinanaturalenvironment,whichcouldleadtoanoverestimationofthemeasureddoseifthiscontribu-tionisnotremoved.Toevaluatethispossiblebias,wefollowedMethod3describedin JoannesBoyau(2013) fortheextractionoftheNOCOR'sfromthemainradiation-inducedESRsignal( Figure12 ).Whenafragmentisrotatedover360 Ê ,theESRintensityvaries.Theratioofthemaximum (Imax)andminimum(Imin)intensities(Imax/Imin)isusedasaproxytoquantifytheincreasedanisotropy potentiallyinducedbyeithergammaorX-rayirradiations.Theevolutionofthisratiowiththeirradia-tiondoseisshownin Figure12 .Whencomparingtheresultsfromthetwolaboratoriesitisclear thatbothobtainsimilarresults.FormeasurementsconductedatCENIEHtheImax/Iminratiosfor Table10. ESRfittingresultsobtainedbySCUandCENIEH-GU.BothlaboratoriesemployedaSingleSaturatingExponential(SSE)fittingfunction.Dmaxwasselectedinaccordancewith DuvalandGru È n(2016) toavoidDEoverestimation.SCUresultsinbracketsshow DEvaluesthatwereobtainedbySCUusingtheCENIEH-GUprocedure(seetextfordetails). SCU Cenieh-gu SampleDE(Gy)Dmax(Gy)Dmax/DEDE(Gy)Dmax(Gy)Dmax/DE 1767194 ± 4( 193 6 )12647±±± 1788232 ± 8( 232 22 )12045159 ± 11164910 1810296 ± 14( 281 34 )27359232 ± 2916497 18411676 ± 127( 1648 500 )35262±±± DOI:10.7554/eLife.24231.022 Dirks etal .eLife2017;6:e24231. DOI:10.7554/eLife.24231 47of59 Researcharticle GenomicsandEvolutionaryBiology

PAGE 48

samples1788and1810remainnear-constant(between1.22±1.24)asthegammadosevaries.FormeasurementsconductedatSCUtheImax/Iminratiosremainnear-constantaswell(andvarywithin uncertainty,between1.6±2.0forsamples1767,1788and1810,andremainaround1.34forsample1841; Figure12 )astheX-raydosevaries.ThedifferencesintheImax/Iminratiosforeachtoothcalculatedbythetwolaboratoriesresultsfromtheorientationchosen,thespatialvariationsoftheESRsignalwithintheenamellayer,andthedifferentcrystalorientationandfragmentpositionduringESRmeasurements.ThefactthattheresultsfrombothlaboratoriesshowconstantratiosindicatesthatnoadditionalNOCOR'shavebeencreatedbybothgammaandX-rayirradiations.Conse-quently,nocorrectionshadtobeappliedtothemeasuredDEvalues. OSLdating BackgroundtoOSLdating Opticallystimulatedluminescence(OSL)datingisawidelyusedmethodforestimatingthelasttimethatsedimentswereexposedtosunlightwithinthepast300±500ka(e.g., Aitken,1985 ; MurrayandWintle,2000 ).Sand-sizedgrainsofquartzandK-feldspar(twocommonlyoccurring minerals)arepreferredforOSLdating.Thesegrainsmaybemeasuredcollectivelyinasinglecom-positesample(analiquot),whichprovidesanaveragesignal,orasindividualgrains,whichprovidesgreaterinsightsintothedepositionalhistoryofthesediments. OSLdatingisaradiationdosimetricdatingtechniquebasedonthetime-dependentaccumulation ofradiationdamageinminerals( AdamiecandAitken,1998 ; MurrayandWintle,2000 ),asaresult ofexposuretolowlevelsofionisingradiationintheenvironment.Theintensityofradiationdamagereflectsthetotalamountofenergy(the`equivalentdose')absorbedfrombackgroundradiationbythemineralovertime.Theradiationdamagecanberemovedfromthemineralbyexposuretoheatorlight,whichisaccompaniedbythereleaseofasmallamountoflightorluminescence.OSLdating Figure11. ESRdoseresponsecurves(DRC's)obtainedforthesamples1767,1788,1810and1841.Tofacilitate comparison,allDRC'shavebeennormalisedtotheintensityofthenaturalpoint(=1). DOI:10.7554/eLife.24231.023 Dirks etal .eLife2017;6:e24231. DOI:10.7554/eLife.24231 48of59 Researcharticle GenomicsandEvolutionaryBiology

PAGE 49

estimatestheagewhensedimentwaslastexposedtolight(`bleached')duringerosion,transportanddepositionofthemineralgrains.Oncesedimentsareburiedandnolongerexposedtosunlight,theluminescencesignalcanstarttobuildupagain( Duller,2007 ). Thelatentluminescencesignalthathasbuiltupinasedimentsamplecanbereleasedand recordedinthelaboratoryusinglight(theOSLtechnique).Thisluminescencesignalisrelatedtotheenvironmentalradiationdosethemineralhasreceivedsincethelastexposuretosunlight.Iftheenvi-ronmentaldoserateisdeterminedtogetherwiththeequivalentdosecontainedinthesample,itispossibletodetermineanageforthesediment,orratheradepositionalagewhichmeasuresthelasttimethesedimentwasexposedtosunlightorheat. MethodologyforOSLdating:UniversityoftheWitwatersrand(Wits) Samplepreparation ± ForsamplesOSL3,4and5,sedimentwasremovedfromthesampletubes undercontrolled,safe-lightlaboratoryconditions.Materiallocatedwithin2cmoftheendsofthetubeswasremovedtoisolateanyquartzgrainspotentiallyexposedtolightduringsampling.Thismaterialwasusedtomeasurewatercontent,andtodeterminethedosimetryofthesample.Theremainingsedimentwastreatedwith33%hydrochloricacidand20%hydrogenperoxidetoremovecarbonateandorganiccomponents.Quartzgrainswereisolatedfromdensermineralsandfeldsparsbyusingsolutionsofsodiumpoly-tungstatewithspecificgravitiesof2.70g/cm3and2.62g/cm3, respectively.Afterrinsing,dryingandsieving,thefinesand(180±212 m m)fractionwasetchedfor40 minin40%hydrofluoricacidtoremovetheouterlayer( ~ 10±15 m m-wide)affectedbyalpharadiation andanyremainingfeldspars.Subsequently,33%hydrochloricacidwasaddedtoremoveacidsolu-blefluorides.Eachsamplewasdriedandre-sievedinpreparationforequivalentdosedetermination. DeterminingtheEquivalentDose(De) ± OSLmeasurementswerecarriedoutonthe180±212 m mgrain-sizefraction.Amonolayerofgrainswaspreparedonsteeldiscsinaliquotsizesof ~ 30 grains.LuminescencemeasurementswereperformedonanautomatedRisùTL/OSL-DA-20reader.Devalueswereobtainedthroughcalibratingthe`natural'opticalsignalacquiredduringburial, against`regenerated'opticalsignalsobtainedbyadministeringknownamountsoflaboratorydose. Figure12. EvolutionoftheImax/Iminratiovstheirradiationdoseforthefourtoothsamples.(seetextfor explanation). DOI:10.7554/eLife.24231.024 Dirks etal .eLife2017;6:e24231. DOI:10.7554/eLife.24231 49of59 Researcharticle GenomicsandEvolutionaryBiology

