Cryptic Striations in the Upper Mantle Revealed by Hafnium Isotopes in Southeast Indian Ridge Basalts

Seediscussions,stats,andauthorprofilesforthispublicationat:http://www.researchgate.net/publication/7253757

CrypticstriationsintheuppermantlerevealedbyhafniumisotopesinSoutheastIndianRidgebasalts

ARTICLEinNATURE·APRIL2006

ImpactFactor:41.46·DOI:10.1038/nature04582·Source:PubMed

CITATIONS

34

READS

71

5AUTHORS,INCLUDING:

DavidW.Graham

OregonStateUniversity

112PUBLICATIONS3,953CITATIONS

SEEPROFILE

JanneBlichert-Toft

EcolenormalesupérieuredeLyon

309PUBLICATIONS9,509CITATIONS

SEEPROFILE

KenHRubin

UniversityofHawaiʻiatMānoa139PUBLICATIONS1,610CITATIONS

SEEPROFILE

FrancisAlbarède

EcolenormalesupérieuredeLyon

426PUBLICATIONS13,785CITATIONS

SEEPROFILE

Availablefrom:JanneBlichert-Toft

Retrievedon:01October2015

© 2006 Nature Publishing Group

Cryptic striations in the upper mantle revealed byhafnium isotopes in southeast Indian ridge basaltsD. W. Graham1, J. Blichert-Toft2, C. J. Russo1, K. H. Rubin3 & F. Albarede2

The Earth’s mantle is isotopically heterogeneous on length scalesranging from centimetres to more than 104 kilometres1,2. Thisheterogeneity originates from partial melt extraction and platetectonic recycling, whereas stirring during mantle convectiontends to reduce it. Here we show that mid-ocean ridge basaltsfrom 2,000 km along the southeast Indian ridge (SEIR) display abimodal hafnium isotopic distribution. This bimodality revealsthe presence of ancient compositional striations (streaks) in theIndian Ocean upper mantle. The number density of the streaks isdescribed by a Poisson distribution, with an average thickness of,40 km. Such a distribution is anticipated for a well-stirred uppermantle, in which heterogeneity is continually introduced by platetectonic recycling, and redistributed by viscous stretching andconvective refolding.

The SEIR stretches from the Rodrigues Triple Junction (25.68 S,70.18 E) to the Macquarie Triple Junction (628 S, 1518 E). Between768–788E it crosses the Amsterdam–St Paul (ASP) plateau, a pro-nounced bathymetric swell associated with relatively hot mantleupwelling beneath the Amsterdam and St Paul islands, while between1208 E and 1288 E it crosses the Australian–Antarctic discordance(AAD), a region of deep bathymetry (.4,000 m) associated withrelatively cold mantle and low melt production3. Notably, over adistance of ,2,500 km, between 868 E and 1208 E, there is a regulareastward decrease in axial depth from 2,300 to 5,000 m, and amorphological transition from axial high to axial valley due todecreasing melt production rate and crustal thickness. This depthgradient occurs at an intermediate and uniform spreading rate(70–75 mm yr21 full rate) and in the absence of large transformoffsets and nearby mantle hotspots. The range in axial depth andridge morphology is similar to the global range for spreading ridgesaway from hotspots, making the SEIR a regional-scale analogue of the50,000-km-long global ocean ridge system. Previous work hasestablished that the He, Pb, Sr and Nd isotope variations along theSEIR are primarily controlled by variation in the depth of melting ofisotopically heterogeneous mantle4–7. Also, all SEIR lavas west of theAAD are true ‘Indian-type’ on the basis of their elevated 208Pb/206Pbratios6.

New Hf isotope results have been obtained for 48 SEIR basaltspreviously analysed for He-Ne-Ar and Pb-Nd-Sr isotope compo-sitions4–7 (see Supplementary Table 1). All samples are fresh mid-ocean ridge basalt (MORB) glasses that were microscopically hand-picked to be free of surface alteration. Between 300 and 600 mg of thisglass was digested and the Hf separated using ultrapure reagents andfollowing established techniques8. The new results show a þ5.5 toþ17.8 range in 1Hf (defined in Fig. 1). The extreme 1Hf values forthe data presented here occur on the ASP plateau (þ5.5) and in thewesternmost AAD (þ17.8), and are within the range of valuesmeasured previously in those areas9–12. Broadly speaking, Hf andNd isotopes in our sample suite show the positive correlation typical

of most oceanic basalts13. However, in detail the Hf isotope variationsare not simply related to axial depth, ridge segmentation, MORB typeor Sr-Nd-Pb-He isotopic variations.

