1

Publishedas:Prave,A.R.etal.GeologyandgeochronologyoftheTanaBasin,Ethiopia:LIP1 volcanism,supereruptionsandEocene-Oligoceneenvironmentalchange.EarthPlanet.Sci.2 Lett.(2016)http://10.1016/j.epsl.2016.03.0093 4 Geology and geochronology of the Tana Basin, Ethiopia: LIP volcanism, super5 eruptionsandEocene-Oligoceneenvironmentalchange6 7 A.R.Prave1*,C.R.Bates1,C.H.Donaldson1,H.Toland2,D.J.Condon3,D.Mark4,T.D.Raub18 1DepartmentEarthandEnvironmentalSciences,UniversityofStAndrews,KY169AL,UK9 2DepartmentGeographyandEarthSciences,AberystwythUniversity,Wales,SY233DB10 3NERCIsotopeGeosciencesLaboratory,BGS,Keyworth,NG125GG,UK11 4Isotope Geoscience Unit, Scottish Universities Environmental Research Centre, East12 Kilbride,G750QF,UK13 *correspondingauthor:emailap13@st-andrews.ac.uk14 15 Abstract16

NewgeologicalandgeochronologicaldatadefinefourepisodesofvolcanismfortheLake17

Tana region in the northern Ethiopian portion of the Afro-Arabian Large Igneous18

Province (LIP): pre-31 Ma flood basalt that yielded a single 40Ar/39Ar age of 34.0519

±0.54/0.56Ma;thickandextensivefelsicignimbritesandrhyolites(minimumvolumeof20

2-3x103km3)eruptedbetween31.108±0.020/0.041Maand30.844±0.027/0.046Ma21

(U-PbCA-ID-TIMSzirconages);maficvolcanismbracketedby 40Ar/39Aragesof28.9022

±0.12/0.14 Ma and 23.75 ±0.02/0.04 Ma; and localised scoraceous basalt with an23 40Ar/39Arageof0.033±0.005/0.005Ma.Thefelsicvolcanismwastheproductofsuper24

eruptionsthatcreateda60-80kmdiametercalderamarkedbykm-scalecaldera-collapse25

faultblocksandasteep-sidedbasinfilledwithaminimumof180mofsedimentandthe26

present-dayLakeTana.Thesenewdata enablemapping,with a finer resolution than27

previously possible, Afro-Arabian LIP volcanism onto the timeline of the Eocene-28

Oligocene transition and show that neither the mafic nor silicic volcanism coincides29

directlywith perturbations in the geochemical records that span that transition. Our30

results reinforce theview that it isnot thedevelopmentof aLIPalonebut its rateof31

effusionthatcontributestoinducingglobal-scaleenvironmentalchange.32

33 Keywords:LIP,Eocene-Oligocenetransition,LakeTana,floodbasalt,supereruption34 35 1.Introduction36

Temporal coincidence between Large Igneous Provinces (LIPs) and worldwide37

environmental perturbation is often considered evidence for causality (e.g. Bond &38

Wignall2014;Burgessetal.2014),yettheAfro-ArabianLIP,estimatedas0.6–1.1Mkm239

2

and0.35–1.0x106km3(MohrandZanettin1988;Dessertetal.2003),seeminglyhad40

littleinfluenceonCenozoicclimate(Hofmannetal.1997;Rochetteetal.1998;Ukstins41

Peate et al. 2003). Here we present new geological and geochronological data that42

provide a refined understanding of the development of that LIP and its temporal43

relationshiptotheclimaticeventsoftheEocene-Oligocenetransition.44

45

1.1.Geologicalbackground.TheEthiopianHighlandsareavolcanicmassifoffloodand46

shieldvolcanobasalt0.5to3kmthickthatformspectaculartraptopography(1500to47

4500maltitudes)flankingtheMainEthiopianRift(Fig.1;Mohr1983).Volcanicactivity48

wasprotractedbutepisodic:floodbasaltat31-29Ma,shieldvolcanoesat30-19and12-49

10Ma, felsic volcanismat30-25, 20-15and12-3Ma, andPliocene-Quaternarybasalt50

(Hofmannetal.1997;Piketal.1998;Rochetteetal.1998;Ayalewetal.2002;Ukstinset51

al.2002;Couliéetal.2003;Riisageretal.2005).Inthemidstofthemassifisa16,500km252

topographic depression that contains Lake Tana, the source of the Blue Nile and53

Ethiopia’slargestlakewithadiameterof60-80km.ThedominantrocktypeintheTana54

regionislow-TitholeiiticMiocene-Pliocenebasaltandlesseramountsoffelsitesandnon-55

marinesedimentaryrocksandlocallyrestrictedbasaltcinderconesandflows(Piketal.56

1998;Abate et al. 1998;Ayalewet al. 2002). Ourpetrological findings echo thoseof57

previousworkersandshowthatcompositionsarebimodal:sub-tomildlyalkaline(MgO58

5.0-9.5 wt%) olivine- to plagioclase-phyric basalt, and dacites to rhyolites including59

banded finely crystalline to glassy rocks, crystal ± lithic ± vitric tuffs, obsidian-rich60

agglomeratesandcoarseignimbrites.61

62

2.Results:newgeologicaldata63

Ourmappingdefinedfivemajorunits(Figs.2and3).Floodbasalt,withabaseat~70064

melevationanditstopat~1600m.Anoverlyingunit,manytenstoseveralhundredsof65

metresthick,offlow-bandedrhyolite,pumice/lithic-richand-poortuffs,coarse-grained66

