+LPDOD\DQ7HFWRQLFV:HDWKHULQJ3URFHVVHVDQGWKH6WURQWLXP,VRWRSH5HFRUGLQ0DULQH /LPHVWRQHV $XWKRU V -0(GPRQG 6RXUFH6FLHQFH1HZ6HULHV9RO1R 'HF SS 3XEOLVKHGE\American Association for the Advancement of Science 6WDEOH85/http://www.jstor.org/stable/2882056 . $FFHVVHG
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use.
Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at . http://www.jstor.org/action/showPublisher?publisherCode=aaas. .
Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].
American Association for the Advancement of Science is collaborating with JSTOR to digitize, preserve and extend access to Science.
http://www.jstor.org Himalayan Tectonics, Weathering Processes, and the Strontium Isotope Record in Marine Limestones
J. M. Edmond The time evolution of the isotopic composition of seawater strontium (the ratio of strontium- these variablesare causallyrelated, it has 87 to strontium-86) over the last 500 million years has the form of an asymmetric trough. not proved possible to develop an unam- The values are highest in the Cambrian and Recent (0.7091) and lowest in the Jurassic biguous and specific interpretationof the (0.7067). Superimposed on this trend are a number of smaller oscillations. Consideration isotope recordin terms of explicit mecha- of the geochemical cycle of strontium and the dynamics of weathering shows that only nisms. The consensusis that variationsin Himalayan-style continental collisions can influence the isotope ratio on the scale ob- the intensity of continental weathering served. The contemporary Himalayan orogeny is by far the largest since the late Pre- are important, although this notion has cambrian Pan-African event that produced the high in the Cambrian. not been developed in any detail. The average isotope ratio of Sr in the fluvial flux from the weathering of lime- stones and evaporites over the period of Variationsin the isotopiccomposition of past (4). Observationsindicate that the record varied between about 0.7075 and strontiumin seawater,recorded in marine fluvialflux is dominatedby the weathering 0.7080, a quite narrow range (4). The limestones, are a proxy of the interplayof of platform carbonatesand evaporites of long residencetime of Sr in the limestone mantle processesand crustal weathering. marine origin. There is only a minor reservoiracts strongly to damp the signal Wickmanoriginally proposed that Rb (the contributionfrom the primaryweathering from the relatively high-frequencymarine radioactiveparent of 87Sr, half-life -48 of the refractoryigneous and metamorphic isotopic oscillations (-50 million years). billion years), as an incompatibleelement, basement rocks (5). Hence, the memory Contemporaryoceanic hot springshave a should be progressivelyenriched in conti- effect is dominant. uniform ratio of about 0.7035, which nental crustalrocks over geologic time rel- A synthetic stratigraphiccolumn com- probablyhas not varied over the last 500 ative to Sr (1). Hence, Sr from fluvial posedof marinelimestones has been assem- million years (3). On the basis of the sea inputsto the ocean shouldbecome increas- bled that is continuousover the past 500 floor spreading record of the last 150 ingly radiogenic, leading to a monotonic million years (6). The Sr isotopicrecord of million years, the hydrothermalflux has increasein the isotopiccomposition of Sr in seawaterderived from this column is char- changed slowly and continuously but seawater.Wickman was the first to recog- acterizedby largeoscillations on a varietyof probablyless than ?30% (8), an amount nize that the record of this increase, pre- time scales (Fig. 2). Values of the 87Srto insufficientto resultin any markedchange servedin marinesediments, could be used 86Sr ratio were high in the Cambrian in the isotopic record. Hence, strong fea- for dating, particularlyof limestones and (-0.7091) but declined irregularlyuntil tures in the record most likely reflect evaporites(1). This latteridea has recently the late Jurassic(-0.7067). Several inter- continental processes.Given that the av- been revivedand appliedwith some success mediate-scalefluctuations are superimposed erage Sr flux from limestones has isotopic (2) (Fig. 1). However, the isotopicsystem- on this trend. The continuation of these ratios between the extrema of the oscilla- atics