13. ISOTOPIC EVIDENCE FOR THE ORIGIN OF BONINITES AND ... › ... › VOLUME › CHAPTERS ›...

Fryer, P., Pearce, J. A., Stokking, L., et al., 1992Proceedings of the Ocean Drilling Program, Scientific Results, Vol. 125

13. ISOTOPIC EVIDENCE FOR THE ORIGIN OF BONINITES AND RELATED ROCKS DRILLED INTHE IZU-BONIN (OGASAWARA) FOREARC, LEG 1251

Julian A. Pearce,2 Matthew F. Thirlwall,3 Gerry Ingram,3 Bramley J. Murton,2,4 Richard J. Arculus,5

and Sieger R. van der Laan6

ABSTRACT

Twenty-six samples representing the wide range of lithologies (low- and intermediate-Ca boninites and bronzite andesites,high-Ca boninites, basaltic andesites-rhyolites) drilled during Leg 125 at Sites 782 and 786 on the Izu-Bonin outer-arc high havebeen analyzed for Sr, Nd, and Pb isotopes. Nd-Sr isotope covariations show that most samples follow a trend parallel to a linefrom Pacific MORB mantle (PMM) to Pacific Volcanogenic sediment (PVS) but displaced slightly toward more radiogenic Sr.Pb isotope covariations show that all the Eocene-Oligocene samples plot along the Northern Hemisphere Reference Line,indicating little or no Pb derived from subducted pelagic sediment in their source. Two young basaltic andesite clasts withinsediment do have a pelagic sediment signature but this may have been gained by alteration rather than subduction. In all isotopicprojections, the samples form consistent groupings: the tholeiites from Site 782 and Hole 786A plot closest to PMM, the boninitesand related rocks from Sites 786B plot closest to PVS, and the boninite lavas from Hole 786A and late boninitic dikes from Hole786B occupy an intermediate position. Isotope-trace element covariations indicate that these isotopic variations can be explainedby a three-component mixing model. One component (A) has the isotopic signature of PMM but is depleted in the moreincompatible elements. It is interpreted as representing suboceanic mantle lithosphere. A second component (B) is relativelyradiogenic (εNd = ca 4-6; 2 0 6Pb/2 0 4Pb = ca 19.0-19.3; εSr = ca -10 to -6)). Its trace element pattern has, among othercharacteristics, a high Zr/Sm ratio, which distinguishes it from the "normal" fluid components associated with subduction andhotspot activity. There are insufficient data at present to tie down its origin: probably it was either derived from subductedlithosphere or volcanogenic sediment fused in amphibolite facies; or it represents an asthenospheric melt component that has beenfractionated by interaction with amphibole-bearing mantle. The third component (C) is characterized by high contents of Sr andhigh εSr values and is interpreted as a subducted fluid component. The mixing line on a diagram of Zr/Sr against ε Sr suggeststhat component C may have enriched the lithosphere (component A) before component B. These components may also be presenton a regional basis but, if so, may not have had uniform compositions. Only the boninitic series from nearby Chichijima wouldrequire an additional, pelagic sediment component. In general, these results are consistent with models of subduction of ridgesand young lithosphere during the change from a ridge-transform to subduction geometry at the initiation of subduction in theWestern Pacific.

INTRODUCTION

During Ocean Drilling Program Leg 125, scientists drilled twosites (782 and 786) on the Izu-Bonin (Ogasawara) outer-arc high(Fig. 1). Site 782 was originally designated as the main site for deepbasement drilling. Two adjacent holes were drilled. Hole 782A wasdrilled to 476.8 mbsf (meters below seafloor), recovering 409.2 m ofsediments and 67.6 m of basement. Hole 782B recovered just 9.6 mof basement between 4593 and 468.9 mbsf. The second site, Site 786,provided a deep, forearc crustal section. Two adjacent holes also weredrilled at this site. Hole 786A was drilledto 167 mbsf with an extendedcore barrel (XCB), recovering 125 m of forearc sediment and about42 m of basement. Hole 786B was drilled with a rotary core barrel(RCB), recovering a further 666 m of basement between 162.5 and828.6 mbsf.

The rocks recovered from these sites have been studied in termsof their stratigraphy and major element contents (Arculus et al., this

Fryer, P., Pearce, J. A., Stokking, L., et al., 1991. Proc. ODP, Sci. Results, 125:College Station, TX (Ocean Drilling Program).