silicicignimbrites,volcanicbrecciasandepiclasticsandstoneandagglomerate(Fig.4);67

clastsizesrangefromlapillitobombsseveralmetresindiameter.Thefelsicrocksoccur68

in fault blocks hundreds of metres in width, 0.2 to 3 km in length and arranged69

centroclinally(dipgenerallyinward)aroundthelake(Fig.5a)althoughareasnorthand70

eastof the lakeexhibitamore jumbledfaultpatternrelativetosouthernandwestern71

areas.ThefaultblocksarepresentonlywithinseveraltensofkilometresofLakeTana72

3

butthefelsicrocksextendwellbeyondthemaparea.73

Thethirdunit(“Chilgabeds”ofYemaneetal.1987)ispatchilypreserved,several74

tens to at least two hundred metres thick, and consists of organic-rich to lignitic75

mudstoneandfine-tocoarse-grainedsandstone, thin felsic tuffsandsilicitebeds,and76

interlayered sparselyolivine-phyric amygdaloidal (quartz-or calcite-filled)basalt and77

andesiticbasalt.Thesedimentaryrocksarelacustrineandfluvialdeposits,andcontain78

fossilboneandplantmaterial that,alongwithassociatedpalynomorphs, indicate that79

climatethenwasmuchwetterthantoday’s(Yemaneetal.1987).80

Thefourthunitisolivine-and/orplagioclase-phyricandvesicle-richto-poorbasalt81

thatform2-3kmhighplateaux.Itsbaseisanunconformityalongwhichsuccessivebasalt82

flowsonlapandburyanirregularpalaeotopographyformedonthefelsicfaultblocks(Fig.83

5b)andsedimentary-basaltunit(“Chilgabeds”).Thickbutvariablydevelopedkaolinite-84

richweatheringprofilesbeneaththeunconformitysurfaceindicateasubstantialhiatus85

priortotheinitialoutpouringofthebasaltformingthehighplateaux.Thefifthunitis86

scoraceousbasaltandcinderconesinareassouthofLakeTana.87

Our geophysical surveys show that Lake Tana overlies a basinmarked by sub-88

verticalwallsandfilledwithatleast180mofflat-lyingsediment(Fig.6).Thetimingof89

sedimentationremainstobedeterminedbutlinearextrapolationusinga14Cageof~1790

kaobtainedat10.2mdepthina90mcorethroughthosesediments(Marshalletal.2011)91

suggeststhatthebaseofthecoreisinexcessof150ka.Ineffect,Tanaisasteep-sided92

bowlfilledwiththickflat-lyingsedimentswhoseultimateageisunknownbutpredate93

thescoraceousbasaltsthatformthepresent-daydamandoutflowofthelake.94

95

2.1.Caldera-formingsupereruptions.PreviousworkersinterpretedtheTanabasinas96

aconsequenceoftheconfluenceofthreeriftstructures(e.g.Chorowiczetal.1998).Our97

newdatasuggestthatitisacaldera:thecentroclinalfaultblocksdecreaseinmagnitude98

(overdistancesoftensofkilometres)awayfromthelakeandpre-datetheunconformably99

overlyingsedimentary-basalt(“Chilgabeds”)andplateaux-formingbasaltunitsthusare100

unrelatedtotheyoungerhigh-angle,largelydip-slipfaultsthattransectalltherockunits101

intheTanabasinandwhicharepartoftheNeogeneopeningoftheMainEthiopiaRift.102

Further,theignimbritesandassociatedfelsicrocksarethickestandcoarsestadjacentto103

Tanaandthinandfineinalldirectionsawayfromthelake.Theserockscoverat least104

10,000km2butextend farbeyond themapareaandaverage200-300m in thickness105

4

hence their minimum eruptive volume would have been 2000-3000 km3. These are106

characteristics of super eruptions (e.g. Bryan and Ferrari 2013) and jointly with the107

seismicdataunderpinour interpretation thatTana isacalderaringedby faultblocks108

formedbycalderacollapse.Today’slandscapereflectsthisancientcalderaexhumedby109

head-warderosionoftheBlueNileRiversystem.110

111

3.Results:newgeochronologydata112

Ourgeochronologystrategywasdesignedtoconstrainthetimingofemplacementofthe113

felsicunit,determinetheagesofthebasaltsboundingthatunitanddefinetheageofthe114

basaltsthatdamtheoutflowofthepresent-daylake(Figs.2and3).Datesareshownwith115

two levels of uncertainty (±A/B) where A is the analytical uncertainty and B is the116

analyticaluncertaintycombinedwithsystematicuncertaintiesrelatedtocalibrationand117

decayconstants;detailsofthesamples,datingmethodologiesandanalyticalresultscan118

befoundintheSupplementalFiles.119

Age constraints for the felsic volcanism are from zircons separated from four120

samples and analysed at the NERC Isotope Geoscience Laboratory, UK, using U-Pb121

chemicalabrasion-isotopedilution-thermalionisationmassspectrometry(CA-ID-TIMS);122

all samples were spiked using the gravimetrically calibrated ET535 or ET2535123

EARTHTIME U-Pb tracer solutions (Condon et al. 2015; McLean et al. 2015; see124

SupplementalFiles). Constraintsforthemaficvolcanismareby40Ar/39Aranalyseson125

foursamplesdoneattheNERCArgonIsotopeFacility,UK,followingmethodsoutlinedin126

Marketal.2010(seeSupplementalFiles).127

128

3.1.Felsicsamples.SampleHydro-1isalapillituffnearthebaseofthefelsiteunit~7129

km southwest of Kunzla; four zirconswere analysed and yield amean age of 31.108130