The striking attribute of this new data set, previously unobservedin MORBs, is the Hf isotopic bimodality for lavas erupted between888E and 1108 E (Fig. 2). Over this 2,000-km length of activelyspreading ridge the two groupings show a ‘gap’ of about one epsilonunit (1Hf ¼ þ9.5 to þ11.5 and þ12.5 to þ14.6, respectively). This1Hf gap is significantly larger than the analytical uncertainty (2jexternal precision is 0.3 epsilon units). Because our sample suite has astrong spatial resolution, the results suggest the presence of striations(streaks) in the upper mantle beneath the SEIR. The 1Hf bimodalityis not observed in other geochemical parameters, including Ndisotope composition, indicating that the streaks carry a crypticmemory of ancient chemical fractionation that is now only apparentin the time-integrated Lu/Hf ratio. Because 1Nd along this section ofthe SEIR is not bimodal, Lu–Hf and Sm–Nd fractionation must havebeen decoupled at some point in the evolutionary history of theunderlying upper mantle.

There are several possible origins for the bimodality in Hf isotopesand decoupling of Nd and Hf isotopes, most of which involve mixingwith recycled mantle components (Fig. 3). One possibility is aremnant of primordial heterogeneity, perhaps resulting from thepresence of a deep magma ocean. Given the efficiency of mantleconvection in eradicating such remnants14, this explanation seems

LETTERS

Figure 1 | Along-axis variations in 1Hf for basalt glasses from theSEIR. 1Hf ¼ ð176Hf=177Hfmeasured=

176Hf=177HfBSE 2 1Þ£ 104; where BSE isthe bulk silicate Earth reference value of 176Hf/177HfBSE ¼ 0.282772 (ref. 16).New data from this study are shown as solid circles; other data are fromrefs 9–12. The dashed box outlines the area shown in Fig. 2a.

1College of Oceanic and Atmospheric Sciences, Oregon State University, Corvallis, Oregon 97331, USA. 2Laboratoire des Sciences de la Terre, Ecole Normale Superieure,69364 Lyon, France. 3SOEST, University of Hawaii, Honolulu, Hawaii 96822, USA.

Vol 440|9 March 2006|doi:10.1038/nature04582

199

© 2006 Nature Publishing Group

unlikely, but it cannot be completely ruled out15. A second possibilityis ancient melting that involved a variable amount of garnet restite,such as during komatiite formation16. The observed north–south 1Hfgradient in depleted Atlantic MORB and its relation to distance fromthe continents can be accounted for in this way17,18.

A third possibility involves mixing of different proportions oftectonically subducted components (altered oceanic crust/lithosphere,pelagic sediment and fluid-modified mantle wedge from subductionzones) with mantle peridotite. Fluid-modified, melt-depleted mantlewedge beneath subduction zones may display 1Hf lying above theHf–Nd mantle array19. Variability in the mixing proportion ofrecycled sediment would produce a covariation in 1Nd–1Hf, andis therefore unlikely, by itself, to account for the observed Hfisotope bimodality. However, variability in the proportion of pelagicsediment plus hydrothermally altered crust might produce thebimodality, if Hf/Nd ratios in the recycled material range to bothhigher and lower values than Hf/Nd in the ambient upper mantle(in which case mixing curves in Hf–Nd isotopic space would divergeand could be strongly hyperbolic).

A fourth possibility is a difference in the mineralogic/lithologicmake-up of the mantle source. For example, clinopyroxene in theresidue of mantle melting can develop radiogenic 1Hf at relativelyinvariant 1Nd over 108-year timescales, owing to its depletion in high-field-strength elements20. However, there is no evidence for covaria-tion between Hf isotopes in SEIR basalts and potential indicators ofmodal clinopyroxene abundance or mantle source fertility, suchas Sc/Nb or CaO/Al2O3 ratios. A fifth possibility is recycling ofmid-crustal amphibolite or high-temperature granulite facies rocksin which rutile and iron–titanium oxides control Lu/Hf and theextent of Hf depletion, as evidenced by significant Lu/Hf–Sm/Nddecoupling in crustal xenoliths from southern African kimberlites21.

