Earth and Planetary Science Letters 193 (2001) 83^98 www.elsevier.com/locate/epsl

Noble gas study of the Reunion hotspot: evidence for distinct less-degassed mantle sources Takeshi Hanyu a;b; *, Tibor J. Dunai a , Gareth R. Davies a , Ichiro Kaneoka b , Susumu Nohda c , Kozo Uto d a

Department of Earth Sciences, Vrije Universiteit Amsterdam, De Boelelaan 1085, 1081 HV Amsterdam, The Netherlands b Earthquake Research Institute, University of Tokyo, Bunkyo-ku, Tokyo 113-0032, Japan c Department of Environmental Sciences, Faculty of Science, Kumamoto University, Kumamoto 860-8555, Japan d Geological Survey of Japan, Tsukuba 305-8567, Japan Received 9 April 2001; received in revised form 22 August 2001; accepted 27 August 2001

Abstract We present an extensive He, Ne and Ar isotope data set from the Reunion hotspot that demonstrates the presence of a homogeneous plume source that has unique isotopic characteristics. 3 He/4 He ratios of the two volcanoes on Reunion (Piton de la Fournaise, 6 0.53 Ma; Piton des Neiges, 2^0.44 Ma) are uniform 12^13.5 Ra regardless of 4 He concentration and sample age. The shield-building Older Series of Mauritius (5^8 Ma) has a constant 3 He/4 He ratio around 11.5 Ra. The similarity of 3 He/4 He and Sr^Nd isotope ratios among them demonstrate that the volcanoes have had a common homogeneous source related to the mantle plume activity over a period of 8 Myr. 20 Ne/22 Ne and 21 Ne/ 22 Ne of these volcanoes define a linear trend on a Ne three-isotope diagram with a slope between the Loihi and MORB correlation lines. There is a clear correlation between 20 Ne/22 Ne and 40 Ar/36 Ar. In contrast, Intermediate (2^3 Ma) and Younger Series ( 6 1 Ma) of Mauritius and Rodrigues (1.5 Ma) have 3 He/4 He ratios similar to MORB and Sr and Nd isotope ratios closer to MORB than lavas from Reunion and Older Series of Mauritius. These Intermediate and Younger Series lavas therefore record a late stage thermal rejuvenation beneath Mauritius derived from a source that is unrelated to the mantle plume. The isotopic characteristics of the source of the Reunion magmatism are relatively low 3 He/4 He (13 Ra), an intermediate slope in a Ne three-isotope diagram and relatively radiogenic Sr isotope ratios. These source characteristics cannot be explained by either crustal contamination or MORB source mixing with Loihi-type primitive mantle. Thus He^Ne^Ar^Sr^Nd isotopes demonstrate that this plume source is distinct from the source of other large plumes (Loihi and Iceland), clearly showing that the mantle contains several relatively less-degassed reservoirs and not a single primitive source. Two possible models can account for the different isotopic signature of Reunion and Hawaii hotspots; (1) the Reunion source contains more recycled material than Loihi source and (2) the Reunion source experienced stronger degassing/differentiation than the Loihi source in the early stage of mantle evolution. In both cases a convection mode in the mantle is required that isolates and preserves several less-degassed reservoirs in the convectively stirred lower mantle. ß 2001 Elsevier Science B.V. All rights reserved.

* Corresponding author. Present address: Earthquake Research Institute, University of Tokyo, Bunkyo-ku, Tokyo 113-0032, Japan. Tel.: +81-3-5841-5700; Fax: +81-3-5802-3391. E-mail addresses: [email protected] (T. Hanyu), [email protected] (T.J. Dunai), [email protected] (G.R. Davies), [email protected] (I. Kaneoka), [email protected] (S. Nohda), [email protected] (K. Uto). 0012-821X / 01 / $ ^ see front matter ß 2001 Elsevier Science B.V. All rights reserved. PII: S 0 0 1 2 - 8 2 1 X ( 0 1 ) 0 0 4 8 9 - 7

EPSL 5991 13-11-01

84

T. Hanyu et al. / Earth and Planetary Science Letters 193 (2001) 83^98

Keywords: noble gases; isotope ratios; Reunion; hot spots

1. Introduction Noble gas studies of mid-ocean ridge basalts and ocean island basalts have demonstrated that the mantle includes two major reservoirs that correspond to degassed and less-degassed components. The isotopic evidence requires that the less-degassed reservoir remain unmixed with other material during mantle evolution, possibly suggesting a form of layered convection in the mantle (e.g., [1]). Tomography, simulation of mantle convection and some radiogenic isotopes (e.g., Hf), however, do not always support the layered convection model and alternative models to maintain isolated less-degassed reservoirs during whole mantle scale stirring have been proposed [2,3]. Mantle plumes are the main mode of material transport from the lower mantle and an understanding of their compositions is required if we are to fully understand how the Earth has di¡erentiated. Here we present new noble gas data from the Reunion hotspot to provide more constraints on the condition of the less-degassed portion of the mantle where some mantle plumes are thought to originate. Reunion Island is characterised by a very homogeneous He, Sr and Nd isotope geochemical signature [4,5], which contrasts with other major hotspots such as Hawaii and Iceland. Hawaiian volcanoes in particular show relatively large variation of isotope ratios in time and space which probably re£ect the complicated melting relationships associated with mantle plumes [6]. This study of Reunion and Mauritius is designed to assess the constancy of the isotopic homogeneity with time and to consider any implications for the origin and structure of the plume source. Hence there are three major objectives : (1) Assessment of the helium isotope constancy of Reunion and Mauritius Islands : the younger volcano on Reunion (Piton de la Fournaise) has a well-constrained and constant 3 He/4 He around 13 Ra (Ra; atmospheric 3 He/4 He ratio) [5]. Due to limited data from the older volcano on Reunion (Piton des Neiges ; [7,8]) and Mauritius Island,

it is still unknown if He isotopes have been constant or record a temporal evolution. Together the islands should record the most recent 8 Myr activity of the hotspot. (2) Source characterisation of Reunion : A question to be assessed is whether the relatively low 3 He/4 He of Reunion compared to Hawaii and Iceland is an isotopic signature recording the composition of the plume source. New Ne and Ar isotope data coupled with He isotopes will characterise the source region of mantle plumes to constrain its genesis. (3) Temporal evolution of Mauritius source domain: Mauritius Island is dominantly composed of Older Series lavas that formed between 5 and 8 Ma. These lavas are followed by Intermediate (2^3 Ma) and Younger (1 Ma to present) Series which are minor in volume [9]. Comparison of noble gas isotope ratios of the three series will reveal any temporal changes in source materials over the long period of volcanic activity. 2. Geological and geochemical background It is generally agreed that Reunion Island, Mauritius Island and Mascarene Plateau have a hotspot/mantle plume origin (Fig. 1). A buoyancy £ux of 1.9 Mg/s [10] means that Reunion is formed from one of the major hotspots on the Earth. It is presently active at Reunion Island where dated rocks record activity from around 2 Ma to historic [11,12]. Mauritius Island was active from at least 8 Ma and it is considered that the island is located on the former site of the Reunion hotspot. Plate reconstruction and geochronological studies indicate that the Deccan £ood basalts are related to the Reunion hotspot, suggesting that the starting plume head caused extensive volcanism at around 65 Ma on the Indian subcontinent [13]. The Reunion hotspot is presently located beneath the Paleocene ocean £oor in the Madagascar Basin on the African plate that migrates northeastward at about 2 cm/yr (Fig. 1). This basin was created by the Central Indian Ridge. In-

EPSL 5991 13-11-01

T. Hanyu et al. / Earth and Planetary Science Letters 193 (2001) 83^98

85

Fig. 1. (A) Bathymetry map around Reunion^Mascarene Plateau topographic lineament (after [14]). Solid lines show the depth contours at 2000 m and 4000 m as well as active spreading ridges. Dotted lines show inactive transform faults. (B, C) Maps of Reunion and Mauritius Islands. Sample localities are shown by black dots with sample names.

active transform faults (Mahanoro fracture zone and Mauritius fracture zone) exist parallel to the Reunion^Mascarene Plateau topographic lineament. Although the Mascarene Plateau has continuous topography, Mauritius and Reunion are separated from each other and from Mascarene Plateau by deep bathymetric gaps (Fig. 1) [14]. Rodrigues Island is situated between Mauritius

and Indian Ocean Ridge 300 km from the Reunion^Mascarene Plateau topographic lineament (Fig. 1). This island is part of an eastward branching ridge of the Mascarene Plateau. The magmatism that formed the island is not simply related to Reunion hotspot activity, but has been interpreted as a consequence of a channeling e¡ect between the Central Indian Ridge and Reunion hotspot [15].

