CENOZOIC CLIMATE OXYGEN ISOTOPE EVIDENCE · CENOZOIC CLIMATE ^ OXYGEN ISOTOPE EVIDENCE Jame’s D....

CENOZOIC CLIMATE ^ OXYGEN ISOTOPEEVIDENCE

Jame’s D. Wright, Rutgers University,Department of Geology, 610 Taylor Road,Piscataway, NJ08854, USA.

Copyright ^ 2001 Academic Press

doi:10.1006/rwos.2001.0252

Discoveries of fossil remains of 50 million year oldalligators on Ellesmere Island and 30}40 millionyear-old forests on Antarctica contrast sharply withour present vision of polar climates. These are notisolated discoveries or quirks of nature. An ever-growing body of faunal, Soral, and geochemicalevidence shows that the Rrst half of the CenozoicEra was much warmer than the present time. Whatmaintained such a warm climate and could it be ananalog for future global warming? To address theseand other questions, one needs more than a qualitat-ive estimate of planetary temperatures. Quantitativetemperature estimates (both magnitudes and rates ofchange) are required to depict how the Earth’s cli-mate has changed through time. One of the mostpowerful tools used to reconstruct past climates dur-ing the Cenozoic (the last 65 million years of Earth’shistory) is the analysis of oxygen isotopes in thefossil shells of marine organisms. The calcium car-bonate shells of the protist foraminifera are themost often analyzed organisms because the differentspecies are distributed throughout surface(planktonic) and deep (benthic) marine environ-ments.

Oxygen Isotope Systematics

The stable isotopes of oxygen used in paleo-oceano-graphic reconstructions are 16O and 18O. There areabout 500 16O atoms for every 18O atom in theocean/atmosphere environment. During the 1940s,Harold Urey at the University of Chicago predictedthat the 18O/16O ratio in calcite (CaCO3) shouldvary as a function of the temperature at which themineral precipitated. His prediction spurred on ex-periments by himself and others at the University ofChicago who measured 18O/16O ratios in CaCO3

precipitated in a wide range of temperatures, lead-ing to the use of stable oxygen isotope measure-ments as a paleo-thermometer.

To determine oxygen isotopic ratios, unknown18O/16O ratios are compared to the known 18O/16Oratio of a standard. The resulting values are ex-pressed in delta notation, d18O, where:

d18O"

18O/16Osample!18O/16Ostandard

18O/16Ostandard]1000 [1]

Carbonate samples are reacted in phosphoric acid toproduce CO2. To analyze water samples, CO2 gas isequilibrated with water samples at a constant tem-perature. Given time, the CO2 will isotopically equi-libriate with the water. For both the carbonate andwater samples, the isotopic composition of CO2 gasis compared with CO2 gas of known isotopic com-position. There are two standards for reporting d18Ovalues. For carbonate samples, the reference stan-dard is PDB, which was a crushed belemnite shell(Belemnitella americana) from the Peedee formationof Cretaceous age in South Carolina. The originalPDB material has been exhausted, but other stan-dards have been calibrated to PDB and are used asan intermediate reference standard through whicha PDB value can be calculated. For measuring theisotopic composition of water samples, StandardMean Ocean Water (SMOW) is used as the refer-ence. The SMOW reference was developed so thatits d18Owater value is 0.0& (parts per thousand) andapproximates the average oxygen isotopic composi-tion of the whole ocean. Deep ocean d18Owater valuesare close to the SMOW value, ranging from !0.2to 0.2&. In contrast, surface ocean d18Owater valuesexhibit a much greater variability, varying between!0.5 and #1.5&.

Oxygen Isotope Paleo-thermometry

Early studies into the natural variations in oxygenisotopes led to the development of a paleotempera-ture equation. The temperature during the precipita-tion of calcite can be estimated by measuring thed18O value in calcite-secreting organisms (foraminif-era, corals, and mollusks) and the value of the waterin which the organisms live. The various paleotem-perature equations all follow the original proposedby Sam Epstein and his colleagues (University ofChicago):

T"16.5!4.3](d18Ocalcite!d18Owater)

0001

0002

0003

0004

CENOZOIC CLIMATE ^ OXYGEN ISOTOPE EVIDENCE 1

Editor: Shivakumar Opertor: Padmini Scan: Jane RWOS 0252

#0.14](d18Ocalcite!d18Owater)2 [2]

where T and d18Owater are the temperature (3C) andoxygen isotope value of the water in which theorganism lived1, and d18Ocalcite is the oxygen isotopevalue of calcite measured in the mass spectrometer.

Eqs. [2] shows that the changes in d18Ocalcite area function of the water temperature and d18Owater

value. A one-to-one relationship between d18Ocalcite

and d18Owater values dictates that a change in thed18Owater term will cause a similar change in themeasured d18Ocalcite value. However, an inverse rela-tionship between d18Ocalcite and T changes dictatesthat for every 13C increase in temperature, there isa 0.23& decrease in the measured d18Ocalcite value.These relationships enable us to interpret d18Ocalcite

changes generated from foraminifera, corals, andmollusks. For many years, the convention was toplot d18Ocalcite values with the axis reversed (highervalues to the left or bottom) so that d18O recordsreSect climate changes (e.g., colder to the left orbottom). More recently, there has been a trendamong some scientists to plot d18Ocalcite values with-out reversing the axis.