Lastly, garnet in some cratonic peridotites from South Africankimberlites shows 1Hf ¼ þ200 to þ400, while 1Nd ranges onlybetween 0 and þ15 (ref. 22). The significant potential for Hf–Ndisotopic decoupling by recycling of continental lithospheric mantlemay account for the isotopic variations observed along the southwestIndian ridge23.

Regardless of their exact origin, the different mantle source regionsbeneath the SEIR sampled by the MORB Hf isotope bimodality haveprimarily witnessed a history of coupled Lu/Hf–Sm/Nd fractionationsimilar to most of the oceanic mantle, because the bimodal popu-lations lie close to the modern 1Hf–1Nd array for oceanic basalts andas a group adhere to the overall trend in this diagram (Fig. 3).

Spatially, the streaks defined by the Hf isotope bimodality are welldescribed as a Poisson distribution, in which the number of Hfisotope ‘toggles’ between the two groupings is proportional to thelength of ridge sampled (Fig. 4a). A Poisson distribution describesthe total number of independent events that occur within a specifiedinterval for a fixed mean value of the ‘arrival rate’; it is characterizedby a number of events proportional to the length of the observationspan (for example, as exemplified by radioactivity). The number ofHf isotope ‘toggles’ in SEIR basalts closely follows this predictionover a ridge length of .2,000 km, as expected for a Poisson distri-bution of cryptic streaks in the underlying mantle. Moreover, whenthe number of striation boundaries follows a Poisson distribution,the intervals between boundaries (that is, the striation thicknesses)follow an exponential distribution, as observed in the current data set(Fig. 4b). The Hf isotope ‘toggles’ are not as well-discerned outsidethe 888 E to 1108 E section of the SEIR, and seem to be absent wherethe geodynamic setting is more complex, such as on top of the ASPplateau and within the AAD. Given the sampling density along theSEIR, the mean thickness of the striations in the underlying mantleappears to be ,40 km (Fig. 4a inset).

Let us define a geochemical heterogeneity as a point on theboundary between two types of mantle, for example, a mantle residueof partial melting versus recycled crust or lithosphere. A Poissondistribution of geochemical heterogeneities should be anticipated asthe natural consequence of a well-stirred upper mantle in whichheterogeneities are continually created by tectonic recycling andredistributed by convective stretching and refolding. The meanstriation thickness is therefore a useful parameter for quantifying

Figure 3 |Global 1Hf–1Nd correlation for,2,100 oceanic basalts. Data arefrom the literature and from J.B.-T.’s unpublished database. The dashed lineis the linear regression for all data (1Hf ¼ 1.31Nd þ 3.3; R2 ¼ 0.70). Thebimodal distribution of SEIR basalts is illustrated by the black circles; basaltsfrom the Amsterdam and St Paul islands11, shown by the triangles, display asimilar bimodality in 1Hf. Examples of potential recycled endmembers thatmay be involved in mantle mixing are included for comparison(AG ¼ Australian granulite, PS ¼ pelagic sediment, SAG ¼ South Africangranulite; for example, see ref. 23).Figure 2 | The Hf isotope bimodality. a, Detailed view of the bimodal 1Hf

region between 888E and 1108E. The data clearly show two groupingsseparated by more than one epsilon unit, with high and low valuesdelineated by the alternating stripes numbered 1 to 18 from west to east. Thesolid curve is a smoothed running mean using a gaussian spatial filter thathas a standard deviation of 150 km. b, The histogram (n ¼ number ofsamples) shows the deviations (D) from the running mean in a. Nd isotopeshave been treated similarly to the Hf isotope data and are shown forcomparison. Student’s t-test indicates that the probability of the Hf isotopesbeing drawn from a single population is ,,0.1%.

LETTERS NATURE|Vol 440|9 March 2006

200

© 2006 Nature Publishing Group

mantle strain rate24. Strain rates associated with mantle convectionare of the order of 10215 s21 (ref. 25). After 40 million years (Myr),recycled material would have been stretched and reduced to about50% of its original thickness via shear strain, and after 300 Myr toabout 10%. (Corresponding timescales for stretching by normalstrain are shorter, ,25 and 80 Myr, respectively; SupplementaryFig. 1.)