PAGE 50

DeestimateswereobtainedusingtheSingle-AliquotRegenerative-dose(SAR)protocolof MurrayandWintle(2000) .Anappropriatepreheattemperaturewasdeterminedusinga`preheatdoserecoverytest'.Atestdoseof27Gywasusedtomonitorsensitivitychange.UptofivedifferentregenerativedosesweregiventodefinethelinearportionofthedoseresponsefromwhichDewas determined.TheOSLsignalswerestimulatedat125 Ê Cfor40swithblueLEDat90%poweranda finalOSLmeasurementwasusedtoclearoutanysignalthatmaybefoundinthermallyunstabletraps.TheDewasevaluatedusinganexponentialandlinearfittoasingleregenerativedosepoint. ErrorsonDeassumepoissonstatistics,withanadditional1%uncertaintyaddedtoeachOSLmeasurement( Galbraith,2002 ; Duller,2007 ). Toobtainaluminescenceage,aburialdose(Db)iscalculatedfromthealiquotmeasurements,in whichindividualvaluesforDearecombinedtoobtainasinglevalueforDb.Giventheoftencomplex distributionofDevalues,Dbiscalculatedusingcomplexstatisticalmodelsincludingthecentralage model(CAM),andtheminimumagemodel(MAM)( Galbraithetal.,2005 ). MeasuredDevaluesforquartzgrainsshowadegreeofvariabilitythatdependsinpartonthe crystallatticestructureofthegrainsandtheirburialhistory(e.g.,completevsincompletebleaching).Forwell-bleachedquartzwherealiquotshavebeenheatedpriortoirradiationthestatisticalvariabil-ity,expressedasthestandarddeviation,inmeasuredDevalueswillbelessthan20%( Olleyetal., 2004 ; Galbraithetal.,2005 ).Thisvariabilityinthedistributionisreferredtoasoverdispersion ( Galbraithetal.,1999 ).Anoverdispersionvalueof20%isusedtoguidetheuseofappropriatestatisticalmodelstocalculateaburialage:MAMisusedtodeterminetheburialagewhenoverdisper-sionexceeds20%;CAMisusedwhenoverdispersionisbelow20%.However,thisisonlyaroughguideandenvironmentalconditionsneedtobeconsideredindecidingthestatisticalmodeltobeused.TheCAMandMAMageswerecalculatedwithsoftwareprovidedbyGeoffDuller(UniversityofAberystwyth,UK). TheCentralAgeModel(CAM)computesanagevaluemuchlikeaweightedaverage,butit assumesanaturaldistributionofDevaluesratherthanasinglevaluefromwhichtocalculatethe age.ACAMisusedwhenitisassumedthatsamplesexperiencedhomogenousbleaching.TheMini-mumAgeModel(MAM)isusedifincompletebleachingisexpectedtobethereasonforanobservedskeweddosedistribution,orif(asinthiscase)thesamplecontainsmineralgrainsthathaveneverbeenatsurface( Galbraithetal.,1999 ).TheMAMusesatruncatednormaldistribution andfitsittotheindividualDedatapointstodeterminetheproportionofgrainsthatwerefully bleachedbeforetheyweredeposited( GalbraithandLaslett,1993 ). Environmentaldoseratedeterminations ± TheconcentrationsofU,ThandKinthesediment samplesofUnit1(sub-units1aand1b)weredeterminedattheUniversityofJohannesburgbyiso-topedilutiononaquaregialeachingtechniques.Amixed236U-229Thspike(allowingmeasurement of(U,Th)disequilibrium)wasaddedbeforeleaching.UandThwereseparatedfrommajorelementsusingionexchangecolumnswithDowex1 8anionexchangeresin;sampleswereloadedwith2ml of80%ethanol+20%5MHNO3,washedwith3mlofthesamesolutionandelutedwith0.5NHCl. UraniumandThwereanalysedtogetheronaNuInstrumentsPlasmaIImulticollectorICPmassspectrometer,usinganAPEX-Qdesolvatingnebulizer.Strongexcessesof234Uand230Th(234U/238U activityratiosof2.2±4.5,230Th/238Uactivityratiosof3.1±4.1)necessitatedsignificantupwardcorrectionstothedoserateforU,yieldingage-dependent`effectiveUconcentrations'tobeusedforthedoseratecalculations. Doseratecalculations(after Aitken,1985 )incorporatedbeta-attenuationfactors( Mejdahl,1979 ), doserateconversionfactors( AdamiecandAitken,1998 )andanabsorptioncoefficientusingthe presentwatercontent( Zimmerman,1971 )witha5%relativeuncertaintytoreflectpotentialtemporalvariationsinpastmoisturecontent.Thewatercontentmayaffectthevalidityoftheagesasitispossiblethatadifferent,possiblywetter,climateregimemayhaveprevailed.ThisissuggestedbythedisequilibriumintheU-seriesmeasurements.Thecosmicdoseratewasdeterminedasafunctionofaltitude,latitude,longitudeanddepth,accordingto PrescottandHutton(1994) equations,plus thesoftcomponentfrom Madsenetal.(2005) . Luminescenceagedeterminations ± Theluminescenceagewasobtainedbydividingthepalaeodosewiththetotaldoserate.TheerrorontheluminescenceageestimatesrepresentsthecombinedsystematicandexperimentalerrorassociatedwithboththeDeanddoseratevalues.Thereisno datumforluminescencedates,thereforetheagereportedistakenfromdateofsampling(i.e.,AD2015). Dirks etal .eLife2017;6:e24231. DOI:10.7554/eLife.24231 50of59 Researcharticle GenomicsandEvolutionaryBiology