EPSL 5991 13-11-01

86

T. Hanyu et al. / Earth and Planetary Science Letters 193 (2001) 83^98 3

2.1. Reunion There are two volcanoes on the island of Reunion. The older volcano, Piton des Neiges, is located northwest of the younger volcano, Piton de la Fournaise (Fig. 1). The age of the oldest dated rock of Piton des Neiges is about 2 Ma [11] and a morphorological study suggests that volcanic activity started at around 5 Ma [16]. The volcanic rocks can be separated into two groups according to their chemical compositions. A di¡erentiated Series is characterised by evolved chemical compositions with high SiO2 , low MgO and high alkali concentrations [4]. Their ages are younger than 0.35 Ma [11]. The shield of the volcano mainly consists of the Oceanite Series composed dominantly of olivine basalts. Most of the £anks of the island are covered by Di¡erentiated Series lavas, Oceanite Series rocks thus appear in the three central calderas (cirques), along eroded valleys and north and west margins of the island. McDougall et al. [11] recognised three phases of volcanic activity on the basis of K^Ar dating (Fig. 1). The ¢rst phase occurred at about 2 Ma and these lavas appear only at La Montagne massif, northern part of the island. The second (1^1.2 Ma) and the third phase lavas (0.57^0.43 Ma) are found mostly on the southeastern side of the volcano but also some exposed on the western side. The rocks in the cirques called agglomerate material are presumably the oldest among the exposed rocks although they are not dated due to extensive alteration [11]. Piton de la Fournaise is the presently active volcano on the island of Reunion. There are three major U-shaped calderas that open to the east. The oldest dated rock of Piton de la Fournaise is 0.53 Ma [12] and younger rocks generally appear in the west^southwestern part of the volcano. Rocks of Piton de la Fournaise are petrologically similar to those of Oceanite Series of Piton des Neiges (e.g., [4]). A signi¢cant characteristic of Reunion magmatism is that Sr and Nd isotope ratios are extremely constant regardless of age and magmatic series [4]. This characteristic is in marked contrast to many other ocean islands. Graham et al. [5] showed that Piton de la Fournaise has a constant

He/4 He ratio of around 13 Ra among samples from 0.36 Ma to historic. Piton des Neiges [7,8] rocks have comparable 3 He/4 He to Piton de la Fournaise, although limited data are available. 2.2. Mauritius and Rodrigues The volcanic activity of Mauritius Island includes three magmatic series at 5^8 Ma (Older Series), 2^3 Ma (Intermediate Series) and 6 1 Ma to present (Younger Series) [9]. Younger Series lavas cover most of the surface of the island, thereby exposures of Older and Intermediate Series are limited (Fig. 1). The Younger Series lavas are thin and represent the least voluminous of the three series. The bulk of the island consists of Older Series lavas which are divided into two groups [9]. The ¢rst stage, which formed around 7 Ma, dominantly comprises oceanites and alkali olivine basalts. In contrast, more di¡erentiated magmas appeared in the second stage at 5^6 Ma. Intermediate Series lavas are only exposed in the south part of the island and consist of alkali olivine basalts, basanites and nephelinites. The Younger Series also consist of alkali olivine basalts. A unique characteristic of this series is low concentration of incompatible elements. Compared to Older Series lavas this suggests formation by higher degrees of melting or derivation from a more trace element depleted source [17]. Older Series lavas have comparable Sr, Nd and Pb isotope ratios to Reunion lavas (Nohda et al., in preparation). Intermediate and Younger Series, however, have less radiogenic Sr and more radiogenic Nd isotope ratios than Older Series. This isotopic relationship among the volcanic series is apparently similar to changes between the shieldbuilding and post-shield/rejuvenated stages of Hawaiian volcanoes [18]. Two K^Ar dates are available for Rodrigues Island, 1.58 and 1.30 Ma [19]. The accessible part of the island predominantly consists of transitional-mildly alkali^olivine basalts with some di¡erentiated rocks [20]. 2.3. Samples Sampling sites are shown by dots in Fig. 1.

EPSL 5991 13-11-01

T. Hanyu et al. / Earth and Planetary Science Letters 193 (2001) 83^98

Samples were collected from the three phases of the Oceanite Series from Piton des Neiges on Reunion. Most of the samples were lava £ows except for RN95-15 (dyke), RN95-29 (wehrlite nodule) and RN95-37 (block in pyroclastic £ow). Noble gases were not measured for the Di¡erentiated Series rocks because large olivines are generally scarce. All the samples from Piton de la Fournaise are Oceanite lavas except for RF94-01 which is a dunite nodule from Piton de Chisney. Olivine ( þ clinopyroxene) bearing lava £ows were collected from the three series on Mauritius Island. Some of samples used in this study were dated by K^Ar dating (Uto et al., in preparation). Three samples from the Rodrigues Island were used for comparison with Reunion and Mauritius samples. These samples were collected from olivine and plagioclase ( þ clinopyroxene) bearing lava £ows.

87

SAES getters to remove active gases. Helium, neon and argon were separated by activated charcoal cooled by liquid nitrogen and a cryogenic pump (at 45 K). Elemental abundances and isotope ratios of helium, neon and argon were measured on a VG5400 mass spectrometer at Vrije Universiteit Amsterdam. Neon isotope ratios were corrected for interferences of 40 Ar2‡ and 20 CO2‡ Ne‡ and 22 Ne‡ , respectively. Typical 2 on blanks of the crushing experiments were 1^4 pcm3 STP of 20 Ne and 2^7 pcm3 STP of 36 Ar. 4 He in the blank was not detected by a Faraday cup but estimated to be less than 0.1 ncm3 STP. Full details of noble gas puri¢cation and analysis subsequent to crushing are described elsewhere [22]. 4. Results 4.1. Stepwise crushing test

3. Experimental methods Rock samples were crushed by a jaw crusher and sieved to collect olivine grains of 0.5^1.2 mm or 0.25^0.5 mm (L or S, respectively, as noted in Table 1). Larger grains were used for noble gas analyses when available, but most of the samples from Intermediate and Younger Series of Mauritius only yielded mineral grains of the smaller size fraction. Following washing in H2 O, olivines were handpicked using a binocular microscope. Samples were washed in ethanol before loading into a sample holder. Gases were extracted from sample grains by crushing in vacuum. The crusher consists of a metal tube with £at bottom and a piston that is lifted by a solenoid coil outside the vacuum system. Samples were loaded into a rotatable bucket above the crushing tube. The crushing tube and sample holder were baked overnight in vacuum at 250³C and 150³C, respectively. Most of the samples were crushed in a single step of 200 times. Multi-stepwise crushing was applied for some samples to assess the possible in£uence of cosmogenic and radiogenic components in the samples [21]. Extracted gases were puri¢ed by hot and cold