±0.020/0.041Ma.SampleZege-1isaflow-bandedstonyrhyoliteinthecentralpartof131

thefelsiteunitexposedinanerosionalwindowthroughQuaternarybasalt~7kmwest132

ofBahirDar;fourzirconswereanalysedandyieldameanageof31.033±0.018/0.041133

Ma.SamplesYifag-2andAby-1areflow-bandedfeldspar-phyricrhyolitesfromtheupper134

partofthefelsiteunit~1.5and1kmwestofYifag,respectively;fivezirconsfromeach135

wereanalysedandyieldmeanagesof30.858±0.024/0.044Maand30.844±0.027/0.046136

Ma,respectively.137

138

5

3.2.Mafic samples. Allmafic samples are olivine- and/or plagioclase-phyric basalt.139

SampleBNG1wascollectednearthetopofthelowerfloodbasalt~0.5kmsoutheastof140

theBlueNileFallsandyieldedatotalfusionageof34.05±0.54/0.56Ma.SampleWD1is141

fromthesedimentary-basaltunit~8kmwestofAykelandyieldedaplateauageof28.90142

±0.12/0.14Ma.SampleKNZ1isfromthelowerpartoftheupperbasaltunit~3kmsouth143

ofKunzlaandyieldedaplateauageof23.75±0.02/0.04Ma.SampleBHD1isfromthe144

outflow of the Blue Nile River in Bahir Dar and yielded a plateau age of 0.033145

±0.005/0.005Ma.146

147

3.3.Geochronologysummary.Thenewgeochronologydefinesfourdistinctepisodes148

of volcanism: (i) ~1 km thick flood basalt likely as old as ~34 Ma but of unknown149

duration;(ii)caldera-formingsilicicsupereruptionsat~31Maspanning~250kyrwith150

aminimumeruptedvolumeof2000-3000km3;(iii)asecondepisodeofmaficvolcanism151

startingperhapsasearlyas~29Maandextendingtoatleast~24Ma;and(iv)localised152

basaltflowsandcindercones33kainage.The34Maageisolderthanthe31-29Maages153

typicallyattributedtoEthiopianfloodbasalt(Hofmannetal.1997;Rochetteetal.1998)154

and,evenifthatageisdownplayed,thisfloodbasaltcanbenoyoungerthanthec.31Ma155

ageoftheoverlyingfelsicrocksandhenceolderthanmostpreviouslydocumentedAfro-156

Arabian flood basalt. Determining if these phases are distinct requires further157

geochronologyandmapping.158

159

4.Discussion:TanavolcanismandEocene-Oligoceneenvironmentalchange160

StableisotopeandorganicgeochemistrytrendsfortheCenozoicdefineoneofthefinest161

timelinesofglobalchangeforanytimeinEarthhistory(Fig.7).Ournewdatapermita162

stricterevaluationthanpreviouslypossibleofAfro-ArabianLIPvolcanismtothosedata.163

164

4.1. Atmospheric CO2 drawdown and C-O isotopes. Following a zenith of global165

warmthat~55Maclimateunderwentadescentintothecurrentstateofbipolaricecaps166

(e.g. Zachos et al. 2001; Pälike et al 2006). The late Eocene-early Oligocene receives167

particular attention because of temporally distinct, sharp steps in C-O isotopes (~1-168

1.5‰, e.g. Oi1 and Oi-2 on Fig. 7; e.g. Miller et al. 1991; Coxall &Wilson 2011 and169

referencestherein)anddeclinesinatmosphericCO2(~300-500ppm;e.g.Zhangetal.170

2013;ArmstrongMcKayetal.2016andreferencestherein),eachexperiencingdurations171

6

estimatedtobebetweenseveral104and105years.172

LIPemissionscangenerateglobalwarmingbutcanalsocauseglobalcoolingdueto173

drawdownofthosegasesviaweatheringoffreshbasalt.TheAfro-ArabianLIPcomprises174

~13%of today’s exposedbasalt but only accounts for~4%of the annual fluxof CO2175

drawdownbecauseofitsaridsetting(Dessertetal.2003).However,thatregion’sclimate176

wasmuchwetterduringtheOligo-Miocene(e.g.Yemaneetal.1987)and,althoughno177

quantitative estimates exist for how much wetter, the then rate of consumption of178

atmosphericCO2offreshbasaltwouldhavebeenhigherthantoday’s.179

Aback-of-the-envelopecalculationcanbeundertakentoassesswhatthepotential180

effectof theTanaeruptionsmighthavebeenonatmosphericCO2drawdown,andwe181

make two assumptions: (i) that the present-day preserved outcrop area of the Afro-182

ArabianLIPapproximatestheoriginalextentoffloodbasalt;and(ii)afluxof4x1012183

molCO2/year,representativeoftoday’sotherlow-latitudefloodbasaltprovinces(Parana,184

Deccan,CentralAmerican;Dessert et al. 2003), is a reasonableapproximation for the185

weatheringofthosebasaltsduringPalaeogenetime.Thelatterassumptionisjustifiable186

becausefossilevidenceindicatesrelativelyhighprecipitationduringemplacementofthe187

Afro-ArabianLIP,whichhas30-50%moresurfaceareathanthoseprovinces.Further,188

estimatesofgloballyaveragedCO2consumptionbybasaltweatheringvaryby~4x,from189

1.75 x 1013 molCO2/year (Navarre-Stichler and Brantley 2007) to 4.08 x 1012190

molCO2/year (Dessert et al. 2003) hence the chosen flux is a reasonable estimate.191