PAGE 51

Palaeomagneticdating Methodologyforpalaeomagneticdating:LaTrobeUniversity(LTU) TheproceduresforpalaeomagneticanalysisofthespeleothemsamplefromRisingStarCavefollowthoseoutlinedin HerriesandShaw(2011) andacomprehensivereviewofspeleothemmagnetism canbefoundin LascuandFeinberg(2011) . Samplepreparation ± Thesamplewasdrilledvertically,acrossthelayering,usinganon-magnetic rockdrilltoproducethree2.5cmby ~ 5±6cmcoresfromtheupper,purerpartoftheflowstone. Carewastakentoremoverecentcontamination.Thesurfaceoftheflowstonehasacalcifiedcover-ingoffineclasticdustthatislikelytobemuchyounger.Therefore,theupperendsofthecoreswereremovedtomakesurethatonlyprimaryflowstonewasmeasured.Thecoreswerethencutinhalfattheirmidpointwheremorerecentcontaminationhasseeminglyoccurredwithinthesample(attheinterfacebetweendifferentgrowthphases; Figure13 ).ThiswasdoneusinganASCScientificnonmagneticsawwithbronzesawbladestoproduceatotalofsix2.5by2.5cmsubsamplecoresfromtheupperpartoftheflowstone;3fromtheuppermost2.5cm(phaseCflowstone)and3fromthelower2.5cm(phaseBflowstone).Thinnersamplescouldnotbeproducedasthesampleswereweak.Onlyasmallpartofthelowest,clasticrichlayer(phaseAflowstone)waspreservedonthebaseoftheblocksample( Figure13a )andthislayerwascutintotwo2 2cmcubesforanalysis. Palaeomagneticfielddeterminations ± Becausethesamplesallconsistofflowstonespeleothem, alternatingfield(AF)demagnetisationwasthesolemethodofmagneticcleaningthatwasusedonthesamples.AFhasbeenshowntobeeffectiveforrecoveringtheprimarypalaeomagneticsignalformedwithinSouthAfricanspeleothemsforbothdetritalinclusionsdepositedduringphasesoffloodingorwithinthewaterformingthespeleothemitself(detritalremanentmagnetisation;DRM),andforchemicalprecipitation(ChemicalRemanentMagnetisation;CRM)ofironphaseswithindrip-water( HerriesandShaw,2011 ; Pickeringetal.,2013 ).Samplesweredemagnetisedusingan AGICOLDA5AFdemagnetiserandmeasuredusinganAGICOJR6spinnermagnetometeronthehighspeedsetting.SampleswereanalysedusingtheprogramPlotcoretoestablishtheprimaryrem-anencedirectionusingprinciplecomponentanalysisofZijderveldplots.MultiplesamplesweretakenfromeachlayerwithintheflowstonespeleothemandanalysedusingFisherstatisticsandthePro-gramFISH2toestablishthePalaeolatitude(Plat.)foreachlayer.Thiswasthenusedtoassignthepolarityofeachlayer,withPlat.Values>+60o/ 60oareconsideredtorepresentnormalorreversed polarity.Radiocarbondating Methodologyforradiocarbondating Radiocarbondatingisacommonlyusedmethodfordatingmaterialsthatcontaincarbon,byusingthedecayoftheradioactivecarbonisotope14Cwithahalf-lifeof ~ 5730years.Becauseoftherelativelyshorthalf-lifeof14C,thistechniquegenerallyonlyreturnsreliableresultsforageslessthan50 ka,anditiswidelyusedinArchaeologicalapplications(e.g., TaylorandBar-Yosef,2014 ). Samplepreparation ± TwobonefragmentscollectedalongtheflooroftheDinalediChamber nearthetopofsub-unit3bweresenttoBetaAnalyticInc.inMiami,Florida.Here,theoutersurfacesofthebonefragmentswereacidetchedwith10%HCl(atroomtemperature)afterwhichsampleswererinsedtoaneutralpHanddried.Sampleswerethengroundtoapowderandpre-treatedwith1Naceticacidfor24hr,rinsed,driedandweighed. Thepre-treatedbonepowderwasacidifiedin85%phosphoricacidat70 Ê Cinaclosedchemistry linethathadbeenpurgedofanyCO2(to<10 15atoms).CO2producedfromthesamplewasintroducedintoareactionvesselcontaininganaliquotofcobaltmetalcatalyst.Hydrogenwasintroducedandthecocktailheatedto500 Ê C,tocrackCO2andformcarbon(graphite).Thegraphitewas pressedintoapelletforanalysis. Analysesanddataprocessing ± Analysesforradiocarbondatingwereperformedusingacceleratormassspectrometry(AMS).AnalyseswerecalibratedwithgraphiteproducedfromtheNIST-4990Cmodernreferencestandard.Reported d13CwasmeasuredrelativetothePDB-1onthesampleitself.TotalfractionationusingtheAMS d13Ccorrectionwasdonetoderiveata`conventional radiocarbonage:'AgesarecalculatedusingBetaCal3.17providedbyBetaAnalyticfollowingproce-duresoutlinedin BronkRamsey(2009) andusingtheSHCAL13database( Hoggetal.,2013 ). Datesarereportedasradiocarbonyearsbeforepresent(`BP'with`present'takenasAD1950). Dirks etal .eLife2017;6:e24231. DOI:10.7554/eLife.24231 51of59 Researcharticle GenomicsandEvolutionaryBiology

PAGE 52

Figure13. SamplesandresultsofpalaeomagneticanalysesforFlowstone1a.( a )Outcropphotoofhanging erosionremnantofFlowstone1afromwhichthepalaeomagneticsamplewastaken.Thethreeflowstonephasesseparatedbydetritalhorizonsareclearlyvisible,andtheirmagneticpolarityhasbeenmarked(N=normal;R=reverse).Thestratigraphictopistowardsthetopofthephoto;( b )close-upofahandsampletakenfor palaeomagneticanalysisfromFlowstone1aintheDinalediChamber.Thesampleislayeredandcomprisesthreedistinctphases(frombasetotop:A-Cmarkedinyellow)separatedbythinclastichorizonsthatmarkdisconformitiesindicatedwithreddashedlines.Thelarger-scaleextentofthethreephasescanbeseenin( a );( c ) intensityspectra,Zijderveldplots,andstereoplotsforsamplesfromphasesAtoCtakenfrom( b ).PhasesBandC shownormalpolarityandphaseAshowsreversedandintermediatepolaritydirections. DOI:10.7554/eLife.24231.025 Dirks etal .eLife2017;6:e24231. DOI:10.7554/eLife.24231 52of59 Researcharticle GenomicsandEvolutionaryBiology

PAGE 53

Acknowledgements WewouldliketothankChrisStringerandfouranonymousreviewersfortheirconstructivecommentsthathavegreatlyhelpedimprovethispaper.Wewouldalsoliketothankthemanyfundingagenciesthatsupportedvariousaspectsofthiswork.InparticularwewouldliketothanktheNationalGeo-graphicSociety,theNationalResearchFoundationandtheLydaHillFoundationforsignificantfund-ingofthediscovery,recoveryandinitialanalysisofthismaterial.FurthersupportwasprovidedbyARC(DP140104282:PHGMD,ER,JK,HHW;FT120100399:AH).TheESRdosimetrystudyunder-takenbyCENIEHandGriffithUniversityhasbeensupportedbyaMarieCurieInternationalOutgo-ingFellowship(underREAGrantAgreementn Ê PIOF-GA-2013±626474)oftheEuropeanUnion's SeventhFrameworkProgramme(FP7/2007-2013)andanAustralianResearchCouncilFutureFellow-ship(FT150100215).ESRandU-seriesdatingundertakenatSCUweresupportedbyARC(DP140100919:RJB).WewouldalsoliketothanktheUniversityoftheWitwatersrand,theEvolution-aryStudiesInstituteandtheSouthAfricanNationalCentreofExcellenceinPalaeoSciencesforhost-ingmanyoftheauthorswhilestudyingthematerial,andallowingoriginalmaterialtobemadeavailablefordating.WewouldliketothanktheSouthAfricanHeritageResourceAgencyforthenec-essarypermitstoworkontheRisingStarsite;theJacobsfamilyandlatertheLeeBergerFoundationforgrantingaccess.TheassistanceofmembersoftheSpeleologicalExplorationClub,invarious 050100150200250300350400450500550 600650700750 800 850 050100150200250300350400450500550 600650700750 800 850 Age (ka)Age (ka) 900 950 1000 900 950 1000 Estimated Deposition of Unit 2 in debris cone Estimated age of sub-unit 1a on the chamber floor Estimated deposition of Flowstone Group 2 Estimated age of Homo naledi Estimated formation of Flowstone Group 3 Estimated Continuous Formation of Unit 1 Estimated deposition of fossilbearing sub-unit 3b ESR-1810 age model ESR-1788 age model Youngest possible age of H. naledibearing sub-unit 3b based on superposition of dated flowstones on cave floor Estimated deposition of Flowstone 1c Normal Reverse Normal OSL Age (MAM: minimum age model) ESR Age (maximum age model = preferred) U-Series Age ( in situ flowstone) Error (2!)* lack of error bars indicates error contained within symbol width. LEGEND Papio tooth reworked into sub-unit 3a Estimated deposition of sub-unit 3a Paleomagnetic reversal in Flowstone 1a = youngest possible age of Unit 2 Sub-unit 1bOSL-3Sub-unit 1bOSL-4 Sample Description Sub-unit 1aOSL-5 Flowstone 1a (top): RS22 Flowstone 1c: RS18 Flowstone Group 2 (top): RS16 Flowstone Group 2 (base): RS17 Flowstone Group 2 (base): RS21 Flowstone Group 3 (top): RS11 Flowstone Group 2: RS15 Flowstone 1a (mid): RS23 Paleomag samples: Flowstone Group 1 H. naledi tooth-UW1810: sub-unit 3b H. naledi tooth-UW1788: sub-unit 3b Papio tooth-UW1841: sub-unit 3a Flowstone Group 2: RS10 Flowstone Group 2: RS8 Flowstone Group 2: RS1 Flowstone Group 2: RS14 Flowstone Group 2: RS6 Flowstone Group 2: RS20Flowstone Group 2: RS19 Estimated formation of Flowstone 1a covering Unit 2 debris cone age mo age mo ESR Age (minimum age model) Oldest possible age of H. naledi bearing sub-unit 3b Estimated age of sub-unit 1b on the chamber floor H. naledi bones covered by Flowstone 1c (provides interpreted minimum age for fossils) Sub-unit 1a and sub-unit 1b eroded to form sub-unit 3b Flowstone Group 2: RS13 Hypothesized older, undifferentiated portion of Unit 1 eroded from locations on fractures and on shelves above entry to form Unit 2 ? ? Figure14. Chronostratigraphicsummaryofradio-isotopicdatingresults,andinterpretationofthedepositional rangesofstratigraphicunits,flowstonesand H.naledi fossilsintheDinalediChamber.Followingthepreferred US-ESRmaximumagemodelandassociateduncertaintiesforESRsamples1788and1810,amaximumdepositionalageof335Mawasdetermined,whiletheminimumdepositionalageof236kawasconstrainedbyFlowstone1c(sampleRS18),whichcovers H.naledi materialintheentryzone. DOI:10.7554/eLife.24231.026 Dirks etal .eLife2017;6:e24231. DOI:10.7554/eLife.24231 53of59 Researcharticle GenomicsandEvolutionaryBiology

