Variation of seawater 87Sr/86Sr throughout Phanerozoic time

W. H. Burke, R. E. Denison,* E. A. Hetherington, R. B. Koepnick, H. F. Nelson,* J. B. Otto

Vtobil Research and Development Corporation, Field Research Laboratory, Dallas, Texas 75221

ABSTRACT Precise measurements of 786 marine carbonate, evaporite, and phosphate samples of known age provide a curve of seawater 87Sr/86Sr versus geologic time through the Phanerozoic. Many episodes of increasing and decreasing values of 87Sr/86Sr of seawater have occurred through the Phanerozoic. The Late -Early seawater ratios are approximately equal to the modern ratio of 0.70907. The lowest ratios, ~ 0.7068, occurred during the and Late . The configuration of the curve appears to be strongly influenced by the history of both plate interactions and seafloor spreading throughout the Phanerozoic. The curve provides a basis for dating many marine carbonate, evaporite, and phosphate samples. Furthermore, diagenetic modifications of original marine 87Sr/86Sr values are often interpretable. Analysis of 87Sr/86Sr data, therefore, may provide useful information on regional diagenetic patterns and processes. All of the Cenozoic samples and some of the samples are from Deep Sea Drilling Project (DSDP) cores. With the exception of the DSDP samples, the curve was constructed only from samples containing at least 200 ppm Sr and not more than 10% dilute acid insoluble material. All measurements are made by comparison with standard 87 86 SrC03 (NBS SRM 987) for which a Sr/ Sr of 0.71014 is assumed. Precision is estimated to be ± 0.00005 at the 95% confidence level. Measured ratios of 42 modern marine samples average 0.70907, with a standard deviation of 0.00004.

INTRODUCTION of 87Rb to 87Sr would be reflected in an (1975), Clauer (1976), Faure and others We present here our estimate of the increase with time of 87Sr/86Sr within sea- (1978), and Kovach (1980). curve of seawater 87Sr/86Sr versus geologic water and within strontium-bearing marine time (Holocene through Late Cambrian), precipitates. Peterman and others (1970) SAMPLES which is based on measurements of 786 determined that 87Sr,86Sr of Phanerozoic Surface and subsurface samples of samples of marine origin. This data set marine fossils both increased and decreased marine origin were obtained from North enables us to define the curve more pre- with time. During the early 1970s, addi- America, Europe, Africa, and Asia. Addi- cisely and more completely than was possi- tional reports on the temporal variation of tional samples from the Atlantic, Indian, ble heretofore and to recognize structure seawater 87Sr/86Sr (Dasch and Biscaye, and Pacific Oceans and the Caribbean Sea not previously apparent. The curve reveals 1971; Veizer and Compston, 1974) gener- were obtained from Deep Sea Drilling Pro- systematic temporal variation of seawater ally supported the earlier findings of ject (DSDP) cores. All Cenozoic and some 87Sr/86Sr. Although few ages can be Peterman and others (1970). Veizer and Cretaceous samples are foram-nannofossil assigned on the basis of unique seawater Compston (1974) included a summary of chalk from DSDP cores. Non-DSDP sam- strontium isotopic values, the curve pro- earlier work and presented reasons for ples used to construct the Mesozoic and vides a basis for resolving specific problems believing that seawater 87Sr/ 86Sr is uniform Paleozoic parts of the curve are predomi- in temporal correlation of marine strata at any given time, that marine precipitates nantly limestone, but also include dolo- and constitutes an important source of will have the seawater ratio when formed, stone, carbonate megafossils, evaporites, information on large-scale geologic forces, and that diagenesis will generally either and conodont samples. that control curve shape. increase the ratio or leave it unaffected. To obtain relevant data for construction Wickman (1948) predicted that decay They pointed out, however, that under cer- of the 87Sr/ 86Sr seawater curve, nonmarine tain circumstances diagenesis can decrease samples and extensively altered marine the ratio (Veizer and Compston, 1974, p. samples were excluded from the data set. 1464). Since 1974, additional work on the For the Cenozoic part of the curve, we •Present addresses: (Denison) Suite 616, One 87 86 temporal variation of Sr/ Sr of marine used only DSDP samples because they are Energy Square, 4925 Greenville Avenue, Dallas, carbonates and phosphates has been pub- Texas 75206; (Nelson) 2516 West Five Mile better dated and are more likely to have Parkway, Dallas, Texas 75233. lished by Brass (1976), Tremba and others retained their original marine ratios than

