RESEARCH ARTICLE SUMMARY

TECTONICS

Revised paleoaltimetry datashow low Tibetan Plateau elevationduring the EoceneSvetlana Botsyun Pierre Sepulchre Yannick Donnadieu Camille RisiAlexis Licht Jeremy K Caves Rugenstein

INTRODUCTION The uplift history of theTibetan Plateau (TP) is critical for understand-ing the evolution of the Asian monsoons andthe geodynamic forces involved in collisionalorogens The early topographic history of theTP is uncertain and the timing of the initia-tion of uplift remains controversial The ma-jority of studies that find evidence for an oldplateau (as early as the Eocene~40 million years ago) rely onstable isotope paleoaltimetryThis method is based on ob-servations andmodels that showdepletion in heavy oxygen iso-topes in rainfall during orographicascent Ancient carbonates ofthe TP record past rainfall iso-topes when available their iso-topic signature can be comparedwith isotopic lapse rates in orderto estimate atwhat elevation theygrew in the past but the use ofthis method in deep time hasmany uncertainties

RATIONALE Applying stable-isotope paleoaltimetry in green-house climatesmakes the implicitassumption that the factors thatcontrol atmospheric distillationand rainfall oxygen isotopic com-position (d18Ow) have remainedconstant over millions of yearsHowever the impact of past cli-mate change on d18Ow values isunclear In particular for theEocene higher atmospheric CO2

concentration (PCO2) and mark-edly different Asian paleogeog-raphy including awide and shallow ParatethysSea in central China and a latitudinal shift ofthe southern Tibet margin ~10deg to the southhave been hypothesized to modify the Asianclimate and regional d18Ow values In additionthe carbonate formation temperature is oftenunknown increasing the uncertainty in thereconstructed d18Ow

RESULTSWe ran climate simulations withEocene boundary conditions and varying TPelevation We accounted for changing PCO2land surface albedo orbital variation andsea-surface temperatures which potentiallycause shifts in d18Ow by changing the rel-ative contribution of different air masses andthe local hydrological cyclemdashthe evaporation-

to-precipitation ratio and the fractioning be-tween convective and large-scale rainfalls Inour simulations the south-shifted locationof the entire Indian foreland induces strongconvection over the southern flank of the TPand a radically different pattern of water re-cycling compared with that of present dayOur simulations reproduce monsoonlike sea-

sonal precipitation over the Indian forelandwith summer convective rainfall reaching theTP An intense anticyclonic circulation duringsummer months induces widespread aridityon the northern part of the Plateau This pecu-liar atmospheric circulation together with in-tensified water recycling and multiple moisturesources results in a reversed isotopic lapse rateacross the southern flank of the TP with themost negative d18Ow over northern India and

increased d18Ow northwardOn the basis of Eoceneexperiments with variedboundary conditions weargue that this is a robustfeature of Eocene climateover the region Further-

more this pattern is the opposite of the present-day d18Ow over the Himalayas which decreaseswith elevation driven by orographic rainoutfollowing a Rayleigh distillation process Lastusing our simulated temperatures and d18Owwe derived virtual carbonates d18O for dif-ferent elevation scenarios and compared themwith the geological record Statistical analysis

shows that a low TP topographyduring the Eocene is the scenariothat provides the best match be-tween model and data

CONCLUSION Our simulationsindicate that standard stable iso-tope paleoaltimetry methods arenot applicable in Eocene Asiabecause of a combination of in-creased convective precipitationmixture of air masses of differentorigin and widespread aridityIn the Eocene the presence ofa reversed isotopic lapse rate pre-cludes use of any of the previ-ously developed d18Ow-elevationrelationships for estimating theEocene elevationof theTPRathera model-data comparison on thecarbonates d18O suggests thatthe TP reached only low to mod-erate (lt3000 m) elevations dur-ing the Eocene reconciling oxygenisotope data with other proxiesof elevation andwith geodynamicmodels that propose a recent(Neogene) uplift More gener-ally we suggest that using cli-mate models in conjunction withstable isotope data from the geo-

logical archives provides a powerful tool toincorporate climatic changes into the analysisof paleoelevations

RESEARCH

Botsyun et al Science 363 946 (2019) 1 March 2019 1 of 1

The list of author affiliations is available in the full article onlineCorresponding author Email svetlanabotsyununi-tuebingendeCite this article as S Botsyun et al Science 363 eaaq1436(2019) DOI 101126scienceaaq1436

50ordm E 70ordm E 90ordm E 110ordm E

0ordm N

10ordm N

20ordm N

30ordm N

40ordm N

Paci

fic O

cean

Paratethys Sea

Indian Ocean-5

-5

-5

-5

-3 -7

-3

-9

-11

-7

-9

-3

-3

-7

-7

-5

-5

9-9-9

Simulated stable oxygen isotope composition of summerprecipitation for the Eocene (42 million years ago) Asia Thecombination of changes in water recycling and the source ofair masses induces the reversal of isotopic lapse rate Summer windstrajectories highlight intense westerlies and easterlies at eachborder of the TP Circles indicate localities of paleoaltimetry sitesused in this study

ON OUR WEBSITE

Read the full articleat httpdxdoiorg101126scienceaaq1436

on March 11 2020

httpsciencesciencem

agorgD

ownloaded from

RESEARCH ARTICLE

TECTONICS

Revised paleoaltimetry datashow low Tibetan Plateau elevationduring the EoceneSvetlana Botsyun12 Pierre Sepulchre1 Yannick Donnadieu13 Camille Risi4Alexis Licht5 Jeremy K Caves Rugenstein67

Paleotopographic reconstructions of the Tibetan Plateau based on stable isotopepaleoaltimetry methods conclude that most of the Plateaursquos current elevation was alreadyreached by the Eocene ~40 million years ago However changes in atmospheric andhydrological dynamics affect oxygen stable isotopes in precipitation and may thus biassuch reconstructions We used an isotope-equipped general circulation model to assessthe influence of changing Eocene paleogeography and climate on paleoelevationestimates Our simulations indicate that stable isotope paleoaltimetry methods are notapplicable in Eocene Asia because of a combination of increased convective precipitationmixture of air masses and widespread aridity Rather a model-data comparisonsuggests that the Tibetan Plateau only reached low to moderate (less than 3000 meters)elevations during the Eocene reconciling oxygen isotope data with other proxies

Because the Tibetan Plateau (TP) is thelargest orogen on Earth its uplift historyis critical for understanding the evolutionof the Asian monsoon systems and thegeodynamic forces involved in continen-

tal collisions (1 2) Previous reconstructionshave proposed a wide range of paleogeographiesfor the late Eocene of Tibet which all accommo-date India-Asia plate motion through a combi-nation of Indian plate subduction and Asianshortening and extrusion (3) but the altitudeof the TP through time remains disputed (4)Numerous uplift scenarios for the TP have beensuggested from a precollision highly elevatedPlateau [for example ~100 million years (Ma)ago] (5) through stepwise uplift since initial col-lision (~55 Ma ago) (6) to more recent rapiduplift particularly of the northern and easternTP (lt8 Ma ago) (7) Although any of these upliftscenarios are possible they imply different geo-dynamic mechanisms and climate responses