PAGE 54

safetyaspectswithinthecaveduringexcavationsisgratefullyacknowledged.ZenobiaJacobsandBertRobertsoftheUniversityofWollongongarethankedforhelpingwithinitialOSLtestsforsinglegrainanalysesandforhelpfuldiscussions.WewouldalsoliketothankWilmaLawrence,BonitaDeKlerk,NatashaBarbolini,MerrillvanderWalt,WayneCrichtonandJustinMukankufortheirassis-tanceduringallphasesoftheproject.RGandMDthankLesKinsley,ANU,forhisinvaluablehelptokeeptheNeptuneintune.MEwouldliketothankStephanWoodborne(iThembaLABS,Gauteng)forhishelpinunravellingthespectrometerresultsrelatedtoOSLandESRdating. Additionalinformation Funding FunderGrantreferencenumberAuthorAustralianResearchCouncilDP140104282PaulHGMDirks EricMRobertsHannahHilbert-WolfJanDKramersCarlSpandlerLeeRBerger AustralianResearchCouncilFT120100399AndyIRHerriesAustralianResearchCouncilDP140100919RenaudJoannes-BoyauMarieCurieInternationalOut-goingFellowship PIOF-GA-2013-626474MathieuDuval AustralianResearchCouncilFT150100215MatthieuDuvalNationalGeographicSocietyLeeRBergerNationalResearchFoundationLeeRBergerLydaHillFoundation LeeRBerger Thefundershadnoroleinstudydesign,datacollectionandinterpretation,orthedecisiontosubmittheworkforpublication. Authorcontributions PHGMD,Conceptualization,Formalanalysis,Supervision,Fundingacquisition,Validation,Investiga-tion,Visualization,Methodology,WritingÐoriginaldraft,Projectadministration,WritingÐreviewandediting;EMR,Conceptualization,Formalanalysis,Supervision,Fundingacquisition,Investiga-tion,Visualization,Methodology,Projectadministration,WritingÐreviewandediting;HH-W,Formalanalysis,Validation,Investigation,WritingÐreviewandediting,Visualisation(U-Th;geochemistry;samplepreparation);JDK,Resources,Formalanalysis,Validation,Investigation,Writing(U-Th,OSL,geochemistry);JHa,Conceptualization,Visualization,WritingÐreviewandediting;AD,JHe,CJP,Formalanalysis,Investigation(U-Th);MD,Resources,Formalanalysis,Validation,Investigation,Writ-ingÐreviewandediting,Visualisation(ESR);MEl,Supervision,Investigation,Visualisation(archaeol-ogy,speleology);MEv,Formalanalysis,Investigation(OSL);RG,Formalanalysis,Validation,Investigation,Resources(ESR);AIRH,Formalanalysis,Investigation,Writing(Palaeomagnetism);RJ-B,Formalanalysis,Validation,Investigation,WritingÐreviewandediting,Visualisation(ESR);TVM,Formalanalysis,Investigation(geochemistry);JR,JWi,Investigation(fieldsedimentology);CS,Vali-dation,WritingÐreviewandediting;JWo,Formalanalysis,Investigation,Resources(U-Th);LRB,Conceptualization,Resources,Fundingacquisition,Projectadministration,WritingÐreviewandediting AuthorORCIDs PaulHGMDirks, http://orcid.org/0000-0002-1582-1405 RenaudJoannes-Boyau, http://orcid.org/0000-0002-0452-486X LeeRBerger, http://orcid.org/0000-0002-0367-7629 ReferencesAdamiecG ,AitkenMJ.1998.Doserateconversionfactors:update. Ancient-TL 16 :37±49. Dirks etal .eLife2017;6:e24231. DOI:10.7554/eLife.24231 54of59 Researcharticle GenomicsandEvolutionaryBiology