516 GEOLOGY, v. 10. p. 516-519, OCTOBER 1982 MILLIONS OF YEARS

Figure 1. Plot of 87Sr/86Sr age for 744 of 786 marine samples. 87Sr/86Sr values for the 42 modern marine samples (Table 1) are not shown. Mod- ern values, however, were accounted for in drawing band and line. For any given time, correct seawater ratio probably lies within band. Line represents our best estimate of seawater ratio versus time. Pre-Cenozoic ages are based on van Eysinga (1975). Cenozoic ages are based on time scale provided by L. B. Gibson (1980, personal commun.). Pliocene-Pleistocene boundary is at 1.62 m.y. B.P., and Tertiary stage boundar- ies are at 5.0, 23.5, 37.0, and 53.5 m.y. B.P.

other samples available to us. For the most MEASUREMENTS RESULTS AND DISCUSSION part, our Mesozoic and Paleozoic samples The samples were dissolved with either Our data are presented graphically in are of lower quality because they have been hydrochloric or nitric acid, and Sr was Figure 1 (age, isotopic ratio, and other subjected to more complex postdeposi- separated from the bulk of other elements data for the individual samples will be tional alteration than our Cenozoic sam- either by means of ion exchange columns presented in a more complete future ples. We have found empirically that or by precipitation from approximately paper). Seawater Sr is assumed to have tighter clustering of 87Sr/86Sr values among 90% nitric acid. Isotope ratios were meas- been sufficiently well mixed at all times so coeval Mesozoic and Paleozoic samples is ured by comparison with standard SrCOi that for any particular time, seawater achieved when samples with low strontium (NBS SRM 987) for which a ratio of 87Sr/86Sr would have been essentially contents or high insoluble residues are 0.71014 has been assumed. Measurements constant throughout the oceans of the eliminated. Thus, our Mesozoic and were made using a second-order double world. General isotopic agreement among Paleozoic data are limited to samples that focusing Nier-Johnson-type mass spec- coeval samples from diverse geographic contain at least 200 ppm Sr and not more trometer with a 60°, 13-in. radius of curva- locations lends support to this assumption than 10% dilute acid insoluble residue. The ture magnetic sector and a 91°, 15.8-in. (Peterman and others, 1970; Veizer and probable explanation for the improved radius of curvature electric sector. Masses Compston, 1974; this paper). In addition, clustering is that the restriction decreases 85, 86, 87, and 88 were collected simul- measurements of 42 modern marine sam- the fraction of samples that have not taneously in four separate faraday cups. ples (Table 1) reaffirm the homogeneity retained the marine 87Sr/86Sr value charac- 87Sr/86Sr values have been corrected of 87Sr/86Sr in modern seawater. 87 teristic of their time of deposition. While for the presence of Rb and have been The gray band in Figure 1 includes most 86 88 this restriction reduces the amount of scat- normalized to Sr/ Sr = 0.1194. Precision of the data points (93%). For any given ter in the Mesozoic and Paleozoic data, it is estimated to be ± 0.00005 at the 95% time, the correct seawater ratio probably does not compensate for a basic difference confidence level. lies within this band. Points outside the in quality between DSDP samples and On the basis of a ratio of 0.71014 for band (7%) probably represent either non- other samples. This contrast is shown in standard SrC03 (NBS SRM 987), meas- marine samples that did not obtain the Figure I by a tighter clustering of coeval ured ratios of 42 modern marine samples 87Sr/86Sr of contemporaneous seawater data points over the Cenozoic part of the average 0.70907, with a standard deviation or diagenetically altered marine samples curve as compared with clustering of co- from the mean of 0.00004 (Table 1), and 36 whose strontium was contaminated by for- eval data points over the pre-Cenozoic part measurements of Eimer and Amend SrCOi eign strontium with a different ratio. of the curve. Analysis of the remaining lot 492327 average 0.70797, with a stand- Because most samples can be confidently sample variation is presented below. ard deviation from the mean of 0.00003. interpreted to be of marine origin, on the