2 Department of Geological Sciences, University of Durham, Durham DH1 3LE,United Kingdom.

3 Department of Geology, RHBNC, Egham Hill, Egham, Surrey, TW20 OEX, UnitedKingdom.

4 Institute of Oceanographic Sciences, Wormley, Godalming, Surrey GU8 5UB,United Kingdom.

5 Department of Geology, University of New England, Armidale, N.S.W. 2351,Australia.

6 Department of Geology and Geophysics, University of Hawaii, Honolulu, Hawaii96822, USA.

volume), their structure (Lagabrielle et al., this volume), their age(Mitchell et al., this volume), their petrography and phase relations(van der Laan et al., this volume) and their trace element charac-teristics (Murton et al., this volume). Site 782 yielded a small numberof clasts of tholeiitic andesites, dacites, and rhyolites which havebeen grouped as a single lithostratigraphic unit. At Site 786, however,drilling penetrated a complex boninitic volcanic edifice with a well-defined stratigraphy (Fig. 2). The top of the edifice, a volcanic breccia,was drilled at Hole 786A. The remainder of the edifice was drilled atHole 786B. In its upper and middle parts, the edifice comprises aseries of lavas, dikes, breccias, and pyroclastic flows of intermediate-Ca boninite, intermediate-Ca bronzite andesite, and andesite-rhyolitecomposition. These overlie a series of low-Ca boninite pillow lavasand hyaloclastite breccias which in turn overlie a swarm of low-Cabronzite andesite dikes. This basal sequence is cut by feeder dikes tothe main part of the edifice. This volcanic structure subsequently wasintruded by a series of dikes and/or sills of intermediate- to high-Caboninite composition. The edifice has been subdivided into 35 litho-stratigraphic units (Unit IV-0 in Hole 786A and Units IV1-IV34 inHole786B).

K-Ar isochrons suggest an age of 41.3 ± 0.5 Ma (middle-lateEocene) for the main constructional phase of the edifice and an ageof 34.6 ± 0.7 Ma (earliest Oligocene) for the later intrusive phase(Mitchell et al., this volume). A number of ages between 17 and 9 Mamay be evidence for one or more subsequent intrusive episodes ormay indicate that alteration of the basement continued for a longperiod after its formation. The tholeiitic lavas at Site 782 gave noconsistent age, although the oldest age of 36 Ma suggests that thesewere broadly contemporaneous with the boninite series volcanicsdrilled at Site 786.

237

J. A. PEARCE ET AL.

40-N

130 140° 150 142°

Figure 1. Regional setting of the Izu-Bonin (Ogasawara) and Mariana forearc terranes and location of ODP Leg 125 outer-arc high Sites 782 and 786. Contoursare in kilometers below sea level.

The aims of this study are to use isotopic data to characterize thesources of the igneous rocks recovered at these two sites and toexamine the results in the light of published isotope data from theIzu-Bonin-Mariana forearc terrane.

SAMPLE SELECTION

Twenty-six samples from the various lithostratigraphic units inSites 782 and 786 were analyzed for Sr, Nd, and Pb isotopes. Sampleselection was carried out mainly during the leg from the mosthomogeneous and apparently least altered of the rock types recovered.The pervasive alteration meant, however, that no truly fresh rockscould be sampled. Classification of the samples chosen was carried outon the basis of the MgO-SiO2 and CaO-SiO2 covariation diagramsdescribed in Arculus et al. (this volume) and reproduced in Figure 3.This figure also demonstrates that the samples selected for isotopeanalysis are representative of the range of compositions encounteredat the two sites. For convenience of presentation, the samples have beenclustered into seven groups in this and subsequent diagrams on thebasis of rock type, location and age. These groups are as follows:

1. low-Ca boninites and low-Ca bronzite andesites from the baseof the 41 Ma edifice in Hole 786B (the 786B-LCB group);

2. intermediate-Ca boninites and intermediate-Ca bronzite ande-sites from the lower/middle to upper part of the 41 Ma edifice in Hole786B (the 786B-ICB group);

3. andesites, dacites, and rhyolites from the middle-upper part ofthe 41 Ma edifice in Hole 786B (the 786B-ADR group);

4. intermediate- and high-Ca boninites forming the late (34-35Ma) dikes or sills in Hole 786B (the 786B-late dike group);

5. intermediate-Ca boninites from the basement in Hole 786A (the786A-ICB group);

6. basaltic andesites from intra-sediment clasts in Hole 786A (the786A-BA group);

7. tholeiitic andesites and dacites from the basement in Holes 786Aand 786B (the 782-ThAD group).

The mineralogies and ages of the samples are relevant to thetreatment of the isotopic data, and thus are described briefly below.

1. 786B-LCB group. Samples of low-Ca boninite pillow lavas weretaken from Cores 57R-4 and 62R-3 for isotopic analysis. Both samplescontain 5%-10% of phenocrysts, the sample from Core 57R-4 contain-ing olivine and orthopyroxene, that from Core 62R-3 containing ortho-pyroxene only. Both have a cryptocrystalline to glassy groundmass. Bothrocks are very altered, in common with all samples from this group.Orthopyroxenes and glass are largely replaced by brown clay, olivine isreplaced by serpentine, oxide, and calcite, and vesicles and small fracturescontain zeolite and brown clay. K-Ar dating gave an age of 38.4 ± 2.3Ma for 57R-4, close to the 41 Ma isochron taken as the edifice age.

The bronzite andesite dike, taken from Core 70R-1, has abundantphenocrysts of orthopyroxene, clinopyroxene, and, less commonly,Plagioclase in a fine-grained groundmass of Plagioclase laths, clino-pyroxene, and opaque minerals. Orthopyroxene and most of thegroundmass are replaced by chlorite, clay minerals, and carbonate.K-Ar dating gave an age of 38.7 ± 0.8, close to the 41 Ma isochron.