PAGE 55

AitkenMJ .1985. ThermoluminescenceDating .London:AcademicPress.p.359 BarboutiAI ,RastinBC.1983.Astudyoftheabsoluteintensityofmuonsatsealevelandundervarious thicknessesofabsorber. JournalofPhysicsG:NuclearPhysics 9 :1577±1595. doi:10.1088/0305-4616/9/12/018 BergerLR ,deRuiterDJ,ChurchillSE,SchmidP,CarlsonKJ,DirksPH,KibiiJM.2010. Australopithecussediba :a newspeciesofHomo-likeaustralopithfromSouthAfrica. Science 328 :195±204. doi:10.1126/science.1184944 , PMID:20378811 BergerLR ,HawksJ,deRuiterDJ,ChurchillSE,SchmidP,DelezeneLK,KivellTL,GarvinHM,WilliamsSA, DeSilvaJM,SkinnerMM,MusibaCM,CameronN,HollidayTW,Harcourt-SmithW,AckermannRR,BastirM,BoginB,BolterD,BrophyJ,etal.2015. Homonaledi ,anewspeciesofthegenus Homo fromtheDinaledi Chamber,SouthAfrica. eLife 4 :e09560. doi:10.7554/eLife.09560 BergerLR ,HawksJ,DirksP,ElliotM,RobertsE.2017. Homonaledi andpleistocenehomininevolutionin subequatorialAfrica. eLife 6 :e24234. doi:10.7554/eLife.24234 BergerLR ,LacruzR,DeRuiterDJ.2002.RevisedageestimatesofAustralopithecus-bearingdepositsat Sterkfontein,SouthAfrica. AmericanJournalofPhysicalAnthropology 119 :192±197. doi:10.1002/ajpa.10156 , PMID:12237940 BrainCK .1993. Swartkrans:ACave'sChronicleofEarlyMan .Pretoria:TransvaalMuseum. BronkRamseyC .2009.BayesiananalysisofRadiocarbondates. Radiocarbon 51 :337±360. doi:10.1017/ S0033822200033865 BruxellesL ,ClarkeRJ,MaireR,OrtegaR,StratfordD.2014.StratigraphicanalysisoftheSterkfonteinStW573 Australopithecusskeletonandimplicationsforitsage. JournalofHumanEvolution 70 :8±48. doi:10.1016/j. jhevol.2014.02.014 , PMID:24698198 ChengH ,EdwardsRL,HoffJ,GallupCD,RichardsDA,AsmeromY.2000.Thehalf-livesofuranium-234and thorium-230. ChemicalGeology 169 :17±33. doi:10.1016/S0009-2541(99)00157-6 ChengH ,LawrenceEdwardsR,ShenC-C,PolyakVJ,AsmeromY,WoodheadJD,HellstromJ,WangY,KongX, Spo È tlC,WangX,CalvinAlexanderE.2013.Improvementsin230thdating,230thand234Uhalf-lifevalues,and U±Thisotopicmeasurementsbymulti-collectorinductivelycoupledplasmamassspectrometry. Earthand PlanetaryScienceLetters 371±372 :82±91. doi:10.1016/j.epsl.2013.04.006 deRuiterDJ ,PickeringR,SteiningerCM,KramersJD,HancoxPJ,ChurchillSE,BergerLR,BackwellL.2009. New Australopithecusrobustus fossilsandassociatedU-PbdatesfromCooper'sCave(Gauteng,SouthAfrica). JournalofHumanEvolution 56 :497±513. doi:10.1016/j.jhevol.2009.01.009 , PMID:19443017 DemboM ,RadovcÏicÂD,GarvinHM,LairdMF,SchroederL,ScottJE,BrophyJ,AckermannRR,MusibaCM,de RuiterDJ,MooersAé,CollardM.2016.Theevolutionaryrelationshipsandageof Homonaledi :anassessment usingdatedbayesianphylogeneticmethods. JournalofHumanEvolution 97 :17±26. doi:10.1016/j.jhevol.2016. 04.008 , PMID:27457542 DirksP ,BergerLR,RobertsEM,KramersJD,HawksJ,Randolph-QuinneyPS,ElliottM,MusibaCM,ChurchillSE, deRuiterDJ,SchmidP,BackwellLR,BelyaninGA,BoshofP,EavesKL,FeuerriegelEM,GurtovA,HarrisonJ,duG,HunterR,etal.2015.Geologicalandtaphonomicevidencefordeliberatebodydisposalbytheprimitivehomininspecies Homonaledi fromtheDinalediChamber,SouthAfrica. eLife 4 :e09561. doi:10.7554/eLife. 09561 DirksP ,PlaczekC,FinkD,DossetoA,RobertsEM.2016b.Using10bedatingtoreconstructtheevolving landscapeintheCradleofHumankind,SouthAfrica. JournalofHumanEvolution 96 :19±34. doi:10.1016/j. jhevol.2016.03.002 DirksPH ,KibiiJM,KuhnBF,SteiningerC,ChurchillSE,KramersJD,PickeringR,FarberDL,MeÂriauxAS,Herries AI,KingGC,BergerLR.2010.Geologicalsettingandageof Australopithecussediba fromsouthernAfrica. Science 328 :205±208. doi:10.1126/science.1184950 , PMID:20378812 DirksP ,BergerLR,HawksJ,Randolph-QuinneyPS,BackwellLR,RobertsEM.2016a.Deliberatebodydisposal byhomininsintheDinalediChamber-areply. JournalofHumanEvolution 96 :145±148. DirksPHGM ,BergerLR.2013.Hominin-bearingcavesandlandscapedynamicsintheCradleofHumankind, SouthAfrica. JournalofAfricanEarthSciences 78 :109±131. doi:10.1016/j.jafrearsci.2012.09.012 DrysdaleRN ,PaulBT,HellstromJC,CouchoudI,GreigA,BajoP,ZanchettaG,IsolaI,Spo È tlC,BaneschiI, RegattieriE,WoodheadJD.2012.PrecisemicrosamplingofpoorlylaminatedspeleothemsforU-seriesdating.QuaternaryGeochronology 14 :38±47. doi:10.1016/j.quageo.2012.06.009 DullerGAT .2007.Assessingtheerroronequivalentdoseestimatesderivedfromsinglealiquotregenerative dosemeasurements. Ancient-TL 25 :15±24. DullerGAT .2008.Single-grainopticaldatingofQuaternarysediments:whyaliquotsizemattersinluminescence dating. Boreas 37 :589±612. doi:10.1111/j.1502-3885.2008.00051.x DuvalM ,FalgueÁresC,BahainJ-J,Gru È nR,ShaoQ,AubertM,DoloJ-M,Agustõ Â J,Martõ Ânez-Nav arroB,Palmqvist P,Toro-MoyanoI.2012.OnthelimitsofusingcombinedU-series/ESRmethodtodatefossilteethfromtwoEarlyPleistocenearchaeologicalsitesoftheOrcearea(Guadix-Bazabasin,Spain). QuaternaryResearch 77 : 482±491. doi:10.1016/j.yqres.2012.01.003 DuvalM ,Gru È nR.2016.ArepublishedESRdoseassessmentsonfossiltoothenamelreliable? Quaternary Geochronology 31 :19±27. doi:10.1016/j.quageo.2015.09.007 EgginsS ,Gru È nR,PikeAWG,ShelleyM,TaylorL.2003.238U,232ThprofilingandU-seriesisotopeanalysisof fossilteethbylaserablation-ICPMS. QuaternaryScienceReviews 22 :1373±1382. doi:10.1016/S0277-3791(03) 00064-7 Dirks etal .eLife2017;6:e24231. DOI:10.7554/eLife.24231 55of59 Researcharticle GenomicsandEvolutionaryBiology