GEOLOGY, OCTOBER 1982 517 basis of other geologic criteria, diagenetic modification is the most likely cause of the scatter. Points occur above or below the TABLE I. 87Sr/86Sr VALUES FOR MODERN MARINE SAMPLES band, depending on the strontium isotopic composition of the terrain from which the Sample type Location 87Sr Sr/66SR diagenetic strontium was derived. Most of (ppm) Pelecypods these points occur above the band and probably reflect the predominance of sam- Donax variabiles texasiana Mustang Island, Texas 1,564 0. 70912 ples from cratonic areas. Old sialic rocks Donax variabiles Gulf Shores, Alabama 1,424 0. 70905 from cratons are important sources of Tel 1 ina radiata JouIters Cay, Bahamas 2,324 0. 70900 radiogenic 87Sr, and weathering of these Tel 1ina tayloriana Mustang Island, Texas 1,741 0. 70905 Myt i1 us edu1i s San Luis Obispo County, Cal ifornia 1,092 0. 70907 rocks yields strontium with a high Myt i1 us edu1i s Brighton Beach, England 922 0. 70907 87Sr/86Sr value. Incorporation of this Myti1 us edu1 is Thesa1 on i k i, Greece 978 0. 70904 strontium into carbonates during diagene- Dosina discus Mustang Island, Texas n.d. 0. 70902 sis alters the original marine ratios and Anandra transversa South Padre Island, Texas 1,875 0. 70909 87 86 Crassostrea virginica South F'adre Island, Texas 835 0. 70907 may lead to higher Sr/ Sr values. Points located below the band represent princi- Gastropods pally samples from convergent plate mar- gins and oceanic settings. These areas Crepi du la fornicata Mustang Island, Texas 1,443 0. 70906 contain mafic volcanic rocks and mafic Sinum perspectivum Mustang Island, Texas 2,498 0. 70901 Arch i tecton i ca nob i1 Mustang Island, Texas 1,357 0. 70904 intrusive rocks that are major sources of 01 iva sayana Mustang Island, Texas 1,334 0. 70903 low-ratio strontium. Weathering of such Turritella gonostoma Gulf of California, Sonora, Mexico 1,288 0. 70906 rocks yields strontium with a low 87Sr/86Sr Turritella sp. Island, Indonesia 0. 70908 Penane n.d. value. Incorporation of this strontium into Murex trunculus Farwa sland, Tunesia 1,665 0. 70908 carbonates during diagenesis may lead to Worms development of lower ratios than the origi- nal marine values. Spirula spirula Mustang Island, Texas 3,472 0.70902 For most of the Phanerozoic, the width BarnacI es of the band is significantly greater than can be attributed to experimental error. Four Bal anus sp. Eugene Island, Louisiana 3,472 0.70907 possible reasons for this are the following: Bal anus sp. Offshore Platform, Nigeria 3,847 0.70909 (1) Diagenesis may have changed the ratio Ech i noderms of many of our samples by small amounts even though it produced large changes in Mel Iita quinquiesper Mustang Island, Texas 1,912 0.70904 only a few. The lower boundary of the Shark and ray teeth cluster of data points is more clearly defined than the upper boundary. If dia- Sphyrna lewini Guymas, Mexico 1,100 0.70909 genesis is the primary cause of the spread, Lamna nausu Venice, Italy 1,090 0.70907 then this boundary relationship would be My Iiobates californicus San Francisco Bay, California n.d. 0.70910 expected in view of the predominance of Raja binoculata San Diego, California 1,162 0.70909 samples from old cratonic areas. (2) There may be errors in assigning relative ages to Carbonate sediments various samples. Such errors would be Coral Couch Key, Florida 8,407 0.70911 most significant during intervals when the Coral Molasses Reef, Florida n.d. 0.70899 ratio is changing rapidly. We believe that Coral Miami, Florida 6,850 0.70903 this problem does contribute to the spread Mol 1 uses Santa Barbara, California 2,262 0.70901 Mol 1 uses Gulf of Mirabella, Crete 1,617 0.70911 Of values for these intervals. (3) Small but Mo 11 uses Island of Rhodes 2,389 0.70902 rapid changes in the ratio may exist that Mol 1 uses ThesaIoniki, Greece 1,555 0.70909 can only be resolved with greater precision 0:70914 Mol 1 uses Penane Island, Indonesia 1,752 in specifying relative ages and, at least for Mol 1 uses Penane Island, Indonesia 1,655 0.70909 some parts of the column, with more sam- Mol 1 uses Cape Blanco, Oregon 4,230 0.70909 Ooliths South Cat Cay, Bahamas 10,531 0.70909 ples per unit of geologic time. (4) The Sr in Ooliths Yucatan, Mexico 799 0.70902 the water from which some samples formed Ool iths JouIters Cay, Bahamas 10,346 0.70899 may not have been perfectly mixed with Sr Lime Mud Williams Island, Bahamas 6,867 0.70908 of the open seas. We believe this potential Lime Mud Florida Bay, Florida 4,790 0.70914 L i me Mud Florida Bay, Florida 5,253 0.70913 effect is not a significant cause of spread in Seawater Coastal Saudi Arabia, the data. The line through the data points Arabian Gulf n.d. 0.70911 represents our best present estimate of the actual seawater ratio as a function of geo- Note: Mean = 0.70907; standard deviation from mean = 0.00004. logic time. It is based on knowledge of the geology of the sampled areas, petrography of the samples, and the Sr content and the dilute acid insoluble content of the sam-