The elevation of the TP since the Eocene asreconstructed in stable isotope paleoaltimetrystudies is close to modern values (gt4000 m)(8ndash10) However other lines of evidence frompaleobotany (11 12) and paleontology (13 14)suggest much lower elevations of the TP in thepast Improved constraints on the early topo-graphic history of the TP are critical to betterunderstand the geodynamic evolution of India-Eurasian collision and its consequences on theclimate systemmdashnamely the Asian monsoon onsetStable isotope paleoaltimetry is based on the

observed and theoretically predicted decreaseof precipitation water d18O [d18Ow expressed inper mil (permil)] with increasing elevation across

mountain ranges (15) In the Himalayas andsouthern Tibet the present-day isotopic lapserate is ~ndash28permilkm (16) which is in agreementwith distillation models that predict decreasingd18Ow in response to decreasing atmospherichumidity and temperature during orographicascent (15) This altitude effect is archived in thed18O of continental carbonates (d18Oc) offsetfrom d18Ow by a fractionation factor that de-pends on the carbonate formation temperature(17) Stable isotope paleoaltimetry studies com-monly provide d18Oc at sites of unknown ele-vation make assumptions about the temperatureat the site the near-sea level d18Ow and theatmospheric relative humidity and from thisreconstruct the paleoelevation by using mod-ern empirical or modeled lapse rates (15 16)However stable isotope paleoaltimetry relies

on the key assumption that the main atmosphericfactors that control atmospheric distillation andair mass d18O values have remained constantover millions of years The impact of past cli-mate changemdashincluding changes in lapse rates(18) sources of precipitation (9) and in mon-soonal regime (19)mdashon d18Ow is uncertain but hasbeen hypothesized to affect d18Ow over the TP(20) In addition carbonate growth temperatureis often unknown increasing the uncertainty inthe reconstructed d18Ow (17) All these factors mayimportantly bias deep-time paleoelevation esti-mates particularly during the Eocene Greenhouseworld Higher atmospheric CO2 concentration(PCO2) at that time (21) and markedly differentAsian paleogeography with a wide and shallowParatethys Sea in central China (22) and a latitu-dinal shift of the southern Tibetan margin ~10degto the south (23) might have substantially modi-fied the Asian hydrological cycle (24 25)We tested the impact of late Eocene paleo-

geography and topography variations on watervapor and precipitation d18O using an isotope-enabled version of the LMDZ atmospheric cir-culation model and compared simulated d18Oc

values with previously published data LMDZ-iso properly simulates atmospheric dynamics

RESEARCH

Botsyun et al Science 363 eaaq1436 (2019) 1 March 2019 1 of 9

1Laboratoire des Sciences du Climat et de lrsquoEnvironnement(LSCE)Institute Pierre Simon Laplace (IPSL) Commissariataacute lrsquoEacutenergie Atomique et aux Eacutenergies Alternatives (CEA)-CNRSndashUniversiteacute de Versailles Saint-Quentin-en-Yvelines(UVSQ) Universiteacute Paris-Saclay Gif-sur-Yvette France2Department of Geosciences University of TuumlbingenTuumlbingen Germany 3Aix-Marseille Universiteacute CNRS Institutpour la Recherche et le Deacuteveloppement (IRD) Collegravege deFrance Centre de Recherche et drsquoEnseignement deGeacuteosciences de lrsquoEnvironnement (CEREGE) Aix-en-ProvenceFrance 4Laboratoire de Meacuteteacuteorologie Dynamique IPSLSorbonne Universiteacute CNRS Paris France 5Department ofEarth and Space Sciences University of Washington SeattleWA USA 6Department of Earth System Science StanfordUniversity Stanford CA USA 7Department of EarthSciences ETH-Zuumlrich Zuumlrich SwitzerlandCorresponding author Email svetlanabotsyununi-tuebingende

Movie 1 Model-simulated d18Ow and precipitation amount for present-day (CTR) and Eocene(EOC) cases monthly time slices Vector arrows show wind speed and directions

on March 11 2020

httpsciencesciencem

agorgD

ownloaded from

and reproduces rainfall and d18Ow patternsconsistent with present-day data over Asia in-cluding the substantial depletion that occursacross the Himalayan flanks and the stark south-to-north positive gradient in d18Ow across thePlateau (Fig 1 A and B) In order to fully ac-count for uncertainties regarding our knowl-edge of Asian paleogeography during the Eocenewe varied the elevation of the TP the latitudeof the TP and the land-sea mask (Fig S1) Ad-ditionally we accounted for changing PCO2 landsurface albedo orbital variation and sea-surfacetemperatures These variables potentially causeshifts in d18Ow by changing the relative contri-

bution of different air masses and the local hy-drological cycle (26ndash28)

Eocene Asian climate and water d18O

Simulated Eocene large-scale circulation is char-acterized by monsoonlike seasonal precipitationover the Indian foreland (fig S2) By contrastover the TP and over central Asia westerliesmore intense than those of today interactedwith large-scale atmospheric subsidence (figs S2and S3) causing widespread aridity (Fig 1C andfig S2) The onset of an anticyclonic circulationduring summer months over the eastern part ofthe TP (Movie 1) induces intense westerlies and

easterlies at each border of the TP This featureis in stark contrast with the modern simulationsin which the southern TP receives a greaterproportion of precipitation from air massesoriginating in the Bay of Bengal and propagat-ing northward (Movie 1) These processes occuracross all of our different paleogeography andpaleotopography scenarios High rainfall overthe Indian foreland (Fig 1 C and D) decreasesd18Ow from the southern tip of India (~ndash4permil)to a minimum at 15degN over the Indian foreland(from ndash8 to ndash11permil depending upon TP topo-graphy) (Fig 2 and fig S2) The multiple sourcesfeeding precipitation in the southern part of

Botsyun et al Science 363 eaaq1436 (2019) 1 March 2019 2 of 9

Fig 1 Present-day and Eocene simulations (A) Simulated MJJAS precipitation-weighted present-day d18Ow Triangles show d18Ow fromGNIP stations (70) circles represent d18O in streams rivers and springs from [(37) and references therein] compilation (B) The d18O south-northprofile averaged between 70degE and 90degE Gray lines indicate minimum and maximum model d18Ow 70degE and 90degE (C and D) Simulated Eoceneprecipitation and isotopic patterns (C) LMDZ-simulated MJJAS precipitation amount (millimeters per day) and (D) MJJAS precipitation-weightedd18Ow (permil) and wind velocities (meters per second) for the Eocene (EOC-L) experiment

RESEARCH | RESEARCH ARTICLEon M

arch 11 2020

httpsciencesciencemagorg

Dow

nloaded from

India explain the absence of a correlation be-tween the rainfall and the value of d18Ow in theregion (Movie 1) Conversely more negative d18Ow

values over the northern part of the Indiancontinent arise from distillation processes asspecific humidity decreases as air masses pro-gress northward across India owing to highprecipitation values In addition extremely arid

environments (lt05 mmday) over the northernTP and central Asia are associated with higherd18Ow values (from ndash3 to ndash1permil) (Fig 1 C and D)The most surprising and counterintuitive resultin our Eocene simulations is an isotopic gradientacross the southern flank of the TP with d18Ow

values ranging from ndash8permil at the foot of the TP tondash5permil at the top of the southern flank of the TP