The paleotemperature equation contains two un-knowns (temperature, d18Owater). Although temper-ature is the main target in reconstructions, onecannot ignore the d18Owater term. In the modernocean, the equator-to-pole gradient measured inplanktonic foraminifera d18Ocalcite values is &5.0&and largely reSects the temperature gradient(&283C). However, if temperature were the soleinSuence on modern d18Ocalcite values, the equator-to-pole gradient would be &6.5&(283C]0.23&/3C). The attenuated d18Ocalcite gradi-ent measured in planktonic foraminifera reSects thesurface ocean d18Owater variability. Therefore, a keyto using d18Ocalcite records as indicators of past cli-mates is to understand the hydrographic parametersthat produce the modern d18Ocalcite gradient. Forinstance, ignoring the d18Owater term results ina 5}63C underestimation compared to the observedtemperature gradient. This occurs largely becausetropical temperature estimates will be too cold(&43C) whereas polar estimates will be warm(&1}23C).

d18O Variation in the NaturalEnvironment

d18Owater values in the ocean/atmosphere systemvary both spatially and temporally because frac-tionation between the H2

18O and H216O molecules

is temperature-dependent in the hydrologic

cycle and follows the Rayleigh Distillation model(Figure 1). In general, water vapor evaporates atlow latitudes and precipitates at higher latitudes.Fractionation during evaporation concentrates thelighter H2

16O molecule in the water vapor, leavingthe water enriched in H2

18O. On average, thed18Owater value of water vapor is 9& lower than itssource water (Figure 1). Fractionation during con-densation concentrates the H2

18O molecules in theprecipitation (rain/snow) by &9&. Therefore, if allof the water evaporated in the tropics rained backinto the tropical oceans, there would be no netchange in the d18Owater term. However, some watervapor is transported to higher latitudes. If the cloudsremain a closed system (i.e., mid-to-high latitudeevaporation does not inSuence the d18Owater value inthe clouds2), then precipitation will further depletethe clouds (water vapor) in H2

18O relative to H216O.

Consequently, the d18O value of water vapor de-creases from the original value as water vapor con-denses into precipitation (Figure 1) and the cloudthat formed from the evaporation in the tropics willeventually lose moisture, fractionating the d18Owater

value of the remaining water vapor (Figures 1 and2). By the time 50% of the initial moisture precipi-tates, the d18O value of the water vapor will be&!15%, while precipitation will be &!6&.Once the cloud reaches the poles, over 90% of theinitial water vapor will have been lost, producingd18O values of snow less than !20&. Snow at theSouth Pole approaches !60&. There is a strongrelationship between d18O values in precipitationand air mass temperatures because air temperaturedictates how much water vapor it can hold, and thed18O values of the precipitation is a function of theamount of water remaining in the clouds (Figure 3).

Spatial Variations in d18Owater ofModern Sea Water

The evaporation/precipitation process that deter-mines the d18Owater values of precipitation (e.g., Fig-ure 1) also controls the d18Owater values in regions inthe ocean. At any one time, the volume of waterbeing transported through the hydrologic cycle (e.g.,atmosphere, lakes, rivers, and groundwater) is smallcompared to the volume of water in the oceans(1:130). Therefore, the hydrologic cycle can inSu-ence the whole ocean d18Owater value only by cre-ating a new or enlarging an existing reservoir (e.g.,glacier/ice sheets). In contrast, evaporation/precipi-tation processes will change the d18Owater and salin-ity values in the surface waters because only the thinsurface layer of the ocean communicates with the

0005

0006

0007

0008

2 CENOZOIC CLIMATE ^ OXYGEN ISOTOPE EVIDENCE


atmosphere. As noted above, the process of evapor-ation enriches surface water in H2

18O molecules andsalt because the water vapor is enriched in H2

16Omolecules. For this reason, high salinity sea waterhas a high d18Owater value and vice versa. MorespeciRcally, tropical and subtropical surface waterd18Owater values are &1& higher than mean oceanwater values (Figure 4). Interestingly, subtropicald18Owater values are higher than tropical values eventhough evaporation is higher in the tropics. Atmo-spheric circulation patterns produce intense rainfallin the tropics to offset some of the evaporation,whereas very little rain falls in the subtropical re-gions. Because evaporation minus precipitation(E!P) is greater in the subtropics, these regionshave higher salinity and d18Owater values. In contrast,subpolar and polar regions have greater precipita-tion than evaporation; hence, high-latitude surfacewaters have low salinity and d18Owater values thatapproach !0.5& (Figure 4).

Temporal Variations

Variations in the amount of water stored on landthrough time, usually in the form of ice, can havea signiRcant effect on the mean ocean d18Owater

value, and hence, the marine d18Ocalcite record. Atpresent, high-latitude precipitation returns to theoceans through summer ice/snow melting. Duringglacial periods, snow and ice accumulate into largeice sheets. Because the difference in ice sheet andmean ocean values is large (d18Oice"!35 to!40& vs. d18Owater mean ocean"&0&), icesheet Suctuations are reSected in mean oceanicd18Owater values. This relationship can be illustratedby examining how the mean ocean d18Owater valueincreased during the last glacial maximum (LGM)relative to the present (Figure 5). During the LGM,water stored in continental ice lowered global sealevel by 120 m, removing &3% of the ocean’s vol-ume. Thus, the mean ocean d18Owater value increasedby 1.2& during the LGM relative to the present(Figure 5).

Pleistocene Oxygen IsotopeVariations

The Rrst systematic downcore examination of themarine stable isotope record was made by CesareHEmiliani during the 1950s on d18Ocalcite records gen-erated from planktonic foraminifera in Caribbeandeep-sea cores. Emiliani recognized the cyclic pat-tern of low and high d18Ocalcite values and concludedthat these represented glacial}interglacial intervals.