518 GEOLOGY. OCTOBER 1982 pies. Also, because of diagenetic processes of continents, extent of continental inunda- REFERENCES CITED and because of a predominance of samples tion by epeiric seas, variation in climate, Brass, G. W., 1976, The variation of marine 87 86 from cratonic areas, the correct curve and extent of volcanic activity. Many or all Sr/ Sr ratio during Phanerozoic time: Interpretation using a flux model: Geochim- probably lies toward the bottom of the of these factors may be related to global ica et Cosmochimica Acta, v..40, band unless there are contrary indications. tectonic processes, yet their combined p. 721-730. The 87Sr/86Sr curve (Fig. 1) exhibits effect on the temporal variations of seawa- Clauer, N., 1976,87Sr/86Sr composition ofevap- complex structure. Strontium isotopic ter 87Sr/86Sr may hamper a plate-tectonic oritic carbonates and sqjphates from Mio- cene sediment cored in the Mediterranean ratios of seawater during Late Cambrian to interpretation of some parts of the curve Sea (DSDP, Leg 13): Sedimentology, v. 23, Early Ordovician time are close to the (for example, Late Permian). Consequently, p. 133-140. modern value of 0.70907. Within the inter- a systematic evaluation of the global geo- Dasch, E. J., and Biscaye, P. E., 1971, Isotopic vening span of time, the seawater ratio logic history of specific time intervals is composition of strontium in Cretaceous-to- fluctuated below the modern value, with required to reveal the tectonic implications Recent pelagic foraminifera: Earth and Planetary Science Letters, v. 11, p. 201-204. lowest ratios, ~ 0.7068, occurring in Juras- of the 87Sr/86Sr curve. The curve of sea- 87 86 Faure, G., Hurley, P. M., and Powell, J. L., sic and Late Permian time. water Sr/ Sr versus geologic time pro- 1965, The isotopic composition of stron- Seawater continually receives strontium vides a basis for dating marine carbonate, tium in surface water from the North Atlan- from the weathering of continents and evaporite, and phosphate samples. How- tic Ocean: Geochimica et Cosmochimica 87 86 Acta, v. 29, p. 209-220. from the weathering and hydrothermal ever, success of Sr/ Sr dating depends Faure, G., Assereto, R., and Tremba, E. L., on the availability of criteria for recogniz- alteration of the seafloor, particularly sea- 1978, Strontium isotope composition of floor basalts at spreading centers. Tem- ing samples that have retained their origi- marine carbonates of Middle to 87 86 poral variations in the 87Sr/86Sr of nal Sr/ Sr values. We have found that Early Jurassic age, Lombardic Alps, Italy: seawater represent changes in the contribu- the most reliable carbonate samples in our Sedimentology, v. 25, p. 523-543. Kovach, J., 1980, Variations in the strontium tion of strontium from various isotopically data set are lime mudstones and lime isotopic composition of seawater during wackestones with high strontium contents distinct sources within the crust. The most Paleozoic time determined by analysis of important of these are old sialic rocks from and low insoluble residues. These findings, conodonts: Geological Society of America continental interiors with 87Sr/86Sr values however, do not constitute an adequate Abstracts with Programs, v. 12, p. 465. that average 0.720, mafic volcanic and basis for evaluation of the full range of Peterman, Z. E., Hedge, C. E., and Tourtelot, H. A., 1970, Isotopic composition of stron- intrusive rocks from active plate margins marine samples that might be dated by tium in seawater throughout Phanerozoic 87 86 the 87Sr/86Sr method. and ocean basins with Sr/ Sr values that time: Geochimica et Cosmochimica Acta, v. average 0.704, and marine carbonates and 34, p. 105-120. 