Mantle heterogeneities beneath the South Atlantic - ScienceDirect

3 downloads 0 Views 1MB Size Report
Denis Fontignie a3 * , Jean-Guy Schilling b31 a Dipartement de Miniralogie,. Universite' de GenPue, rue des Maraichers 13, CH-1211 GenPve 4, Switzerland.
EPSL ELSEVIER

Earth and Planetary

Science Letters 142 (1996) 209-221

Mantle heterogeneities beneath the South Atlantic: a Nd-Sr-Pb isotope study along the Mid-Atlantic Ridge ( 3”S-46’S) Denis Fontignie a3*, Jean-Guy Schilling b31 a Dipartement h Graduate

de Miniralogie,

Universite’ de GenPue, rue des Maraichers

School of Oceanography,

University

Received 4 December

13, CH-1211

of Rhode Island. Narragansett,

1995; accepted

GenPve 4, Switzerland

RI 02882-l

197, USA

13 April 1996

Abstract We report on 55 Nd and Sr isotope analyses of Mid-Atlantic Ridge (MAR) basalt glasses from 3”s to 46”S, using the same samples on which Pb and He isotope ratios were reported earlier (Hanan et al. [l] and Graham et al. [2]). Eighteen new Pb, Sr, and Nd isotope analyses are also reported on basalt glasses from 17 stations from the same region. “Sr/ 86Sr ratios range from 0.70212 to 0.70410 and ‘43Nd/‘44Nd from 0.51285 to 0.51331. The along-ridge long wavelength 87Sr/86Sr variation delineated by light-REE depleted basalts increases progressively southward toward Tristan da Cunha. Short wavelength, spike-like, positive 87Sr/86Sr anomalies composed of light-REE enriched basalts are found opposite the Tristan, St. Helena, and Circe (Ascension) hotspots (as for the Pb isotopes). The short and long wavelength ‘43Nd/ lUNd variations anti-correlate with those of *‘Sr/s6Sr. The 17 new Pb isotope analyses confirm both the short and long wavelength trends previously reported by Hanan et al. [I]. These spatial variations, as well as the variations in Nd-Sr-Pb isotopic space fully confirm the mantle plume-ridge interaction model and upper mantle mixing conditions beneath the South Atlantic inferred previously on the basis of Pb isotopic data only (Hanan et al. [l]). However, in Nd-Sr isotopic subspace the Circe and the St. Helena mixing vector are not distinguishable. The short wavelength Nd-Sr-Pb-He anomalies suggest recent lateral sublithospheric channeled flows from these off-ridge plumes to the migrating MAR axis. The long wavelength variations reflect a broad pollution of the asthenosphere by Pb and Sr radiogenic, isotope-rich, mantle material, which has been partly depleted of incompatible elements relatively recently. This broad pollution may be related to the partial melting and dispersion of the Tristan and St. Helena plume heads into the subcontinental asthenosphere, prior to the opening of the South Atlantic. In this two-stage model, the mixing relations in Nd-Sr-Pb space further require that the incompatible element depletion and dispersion of the St. Helena plume into the asthenosphere occurred before that of the Tristan plume head. We emphasize that this two-stage model is based purely on isotope systematics. The implied thermal and dynamic aspects of this model remain to be evaluated and tested. Keywords:

South Atlantic; Mid-Atlantic

Ridge; mantle plumes; Sr-87/Sr-86;

Nd-144/Nd-143;

Pb-208/Pb-204

1. Introduction

Corresponding author. Fax: +41 [email protected] ’ E-mail: jgs@gsosunl .gso.uri.edu l

0012-821X/96/%12.00 PII

Copyright

SOOl2-821X(96)00079-9

22

320

5732.

E-mail:

Ocean island basal& from the South Atlantic have played a key role in the development and understanding of mantle heterogeneities in heat-producing

0 1996 Elsevier Science B.V. All rights reserved.

210

D. Fontignie. J.-G. Schilling/Earth

and Planetary Science Letters 142 (1994) 209-221

and other incompatible elements [3,4]. The isotopic composition of St. Helena and Tristan da Cunha hotspots are extreme and represent two of the 4 or 5 end-member mantle source types so far recognized. On a time-integrated basis over the age of the Earth, the St. Helena source evolved with high U/Pb, low Rb/Sr and intermediate Sm/Nd parent/daughter ratios, representing the ocean island mantle source type referred to as HIMU. In contrast, Tristan da Cunha is close to the high Rb/Sr, low U/Pb and Sm/Nd component referred to as EM1 [51. The composition and spatial variation of Pb isotope ratios in dredged basalts collected along the southern MidAtlantic Ridge (MAR), and their relation to these islands have, so far, also provided the strongest evidence for the mantle plume source-migrating ridge sink model (MPS-MRS) [6,7]. Spike-like geo30”

40”

10

I\\



I

I

chemical anomalies are found on ridge-axis segments located opposite Gough-Tristan da Cunha, St. Helena, and Ascension (Circe> hotspots or mantle plumes. The Pb isotope signals of these anomalies are diagnostic of the corresponding plumes located off the ridge axis and support the sublithospheric flow channel connecting the plume with the westward migrating MAR axis [1,8]. More recently, the MPS-MRS model has also received support from seismic tomography imaging [9]. Because of the importance of further testing of the validity of the MPS-MRS model and gaining further insights into mantle mixing conditions and dynamics, we now report on the Sr and Nd isotopic composition of the very same South Atlantic MORB samples previously studied for Pb and He isotope ratios [1,2] (Fig. 11, and discuss their implications for the dispersal of

20”

10

I

I

I

WO”E

I

I

IO A

I

20”

RICA I

0 D St. Helena

20”

50

I I

1

40”

I

30”

I

20”

I

lo”

\\

osouver WO”E

10”

Fig. 1. Schematic map of the South Atlantic showing the location of dredged basalt glasses along the Mid-Atlantic Ridge which were analyzed for Nd and Sr isotopes (dots). Arrows show locations of samples for which new F’bisotope data are also reportedin this study. Shaded regions show aseismic ridges (reflecting plume tracks).

D. Foniignie,

Table

J.-G. Schilling

/Earth

and Planetary

211

Science Letters 142 (1996) 209-221

1

(La/Sm),.