Multiple stepwise crushing was performed on samples RN95-36, RN95-40b, RF94-01 and MR94-0202. The largest amount of He, Ne and Ar was always released in the ¢rst step. 3 He/4 He ratios were identical among the steps of each sample. Signi¢cantly MR94-0202, one of the oldest samples in this study (5.4 Ma), did not show any indication of the progressive release of radiogenic 4 He, demonstrating that radiogenic 4 He is minimised by the crushing approach followed in this study. Previously the validity of a crushing method to extract `primary' gases and to avoid the release of in situ produced radiogenic 4 He has been reported when crushing times are short [21]. In contrast, Ne and Ar isotope ratios were not always constant in the di¡erent steps. RF9401, for example, has higher 20 Ne/22 Ne, 21 Ne/22 Ne and 40 Ar/36 Ar in the second and third steps than in the ¢rst step. These data appear to be a consequence of less atmospheric contribution to the later extracted gases. 4.2. Grain size e¡ect To check for a possible grain size e¡ect on isotopic measurements, two di¡erent size fractions of

EPSL 5991 13-11-01

88

T. Hanyu et al. / Earth and Planetary Science Letters 193 (2001) 83^98

Table 1 He, Ne and Ar concentrations and isotope ratios in Reunion, Mauritius and Rodrigues samples Sample

Weight Crushing (g) Times

Grain Age Size

4

He (ncm3 /g)

20

36 3 Ne Ar He/4 He 3 3 (pcm /g) (pcm /g) (Ra)

Reunion Piton des Neiges RN94-03 0.663 200

L

A

7.51

RN94-0501

1.035

200

L

R

0.83

RN94-07

1.847

200

L

R

RN94-0801

1.054

200

L

R

3.36

26.4

36.4

RN94-1103

0.933

200

L

A

5.62

12.4

43.4

RN94-13

0.922

200

L

A

4.72

12.8

35.5

RN95-1001

1.480

200

L

A

10.0

RN95-15

1.316

200

L

R

16.5

RN95-2301

0.927

200

L

R

RN95-2701a 1.635

200

L

R

91.1

63.4

43.8

RN95-2701b 1.168

200

S

R

27.2

38.4

48.0

RN95-28

1.187

200

L

R

13.2

63.4

RN95-29a (nodule) RN95-29b (duplicate) RN95-36

1.035

200

L

66.5

39.7

0.110

200

L

1.504

50

L

139

9.78

1.92 103 59.0

R

add 150

8.55 1.64

8.96 159

9.87 20.3 40.7

158 100 6.28

248.5 21.2 222

18.9 77.0 103

144 16.3 11.6

RN95-37

1.293

200

L

R

RN95-40a

1.447

200

L

R

78.2

RN95-40b (duplicate)

1.714

50

L

R

24.6

5.86

14.2

4.02

8.06

1.81

3.92

add 150 add 200

RN95-43

2.043

200

Reunion Piton de la Fournaise RF94-01 1.78 50 (nodule) add 150 add 200

1.87

23.3

1.16 L L

A

8.82

38.0

31.1

19.8

24.3 10.2

40.2

17.8

354

65.1

79.9

271

40.0

42.2

139

20.1

17.4

EPSL 5991 13-11-01

20

Ne/22 Ne

21

Ne/22 Ne 38 Ar/36 Ar

12.03 0.19 13.18 0.81 12.56 0.13 13.52 0.35 12.82 0.22 13.38 0.29 12.32 0.16 12.66 0.13 12.69 0.18 13.03 0.06 12.71 0.10 12.08 0.44 12.24 0.09 11.76 0.24 13.03 0.13 13.00 0.31 12.54 0.48 12.60 0.05 12.49 0.10 12.48 0.12 14.09 1.94 13.52 0.15

9.73 0.18 10.44 0.28 10.15 0.04 9.72 0.17 9.40 0.20 9.45 0.29 9.84 0.20 9.88 0.15 9.82 0.16 10.38 0.06 9.86 0.13 9.58 0.16 11.64 0.13 11.29 0.36 10.19 0.20

0.0288 0.0013 0.0328 0.0019 0.0304 0.0005 0.0293 0.0012 0.0299 0.0015 0.0316 0.0013 0.0282 0.0025 0.0307 0.0010 0.0286 0.0008 0.0312 0.0003 0.3020 0.0048 0.0290 0.0008 0.0355 0.0009 0.0349 0.0014 0.0291 0.0013

9.83 0.09 11.78 0.22 11.29 0.18 11.35 0.33

13.02 0.06 13.26 0.19 13.34 0.25

40

Ar/36 Ar

0.0295 0.0005 0.0386 0.0017 0.0351 0.0015 0.0360 0.0023

0.1875 0.0009 0.1866 0.0062 0.1888 0.0007 0.1900 0.0040 0.1879 0.0025 0.1901 0.0040 0.1828 0.0032 0.1884 0.0011 0.1905 0.0012 0.1877 0.0016 0.1878 0.0017 0.1904 0.0013 0.1876 0.0043 0.1969 0.0075 0.1862 0.0036 0.1807 0.0049 0.1883 0.0024 0.1893 0.0044 0.1947 0.0083 0.1887 0.0089

333.2 3.4 342.6 10.9 684.1 6.3 343.7 7.2 426.6 14.2 421.6 8.3 908.4 34.3 461.2 6.0 352.3 5.5 1361 28 555.5 15.7 372.3 7.0 3858 92 3106 120 847.8 48.8 448.1 34.1 389.7 13.0 2428 53 2437 98 1903 90

9.84 0.07

0.0295 0.0012

0.1812 0.0029

686.7 22.7

11.22 0.07 12.68 0.12 12.64 0.19

0.0355 0.0008 0.0395 0.0009 0.0392 0.0013

0.1891 0.0016 0.1934 0.0029 0.2069 0.0070

2928 73 4408 187 4634 427

T. Hanyu et al. / Earth and Planetary Science Letters 193 (2001) 83^98

89

Table 1 (continued) Sample

Weight Crushing (g) Times

Grain Age Size

RF94-04

1.835

200

L

RF94-05

1.721

200

L

RF94-0602

1.106

200

RF95-2903

1.724

RF95-35

4

He (ncm3 /g) 14.6

20

36 3 Ne Ar He/4 He (pcm3 /g) (pcm3 /g) (Ra)

12.6

16.8

7.32

13.5

16.8

L

7.68

47.3

60.9

200

L

3.92

15.8

34.9

1.928

200

L

36.1

32.1

RF95-72

1.732

200

L

9.95

51.2

40.1

RF95-74

1.884

200

L

6.97

29.8

36.7

RF95-77

2.051

200

L

13.6

28.4

32.9

RF95-79

1.97

200

L

16.3

43.2

36.1

Mauritius MR94-0202

1.538

50

L

12.4

16.6

31.8

13.9

O

add 150

8.28

add 200

3.56

1.13

24.5

MR94-0501

0.951

200

S

Y

0.09

48.1

76.5

MR94-06

1.539

200

L

O

0.13

13.8

25.9

MR94-08

1.308

200

L

O

7.94

11.7

80.7

MR95-01

1.498

200

S

Y

0.200

MR95-02

1.586

200

L

O

MR95-03

0.911

200

L

O

7.39

26.1

83.8

MR95-09

1.088

200

S

Y

0.124

19.6

62.1

MR95-10

1.555

200

S

Y

1.09

67.4

MR95-12

1.004

200

S

Y

0.260

38.0

75.0

MR95-14

1.925

200

S

Y

0.615

23.2

45.5

MR95-18

1.024

200

S

Y

33.3

43.9

MR95-2102

0.486

200

S

I

0.283

28.9

101

MR95-23

1.359

200

S

I

3.75

69.8

428.2

36.9

9.56 43.2

24.0 187

102

EPSL 5991 13-11-01

20

Ne/22 Ne

21

Ne/22 Ne 38 Ar/36 Ar

40

Ar/36 Ar

13.54 0.12 13.02 0.18 12.48 0.23 12.60 0.20 12.67 0.11 12.49 0.16 12.74 0.15 12.97 0.11 13.07 0.12