2. 786B-ICB group. Intermediate-Ca boninites were sampledfrom Core 51R-1, in a breccia unit just above the low-Ca boninitebasement, and from Core 11R-1, in a polymict breccia near the top ofthe sequence. Both rocks carry abundant phenocrysts of Plagioclase,orthopyroxene, and clinopyroxene in a groundmass of Plagioclase,pyroxenes, an opaque phase, and interstitial glass and contain about15% of unfilled vesicles. The orthopyroxene phenocrysts and thematrix glass are replaced by brown clays. Both samples are vesicular,the vesicles from the sample in Core 11R-1 having a green chloriticlining. The K-Ar age of the sample from 51R-1 is 33.6 ±0.6 Ma,interpreted as having been reset from a 41 Ma age by the 34-35 Madike intrusion episode. The age of the sample from Core 11R-1 is 25.7± 1.2 Ma, again a reset age.

238

ORIGIN OF BONINITES AND RELATED ROCKS

Intermediate-Ca bronzite andesites are taken from breccia units inCores 1R-1 and 30R-1. They are both highly plagioclase-clinopyroxene-orthopyroxene phyric with rare olivine in the former. Groundmasses aredominated by Plagioclase microliths and pyroxenes with interstitial glass.The glass, and many orthopyroxene crystals, have been replaced by clayminerals. A sample close to that from Core 1R-1 and the sample fromCore 30R-1 gave ages of 41.0 ± 1.1 and 39.8 + 1.3 Ma respectively.

3. 786B-ADR group. Andesites were sampled at four localities inthe top to middle part of Hole 786B, in Cores 8R-1, 19R-1, 21R-1,and 49R-4. All rocks contained abundant phenocrysts of Plagioclaseand clinopyroxene and rare orthopyroxene, in groundmasses dom-inated by Plagioclase and opaque phases with interstitial glass. Thelatter is now altered to clay minerals, totally in the samples fromCores 8R-1 and 21R-1 (129-131 cm) and partially in samples fromCores 19R-1 and 49R-4. The sample from Core 49R-4 also containsrare, replaced olivine phenocrysts and more clinopyroxene in thegroundmass. K-Ar dating gave reset ages of 9.2 ± 0.4 and 16.7 ±0.4 Ma for the samples from Cores 8R-1 and 21R-1 respectively.

Dacites and rhyolites were sampled from lavas in the upper part ofthe Hole 786B (from Cores 21R-1 and 35R-1) and from dikes within thelow-Ca boninite series basement (from Cores 63R-2 and 66R-1). Thelava samples are highly vesicular and contain phenocrysts of Plagioclaseand clinopyroxene in a matrix similar to that of the andesites. The dikesmay have had a similar mineralogy but have been intensely altered to anassemblage dominated by clay minerals and chlorite. K-Ar dating of adike in Core 66R-2 gave an age of 41.3 ± 0.8 Ma.

4. 786B-late dike group. Samples of four late dikes have been takenfrom Hole 786B, from Cores 5R-2, 21R-2, 40R-2, and 67R-1. Twosamples (from Cores 21R-2 and 67R-1) classify as intermediate-Caboninite. They are olivine-orthopyroxene-clinopyroxene phyric in afine-grained groundmass of Plagioclase, clinopyroxene, an opaquephase, and cryptocrystalline-glassy groundmass. The other two sam-ples (from Cores 5R-2 and 40R-2) classify as high-Ca boninites. Theyare olivine-orthopyroxene and olivine-orthopyroxene-clinopyroxenephyric, respectively, in a groundmass similar to that of the inter-mediate-Ca boninites. All samples contain a small proportion of un-filled vesicles. Variable clay mineral alteration of the glassygroundmass and of many olivine and orthopyroxene phenocrysts iscommon to all samples. Olivine phenocrysts in the sample from Core21R-2 have been replaced by carbonate, resulting in loss of Mg andgain in Ca. The MgO and CaO values in Table 1 have been adjusted toshow the pre-carbonation composition.

All four samples have been dated by the K-Ar method. Those fromCores 21R-2 and 40R-2 both plot on the 34-5 Ma isochron. Core67R-1 gave an older age of 41 ± 1.0 Ma and would have been classifiedwith the 786B-ICB group were it not for its isotope and trace elementcomposition (see later). A sample in the same unit as Core 5R-2 gavean age of 17.3 ±0.9 Ma. It is not clear at present whether this unitrepresents a separate intrusive phase or whether it had an older(34-35 Ma) age that has been reset.

5.786A-1CB group. The intermediate-Ca boninite samples were takenfrom near the base of Hole 786A, from Cores 16X-1 and 17X-1. Theyare both moderately vesicular and contain phenocrysts of olivine, ortho-pyroxene, clinopyroxene, and rare Plagioclase in a matrix of plagio-clase,clinopyroxene, opaque minerals, and altered glass. The sample from Core16X-1 gave a K-Ar age of 15.3 ± 1.0, taken as reset in view of its Eocenebiostratigraphic age. In the absence of more definitive data, ages of 35Ma have been assumed for these samples.