Dependingonestimatesofatmosphericcomposition,1ppmofatmosphericCO2equates192

tobetween~3to7.8Gt.Generatingthestepwisedecreasesof300-500ppmCO2would193

haverequireddrawingdown2300-3900Gt.Simplifyingthecalculationtofocussolelyon194

weathering,atime-averagedfluxof4x1012molCO2/yearwouldhavedrawn-down~0.02195

GtCO2/year, requiring ~50-200 kyr to generate the estimated magnitude of change196

indicated by each of the step-wise declines of CO2. Volcanic degassing and/or cryptic197

degassing(ArmstrongMckayetal.2014)wouldhaveactedtoreducethedeclinesbut,198

giventheworldwidecoolingtrendatthattime,thatinfluencemusthavebeenminimal.199

Thisthoughtexperimentshowsthatarateofweathering-inducedCO2drawdown200

canbeobtainedthat iscompatiblewith that impliedbypreviousworkers for the two201

dropsinEocene-OligoceneCO2.However,mappingourgeochronologicaldataontothe202

timeline of Eocene-Oligocene environmental change (Fig. 7) refutes that possibility.203

Assumingthat(i)thec.104-5yearestimateddurationsrequiredtodrawdownCO2are204

7

correcttowithinanorder-of-magnitudeand(ii)theabsoluteagesofthosedrawdowns205

basedontheproxyrecordsarebroadlycorrect,thenthec.34Maepisodeoffloodbasalt206

predatesbyatleast0.5-1MyrtheproposedtimingoftheearlierdropinatmosphericCO2207

thatbeganpriortotheEocene-Oligoceneboundary(Fig.7).Further,theseconddropin208

CO2occurred~3Myrafterthec.29Maonsetoftheyoungerphaseofmaficvolcanism.209

ThemodelledpCO2dataoverthattimeare,withinerror, largelyinvariantfor3-4Myr210

throughtheoutpouringofthec.34-31Mafloodbasaltandc.31Masiliciceruptions,yet211

showaslight increaseduringtheapparentgapinvolcanicactivity intheTanaregion.212

Thus, there is a temporal disconnect between the timing of volcanic activity and the213

pattern of CO2 fluctuations indicated by the proxy data records (Fig. 7). Although214

weatheringoftheAfro-Arabianbasaltsnodoubtcontributedtolong-termCO2drawdown215

(e.g.Lefebvreetal.2013;Kent&Muttoni2013),ourfindingsruleoutvolcanismasakey216

influenceontheshorter-term,104-5yeardraw-downtimescalesimpliedbythevarious217

CO2proxydata.218

Ourgeochronologyalsorevealsthatthereisnoconsistentrelationshipbetweenthe219

trendofO-andC-isotopesandthetimingofmaficvolcanism.Oisotopesincrease,and220

sharplyso(i.e.atOi-1,Fig.7),enteringthec.34-31Maphaseofvolcanismbutmaintain221

a broadly invariant trend throughout that phase. In contrast, O-isotopes remain222

unperturbedacrosstheinitiationofthec.29Maphaseofmaficvolcanism,reachazenith223

(marked by Oi-2; Fig. 7) several millions of years after that onset, and then decline224

throughouttheremainingphaseofvolcanicactivity.Wedonotknowtheexacttimingof225

onsetofthepre-31Mafloodbasalt,anditistemptingtoconsiderapossiblelinktothe226

sharpincreaseinslopeinOisotopesattheEocene-Oligocenetransition,butadditional227

geochronology is needed to test this plausibility. The C-isotope profile also shows228

contraryrelationshipsbetweenvolcanicepisodes:C isotopesrise irregularlyandthen229

fallsteadilythroughthec.34-31Maphaseofmaficvolcanism,areunperturbedduring230

thec.31Masupereruptions,andriseirregularlythroughthec.29-23Maphaseofmafic231

volcanism.Thus,thec.34-23MyrspanofAfro-ArabianLIPvolcanismintheTanaregion232

failed to imprint significantly on the isotopic compositions of the Eocene-Oligocene233

oceansandatmosphere.234

235

4.2.Volatilesaskillmechanisms.Therateandmagnitudeofinjectionofsulphurand236

halogenmoleculesintotheatmospherefromLIPvolcanismhasbeenhighlightedaskill237

8

mechanisms for mass extinction especially when devolitalisation of evaporite- and238

organic-richcountryrockoccurs(e.g.Chenetetal.2007;Svensenetal.2009;Brandetal.239

2012;Blacketal.2012).TheAfro-ArabianLIPsilicicvolcanismreleased~45GtSand240

~224GtCl(Ayalewetal.2002),inpartattributabletovolatilesderivedfromthe1-2km241

thickMesozoicsuccessionofwhich~50%consistsofmarinecarbonateandevaporite242

rocksandminor lignitesandcoals (Wolelaet al. 2008).No suchestimateshavebeen243

madeforthemaficvolcanism,butafirst-orderapproximationforthepotentialamount244

ofdegassedvolatilesfromboththesilicicandmaficlavascanbeattempted.Wemakea245

suppositionthattheamountsofSandCldevolitalisedbybothtypesofvolcanismscaled246

proportionallytotheirrespectivevolumes.ThevolumeofbasaltintheAfro-ArabianLIP247

is ~6 x 104 km3, which is 5-15x the total volume that has been estimated for felsic248