10.58 0.17 10.00 0.17 9.78 0.10 9.70 0.15 10.00 0.08 9.98 0.08 9.86 0.09 9.77 0.10 9.88 0.07

0.0331 0.0013 0.0288 0.0010 0.0291 0.0006 0.0319 0.0010 0.0305 0.0004 0.0302 0.0007 0.0294 0.0007 0.0295 0.0006 0.0292 0.0007

0.1854 0.0030 0.1846 0.0033 0.1882 0.0026 0.1884 0.0016 0.1866 0.0018 0.1884 0.0015 0.1874 0.0016 0.1881 0.0016 0.1919 0.0016

1630 61 703.8 25.9 523.3 7.0 403.4 9.7 512.2 12.8 468.6 10.1 442.1 9.3 427.5 9.4 507.7 11.9

11.59 0.20 11.31 0.18 11.36 0.24

9.82 0.11

0.0291 0.0010

0.1825 0.0022 0.1832 0.0026

602.7 18.1 622.3 24.6

9.71 0.15 9.38 0.10 9.88 0.22 9.86 0.17 10.47 0.16 9.73 0.22 9.93 0.18 9.90 0.06 9.66 0.10 9.62 0.10 10.30 0.17 9.73 0.18 9.96 0.06

0.0274 0.0008 0.0285 0.0017 0.0284 0.0013 0.0286 0.0012 0.0304 0.0006 0.0284 0.0019 0.0300 0.0007 0.0301 0.0003 0.0279 0.0005 0.0279 0.0007 0.0330 0.0015 0.0294 0.0022 0.0302 0.0002

0.1887 0.0015 0.1908 0.0025 0.1879 0.0011 0.1820 0.0027 0.1942 0.0077 0.1891 0.0015 0.1886 0.0015 0.1873 0.0008 0.1862 0.0014 0.1880 0.0011 0.1870 0.0024 0.1882 0.0019 0.1821 0.0006

774.8 13.7 324.1 9.4 370.1 4.6 529.6 19.3 1333 52 440.7 7.3 497.8 10.2 556.4 5.2 421.0 8.0 467.7 7.9 647.9 17.3 403.4 11.5 446.4 4.0

11.79 0.12 11.81 0.10 11.73 0.19 8.39 0.56 7.93 0.47

6.49 0.17

90

T. Hanyu et al. / Earth and Planetary Science Letters 193 (2001) 83^98

Table 1 (continued) Sample

4

He (ncm3 /g)

20

36 3 Ne Ar He/4 He (pcm3 /g) (pcm3 /g) (Ra)

Weight Crushing (g) Times

Grain Age Size

Rodrigues RD95-01

0.517

200

S

2.01

35.6

65.4

RD95-02

1.500

200

L

1.30

23.0

20.9

RD95-10

0.564

200

L

4.84

24.0

98.8

7.85 0.45 7.94 0.50

20

Ne/22 Ne

9.91 0.16 10.03 0.20 9.33 0.15

21

Ne/22 Ne 38 Ar/36 Ar

0.0309 0.0020 0.0283 0.0013 0.0303 0.0013

0.1852 0.0031 0.1852 0.0032 0.1874 0.0017

40

Ar/36 Ar

456.8 19.3 382.7 13.8 361.6 9.0

L and S in the column of grains size denote larger (0.5^1.2 mm) and smaller (0.25^0.5 mm) olivine grains used for analyses. For age, A and R for Piton des Neiges denote ancient (about 2 Ma) and recent (1.2^0.43 Ma) rocks, respectively. O, I and Y for Mauritius indicate Older (5^8Ma), Intermediate (2^3Ma) and Younger Series (and 6 1 Ma), respectively. Helium isotope ratios are normalised by atmospheric ratio. Errors in the second line denote 1c. Volumes at STP.

olivines (denoted as L and S in Table 1) were analysed from RN95-2701(a,b). This test is to assess any potential bias of isotopic ratios of Intermediate and Younger Series samples of Mauritius which have only smaller grains of olivines. The di¡erent grain sizes did not show any signi¢cant di¡erence in 3 He/4 He ratios. The smaller grains, however, have lower 20 Ne/22 Ne, 21 Ne/22 Ne and 40 Ar/36 Ar than larger grains. This observation is ascribed to smaller grains having greater atmospheric contamination for Ne and Ar isotopes, although helium isotopic ratios are not signi¢cantly changed. 4.3. Reunion He/4 He ratios of Piton de la Fournaise are between 12.5 and 13.5 Ra for both Oceanites and dunite nodules (RF94-01) (Fig. 2). These data are in good agreement with the previous work [5,7,8,23]. 3 He/4 He of Piton des Neiges are also constant regardless of 4 He concentration and sample age and indistinguishable from the 3 He/ 4 He ratio of Piton de la Fournaise. Some samples from both Reunion volcanoes have higher 20 Ne/22 Ne and 21 Ne/22 Ne ratios than the atmospheric value. In particular, two nodules (RN95-29 and RF94-01) and one oceanite sample (RN95-40) have large excess of 20 Ne/22 Ne and 21 Ne/22 Ne ratios, with ratios up to 12.6 and 0.0395 respectively. These data, along with previous data [8], de¢ne a positive correlation that lies between the Loihi^Kilauea and MORB corre-

lation lines [24,25] on the Ne three-isotope diagram (Fig. 3). Argon isotope ratios vary from the atmospheric ratio up to 4600. There is a clear correlation between 20 Ne/22 Ne and 40 Ar/36 Ar (Fig. 4). 20 Ne/ 22 Ne ratios higher than the atmospheric value are always accompanied by 40 Ar/36 Ar of more than 1000. 4.4. Mauritius and Rodrigues The Older Series of Mauritius has a constant He/4 He ratio around 11.5 Ra regardless of 4 He concentration (Table 1, Fig. 2). This ratio is slightly lower than that of Reunion. In contrast,

3

3

Fig. 2. 3 He/4 He ratios plotted against 4 He concentrations for all the measured samples. Range of 3 He/4 He of MORBs and Piton de la Fournaise [5] are shown by hatching. Error bars are 1c.

EPSL 5991 13-11-01

T. Hanyu et al. / Earth and Planetary Science Letters 193 (2001) 83^98

Intermediate and Younger series of Mauritius and Rodrigues have signi¢cantly lower 3 He/4 He than the samples from Reunion and the Older Series of Mauritius. 3 He/4 He ratios of the Younger Series of Mauritius and Rodrigues are indistinguishable from those of MORB (Fig. 2). A sample from the Intermediate Series of Mauritius has an even lower 3 He/4 He of 6.5 Ra. He isotope ratios could not be determined for some samples from Mauritius and Rodrigues Islands because of low helium contents. In particular, Intermediate and Younger Series samples of Mauritius generally have small amounts of helium. 4 He/40 Ar* ratios (40 Ar* = (40 Ar/36 Ar-296)U (36 Ar)) of Intermediate and Younger Series samples are systematically lower than those of samples from Reunion and Older Series of Mauritius and the estimated mantle production ratio, suggesting that Intermediate and Younger Series samples degassed more He than Ar by di¡usive loss either in the magma chamber or after eruption because these samples are ¢ner grained than samples from Reunion and Older Series of Mauritius. One Older Series sample (MR95-02) has Ne isotope ratios that are distinct from atmospheric values and the data plot close to the correlation line of Reunion in the Ne three-isotope diagram (Fig. 3). This sample has relatively high 40 Ar/36 Ar

91

Fig. 4. 20 Ne/22 Ne against 40 Ar/36 Ar for Reunion data. MORB correlation line is from [30]. Although 20 Ne/22 Ne^ 40 Ar/36 Ar correlations of each hotspot are more scattered than 20 Ne/22 Ne^21 Ne/22 Ne correlation in Fig. 3, Reunion, Loihi and Iceland are distinct from MORB trend. Possible endmembers 20 Ne/22 Ne of 13.8 (Solar) or 12.5 (Neon B) are shown on the left axis.