Emiliani identiRed the seven most recent climatecycles and estimated that they spanned the last280 000 years. Current age estimates indicate thatthe duration of the cycles is approximately 525 000years. To apply the paleotemperature equation tothese records, Emiliani estimated that ice sheet-induced ocean d18Owater variability was relativelysmall, 0.3&. As showed above, the maximum gla-cial}interglacial ice sheet signal was closer to 1.2&.Therefore, most of the d18Ocalcite variability betweenglacial and interglacial intervals represented temper-ature changes of 5}103C. Emiliani divided thed18Ocalcite record into warm stages (designated withodd numbers counting down from the Holocene)and cold stages (even numbers). Hence, ‘IsotopeStage 1’ refers to the present interglacial intervaland ‘Isotope Stage 2’ refers to the LGM (Figure 6).During the 1960 s and 1970s, many argued thatmost of the glacial to interglacial difference ind18Ocalcite values resulted from ice volume changes.Nicholas Shackleton of Cambridge University madethe key observation that benthic foraminiferal d18Ovalues show a glacial to interglacial difference of&1.8&. If the ice volume contribution was only0.3& as argued by Emiliani, then the deep oceantemperatures would have been 6}73C colder thanthe present temperatures of 0}33C. Sea water freezesat !1.83C, precluding Emiliani’s ‘low’ ice volumeestimate. By the early 1970s, numerous d18O recordshad been generated and showed a cyclic variationthrough the Pleistocene and into the late Pliocene.One hundred oxygen isotope stages, representing 50glacial}interglacial cycles, have been identiRed forthe interval since 2.6 million years ago (Ma) (Figure6).

Cenozoic d18O Records

The Rrst Cenozoic d18O syntheses based onforaminiferal d18O records were produced duringthe mid-1970s. Nicholas Shackleton and JamesKennett produced a composite benthic d18O recordfor the Cenozoic from cores to the south of Austra-lia. A second group led by Samuel Savin generatedlow-latitude planktonic and benthic foraminiferald18O syntheses. Both records are important to un-derstanding Cenzoic climate changes. Benthicforaminiferal records best reSect global temperatureand ice volume changes. Additional advantages ofthe benthic foraminiferal composite include:(1) deep-ocean temperatures are more uniform withrespect to horizontal and vertical gradients;(2) deep-ocean d18Owater values are less variablecompared to the large surface water changes; (3) thedeep ocean approximates high-latitude surface

0009

0010

0011



water conditions where deep waters originated dur-ing the Cenozoic (i.e., Antarctica, northern NorthAtlantic); and (4) many benthic foraminifera taxaare long-lived so that one species can be used toconstruct records spanning several millions of yearsin contrast to planktonic taxa which have shorterdurations and require records to be spliced togetherfrom several species.

Low-latitude planktonic foraminiferal d18O re-cords are good proxies for tropical sea surface tem-peratures. Tropical temperatures are an importantcomponent of the climate system because they inSu-ence evaporation, and hence, total moisture in theatmosphere. Planktonic and benthic foraminiferald18O comparison allow one to assess equator-to-pole as well as vertical temperature gradients duringthe Cenozoic, and thus, to determine planetary tem-perature changes. Finally, much of the climaticchange in the last 65 million years has been ascribedto poleward heat transport or greenhouse gas Suctu-ations. General circulation models indicate that eachmechanism should produce different temperaturepatterns that can be approximated with the plank-tonic and benthic d18O records.

The Rrst benthic d18O syntheses generated, as wellas more recent compilations, show the same long-term patterns. After the Cretaceous}Tertiary (K/T)boundary events, deep-water d18O values remainedrelatively constant for the Rrst 7 million years ofthe Paleocene (Figure 7A). At 58 Ma, benthicforaminiferal d18O values began a decrease over thenext 6 myrs that culminated during the early Eocenewith the lowest recorded value (!0.5&) of theCenozoic. Following this minimum at 52 Ma, d18Ovalues increased by 5.5&, recording maximumvalues (&5&) during the glacial intervals of Pleis-tocene (Figure 6). The Rrst part of this long-termchange was a gradual increase of 2& through theend of the Eocene (52}34 Ma). The remainder of theincrease was accomplished through large steps atthe Eocene/Oligocene boundary (&33.5 Ma), dur-ing the middle Miocene (ca. 15}13Ma) and latePliocene (ca. 3.2}2.6 Ma). After 2.6 Ma, the ampli-tude of the high-frequency signal increasedto'1&, reaching 1.8& over the past 800 thousandyears.

Planktonic and benthic foraminiferal d18O valuesco-varied in general during the early Cenozoic(6.5}34Ma). Values averaged about !1& between65 and 58 Ma, before decreasing to !2.5& duringthe early Eocene, recording the lowest values of theCenozoic (Figure 7A). From 52 to 33 Ma, plank-tonic foraminiferal values increased by 2&. In spiteof a break in the latest Eocene record, it appearsthat the tropical ocean differed from the deep ocean

across the Eocene/Oligocene boundary. Formuch of the Oligocene (&33}25 Ma), planktonicforaminiferal d18O values remained unusually high,averaging !0.5&. Beginning around theOligocene/Miocene boundary (&25 Ma), plank-tonic foraminiferal d18O values began a long-termdecrease, culminating in the Pleistocene with aver-age values of !1.5&. In contrast, the benthic d18Orecord permanently changed during the middleMiocene d18O shift and late Pliocene increase.