6. 786A-BA group. Sample 12X-1 and 6X-5 represent the basalticandesite clasts recovered from Units I and III of the sedimentsoverlying the basement at Site 786A. Both contain clinopyroxene andPlagioclase microliths in an altered glassy matrix. The sample fromCore 12X-1 gave a K-Ar age of 9.3 ±0.4 Ma. This could be a true ageif the sample is from a dike or sill rather than a lava or if the enclosinglate Eocene sediments have been reworked. More probably, however,the age has been reset. An arbitrary age of 17 Ma is taken for theisotope calculations. The sample from Core 6X-5 has not been dated

but must be equal to or less than its stratigraphic age of upper MiddleMiocene (Fryer, Pearce, Stokking, et al., 1990). An age of 10 Ma hasbeen assumed for this sample.

7. Site 782-ThAD group. Three samples were analyzed fromHole 782A, andesites from Cores 49X-CC and 50X and a dacite fromCore 45X-1. These samples are all highly vesicular and containPlagioclase and clinopyroxene phenocrysts in an altered (clay-rich)glassy matrix. The samples from Cores 45X-1 and 50X-1 have beendated, giving ages of 18.1 ± 0.6 and 30.7 ±1.3 Ma respectively.Another sample from Site 782 gave an age of 36 ± 1.1 Ma, probablycloser to the real age of the basement at this site.

ANALYTICAL TECHNIQUES

All samples chosen for isotope analysis have been analyzed for majorelements and 32 trace elements, including those relevant to the isotopesystems (Table 1). The major element data are taken from Arculus et al.(this volume). Trace element analyses were carried out by ICP-MS at theUniversity of Durham, using duplicate preparations to improve thequality of the data Because the samples were prepared on board theJOIDES Resolution using a tungsten carbide tema mill, they haveexperienced Ta, Nb, and Hf contamination. The concentrations of theseelements have therefore been obtained by analyses of adjacent samplescrushed using an agate tema mill. These analyses were carried out bothby INAA and ICP-MS. The data obtained for this paper are also incor-porated into the full trace element data set in Murton et al. (this volume),to which the reader is referred for details of the analytical techniques.Note that there may be slight discrepancies between Table 1 and theequivalent table in Murton et al. (this volume) because of small differ-ences in weighting duplicate analyses. Any differences should be withinanalytical error, which is shown as error bars on all diagrams that usetrace elements. Many of the same samples have been dated by the K-Armethod. These dates have been summarized above and are discussedfurther in Mitchell et al. (this volume).

The isotopic analyses were carried out at Royal Holloway andBedford New College (RHBNC), London, using a VG354 five-collector mass-spectrometer. Multidynamic data collection was usedfor Sr and Nd and static data collection for Pb. Most Nd analyses weredetermined as NdO+ on single Re filaments using the technique ofThirlwall (1991a). Conventional ion exchange methods were used toeffect element separations. The data are presented in Table 2.

87Sr/86Sr ratios were normalized by exponential law to 86Sr/88Sr =0.1194. SRM987 gave a mean value of 0.710244 ±23 (2 s.d., N = 48)over the period of analysis. Because of alteration, all except the freshestand most acidic samples were subjected to heavy leaching in 6N HC1 forseveral hours at 180°C followed by careful rinsing with pure water.143Nd/144Nd ratios were normalized to 146Nd/144Nd = 0.7219. Values of0.511420 ±9 (2 s.d., N = 23) were obtained for a laboratory standard overthe period of analysis, equivalent to a value for the La Jolla standard of0.511857 (Thirlwall, 1991 a,b). Pb isotope ratios were corrected for massfractionation by ca 0.11%/amu by normalization to SRM981. Internalerrors are estimated at better than 0.005%/amu (2 s.e.) and reproducibilityat better than 0.05%/amu (2 s.d.).

Sr isotope measured ratios for the leached samples were treated asequivalent to initial values, following optical examination whichshowed that the residue from leaching was dominated by clino-pyroxene. Unleached samples were corrected using Rb and Sranalyses from Table 1 assuming that alteration did not post-datesignificantly the eruption age. Nd analyses were converted to initialratios using Sm and Nd analyses from Table 1, again assuming thatalteration and eruption ages were similar. Epsilon-values for bothelements, also given in Table 1, are used as the basis for isotopecomparisons. They are based on present-day bulk earth compositionsof 143Nd/144Nd = 0.512637,147Srn/144Nd = 0.1967, ̂ Sr/^Sr=0.7047, and87Rb/86Sr = 0.0847 (Hawkesworth and Norry, 1983).

Lead isotopes have been reported as both measured and initialratios in Table 2. Because of the need to compare lead isotopes with

239

J. A. PEARCE ET AL.

Table 1. Major and trace element analyses of samples chosen for isotopic analysis. For details, see the text."