The timescale of 300 Myr is similar to that inferred from com-parative U–Pb, Sm–Nd and Rb–Sr isotope systematics of enrichedand depleted MORB worldwide. This timescale potentially repre-sents a convective cycling time within the mantle, either for oceaniclithosphere and metasomatized mantle subducted beneath islandarcs or for continental lithosphere delaminated during rifting, to theeventual resampling by partial melting beneath mid-oceanridges26–28. Notably, the Pb, Nd, Sr and Hf isotopes in SEIR basaltsare correlated with their respective parent/daughter ratios, and eachof these correlations also imply ‘ages’ of 200–400 Myr (ref. 6 andSupplementary Fig. 2). However, these age estimates have largeuncertainties, and probably represent minimum values becausethe slopes of the correlations could also have been affected by varyingU/Pb, Sm/Nd, Rb/Sr and Lu/Hf ratios during melting.

The narrow range of 1Hf of ,5 units in the bimodal region makesit currently impossible to discriminate between the possible originsof the streaks, because diagnostic geochemical signatures associatedwith the small extent of Lu/Hf–Sm/Nd decoupling are extremelyweak in this sample suite. Nevertheless, the fact that the Amsterdamand St Paul islands display the same Hf isotope bimodality11 as theSEIR MORBs studied here (Fig. 3) suggests a significant role forupwelling plumes in convective dispersion and refolding of suchmantle streaks, and is further evidence that regions of the uppermantle far from any active hotspot influence still retain a record ofpast pollution by mantle plumes29.

The isotopic compositions of mantle-derived materials provideconstraints on crust/mantle differentiation and planetary evolution,while their spatial distribution is fundamentally linked to mantledynamics. The Hf isotope bimodality along the SEIR represents the

first observational evidence that a Poisson distribution of hetero-geneous streaks characterizes large sections of the upper mantle. Wespeculate that this distribution is more easily recognized along theSEIR than along other sections of the mid-ocean ridge system,because the SEIR is uncomplicated by the effects of nearby mantlehotspots, continental land masses, or large fracture zone offsets.Striation thickness distributions in realistic, time-periodic flows ofviscous fluids, such as those appropriately describing aspects of theEarth’s mantle, are extremely difficult to model numerically giventhe computational state-of-the-art. However, striation thicknessdistributions can be accurately predicted from a knowledge of thestretching values30. A Poisson distribution of mantle heterogeneities,and the associated exponential distribution of striation thicknesses, isa fundamental mantle property that should be taken into account inforthcoming models of mantle convection and rheology.

Received 3 August 2005; accepted 12 January 2006.

1. Hart, S. R. A large-scale isotope anomaly in the Southern Hemisphere mantle.Nature 309, 753–-757 (1984).

2. Allegre, C. J. & Turcotte, D. L. Implications of a two-component marble-cakemantle. Nature 323, 123–-127 (1986).

3. Sempere, J.-C., Palmer, J., Christie, D., Morgan, J. P. & Shor, A. Australian–-Antarctic discordance. Geology 19, 429–-432 (1991).

4. Graham, D. W., Lupton, J. E., Spera, F. J. & Christie, D. M. Upper mantledynamics revealed by helium isotope variations along the southeast Indianridge. Nature 409, 701–-703 (2001).

5. Nicolaysen, K. E. The Cretaceous to Recent History of the Southeastern IndianOcean: The Role of Mantle Plumes Examined by the Geochronology and Sr, Nd, Pband He Isotopic Geochemistry. 1–-274, PhD dissertation, Mass. Inst. of Technol.(2001).

6. Mahoney, J. J. et al. Between a hot spot and cold spot: isotopic variation in theSoutheast Indian Ridge asthenosphere, 868-1188E. J. Petrol. 43, 1155–-1176(2002).

7. Burnard, P., Graham, D. W. & Farley, K. A. Fractionation of noble gases (He, Ar)during MORB mantle melting: a case study on the Southeast Indian Ridge.Earth Planet. Sci. Lett. 227, 457–-472 (2004).

8. Blichert-Toft, J., Chauvel, C. & Albarede, F. Separation of Hf and Lu for high-precision isotope analysis of rock samples by magnetic sector-multiplecollector ICP-MS. Contrib. Mineral. Petrol. 127, 248–-260 (1997).