Such a gradient is at odds with modern observedand simulated values that depict a decrease ind18Ow from ndash8 to ndash18permil as altitude increases onthe southern flank of the TP (Fig 2 and fig S4)The combination of changes in water recyclingand the source of air masses explains the reversalof the isotopic lapse rate during orographic as-cent in the Eocene

Back-trajectory analysis

Our back-trajectory analysis confirms that inthe Eocene sites across the TP are influencedby multiple air masses that originate from boththe Indian Ocean and the Paratethys Sea (Fig 3and Movie 1) In detail stronger westerlies duringthe Eocene are divided between a southernbranch following the Himalayan flanks anda northern branch that deviates to the southand joins the easterlies around 95degE (Movie 1)The presence of multiple air masses over theEocene southern TP invalidates two commonassumptions (29 30) in stable isotope paleo-altimetry that (i) southern TP sites receive mois-ture from a single air mass and (ii) air massesprogress over the southern margin and onto theTP permitting the resulting d18Ow to be treatedas a Rayleigh distillation process Instead d18Ow

over the southern TP appears to be a mixture ofmultiple air masses having followed along-striketrajectories (eastward and westward winds)Therefore the assumption of Rayleigh distilla-tion along a single air mass trajectory cannotbe used to explain d18Ow patterns over elevatedparts of the Eocene topography Our modelingresults are also supported by present-day satelliteobservations showing that mixing of various airmasses tends to induce higher d18Ow than apurely Rayleigh regime (31)

Evaporation to precipitation ratio andprecipitation regimes

The south-shifted location of the Eocene TPinduces strong convection over its southernflank and a radically different pattern of waterrecycling compared with the modern pattern(Fig 4) These features persist even in a sce-nario in which the TP has been latitudinallyshifted (fig S5) Our present-day simulationreproduces the large decrease in convectiveprecipitation that exists from the Indian forelandup to the southern flank of the TP [evaporationprecipitation ratio (EP) diminishes along thesame transect] (Fig 4A) However during theEocene convective precipitation (CP) prevailsover large-scale precipitation (LSP) (CPLSP gt 1)with no clear decrease in EP over this region(Fig 4B and fig S5) The negligible decreasein d18Ow in the Eocene between the Indianforeland and the southern flank of the TP to-gether with high convective precipitation rateare therefore consistent with observationaland modeling studies (32ndash34) that show thatstratiform precipitation regimes are associatedwith more depleted isotopic values than con-vective regimes An additional simulation usinga TP of the equivalent elevation as today for theEocene does not change zonal profiles of EP

Botsyun et al Science 363 eaaq1436 (2019) 1 March 2019 3 of 9

Fig 2 Precipitation-weighted MJJAS d18Ow in simulations (A) EOC-L (B) EOC-XL (C) EOC-S(D) EOC-Him (E) EOC-M (F) EOC-sea (G) EOC-North (H) EOC-South Vectors show MJJASvertically integrated moisture transport (kilograms per meter per second)

RESEARCH | RESEARCH ARTICLEon M

arch 11 2020

httpsciencesciencemagorg

Dow

nloaded from

and CPLSP suggesting that our hypothesizedmechanism to explain reverse isotopic gradientsin the Eocene over the southern flank of the TPis robust

Reversed Eocene isotopic lapse rate

The increased convective precipitation rates andmultiple air masses over the southern TP resultin a latitudinal dipole in d18Ow with low d18Ow

over the India foreland and higher d18Ow overthe TP This dipole creates an apparent positiveisotopic lapse rate across the southern edge ofthe TP with increased d18Ow as elevation in-creases south to north This positive isotopiclapse rate is a feature of all Eocene simulationsregardless of the prescribed topography of thelatitudinal position of the TP and of the po-sition of the coastline (Fig 2) Furthermorethis pattern is opposite to the expected d18Ow-elevation relationship derived from Rayleighdistillation and driven by orographic rainoutwhich predicts that d18Ow should decrease withelevation (15) Flat isotopic lapse rates are foundtoday in areas of high convective precipitationsuch as Ethiopia and Kenya (35) because of con-vective instability over the region that producesenriched d18O values at altitude Similar negligi-ble isotopic lapse rates have also been observedin the Andes (36) For the Eocene this unusualisotopic lapse rate precludes use of any of thepreviously developed d18Ow-elevation relation-ships for the Eocene TP

Discussion and comparison with data

For the purpose of a model-data comparisonwe derived virtual d18Oc values using simulatedd18Ow and surface temperatures and comparedthese derived values with measured d18Oc from

previous studies (Table 1 and Figs 5 A and Band 6) The d18Oc data show a reasonable agree-ment with the simulated south-north isotopicgradient in d18Oc for the Eocene experiments(Figs 5 A and B and 6) However simulationswith relatively low Asian topography (EOC-Sand EOC-M) are a better fit according to aranking of the model simulations based on thesum of squared residuals (Fig 5C) The largeuncertainties in the d18Oc data preclude anyprecise paleoaltimetry estimates but do allow usto discriminate between different paleoelevationscenarios In simulations with higher TP to-pography d18Oc data are systematically morenegative than simulated d18Oc values (Figs 5Aand 6) A decomposition of the effects on d18Oc

values shows that this discrepancy is mainly dueto the decrease in temperature over the TP aselevation increases because of adiabatic cooling(figs S7 to S9) Lower carbonate formation tem-peratures result in less negative d18Oc values overthe TP This effect is enhanced by more intenserainfall over the Indian foreland associated withmore depleted d18Ow values (figs S2 and S3) asTP elevation increases These two mechanismsreinforce the positive isotopic lapse rate at thesouthern edge of the TP resulting in increasedd18Oc with increasing elevationDespite uncertainties in our modeling ap-

proach (Materials andmethods) our simulationshighlight (i) the permanence of a latitudinald18Ow dipole in agreement with conclusionsfrom recent studies (37 38) and (ii) the limitof standard atmospheric distillation modelsbased on modern calibrations We show thatmost of the assumptions that underlie stableisotope paleoaltimetrymdashsuch as commonly usedmodern temperature lapse rates single and in-

variant moisture sources and similar convec-tive precipitation activitymdashare not valid for theEocene TP By contrast our simulations suggestthat previously published d18Oc are more com-patible with modest Eocene TP elevations (nothigher than 3000 m) These estimates reconcilestable isotopic paleoaltimetry estimates with al-ternative paleontological (13 14) and paleobo-tanical (11 12) data Our results imply thatsubstantial uplift (gt2000 m) of at least someparts of the TP could have occurred after theEocene The complexity of atmospheric pro-cesses within greenhouse climates combinedwith differing paleogeographies highlight thenecessity of using dedicated climate simulationsfor interpreting d18Oc data

Materials and methodsNumerical modeling approach

In order to assess spatial and temporal varia-tions of climatic parameters and the isotopiccomposition of vapor and precipitation we usethe LMDZ-iso Atmospheric General Circulation

Botsyun et al Science 363 eaaq1436 (2019) 1 March 2019 4 of 9

Fig 3 Back trajectory analysis for present-day (CTR) and Eocene winds (A to F) Theprobability distribution of air masses along back trajectories for each of the six points for [(A) to (C)]CTR and [(D) to (F)] EOC-L Back trajectories are calculated daily from May to September at900 hPa level backward for 6 days each starting from the black circle