HoleCoreInterval (cm)UnitMagma type

SiO2

TiO2

A12O3

Fe2O3

MnOMgOCaONa2OK2OP 2 O 5

LOI

786B57R-469-77IV-27LCB

50.280.21

10.907.390.13

12.194.902.880.480.01

10.50

786B62R-340-42IV-27LCB

53.470.20

12.058.470.13

13.045.273.190.670.034.33

786B70R-168-73IV-33

LCBrzA

58.440.23

11.436.150.137.404.953.230.200.044.66

786B51R-151-55IV-26ICBa

53.250.22

12.748.490.16

12.367.782.690.260.242.01

768B11R-1

122-126IV-4ICB

53.640.22

12.897.620.14

10.886.502.730.420.012.80

786B30R-129-31IV-13

ICBrzA

56.030.21

12.206.980.096.529.832.590.870.024.23

786B1R-1

61-64IV-1

ICBrzA

60.110.21

12.397.030.117.806.093.060.510.022.19

786B8R-1

45^t7IV-3Aa

54.370.31

17.487.760.123.987.693.610.650.052.71

786B19R-191-94IV-6

A

58.130.29

16.257.490.123.577.143.130.710.042.34

786B49R-432-37IV-24

A

58.340.31

15.037.000.133.627.293.100.730.042.89

786B21R-126-30IV-6Aa

56.850.36

16.957.700.101.856.563.900.850.312.91

786B35R-173-76IV-20

D

64.800.28

14.725.390.042.365.003.820.690.081.23

786B21R-1

129-131IV-8

R

66.900.28

12.623.490.070.312.303.992.050.076.50

786B66R-188-90IV-29

R

68.950.26

13.414.070.080.822.263.703.120.082.36

Total 99.87 100.85 96.86 100.20 97.85 99.57 99.52 98.73 99.21 98.34 98.41 98.58 99.11

SeCrVNiCoCuZn

RbSrYZrNbBa

LaCePrNdSmEuGdTbDyHoErTmYbLu

HfTaPbThU

30.7111313428141.61861

7.0115

7.4360.70

23

1.673.940.643.050.810.280.940.161.090.240.700.110.720.12

0.960.041.570.270.14

Chondrite normalized

LaCePrNdSmEuGdTbDyHoErTmYbLu

5.394.885.255.084.153.813.633.383.393.343.333.403.443.57

29.01036

16630645.06063

6.4107

7.2350.53

47

1.813.860.592.840.810.270.970.161.050.230.690.110.800.14

0.900.031.500.270.18

5.844.784.844.734.153.673.753.383.263.203.293.403.834.35

27.6670148202

32.07077

1.5144

8.9450.65

50

2.625.520.803.650.970.341.200.211.210.270.770.140.920.15

1.150.054.660.310.26

8.456.836.566.084.974.634.634.433.763.763.674.324.404.66

28.776422019636.54269

4.3156

16.9370.62

19

5.4712.641.325.981.480.431.710.301.840.441.380.251.400.25

1.000.052.000.270.14

17.6515.6410.829.977.595.856.606.335.716.136.577.726.707.76

29.254317115736.34071

5.7165

5.140

0.6219

1.313.370.482.320.610.240.780.140.840.170.560.100.700.11

1.020.032.370.340.13

4.234.173.933.873.133.273.012.952.612.372.673.093.353.42

31.037413212630.92951

20.3193

5.1360.52

44

1.703.270.462.200.630.220.790.140.800.180.490.080.540.09

0.860.041.730.320.23

5.484.053.773.673.232.993.052.952.482.442.332.472.582.70

24.938417711529.16655

5.8166

6.745

0.7036

1.773.900.572.620.790.260.880.150.950.200.640.110.700.12

1.160.051.780.340.18

5.714.834.674.374.053.543.403.162.952.793.053.403.353.73

25.012

2364124.56873

8.5220

13.150

0.770.7

3.026.981.487.152.240.762.290.382.520.531.350.211.300.21

1.370.062.290.370.11

9.748.64

12.1311.9211.4910.348.848.027.837.386.436.486.226.52

20.812

2262219.18372

7.7215

9.249

0.7431

2.555.250.823.901.140.381.380.241.610.290.980.140.990.16

1.290.041.790.460.19

8.236.506.726.505.855.175.335.065.004.044.674.324.744.97

25.0113244

3922.48958

10.7186

9.746

0.7256

2.604.870.823.991.180.401.300.251.700.371.020.171.070.16

1.230.052.300.350.27

8.396.036.726.656.055.375.025.275.285.154.865.255.124.97

21.811

1901711.38072

11.0238

20.7580.90

44

7.4615.771.817.851.820.561.840.372.460.571.900.332.140.33

1.620.062.300.580.40

24.0619.5214.8413.089.337.627.107.817.647.949.05

10.1910.2410.25

20.451

1522613.93448

9.1222

10.162

1.0362

3.026.460.964.641.240.391.590.241.490.321.030.161.040.15

1.670.062.240.550.27

9.748.007.877.736.365.316.145.064.634.464.904.944.984.66

11.44

2467.9

6849

34.2151

11.883

1.0854

3.357.111.085.081.400.371.560.301.900.391.150.201.220.21

2.190.082.800.720.44

10.818.808.858.477.185.036.026.335.905.435.486.175.846.52

11.8333195.6

5092

29.010410.7810.98

388

3.608.101.415.901.390.462.160.261.820.381.210.181.170.19

2.100.083.190.670.37

11.6110.0211.569.837.136.268.345.495.655.295.765.565.605.90

Note: Missing values not analyzed. Magma types: LCB = low-Ca boninite, LCBrzA = low-Ca bronzite andesite, ICB = intermediate-Ca boninite, ICBrzA = intermediate-Ca bronziteandesite, A = andesite, D = dacite, R = rhyolite, HCB = high-Ca boninite, BA = basaltic andesite, ThA = tholeiitic andesite, ThD = tholeiitic dacite.

a Exhibits extreme secondary REE enrichment.