“Sr/ ‘%r and ‘43Nd/ ‘“Nd in basal& from the Mid-Atlantic

Ridge between 3’S and 46”s on the same samples reported in [l]

for Pb isotope ratios Sample ID a

(La/Sm),

s’Sr/

EN061 3D-lg EN061 4D-lg EN061 5D-lg EN061 6D-IQ EN061 7D- Ig EN061 8D-lg EN061 8D- 1Ag EN061 8D-2Ag EN061 9D-1 EN061 lOD-lg EN061 IOD-2 EN061 lOD-3 EN061 1 lD-1 Ag EN061 12D-lg EN061 13D-lg EN061 14D-lg EN061 ISD-I EN061 16D-lg RC16 6D-1A RC16 7D-lg EN061 17D-lg EN061 18D-lg EN061 18D-2Ag EN061 19D-lg EN061 20D-lg EN061 22D-lg EN061 23D-1Ag EN063 2D-5g EN063 7D-5g EN063 9D-5 EN063 IOD-5g EN063 12D-5g EN063 14D-5g EN063 17D-5g EN063 22D-5g EN063 24D-5g EN063 23D-5g AI1 107-7 2%1A AH 107-7 25-23 AI1 107-7 20-3g AI1 107-7 18-288 AI1 107-7 17-71g AI1 107-7 15-3g AI1 107-7 14-43g AI1 107-7 14-77g AI1 107-7 13-lg AI1 107-7 lo-lg AI1 107-7 9-38 AI1 107-7 7-3g AI1 107-7 7-log AI1 107-7 6-20g

0.560 0.530 0.620 0.560 0.714 1.245 1.245 0.793 1.298 1.398 1.162 1.930 1.655 1.459 0.795 0.684 0.820 0.744 0.748 0.49 1 0.730 2.995 0.982 0.841 0.917 0.759 0.688 0.428 0.614 0.486 0.558 0.486 0.567 0.711 0.576 0.695 0.823 0.986 0.871 0.342 0.516 0.96 I 1.044 1.020 1.435 0.633 0.880 0.812 0.528 0.68 1 0.720

0.702 183 0.702225 0.702 1 I8 0.702232 0.702454 0.702456 0.702458 0.702372 0.703007 0.702536 0.702527 0.702584 0.702772 0.7027 IO 0.702448 0.702487 0.702428 0.702575 0.702499 0.702326 0.702376 0.702918 0.702559 0.702313 0.702446 0.70241 I 0.702453 0.702374 0.702340 0.702676 0.7029 I8 0.702489 0.702493 0.7027 19 0.702829 0.702599 0.702579 0.7029 I6 0.702870 0.702832 0.703068 0.703 I47 0.703445 0.703366 0.704096 0.702937 0.703 162 0.7034s 1 0.7029 I I 0.702927 0.703 157

86Sr

( k 2SE)

‘43Nd/ ‘44Nd

( f 2SE)

(16) (26) (22) (26) (20) (24) (18) (36) (22)

0.5 13262 0.5 13286 0.513310 0.5 I3297 0.513159 0.513123 0.513114

(5) (5) (2) (2) (3)

0.5 12960 0.513158 0.513161

(3)

0.513178 0.5 13038 0.513191 0.5 12992 0.513175 0.513157 0.5 I3076

(8) (2) (5) (4) (7) (5) (3)

0.513118 0.512971 0.513124 0.5 13207 0.513195 0.513154 0.5 13075 0.513152 0.513123 0.5 13083 0.5 13057

(4) (5) (4) (5) (5) (3) (2) (3) (7) (8) (3)

0.5 13061 0.513000 0.513033 0.5 13092 0.513123 0.5 13063 0.5 13046 0.513164 0.5 12998 0.513010 0.5 12952

(7) (7) (4) (4) (7) (2) (4) (7) (3) (2) (2)

0.512851 0.513036

(3) (2) (8) (4) (8) (7) (5)

(8) (26) (26) (24) (26) (24) (26) (28) (22) (26) (28) (24) (24) (20) (36) (20) (26) (36) (18) (20) (16) (12) (14) (14) (14) (16) (18) (30) (22) (18) (14) (30) (16)

(20) (24) (14) (26) (16) (28) (14) (18) (16)

0.5 13037 0.513014 0.5 12999 0.5 13041 0.5 12908

(2) (2)

(2) (8)

212

D. Fontignie, J.-G. Schilling/Earth

and Planetary

Science Letters 142 (1996) 209-221

Table 1 (continued) Sample ID a

(La/Sm),

s7sr/ 86%

( k 2SE)

‘43Nd/ ‘44Nd

(+ 2SE)

AI1 107-7 4-4g AI1 107-7 2-388 AH 107-7 2-53g

0.800 0.936 0.987

0.703062 0.703240 0.70325 1

(7) (16) (20)

0.512982 0.5 12970 0.5 12966

(3) (2) (4)

a g in sample identification stands for glass, otherwise pillow interiors. (La/Sm), = La/Sm ratio normalized to chondrites using La = 0.3 and Sm = 0.21 ppm. Errors in parentheses are the within-run 2 standard errors of the mean of the last significant figures. The Sr isotope ratios measured at URI are mass fractionation corrected to an ‘sSr/ 86Sr of 8.375209, and normalized to the E&A standard with s’Sr/%r = 0.708000, using an average of 0.708068 &-8 X 10m6 (2SE; n = 35) measured from 7/83 to 3/85 [461. The average of the SRM987 standard measured during this period was 0.710307 * 13 X 10m6 (2SE; n = 8). The ‘43Nd/ ‘44Nd ratios were measured in static and/or dynamic mode and are mass fractionation corrected to a ‘46Nd/ ‘44Nd= 0.721903 and normalized to the La Jolla Nd standard = 0.511835. In static mode, an average of 0.511909 + 4 X 10e6 (2SE; n = 60) and, in dynamic mode 0.511828 k 6 X IO-” (2SE; n = 41) were measured from 1l/90 to S/91.

mantle plumes in the South Atlantic, and their relationships to the so-called DUPAL anomaly [lo- 121.