Figure 4 | Poisson character of upper mantle streaks beneath the SEIR.a, Cumulative number of Hf isotopic striations versus distance (inkilometres) along the SEIR. The line shown is a linear regression(y ¼ 0.0107x þ 0.487, R2 ¼ 0.992, n ¼ 29; 1j uncertainties on the slopeand intercept are 0.00099 and 0.988, respectively). The mean thickness of thestriations, based on the slope of this diagram, would be 93 km (^9 km, 1j).The inset illustrates how the inferred mean thickness of the striations isaffected by sampling density. A large population with a uniform density forthe probability of a transition between two different mantle ‘flavours’ hasbeen approximated by 1,000 points spread randomly along a line. Thedistance between two consecutive points represents the thickness of anindividual striation. This model has an exponential distribution of striationthicknesses, consistent with the observed relationship between cumulative

number of transitions and distance. As expected, the proportion of striationsrecovered during sampling increases as the sampling density (that is,number of samples per striation) increases. Because the sampling densityalong the SEIR has 1.6 samples per striation, about 40% of the striationspresent in the underlying mantle have been sampled, indicating that the truemean thickness of the striations is ,40 km. b, Probability diagram of thewidth (thickness) of the Hf isotopic striations. The x axis shows thepercentage of striation widths (normalized to a normal probabilitydistribution) whose value is less than the respective value of the y-axis datapoint. The solid curve is an exponential fit (y ¼ 89.71exp[0.476x],R2 ¼ 0.971) and the dashed line is a linear fit (y ¼ 100.4 þ 48.97x,R2 ¼ 0.86). This good exponential relationship is consistent with a Poissondistribution for the number density of upper mantle striations.

NATURE|Vol 440|9 March 2006 LETTERS

201

© 2006 Nature Publishing Group

9. Chauvel, C. & Blichert-Toft, J. A hafnium and trace element perspective onmelting of the depleted mantle. Earth Planet. Sci. Lett. 190, 137–-151 (2001).

10. Kempton, P. L., Pearce, J. A., Barry, T. L., Langmuir, C. H. & Christie, D. M.Sr-Nd-Pb-Hf isotope results from ODP Leg 187: evidence for mantle dynamicsof the Australian–-Antarctic discordance and origin of the Indian MORB source.Geochem. Geophys. Geosyst. 3, doi:10/1029/2002GC000320 (2002).

11. Doucet, S., Weis, D., Scoates, J. S., Debaille, V. & Giret, A. Geochemical andHf-Pb-Sr-Nd isotopic constraints on the origin of the Amsterdam–-St. Paul(Indian Ocean) hotspot basalts. Earth Planet. Sci. Lett. 28, 179–-195 (2004).

12. Hanan, B., Blichert-Toft, J., Pyle, D. G. & Christie, D. M. Contrasting origins ofthe upper mantle revealed by hafnium and lead isotopes from the SoutheastIndian Ridge. Nature 432, 91–-94 (2004).

13. Salters, V. J. M. & White, W. M. Hf isotope constraints on mantle evolution.Chem. Geol. 145, 447–-460 (1998).

14. Blichert-Toft, J. & Albarede, F. Short-lived chemical heterogeneities in theArchaean mantle with implications for mantle convection. Science 263,1593–-1596 (1994).

15. Albarede, F. in Structure, Composition and Evolution of the Earth’s Mantle (edsvan der Hilst, R. D., Bass, J., Matas, J. & Trampert, J.) 27–-46 (Geophys.Monogr. 160, American Geophysical Union, Washington DC, 2005).

16. Blichert-Toft, J., Arndt, N. & Gruau, G. Hf isotope measurements on Barbertonkomatiites: effects of incomplete sample dissolution and importance forprimary and secondary magmatic signatures. Chem. Geol. 207, 261–-275(2004).

17. Andres, M., Blichert-Toft, J. & Schilling, J.-G. Nature of the depleted uppermantle beneath the Atlantic: evidence from Hf isotopes in normal mid-oceanridge basalts from 798N to 558S. Earth Planet. Sci. Lett. 225, 89–-103 (2004).

18. Blichert-Toft, J. et al. Geochemical segmentation of the Mid-Atlantic Ridgenorth of Iceland and ridge-hotspot interaction in the North Atlantic. Geochem.Geophys. Geosyst. 6, doi:10.1029/2004GC000788 (2005).