PAGE 56

EgginsSM ,Gru È nR,McCullochMT,PikeAWG,ChappellJ,KinsleyL,MortimerG,ShelleyM,Murray-WallaceCV, Spo È tlC,TaylorL.2005.InsituU-seriesdatingbylaser-ablationmulti-collectorICPMS:newprospectsfor quaternarygeochronology. QuaternaryScienceReviews 24 :2523±2538. doi:10.1016/j.quascirev.2005.07.006 GalbraithRF ,LaslettGM.1993.Statisticalmodelsformixedfissiontrackages. NuclearTracksandRadiation Measurements 21 :459±470. doi:10.1016/1359-0189(93)90185-C GalbraithRF ,RobertsRG,LaslettGM,YoshidaH,OlleyJM.1999.Opticaldatingofsingleandmultiplegrainsof quartzfromjinmiumrockshelter,northernAustralia:parti,experimentaldesignandstatisticalmodels.Archaeometry 41 :339±364. doi:10.1111/j.1475-4754.1999.tb00987.x GalbraithRF ,RobertsRG,YoshidaH.2005.ErrorvariationinOSLpalaeodoseestimatesfromsinglealiquotsof quartz:afactorialexperiment. RadiationMeasurements 39 :289±307. doi:10.1016/j.radmeas.2004.03.023 GalbraithRF .2002.Anoteonthevarianceofabackground-correctedOSLcount. Ancient-TL 20 :49±51. GoldmannA ,BrenneckaG,NoordmannJ,WeyerS,WadhwaM.2015.Theuraniumisotopiccompositionofthe EarthandtheSolarSystem. GeochimicaEtCosmochimicaActa 148 :145±158. doi:10.1016/j.gca.2014.09.008 GrangerDE ,GibbonRJ,KumanK,ClarkeRJ,BruxellesL,CaffeeMW.2015.Newcosmogenicburialagesfor SterkfonteinMember2 Australopithecus andMember5oldowan. Nature 522 :85±88. doi:10.1038/ nature14268 , PMID:25830884 Gru È nR ,AubertM,Joannes-BoyauR,MoncelM-H.2008a.Highresolutionanalysisofuraniumandthorium concentrationaswellasU-seriesisotopedistributionsinaneanderthaltoothfromPayre(ArdeÁche,France)usinglaserablationICP-MS. GeochimicaEtCosmochimicaActa 72 :5278±5290. doi:10.1016/j.gca.2008.08.007 Gru È nR ,BrumbyS.1994a.TheassessmentoferrorsinpastradiationdosesextrapolatedfromESR/TLdoseresponsedata. RadiationMeasurements 23 :307±315. doi:10.1016/1350-4487(94)90057-4 Gru È nR ,EgginsS,KinsleyL,MoseleyH,SambridgeM.2014.LaserablationU-seriesanalysisoffossilbonesand teeth. Palaeogeography,Palaeoclimatology,Palaeoecology 416 :150±167. doi:10.1016/j.palaeo.2014.07.023 Gru È nR ,Joannes-BoyauR,StringerC.2008b.TwotypesofCO2 radicalsthreatenthefundamentalsofESR datingoftoothenamel. QuaternaryGeochronology 3 :150±172. doi:10.1016/j.quageo.2007.09.004 Gru È nR ,Katzenberger-ApelO.1994b.AnalphairradiatorforESRdating. AncientTL 12 :35±38. Gru È nR ,RhodesEJ.1992.SimulationsofsaturatingexponentialESR/TLdoseresponsecurves-weightingof intensityvaluesbyinversevariance. AncientTL 10 :50±56. Gru È nR ,SchwarczHP,ChadamJ.1988.ESRdatingoftoothenamel:coupledcorrectionforU-uptakeand U-seriesdisequilibrium. InternationalJournalofRadiationApplicationsandInstrumentation.PartD.Nuclear TracksandRadiationMeasurements 14 :237±241. doi:10.1016/1359-0189(88)90071-4 Gru È nR .1987.Alphadoseattenuationinthinlayers. AncientTL 5 :6±8. Gru È nR .1989.Electronspinresonance(ESR)dating. QuaternaryInternational 1 :65±109. doi:10.1016/1040-6182 (89)90010-4 Gru È nR .1997.Electronspinresonancedating.In:TaylorR.E,AitkenM.J(Eds). ChronometricDatingin Archaeology .NewYork:PlenumPress.p.217±260. Gru È nR .2000a.MethodsofdosedeterminationusingESRspectraoftoothenamel. RadiationMeasurements 32 : 767±772. doi:10.1016/S1350-4487(99)00281-4 Gru È nR .2000b.AnalternativeformodelforopensystemU-series/ESRagecalculations:(closedsystemU-series)ESR,CSUS-ESR. AncientTL 18 :1±4. Gru È nR .2009.TherelevanceofparametricU-uptakemodelsinESRagecalculations. RadiationMeasurements 44 :472±476. doi:10.1016/j.radmeas.2009.02.011 Gue  rinG ,MercierN,AdamiecG.2011.Doserateconversionfactors:update. AncientTL 29 :5±8. HawksJ ,BergerLR.2016.Theimpactofadateforunderstandingtheimportanceof Homonaledi . Transactions oftheRoyalSocietyofSouthAfrica 71 :125±128. doi:10.1080/0035919X.2016.1178186 HawksJ ,ElliottM,SchmidP,ChurchillSE,deRuiterDJ,RobertsEM,Hilbert-wolfH,GarvinHM,WilliamsSA, DelezeneLK,FeuerriegelEM,Randolph-QuinneyP,KivellTL,LairdMF,TawaneG,DeSilvaJM,BaileySE,BrophyJK,MeyerMR,SkinnerMM,etal.2017.Newfossilremainsof Homonaledi fromtheLesediChamber, SouthAfrica. eLife 6 :e24232. doi:10.7554/eLife.24232 HellstromJ ,PickeringR.2015.RecentadvancesandfutureprospectsoftheU±ThandU±Pbchronometers applicabletoarchaeology. JournalofArchaeologicalScience 56 :32±40. doi:10.1016/j.jas.2015.02.032 HellstromJ .2003.RapidandaccurateU/Thdatingusingparallelion-countingmulti-collectorICP-MS. Journalof AnalyticalAtomicSpectrometry 18 :1346±1351. doi:10.1039/b308781f HellstromJ .2006.U±Thdatingofspeleothemswithhighinitial230Thusingstratigraphicalconstraint. QuaternaryGeochronology 1 :289±295. doi:10.1016/j.quageo.2007.01.004 HerriesAI ,AdamsJW,KuykendallKL,ShawJ.2006.Speleologyandmagnetobiostratigraphicchronologyofthe GD2localityoftheGondolinhominin-bearingpaleocavedeposits,NorthWestProvince,SouthAfrica. Journal ofHumanEvolution 51 :617±631. doi:10.1016/j.jhevol.2006.07.007 , PMID:16949648 HerriesAI ,AdamsJW.2013.Clarifyingthecontext,datingandagerangeoftheGondolinhomininsand Paranthropus inSouthAfrica. JournalofHumanEvolution 65 :676±681. doi:10.1016/j.jhevol.2013.06.007 , PMID:23911292 HerriesAI ,ShawJ.2011.PalaeomagneticanalysisoftheSterkfonteinpalaeocavedeposits:implicationsforthe ageofthehomininfossilsandstonetoolindustries. JournalofHumanEvolution 60 :523±539. doi:10.1016/j. jhevol.2010.09.001 , PMID:21392817 HerriesAIR ,CurnoeD,AdamsJW.2009.Amulti-disciplinaryseriationofearly Homo andParanthropusbearing palaeocavesinSouthernAfrica. QuaternaryInternational 202 :14±28. doi:10.1016/j.quaint.2008.05.017 Dirks etal .eLife2017;6:e24231. DOI:10.7554/eLife.24231 56of59 Researcharticle GenomicsandEvolutionaryBiology