2. Results Table 1 lists the ‘43Nd/ ‘44Nd and *7Sr/ @Sr isotope analyses of the same MORB samples on which we previously reported Pb and He isotope ratios [ 1,2]. The Sr and Nd were extracted from the same sample dissolution as the Pb initially was. Table 2 lists new Pb, Sr, and Nd analyses of 18 MORB from 17 dredge stations in the same region. These stations are highlighted in Figs. 1 and 2. The Pb and most of the Sr isotope analyses were made on a VG Micromass 30B single collector, double focusing, thermal ionization mass spectrometer at the University of Rhode Island. The Nd isotope measurements were made at the University of Geneva on a 7-collector Finnigan Mat 262 thermal ionization mass spectrometer with extended geometry and stigmatic focusing, using double Re filaments. All the separations were conducted in class A-100 clean rooms at URI following the same method as described in Schilling et al. [13]. Reagents blanks for Nd, Sr, and Pb are negligible. Mass fractionation and normalization procedures are given in the footnotes of Tables I and 2.

3. Latitudinal

variations

and segmentation

The latitudinal variation in Pb-Nd-Sr isotope ratios are compared with that of chondrite-normal-

ized (La/Sm), ratios in Fig, 2. The basalt samples with pronounced light-REE depleted patterns ((La/Sm), < 0.65) are shown as filled symbols. Fig. 2 illustrates the close coupling of (La/Sm), with the Pb-Nd-Sr isotope ratios for the short wavelength spike-like anomalies, and the marked decoupling for the long wavelength base line trends (heavy lines). Despite the fact that the long wavelength base line in (La/Sm), is essentially constant, the corresponding long wavelength trends in Pb, Nd, and Sr isotope ratios are highly structured and quite distinct from each other. 87Sr/ 86Sr and *08Pb/ ‘04Pb increase southward towards Tristan da Cunha, *06Pb/ *04Pb and *07Pb/ *04Pb go throu h broad maxima near St. Helena (- 209) and I4$Nd/ ‘44Nd linearly decreases southward and anti-correlates with 87Sr/ 86Sr. The short wavelength, spike-like, positive anomalies in (La/Sm),, 206Pb/ *04Pb, and *‘*Pb/ ‘04Pb observed opposite Ascension, St. Helena, and Tristan-Gough islands have been attributed to recent (current) plume flows along sublithospheric channels connecting the westward migrating MAR axis with the three plume conduits in off-ridge positions [1,7,8]. The Tristan channel would have been established some 70 Ma, when the ridge started migrating away from the Tristan plume, as is evident from the record of the Rio Grande and Walvis plume tracks (e.g. [ 141). The timing of formation of the St. Helena and Circe channels remain unknown. The fact that the long wavelength trend observed in *06Pb/ *04Pb reaches a broad maximum in the vicinity of St. Helena, whereas 208Pb/204Pb peaks near TristanGough has been attributed to the activity and dispersal into the upper mantle of the St. Helena and

4.35 7.87 7.87 12.90 14.19 19.99 20.32 20.79 21.47 22.01 22.96 24.01 25.40 26.52 27.56 28.01 28.91 28.96

12.35 13.45 13.45 14.73 13.95 11.88 II.71 11.60 I 1.58 10.96 13.49 13.30 13.27 13.64 13.44 13.20 12.97 13.50

Longitude

CW

Latitude

(“S)

2303 3282 3560 3270 3460 2329 4440 2105 3370 3270 3935 3640 3410 3890 4140

(ml

Depth

0.585 0.73 I 0.754 0.683 0.640 0.630 0.498 0.504 0.586 0.611 0.561 0.589 0.562 0.550 0.627 0.671 0.727 0.676

(La/Sm),

17.726 18.352 18.413 18.151 18.206 18.542 18.523 18.533 18.435 18.533 18.152 18.287 18.157 18.298 18.317 18.324 18.054

(11) (13) (18) (9) (36) (8) (12) (23) (16) (23) (8) (27) (15) (21) (IO) (23) (14)

(11) (13) (15) (7) (31) (8) (13) (24) (8) (22) (9) (21) (13) (19) (8) (16) ( I 4)

15.494 15.515 15.468

ro4Pb (+ 2SE)

15.462 15.485 15.531 15.461 15.496 15.505 15.523 15.511 15.520 15.519 15.495 15.487 15.500 15.490

*“Pb/

Ridge between 3% and 46%

*06Pb/ *04Pb ( f 2SE)

basalts from Mid-Atlantic

37.237 37.835 37.987 37.737 37.724 37.937 37.960 37.969 37.933 38.042 37.683 37.783 37.872 37.895 37.920 38.012 37.684

“‘Pb/

(7) (6) (7) (131 (7) (6) (7) (9) (6) (7) (6) (4) ( 12) (8) (7) (9) (8) (8)

(+ 2SE)

(5) (12) (3) (4) (4) (3) (3) (4)

0.513140 0.513000 0.5 13058 0.5 13055 0.5 1298 I 0.51301 I 0.5 13074 0.513063

(4) (4) (3) (2) (41 (3)

(k 2133

(3) (8) (3)

‘44Nd

0.513275 0.513210 0.5 13202 0.513199 0.513120 0.5131 I8 0.5 13089 0.513073 0.5 13068

ld3Nd/

account within-run precision and, for Pb, the uncertainty in SRM981, using the values of [47]. The discrimination factor The Sr isotope ratios measured in Geneva in static mode are = 0.708000, using an average of 0.708007 + 6~ IOU6 (2SE; was 0.710249& 16x 10e6 (2SE; n = 17). See Table I for

(26) (26) (39) (19) (72) (17) (38) (70) (23) (58) (27) (53) (38) (40) (20) (47) (29)

s%r

b.7021 I7 0.702246 0.702250 0.702481 0.702623 0.702214 0.702413 0.702397 0.702607 0.70242 I 0.7023 15 0.702509 0.702608 0.702544 0.702590 0.702755 0.702654 0.702747

204Pb (f 2SE) %/

a g in sample identification stands for glass, otherwise pillow interiors. Errors in parentheses are the within-run 2 standard errors of the mean of the last significant figures, taking into discrimination correction. The Pb isotope ratios were normalized on the basis of replicate measurements of NBS averaged 1.00_+0.07 (2SE& per mass unit. Pb blank is < 0.12 ng on sample sizes up to 0.4 g and is negligible. mass fractionation corrected to an s8Sr/s6Sr of 8.375209 and normalized to the E&A standard with “Sr/ 86Sr n = 52) measured from I i/94 to 5/95. The average of the SRM987 standard measured during this period 14’Nd/ ‘44Nd analytical conditions and standardization.