19. Woodhead, J. D., Hergt, J. M., Davidson, J. P. & Eggins, S. M. Hafnium isotopeevidence for ‘conservative’ element mobility during subduction zone processes.Earth Planet. Sci. Lett. 192, 331–-346 (2001).

20. Salters, V. J. M. & Zindler, A. Extreme 176Hf/177Hf in the sub-oceanic mantle.Earth Planet. Sci. Lett. 129, 13–-30 (1995).

21. Schmitz, M. D., Vervoort, J. D., Bowring, S. A. & Patchett, P. J. Decoupling ofthe Lu–-Hf and Sm–-Nd isotope systems during evolution of granulitic lowercrust beneath southern Africa. Geology 32, 405–-408 (2004).

22. Bedini, R. M., Blichert-Toft, J., Boyet, M. & Albarede, F. Isotopic constraints onthe cooling of the continental lithosphere. Earth Planet. Sci. Lett. 223, 99–-111(2004).

23. Janney, P. E., le Roex, A. P. & Carlson, R. W. Hafnium isotope and traceelement constraints on the nature of mantle heterogeneity beneath the central

Southwest Indian Ridge (138E to 478E). J. Petrol. 46, 2427–-2464 (2005).

24. Olson, P. L., Yuen, D. A. & Balsiger, D. S. Mixing of passive heterogeneities bymantle convection. J. Geophys. Res. 89, 425–-436 (1984).

25. Schubert, G., Turcotte, D. L. & Olson, P. Mantle Convection in the Earth andPlanets 1–-940 (Cambridge Univ. Press, Cambridge, Massachusetts, 2001).

26. Allegre, C. J. & Lewin, E. Isotopic systems and stirring times of the Earth’smantle. Earth Planet. Sci. Lett. 136, 629–-646 (1995).

27. Albarede, F. Radiogenic ingrowth in systems with multiple reservoirs:applications to the differentiation of the mantle-crust system. Earth Planet. Sci.

Lett. 189, 59–-73 (2001).

28. Donnelly, K. E., Goldstein, S. L., Langmuir, C. H. & Spiegelman, M. Origin ofenriched ocean ridge basalts and implications for mantle dynamics. Earth

Planet. Sci. Lett. 226, 347–-366 (2004).

29. Hanan, B. B., Kingsley, R. H. & Schilling, J.-G. Pb isotope evidence in the SouthAtlantic for migrating ridge–-hotspot interactions. Nature 322, 137–-144 (1986).

30. Muzzio, F. J., Alvarez, M. M., Cerbelli, S., Giona, M. & Adrover, A. Theintermaterial area density generated by time- and spatially periodic 2D chaotic

flows. Chem. Eng. Sci. 35, 1497–-1508 (2000).

Supplementary Information is linked to the online version of the paper atwww.nature.com/nature.

Acknowledgements D.W.G. thanks A. Bouvier, A. Agranier, P. Beck, F. Moynier,K. Koga and E. Koga for their help and hospitality during two extended visits toLyon. We thank P. Telouk for his help with the P54 mass spectrometer.D. Christie, B. Hanan, K. Johnson, J. Mahoney and D. Pyle provided help onmany aspects of SEIR geochemistry. N. Pisias provided insight into some of thenuances of statistical distributions. D.W.G., C.J.R. and K.H.R. were supported bythe Marine Geology division of the NSF, and J.B.-T. and F.A. by the FrenchInstitut National des Sciences de l’Univers.

Author Contributions D.W.G. and J.B.-T. performed the Hf isotopemeasurements. All authors contributed to data analysis.

Author Information Reprints and permissions information is available atnpg.nature.com/reprintsandpermissions. The authors declare no competingfinancial interests. Correspondence and requests for materials should beaddressed to D.W.G. ([email protected]).

LETTERS NATURE|Vol 440|9 March 2006

202

Supplementary Figure Legends.

Fig. S1. Two diagrams are shown illustrating the stretching of a viscous band by shearstrain (A) and normal strain (B). The lines correspond to different degrees of stretching(δ/δ0) of a band of initial thickness δ0. Three cases are illustrated that represent 10%stretching (blue), 50% stretching (green) and 90% stretching (red). As an example, after40 Myrs recycled crust or lithosphere would have been stretched and thinned to about50% of its original thickness via shear strain, and after 300 Myrs to about 10%. Timescales for the same stretching by normal strain are correspondingly shorter, at ~25 and 80Myrs, respectively. Layer thinning is described by δ/δ0=1/(1+ξt) for shear strain and byδ/δ0=exp(-ξ t) for normal strain, where ξ is strain rate and t is time.