(1330) compared to other Mauritius and Rodrigues samples and it lies on the Ne^Ar isotopic correlation line of Reunion. In contrast, MR95-18 from the Younger Series, which has a MORB-like He isotope signature, seems to plot on the MORB correlation line (Fig. 3). This sample is characterised by a small Ne excess but the analytical errors make it di¤cult to determine whether it has a MORB-like or Reunion-like Ne isotope signature. 5. Discussion 5.1. Constancy of helium isotope ratios of the Reunion hotspot

Fig. 3. Ne three-isotope diagram for all the measured samples. Previous data from [8] are also shown. Loihi and MORB correlation lines are from [24,25]. Error bars are 1c. The entire dataset de¢nes a linear correlation in 20 Ne/22 Ne and 21 Ne/22 Ne between Loihi and MORB trends.

It has been recognised that 3 He/4 He is uniform among the samples from Piton de la Fournaise [5]. Constancy of 3 He/4 He, as well as Sr and Nd isotope ratios [4], suggests a homogeneous source region of Piton de la Fournaise over a period of 0.36 Myr. Due to the limited database it was not clear, however, if the older volcano on Reunion, Piton des Neiges, also had similar helium isotope characteristics [7,8]. The samples in this study cover a wide area of the Piton des Neiges and include the oldest (about 2 Ma) and youngest (0.44 Ma) lavas. The present data establish that

EPSL 5991 13-11-01

92

T. Hanyu et al. / Earth and Planetary Science Letters 193 (2001) 83^98

the volcano had a constant 3 He/4 He. The average 3 He/4 He of Piton des Neiges appears to be slightly lower than that of Piton de la Fournaise, however, statistically these data can be treated as samples of a single population. The similarity of 3 He/ 4 He and Sr^Nd isotopes between the two volcanoes of Reunion demonstrates that they have had a common homogeneous source over a 2 Myr period. The age di¡erence between Mauritius and Reunion Islands is consistent with both islands being produced by the same mantle plume on the oceanic plate moving at about 2 cm/yr. The Older Series, corresponding to the shield-building stage of Mauritius, have comparable 3 He/4 He ratios to those of Reunion as well as Sr and Nd isotope ratios (Nohda et al., in preparation). These data suggest that both islands were produced by the same hotspot activity and that there was minimal temporal change in plume composition over an 8 Myr period. The Older Series lavas of Mauritius, in fact, have slightly lower 3 He/4 He ratios than Reunion by 0.5^1 Ra units. Since 3 He/4 He ratios of both islands are constant regardless of 4 He concentration (Fig. 2), any di¡erence in 3 He/4 He between them cannot be simply explained by radiogenic ingrowth of 4 He or atmospheric contamination after eruption. Scarsi [21] has shown that a controlled crushing method is a particularly powerful way to resolve trapped He in the inclusions from radiogenic 4 He. The constant 3 He/4 He in the stepwise crushing test (MR94-0202) supports this idea. Relatively uniform 4 He/40 Ar* ratios of Reunion and Older Series of Mauritius are close to predicted mantle production ratios, demonstrating that degassing processes were insigni¢cant. Crustal contamination is another possible process that could potentially decrease 3 He/4 He ratios of ocean island basalts as is the case of Heard Island [26]. Although there is no systematic di¡erence in Sr^Nd isotope ratios between Reunion and Mauritius as seen at Heard Island, contamination of aged oceanic crust could have potentially reduced 3 He/4 He ratios by 0.5^1 Ra units. Another possible explanation for the small di¡erence in the He isotope ratios of Mauritius and Reunion is a small heterogeneity in the source region of mantle

plumes. There are no clear di¡erences in radiogenic isotopes between Mauritius and Reunion but the He isotopes record a small heterogeneity in the source reservoirs. The relatively constant isotopic composition of the hotspot activity establishes that there was minimal temporal change in plume composition over an 8 Myr period. This conclusion contrasts with some models that argue for signi¢cant temporal evolution in plume sources due to radioactive decay [27]. In fact the He isotope ratios increase with time, the opposite of that predicted by such models. On Hawaii the marked He isotope variations are spatially and temporally related to proximity to the centre of the inferred plume track [28]. We therefore argue that volcanism on Reunion and Mauritius records a slight di¡erence in the proportion of melts derived from plume and depleted mantle reservoirs. 5.2. Temporal variation of helium isotope ratios among the three series of Mauritius Island The Intermediate and Younger Series lavas of Mauritius have systematically lower 3 He/4 He ratios than lavas of Reunion and Older Series of Mauritius (Fig. 2). Higher 3 He/4 He of the Older Series are attributed to the mantle plume that is compositionally similar to that of Reunion as discussed above. The lower 3 He/4 He ratios of Intermediate and Younger Series, as well as Rodrigues samples, demonstrate that they are not directly related to the mantle plume activity. In particular, MORB-like 3 He/4 He ratios of the Younger Series suggest that their source is similar to the degassed mantle component (MORB source). This idea is supported by Sr, Nd and Pb isotope data. Older Series lavas have similar isotope ratios to Reunion but isotope ratios of the Younger Series shift towards MORB values (Nohda et al., in preparation). These isotopic characteristics of the Younger Series are similar to the rejuvenated stage of the Hawaiian volcanism [18]. The Intermediate Series has slightly lower 3 He/ 4 He ratio than MORB. Although a single datum is available for this series, it might suggest involvement of another component with a lower 3 He/4 He as recycled components (e.g., [29]) or crustal contamination [26].

EPSL 5991 13-11-01

T. Hanyu et al. / Earth and Planetary Science Letters 193 (2001) 83^98

It is notable that volcanism of the Younger Series of Mauritius happened at the same time as Piton des Neiges and Piton de la Fournaise of Reunion. Although the onset of Piton des Neiges is not certain, it is estimated that the beginning of the activity was at 3^5 Ma on the basis of topography and geoid data [14]. Extrapolation of eruptive emissivity and the total volume of the volcano also implies initiation of activity at 5 Ma [16]. Hence production of the Intermediate Series of Mauritius also occurred simultaneously with some activity on Piton des Neiges which started its activity well after waning of the Older Series of Mauritius. By combining the temporal constraints with the isotopic data, it is possible to establish that plume-derived magmas, with higher 3 He/4 He than MORB, was not simultaneously supplied to Mauritius and Reunion. The feeding system shifted from Mauritius to Reunion at around 5 Ma. The source of the later stage volcanism on the Mauritius Island (Intermediate and Younger Series) was unrelated to the mantle plume. Melting was probably due to thermal rejuvenation of lithosphere/asthenosphere beneath the island. 5.3. The primitive component of the Reunion mantle plume The Reunion hotspot has a helium isotope signature that is distinct from other major hotspots such as Hawaii and Iceland. 3 He/4 He of the Reunion hotspot are higher than the MORB value and demonstrate that the mantle plume contains a less-degassed component, but 3 He/4 He ratios are not as high as the other plumes with comparable or higher buoyancy £ux (e.g., Hawaii and Iceland). Sr and Nd isotope ratios of Reunion are also constant and 87 Sr/86 Sr ratios are moderately radiogenic (0.7040^0.7044). These constraints were previously examined by Graham et al. [5] using the data from Piton de la Fournaise. The new He, Ne and Ar data from Piton des Neiges and Mauritius provide stronger constraints on the nature of the plume source. There are two general models that could potentially explain the origin of the Reunion plume. First a Loihi-like primitive component with an isotopic signature changed by (i) atmospheric and crustal contamination or