87 86 sulfates with Sr/ Sr values that average SUMMARY AND CONCLUSIONS Pitman, W. C., Ill, 1978, Relationship between eustacy and stratigraphic sequences of pas- 0.708 (Faure and others, 1965). While the 87 86 The curve of seawater Sr/ Sr versus sive margins: Geological Society of America last source is important in terms of the geologic time (Fig. 1) exhibits an interest- Bulletin, v. 89, p. 1389-1403. quantity of strontium contributed to sea- ing structure, with many episodes of Tremba, E. L., Faure, G., Katsikatsos, G. C., water, it is much less important as a cause increasing and decreasing values. The con- and Summerson, C. H., 1975, Strontium isotope composition in the Tethys Sea, of temporal variation of the seawater ratio figuration of the curve appears to be (for example, see Brass, 1976) because the Euboea, Greece: Chemical Geology, v. 16, strongly influenced by the history of plate p. 109-120. original ratios of marine carbonates and interactions and seafloor spreading van Eysinga, F.W.B., 1975, Geological time sulfates of all Phanerozoic ages are con- throughout the Phanerozoic. Diagenetic scale (third edition): Amsterdam, Elsevier. R7 86 fined within the narrow range of 0.7067 to modifications of original marine 87Sr/86Sr Veizer, J., and Compston, W„ 1974, Sr/ Sr composition of seawater during the Phane- 0.7091, as shown in Figure 1. High ratios, values are often interpretable within a spe- rozoic: Geochimica et Cosmochimica Acta, then, indicate a dominance of old sialic cific geologic province. Analysis of 87 86 v. 38, p. 1461-1484. sources in determining the marine Sr/ Sr 87 86 Sr/ Sr data, therefore, may provide Wickman, F. W., 1948, Isotope ratios—A clue 87 86 values, and low Sr/ Sr values indicate a useful information on regional diagenetic to the age of certain marine sediments: dominance of mafic sources (Faure and patterns and processes. The configuration Journal of Geology, v. 56, p. 61-66. 87 86 others, 1978). The shape of the Sr/ Sr of the curve provides a basis for dating curve, therefore, can be influenced signifi- ACKNOWLEDGMENTS many marine carbonate, evaporite, and We thank Mobil Research and Development cantly by the history of plate interactions phosphate samples. However, broad appli- Corporation for permission to publish this and by rates of seafloor spreading cation of this method depends upon devel- paper. throughout the Phanerozoic. For example, opment of criteria for recognizing samples 87 86 Manuscript received March 22, 1982 the systematic rise in seawater Sr/ Sr that have retained their original marine Revised manuscript received June I, 1982 through the Tertiary indicates a relative 87 86 Sr/ Sr values. Manuscript accepted June 2, 1982 decrease in contribution of strontium from oceanic sources and a relative increase in contribution of strontium from continental sources. This change may reflect an overall decrease in both oceanic ridge volumes and Reviewer's comment spreading rates throughout the Tertiary (Pitman, 1978). The actual circumstances A well-written summary of an extraordinary data set. leading to an increase or decrease in the ratio, however, probably involve a variety of factors, such as variation in lithologic composition of the crust exposed to weath- ering, configuration and topographic relief

GEOLOGY. OCTOBER 1982 PRINTED IN U SA 5)9