Fig. S2. Hf ‘isochron’ diagram for Southeast Indian Ridge MORB glasses and basaltsfrom Amsterdam and St. Paul islands. The trendline corresponds to an age of 400 Myr,equivalent to the linear regression of the new SEIR data (slope=0.00769, r2=0.23). Datasources are [1-2] for the Australian-Antarctic Discordance (AAD), [3] for the SEIR westof the ASP Plateau, and [4] for Amsterdam and St. Paul islands. SEIR data in dark blueare from this study, with Lu and Hf concentration data from [5].

1. Kempton, P. L., Pearce, J. A., Barry, T. L., Langmuir, C. H. & Christie, D. M. Sr-Nd-Pb-Hf isotope results from ODP Leg 187: evidence for mantle dynamics of theAustralian-Antarctic Discordance and origin of the Indian MORB source. Geochem.Geophys. Geosyst. 3, doi:10/1029/2002GC000320 (2002).

2. Hanan, B., Blichert-Toft, J., Pyle, D. G. & Christie, D. M. Contrasting origins of theupper mantle revealed by hafnium and lead isotopes from the Southeast Indian Ridge.Nature 432, 91-94 (2004).

3. Chauvel, C. & Blichert-Toft, J. A hafnium and trace element perspective on melting ofthe depleted mantle. Earth and Planet. Sci. Lett. 190, 137-151 (2001).

4. Doucet, S., Weis, D., Scoates, J. S., Debaille, V. & Giret, A. Geochemical and Hf-Pb-Sr-Nd isotopic constraints on the origin of the Amsterdam-St. Paul (Indian Ocean)hotspot basalts. Earth Planet. Sci. Lett. 28, 179-195 (2004).

5. Christie, D. M., Pyle, D. G. and Sylvander, B. A. Major and trace element compositionof basalts from the Southeast Indian Ridge (Westward cruise, leg 10). LDEOPetrological Database, submitted data set (2004).

StrainRate(s-1)

A. Stretching by Shear Strain

1 10 100 1000

d/d0=0.90

d/d0=0.50

d/d0=0.10

10-14

10-15

10-16

10-17

typical strain rates in mantle convection models

Time (Myr)

Time (Myr)

B. Stretching by Normal Strain

typical strain rates in mantle convection models

StrainRate(s-1)

10-14

10-15

10-16

10-17

1 10 100 1000

Fig. S1

d/d0=0.90

d/d0=0.50

d/d0=0.10

0.2829

0.2830

0.2831

0.2832

0.00 0.01 0.02 0.03 0.04

AADSEIR (ASP Plateau & west)Amst/SP IslandsSEIR (this study)ASP Plateau (this study)

Fig. S2

176Lu/177Hf

176 H

f/177

Hf

T=400 Myr

Supplementary Table 1. Hf Isotope results for Southeast Indian Ridge MORB Glasses

Sample Latitude(°S)

Longitude(°E)

Depth(m)