Fig 4 Hypothesized mechanisms that drivedifferences between Eocene and present-day d18Ow (A and B) The ratio betweenconvective precipitation and large-scale precip-itation (CPLSP blue lines) and the ratiobetween evaporation and precipitation (EP redlines) along south-north TP cross sections for(A) CTR and (B) Eocene experiments In (B)dashed lines indicate CPLSP and EP for theEOC-XL experiment and the solid line indicatesthe same for the EOC-L experiment Stratiformprecipitation is mimicked in the model withlarge-scale precipitation Gray shading indicatesthe topography for the correspondingsimulation Profiles are averaged between 85degEand 90degE for the CTR case and between 73degEand 78degE for the Eocene cases

RESEARCH | RESEARCH ARTICLEon M

arch 11 2020

httpsciencesciencemagorg

Dow

nloaded from

model (AGCM) (39) a derivative of LMDZ(40) LMDZ was developed at Laboratoire deMeacuteteacuteorologie Dynamique (LMD) Paris Franceand is the atmospheric component of the In-

stitute Pierre SimonLaplace (IPSL) Earth SystemModel (ESM) (40) which participates in theCoupled Model Intercomparison Project (CMIPs)program (41) LMDZ incorporates many processes

decomposed into a dynamical core calculatingthe numerical solutions of general equations ofatmospheric dynamics and a physical part cal-culating the details of the climate in each grid

Botsyun et al Science 363 eaaq1436 (2019) 1 March 2019 5 of 9

Table 1 Compilation of measured Eocene d18Oc from published materials Corresponding sample sites geographic locations are provided in fig S6(8 10 13 24 57ndash66) d18Oc are reported relative to V-PDBWe report the minimum measured values for lacustrinebiotic data and minimum and averaged values

for paleosols Numbers in parentheses are average values for paleosols Dashes indicate absence of corresponding data in the original study For d18Oc dashes are

shown where 1 SD calculation is not applicable PETM PaleocenendashEocene Thermal Maximum Min minimum

Locality Age Carbonate

nature

d18Oc

(V-PDB)

(permil) min

values

d18Oc

(1 SD)

Chosen

d18Oc

Temperature

used in

the original

study (degC)

Elevation

reconstructed

in the original

study (m)

Reference to

isotopic data

Aertashi 40 to 343 Ma ago Undifferentiated

(pedogenic nodule or

sediment matrix)

ndash77 118 Min mdash mdash (60)

Central Myanmar Eocene Gastropods ndash14 mdash Min 34 plusmn 5 mdash (24)

Chake Basin Eocene Pedogenic

nodules

ndash104 (ndash88) 075 Min and

average

27 121 plusmn 700 (61)

Erlian Basin PETM Paleosol ndash126 (ndash93) 107 Min and

average

mdash mdash (62)

Gerze Basin 42 to 34 Ma ago Lacustrine

carbonate

ndash81 11 Min 25 plusmn 10 Close to

sea level

(13)

Hengyang Basin Eocene Paleosol ndash99 (ndash72) 087 Min and

average

mdash mdash (63)

Hoh Xil Basin 375 Ma ago Lacustrine

carbonates

ndash117 035 Min and

average

15 to 30 2000 + 850ndash1050 m (64)

Janggalsay 34 Ma ago Undifferentiated

(pedogenic nodule or

sediment matrix)

ndash65 mdash Min mdash mdash (60)

Lake Mahai 50 to 34 Ma ago Undifferentiated

(pedogenic nodule or

sediment matrix)

ndash111 156 Min mdash mdash (60)

Lenghu Eocene Undifferentiated

(pedogenic nodule or

sediment matrix)

ndash75 mdash Min mdash mdash (60)

Liming Eocene Pedogenic

nodules

ndash144 (ndash115) 145 Min and

average

26 and 21 2650 plusmn 300 (61)

Linzhou Basin

Pana Formation

51 to 48 Ma ago Pedogenic

nodules

ndash162 (ndash137) 136 Min and

average

138 plusmn 8 4000 plusmn 500 (10)

Lulehe Eocene Undifferentiated

(pedogenic nodule or

sediment matrix)

ndash103 mdash Min mdash mdash (60)

Lunpola Basin

Middle Niubao

and Upper Niubao

55 to 375 Ma ago Lacustrine and

pedogenic nodules

ndash155 236 Min and

average

10 plusmn 10 4050 + 510ndash620 (8)

Oytag Eocene Mudstone ndash84 mdash Min 19 (58)

Puska Eocene Undifferentiated

(pedogenic nodule or

sediment matrix)

ndash88 194 Min mdash mdash (60)

Taatsin Gol 36 to 34 Ma ago Paleosol ndash96 (ndash9) 039 Min and

average

25 (65)

Tangra Yum Co ~46 Ma ago Paleosol ndash183 (ndash127) 169 Min and

average

15 plusmn 8 2590 (+730ndash910) (66)

Xiao Qaidam Eocene Undifferentiated

(pedogenic nodule or

sediment matrix)

ndash105 085 Min mdash mdash (60)

Xorkol Eocene Undifferentiated

(pedogenic nodule or

sediment matrix)

ndash86 mdash Min mdash mdash (60)

RESEARCH | RESEARCH ARTICLEon M

arch 11 2020

httpsciencesciencemagorg

Dow

nloaded from

point and containing parameterizations forprocesses such as the effects of clouds convec-tion and orographyIn the version of LMDZ used in this work the

land surface is represented as a simple bucketmodel and land surface evaporation is calcu-lated as a single flux no distinction is madebetween transpiration bare soil evaporationor evaporation of intercepted water bythe canopy However despite these lim-itations the ability of this model tosimulate atmospheric dynamics and re-produce rainfall patterns over Asia hasbeen demonstrated in numerous studies(20 24 42ndash44)LMDZ-iso represents isotopic processes

associated with all phase changes andis documented in (39) The implementationof water isotopes in the convection schemeis described in (45) Water species are trans-ported and mixed passively by large-scaleadvection (Van Leer advection scheme (46))and various air mass fluxes Equilibriumfractionation coefficients are calculatedafter Merlivat and Nief (47) and Majoube(48) and kinetic effects are taken into ac-count following Merlivat and Jouzel (49)and Jouzel and Merlivat (50) The modeltakes into account the evolution of the com-position of both the rain and surroundingvapor as raindrops re-evaporate (45) How-ever the model does not account for rain-drop size during evaporation as describedin Lee and Fung (51)Here we use a model configuration

with 144 grid points in longitude 142 inlatitude 39 vertical layers and a stretch-able grid (40) that permits greater regionalspatial resolution and yields an averageresolution of ~50 km over central Asia(fig S10) We force LMDZ-iso with realisticsea-surface temperatures (SSTs) simulatedby FOAM (52) and by realistic albedo androughness values obtained with the vege-tation distribution simulated by the bio-physics vegetation model LPJ (fig S11)(53) We choose this approach rather than(i) using LMDZ-iso with pre-industrialboundary conditions because it permitsaccounting for the greenhouse gases con-centration and related changes in SSTs and(ii) use of a fully coupled GCM becausestable isotopes have not been implementedyet in the IPSL ESM and present-day cal-culation capabilities do not allow per-forming fully coupled experiments in highresolution with the isotopes implemen-tation We should note however thatFOAM and similar coarse resolution mod-els (for example MPIOM T31) have beensuccessfully applied in many other paleo-climate studies (54) In order to show thecapability of the FOAM model to simulateclimate we provide here a rough compar-ison between the SST and the field data[see the data compilation in (55)] for theLate Eocene from 38 to 34 Ma (fig S12)