volcanismacrosstheentireLIP(0.3–1x106km3;UkstinsPeateetal.2003). Assuch,249

Afro-Arabianvolcanismcouldhave released250 to750GtS and1300 to3600GtCl,250

amountsthatarecomparableatorder-of-magnitudeuncertaintiestothoseforotherLIPs251

suchastheDeccanandSiberiantraps(e.g.Blacketal.2012;Selfetal.2008).However,252

Afro-Arabian volcanism postdates the c. 34–33.7 Ma acme of progressive/stepwise253

plankton extinctions (Keller 2005; Pearson et al. 2008; Peters et al. 2013), which254

highlightsthat,althoughthetotalhalogeneffusionoftheAfro-ArabianLIPmayhavebeen255

large enough to affect deleteriously the biosphere, the rate of effusion was below a256

thresholdnecessarytocausemajordisruptionsinecosystems.Atemporallycalibrated257

example is the Central AtlanticMagmatic Provincewhere only the first (~600 kyr in258

duration)offourvolcanicpulsescanbelinkedtoanextinctioneventandthethreelater259

pulseshadlittletonoeffectonthebiosphere(Blackburnetal.2013).260

261

4.3.Silicicsupereruptions. Ournewdatashowthat theTana felsiceruptionswere262

comparable to some of the largest explosive volcanic events in Earth history. These263

spanned2-3Myrasevident from 40Ar/39Arplateauand single crystal agesandRb-Sr264

isochronagesonthick,widespreadignimbritesandtuffselsewhereinEthiopia,andon265

correlativetephralayers2700kmdistantintheIndianOcean(Bakeretal.1996;Coulié266

etal.2003;UkstinsPeateetal.2003,2008;Riisageretal.2005).ThetimingofOi-1and267

thec.31Masupereruptionsisinteresting(Fig.7)butsimilarspikesinO-isotopesoccur268

episodicallythroughtheEocene-Oligocene,whichurgescautionandconsiderationofthat269

relationshipasacoincidenceratherthanasacause-and-effect,particularlygiventhatthe270

9

O-isotopeprofilethroughtheshortintervalofTanasupereruptionsexhibitsjaggedrises271

andfallsnotunlikethosemarkingtheseveralmillionsofyearsprecedingandpostdating272

thesupereruptions.Asfortheotherisotopicproxies,theyremainunperturbedbythose273

eruptions(Fig.7).Perhapsthisisunsurprisinggiventhatthemid-Oligoceneeruption(c.274

28Ma) of the LaGarita caldera that formed the Fish CanyonTuff, the volumetrically275

largestwell-constrainedsilicicsupereruptionknownforthePhanerozoic(e.g.Wotzlaw276

etal.2014),likewisedoesnotcoincidewithaperturbationinthestableisotopedata.277

278

5. Conclusion. Mapping of our new age constraints onto the timeline for Eocene-279

Oligocene environmental change reveals that there is neither a synchronicity in the280

initiationofnoraconsistency inthetrendof thechanges-in-slopesofmarine isotopic281

compositionsandatmosphericCO2recordsfromonevolcaniceventtoanotherduring282

theAfro-ArabianLIP.This reconfirms thatAfro-ArabianLIPvolcanismwasseemingly283

ineffectual incausingsignificantdeviationof thetrajectoryofclimatechangethathad284

begun in the early Palaeogene. Unlike the <1-2 Myr timescales documented for the285

duration of other Phanerozoic LIPs, Afro-Arabian LIP volcanism was protracted and286

lackedthenecessaryandsufficientratesofeffusionandvolatileemissionto influence287

global-scaleenvironmentalchange.Further,assumingthattheTanasupereruptionsare288

typical of those at other times inEarth history, our data imply that the role of super289

eruptionsininducingEarthSystemchangemostlikelyhastobeaspartofanensemble290

ofenvironmentalcircumstancesratherthanasasoloact.291

292

Acknowledgements.WethanktheUniversityofAddisAbabaandHDibabeandthelate293

Prof M Umer for logistical support, and Prof H Lamb and D Clewley, University of294

Aberystwyth,andDHerd,ACalderandAMackie,UniversityofStAndrews,forassistance.295

DrNAtkinsonattheNERCIsotopeGeosciencesLaboratoryperformedtheU-Pbdating296

analyses. This work was supported by NERC Grants NE/D012996/1 and297

NER/B/S/2002/00540andNIGFSCIP/1024/0508.Threeanonymousreviewershelped298

improvethismanuscript.299

300 References301 Abate,B.,Koeberl,C.,Buchanan,P.C.,Körner,W.1998.Petrographyandgeochemistryof302 basalticandrhyodaciticrocksfromLakeTanaandtheGimjabet-Kosoberareas(North303 CentralEthiopia).J.AfricanEarthSciences26,119-134.304 305