240


Table 1 (continued).

786B63R-272-74IV-29

R

74.290.24

12.411.920.010.000.454.854.980.080.27

99.50

8.21

2211.0

6341

44.04411.785

1.00578

4.839.121.306.201.870.371.770.311.930.481.270.201.240.21

2.410.072.840.650.47

15.5811.2910.6610.339.595.036.836.545.996.696.056.175.936.52

786B21R-272-76IV-9ICB

50.620.23

11.808.140.14

13.075.902.380.390.075.75

98.49

26.1889185223

39.94054

6.8145

7.628

0.5418

1.472.880.462.190.630.230.890.161.090.270.740.120.760.11

0.780.030.790.190.16

4.743.563.773.653.233.133.443.383.393.763.523.703.643.42

786B67R-156-59IV-32ICB

53.000.25

12.467.540.14

13.086.702.760.210.042.56

98.74

33.7786202296

38.55658

3.1136

6.527

0.447.6

1.152.570.371.970.610.220.810.151.000.210.720.120.750.11

0.700.030.720.17

0.12

3.713.183.033.283.132.993.133.163.112.923.433.703.593.42

786B40R-254-57IV-23HCB

47.230.31

12.068.050.14

11.4011.902.120.260.045.46

98.97

27.1762193374

44.62757

7.6139

8.226

0.5315

1.503.050.502.610.800.311.060.201.300.300.840.140.870.13

0.700.020.840.180.14

4.843.774.104.354.104.224.094.224.044.184.004.324.164.04

786A5R-2

70-72IV-2HCB

50.230.37

14.408.550.117.67

10.892.561.250.103.40

99.53

30.340423717040.76363

5.8147

10.633

0.3917

2.224.130.783.801.070.401.410.261.700.371.100.171.140.17

0.890.020.700.180.20

7.165.116.396.335.495.445.445.495.285.155.245.255.455.28

786A17X-CC

4-6

rv-oICB

48.680.20

11.439.230.12

14.386.582.710.700.016.16

100.20

24.86081942541.4

16073

9.593

6.025

0.427.8

0.521.430.211.110.380.140.510.090.690.150.410.080.560.08

0.720.020.680.120.29

1.681.771.721.851.951.901.971.902.142.091.952.472.682.48

786A16X-CC25-28IV-0ICB

51.000.24

14.627.810.13

11.538.063.030.320.012.41

99.16

29.355722619040.23768

3.1143

6.228

0.3927

1.122.600.442.170.610.230.770.141.030.210.640.110.770.12

0.830.021.410.170.12

3.613.223.613.623.13

3.13

2.972.953.202.923.053.403.683.73

786A12X-1

140-142

πBA

51.430.40

17.278.340.095.348.202.821.380.022.92

98.21

31.1320330

8526.24062

14.0171

6.146

0.6618

0.723.880.321.950.560.290.810.140.990.210.580.10.630.09

1.240.044.800.250.22

2.324.802.623.252.873.953.13

2.95

3.072.922.763.093.012.80

786A6X-524-27

IIBA

52.270.82

16.1212.230.204.109.651.590.340.010.84

98.17

41.223

42818

166113

6.118022.040

0.2577

1.244.230.804.541.630.632.310.412.940.631.900.282.030.38

1.140.023.140.060.07

4.005.246.567.578.368.578.928.659.138.779.058.649.71

11.80

782A50X-137-42

πThA

56.100.78

14.9610.920.153.597.853.350.510.071.56

99.84

13.913

2861421.06075

8.3188.4

16.161

0.5121

2.015.830.975.301.720.642.210.342.500.471.530.241.590.27

1.680.051.740.200.20

6.487.227.958.838.828.718.537.177.766.557.297.417.618.39

782A49X-CC

31-34II

ThA

56.000.75

14.6010.180.142.856.873.880.530.052.18

98.03

4598

11590

6.1175

16.557

0.5025

2.976.431.025.251.560.552.030.352.550.531.640.291.810.30

1.660.041.480.170.11

9.587.968.368.75,-8.007.487.847.387.927.387.818.958.669.32

782A50X-132-35

IIThD

66.970.45

14.254.050.100.683.373.951.080.073.49

98.46

12.7

3234.0

11104

17.6160

19.8100

0.7835

4.4710.7

1.929.662.560.743.030.543.440.682.070.312.080.30

2.800.072.760.400.21

14.4213.2415.7416.1013.1310.0711.7011.3910.689.479.869.579.959.32