LC-B 3D-003g LC-B 2D- 103g LC-B 2D-202g RC16 3D-1 2IID 19D-2g 2flD 26D-2g EN061 25D-lg EN063 ID-6g 2fID 25D-3 EN063 3D-5g EN063 5D-6g EN063 8D-5g EN063 1 ID-5g EN063 13D-5g EN063 l5D-5g EN063 16D-6g EN063 18D-5g EN063 19D-5g

Sample ID a

Table 2 Pb, Sr and Nd isotope ratios in additional

?

? 5

?i 2 3 2

5 2

k? $

?B’ ,m 2 3 Sk z1 F z n’ 4

% s 5

;

* 3. .? c

? 4

214

D. Fontignie. J.-G. Schilling/ Earth and Planetary Science Letters 142 (1996) 209-221

Tristan-Cough plume heads prior to the opening of the South Atlantic. The reader is referred to Hanan et al. [ll for a full account of this two-stage plume dispersion model. The short and long wavelength Sr and Nd isotope variations reported here are fully consistent with the interpretation based on Pb isotopes only, and thus strengthen the two-stage plume dispersion model of Hanan et al. [l]. The MORB population from the Circe and St. Helena region is most enriched in 206Pb/ *04Pb, intermediate in 143Nd/ 144Nd, and reIatively low in *‘Sr/ 86Sr, as the adjacent St. Helena island is (and Ascension). In contrast, the MORB population near Tristan-Gough islands is most enriched in “Sr/ 86Sr, and lowest in ‘43Nd/ ‘44Nd, as these two islands are. The 17 new analyses listed in Table 2, and additional ones near 26”N reported by Castillo and Batiza [ 151, have also been plotted in the 3 Pb isotope profiles in Fig. 2. These new data points fit well with the previously established latitudinal variation, thus adding robustness to these trends.

4. Tests in Pb-Nd-Sr

Fig. 2. Comparison of latitudinal variations in s7Sr/86Sr, ‘43Nd/ ‘“Nd (note scale increasing downward) *08Pb/ ‘“Pb, (La/Sm), *“Pb/ *04Pb, ‘06 Pb/ *04Pb, and chondrite-normalized ratio [8] in basaltic glasses from the Mid-Atlantic Ridge. Note the short wavelength, spike-like anomalies (thin dashed lines) present opposite Ascension/Circe (C), St. Helena (SH), and Tristan (T), and the well structured, long wavelength base line trends in the Nd, Sr, and Pb isotope profiles compared to that of (La/Sm), (shaded band), which remains essentially constant from 3”s to 46% 0 = samples previously reported for Pb isotopes by Hanan et al. [l]; triangles = new Pb isotope data from 17 additional dredged stations. The thick, long wavelength lines represent a degree 2 polynomial best fit through the basalt population with (La/Sm), < 0.65 (shaded region in the (La/Sm), profile or dots and triangles on the Pb, Sr, and Nd isotope profiles). Vertical bars represent range of isotope ratios in basalts from the 25%26.5”s MAR segment reported by Castillo and Batiza [15].

space

Our previous study [ll has revealed three pseudobinary mixing trends in ‘08Pb/ *04Pb versus *06Pb/ *04Pb space (Fig. 3); namely: the dominant St. Helena-depleted mantle (DM) binary mixing for the MORB population located between 12% and 24% (and perhaps 3-7”S), which reflects the influence of the St. Helena plume; the Ascension (Circe) pseudo-binary mixing trend composed of the MORB population from 7% to 12”S, which reflects, presumably, the influence of the Circe plume. This pseudo-binary trend branches out from the DM-St. Helena mixing trend at a *06Pb/ *04Pb of about 18.75 and has a steeper positive slope than the latter; the MORB population from 24”s to 46”s forms a trend which points towards the Pb isotopic composition of Tristan and Gough islands, which again reflects the influence of this plume. This broad and steep MORB trend also branches out from the St. Helena-DM binary mixing trend, but at a rather lower *“Pb/ 204Pb than the Circe trend, namely, around 1% 18.25.

D. Fontignie,

J.-G. Schilling/Earth

and Planetary

Hanan et al. [l] have pointed out that the Ascension (Circe) and Tristan-Gough-Discovery pseudobinary trends branch out from the St. Helena-DM trend (Fig. 3, model 31, instead of converging with the DM source in a fantail fashion (Fig. 3, model 1). This observation implies a possible chronology of mixing and pollution of the asthenosphere, with the St. Helena plume pollutant appearing first. Similar mixing relationships were also confirmed in 3He/ 4He versus ‘06Pb/ 204Pb or 87Sr/ 86Sr [2]. The 18 new Pb isotope analyses of MORB from 7’S, 14”s and 20-24”s (Fig. 3), as well as from 26”s [15], are fully consistent with such groupings and add robustness to these mixing trends. The consistency of Hanan et al.‘s [l] model is also evident in Nd versus Sr or Sr versus Pb and Nd versus Pb isotope ratio projections (Fig. 4). The main binary mixing trend observed in these diagrams comprise the 3-24”s DM-Ascension-St. Helena mixing trend. The broad Tristan-Gough trend branches out from the former at some intermediate values (DM source polluted by St. Helena), rather than being

I 4O _

Science Letters 142 (1996) 209-221

directed towards the 3-7”s non-radiogenic MORB population (pure DM source). The radiogenic Pb and Sr end of this mixing vector points broadly toward Tristan-Gough-Discovery compositions, again in full accord with our two-stage model interpretation. These groupings are the least discernible in the ‘43Nd/ ‘44Nd versus “Sr/ 86Sr projection, because of the small dynamic range of the DM-AscensionSt. Helena mixing vector and the low angle it makes with that of the 24-46”s Tristan-Gough-MORB mixing trend. However, the dynamic range of the Tristan-Gough mixing vector is maximized in the Nd-Sr projection and clearly points towards the Tristan-Gough Island composition. Also important, the 12-24”s St. Helena-MORB mixing vector does not overshoot the position of the St. Helena hotspot composition in this projection, and thus is consistent and fully supports the binary mixing model initially roposed on the basis of the 2o8Pb/ ‘04Pb versus P 06Pb/ 204Pb projection. Because of the strong correthe mixing lation of ‘43Nd/ ‘44Nd with “Sr/s6Sr, relationship in 3He/ 4 He versus “Sr/ 86Sr space

x 3=7”s s

+ 7=24”S n

24=30%

215

/,’

C?