PAGE 57

HerriesAIR ,KappenP,KegleyA,PatersonD,HowardDL,deJongeMD,PotzeS,AdamsJW.2014. Palaeomagneticandsynchrotronanalysisof>1.95MafossilbearingpalaeokarstatHaasgat,Gauteng. South AfricanJournalofScience 110 :1±12. doi:10.1590/sajs.2014/20130102 HerriesAIR ,PickeringR,AdamsJW,CurnoeD,WarrG,LathamAG,ShawJ.2013.Amulti-disciplinary perspectiveontheageof Australopithecus insouthernAfrica.In:ReedK.E,FleagleJ.G,LeakeyR(Eds). PaleobiologyofAustralopithecus.VertebratePalaeobiologyandPaleoanthropologySeries .p.21±40. HoggAG ,HuaQ,BlackwellPG,NiuM,BuckCE,GuildersonTP,HeatonTJ,PalmerJG,ReimerPJ,ReimerRW, TurneyCSM,ZimmermanSRH.2013.SHCal13SouthernHemisphereCalibration,0±50,000YearscalBP.Radiocarbon 55 :1889±1903. doi:10.2458/azu_js_rc.55.16783 HorwitzEP ,ChiariziaR,DietzML,DiamondH,NelsonDM.1993.Separationandpreconcentrationofactinides fromacidicmediabyextractionchromatography. AnalyticaChimicaActa 281 :361±372. doi:10.1016/00032670(93)85194-O HorwitzEP ,DietzML,ChiariziaR,DiamondH,EsslingAM,GraczykD.1992.Separationandpreconcentrationof uraniumfromacidicmediabyextractionchromatography. AnalyticaChimicaActa 266 :25±37. doi:10.1016/ 0003-2670(92)85276-C Joannes-BoyauR ,Gru È nR,BodinT.2010.Decompositionofthelaboratorygammairradiationcomponentof angularESRspectraoffossiltoothenamelfragments. AppliedRadiationandIsotopes 68 :1798±1808. doi:10. 1016/j.apradiso.2010.03.015 , PMID:20409724 Joannes-BoyauR ,Gru È nR.2011.AcomprehensivemodelforCO2 radicalsinfossiltoothenamel:implications forESRdating. QuaternaryGeochronology 6 :7:82±97. doi:10.1016/j.quageo.2010.09.001 Joannes-BoyauR .2013.Detailedprotocolforaccuratenon-destructivedirectdatingofhumanremains. Geochronometria 40 :322±333. doi:10.2478/s13386-013-0132-7 KramersJD ,DirksPHGM,DepartmentofGeosciences,CollegeofScienceandEngineering,JamesCook University,Townsville,Queensland,Australia.2017.TheageoffossilStW573(`LittleFoot'):Analternativeinterpretationof26Al/10Beburialdata. SouthAfricanJournalofScience 113 :1±8. doi:10.17159/sajs.2017/ 20160085 KronfeldJ ,VogelJC,TalmaAS.1994.Anewexplanationforextreme234u238udisequilibriainadolomitic aquifer. EarthandPlanetaryScienceLetters 123 :81±93. doi:10.1016/0012-821X(94)90259-3 KrugerA ,Randolph-QuinneyP,ElliottM,EvolutionaryStudiesInstitute,SchoolofGeosciences,Universityofthe Witwatersrand,Johannesburg,SouthAfrica.2016.MultimodalspatialmappingandvisualisationofDinalediChamberandRisingStarCave. SouthAfricanJournalofScience 112 :11. doi:10.17159/sajs.2016/20160032 LahrMM ,FoleyR.2001.Mode3, Homo helmei,andthepatternofhumanevolutionintheMiddlePleistocene. In: HumanRoots,AfricanandAsiaintheMiddlePleistocene .Bristol:WesternAcademic.p.23±39. LascuI ,FeinbergJM.2011.Speleothemmagnetism. QuaternaryScienceReviews 30 :3306±3320. doi:10.1016/j. quascirev.2011.08.004 LouX ,RehkamperM,LeeD-C,HallidayAN.1997.Highprecision230th/232Thand234U/238Umeasurements usingenergy-filteredICPmagneticsectormultiplecollectormassspectrometery. InternationalJournalofMass Spectrometry 171 :105±117. doi:10.1016/S0168-1176(97)00136-5 LudwigKR .2012.User'smanualforIsoplot3.75.BerkeleyGeochronologyCenter. SpecialPublication 5 :1±75. MadsenAT ,MurrayAS,AndersenTJ,PejrupM,Breuning-MadsenH.2005.Opticallystimulatedluminescence datingofyoungestuarinesediments:acomparisonwith210pband137csdating. MarineGeology 214 :251± 268. doi:10.1016/j.margeo.2004.10.034 MakhubelaTV ,KramersJD,BelyaninGA,DirksP,RobertsEM.2017.Proterozoic40ar/39Aragesfromcave depositsofthemalapa,sterkfontein1andDinaledifossilsites,CradleofHumankind,SouthAfrica. South AfricanJournalofGeology .Inpress. MartiniJ ,WipplingerP,MoenH,KeyserA.2003.ContributiontothespeleologyofSterkfonteinCave,Gauteng Province,SouthAfrica. InternationalJournalofSpeleology 32 :43±69. doi:10.5038/1827-806X.32.1.4 McbreartyS ,BrooksAS.2000.Therevolutionthatwasn't:anewinterpretationoftheoriginofmodernhuman behavior. JournalofHumanEvolution 39 :453±563. doi:10.1006/jhev.2000.0435 , PMID:11102266 MejdahlV .1979.Thermoluminescencedating:beta-doseattenuationinquartzgrains. Archaeometry 21 :61±72. doi:10.1111/j.1475-4754.1979.tb00241.x MetropolisN ,RosenbluthAW,RosenbluthMN,TellerAH,TellerE.1953.Equationofstatecalculationsbyfast computingmachines. TheJournalofChemicalPhysics 21 :1087±1092. doi:10.1063/1.1699114 MorwoodMJ ,SoejonoRP,RobertsRG,SutiknaT,TurneyCS,WestawayKE,RinkWJ,ZhaoJX,vandenBergh GD,DueRA,HobbsDR,MooreMW,BirdMI,FifieldLK.2004.ArchaeologyandageofanewhomininfromFloresineasternIndonesia. Nature 431 :1087±1091. doi:10.1038/nature02956 , PMID:15510146 MurrayAS ,WintleAG.2000.Luminescencedatingofquartzusinganimprovedsingle-aliquotregenerative-dose protocol. RadiationMeasurements 32 :57±73. doi:10.1016/S1350-4487(99)00253-X NaidooM .2016. GeophysicsoftheRisingStarCaveSystem .Honsthesis:UniversityoftheWitwatersrand.p.34 O'ReganHJ ,ReynoldsSC.2009.Anecologicalreassessmentofthesouthernafricancarnivoreguild:acasestudy fromMember4,sterkfontein,SouthAfrica. JournalofHumanEvolution 57 :212±222. doi:10.1016/j.jhevol.2009. 04.002 , PMID:19647856 OlleyJM ,PietschT,RobertsRG.2004.OpticaldatingofHolocenesedimentsfromavarietyofgeomorphic settingsusingsinglegrainsofquartz. Geomorphology 60 :337±358. doi:10.1016/j.geomorph.2003.09.020 PartridgeTC ,GrangerDE,CaffeeMW,ClarkeRJ.2003.LowerPliocenehominidremainsfromSterkfontein. Science 300 :607±612. doi:10.1126/science.1081651 , PMID:12714736 Dirks etal .eLife2017;6:e24231. DOI:10.7554/eLife.24231 57of59 Researcharticle GenomicsandEvolutionaryBiology