241

J. A. PEARCE ET AL.

HOLE 786A

5 0 -

100-1 1 X

1H

4X

7X

8X

4X

5X

üLU

AA

LITHOLOGY

, ; . ; j >;;.;..; i i i . .;i

UNITBOUNDARY

ANDROCK TYPE

BA

9X-5 66cm

11X-5 125cm

BA/ICB

13X-CC 4cm

ICB

ittz

- 6X-5, 24-27cm

V O

SAMPLE LOCATION

- 12X-1,140-142cm

- 16X-CC, 25-28cm- 17X-CC, 4-6cm

Figure 2. Stratigraphic section for Site 786 (Arculus et al., this volume) showing locations of samples chosen forisotope analysis.

recent samples, we decided to use present-day values for regionalcomparisons. However, the mobility of U and Pb (see next section)has caused many of the present-day Pb isotope ratios to have evolvedfrom their initial ratios along alteration vectors. The high present-day238U/235U ratio means that 206Pb is more affected by alteration than207Pb. Although alteration-induced variations in 206Pb are small inabsolute terms, they are significant compared with analytical errorsand can distort intra-sample variations. The lead isotope values havetherefore been alteration-corrected in three stages: (1) by extrapola-

tion back to the assumed age of alteration using present-day U/Pb andTh/Pb ratios; (2) further extrapolation back to the age of eruptionalong inferred primary ratios; and (3) projection from the initial ratiosto present-day values using the calculated ratios of their mantlesources. Ages of metamorphism and eruption or intrusion are inferredin part from K-Ar ages discussed in the previous section and thealteration corrections are discussed in the next section. The correctedvalues are given in Table 3 with the subscript "r." Although thismethod makes a number of assumptions, the result is a slightly more

242


HOLE 786B

200-

250-

E

L±JQ

300-

350-

4 0 0 -

UNIT BOUNDARYAND ROCK TYPE

27R-l,0cm27R-l,l00-l40cm

Figure 2 (continued).

243

J. A. PEARCE ET AL.

HOLE 786B (continued)

4 5 0

500-

ε

Q_LuQ

5 5 0 -

6 0 0 -

650-

27R

28 R

45R

46R


A-AA

LITHOLOGY

XoXo×o×oX


27R-|,IOO-l4Ocm-27R-2,0cm

ICBrzA

fm-\ yr-r-rtyr-TTiwn 29R-1,40"70cm

π 30R-2,75cm H C B

30R-3,2lcm

ICBrzA

32R-2,55cm-R

- o —-•o —

• o o

R34R-l,64-85cm34R-2,l5cm

HCB34R-4,IOOcm

D-R

37R-|,0cm

37R-3,3lcmICB

40R-l,65cm

40R-4,30cm'

ICBrzA

44R-|,0cm

ICB44R-2,IOcm-

HCB

ICBrzA

5IR-l,0cm

ICB

13

13

13

16

16

19

20

21

22

23

24

25

24

26

SAMPLELOCATION

12

J4

)5

J 7

1816

35R-l,73-76cm

— 40R-2,54-57cm*

— 49R-4,32-37cm

— 5IR-l ,5 l-55cm

244


HOLE 786B (continued)

700-

E

\- 750-Q_LuQ

800-


ICB

55R-2,60cm-

LCB

Sed.

R

57R-1,104cm

LCB/BrzA

6IR-4,42cm

6IR-6,40cm

LCB/BrzA63R-2,0cm —

R64R-2,128cm___. . -,c Other65R-I,75cm

R

67R-l,8-l0cm67R-l,50-80cm

68R-l,0cm B

LCBrzA

72R-l,0-l2cm jCB

LCBrzA

26

27—57R-4,69-77cm

28

27

29

30

29

33

33

SAMPLELOCATION

— 62R-3,40-42cm

63R-2,72-74cm

— 66R-l ,88-90cm

31'29~32-67R-l ,63-65cm*29

— 70R-l,68-73cm

34

FLOWS BRECCIAS

HCB

ICB

LCB

ICBrzA

LCBrzA

A

D

V V "v?

'ék:•-•A-A

A A a

r f

ΛΛΛ

(vXv×v)

High-Ca boninite

Intermediate-Caboninite

Low-Ca boninite

Intermediate-Cabronzite andesite

Low-Cabronzite andesite

Andesite

Dacite


245

J. A. PEARCE ET AL.

Table 2. Isotopic analyses and initial isotopic ratios of samples from Leg 125. For analytical details, see the text.