Apportioning the d18O changes recorded by thebenthic and planktonic foraminifera between tem-perature and ice volume changes requires knowledgeof, or reasonable estimates for, one of theseparameters. One promising tool that may help dis-criminate between each effect is the Mg/Ca ratiomeasured in benthic foraminifera. Initial studies us-ing Mg/Ca ratios conRrmed the long-term temper-ature changes during the Cenozoic calculated usingthe d18O record and other climate proxies. If veriR-ed, this record implies that small ice sheets grewduring the middle and late Eocene and Suctuated insize throughout the Oligocene to Miocene. At pres-ent, the Mg/Ca record lacks the resolution for keyintervals and still requires veriRcation of interspeciesoffsets before it can be applied unequivocally toisolate the ice volume-induced d18Owater componentin the foraminiferal d18O records. For the discussionthat follows, glacialogical evidence is used to esti-mate the ice volume/d18Owater variations.

The Greenhouse World

The oldest unequivocal evidence for ice sheets onAntarctica, ice-rafted detritus (IRD) deposited byicebergs in the ocean, places the Rrst large ice sheetin the earliest Oligocene. Thus, it is reasonable toassume that ice sheets were small to absent and thatsurface and deep-water temperature changes con-trolled much if not all the d18O change prior to34 Ma. The modern Antarctic and Greenland icesheets lock up &2% of the total water in theworld’s ocean. If melted, these ice sheets would raiseglobal sea level by &70}75 m and mean oceand18Owater value would decrease to !0.9& PDB (seeabove). One can then apply eqn [2] to the benthicand planktonic foraminiferal d18O records to esti-mate deep- and surface-ocean temperatures for theRrst half of the Cenozoic (ca. 65}34 Ma).

During the early to middle Paleocene, deep-watertemperatures remained close to 103C (Figure 7B).The 1& decrease between 58 and 52 Ma translatesinto a deep-water warming of 43C, reaching a highof 143C. This is in sharp contrast to the moderndeep-water temperatures, which range between

0012

0013

0014

0015

0016

0017



0 and 33C. Following the peak warmth at 52 Ma,the 2& increase in benthic foraminiferal values indi-cates that the deep waters cooled by 73C and were73C by the end of the Eocene. If small ice sheetsexisted during the Paleocene and Eocene, then tem-perature estimates would be on the order of 13Cwarmer than those calculated for the ice-free as-sumption. Some data indicate that smaller ice sheetsmay have existed on the inland parts of Antarcticaduring the late Eocene. However, these were notlarge enough to deposit IRD in the ocean. There-fore, their effect on the d18O values of the ocean wasprobably less than 0.3&.

Tropical surface water temperatures warmedfrom 22 to 243C, based on eqn [2], at the beginningof the Cenozoic to 283C during the early Eocene(52 Ma; Figure 7B). The higher estimate is similar totemperatures in the equatorial regions of the mod-ern oceans. Planktonic foraminiferal d18O values re-corded a long-term increase of by 2& (!2.5 to!0.5&) through the remainder of the Eocene. Justprior to the Eocene/Oligocene boundary, tropicalsurface water temperatures were &213C, endingthe long-term tropical cooling of 73C from 52 to34 Ma.

The Ice House World of theLast 33 Million Years

As mentioned above, southern ocean cores containIRD at and above the Eocene/Oligocene boundary.Widely distributed IRD and glacial tills on parts ofthe Antarctic continental margin representing theOligocene to Recent mark the onset of large icesheets. Whether these sediments represent persistentor periodic ice cover is uncertain. At least some icewas present on Antarctica during the Oligocene toearly Miocene. The Antarctic ice sheet has beena Rxture since the middle Miocene (&15 Ma). Ourrecord of Northern Hemisphere ice sheets suggeststhat they were small or nonexistent prior to the latePliocene. For the purpose of estimating surface anddeep temperatures, an ice volume estimate slightlylower than the modern will be applied for the inter-val that spans from the Oligocene into the middleMiocene (33}15Ma). For the interval between 15and 3 Ma, ice volumes were probably similar tothose of today. From 3 Ma, ice volumes rangedbetween the modern and LGM. Using these broadestimates for ice volumes, mean ocean d18Owater

values for those three intervals were !0.5, !0.22,and 0.4& PDB, respectively. The 0.4& estimatereSects the average between the maximum and min-imum conditions during the Plio-Pleistocene. As

noted above, the largest portion of the high fre-quency signal is controlled by ice volume changes.

The benthic foraminiferal d18O increase at theEocene/Oligocene boundary occurred rapidly(&10 000 years; Figure 8A). At the peak of theEocene/Oligocene boundary event, benthicforaminifera recorded d18O values similar to modernvalues. Using the ice volume assumption fromabove, deep-water temperatures approached moderndeep-ocean temperatures (33C). This marks an im-portant transition from the relatively warm oceansof the Paleocene and Eocene to the cold deep watersof the Oligocene to present. This switch to a coldocean where bottom waters formed at near-freezingtemperatures heralded the development of thepsychrosphere. Following the Eocene/Oligoceneboundary, deep-water temperatures began a long-term warming over the next 18 million years(33}15Ma). The coldest deep-water temperatures of33C were recorded at 33 Ma, while temperaturesreached 93C at &25 and &15 Ma (Figure 7B).