are much more complex than Wick- oscillationsin the Cretaceousand Tertiary man could have realized. was interruptedin the mid-Tertiaryby a The isotopic composition of Sr in sea- steep rise that has continued until the 0.7093 water at any time is determined by the present time (Fig. 1). The contemporary balance between the deep-sea hydrother- seawaterratio (0.7091) is the same as that mal input of nonradiogenicSr of mantle at the height of the Cambrianand is sub- origin (-0.7035) and the more radiogenic stantially higher than at any intervening fluvial flux derived from continental time. These observationsare completelyat weathering(-0.712) (3). The majorsink variancewith the originallinear evolution for Sr is marine carbonaterocks; incorpo- model (1). ration in the carbonatesinvolves no iso- In attempts to explain the features in topic fractionation.The residencetime of the isotope curve, correlationshave been Sr in this limestone reservoir is on the claimedwith many events contained in or (0*0 orderof 150 million years (and in seawater inferredfrom the geologic record (2, 7). 0~~~~~~~~~~~~~~ -3 million years) (3). Variations in the Changes in the fluvial inputs have been continental signal, caused by the dissolu- attributedto variationsin continental gla- 0~~~~~~~~~~~~~~ tion of igneous and metamorphicrocks, ciation, eustatic sea level, atmospheric OR are thereforedamped by the weatheringof CO2 levels, climate, vegetation, and these carbonatesafter exposurecaused by mountain building, individually or in changesin eustaticsea level or by tectonic combination. For time periods before uplift. Because of the weathering of car- about 100 million years ago where the 0.707",'0 20 40 60 80 100 bonates, the fluvial flux retains a "memo- paleomagneticrecord provides little con- Age (millions of years) ry"of isotopic fluctuationsin the geologic straint, changes have been claimed be- Fig. 1. High-resolutionisotope ratio curve for cause of substantialvariations in the oce- The author is in the Department of Earth, Atmospheric, abyssal carbonates over the past 100 million and Planetary Sciences, Massachusetts Institute of anic hydrothermalflux and hence the sea years based on drill core samples [modified Technology, E34-201, Cambridge, MA 02139. floor spreading rate. Because many of from (2)]. The chronometricpotential is clear.
1594 SCIENCE * VOL. 258 * 4 DECEMBER1992 tions and that deviations, such as the Strontium in Rivers global reconnaissance(3, 12). The funda- recent one, extend well above this range, mental resultis that exposedigneous base- the fundamentalvariable is the weather- A large amount of data exists on the Sr ment terrains, in continental shields, do ing yield of radiogenic Sr from old base- systematicsin rivers that characterizethe not yield radiogenicSr in high flux, even ment rocks. To the first order, the prob- relationsamong the Sr flux, the Sr isotope when the rocks are highly alkalic and of lem is reduced to the identification of ratio of the riversand oceans, the geologic greatage. This is trueof both deeplyweath- distinct processes that produce a high events, and weathering environments ered, lateritic soils from basement in the fluvial flux of radiogenic Sr and operate where Sr is releasedfor transport.A num- Tropics,such as that foundin the Guyana, repeatedlyover discrete intervals of geo- ber of local studiesare available(9), along Brazilian,and southernAfrican shields (3, logic time. with some regionalones (10, 11) and one 12), and recently glaciated terrains, as foundin easternCanada (10). Of the other siliceousrocks (12), black shalescan give a high dissolved yield as a result of acid 0.71019]Tertiary Cretaceous Jurassic TdasIc PerrmanPenn] Mbis.sDevonianSilurian Ordoviciancam weathering associatedwith sulfide oxida- tion; basic rocks also weather rapidlybe- cause of the oxidative breakdownof mafic minerals.However, the Sr that is releasedis 0.709 _ not radiogenicbecause of the usualadmix- ture of carbonaterocks in black shales and the mantle affinitiesof volcanic rocks. The only observedexception to this generaliza- 0.708~ - tion is the metamorphiccore complexfrom the centralHimalaya, which influencesthe CD Gangesand Brahmaputrarivers (3, 11, 12). 00 0 Data from a wide varietyof geologic envi- ronments (Fig. 3, A and B) define a uni- form shallowincrease in the isotopic com- 0.70671 position as Sr concentration decreases. Only the datafrom the Himalayafall mark- 0 100 200 300 400 500 edly off this trend; the main relation is geologicallynormal. The modeling studies Age (millions of years) (7) have shown, in effect, that this trend Fig. 2. The variations in the ratio of 87Srto 86Sr in marine limestones over the past 500 million years does not supportthe observedchanges in [modified from (6)]. the isotope curveif the hydrothermalinput