0

6/4 Pb Fig. 3. *‘sPb/ *04Pb versus 206Pb/204Pb diagram comparing the new MAR basaltic glass delineated previously by Hanan et al. [I] for the South Atlantic MAR (see Fig. 1 and Fig. 2 data fully confirm the N-MORB-St. Helena mixing vector previously established north of between 245 and 30% tend towards the field of the Tristan-Cough-Discovery anomaly respectively). Fields marked AK, S, 7’. G, and D delineate the range of basalt Pb isotope Tristan da Cunha, and Discovery off-ridge hotspots repotted by Sun [42].

analyses reported in Table 2 with the fields for location of these new samples). The new 24%. Some of the new Pb isotope analyses observed south of 30’S (fields 1, 2 and 3, compositions of the Ascension, St. Helena,

216

D. Fontignie, J.-G. &hilling/Earth

0.5135

,

and Planetary Science Letters 142 (19%) 209-221

7

cDM

discussed by Graham et al. [2] is also readily apparent in 3He/ 4 He versus 143Nd/ ‘44Nd space (not shown).

5. South Atlantic light-REE depleted MORB population

W 0.702

0.703

0.704 WWWr

0.705

0.702

18

19

20

, 18

21

1” 20

Fig. Scatter in Sr, isotope for basalt from to Note consistently, the population from close Tristan (T)-Gough (G)-Dis(D) off-ridge hotspots a vector broadly the Sr, composition of or Tristan-Gough hotspot (Walvis WI, the MORB to St. However, this case, contrast Pb space, Circe St. populations cannot T, and = averages data [35,42-441, and = average DSDP from Walvis reported Richardson et 1451. = mantle.

We now focus on the light-REE depleted MORB populations with (La/Sm), < 0.65 which spread along the entire South Atlantic MAR profile, in between the spike-like positive anomalies (Fig. 2). This population delineates mostly the long wavelength isotopic trend of the asthenosphere beneath the South Atlantic and provides a means of constraining its evolution and origin. The following observations are noted: (1) On the basis of their REE patterns, this MORB population would be considered to be simply derived from the normally depleted asthenosphere (e.g. [ 161). Yet, Fig. 5a-d shows that, while (La/Sm), stays essentially constant in this population (i.e. 0.410.65), it covers a wide range of values in radiogenic Sr and Nd, which define a smooth long wavelength trend along the ridge (Fig. 3). The s’Sr/ 86Sr ratio varies from 0.7021 to 0.703 1 and aNd from 13.3 to 7.0. Such a large range exceeds that found, for example, along the 400 km long Reykjanes Ridge gradient south of the Iceland plume, where the (La/Sm), changes from 0.4 to > 1.5 [17] and positively correlates with 87Sr/ 86Sr [18] or Pb isotope ratios [19]. It is also in marked contrast to the basalt population which form the spikes in the South Atlantic (Fig. 2) or to other MAR segments located over or close to other mantle plumes in the North Atlantic or Pacific (e.g. [20-221) where 87Sr/86Sr also correlates positively with (La/Sm),. The South Atlantic long wavelength Pb-Nd-Sr isotope ratio trends would require long-term Th/Pb, U/Pb, Sm/Nd, and Rb/Sr parent/daughter differences if the Pb-Nd-Sr radiogenic isotopes built up directly in the asthenosphere. However, this would not be compatible with the uniformly low (La/Sm), (high Sm/Nd) observed in these MORBs. This observation suggests that the long wavelength isotopic trends must reflect relatively recent injections of radiogenie-rich Pb and Sr (radiogenic-poor Nd) mantle material into the asthenosphere, and that the large

D. Fontignie, J.-G. Schilling/

high (i.e. high 206Pb/ 204Pb near St. Helena, “Sr/ *6Sr and low ‘43Nd/ ‘44Nd near TristanGough) suggests that the St. Helena and Tristan plumes were the two centers of injections for this broad scale pollution of the asthenosphere.

scale, uniform, light-REE depletion characterizing these MOREk must be of relatively recent vintage. (2) The fact that each of the long wavelength isotopic profiles reach maxima or minima near offridge hotspots with the same isotopic characteristics

17.6

0.7020

18.4 18.0 182 206Pb/204Pb

17.8

0.7025

18.6

37.5 38.0 208Pbp04Pb

0.7030

0.5130

%r/Qr

“.705

$ 0.704 L

0.5132

0.5134

G

T

(@

o.,05

m.’

D W

E L

L

38.5

‘43Nd/‘uNd

t cl

217

Earth and Planetary Science Letters 142 (1996) 209-221

L

$ 0.704

-

%*

-

6G

/

t S

S I

I

37

38

39 20BPb/2wPb

40

I

I

I

I

I

19

18

I,

1 I

I

,

21

20

(h)

S

5

4 D

0.5125

TG

W I

0.702

I,,,

111,11,,111,

0.703

0.704

%M%r

?

0.5125

-

D%

w I

0.705

18

I

,,,I,

I

19

I

I

I

20

I

I

I,

I

21

206Pb/2@‘Pb

Fig. 5. (a-d) (La/Sm), versus Pb, Sr. and Nd isotope ratios for the light-REE depleted MORB glass population with (La/Sm), < 0.65, which delineate the long wavelength MAR trend shown in Fig. 2. A significant Pb-Sr-Nd isotope variation is observed at essentially constant (La/Sm),. (e-h) Scatter diagrams in Sr, Nd, and Pb isotope space for the same glass population as in (a-d). Note that the 3-24’S and 24-46”s populations show two mixing vectors pointing consistently toward the average composition of off-ridge hotspots of St. Helena (S) and Tristan CT)-Gough (G)-Discovery (D), respectively. and the Walvis Ridge (W). For position of DM see Fig. 4.

218

D. Fontignie, J.-G. Schilling/ Earth and Planetary Science Leners I42 119%) 209-221

(3) Consistent with point 2, in Pb-Nd-Sr isotope space the light-REE depleted MORB population reduces to two distinct linear trends (Fig. 5e-h). In the Pb versus Sr isotope ratio space, the two normal segments 3-7”s and 14-24”s form a single linear mixing vector of the DM-St. Helena type but its length is smaller than for the spike population (compare Fig. 4 with Fig. 5e,f). The normal segments south of 24”s to 47”s form another small binary mixing vector directed toward Tristan compositions. However, the origin of this vector is not at DM but, instead, close to the apex of the DM-St. Helena small mixing vector. Consistent relationships are observed in the Nd versus Pb isotope space representation (Fig. 5g,h), despite the smaller dynamic range of Nd isotope systematics. These relationships suggest that the broad scale pollution of the asthenosphere by the St. Helena plume occurred prior to that of the Tristan plume.