There is a gap in the planktonic foraminiferald18O record for the latest Eocene that hampers ourassessment of tropical response during Eocene/Oligocene climate event. However, it is clear fromthe data that do exist that the planktonic responseacross the Eocene/Oligocene boundary differed fromthe benthic response. The planktonic foraminiferald18O values for the early Oligocene are similar tolate Eocene values, whereas the benthic values re-corded a 1.5& increase. Planktonic foraminiferalrecords from other regions that span the Eocene/Oligocene boundary indicate that the surface waterd18O increase was on the order of 0.5&. Thischange is approximately equal to the effect of themodern Antarctic ice sheet. Combined with thephysical evidence, it seems probable that the plank-tonic foraminiferal d18O increase at the Eocene/Oligocene boundary recorded the ice volume inSu-ence with little temperature effect. Therefore, tropi-cal surface temperatures remained around 223Cwhile the deep ocean cooled during this d18O shift.Following the boundary event, planktonicforaminifera d18O record during the Oligocene andearly Miocene mirrored the benthic record in manyrespects. For much of the Oligocene and earlyMiocene, the absolute values are close to !0.5&,which translates into a temperature estimate of213C (Figure 7B). By 15 Ma, tropical surface watershad warmed to 263C.

The middle Miocene d18O shift represents an in-crease of 1.5& in the benthic record between 15and 13 Ma. This transition is composed of twosharp increases around 14 and 13 Ma (Figure 8B).These d18O steps occurred in less than 200 000 years

0018

0019

0020

0021

0022



with each recording an increase of &1& followedby a small decrease. During these two shifts, deepwaters cooled from 9 to 53C. The planktonicforaminiferal d18O record from 15 to 13 Ma showstwo increases as recorded in the benthicforaminiferal record (Figure 7). However, it doesnot show the large permanent shift recorded by thebenthic foraminifera, indicating a small coolingfrom 26 to 243C. From 13 to 3 Ma, the deep oceancooled slightly from 5 to 33C while the surfacewaters warmed from 24 to 263C (Figure 7B).

The last of the large d18O steps in the Cenozoicwas recorded during the late Pliocene from 3.2 to2.6 Ma. This ‘step’ is better characterized as aseries of d18O cycles with increasing amplitudes andvalues over this interval (Figures 6 and 8C). Thecycles have been subsequently determined to be40 kyr cycles related to variations in the solar radi-ation received in the high latitudes. This intervalushered in the large-scale Northern Hemisphere icesheets that have since dominated Earth’s climate. At2.6 Ma, the Rrst IRD was deposited in the openNorth Atlantic and was coeval with the d18O max-imum. Prior to 2.6 Ma, IRD was conRned to themarginal basins to the north, Greenland’s and Ice-land’s continental margins. Subsequent d18O maxi-ma were associated with IRD. Between 2.6 and1 Ma, large Northern Hemisphere ice sheets waxedand waned on the 40 000 year beat. Beginningaround 1 Ma, the ice sheets increased in size andswitched to a 100 000 year beat (Figure 6). Duringthis interval, deep-water temperatures remainedsimilar to those in the modern ocean (0 to 33C).

The planktonic foraminiferal d18O response dur-ing the late Pliocene event shows the cyclic behav-ior, but not the overall increase recorded by thebenthic foraminifera. As with the middle Miocened18O shift, the late Pliocene increase represents thecyclic build-up of ice sheets accompanied by deepwater cooling. The tropical surface water temper-atures, however, varied between 26 and 283C.

Mechanisms for Climate Change

Most climate change hypotheses for the Cenozoicfocus on either oceanic heat transport and/or green-house gas concentrations. Each mechanism producesdifferent responses in the equatorial-to-pole and sur-face-to-deep temperature gradients. An increase inthe meridional heat transport generally cools thetropics and warms the poles. If poleward transportof heat decreases, then the tropics will warm andthe poles will cool. Variations in greenhouse gasconcentrations should produce similar changes inboth the tropical and polar regions.

Tropical surface water and deep-ocean recordsco-varied for the Rrst part of the Cenozoic. Thewarming and subsequent cooling between 65 and34 Ma are most often ascribed to changing green-house gas concentrations. The interval of warmingthat began around 58 Ma and peaked at 52 Macoincided with the release of large amount of CO2

into the atmosphere as a consequence of tectonicprocesses. The eruption of the Thulean basalts inthe north-eastern Atlantic Ocean began during thePaleocene and peaked around 54 Ma. It is also rec-ognized that there was a large-scale reorganizationof the midocean ridge hydrothermal system whichbegan during the late Paleocene and extended intothe Eocene. Both tectonic processes acceleratemantle degassing which raises atmospheric levels ofCO2. Recently, evidence for another potentiallylarge CO2 reservoir was found along the easterncontinental margin of North America. Methane hy-drates frozen within the sediments appear to havereleased catastrophically at least once and possiblymultiple times during the latest Paleocene and earlyEocene (&58}52 Ma). One or all of these sourcescould have contributed to the build-up of green-house gases in the atmosphere between 58 and52 Ma. Following the thermal maximum, the long-term cooling in both the surface and deep watersimplies that greenhouse gas concentrations slowlydecreased. Proxies for estimating pCO2 concentra-tions (d13C fractionation within organic carbon andboron isotopes) are still being developed and re-Rned. However, preliminary indications are that at-mospheric pCO2 levels were high ('1000 p.p.m.)during the early Eocene, dropped to &400}500 p.p.m. during the middle to late Eocene, andreached late Pleistocene concentrations (200}300 p.p.m.) by the early Oligocene.