1.0 0.80- A A B A
0 0 0.97 0 0 0
0 0 Co ~~~~~~~~~~~~~~~~~~~~A0 Co ~~~~~~~~~~~~~~~~~~~~~~~~~A~~ AA 0 0A 0 0 0.75- A A 0~~~~~~0c20 0~~~~~ 0A~ o 08 0 0
0 0 0 0 0 ZP 00 000 00 g+xo
0 50 0100 150 10 20
ll 1/[Sr] (umol/kg)-l 1 /Sr!: lmolikg)l Fig. 3. (A) Summary plot of 87Sr/86Sr versus 1/[Sr] (concentration in There are several data sets for which no concentrationsare reported.For micromoles per kilogram) in river water using all the available data (-425 the CanadianShield (mollusks),southern Africa (elephant tusk and bone), points) for which both isotopic composition and concentration were central Europe (Rhine drainage), and the Baltic Shield (Gulfof Bothnia reported (3, 4, 9-12). The data form several groups: (0) Shields (Guyana, drainage),the data are consistent withcomplete data sets fromthe rivers Brazilian, southeastern Canada, east Greenland, western Australia). of the regions as displayed here. (B) The flow-weightedmean concentra- These are generally highly radiogenic with very low concentrations. ([1) tion of fluvialSr is -1 ,mol/kg, so the global plot (A) is truncatedat 0.05 North America (with the exception of southeastern Canada). The geo- ,mol/kg (1/[Sr]= 20) to show in moredetail the relationsfor riverscarrying graphical coverage is essentially complete. (A) Himalayan streams significantflux. The separationof the Himalayanstreams fromthe others is (Ganges left bank; Brahmaputra in Assam). (x) The Andean arc in the clear. The positive, shallow isotope-concentrationgradient is followed by drainages of the Amazon and Orinoco (southern Bolivia to northwestern all non-Himalayanrivers, regardless of geologic setting and weathering Venezuela). (+) Other point samples collected in reconnaissance efforts environment,for reasons discussed in the text. Data defining the lower (the Amazon flood plain, western and central Europe, east Africa, central envelope of this trend are from basaltic terrainsat convergent margins. India, eastern Australia, southeast Asia, China, Philippines, and Japan). The upper envelope is defined by the data fromthe shields.
SCIENCE * VOL. 258 * 4 DECEMBER1992 1595 is fairlyconstant. Excludingthe Himalayan from the Sr isotope-concentrationrelation rangebetween 0.7 10 and 0.713. This semi- rivers, the averageisotopic compositionof (Fig. 3B). On a global basis, the fluvial arid terrainhas not been glaciated,despite fluvial Sr is about 0.710, a value that isotope ratio increasesslowly with decreas- regionalelevations of between 1 and 2 km, probablydoes not (and from the above ing concentrationfrom the marinerange at but has all the features of a periglacial discussion cannot) vary appreciablywith high concentration (limestones, partial regime, including extensive frost shatter- time. weatheringof silicates) to more radiogenic ing, frozen and patternedground, and so Severalfactors contribute to this behav- values at low concentration (progressive forth. The riversare bouldertorrents that ior. Carbonaterocks and evaporitesweath- weatheringof more refractoryaluminosili- transportlarge amountsof immaturesand er much more rapidlythan massive base- cates) (3). In the Ganges-Brahmaputra,the but negligiblequantities of fine-grainedma- ment rocks, and shales rarely have the rise is much steeper.Extensive sampling of terial. Weatheringyields are quite high (Sr mechanical competence to sustain signifi- Himalayanheadwater tributaries in the In- values up to 3 pumol/kg),comparable to cant chemicalweathering at outcropin the dian sector of the chain demonstratesthat those of the Himalaya,but the intensity, as face of rapidphysical erosion and transport. this anomalousbehavior is driven by the reflectedin the isotopes, is very low. In addition, these three rock types domi- weatheringyield from the unroofedmeta- The