As in similar recent climate model studies (56)we find that our simulation fits some of thedata but not all and that our latitudinal gra-dient remains too steep when compared to dataWe stress that this is a crude comparison aswe include data from a long time interval (LateEocene) during which global climate changessubstantially

Experimental designWe provide experiments depicting both present-day and Eocene climate For the Eocene ex-periments we use the Eocene paleogeographyreconstruction of Licht et al 2014 (24) whichincludes a southward shift of the Himalayanfront and a large Paratethys Sea to the north ofthe TP To understand the role of topography on

Eocene climate and the distribution ofd18Oc we perform a series of experimentsby varying the elevation of the TP TheEocene TP is as high as 3500 m in theinitial reconstruction (EOC-L simulation)(fig S1D) (24) The elevation is increasedto 5000 m in the EOC-XL simulation (figS1E) reduced to 1750 m in the EOC-Msimulation (fig S1C) and lowered to 200min the EOC-S simulation (fig S1B) Addi-tionally we modify the TP geometry to ob-tain steep narrow mountains analogous tothe present-day Himalayas over the south-ern flank of the elevated topography in theEOC-Him experiment (fig S1F) Lastly weperform a series of complementary experi-ments to test different possible Eocenepaleogeographies for the studied area(i) with the TP moved to the south EOC-South experiment (fig S1G) (ii) to the northEOC-North experiment (fig S1H) and(iii) with a water body over northernIndia EOC-sea experiment (fig S1J)For all Eocene simulations we use PCO2

of 1120 ppm which lies within publishedestimates (21) and identical SSTs to iso-late the effect of topography on atmosphericdynamics and d18Ow We perform one ad-ditional experiment with the Eocene paleo-geography with pre-industrial PCO2 (EOC-1xexperiment) in order to estimate its impacton resulting d18OcWe use present-day orbital parameters

which is a standard technique for deep-time paleoclimate modeling studies giventhat we provide snapshots to illustrate theclimatic state during the Eocene periodand do not focus on representing transientclimates due to orbital variations Howeverin some other studies it has been shownthat such variations may have a significantimpact on the hydrological cycle over Asia(24 25) In order to address this issue wehave tested the impact of orbital param-eters variations on the isotopic compositionof precipitation with a series of additionalpaleo-climate experiments (fig S4 B andC EOC-CO and EOC-WO experiments) Inthese experiments we test two extreme or-bital configurations for the Eocene case inorder to produce either (i) ldquowarm australrdquoorbit (or ldquocold borealrdquo orbit) with eccen-tricity 005 obliquity 245 and Earth atperihelion in January (EOC-CO) and (ii) acold austral summer (and thus a ldquowarmborealrdquo summer) with eccentricity 005obliquity 225 and Earth at perihelion inJuly (EOC-WO) Despite some differencesin simulated MJJAS precipitation amounts

Botsyun et al Science 363 eaaq1436 (2019) 1 March 2019 6 of 9

Fig 5 Model-data comparison for the Eocene (A andB) Maps of d18Oc calculated from model-simulated MJJASprecipitation-weighted d18Ow and model MJJAS meantemperatures for (A) EOC-XL and (B) EOC-S experimentsPoints indicate measured d18Oc from published studies(Materials and methods and Table 1) consisting of the mostnegative values for lacustrine carbonates and averagevalues for paleosols (C) The sum of squared residuals (SSR)for the Eocene simulations (y axis) against modeled TPelevation (Materials and methods) The best model-data fit isobtained when SSR is minimized [cases with relatively lowAsian topography (EOC-S and EOC-M)]

RESEARCH | RESEARCH ARTICLEon M

arch 11 2020

httpsciencesciencemagorg

Dow

nloaded from

for these ldquowarmborealrdquoand ldquocoldborealrdquo scenariosd18Op summer [May June July August andSeptember (MJJAS)] patterns appear to be fairlysimilar for the three orbital forcings tested hereFor all cases more negative values of d18Op aresimulated over India and the southern slopesof the HTP while an area between 20degN and40degN is characterized by more positive values ofd18Op This gradient is strongest for the ldquowarmborealrdquo scenario case

Model-data comparison

We compare Eocene precipitation-weighted d18Oc

derived from our simulations with published

Eocene d18Oc data (Table 1 and Figs 5 and 6)(8 10 13 24 57ndash66) Pedogenic carbonates onthe TP commonly form during the summer sea-son (67) thus to derive d18Oc from our simu-lated d18Ow values we use MayndashSeptember meansurface temperatures and use the fractionationfactors of (17)With an objective of adjusting the param-

eters of a model function [y = f(x)] to best fit adata set (where x is d18Oc from published datasets and y is d18Oc simulated by LMDZ-iso) wecalculate the sum of squared residuals (SSR) for20 data points which permits us to define theoptimum model-data fit which is when the SSR

is minimized For EOC-S and EOC-M simula-tions SSR is estimated to be the smallest 658and 160 respectively while EOC-XL and EOC-Him simulations show the highest values 3697and 45363 respectively SSR for the EOC-seaexperiment is comparable with those for theEOC-L experiment and both EOC-North andEOC-south show relatively high SSR similarto the EOC-Him case Note that for SSR cal-culation for the EOC-North and EOC-South wehave adjusted the location of the data points5deg to the south and to the north respectivelyThe difference in the SSR between two dif-ferent elevations scenarios is larger than the

Botsyun et al Science 363 eaaq1436 (2019) 1 March 2019 7 of 9

Fig 6 Maps of d18Oc

calculated frommodel simulatedMJJAS precipitation-weighted d18Ow andmodel MJJAS meantemperatures forEocene cases(A) EOC-L (B) EOC-Him (C) EOC-M(D) EOC-sea(E) EOC-North(F) EOC-South For allEocene maps pointsshow measured d18Oc

from published datasets consisting of themost negative valuesfor lacustrine carbo-nates and averagevalues for paleosols(Table 1)

RESEARCH | RESEARCH ARTICLEon M

arch 11 2020

httpsciencesciencemagorg

Dow

nloaded from

difference attributable to the variability in thed18O dataset (Table 1) Consequently we dismissvariability in the carbonate d18O as excludingthe firm conclusions that the best data-modelmatch is with lower elevation scenarios

Uncertainties and limits of ourmodeling approach

A major uncertainty in this study is due to thereconstruction of the geographical position ofeach sitersquos d18Oc data during Eocene Based onpaleomagnetic data for the Hoh Xil Basin (mod-ern coordinates 3460degN 93degE) (68) we recon-struct its Eocene paleolatitude as approximately21deg N (as a result of northward convergenceof the Hoh Xil Basin with respect to Eurasia(Siberia) since Early EocenendashLate Oligocenetime) Thus we use a constant correction of 13degfor positioning all of the geologic d18O samplingsites over our Eocene paleogeography Suchpaleoaltitude estimates and corrections of about13 plusmn 5deg are in general consistent with the palaeo-latitudes derived from a large number of high-quality palaeomagnetic studies fromeastern Chinato Kyrgyzstan (23)Though GCMs produce realistic Eocene tem-

peratures over Asia and simulate the associatedresponse of d18Ow numerical modeling withGCMs involves several uncertainties includingthe assumption of constant seawater d18O andthe accuracy of the paleogeographic reconstruc-tion The latitudinal position of the Indian sub-continent in the Eocene and the height of otherorogens in Asia as well as their spatial extentremain a point of contention (23 69) and mayimpact the results of our experiments To ad-dress this limitation additional paleogeographicscenarios should be tested in future work Otherpotential limitations of using isotope-enabledGCMs include complexity associated with ourrepresentation of soils evapotranspiration andvegetation However results of our current studypermit a first-order estimate of Eocene d18Ow

and its low sensitivity to topographic change

Decomposing d18Oc differences

Our goal is to understand to what extent d18Ow

changes explain the d18Oc signal over the EoceneTP and what part of this signal depends on thetemperature of carbonate formation First basedon an assumption of equilibrium between waterand calcite we relate d18Oc to d18Ow using therelationship of (17)