93

(ii) MORB source mixing during upwelling of the mantle plume. A second model is that the observed isotopic signature re£ects a source that is distinct from a Loihi-like primitive component. There are three noble gas isotope characteristics of Reunion that require explanation. (1) Constant and moderately high 3 He/4 He (13 Ra; Fig. 1). (2) A slope of the Ne isotope data intermediate between the Loihi and MORB correlation lines (Fig. 3). Although some Ne isotope data were reported previously from Reunion [7,8], a linear correlation between 20 Ne/22 Ne and 21 Ne/22 Ne is clearly identi¢ed by the new data. The highest value of 20 Ne/ 22 Ne, 12.6, is comparable to the maximum value reported from OIBs and MORBs [30,31]. (3) Correlation between 20 Ne/22 Ne and 40 Ar/36 Ar (Fig. 4). Although possible atmospheric and crustal contamination will alter Ne and Ar isotope ratios, the correlation between them enables the isotope ratios of the plume endmember to be inferred. On the assumption that the maximum 20 Ne/22 Ne ratios of the plume endmember is the solar value of 13.8, the 40 Ar/36 Ar of the Reunion plume is estimated to be 6100 þ 800 (1c), which is within the range of 40 Ar/36 Ar of £uid inclusions in Reunion dunites (500^8500; [32]). This 40 Ar/36 Ar ratio is distinct from that of the MORB source [30] but overlaps with that of other hotspots (Fig. 4), for example Loihi, 2500^6000 [33] or V8000 [31]. 5.3.1. Atmospheric and crustal contamination Although atmospheric and crustal contamination to Loihi-like primitive component will reduce 3 He/4 He, the slope of the correlation trends in the Ne three-isotope diagram cannot be altered by these processes (Fig. 3). In addition it is unrealistic to argue that constant 3 He/4 He for millions of years, regardless of He concentration, was maintained by variable atmospheric and crustal contamination. By its very nature shallow-level contamination processes should act to di¡erent degrees in each magma batch and eruption. 5.3.2. MORB source mixing Mixing of a Loihi-like primitive component and a MORB source may produce moderately high 3 He/4 He and an intermediate slope of the 20 Ne/ 22 Ne^21 Ne/22 Ne correlation. The atmosphere-free

EPSL 5991 13-11-01

94

T. Hanyu et al. / Earth and Planetary Science Letters 193 (2001) 83^98

21

Ne/22 Ne ratio, (21 Ne/22 Ne)m , can be calculated by extrapolating the Ne data point from atmospheric value to the solar 20 Ne/22 Ne (13.8), and 4 He/3 He versus (21 Ne/22 Ne)m data are plotted in Fig. 5. Reunion, Iceland and Samoa lie between the data of Hawaii and MORBs. Bulk mixing calculations demonstrate that more than 99.5% degassed mantle (MORB source) is required to explain He and Ne isotopes of Reunion by assuming endmember parameters from the steady state model [34]. If [3 He]plume /[3 He]MORB source is 5 which is much lower than the estimated value (200) by the steady state model, as suggested by [35], the calculated mixing ratio for He and Ne is modi¢ed, but the MORB source still should contribute V85% to the Reunion mantle plume. This conclusion is in contradiction to the Sr and Nd isotope ratios of Reunion with 87 Sr/86 Sr higher than MORB and Loihi. The calculated mixing proportion can be reduced if melted/unmelted plume material is assumed to mix with incipient melts from a

Fig. 5. (21 Ne/22 Ne)m against 4 He/3 He for Reunion, MORB, Hawaii, Samoa and Iceland [24,25,30,31,33,38,49^52]. (21 Ne/ 22 Ne)m denotes the atmosphere-free 21 Ne/22 Ne ratio which can be calculated by extrapolating the Ne data point from atmospheric value to the plume endmember at 20 Ne/ 22 Ne = 13.8. The mixing calculation between primordial mantle (P Mantle) and MORB source is carried out by assuming the endmember isotopic composition and elemental abundances from [34], but 4 He/3 He is modi¢ed so that 3 He/4 He is 50 Ra. Tics and numbers on the mixing line denote the mixing ratio of P Mantle against MORB source. The arrow schematically shows the evolution curve of He and Ne isotopic ratios in the source material with lower 3 He/22 Ne than Porcelli and Wasserburg's model [34].

Fig. 6. Overall correlation between 4 He/21 Ne* and 4 He/40 Ar* (21 Ne* is non-atmospheric components of 21 Ne) among OIBs and MORBs demonstrate elemental fractionation, but most of Reunion samples plot close to the mantle production ratios.

MORB source. This is because He and Ne behave like incompatible elements. This model has previously been proposed to account for the hyperbolic trend between 4 He/3 He and (21 Ne/22 Ne)m where plume^ridge interaction occurs on both a local scale (Shona Ridge, [36]; southern East Paci¢c Rise, [37]) and regional scale (Iceland, [38]). Sr and Nd, however, are also incompatible elements and are concentrated in MORB melt. This mixing scenario therefore does not explain the combined noble gas and Sr and Nd isotope ratios of Reunion unless there is a strong decoupling between noble gases and solid elements. Sarda et al. [39] explained the hyperbolic trend at Shona and Discovery Ridge as a consequence of fractionation of He/Ne by vesicle loss from a MORB source melt prior to mixing with a plume-derived melt. There is, however, no evidence of elemental fractionation in 4 He/21 Ne* or 4 He/40 Ar* for Reunion samples (Fig. 6; 21 Ne* and 40 Ar* are non-atmospheric component of 21 Ne and 40 Ar, respectively). Consequently He^Ne isotope decoupling in Reunion lavas cannot be explained in the same way as has been proposed for some cases of plume^ridge interaction. By a process of elimination it follows that the observed He^Ne^Ar isotope ratios dominantly re£ect the isotopic characteristics of the source domain of the Reunion mantle plume. This source

EPSL 5991 13-11-01

T. Hanyu et al. / Earth and Planetary Science Letters 193 (2001) 83^98

has di¡erent isotopic characteristics to the Loihi source. The Reunion source does, however, have a primitive signature of high 3 He/4 He and 20 Ne/ 22 Ne. These data imply that the lower mantle is not homogeneous and contains several relatively less-degassed reservoirs. The Reunion source has lower time-integrated He/(U+Th) and Ne/(U+Th) ratios than the Loihi source. Possible explanations to explain the di¡erent noble gas vs. lithophile elements ratios of Reunion and Loihi-like mantle sources include; (1) addition of subducted slabs into Loihi-like mantle source and (2) di¡erent degrees of degassing in the early stage of mantle evolution. Irrespective of the explanation, the distinct sources imply that several less-degassed reservoirs have been generated in the past and that reservoirs with di¡erent He and Ne isotope ratios have evolved and remained isolated in the convectively stirred mantle. The recycling model requires that a primitive (less-degassed) component is mixed with a subducted component to form the source of the plume. The noble gas composition of recycled oceanic lithosphere will depend on the mobility of noble gases, U, Th and K during subduction. He and Ne will not be extensively delivered into the mantle with subducted slabs although Ar may be in part recycled [40,41]. Due to their lithophile nature elements such as K and U will be e¤ciently stripped from subducted slabs by dehydration due to high solubility in £uids [42]. It is probable, however, that some part of U and Th budget of subducted materials can be recycled into the mantle [43]. If this is the case, recycled subducted material supply U and Th to the mantle, resulting in reduced (He,Ne)/(U+Th) ratios. Due to the relatively constant K/U ratio of MORB and OIB [44], one can assume that a proportion of the K budget in oceanic lithosphere was also recycled into the mantle. The exact e¡ect on Ar/K ratios is di¤cult to constrain because Ar recycling is still in debate [40,41,45]. The moderately radiogenic 87 Sr/86 Sr of Reunion are compatible with a recycling model. 187 Os/186 Os ratios of Reunion, however, are indistinguishable from those of MORBs [46] suggesting that any recycled component cannot include oceanic crust with a high Re/Os ratio. We therefore conclude that a recycled model does