key to the high radiogenicyields nate shelf sequencesand are thus preferen- morphiccore complexof the collision zone from the rocks of the complex lies in the tiallyexposed to weatheringduring eustatic (3, 11, 12) and not from diagenetically disturbed Sr-Rb systematics (17). High- sea-level loweringor after tectonic transi- altered carbonates, as recently proposed grade metamorphismand partial melting tion frompassive to active marginregimes. (12). Streamswith Sr concentrationschar- of the rocks during the recent Tertiary The constituentminerals of basementrocks acteristicof limestone terrains(>2 ,umol/ collision of India with Asia have been weatherchemically at greatlydifferent rates kg) displayisotope ratios of 0.8 and greater. sufficientto mobilize radiogenicSr out of (13). In particular,sodic and calcic alumi- Modeling studiesby variousworkers using the K-rich minerals into phases that nosilicates are much more reactive than these data from the Himalaya(7) identify weather more rapidly. With reportedini- K-containingminerals. Because Sr is parti- the weatheringyield fromthis orogenas the tial isotope ratios as high as 0.769 in both tioned into Na- and Ca-rich minerals, driver of the dramaticrise in the marine the leucogranites and the gneisses (20), whereasRb is in resistantK-rich minerals, isotope ratio over the last 20 to 40 million the Na- and Ca-rich minerals have been the dissolvedyield duringpartial chemical years. The cause of this distinct source of transformedinto a sourceof radiogenicSr. weatheringis predominantlynonradiogenic radiogenic Sr in high flux has not been In strikingcontrast, the Aldan Shield has Sr with an isotopic composition between clear. been tectonically quiescent for at least the the whole-rockvalue and the initial ratioat The metamorphiccore complexextends last 2 billion years. Rapid erosion does the time of emplacement,before the onset along the axis of the mountainbelt, from sustainthe flux in both regionsbut only by of in situ Rb decay. Becausethe Na- and Assam to Pakistan(14). It consists of am- acceleratingthe productionof freshsurfac- Ca-rich minerals are abundant in silicic phibolite-gradegneiss and associatedlarge es. The intense mechanical action associ- rocks, their preferentialdissolution destabi- leucogranitemasses (15). The complex is ated with glacial and periglacialprocesses lizes the rock surface mechanically, en- thought to representthermally remobilized means that, while the total erosion rate is hancingthe removalof the resistantK-rich basement of the Indian plate, possiblyof increased, the intensity of chemical mineralsand quartzby physicalprocesses. Archean and EarlyPrecambrian age, that weatheringis not; it may even decrease. In areasof subduedlocal relief, where hy- was buried to great depth early in the Rock flour,the dominantparticulate prod- drologic transportis very low, complete collision and then tectonicallyexhumed to uct from glaciated igneous and metamor- weatheringof basement rocks does occur high altitude and rapidly unroofed (16). phic terrains, is much more immature and resultsin a residualmantle of kaolinite, The age of the complex, in particularthe than the suspendedmaterial derived from gibbsite,quartz, and iron oxyhydroxides-a time of emplacementof the leucogranites, rocks of similar composition in nonglaci- laterite. However, the dissolutionrates of is quite uncertainbut appearsto be -20 ated terrains. The result is Sr isotopic the K minerals are sufficientlyslow that, million years(17). The thermalremobiliza- values in the flux that are close to the although very radiogenic Sr is released, tion of the basement rocks during deep initial ratio in the rock masses and char- ratios >0.9 in some cases (Fig. 3A), the burial was sufficientto disrupt the Rb-Sr acteristic of the most labile minerals. flux is extremelylow (12). Partialweather- systematicsbut not to reset the systemby ing of even highly radiogenic basement homogenizationthroughout the rock mass- Implications for Past Variations rocks tends to give isotope values not sig- es (17). nificantly greater than those from lime- The complex is exposed over a huge The major oscillations in the Sr isotope stones; hence, the data from shield and elevationrange in an areaof intense glacial recordin marinelimestones can be caused