Rc = Rw10309 ndash 2998 + 17491000T ndash 3145 (1)

where T is the temperature of carbonate for-mation Rc is the isotopic ratios in carbonateand Rw is the isotopic ratios in paleo surfacewater or in a simpler form

Rc = Rc (Rw T) (2)

In order to understand why Rw varies fromone climatic state to another we decomposeDRc = Rc2 minus Rc1 into contribution from dRw

and dT terms referring to the climatic statesusing subscript 1 and 2 and to their differenceusing the D notation

DRc = DRcdRw + DRcdT + N (3)

where DRcdRw and DRcdT are the contributionsof the part of the Rc signal related to the changeof Rw and temperature accordingly Nonlinearterms of decomposition are gathered into theresidual term NSimilar to the isotopic signal decomposition sug-

gested in (20) contributions could be estimated as

DRcdRw = Rc (Rw2 dT prime) ndash Rc (Rw1 dTprime) (4)

and

DRcdT = Rc (dRwprime T2) ndash Rc (dRwprime T1) (5)

where Rw2 Rw1 T2 and T1 are d18Ow and tem-perature from two climatic states respectivelyand dRwprime dT are centered differences

dRwprime = (Rw1 + Rw2)2 (6)

dT prime = (T1 + T2)2 (7)

From the Eqs 1 3 4 and 5 we have

DRcdRw = (Rw2 ndash Rw1)10309 (8)

DRcdT = 1749 (1000T2 ndash 1000T1) (9)

N = DRc ndash (DRcdRw + DRcdT) (10)

REFERENCES AND NOTES

1 P Molnar P England J Martinod Mantle dynamics uplift ofthe Tibetan Plateau and the Indian monsoon Rev Geophys31 357ndash396 (1993) doi 10102993RG02030

2 M E Raymo W F Ruddiman Tectonic forcing of lateCenozoic climate Nature 359 117ndash122 (1992) doi 101038359117a0

3 M Ingalls D B Rowley B Currie A S Colman Large-scalesubduction of continental crust implied by India-Asia mass-balance calculation Nat Geosci 9 848ndash853 (2016)doi 101038ngeo2806

4 C Wang et al Outward-growth of the Tibetan Plateau duringthe Cenozoic A review Tectonophysics 621 1ndash43 (2014)doi 101016jtecto201401036

5 M A Murphy et al Did the Indo-Asian collision alone createthe Tibetan plateau Geology 25 719ndash722 (1997)doi 1011300091-7613(1997)025lt0719DTIACAgt23CO2

6 P Tapponnier et al Oblique stepwise rise and growth of theTibet plateau Science 294 1671ndash1677 (2001) doi 101126science105978 pmid 11721044

7 D Zheng et al Rapid exhumation at ~ 8 Ma on the LiupanShan thrust fault from apatite fission-track thermochronologyImplications for growth of the northeastern Tibetan Plateaumargin Earth Planet Sci Lett 248 198ndash208 (2006)doi 101016jepsl200605023

8 D B Rowley B S Currie Palaeo-altimetry of the late Eoceneto Miocene Lunpola basin central Tibet Nature 439 677ndash681(2006) doi 101038nature04506 pmid 16467830

9 Q Xu et al Paleogene high elevations in the QiangtangTerrane central Tibetan Plateau Earth Planet Sci Lett 36231ndash42 (2013) doi 101016jepsl201211058

10 L Ding et al The Andean-type Gangdese MountainsPaleoelevation record from the Paleocene-Eocene LinzhouBasin Earth Planet Sci Lett 392 250ndash264 (2014)doi 101016jepsl201401045

11 X Y Song R A Spicer J Yang Y F Yao C-S Li Pollenevidence for an Eocene to Miocene elevation of centralsouthern Tibet predating the rise of the High HimalayaPalaeogeogr Palaeoclimatol Palaeoecol 297 159ndash168 (2010)doi 101016jpalaeo201007025

12 H Wang et al Amber fossils reveal the Early Cenozoicdipterocarp rainforest in central Tibet Palaeoworld 27506ndash513 (2018) doi 101016jpalwor201809006

13 Y Wei et al Low palaeoelevation of the northern Lhasa terraneduring late Eocene Fossil foraminifera and stable isotopeevidence from the Gerze Basin Sci Rep 6 27508 (2016)doi 101038srep27508 pmid 27272610

14 F Wu D Miao M M Chang G Shi N Wang Fossil climbingperch and associated plant megafossils indicate a warm andwet central Tibet during the late Oligocene Sci Rep 7 878(2017) pmid 28408764

15 D B Rowley R T Pierrehumbert B S Currie A new approachto stable isotope-based paleoaltimetry Implications forpaleoaltimetry and paleohypsometry of the High Himalayasince the late Miocene Earth Planet Sci Lett 188 253ndash268(2001) doi 101016S0012-821X(01)00324-7

16 J Quade D O Breecker M Daeumlron J Eiler The paleoaltimetryof Tibet An isotopic perspective Am J Sci 311 77ndash115(2011) doi 10247502201101

17 S T Kim J R OrsquoNeil Equilibrium and nonequilibrium oxygenisotope effects in synthetic carbonates Geochim CosmochimActa 61 3461ndash3475 (1997) doi 101016S0016-7037(97)00169-5

18 C J Poulsen M L Jeffery Climate change imprinting onstable isotopic compositions of high-elevation meteoric watercloaks past surface elevations of major orogens Geology 39595ndash598 (2011) doi 101130G320521

19 M J Winnick C P Chamberlain J K Caves J M WelkerQuantifying the isotopic ldquocontinental effectrdquo Earth PlanetSci Lett 406 123ndash133 (2014) doi 101016jepsl201409005

20 S Botsyun P Sepulchre C Risi Y Donnadieu Impacts ofTibetan Plateau uplift on atmospheric dynamics andassociated precipitation d18O Clim Past 12 1401ndash1420 (2016)doi 105194cp-12-1401-2016

21 E Anagnostou et al Changing atmospheric CO2concentration was the primary driver of early Cenozoicclimate Nature 533 380ndash384 (2016) doi 101038nature17423 pmid 27111509

22 R Bosboom et al Timing cause and impact of the late Eocenestepwise sea retreat from the Tarim Basin (west China)Palaeogeogr Palaeoclimatol Palaeoecol 403 101ndash118 (2014)doi 101016jpalaeo201403035