10

ArmstrongMcKay,D.I.,Tyrell,T.,Wilson,P.A.,Foster,G.L.2014.Estimatingtheimpact306 ofcrypticdegassingofLargeIgneousProvinces:amid-Miocenecasestudy.Earthand307 PlanetaryScienceLetters403,254-262.308 309 Ayalew,D.,Barbey,P.,Marty,B.,Reisberg,L.,Yirgu,G.,Pik,R.2002.Source,genesisand310 timingofgiantignimbritedepositsassociatedwithEthiopiancontinentalfloodbasalts.311 Geoch.Cosmoch.Acta66,1429-1448.312 313 Baker,J.,Snee,L.,Menzies,M.1996.AbriefOligoceneperiodoffloodvolcanismin314 Yemen:implicationsforthedurationandrateofcontinentalfloodvolcanismatthe315 Afro-Arabiantriplejunction.EarthandPlanetaryScienceLetters138,39-55.316 317 Black,B.A.,Elkins-Tanton,L.T.,Rowe,M.C.,UkstinsPeate,I.2012.Magnitudeand318 consequencesofvolatilereleasefromtheSiberianTraps.EarthandPlanetaryScience319 Letters317-318,363-373.320 321 Blackburn,T.J.,Olsen,P.E.,Bowring,S.A.,McLean,N.M.,Kent,D.V.,Puffer,J.,McHone,G.,322 Rasbuty,E.T.,Et-Touhami,M.2013.ZirconU-Pbgeochronologylinkstheend-Triassic323 ExtinctionwiththeCentralAtlanticMagmaticProcess.Science340,941-945.324 325 Bond,D.P.G.,Wignall,P.B.2014.Largeigneousprovincesandmassextinctionsan326 update,inKeller,G.andKerr,A.C.,eds.Volcanism,Impacts,andMassExtinctions:327 CausesandEffects.Geol.Soc.Amer.Spec.Pap.505,29-55.328 329 Brand,U.,Pesenato,R.,Came,R,Affek,H.,Angiolini,L.,Azmy,K.,Farabegoli,E.2012.The330 end-Permianmassextinction:arapidvolcanicCO2andCH4-climaticcatastrophe.331 ChemicalGeology322-323,121-144.332 333 Bryan,S.E.,Ferrari,L.2013.Largeigneousprovincesandsiliciclargeigneousprovinces:334 progressinunderstandingoverthelast25years.Geol.Soc.Amer.Bull.125,1053-1078.335 336 Burgess,S.,Bowring,S.,Shen,S-Z.2014.High-precisiontimelineforEarth’smostsevere337 extinction.ProceedingsNat.Acad.Sci.USA.111,3316-3321.338 339 Chenet,A.L.,Quidelleur,X.,Fluteau,F.,Courtillot,V.,Bajpal,S.2007.40K-40Ardatingof340 themainDeccanlargeigneousprovince:furtherevidenceofKTBageandshort341 duration.EarthPlanet.Sci.Lett.263,1-15.342 343 Condon,D.J.,Schoene,B.,McLean,N.,Bowring,S.A.,Parrish,R.2015.Metrologyand344 traceabilityofU-Pbisotopedilutiongeochronology(EARTHTIMETracerCalibration345 PartI):GeochimicaetCosmochimicaActa,doi:10.1016/j.gca.2015.05.026346 347 Chorowicz,J.,Collet,B.,Bonavia,F.F.,Mohr,P.,Parrot,J.F.,Korme,T.1998.TheTana348 basin,Ethiopia:intra-plateauuplift,riftingandsubsidence.Tectonophys.295,351-367.349 350 Coulié,E.,Quidelleur,X.,Gillot,P-Y.,Courtillot,V.,Lefèvre,J-C.,Chiesa,S.2003.351 ComparativeK-ArandAr/ArdatingofEthiopianandYemeniteOligocenevolcanism:352 implicationsfortiminganddurationoftheEthiopiantraps.EarthandPlanetaryScience353 Letters206,477-492.354

11

355 Coxall,H.K.,Wilson,P.A.,Pälike,H.,Lear,C.H.,Backman,J.2005.Rapidstepwiseonsetof356 AntarcticglaciationanddeepercalcitecompensationdepthinthePacificOcean.Nature357 433,53-57.358 359 Cramer,B.S.,Toggweiler,J.R.,Wright,J.D.,Katz,M.E.,Miller,K.G.2009.Ocean360 overturningsincetheLateCretaceous:inferencesfromanewbenthicforaminiferal361 isotopecompilation.Paleoceanography24,doi:10.1029/2008PA001683.362 363 Dessert,C.,Depré,B.,Gaillardet,J.,François,L.M.,Allègre,C.2003.Basaltweathering364 lawsandimpactofbasaltweatheringontheglobalcarboncycle.Chem.Geol.202,257-365 273.366 367 Hofmann,C.,Courtillot,V.,Féraud,G.,Rochette,P.,Yirgu,G.,Ketefo,E.,Pik,R.1997.368 TimingoftheEthiopianfloodbasalteventandimplicationsforplumebirthandglobal369 change.Nature389,838-841.370 371 Keller,G.2005.Impacts,volcanismandmassextinction:randomcoincidenceorcause372 andeffect.AustralianJ.EarthSci.52,725-757.373 374 Kent,D.V.,Muttoni,G.2013.ModulationofLateCretaceousandCenozoicclimateby375 variabledrawdownofatmosphericpCO2fromweatheringofbasalticprovinceson376 continentsdriftingthroughtheequatorialhumidbelt.Clim.Past9,524-546.377 378 Lefebvre,V.,Donnadieu,Y.,Godderis,Y.,Fluteau,F.,Hubert-Theuo,F.2013.Wasthe379 AntarcticdeglaciationcausedbyahighdegassingrateduringtheearlyCenozoic.Earth380 andPlanetaryScienceLetters371-372,203-211.381 382 Mark,D.F.,Gonzalez,S.,Huddart,D.,Bohnel,H.2010.DatingoftheValsequillovolcanic383 deposits:ResolutionofanongoingcontroversyinCentralMexico.JournalofHuman384 Evolution58,441-445.385 386 Marshall,M.H.,Lamb,H.F.,Huws,D.,Davies,S.J.,Bates,R.,Bloemendal,J.,Boyle,J.,Leng,387 M.J.,Umer,M.,Bryant,C.2011.LatePleistoceneandHolocenedroughteventsatLake388 Tana,thesourceoftheBlueNile.GlobalandPlanetaryChange78,147-161.389 390 McLean,N.,Condon,D.J.,Schoene,B.,Bowring,S.A.2015.Evaluatinguncertaintiesinthe391 calibrationofisotopicreferencematerialsandmulti-elementisotopictracers392 (EARTHTIMETracerCalibrationPartII):GeochimicaetCosmochimicaActa,393 doi:10.1016/j.gca.2015.02.040394 395 Miller,K.G.,Wright,J.D.,Fairbanks,R.G.1991.UnlockingtheIceHouse:Oligocene–396 Mioceneoxygenisotopes,eustasy,andmarginerosion.J.Geophys.Res.96,6829–6848.397 398 Mohr,P.1983.Ethiopianfloodbasaltprovince.Nature303,577-584.399 400 Mohr,P.,Zanettin,B.1988.TheEthiopianfloodbasaltprovince,in.J.D.Macdougall(Ed)401 ContinentalFloodBasalts.Kluwer,Dordrecht,63-110.402 403