Distancefrom St. Paul

(km)MORBType

176Hf/177Hf ± εHf

BandNumber

BMRG06 75-1 35.282 78.595 3269 -205.0 N 0.283177 0.000007 14.32BMRG06 73-3 36.066 78.825 2659 -129.3 E 0.283047 0.000008 9.73BMRG06 WC46 36.182 78.953 2734 -112.1 E 0.283086 0.000009 11.10BMRG06 63-1 38.199 78.367 1838 9.4 E 0.282935 0.000005 5.76BMRG06 59-1 39.124 78.144 1903 69.51 E 0.282927 0.000007 5.48BMRG06 43-2 41.252 79.105 2785 296.2 N 0.283124 0.000008 12.45BMRG06 39-1 41.242 81.153 2776 416.5 N 0.283111 0.000005 11.99BMRG06 WC04 42.863 83.290 2953 668.4 N 0.283202 0.000006 15.21BMRG06 36-1 41.098 86.230 2500 722.3 N 0.283133 0.000003 12.77BMRG06 34-1 41.517 87.092 2480 807.4 N 0.283097 0.000005 11.49BMRG06 33-1 42.115 88.042 2450 910.3 N 0.283042 0.000019 9.55WW10 70-32 42.568 90.204 2575 1081.7 N 0.283126 0.000006 12.52 0WW10 71-1 42.886 90.795 2350 1141.6 N 0.283079 0.000009 10.86 1WW10 65-9 43.082 91.095 2494 1174.1 N 0.283121 0.000008 12.34 2WW10 73-8 43.468 91.684 2738 1237.9 N 0.283141 0.000005 13.05 2WW10 75-4 43.579 92.680 2585 1310.4 N 0.283043 0.000008 9.58 3WW10 76-1 43.882 93.114 2653 1358.3 N 0.283081 0.000005 10.93 3WW10 77-7 44.117 93.773 2795 1416.4 N 0.283161 0.000004 13.76 4WW10 78-2 44.833 94.833 2719 1530.6 N 0.283074 0.000008 10.68 5WW10 82-35 45.177 95.702 3164 1608.6 N 0.283153 0.000005 13.47 6WW10 84-7 45.111 95.932 2660 1619.8 N 0.283139 0.000007 12.98 6WW10 WC48 46.090 95.926 2990 1678.4 E 0.283046 0.000005 9.69 7WW10 87-38 46.778 96.368 2520 1748.2 N 0.283143 0.000005 13.12 8WW10 88-1 47.076 96.833 2568 1795.8 N 0.283137 0.000007 12.91 8WW10 89-107 47.443 97.512 2470 1860.8 E 0.283078 0.000006 10.82 9WW10 91-3 47.909 98.601 2900 1957.4 N 0.283126 0.000005 12.52 10WW10 92-1 48.101 98.943 2668 1990.2 N 0.283074 0.000008 10.68 11WW10 96-1 47.335 100.672 2465 2059.3 E 0.283070 0.000006 10.54 11WW10 100-1 47.630 101.530 2857 2131.6 E 0.283060 0.000005 10.18 11WW10 102-1 47.876 102.140 2986 2184.7 N 0.283146 0.000006 13.23 12WW10 105-1 47.765 103.038 2783 2238.1 E 0.283087 0.000007 11.14 13WW10 106-4 47.887 103.353 2985 2265.2 E 0.283150 0.000006 13.37 14WW10 110-4 48.103 103.930 3480 2314.3 N 0.283136 0.000007 12.87 14WW10 111-18 48.212 104.662 3060 2368.3 E 0.283094 0.000005 11.39 15WW10 113-7 48.752 105.224 3630 2431.9 N 0.283097 0.000007 11.49 15WW10 115-3 49.229 105.868 3696 2497.2 E 0.283087 0.000006 11.14 15WW10 116-2 48.873 106.494 4835 2521.6 E 0.283130 0.000005 12.66 16WW10 116-15 48.873 106.494 4835 2521.6 E 0.283135 0.000004 12.84 16WW10 118-1 48.428 107.527 2673 2570.0 E 0.283050 0.000006 9.83 17WW10 125-1 49.450 109.105 3475 2721.0 N 0.283119 0.000009 12.27 18WW10 130-1 49.775 111.133 3520 2869.8 N 0.283146 0.000007 13.23 18WW10 132-1 50.212 111.783 3328 2930.6 E 0.283095 0.000006 11.42WW10 133-1 50.323 112.325 3484 2971.0 N 0.283141 0.000005 13.05WW10 134-1 50.295 112.490 3650 2980.9 N 0.283185 0.000013 14.61WW10 138-1 50.187 112.855 3845 3001.1 N 0.283126 0.000007 12.52WW10 141-1 50.347 113.615 3002 3057.9 E 0.283067 0.000006 10.43WW10 145-1 49.273 116.717 4665 3231.3 E 0.283127 0.000007 12.55WW10 145-7 49.273 116.717 4665 3231.3 E 0.283274 0.000005 17.75

Hf chemical separation, and mass spectrometric analyses by MC-ICP-MS (VG Plasma 54), were carried out at ENSLfollowing procedures outlined in ref. 8. 176Hf/177Hf was normalized for mass fractionation relative to 179Hf/177Hf = 0.7325.176Hf/177Hf of the JMC-475 Hf standard = 0.282160 ± 0.000010 (2σ). Hf standards were analyzed for every second sample.Reported uncertainties are the within-run 2-standard errors.