23 P C Lippert D J J van Hinsbergen G Dupont-Nivet EarlyCretaceous to present latitude of the central proto-TibetanPlateau A paleomagnetic synthesis with implications forCenozoic tectonics paleogeography and climate of AsiaSpec Pap Geol Soc Am 2507 1ndash21 (2014)

24 A Licht et al Asian monsoons in a late Eocene greenhouseworld Nature 513 501ndash506 (2014) doi 101038nature13704pmid 25219854

25 R Zhang et al Changes in Tibetan Plateau latitude as animportant factor for understanding East Asian climate sincethe Eocene A modeling study Earth Planet Sci Lett 484295ndash308 (2018) doi 101016jepsl201712034

26 J E Kutzbach P J Guetter W F Ruddiman W L PrellSensitivity of climate to late Cenozoic uplift in southern Asiaand the American west Numerical experiments J GeophysRes Atmos 94 (D15) 18393ndash18407 (1989) doi 101029JD094iD15p18393

27 A J Broccoli S Manabe The effects of orography onmidlatitude northern hemisphere dry climates J Clim 51181ndash1201 (1992) doi 1011751520-0442(1992)005lt1181TEOOOMgt20CO2

28 X Liu Z Y Yin Sensitivity of East Asian monsoon climate tothe uplift of the Tibetan Plateau Palaeogeogr PalaeoclimatolPalaeoecol 183 223ndash245 (2002) doi 101016S0031-0182(01)00488-6

29 D B Rowley C N Garzione Stable isotope-basedpaleoaltimetry Annu Rev Earth Planet Sci 35 463ndash508(2007) doi 101146annurevearth35031306140155

30 C N Garzione J Quade P G DeCelles N B EnglishPredicting paleoelevation of Tibet and the Himalaya from d18Ovs altitude gradients in meteoric water across the NepalHimalaya Earth Planet Sci Lett 183 215ndash229 (2000)doi 101016S0012-821X(00)00252-1

31 J Worden D Noone K Bowman Tropospheric EmissionSpectrometer Science Team and Data contributorsImportance of rain evaporation and continental convection in

Botsyun et al Science 363 eaaq1436 (2019) 1 March 2019 8 of 9

RESEARCH | RESEARCH ARTICLEon M

arch 11 2020

httpsciencesciencemagorg

Dow

nloaded from

the tropical water cycle Nature 445 528ndash532 (2007)doi 101038nature05508 pmid 17268467

32 P K Aggarwal et al Proportions of convective and stratiformprecipitation revealed in water isotope ratios Nat Geosci 9624ndash629 (2016) doi 101038ngeo2739

33 N Kurita et al Intraseasonal isotopic variation associated withthe Madden-Julian Oscillation J Geophys Res Atmos 116D24101 (2011) doi 1010292010JD015209

34 O A Tuinenburg et al J Geophys Res Atmos 1010022015JD023461 (2015) doi 1010022015JD023461

35 N E Levin E J Zipser T E Cerling Isotopic composition ofwaters from Ethiopia and Kenya Insights into moisturesources for eastern Africa J Geophys Res Atmos 114D23306 (2009) doi 1010292009JD012166

36 A Rohrmann et al Can stable isotopes ride out thestorms The role of convection for water isotopes in modelsrecords and paleoaltimetry studies in the central AndesEarth Planet Sci Lett 407 187ndash195 (2014) doi 101016jepsl201409021

37 J K Caves et al Role of the westerlies in Central Asia climateover the Cenozoic Earth Planet Sci Lett 428 33ndash43 (2015)doi 101016jepsl201507023

38 A Licht et al Resilience of the Asian atmosphericcirculation shown by Paleogene dust provenanceNat Commun 7 12390 (2016) doi 101038ncomms12390pmid 27488503

39 C Risi S Bony F Vimeux J Jouzel Water-stable isotopes inthe LMDZ4 general circulation model Model evaluation forpresent-day and past climates and applications to climaticinterpretations of tropical isotopic records J Geophys Res Atmos115 (D24) 1ndash27 (2010) doi 1010292010JD015242

40 F Hourdin et al The LMDZ4 general circulation model Climateperformance and sensitivity to parametrized physics withemphasis on tropical convection Clim Dyn 27 787ndash813(2006) doi 101007s00382-006-0158-0

41 K E Taylor R J Stouffer G A Meehl An overview of CMIP5and the experiment design Bull Am Meteorol Soc 93485ndash498 (2012) doi 101175BAMS-D-11-000941

42 J E Lee et al Asian monsoon hydrometeorology from TESand SCIAMACHY water vapor isotope measurements andLMDZ simulations Implications for speleothem climate recordinterpretation J Geophys Res Atmos 117 1ndash12 (2012)doi 1010292011JD017133

43 R Krishnan et al Deciphering the desiccation trend of theSouth Asian monsoon hydroclimate in a warming worldClim Dyn 101007s00382-015-2886-5 (2015) doi 101007s00382-015-2886-5

44 J-B Ladant Y Donnadieu V Lefebvre C Dumas Therespective role of atmospheric carbon dioxide and orbitalparameters on ice sheet evolution at the Eocene-Oligocenetransition Paleoceanography 29 810ndash823 (2014)doi 1010022013PA002593

45 S Bony C Risi F Vimeux Influence of convective processeson the isotopic composition (d18O and dD) of precipitation andwater vapor in the tropics 1 Radiative-convective equilibriumand Tropical Ocean-Global-Atmosphere-Coupled Ocean-Atmosphere Responce Experiment (TOGA-COARE)J Geophys Res Atmos 113 D19305 (2008) doi 1010292008JD009942

46 B Van Leer Towards the ultimate conservative differencescheme IV A new approach to numerical convectionJ Comput Phys 23 276ndash299 (1977) doi 1010160021-9991(77)90095-X

47 L Merlivat G Nief Fractionnement isotopique lors deschangements drsquoeacutetat solide-vapeur et liquide-vapeur de lrsquoeau agravedes tempeacuteratures infeacuterieures agrave 0degC Tellus 19 122ndash127 (1967)doi 103402tellusav19i19756

48 M Majoube Fractionnement en oxygene-18 et en deuteriumentre lrsquoeau et sa vapeur J Chim Phys 68 1423ndash1436 (1971)doi 101051jcp1971681423

49 L Merlivat J Jouzel Global climatic interpretation of thedeuterium-oxygen 18 relationship for precipitation J GeophysRes Oceans 84 (C8) 5029ndash5033 (1979) doi 101029JC084iC08p05029

50 J Jouzel L Merlivat Deuterium and oxygen 18 in precipitationModeling of the isotopic effects during snow formationJ Geophys Res Atmos 89 (D7) 11749ndash11757 (1984)doi 101029JD089iD07p11749

51 J E Lee I Fung ldquoAmount effectrdquo of water isotopes andquantitative analysis of post-condensation processesHydrol Processes 22 1ndash8 (2008) doi 101002hyp6637

52 R L Jacob Low frequency variability in a simulatedatmosphere ocean system thesis University of WisconsinMadison WI (1997)

53 S Sitch et al Evaluation of ecosystem dynamics plantgeography and terrestrial carbon cycling in the LPJ dynamicglobal vegetation model Glob Change Biol 9 161ndash185 (2003)doi 101046j1365-2486200300569x