platformcarbonate terrains are superposed erosion. Earlier interpretationshave as- only by continentalcollisions of the Hima- on the isotope-concentrationdiagram (Fig. cribedthe high flux of radiogenicSr to this layan type. This is the sole mechanismby 3A). The observationaldata and the mech- weatheringenvironment (7); however, ob- which ancient basement rocks are down- anistic understandingof weathering pro- servationsfrom other glaciatedregions do thrust to great depth, thermally remobi- cesses are both sufficientlygood to allow not support this inference. Data from lized, heated, even partiallymelted such one to concludethat, undernormal circum- streamsdraining the essentiallycontinuous that radiogenicSr is redistributedinto the stances,it is difficultto sustaina high fluxof fluvio-glacialcover of the stable shields in Na- and Ca-rich phases, and rapidlyex- radiogenicSr, regardlessof the intensityof easternCanada (10) and the Baltic region humedand eroded. Orogeniccollapse (21) the weatheringprocess. The cause of the (18) give low concentrations(<0.2 ,umol/ determinesthe relativelyuniform time scale large deviations from the isotope curve kg for eastern Canada; data from Baltic of discrete events in the record and, in must be soughtelsewhere. Shield not reported)and an isotope ratio particular,the steepnessof the decline in around0.73, both much lower than in the the signal. Recent work has extended the Effectsof the HimalayanOrogeny streams from the complex. Recent data isotope record, discontinuously, back to (19) for streams draining the Early Archean about 950 million years ago (22). The The Ganges-Brahmaputrais the only major Aldan Plateau, in the drainage of the Up- record shows a rise in the late Precambrian, river system to show a significant departure per Lena River in the Russian Far East, which resulted in the Cambrian high, that
1596 SCIENCE * VOL. 258 * 4 DECEMBER1992 ARTICLES is comparableto the rise being producedby the fluvial Sr isotopic value cannot be Am. J. Sci. 292, 136 (1992); F. M. Richter, D. B. Rowley, D. J. DePaolo, Earth Planet. Sci. Lett. 109, the Himalayanevent (Fig. 2). It is almost expected because the normal weathering 11 (1992): certainly associatedwith the Pan-African yield of radiogenicSr is transport-limited 8. S. Gaffin, Am. J. Sci. 287, 596 (1987). orogeny (22). If the magnitudeand steep- (3, 12). 9. R. S. Fisher and A. M. Stueber, Water Resour. ness of the isotopeshift areindicative of the This interpretationof the recordof the Res. 12, 1061 (1976); F. Albarede and A. Mi- chard, Chem. Geol. 64, 55 (1987); D. Buhl et al., scale of a particularcollision, then, from isotopiccomposition of marineSr illustrates Naturwissenschaften 78, 337 (1991); B. L. Ingram Fig. 2, it can be concluded that nothing the difficultiesin developingand interpret- and D. Sloan, Science 255, 68 (1992). comparablein size with these two events ing geochemicalproxy recordsin the ab- 10. M. A. Wadleigh, J. Veizer, C. Brooks, Geochim. has occurredin the intervening600 million Cosmochim. Acta 49,1727 (1985); S. J. Goldstein sence of a completeunderstanding of all of and S. B. Jacobsen, Chem. Geol. Isotope Geosci. years.Similarly, the time spanof the Cam- the processesaffecting the behaviorof the Sect. 66, 245 (1987). brianhigh impliesthat the Himalayanoro- elementsand isotopesin question.The best 11. S. Krishnaswami et al., Earth Planet. Sci. Lett. 109, genic cycle of basement reactivation, un- developedproxy records are those used for 243 (1992). 12. M. R. Palmer and J. M. Edmond, Geochim. Cos- roofing, and erosion should continue for radiometricage dating. All unknowns in mochim. Acta 56, 2099 (1992). tens of millionsof years,sufficient to rework the behaviorof the chronometercontami- 13. S. S. Goldich and P. W. Gast, Earth Planet. Sci. much of the IndianShield. nate the derivedage. The obviousproblem Lett. 1, 372 (1966); E. J. Dasch, Geochim. Cos- mochim. Acta 33, 1521 (1969); G. W. Brass, ibid. In the absenceof constraintson the sea is the uncertaintyin the half-life.However, 39, 1647 (1975). floor spreadingrate before -100 million changes in the source function, for exam- 14. B. F. Windley, Philos. Trans. R. Soc. London Ser. yearsago, the primaryinformation that can ple, for the cosmogenic isotopes, initial A 326, 3 (1988). be