6. St. Helena and Tristan plume head dispersions The mixing and light-REE depletion chronology discussed in the previous section is fully consistent with the two-stage model of Hanan et al. [I]. This model suggests that the St. Helena plume head flattened against the continental lithosphere, was partially melted and partly drained of its melts, dispersed and mixed with the depleted asthenosphere before the Tristan plume head went through the same process. The particular partial melting and mixing conditions required to model quantitatively the uniform depletion in La/Sm (and other incompatible elements) and the Pb-Nd-Sr variation observed both spatially and in isotope space is complex and beyond the scope of this paper. We will merely outline qualitutiuely the predominant controlling factors. First we note that, since the mixing trend observed in the Pb, Sr, and Nd isotope space projections of Figs. 4 and 5 are all practically linear, the Pb, Sr, and Nd end-member concentrations are not likely to differ from each other by very much (perhaps a factor < 2). The extents of partial melting by decompression and melt removal during the flattening of the two plume heads are likely to have been controlled by their excess temperature and water content, volumetric fluxes, and the existing thickness and rate of

extension and thinning of the two lithosphere lids present over them. Thus, the timing of the flattening of the St. Helena and Tristan plume heads relative to the tectonic stage at which the opening of the South Atlantic took place is critical but poorly constrained. The opening of the South Atlantic has slowly propagated from south to north in the 140- 118 Myr period (e.g. [23]). The early depletion event in the case of Tristan is evident from the 137-127 Ma old [24], large (- 1 X lo6 km3), ParanCEtendeka tholeiitic flood basalt outpouring, which apparently took place as the result of the interaction of this plume head with the overlying lithosphere [25,26]. Not all this magma was extracted entirely from the plume head, some also originated from the overlying continental lithosphere [27,28]. Thus, the residual plume head need not be totally depleted in basaltic components nor in incompatible elements. Although there is no comparable evidence of extensive melt extraction associated with the presumed impingement and dispersion of the St. Helena plume head, the relative chronology of mixing previously inferred requires that it took place prior to that of the Tristan plume head dispersion (ca. 130 Ma), and thus significantly earlier than the opening of the South Atlantic at this latitude (ca. 118 Ma, e.g. [23]). Halliday et al. [29] suggested that the fossil St. Helena plume head may lie in part beneath Cameroon. Consequently, we can infer that the St. Helena plume head would probably have flattened against an unstretched, and thus thicker, continental lithosphere lid, resulting in less decompression melting, and extraction of magma produced by smaller degrees of melting than what is likely to be the case for the Tristan plume. This would predict higher La/Sm ratios in the MORB derived from the St. Helena plume head than those derived from the Tristan plume head, if the initial La/Sm ratio in the two plume heads were to be the same. However, it must also be noted that the St. Helena plume is likely to be less enriched in incompatible elements than the Tristan plume. This is dictated by isotope systematits, which require that the time-integrated Rb/Sr, Nd/Sm (thus La/Sm), and Th/U ratios of the St. Helena mantle source be intermediate between that of Tristan and the unpolluted depleted asthenosphere, such as present beneath the 3-7”s MAR segment. Thus, at least qualitatively, it is conceivable that the

D. Fontignie.

J.-G. Schilling

/ Earth and Planetary

polluted asthenosphere in the two regions have comparably similar La/Sm depletions, as required by the data shown in Fig. 2. The fact that the St. Helena plume has left a much smaller track of constructional volcanism on the seafloor (hotspot track) than that of Tristan (i.e., the Walvis-Rio Grande rise), as evident on the Geosat/Seasat gravity map of Sandwell and Smith [30], is also consistent with, and further supports, this model scenario, as does the independently inferred present day volumetric fluxes of the Tristan and St. Helena plumes [7,31,32]. Finally, as previously suggested [ 1,8], we envision that the development of the sublithospheric channels connecting the MAR with the Tristan, St. Helena, and Circe plume conduits, and the resulting mixing with the broadly polluted asthenosphere taking place by entrainment, would have begun some 60 Ma, at least for Tristan, when the MAR began migrating westward and the production of Rio Grande rise by constructional volcanism ceased (e.g. [14]).

7. Continental

lithosphere

delamination

As an alternative but less likely hypothesis, the long wavelength trend in 87Sr/ 86 and 143Nd/ 144Nd (Fig. 2, light-REE depleted population), could be explored within the context of the DUPAL anomaly, using Hart’s [ 111 corresponding Azo8Pb/ 204Pb parameter (Fig. 6). The large change in slope in these parameters south of 24”s could be interpreted as the northern boundary of the east-west trending DUPAL anomaly of semi-hemispherical scale based on ocean island data [ 11,123. Recent MAR sampling south of Tristan to the Bouvet triple junction suggests that the long wavelength A*‘*Pb/ 204Pb trend (Fig. 6) abruptly drops between 49% and 50.5”s to values comparable to that found north of 24”s [33]. Thus, to a first order, these trends would appear consistent with the prediction made by the DUPAL isotope contouring of Hart [ 1 l] and Castillo [ 121. However, it must also be realized that south of 50.5’S, a major gravity/topographic and geochemical anomaly of high 3He/ 4 He and HIMU character, similar to Bouvet [33-351, also occurs opposite the Shona/Meteor hotspot track; suggesting, by itself, the influence of another plume in the region [36]. The cause of the DUPAL anomaly has been attributed to either delam-

Science Letters

142 (I 9%) 209-221

219

Fig. 6. A2a8Pb/ 2”4Pb (A8/4) latitudinal MAR profile for MORB glasses The Aro8Pb/ *04Pb parameter represents the excess *“Pb/ ‘04 Pb at a given *06Pb/ 204Pb relative to the Northern Hemisphere reference line, as given by Hart [l I]. Note the minimum in the region close to St. Helena and the rapid rate of increase southward towards the Tristan plume, with the so-called DUPAL characteristics. The low over the 20-243 region may be considered either as reflecting the pollution of the depleted asthenosphere by the St. Helena plume, or as the diffused northern boundary of the DUPAL anomaly, whatever its cause. The vertical bar represents the range of isotope ratios in basahs from the 25.5-26.5”s MAR segment reported by Castillo and Batiza [15]. Symbols as in Fig. 2.