54 Y Donnadieu E Puceacuteat M Moiroud F GuillocheauJ-F Deconinck A better-ventilated ocean triggered by LateCretaceous changes in continental configurationNat Commun 7 10316 (2016) doi 101038ncomms10316pmid 26777897

55 M Baatsen et al Equilibrium state and sensitivity of thesimulated middle-to-late Eocene climate Clim Past Discuss1ndash49 (2018) doi 105194cp-2018-43

56 D K Hutchinson et al Climate sensitivity and meridionaloverturning circulation in the late Eocene using GFDL CM21Clim Past 14 789ndash810 (2018) doi 105194cp-14-789-2018

57 M T Hren B Bookhagen P M Blisniuk A L BoothC P Chamberlain d18O and dD of streamwatersacross the Himalaya and Tibetan Plateau Implications formoisture sources and paleoelevation reconstructionsEarth Planet Sci Lett 288 20ndash32 (2009) doi 101016jepsl200908041

58 J Bershaw S M Penny C N Garzione Stable isotopes ofmodern water across the Himalaya and eastern TibetanPlateau Implications for estimates of paleoelevation andpaleoclimate J Geophys Res Atmos 117 1ndash18 (2012)doi 1010292011JD016132

59 J Bershaw C N Garzione L Schoenbohm G Gehrels L TaoCenozoic evolution of the Pamir plateau based on stratigraphyzircon provenance and stable isotopes of foreland basinsediments at Oytag (Wuyitake) in the Tarim Basin (westChina) J Asian Earth Sci 44 136ndash148 (2012) doi 101016jjseaes201104020

60 M L Kent-Corson et al Stable isotopic constraints on thetectonic topographic and climatic evolution of the northernmargin of the Tibetan Plateau Earth Planet Sci Lett 282158ndash166 (2009) doi 101016jepsl200903011

61 G D Hoke J Liu-Zeng M T Hren G K WissinkC N Garzione Stable isotopes reveal high southeast TibetanPlateau margin since the Paleogene Earth Planet Sci Lett394 270ndash278 (2014) doi 101016jepsl201403007

62 G J Bowen P L Koch J Meng J Ye S Ting Age andCorrelation of Fossiliferous Late PaleocenendashEarly Eocene

Strata of the Erlian Basin Inner Mongolia China Am MusNovit 3474 1 (2005) doi 1012060003-0082(2005)474[0001AACOFL]20CO2

63 S Ting et al Biostratigraphic chemostratigraphic andmagnetostratigraphic study across the Paleocene-Eoceneboundary in the Hengyang Basin Hunan China Geol Soc AmSpacial Pa 521ndash535 (2003)

64 A J Cyr B S Currie D B Rowley Geochemical evaluationof Fenghuoshan group lacustrine carbonates north-centralTibet Implications for the paleoaltimetry of the EoceneTibetan Plateau J Geol 113 517ndash533 (2005) doi 101086431907

65 J K Caves D J Sjostrom H T Mix M J WinnickC P Chamberlain Aridification of Central Asia and uplift of theAltai and Hangay Mountains Mongolia Stable isotopeevidence Am J Sci 314 1171ndash1201 (2014) doi 10247508201401

66 Q Xu L Ding R Hetzel Y Yue E F Rades Low elevationof the northern Lhasa terrane in the Eocene Implicationsfor relief development in south Tibet Terra 27 458ndash466(2015)

67 J Quade J Eiler M Daeron H Achyuthan The clumpedisotope geothermometer in soil and paleosol carbonateGeochim Cosmochim Acta 105 92ndash107 (2013) doi 101016jgca201211031

68 Z Liu X Zhao C Wang S Liu H Yi Magnetostratigraphyof Tertiary sediments from the Hoh Xil Basin Implicationsfor the Cenozoic tectonic history of the Tibetan PlateauGeophys J Int 154 233ndash252 (2003) doi 101046j1365-246X200301986x

69 Y Najman et al Timing of India-Asia collision Geologicalbiostratigraphic and palaeomagnetic constraints J GeophysRes Solid Earth 115 B12416 (2010) doi 1010292010JB007673

70 GNIP IAEAWMO Global Network of Isotopes in PrecipitationThe GNIP Database wwwiaeaorgservicesnetworksgnip(2016)

ACKNOWLEDGMENTS

Funding This work was funded by the FP7-PEOPLE Marie-CurieActions [grant 316966 iTECC (interaction Tectonics-Erosion-Climate-Coupling) project] This work was granted access tothe HPC resources of IDRIS under the allocation 2017-0292made by GENCI SB was also supported by the GermanMinistry for Education and Research (BMBF) grant 03T0863Ato T Ehlers JKCR is funded by an ETH Fellowship Authorcontributions SB PS and YD developed the conceptualidea SB designed and carried out simulations in closecollaboration with PS CR and YD JKCR and SB didstatistical analysis SB AL and JKCR analyzed observedcarbonate data All authors discussed the results andimplications and commented on the manuscript Competinginterests None declared Data and materials availabilityAll data are available in the manuscript or in thesupplementary materials

SUPPLEMENTARY MATERIALS

wwwsciencemagorgcontent3636430eaaq1436supplDC1Figs S1 to S12

5 October 2017 resubmitted 6 September 2018Accepted 22 January 2019101126scienceaaq1436

Botsyun et al Science 363 eaaq1436 (2019) 1 March 2019 9 of 9

RESEARCH | RESEARCH ARTICLEon M

arch 11 2020

httpsciencesciencemagorg

Dow

nloaded from

Revised paleoaltimetry data show low Tibetan Plateau elevation during the EoceneSvetlana Botsyun Pierre Sepulchre Yannick Donnadieu Camille Risi Alexis Licht and Jeremy K Caves Rugenstein

DOI 101126scienceaaq1436 (6430) eaaq1436363Science

this issue p eaaq1436 see also p 928Sciencefor a Tibetan Plateau that was about 1000 meters lower than it is todayPerspective by van Hinsbergen and Boschman) The results harmonize the isotopic record with other proxies and arguea model to show that several previously overlooked factors contribute to the isotopic record from the Eocene (see the

usedet al40 million years ago) other lines of evidence suggest a lower elevation in the distant past Botsyun simEocene (patterns Although some isotope proxies have suggested a roughly equivalent height for the plateau as far back as the

The elevation of the Tibetan Plateau has a major impact on climate affecting the monsoons and regional weatherAncient height of the Tibetan Plateau

ARTICLE TOOLS httpsciencesciencemagorgcontent3636430eaaq1436

MATERIALSSUPPLEMENTARY httpsciencesciencemagorgcontentsuppl201902273636430eaaq1436DC1

CONTENTRELATED httpsciencesciencemagorgcontentsci3636430928full

REFERENCES

httpsciencesciencemagorgcontent3636430eaaq1436BIBLThis article cites 67 articles 5 of which you can access for free

PERMISSIONS httpwwwsciencemagorghelpreprints-and-permissions

Terms of ServiceUse of this article is subject to the

is a registered trademark of AAASScienceScience 1200 New York Avenue NW Washington DC 20005 The title (print ISSN 0036-8075 online ISSN 1095-9203) is published by the American Association for the Advancement ofScience

Science No claim to original US Government WorksCopyright copy 2019 The Authors some rights reserved exclusive licensee American Association for the Advancement of

on March 11 2020

httpsciencesciencem

agorgD

ownloaded from