95

not easily explain the observed noble gas^Sr^Nd^ Os isotope systematics. The alternative explanation of the source characteristics of the Reunion hotspot is that the Reunion source experienced a larger degree of degassing than the Loihi-like source in the early evolutionary stage of the Earth. If He and Ne degassed more extensively in the Reunion source than in the Loihi source, the former would have lower time-integrated (He,Ne)/(U+Th) ratios. This interpretation requires that primitive portions of mantle which experienced various degrees of degassing/di¡erentiation should survive in the convective mantle as small-scale blobs [47] or large-scale piles [48]. Possible decoupling between He and Ne isotope ratios in Reunion would be in favour of the second option. He and Ne isotope evolution of Loihi and MORB sources are well explained by radiogenic 4 He and nucleogenic Ne ingrowth from primordial solar component [49]. Provided all the mantle reservoirs have uniform 3 He/22 Ne ratio, all the data from OIBs and MORBs should de¢ne a straight line in Fig. 5 according to the model [49]. Reunion and some OIBs, however, plot below the line connecting Loihi and MORBs in this ¢gure. This clearly indicates that the Reunion source has lower 3 He/22 Ne ratios than the Loihi and MORB sources. We interpret this distinction to be a consequence of variable elemental fractionation by degassing in the early stage of mantle evolution. Subsequent nucleogenic ingrowth of 4 He and Ne will change He isotope ratios faster than 21 Ne/22 Ne, as schematically shown in Fig. 5. If various 3 He/22 Ne ratios were produced by heterogeneous early degassing of the mantle this would have caused di¡erent rates of change in 3 He/4 He and 21 Ne/22 Ne among mantle reservoirs. This scenario explains the He^Ne isotopic decoupling of Reunion and possibly some other OIBs. Due to large atmospheric contamination 40 Ar/ 36 Ar ratios of OIBs are not well constrained. Current estimates of endmember 40 Ar/36 Ar of Reunion and Loihi, however, suggest similar timeintegrated Ar/K ratios, thereby the extent of degassing of Ar in the two sources would be equivalent. Further heavy noble gases investigations will provide more constraints on the deple-

EPSL 5991 13-11-01

96

T. Hanyu et al. / Earth and Planetary Science Letters 193 (2001) 83^98

tion of light noble gas attributed to various degrees of degassing in the early stage of mantle evolution.

Promotion of Science to T.H. (9034). This is NSG publication No. 20010803.[AH] References

6. Conclusion An extensive data set of He, Ne and Ar isotopes are reported from two volcanoes on Reunion Island and three magmatic series from Mauritius Island. 3 He/4 He ratios of Reunion and Mauritius (Older Series) are almost constant and higher than MORB value, indicating that the Reunion mantle plume includes a less-degassed mantle component and has been homogeneous for an 8 Myr period. Combined He, Ne and Ar isotopes demonstrate that the less-degassed component of the Reunion plume is distinct from that of Hawaii. This conclusion establishes that the mantle consists of several relatively less-degassed reservoirs that are isolated in the lower mantle. Recycling of oceanic lithosphere into the lower mantle could potentially account for some of the isotopic di¡erences between the Hawaiian and Reunion plume. The most probable explanation of the isotopic data, however, is variable degrees of degassing of the lower mantle in the early stage of Earth's evolution, followed by isolated convection to preserve chemically heterogeneous domains in the lower mantle. Acknowledgements Sampling was carried out as an International Scienti¢c Research Program in 1994 and 1995 sponsored by the Ministry of Education, Science, Culture and Sports, Japan (06041026). T. Fujii, S. Nakada, Y. Tatsumi and A. Yasuda are acknowledged for assistance with ¢eld work. We thank I. Nikogosian for determining the major element composition of olivines and melt inclusions. T.H. is grateful to S. Nakai for his support in the course of the study. P. Burnard and an anonymous reviewer are thanked for their thoughtful reviews. This study was supported in part by the Research Fellowships of the Japan Society for the

[1] C.J. Alle©gre, A.W. Hofmann, K. O'Nions, The argon constraints on mantle structure, Geophys. Res. Lett. 23 (1996) 3555^3557. [2] M. Gurnis, G.F. Davies, The e¡ect of depth-dependent viscosity on convective mixing in the mantle and the possible survival of primitive mantle, Geophys. Res. Lett. 13 (1986) 541^544. [3] F. Albare©de, Time-dependent models of U^Th^He and K^Ar evolution and the layering of mantle convection, Chem. Geol. 145 (1998) 413^429. [4] M.R. Fisk, B.G.J. Upton, C.E. Ford, Geochemical and experimental study of the genesis of magmas of Reunion Island, Indian Ocean, J. Geophys. Res. 93 (1988) 4933^ 4950. [5] D. Graham, J. Lupton, F. Albare©de, M. Condomines, Extreme temporal homogeneity of helium isotopes at Piton de la Fournaise, Reunion Island, Nature 347 (1990) 545^548. [6] J.C. Lassiter, E.H. Hauri, Osmium-isotope variations in Hawaiian lavas: evidence for recycled oceanic lithosphere in the Hawaiian plume, Earth Planet. Sci. Lett. 164 (1998) 483^496. [7] I. Kaneoka, N. Takaoka, B.G.J. Upton, Noble gas systematics in basalts and a dunite nodule from Reunion and Grand Comore Islands, Indian Ocean, Chem. Geol. 59 (1986) 35^42. [8] T. Staudacher, P. Sarda, C.J. Alle©gre, Noble gas systematics of Reunion Island, Indian Ocean, Chem. Geol. 89 (1990) 1^17. [9] I. McDougall, F.H. Chamalaun, Isotopic dating and geomagnetic polarity studies on volcanic rocks from Mauritius, Indian Ocean, Geol. Soc. Am. Bull. 80 (1969) 1419^ 1442. [10] N.H. Sleep, Hotspots and mantle plumes: some phenomenology, J. Geophys. Res. 95 (1990) 6715^6736. [11] I. McDougall, The geochronology and evolution of the young volcanic island of Reunion, Indian Ocean, Geochim. Cosmochim. Acta 35 (1971) 261^288. [12] P.-Y. Gillot, P. Nativel, Eruptive history of the Piton de la Fournaise volcano, Reunion Island, Indian Ocean, J. Volcanol. Geotherm. Res. 36 (1989) 53^65. [13] M.A. Richards, R.A. Duncan, V.E. Courtillot, Flood basalts and hot-spot tracks plume heads and tails, Science 246 (1989) 103^107. [14] A. Bonneville, J.P. Barriot, R. Bayer, Evidence from geoid data of a hotspot origin for the southern Mascarene Plateau and Mascarene Islands (Indian Ocean), J. Geophys. Res. 93 (1988) 4199^4212. [15] W.J. Morgan, Rodriguez, Darwin, Amsterdam, T, A sec-

EPSL 5991 13-11-01

T. Hanyu et al. / Earth and Planetary Science Letters 193 (2001) 83^98

[16] [17] [18]

[19] [20] [21]

[22]

[23]

[24] [25] [26] [27] [28]