The deep-water temperature cooling across theEocene/Oligocene boundary (Figure 7B) was notaccompanied by tropical cooling, and resulted fromthe Rrst step in the thermal isolation of Antarctica.In modern ocean, the Antarctic Circumpolar Cur-rent is a vigorous surface-to-bottom current thatprovides an effective barrier to southward-Sowingwarm surface waters. The development of this cur-rent during the Cenozoic hinged on the deepeningfor the Tasman Rise and opening of the DrakePassage. Recent drilling indicates that marine con-nections developed across the Tasman Rise at ornear the Eocene/Oligocene boundary (33.5 Ma).Tectonic constraints on the separation of the DrakePassage are less precise. Estimates range from 35 to22 Ma for the opening of this gateway. The uncer-tainty lies in the tectonic complexity of the regionand what constitutes an effective opening for water

0023

0024

0025

0026

0027



to Sow through. The climate consequence of cre-ating a circumpolar Sow was to thermally isolateAntarctica and promote the growth of the Antarcticice sheet. As noted above, the Rrst large ice sheetgrew at beginning at the Eocene/Oligocene bound-ary.

The most notable divergence in the d18O recordsoccurred during the middle Miocene (&15 Ma).For the Rrst time during the Cenozoic, the tropicalsurface and deep waters recorded a clear divergencein d18O values, a trend that increased in magnitudeand reached a maximum in the modern ocean. Anypoleward transport of heat appears to have beeneffectively severed from Antarctica by 15 Ma, pro-moting further cooling. On the other hand, thetropics have been warming over the past 15 myrs.A combination of different factors fueled this warm-ing. First, less heat was being transported out of thelow- and mid-latitude regions to the high southernlatitudes. Second, the opening of the SouthernHemisphere gateways that promoted the formationof the circumpolar circulation led to the destructionof the circumequatorial circulation. The effects ofthe closure of the Tethys Ocean (predecessor tothe Mediterranean), shoaling of the PanamanianIsthmus (4.5}2.6 Ma), and constriction in theIndonesian Passage (&3 Ma to present) allowed theeast-to-west Sowing surface waters in the tropics to‘pile’ up and absorb more solar radiation. A conse-quence of the equatorial warming and high-latitudecooling was an increase in the equator-to-pole tem-perature gradient. As the gradient increased, windsincreased, promoting the organization of the surfaceocean circulation patterns that persist today.

Some Caveats

A concern in generating marine isotope records isthat the isotopic analyses should be made on thesame species. This is important because d18Ocalcite

values can vary among the different species of or-ganisms. Coexisting taxa of benthic foraminiferarecord d18O values that can differ by as much as1&. In planktonic foraminifera, variations betweenspecies can be as great as 1.5&. For both the plank-tonic and benthic foraminifera, interspeciRc differ-ences are as large as the glacial}interglacial signal.These interspeciRc d18Ocalcite variations are often as-cribed to a vital effect or kinetic fractionation of theoxygen isotopes within the organism. However,some of the difference in the planktonic taxa resultsfrom different seasonal or depth habitats and there-fore provides important information about proper-ties in the upper part of the water column. It isnoteworthy that the Rrst d18O syntheses were based

on mixed species analyses and yet basic featurescaptured in these curves still persist today. Thisattests to the robustness of these records andmethod for reconstructing climate changes in theocean.

The high-frequency signal that dominates the latePliocene to Pleistocene records is also present in theMiocene and Oligocene intervals. The cloud ofpoints about the mean shown in Figure 7(A) reSectsrecords that were sampled at a resolution sufRcientto document the high frequency signal. For the in-terval between 35 and 1 Ma, the benthicforaminiferal d18O record has a 40 000 year fre-quency superimposed on the long-term means thatare represented by the smoothed line. The origin ofthe 40 000 year cycles lies in variations in the tilt ofthe earth’s axis that inSuences the amount of solarradiation received in the high latitudes. This insola-tion signal is transmitted to the deep ocean becausethe high latitudes were the source regions for deepwaters during much, if not all, of the Cenozoic. Therecord prior to 35 Ma is unclear with regard to thepresence or absence of 40 000 year cycles.

See also

Holocene Climate Variability (246). Icebergs (2).Ice-induced Gouging of the Sea]oor (9). OxygenIsotopes in Seawater (163). Sea Ice (1). Sub Ice-shelf Circulation and Processes (10). 254.

Further ReadingCraig H (1957) Isotopic standards for carbon and oxygen

and correction factors for mass spectrometric analysisof carbon dioxide: Geochemica et Cosmochemica Acta12: 133}149.

Craig H (1965) The measurement of oxygen isotopepaleotemperatures. In: Tongiorgi E (ed.) Stable Iso-topes in Oceanographic Studies and Paleotempera-tures, pp. 161}182. Spoleto: Consiglio Nazionale delleRicerche, Laboratorio di Geologica Nucleare, Pisa.

Craig H and Gordon LI (1965) Deuterium and oxygen-18variations in the oceans and marine atmosphere. In:Tongiorgi E (ed.) Stable Isotopes in OceanographicStudies and Paleotemperatures, pp. 1}122. Spoleto:Consiglio Nazionale delle Ricerche, Laboratorio diGeologica Nucleare, Pisa.

Emiliani C (1955) Pleistocene temperatures. Journal ofGeology 63: 539}578.

Epstein S, Buchsbaum R, Lowenstam H and Urey HC(1953) Revised carbonate-water temperature scale.Bulletin of the Geological Society of America 64:1315}1326.