12

Navarre-Sitchler,A.,Brantley,S.2007.Basaltweatheringacrossscales.EarthPlanet.Sci.404 Lett.261,321-334.405 406 Pälike,H.,Norris,R.D.,Herrle,J.O.,Wilson,P.A.,Coxall,H.K.,Lear,C.H.,Shackleton,N.J.,407 Triparti,A.K.,Wade,B.S.2006.TheheartbeatoftheOligoceneclimatesystem.Science408 324,1894-1898.409 410 Pagani,M.,Huber,M.,Liu,Z.,Bohaty,S.M.,Henderiks,J.,Sijp,W.,Krishnan,S.,DeConto,411 R.M.2011.TheroleofcarbondioxideduringonsetofAntarcticglaciation.Science334,412 1261-1264.413 414 Pearson,P.N.,McMillan,I.K.,Wade,B.S.,Jones,T.D.,Coxall,H.K.,Brown,P.R.,Lear,C.H.415 2008.ExtinctionandenvironmentalchangeacrosstheEocene-Oligoceneboundaryin416 Tanzania.Geology36,179-182.417 418 Peters,S.E.,Kelly,D.C.,Fraass,J.2013.Oceanographiccontrolsonthediversityand419 extinctionofplanktonicforaminifera.Nature493,398-01.420 421 Pik,R.,Deniel,C,Coulon,C,Yirgu,G.,Hofmann,Ayalew,D.1998.Thenorthwestern422 EthiopianPlateaufloodbasalts:classificationandspatialdistributionofmagmatypes.J.423 VolcanologyandGeothermalResearch81,91-111.424 425 Pik,R.,Deniel,C,Coulon,C,Yirgu,Marty,B.1999.Isotopicandtraceelementsignatures426 ofEthiopianfloodbasalts:evidenceforplume-lithosphereinteractions.Geochim.427 Cosmoch.Acta63,2263-2279.428 429 Riisager,P.,Knight,K.B.,Baker,J.A.,UkstinsPeate,I.,Al-Kasadi,M.,Al-Subbary,A.,430 Renne,P.R.2005.Paleomagnetismand40Ar/39ArgeochronologyofYemeniOligocene431 volcanics:implicationsfortiminganddurationofAfro-Arabiantrapsandgeometryof432 theOligocenepaleomagneticfield.EarthPlanet.Sci.Lett.237,647-672.433 434 Rochette,P.,Tamrat,E.,Féraud,Pik,R.,Courtillot,V.,Ketefo,E.,Coulon,C.,Hoffmann,C.,435 Vandamme,D.,Yirgu,G.1998.MagnetostratigraphyandtimingoftheOligocene436 Ethiopiantraps.EarthPlanet.Sci.Lett.164,497-510.437 438 Self,S.,Blake,S.,Sharma,K.,Widdowson,M.,Sephton,S.2008.Sulfurandchlorinein439 LateCretaceousDeccanmagmasanderuptivegasrelease.Science319,1654-1657.440 441 Svensen,H.,Planke,S.,Polozov,A.G.,Schmidbauer,N.,Corfu,F.,Podladchikov,Y.Y.,442 Jamveit,B.2009.SiberiangasventingandtheendPermianmassextinction.Earthand443 PlanetaryScienceLetters277,490-500.444 445 Ukstins,I.A.,Renne,P.R.,Wolfenden,E.,Baker,J.,Ayalew,D.,Menzies,M.2002.Matching446 conjugatevolcanicriftedmargins:40Ar/39Archronostratigraphyofpre-andsyn-rift447 bimodalfloodvolcanisminEthiopiaandYemen.EarthPlanet.Sci.Lett.198,289-306.448 449 UkstinsPeate,I.,Baker,J.A.,Kent,A.J.R.,Al-Kadasi,M.,Al-Subbary,A.,Ayalew,D.,450 Menzies,M.2003.CorrelationofIndianOceantephratoindividualOligocenesilicic451 eruptionsfromtheAfro-Arabianfloodvolcanism.EarthPlanet.Sci.Lett.211,311-327.452