ination of old sub-continental lithosphere by thermal erosion, rafting and recycling in the upper mantle during the break-up of the Gondwana plate [27,37,38]; or, alternatively, to the dispersion of mantle plumes of Kerguelen/Tristan isotopic type [11,39]. Our contention is that, dynamically, these two models need not be mutually exclusive. If, in the previous model, some of the Tristan and St. Helena plume heads underplated the Gondwana continental lithosphere prior to its break-up, this plume material would be delaminated first during the subsequent opening of the South Atlantic. The delamination could have occurred by secondary mantle convection, perhaps similar to such as caused by pull-apart forces and lateral temperature gradients between Archons and oceanic lithosphere (e.g. [40]). On the other hand, the similarity of the long and short wavelength MAR Pb-Nd-Sr isotope variations with that of the off-ridge hotspots appears too coincidental to be explained by delamination of subcontinental material unrelated to the South Atlantic hotspots. In conclusion, the Nd-Sr-Pb isotope data reported here best fit the two-stage plume dispersion/melting model described in Hanan et al. [l] and briefly expanded here. However, we also emphasize that this working hypothesis is based primarily on geochemical and isotopic evidence. The

220

D. Fontignie. J.-G. Schilling/Earth

and Planetary Science Letters 142 (19%)

thermal and dynamic plausibility of the first stage, plume head dispersion, part of this model remains untested at this time and thus speculative, whereas the second stage of the model; that is the lateral channeled plume flow towards the MAR, has received some support from numerical fluid dynamic experiments [41].

Acknowledgements We thank R. Kingsley and L. DiPanni for carrying out the Pb isotope analyses and the Sr/Nd separations, M. Cole for the initial Sr isotope analyses and K. Carey for careful editing. We also thank D.W. Graham, B.B. Hanan and A.P. 1eRoex for very constructive reviews. This work was supported by the National Science Foundation (Grant OCE 8916334) and the Fonds National Suisse de la Recherche Scientifique (Grant 20-39624.93). [FAI

References [iI B.B. Hanan, R.H. Kingsley and J.-G. Schilling, Pb isotope evidence in the South Atlantic for migrating ridge-hotspot interactions, Nature 322, 137- 144, 1986. [21D.W. Graham, W.J. Jenkins, J.-G. Schilling, G. Thompson, M.D. Kurz and S.E. Humphris, Helium isotope geochemistry of mid-ocean ridge basalts from the South Atlantic, Earth Planet. Sci. Lett. 110, 133-147, 1992. I31 P.W. Gast, G.R. Tilton and C. Hedge, Isotopic composition of lead and strontium for Ascension and Gough Islands, Science 145, 1181-1185, 1964. [41 B.L. Weaver, D.A. Wood, J. Tarney and J.L. Joron, Role of subducted sediment in the genesis of ocean-island basalts: Geochemical evidence from South Atlantic Ocean islands, Geology 14, 275-278, 1986. [51 A. Zindler and S. Hart, Chemical geodynamics, Annu. Rev. Earth Planet. Sci. 14, 493-571, 1986. I61 J.-G. Schilhng, Upper mantle heterogeneities and dynamics, Nature 314, 62-67, 1985. 171J.-G. Schilling, Fluxes and excess temperatures of mantle plumes inferred from their interaction with migrating mid-ocean ridges, Nature 352, 397-403, 1991. @I J.-G. Schilling, G. Thompson, R. Kingsley and S. Humphris, Hotspot-migrating ridge interaction in the South Atlantic, Nature 313, 187-191, 1985. [91 Y.-S. Zhang and T. Tanimoto, Ridges, hotspots and their interaction as observed in seismic velocity maps, Nature 355, 45-49, 1992.

209-221

[lOI B. Dupre and C.J. Allbgre, Pb-Sr isotope variation in Indian Ocean basalts and mixing phenomena, Nature 303, 142-146, 1983. S.R. Hart. A large-scale isotope anomaly in the Southern Hemisphere mantle, Nature 309, 753-757, 1984. P. Castillo, The Dupal anomaly as a trace of the upwelling lower mantle, Nature 336, 667-670, 1988. J.-G. Schilling, B.B. Hanan, B. McCully, R.H. Kingsley and D. Fontignie, Influence of the Sierra Leone mantle plume on the equatorial Mid-Atlantic Ridge: A Nd-Sr-Pb isotopic study, J. Geophys. Res. 99, 12,005-12,028, 1994. D41 J.M. O’Connor and R.A. Duncan, Evolution of the Walvis Ridge-Rio Grande Rise hot spot system: Implications for African and South American plate motions over plumes, J. Geophys. Res. 95, 17,475-17,502, 1990. D51 P. Castillo and R. Batiza, Strontium, neodymium and lead isotope constraints on near-ridge seamount production beneath the South Atlantic, Nature 342, 262-265, 1989. 1161J.-G. Schilling, Azores mantle blob: Rare earth evidence, Earth Planet. Sci. Lett. 25, 103-l 15, 1975. iI71 J.-G. Schilling, Iceland mantle plume, geochemical evidence along Reykjanes Ridge, Nature 242, 565-571, 1973. [rsl S.R. Hart, J.-G. Schilling and J.L. Powell, Basalts from Iceland and along the Reykjanes Ridge: Sr isotope geochemistry, Nature Phys. Sci. 236, 104-107, 1973. [191 S.-S. Sun, M. Tatsumoto and J.-G. Schilling. Mantle plume mixing along the Reykjanes Ridge Axis: Lead isotopic evidence, Science 190, 143-147, 1975. DOI W.M. White and J.-G. Schilling, The nature and origin of geochemical variation in Mid-Atlantic Ridge basalts from the central North Atlantic, Geochim. Cosmochim. Acta 42, 1501-1516, 1978. I211S.P. Verma and J.-G. Schilling, Galapagos hotspot-spreading center system, 2. “Sr/ *6Sr and large ion lithophile element variations (SS”W-lOl”WJ, J. Geophys. Res. 87, 10,83810,856, 1982. [221D. Fontignie and J.-G. Schilling, “Sr/ *%r and REE variations along the Easter Microplate boundaries (South Pacific): Application of multivariate statistical analyses to ridge segmentation, Chem. Geol. 89, 209-241, 1991. [231 D. Nimberg and R.D. Miller, The tectonic evolution of the South Atlantic from late Jurassic to present, Tectonophysics 191, 27-53, 1991. [241 S. Turner, M. Regelous, S. Kelley, C. Hawkesworth and M. Mantovani, Magmatism and continental break-up in the South Atlantic: high precision 4oAr-39Ar geochronology, Earth Planet. Sci. Lett. 121, 333-348, 1994. [251 R. White and D. McKenzie, Magmatism at rift zones: The generation of volcanic continental margins and flood basalts, J. Geophys. Res. 94, 7685-7729, 1989. LX1 M.A. Richards, R.A. Duncan and V.E. Courtillot, Flood basahs and hot-spot tracks: Plume heads and tails, Science 246, 103-107, 1989. [271 C.J. Hawkesworth, M.S.M. Mantovani, P.N. Taylor and Z. Palacz, Evidence from the Paran6 of south Brazil for a continental contribution to Dupal basalts, Nature 322, 356359, 1986.