[29] [30] [31]

ond type of hotspot island, J. Geophys. Res. 83 (1978) 5355^5360. P.-Y. Gillot, J.-C. Lefevre, P.-E. Nativel, Model for the structural evolution of the volcanoes of Reunion Island, Earth Planet. Sci. Lett. 122 (1994) 291^302. A.N. Baxter, Geochemistry and petrogenesis of primitive alkali basalt from Mauritius, Indian Ocean, Geol. Soc. Am. Bull. 87 (1976) 1028^1034. P. Stille, D.M. Unruh, M. Tatsumoto, Pb, Sr, Nd, and Hf isotopic constraints on the origin of Hawaiian basalts and evidence for a unique mantle source, Geochim. Cosmochim. Acta 50 (1986) 2303^2319. I. McDougall, B.G.J. Upton, W.J. Wadsworth, A geological reconnaissance of Rodriguez Island, Indian Ocean, Nature 206 (1965) 26^27. A.N. Baxter, B.G.J. Upton, W.M. White, Petrology and geochemistry of Rodrigues Island, Indian Ocean, Contrib. Mineral. Petrol. 89 (1985) 90^101. P. Scarsi, Fractional extraction of helium by crushing of olivine and clinopyroxene phenocrysts: E¡ects on the 3 He/4 He measured ratio, Geochim. Cosmochim. Acta 64 (2000) 3751^3762. M.C. van Soest, D.R. Hilton, R. Kreulen, Tracing crustal and slab contributions to arc magmatism in the Lesser Antilles island arc using helium and carbon relationship in geothermal £uids, Geochim. Cosmochim. Acta 62 (1998) 3323^3335. T. Staudacher, M.D. Kurz, C.J. Alle©gre, New noble-gas data on glass samples from Loihi Seamount and Hualalai and on dunite samples from Loihi and Reunion Island, Chem. Geol. 56 (1986) 193^205. P. Sarda, T. Staudacher, C.J. Alle©gre, Neon isotopes in submarine basalts, Earth Planet. Sci. Lett. 91 (1988) 73^ 88. M. Honda, I. McDougall, D.B. Patterson, A. Doulgeris, D.A. Clague, Possible solar noble-gas component in Hawaiian basalts, Nature 349 (1991) 149^151. D.R. Hilton, J. Barling, G.E. Wheller, E¡ect of shallowlevel contamination on the helium isotope systematics of ocean^island lavas, Nature 373 (1995) 330^333. C. Class, S.L. Goldstein, S.J.G. Galer, D. Weis, Young formation age of a mantle plume source, Nature 362 (1993) 715^721. D.J. DePaolo, J.G. Bryce, A. Dodson, D.L. Shuster, B.M. Kennedy, Isotopic evolution of Mauna Loa and the chemical structure of the Hawaiian plume, Geochem. Geophys. Geostystems 2 (2001) 2000GC000139. T. Hanyu, I. Kaneoka, The uniform and low 3 He/4 He ratios of HIMU basalts as evidence for their origin as recycled materials, Nature 390 (1997) 273^276. M. Moreira, J. Kunz, C. Alle©gre, Rare gas systematics in popping rock: Isotopic and elemental compositions in the upper mantle, Science 279 (1998) 1178^1181. M. Trielo¡, J. Kunz, D.A. Clague, D. Harrison, C.J. Alle©gre, The nature of pristine noble gases in mantle plumes, Science 288 (2000) 1036^1038.

97

[32] P.G. Burnard, F.M. Stuart, G. Turner, C-He-Ar variations within a dunite nodule as a function of £uid inclusion morphology, Earth Planet. Sci. Lett. 128 (1994) 243^ 258. [33] P.J. Valbracht, T. Staudacher, A. Malaho¡, C.J. Alle©gre, Noble gas systematics of deep rift zone glasses from Loihi Seamount, Hawaii, Earth Planet. Sci. Lett. 150 (1997) 399^411. [34] D. Porcelli, G.J. Wasserburg, Mass transfer of helium, neon, argon, and xenon through a steady-state upper mantle, Geochim. Cosmochim. Acta 59 (1995) 4921^ 4937. [35] D.R. Hilton, M.F. Thirlwall, R.N. Taylor, B.J. Murton, A. Nichols, Controls on magmatic degassing along the Reykjanes Ridge with implications for the helium paradox, Earth Planet. Sci. Lett. 183 (2000) 43^50. [36] M. Moreira, T. Staudacher, P. Sarda, J.-G. Schilling, C.J. Alle©gre, A primitive plume neon component in MORB: The Shona ridge-anomaly, South Atlantic (51^52³S), Earth Planet. Sci. Lett. 133 (1995) 367^377. [37] S. Niedermann, W. Bach, J. Erzinger, Noble gas evidence for a lower mantle component in MORBs from the southern East Paci¢c Rise: Decoupling of helium and neon isotope systematics, Geochim. Cosmochim. Acta 61 (1997) 2697^2715. [38] M. Moreira, K. Breddam, J. Curtice, M.D. Kurz, Solar neon in the Icelandic mantle: new evidence for an undegassed lower mantle, Earth Planet. Sci. Lett. 185 (2001) 15^23. [39] P. Sarda, M. Moreira, T. Staudacher, J.-G. Schilling, C.J. Alle©gre, Rare gas systematics on the southernmost MidAtlantic Ridge: constraints on the lower mantle and the Dupal source, J. Geophys. Res. 105 (2000) 5973^5996. [40] P. Sarda, M. Moreira, T. Staudacher, Argon^lead isotopic correlation in Mid-Atlantic ridge basalts, Science 283 (1999) 666^668. [41] P.G. Burnard, Origin of argon^lead isotopic correlation in basalts, Science 286 (1999) 871. [42] H. Keppler, Constraints from partitioning experiments on the composition of subduction-zone £uids, Nature 380 (1996) 237^240. [43] T. Elliott, A. Zindler, B. Bourdon, Exploring the kappa conundrum: the role of recycling in the lead isotope evolution of the mantle, Earth Planet. Sci. Lett. 169 (1999) 129^145. [44] K.P. Jochum, A.W. Hofmann, E. Ito, H.M. Seufert, W.M. White, K, U, and Th in mid-oceanic ridge basalt glasses and heat production, K/U and K/Rb in the mantle, Nature 306 (1983) 431^436. [45] T. Staudacher, C.J. Alle©gre, Recycling of oceanic crust and sediments: the noble gas subduction barrier, Earth Planet. Sci. Lett. 89 (1988) 173^183. [46] M. Roy-Barman, C.J. Alle©gre, 187 Os/186 Os in oceanic island basalts: tracing oceanic crust recycling in the mantle, Earth Planet. Sci. Lett. 129 (1995) 145^161. [47] T.W. Becker, J.B. Kellogg, R.J. O'Connell, Thermal con-

EPSL 5991 13-11-01

98

T. Hanyu et al. / Earth and Planetary Science Letters 193 (2001) 83^98

straints on the survival of primitive blobs in the lower mantle, Earth Planet. Sci. Lett. 171 (1999) 351^365. [48] P.J. Tackley, Mantle convection and plate tectonics: Toward an integrated physical and chemical theory, Science 288 (2000) 2002^2007. [49] M. Honda, I. McDougall, D.B. Patterson, A. Doulgeris, D.A. Clague, Noble gases in submarine pillow basalt glasses from Loihi and Kilauea, Hawaii: A solar component in the Earth, Geochim. Cosmochim. Acta 57 (1993) 859^874.

[50] R.J. Poreda, K.A. Farley, Rare gases in Samoan xenoliths, Earth Planet. Sci. Lett. 113 (1992) 129^144. [51] D. Harrison, P. Burnard, G. Turner, Noble gas behaviour and composition in the mantle: constraints from the Iceland Plume, Earth Planet. Sci. Lett. 171 (1999) 199^ 207. [52] E.T. Dixon, M. Honda, I. McDougall, I.H. Campbell, I. Sigurdsson, Preservation of near-solar neon isotopic ratios in Icelandic basalts, Earth Planet. Sci. Lett. 180 (2000) 309^324.

EPSL 5991 13-11-01