13

453 UkstinsPeate,I.,Kent,A.J.R.,Baker,J.A.,Menzies,M.A.2008.Extremegeochemical454 heterogeneityinAfro-ArabianOligocenetephras:preservingfractionalcrystallization455 andmaficrechargeprocessesinsilicicmagmachambers.Lithos102,260-278.456 457 Westerhold,T.,Rohl,U.,Palike,H.,Wilkens,R.,Wilson,P.A.,Acton,G.2014.Orbitally458 tunedtimescaleandastronomicalforcinginthemiddleEocenetoearlyOligocene.459 ClimateofthePast10,955-973.460 461 Wolela,A.2008.SedimentationoftheTriassic-JurassicAdrigatSandstoneFormation,462 BlueNile(Abay)Basin,Ethiopia.J.Afr.EarthSci.52,30-42.463 464 Wotzlaw,J-F.,Schalteger,U.,Frick,D.A.,Dungan,M.A.,Gerdes,A.,Günther.2014.465 Trackingtheevolutionoflarge-volumesilicicmagmareservoirsfromassemblyto466 supereruption.Geology41,867-870.467 468 Yemane,K.,Robert,C.,Bonnefille,R.1987.Pollenandclaymineralassemblagesofalate469 MiocenelacustrinesequencefromtheNorthwesternEthiopianHighlands.470 Palaeogeography,Palaeoclimatology,Palaeoecology60,123-141.471 472 Zachos,J.C.,Dickens,G.R.,Zeebe,R.E.2008.AnearlyCenozoicperspectiveon473 greenhousewarmingandcarbon-cycledynamics.Nature451,279-283.474 475 Zachos,J.C.,Pagani,M.,Sloan,L.,Thomas,E.,Billups,K.2001.Globalclimate65Mato476 Present.Science292,686-693.477 478 Zhang,Y.G.,Pagani,M,Liu,Z.,Bohaty,S.M.,DeConto,R.2013.A40-million-yearhistory479 ofatmosphericCO2.Philos.Trans.Roy.SocA,1-20.480 481 482

14

Figure1.GeneralisedCenozoicgeologyoftheregionsborderingtheMainEthiopianRift483 system(simplifiedfromtheGeologicalMapofEthiopia).Circled‘A’isthelocationof484 AddisAbaba.485 486 Figure2.SimplifiedgeologicalmapoftheLakeTanaregionbasedonmappingdone487 duringthiswork.488 489 Figure3.StratigraphyoftheLakeTanaregionshowingthefourmapunitsandposition490 ofgeochronologysamples.SeetextfordetailsandSupplementaryFilesforGPS491 locationsofsamples.m.a.s.l.–metresabovesealevel.492 493 Figure4.Felsicvolcanicrocks.A.Rhyolitedomes10-20kmsoutheastofKunzla494 (southwestcornerofLakeTana).B.Tiltedrhyolitedome5kmwestofGorgora(north-495 centralmarginofLakeTana);basalpartconsistsofflow-bandedrhyolitewhereasinner496 partofdomeconsistsofmoremassiverhyolite.C.Flow-bandedrhyoliteofsampleAby-497 1(northeastcornerofLakeTana)thatyieldedaU-Pbzirconageof30.844498 ±0.027/0.046Ma.DandE.Massiveignimbriteflowsandlargebombofpumiceousand499 bandedrhyolite(~12kmwestofYifag,northeastmarginofLakeTana).FandG.500 Fragmentaltextureofignimbritedeposits;Fcontainsavarietyoffelsicclastsandminor501 maficclastswhereasGismostlypumice(areawestofGorgora,north-centralshoreline502 ofLakeTana).H.Cross-beddedvolcaniclasticconglomerate-sandstonewithobsidian503 andfelsicvolcanicclastsandrarebasaltclasts;thisunitoverliessharplythefelsictuffof504 sampleHydro-1(15kmsouthwestofKunzla)thatyieldedaU-Pbzirconageof31.108505 ±0.020/0.041Ma.506 507 Figure5.Structuralfeaturesofthefelsicunit.A.East-tiltedfaultblocksofsilicic508 ignimbriteandtuff(~5kmnorthwestofKunzla,southwestmarginofLakeTana).B.509 Progressiveonlapandburialofc.31Masouthwest-tiltedfelsiticrocksbyflat-lyingc.24510 Mabasalt(~20kmsouthofKunzla).511 512 Figure6.Seismicprofileshowingsteep-sidedbedrockmarginbeneathLakeTanaand513 howthesedimentaryfillabutsagainstandburiesthosemargins.Thesedimentsare514 >180mthick(andlikelymuchthicker)andlinearextrapolationofthe14Cageforthe515 topmostdesiccationsurface(Marshalletal.2011)wouldimplythatthebaseofthe516 seismicallyimagedsedimentisinexcessof150ka.517 518 Figure7.IsotopicandatmosphericCO2envelopesfortheCenozoic;greyandblack519 shadingdenotesagebracketsforvolcanismintheLakeTanaandsurroundingregions.520 C-andO-isotopeenvelopesarefromZachosetal.(2001,2008)andotherproposed521 best-fitsolidlinesarefromCrameretal.(2009).CO2trendsgeneralisedfrommultiple522 proxiesasreportedbyPaganietal.(2011)andZhangetal.(2013).Timescale523 calibrationfollowsWesterholdetal.(2014).OiandMi:oxygenisotopeevents.524 525

15

Figure1526

527 528 529

16

Figure2.530

531 532 533

17

Figure3.534

535 536 537

18

Figure4.538

539 540 541

19

Figure5.542

543 544 545

20

Figure6.546

547 548 549

21

Figure7.550

551