Fairbanks RG, Charles CD and Wright JD (1992) Originof Melt Water Pulses. In: Taylor RE et al. (eds)Radiocarbon After Four Decades pp. 473}500. Spring-

0028

0029

0030

0031



er-Verlag.Imbrie J, Hays JD, Martinson DG et al. (1984) The

orbital theory of Pleistocene climate: support froma revised chronology of the marine d18O record. In:Berger AL, Imbrie J, Hays JD, Kukla G and SaltzmanB (eds) Milankovitch and Climate, part, pp. 269}305.Dordrecht: Reidel.

Lear CH, ElderReld H and Wilson PA (1999) Cenozoicdeep-sea temperatures and global ice volumes fromMg/Ca in benthic foraminiferal calcite. Science 287:269}272.

Miller KG, Fairbanks RG and Mountain GS (1987) Terti-ary oxygen isotope synthesis, sea-level history, andcontinental margin erosion. Paleoceanography, 2,1}19.

Miller KG, Wright JD and Fairbanks RG (1991) Unlock-ing the Ice House: Oligocene}Miocene oxygen iso-topes, eustasy, and margin erosion. Journal ofGeophysical Research 96: 6829}6848.

Pagani M, Arthur MA and Freeman KH (1999) Mioceneevolution of atmospheric carbon dioxide. Paleoceanog-raphy. 14: 273}292.

Palmer MR, Pearson PN and Cobb SJ (1998) Reconstruct-ing past ocean pH-depth proRles. Science 282:1468}1471.

Pearson PN and Palmer MR (1999) Middle Eocene sea-water pH and atmospheric carbon dioxide concentra-tions. Science 284: 1824}1826.

Rozanski K, Araguas-Araguas L and GonRantini R (1993)Isotopic patterns in modern global precipitation. In:Swart PK, McKenzie J and Savin S (eds) ClimateChange in Continental Isotopic Records. GeophysicalMonograph 78, pp. 1}35. Washington, DC: AmericanGeophysical Union.

Rye DM and Sommer MA (1980) Reconstructingpaleotemperature and paleosalinity regimes with oxy-gen isotopes. In: Rhoads DC and Lutz RA (eds) Skel-etal Growth of Aquatic Organisms, pp. 162}202. NewYork: Plenum.

Savin SM, Douglas RG and Stehli FG (1975) Tertiarymarine paleotemperatures. Geological Society ofAmerica Bulletin 86: 1499}1510.

Shackleton NJ (1967) Oxygen isotope analyses andPleistocene temperatures re-assessed. Nature 215:115}117.

Shackleton NJ, Berger A and Peltier WR (1990) An alter-native astronomical calibration of the Lower Pleis-tocene time scale based on ODP Site 677. Transactionsof the Royal Society of Edinburgh, Earth Science 81:251}261.

Shackleton NJ and Kennett JP (1975) Paleotemperaturehistory of the Cenozoic and initiation of Antarcticglaciation. Oxygen and carbon isotopic analysis inDSDP Sites 277, 279, and 281. Initial Report. DeepSea Drilling Project 29, 743}755.

Shackleton NJ and Opdyke ND (1973) Oxygen isotopeand paleomagnetic stratigraphy of equatorial PaciRccore V28-238. Oxygen isotope temperatures and icevolumes on a 105 year and 106 year scale. QuaternaryResearch 3: 39}55.

Tiedemann RM, Sarnthein M and Shackleton NJ (1994)Astronomic calibration for the Pliocene Atlantic d18Oand dust Sux records of Ocean Drilling Program Site659. Paleoceanography 9: 619}638.

Urey HC (1947) The thermodynamic properties of iso-topic substances. Journal of the Chemical Societypp. 562}581.



1 d18Owater term in paleotemperature equations is expressedrelative PDB, not SMOW. To convert the d18Owater values fromthe SMOW to PDB scales, one must subtract 0.22& from theSMOW value.

2 Many island or coastal regions have significantly higherd18Owater values relative to continental locations at similar latit-udes. This occurs because local evaporation increases thed18Owater values, thus, resetting the initial conditions for Rayleighdistillation to occur.

0

_10

_20

_ 30

_ 40

_ 50

_ 60

d18O

%°

1.0 0.8 0.6 0.4 0.2 0Fraction remaining

Liquid

Water vapor

Figure 1 Rayleigh distillation model showing the effects ofevaporation and precipitation on the d18O values in the vaporand liquid phases. The initial conditions are a temperature of253C and d18Owater value of 0&. This model also assumes that itis a closed system, meaning that water vapor is not added oncethe cloud moves away from the source regions. As clouds losemoisture, fractionation during the condensation further lowersthe d18Owater value in the water vapor.

a0252fig0001



100% of water vaporremaining

O = _ 9d18

%°

O = 0d18

rain %° O = _ 6d18

rain %°

O = _ 20d18

snow %°


O = _ 15%d18

°


O = _ 30d18

%°

Eva

pora

tion

EQ 20° 40° 60° 80°Latitude

water vapor water vapor water vapor

Ice sheet Od18

_ 20 to_ 60%°

Figure 2 Illustration of the Rayleigh distillation process on d18O values as clouds move over land and into the polar regions.Decreasing air temperatures cause moisture to rain/snow out of the cloud. Fractionation of the oxygen isotopes during condensationfurther decreases values. By the point that a cloud reaches the high latitudes, less than 10% of the original water vapor remains.Snowfall on Antarctica has values between !20 and !60&.The average d18O value for ice on Antarctica is &!40&.