extractedfrom the isotope curve is the disequilibrium,postformation mobilization, 15. M. Colchen, P. Le Fort, A. Pecher, Recherches Geologique dans l'Himalaya du Nepal (Centre timing and relative scale of previouscolli- and so forth, all lead to spuriousages if not National de la Recherche Scientifique, Paris, sion events. Because the total weathering accountedfor quantitatively(25). In prin- 1986). flux from the Himalayanchain, and the ciple, the sameand other difficultiesattend 16. K. V. Hodges, M. S. Hubbard, D. S. Silverberg, majorelement compositionof this flux, is the proxy recordsfor geological events in Philos. Trans. R. Soc. London Ser. A 326, 257 (1988). similarto that froman Andean-typemargin general. The ultimate utility of environ- 17. P. Le Fort et al., Tectonophysics 134, 39 (1987); P. of an equivalentlength, there is no neces- mental paleorecordsis that they constrain Le Fort, Philos. Trans. R. Soc. London Ser. A 326, sarycorrelation of the Sr isotopecurve with modelsof processesof climatic and geolog- 281 (1988). other geochemicalproxy records (7). Hi- ical change (7). The 18. G. Aberg and F. E. Wickman, Nord. Hydrol. 18, 33 validityof these con- (1987). malayan-styleevents also maynot necessar- straintscan only be establishedby compre- 19. G. Penteleyev, M. R. Kurz, J. M. Edmond, unpub- ily cause large-scaleperturbations of the hensive field observation. lished data. preexistingCO2 cycle, such as would be 20. M. P. Searle and B. J. Fryer, in Collision Tectonics, M. P. Coward and A. C. Ries, Eds. (Geological causedby a globallysignificant increase in REFERENCESAND NOTES Society Special Publication, no. 19, Blackwell the rateof chemicalweathering (23). How- Scientific, London, 1986), p. 185; Ph. Vidal et al., ever, these events, because they generate 1. F. E. Wickman, J. Geol. 56, 61 (1948). Phys. Earth Planet. Inter. 35, 179 (1984); C. De- enormousareas of extremeelevation locat- 2. D. J. DePaolo and B. L. Ingram, Science 227, 938 niel, Ph. Vidal, P. Le Fort, Terra Cognita 3, 266 (1985). (1983); Ph. Vidal, A. Cocherie, P. Le Fort, ed in the interiorof the newly composited 3. M. R. Palmer and J. M. Edmond, Earth Planet. Sci. Geochim. Cosmochim. Acta 46, 2279 (1982). continent, may have effects on climate Lett. 92, 11 (1989). 21. C. Burchfiel and L. Royden, Geology 13, 679 markedlydifferent from those of the nar- 4. G. W. Brass, Geochim. Cosmochim. Acta 40, 721 (1985). (1976). 22. L. A. Derry et al., Geochim. Cosmochim. Acta 53, row, peripheral, Andean-type orogenies 5. R. F. Stallard and J. M. Edmond, J. Geophys. Res. 2331 (1989); L. A. Derry, A. J. Kaufman, S. B. (23). If this is the case, then globalpertur- 88, 9671 (1983); ibid. 92, 8293 (1987); H. Ming- Jacobsen, ibid. 56, 1317 (1992). bations in climate, and hence in the hui, R. F. Stallard, J. M. Edmond, Nature 298, 550 23. M. E. Raymo, W. F. Ruddiman, P. N. Froelich, (1982). Geology 16, 649 (1988); W. F. Ruddiman and J. E. geochemical cycle, may be induced by 6. P. W. Gast, Geol. Soc. Am. Bull. 66,1449 (1955); Kutzbach, Trans. R. Soc. Edinburgh 81, 301 large-scalecontinental collisions. Postulat- Z. E. Peterman, C. E. Hedge, H. A. Tourtelot, (1990). ed increasesin the subduction-relatedCO2 Geochim. Cosmochim. Acta 34, 105 (1970); J. 24. K. Caldeira, Nature 357, 578 (1992). flux at the time of the caused Veizer and W. Compston, ibid. 38, 1461 (1974); 25. G. Faure, Principles of Isotope Geology (Wiley, collision, by W. H. Burke et al., Geology 10, 516 (1982); J. New York, ed. 2, 1986). the shift fromshelf to pelagicaccumulation Hess, M. L. Bender, J.-G. Schilling, Science 231, 26. The ideas in this article were clarified by discus- of carbonate(24) due to the associatedsea 979 (1986). sions with P. Molnar. The supporting data depend level drop, merely move the Sr isotope- 7. E. T. C. Spooner, Earth Planet. Sci. Lett. 31, 167 largely on collaborations with M. R. Palmer, S. R. (1976); L. R. Kump, Am. J. Sci. 289, 390 (1989); Hart, and M. M. Sarin. Much of this work was concentration relation along the shallow D. A. Hodell, G. A. Mead, P. A. Mueller, Chem. supported by the Earth Surface Processes Divi- globaltrend (Fig. 3). Dramaticincreases in Geol. 80, 291 (1990); R. A. Berner and D. M. Rye, sion of the National Science Foundation.
SCIENCE * VOL. 258 * 4 DECEMBER1992 1597