D. Fontignie.

[28]

[29]

[30]

[31]

[32] [33]

[34]

[35]

[36]

[37]

J.-G. Schilling/Earth

and Planetary

C.J. Hawkesworth, K. Gallagher, S. Kelley, M. Mantovani, D.W. Peate, M. Regelous and N.W. Rogers, Pam& magmatism and the opening of the South Atlantic, in: Magmatism and the Causes of Continental Break-up, B.C. Storey. T. Alabaster and R.J. Pankhurst, eds., Geol. Sot. London Spec. Publ. 68, 221-240, 1992. A.N. Halliday, J.P. Davidson, P. Holden, C. DcWolf, D.-C. Lee and J.G. Fitton, Trace-element fractionation in plumes and the origin of HIMU mantle beneath the Cameroon line, Nature 347, 523-528, 1990. D.T. Sandwell and W.H.F. Smith, Global marine gravity from ERS-1, Geosat and Seasat reveals new tectonic fabric, EOS Trans. AGU 73, 133, 1992. G.F. Davies, Ocean bathymetry and mantle convection 1. Large-scale flow and hotspots, J. Geophys. Res. 93, 10,46710,480, 1988. N.H. Sleep, Hotspots and mantle plumes: Some phenomenology, J. Geophys. Res. 95, 6715-6736, 1990. J. Douglass, J.-G. Schilling, R.H. Kingsley and D. Fontignie, Plume-ridge interaction in the South Atlantic from 40”s to 52S”S: 2. Pb isotope evidence, EOS Trans. AGU 75, 723, 1994. M. Moreira, T. Staudacher, P. Sarda, J.-G. Schilling and C.J. Alltgre, A primitive plume neon component in MORB: The Shona ridge-anomaly, South Atlantic (51-52”S), Earth Planet. Sci. Lett. 133, 367-377, 1995. J. Douglass, J.-G. Schilling, R.H. Kingsley and C. Small, Influence of the Discovery and Shona mantle plumes on the southern Mid-Atlantic Ridge: Rare earth evidence, Geophys. Res. L&t. 22(21), 2893-2896, 1995. A.P. Le Roex, H.J.B. Dick, L. Gulen, A.M. Reid and A.J. Erlank, Local and regional heterogeneity in MORB from the Mid-Atlantic Ridge between 54.5’S and 51%: Evidence for geochemical enrichment, Geochim. Cosmochim. Acta 51, 541-555, 1987. E.M. Klein, C.H. Langmuir, A. Zindler, H. Staudigel and B. Hamelin, Isotope evidence of a mantle convection boundary at the Australian-Antarctic discordance, Nature 333, 623629, 1988.

Science Letters 142 (1996) 209-221

221

[38] J.J. Mahoney, J.H. Natland, W.M. White, R. Poreda, S.H. Bloomer, R.L. Fisher and A.N. Baxter, Isotopic and geochemical provinces of the western Indian Ocean spreading centers, 1. Geophys. Res. 94, 4033-4052, 1989. [39] M. Storcy, A.D. Saunders, J. Tarney, I.L. Gibson, M.J. Non-y, M.F. Thirlwall, P. Leat, R.N. Thompson and M.A. Menzies, Contamination of Indian Ocean asthenosphere by the Kerguelen-Heard mantle plume, Nature 338, 574-576, 1989. [40] S.D. King and D.L. Anderson, An alternative mechanism of flood basalt formation, Earth Planet. Sci. Lett. 136, 269-279, 1995. [41] C. Kin&d, J.-G. Schilling and C. Gable, The dynamics of off-axis plume-ridge interaction in the uppermost mantle. Earth Planet. Sci. Len. 137, 29-44, 1996. [42] S.S. Sun, Lead isotopic study of young volcanic rocks from mid-ocean ridges, ocean islands and island arcs, Philos. Trans. R. Sot. London A 297, 409-445, 1980. [43] A. Zindler, E. Jagoutz and S. Goldstein. Nd, Sr, and Pb isotopic systematics in a three-component mantle: a new perspective, Nature 298, 519-523, 1982. [44] D.J. Chaffey, R.A. Cliff and B.M. Wilson, Characterization of the St. Helena magma source, in: Magmatism in the Ocean Basins, A.D. Saunders and M.J. Norry, eds., Geol. Sot. London Spec. Publ. 42, 257-276, 1989. [45] S.H. Richardson, A.J. Erlank, A.R. Duncan and D.L. Reid, Correlated Nd, Sr, and Pb isotope variation in Walvis Ridge basalts and implications for the evolution of their mantle source, Earth Planet. Sci. Lett. 59, 327-342, 1982. [46] M.W. Cole, B.B. Hanan, R. Kingsley and J.-G. Schilling, Isotopic variations in South Atlantic MAR basalts: Implications on mantle plume-migrating ridge dynamics, EOS Trans. AGU 66, 408, 1985. [47] W. Todt, R.A. Cliff, A. Hanser and A.W. Hofmann, *‘*Pb +*“‘Pb double spike for lead isotopic analyses, Terra Cognita 4, 209, 1984.