5

0

_ 5

_10

_15

_ 20

_25

_30

_35

_4090 70 50 30 10 10 30 50 70 90

NSLatitude

d18O

%°

(A)

d18O

of p

reci

pita

tion

SM

OW

(%

) °

0

_10

_ 20

_ 30

_ 40_ 40 _ 30 _ 20 _ 10 10 20 300

Surface air temperature (°C)(B)

Figure 3 Mean annual d18O water of precipitation (rain/snow) versus mean annual temperatures. The correlation between d18Ovalues and latitude (A) is a function of temperature, (B). The rainout/fraction of water remaining, and hence the fraction of d18Ovalues, is determined by the cloud temperatures. Latitude is the dominant effect shown here. The scatter among sites at similarlatitude results from elevation differences as well as differences in the distance from the ocean.

a0252fig0002

a0252fig0003



38

37

36

35

34

33

32

Sal

inity

(%

) °

80°S 60°S 40°S 20°S 20°N 40°N 60°N 80°N0°Latitude(A)

80°S 60°S 40°S 20°S 20°N 40°N 60°N 80°N0°Latitude

1.5

1.0

0.5

0.0

_0.5

_1.0

d18O

%°

(B)

1.5

1.0

0.5

0.0

_ 0.5

_1.032 33 34 35 36 37 38

Salinity

d18O

%°

d18

OSalinity

0.5=

(C)

Figure 4 The salinity (A) and d18Owater values (B) measured inthe open Atlantic and Pacific Oceans. Note the higher values inthe tropical and subtropical region relative to the subpolar andpolar regions. Evaporation and precipitation/runoff processesproduce similar patterns in salinity and d18Owater values. Theocean-to-ocean difference between the Atlantic and Pacific re-sults from a net transfer of fresh water from the Atlantic to thePacific.

a0252fig0004



Present

20000 y BP

Sea water

Sea water

O = 0.0%d18

°

O = 1.2%d18

°

Glacial iced

18O = _ 40%°

∆Sea level = 120 m

120 m/ 3800 m = 0.03

∆d18

O = 1.2%°

0.0 _ ( _ 40 0.03)×% %° °

water

Mean depth of oceans

= 3800 m

d18

O = 0.0%°

Figure 5 The effect of building or removing large ice sheets on the d18O composition of the ocean can be significant. The removalof 3% of the ocean’s water during the last glacial maximum lowered sea level by 120 m. The d18O difference between the ocean andthe ice is 40&, causing a whole ocean d18O change of 1.2&. The reverse process occurs during the melting of large ice sheets. Ifthe Antarctic and Greenland ice were to melt, then sea level would rise &70 m. The volume of water stored in these ice sheets isequivalent to &2% of the water in the ocean. Therefore, the mean d18O value of the ocean would decrease by 0.7}0.8&.

a0252fig0005



Planktonic foraminifera

Benthic foraminifera

2

3

4

5

0 0.5 1.0 1.5Age (Ma)

_ 2

_ 1

0

d18O

%be

nthi

c fo

ram

inife

ra°

d18O

%planktonic foram

inifera°

Holocene Pleistocene

1234 5 6 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37394143 45 4749



2

3

4

5

1.5 2.0 2.5 3.0Age (Ma)

_ 2

_ 1

0

d18O

%be

nthi

c fo

ram

inife

ra°

d18O

%planktonic foram

inifera°

Pleistocene

515355 57 59 61 63 65 67 69 71 7375 77 798183 85 87 89919395

96 98 100

Pliocene

(A)

(B)

Figure 6 Planktonic and benthic foraminiferal d18O records for the last 3 million years. Note the high frequency signals in therecords. For the interval between 3 and 1 Ma, a 40 000 year cycle dominates the records. After 1 Ma, the beat changes to a 100 000year cycle and the amplitudes increase.

a0252fig0006



_2

_1

0

1

2

3

4

0 5 10 15 20 25 30 35 40 45 50 55 60 65



Plio.

Ple

ist.

Miocene Oligocene Eocene Paleocene

d18O

%°

Age (Ma)

0 5 10 15 20 25 30 35 40 45 50 55 60 65

Plio.

Ple

ist.

Miocene Oligocene Eocene Paleocene

Age (Ma)

30

25

20

15

10

5

0

Tem

pera

ture

(°C

)

Figure 7 (A) Planktonic and benthic foraminiferal d18O composite records representing the tropical surface and deep oceanconditions. The thick line through both records was generated using a 1 million year Gaussian filter. (B) Temperature estimatesbased on planktonic and benthic records and ice volume estimates discussed in the text.

a0252fig0007



1.5

2.0

2.5

3.0

33.00 33.25 33.50 33.75 34.00Age (Ma)

d18O

%°

(A)

1.0

1.5

2.0

2.5

3.012.5 13.0 13.5 14.0 14.5

Age (Ma)

d18O

%°

(B)

2.5

3.0

3.5

4.0

4.5

d18O

%°

2.4 2.5 2.6 2.7 2.8 2.9 3.0 3.1 3.2

Increasing No. Hemisphere glaciation

Age (Ma)(C)

Figure 8 High-resolution d18O records representing the Eocene/Oligocene boundary (A), middle Miocene (B), and late Pliocene(C) d18O shifts.

a0252fig0008



CENOZOIC CLIMATE OXYGEN ISOTOPE EVIDENCE · CENOZOIC CLIMATE ^ OXYGEN ISOTOPE EVIDENCE Jame’s D....

Documents

Transcript of CENOZOIC CLIMATE OXYGEN ISOTOPE EVIDENCE · CENOZOIC CLIMATE ^ OXYGEN ISOTOPE EVIDENCE Jame’s D....