PAGE 58

PartridgeTC ,ShawJ,HeslopD,ClarkeRJ.1999.ThenewhominidskeletonfromSterkfontein,SouthAfrica:age andpreliminaryassessment. JournalofQuaternaryScience 14 :293±298. doi:10.1002/(SICI)1099-1417(199907) 14:4<293::AID-JQS471>3.0.CO;2-X PartridgeTC .1973.GeomorphologicaldatingofCaveOpeningsatMakapansgat,Sterkfontein,Swartkransand Taung. Nature 246 :75±79. doi:10.1038/246075a0 PickeringR ,DirksPH,JinnahZ,deRuiterDJ,ChurchilSE,HerriesAI,WoodheadJD,HellstromJC,BergerLR. 2011a. Australopithecussediba at1.977Maandimplicationsfortheoriginsofthegenus Homo . Science 333 : 1421±1423. doi:10.1126/science.1203697 , PMID:21903808 PickeringR ,HancoxPJ,Lee-ThorpJA,Gru È nR,MortimerGE,McCullochM,BergerLR.2007.Stratigraphy,U-Th chronology,andpaleoenvironmentsatGladysvaleCave:insightsintotheclimaticcontrolofSouthAfricanhominin-bearingcavedeposits. JournalofHumanEvolution 53 :602±619. doi:10.1016/j.jhevol.2007.02.005 , PMID:17920104 PickeringR ,JacobsZ,HerriesAIR,KarkanasP,Bar-MatthewsM,WoodheadJD,KappenP,FisherE,Marean CW.2013.Paleoanthropologicallysignificantsouthafricanseacavesdatedto1.1±1.0millionyearsusingacombinationofU±Pb,TT-OSLandpalaeomagnetism. QuaternaryScienceReviews 65 :39±52. doi:10.1016/j. quascirev.2012.12.016 PickeringR ,KramersJD,HancoxPJ,deRuiterDJ,WoodheadJD.2011b.Contemporaryflowstonedevelopment linksearlyhomininbearingcavedepositsinSouthAfrica. EarthandPlanetaryScienceLetters 306 :23±32. doi:10.1016/j.epsl.2011.03.019 PickeringR ,KramersJD,PartridgeT,KodolanyiJ,PettkeT.2010.U±Pbdatingofcalcite±aragonitelayersin speleothemsfromhomininsitesinSouthAfricabyMC-ICP-MS. QuaternaryGeochronology 5 :544±558. doi:10. 1016/j.quageo.2009.12.004 PickeringR ,KramersJD.2010.Re-appraisalofthestratigraphyanddeterminationofnewU-Pbdatesforthe sterkfonteinhomininsite,SouthAfrica. JournalofHumanEvolution 59 :70±86. doi:10.1016/j.jhevol.2010.03. 014 , PMID:20605190 PlaczekC ,PatchettPJ,QuadeJ,WagnerJDM.2006.StrategiesforsuccessfulU-Thdatingofpaleolake carbonates:anexamplefromtheBolivianAltiplano. Geochemistry,Geophysics,Geosystems 7 :n/a. doi:10. 1029/2005GC001157 PrescottJR ,HuttonJT.1994.CosmicraycontributionstodoseratesforluminescenceandESRdating:large depthsandlong-termtimevariations. RadiationMeasurements 23 :497±500. doi:10.1016/1350-4487(94)900868 ReynoldsSC ,KibiiJM.2011.Sterkfonteinat75:reviewofpalaeoenvironments,fauna,datingandarchaeology fromthehomininsiteofsterkfontein(GautengProvince,SouthAfrica). PalaeontologiaAfricana 46 :59±88. RobertsRG ,WestawayKE,ZhaoJX,TurneyCS,BirdMI,RinkWJ,FifieldLK.2009.Geochronologyofcave depositsatLiangBuaandofadjacentriverterracesintheWaeRacangvalley,westernFlores,Indonesia:asynthesisofageestimatesforthetypelocalityof Homofloresiensis . JournalofHumanEvolution 57 :484±502. doi:10.1016/j.jhevol.2009.01.003 , PMID:19254806 SambridgeM ,Gru È nR,EgginsS.2012.U-seriesdatingofboneinanopensystem:thediffusion-adsorptiondecaymodel. QuaternaryGeochronology 9 :42±53. doi:10.1016/j.quageo.2012.02.010 SasowskyID .1998.Determiningtheageofwhatisnotthere. Science 279 :1874. doi:10.1126/science.279.5358. 1874 ShaoQ ,BahainJ-J,DoloJ-M,FalgueÁresC.2014.MonteCarloapproachtocalculateUS-ESRageandage uncertaintyfortoothenamel. QuaternaryGeochronology 22 :99±106. doi:10.1016/j.quageo.2014.03.003 SingerBS .2014.Aquaternarygeomagneticinstabilitytimescale. QuaternaryGeochronology 21 :29±52. doi:10. 1016/j.quageo.2013.10.003 StockGM ,GrangerDE,SasowskyID,AndersonRS,FinkelRC.2005.ComparisonofU±Th,paleomagnetism,and cosmogenicburialmethodsfordatingcaves:Implicationsforlandscapeevolutionstudies. EarthandPlanetary ScienceLetters 236 :388±403. doi:10.1016/j.epsl.2005.04.024 StratfordD ,GrabS,PickeringTR.2014.Thestratigraphyandformationhistoryoffossil-andartefact-bearing sedimentsintheMilnerHall,SterkfonteinCave,SouthAfrica:newinterpretationsandimplicationsforpalaeoanthropologyandarchaeology. JournalofAfricanEarthSciences 96 :155±167. doi:10.1016/j.jafrearsci. 2014.04.002 StringerC .2002.Modernhumanorigins:progressandprospects. PhilosophicalTransactionsoftheRoyalSociety B:BiologicalSciences 357 :563±579. doi:10.1098/rstb.2001.1057 , PMID:12028792 SutiknaT ,TocheriMW,MorwoodMJ,SaptomoEW,Jatmiko,AweRD,WasistoS,WestawayKE,AubertM,LiB, ZhaoJX,StoreyM,AllowayBV,MeijerHJ,MeijerHJ,vandenBerghGD,Gru È nR,DossetoA,JungersWL, JungersWL,etal.2016.RevisedstratigraphyandchronologyforHomofloresiensisatLiangBuainIndonesia.Nature 532 :366±369. doi:10.1038/nature17179 , PMID:27027286 TaylorRE ,Bar-YosefO.2014. RadiocarbonDating .2ndedn.WalnutCreek,California:LeftCoastPress. ThackerayFJ .2016.EstimatingtheageandaffinitiesofHomonaledi. SouthAfricanJournalofScience 111 :1±2. doi:10.17159/sajs.2015/a0124 TobiasPV .2000.Thefossilhominids.In:PartridgeT.C,MaudR.R(Eds). TheCenozoicofSouthernAfrica, OxfordMonographsonGeologyandGeophysics .Oxford:OxfordUniversityPress.p.252±276. VrbaES .1975.SomeevidenceofchronologyandpalaeoecologyofSterkfontein,SwartkransandKromdraaifrom thefossilBovidae. Nature 254 :301±304. doi:10.1038/254301a0 Dirks etal .eLife2017;6:e24231. DOI:10.7554/eLife.24231 58of59 Researcharticle GenomicsandEvolutionaryBiology

PAGE 59

VrbaES .1995.ThefossilrecordofAfricanantelopes(Mammalia,Bovidae)inrelationtohumanevolutionand paleoclimate.In:VrbaE,DentonG,BurckleL,PartridgeT(Eds). PaleoclimateandEvolutionwithEmphasison HumanOrigins .NewHaven:YaleUniversityPress.p.262±288. WalkerJ ,CliffRA,LathamAG.2006.U-PbisotopicageoftheStW573hominidfromSterkfontein,SouthAfrica. Science 314 :1592±1594. doi:10.1126/science.1132916 , PMID:17158326 WilkinsonMJ .1985.Lower-lyingandpossiblyolderfossiliferousdepositsatSterkfontein.In:TobiasP.V(Ed). HominidEvolution:Past,PresentandFuture .p.165±170. WoodroffeCD ,ShortSA,StoddartDR,SpencerT,HarmonRS.1991.StratigraphyandChronologyofLate PleistoceneReefsintheSouthernCookIslands,SouthPacific. QuaternaryResearch 35 :246±263. doi:10.1016/ 0033-5894(91)90071-C ZimmermanDW .1971.Thermoluminescentdatingusingfinegrainsfrompottery. Archaeometry 13 :29±52. doi:10.1111/j.1475-4754.1971.tb00028.x Dirks etal .eLife2017;6:e24231. DOI:10.7554/eLife.24231 59of59 Researcharticle GenomicsandEvolutionaryBiology


printinsert_linkshareget_appmore_horiz

Download Options

close


  • info Info

    There are only PDFs associated with this resource.

  • link PDF(s)



Cite this item close

APA

Cras ut cursus ante, a fringilla nunc. Mauris lorem nunc, cursus sit amet enim ac, vehicula vestibulum mi. Mauris viverra nisl vel enim faucibus porta. Praesent sit amet ornare diam, non finibus nulla.

MLA

Cras efficitur magna et sapien varius, luctus ullamcorper dolor convallis. Orci varius natoque penatibus et magnis dis parturient montes, nascetur ridiculus mus. Fusce sit amet justo ut erat laoreet congue sed a ante.

CHICAGO

Phasellus ornare in augue eu imperdiet. Donec malesuada sapien ante, at vehicula orci tempor molestie. Proin vitae urna elit. Pellentesque vitae nisi et diam euismod malesuada aliquet non erat.

WIKIPEDIA

Nunc fringilla dolor ut dictum placerat. Proin ac neque rutrum, consectetur ligula id, laoreet ligula. Nulla lorem massa, consectetur vitae consequat in, lobortis at dolor. Nunc sed leo odio.