Hole, Core,Interval (cm)

786B-57R-4, 69-77786B-62R-3,40-42786B-70R-1,68-73786B-51R-1, 51-55786B-11R-1, 122-126786B-30R-1, 29-31786B-1R-1, 61-64786B-8R-1, 45-*7786B-19R-1, 91-94786B^9R-4, 32-37786B-21R-1, 26-29786B-35R-1, 73-76786B-21R-1,129-131786B-66R-1, 88-90786B-63R-2, 72-74786B-21R-2,72-76786B-67R-1,63-65786B^0R-2, 54-57786B-5R-2, 70-72786B-17X-CC, 4-6786B-16X-CC, 25-28786A-12X-1, 140-142786B-6X-5, 24-27786A-50X-1, 37-42782A-^9X-CC, 31-34782A-45X-1, 32-35

Rocktype

LCBLCB

LCBrzAICBICB

ICBrzAICBrzA

AAAADRRR

ICBICBHCBHCBICBICBBABA

ThAThAThD

Age(Ma)

4141414141414141414141414141413535351735351710353535

"Sr/^Sr

0.703836 ±90.703740 ± 120.703721 ±90.703598 ±90.703588 ±90.703584 ± 100.703542 ±70.703514 + 110.703577 + 110.703692 ±80.703533 ±120.703547 ±120.704698 ±150.704673 ± 120.706079 ± 100.703404 ± 100.703422 ± 140.703331 ±120.703301 ±9

0.703324 ±120.703263 ±110.703556 ±11

0.703471 ±110.703613 ±13

(87Sr/86Sr)o

0.7038360.7037400.7037210.7035980.7035300.7035840.7035420.7035140.7035170.7035950.7035330.7035470.7043180.7042030.7043900.7034040.7033890.7033310.703301

0.7033240.7032630.703556

0.7034710.703613

εSr

-11.56-12.92-13.19-14.94-15.90-15.15-15.75-16.13-16.09-14.98-15.86-15.66

-4.72-6.35-3.70

-17.71-18.01-18.75-19.25

-18.93-19.80-16.13

-16.84-14.83

1 4 3Nd/1 4 4Nd

0.512948 ±40.512973 ±60.512976 ±40.512982 ±40.512987 ±50.512989 ±60.512988 ±140.512987 ±40.512983 ±70.512980 ±50.512978 ±7

0.512988 ±50.512996 ±6

0.513020 ±50.513024+60.513042 ±30.513034 ±5

0.513037 ±40.513040 ± 60.513114 ±4

0.513073 ±50.513056 ±4

(1 4 3Nd/1 4 4Nd)o

0.5129040.5129260.5129320.5129420.5129450.5129420.5129400.5129370.5129360.5129320.512941

0.5129430.512957

0.5129800.5129810.5130140.513015

0.5129980.5130000.513100

0.5130320.513019

εNd

6.236.656.776.957.026.966.916.866.856.766.94

6.987.26

7.557.587.797.85

7.917.949.26

8.568.31

Note that Core 57R-4 is interpreted as having been contaminated by pelagic sediment on eruption and that the rhyolite Sr-isotopes arealteration-affected. For strongly leached samples, the initial Sr ratios are reported as the same as measured ratios. Calculations of initialratios and epsilon values are based on trace element data in Table 1. The key to the rock names is given in Table 1.

coherent data set than that provided by the measured values. Forexample, 206Pb/204Pb values in samples from Cores 16X-CC and17X-CC converge when alteration-induced U-enrichment in the lattersample is taken into account. The corrected lead isotope data inTable 3 have also been quoted as delta values, relative to the NorthernHemisphere Reference Line of Hart (1984).

ALTERATION EFFECTS

Taylor et al. (in press) and Murton et al. (this volume) have demon-strated that a number of elements that are usually considered to beimmobile may be mobilized during alteration of boninitic glass. Giventhe extent of alteration at Sites 782 and 786, it is necessary, beforeconsidering primary isotopic variations, to take into account the effects ofalteration on their isotope ratios and the associated elements. The plots inFigure 4, of Sr, Nd, and Th against Zr and Pb and U against Th, have beenchosen to demonstrate how alteration may have affected the elements ofinterest They are drawn from the full trace element data set of Murton et al.(this volume), not simply the data in Table 1. Zr is used as the horizontal axisbecause it has been shown by the authors quoted above to have been one ofthe least mobile elements, having experienced only dilutions and concentra-tions caused by the transport of other elements into and out of the system. Itthus gives an alteration-independent index of geochemical variation.

Figure 4 A shows the behavior of Sr. The intermediate-Ca boninitesand bronzite andesites from Site 786 plot along a trend of increasingSr with increasing Zr, consistent with fractional crystallization andcumulation of mafic phases. This trend continues into the andesites,dacites, and rhyolites where Sr decreases in concentration as pla-gioclase becomes a crystallizing phase. The andesites and dacitesfrom Site 782 form a similar trend, but displaced to lower Sr valuesfor a given Zr content. The more altered low-Ca boninites and bronziteandesites form a trend which lies subparallel to that of the inter-mediate-Ca series but also displaced to lower Sr values. The fact that

the main axis of dispersion for this group is in the direction of themafic phase crystallization vector suggests that the low Sr is aprimary, rather than alteration-induced, feature. The late dikes fromHole 786B form a cluster displaced to higher Sr for a given Zr content.

Although the scatter exceeds analytical error, many of the rocksare highly porphyritic (some plagioclase-phyric), which will enhanceprimary scatter. Thus, although Sr is theoretically a highly mobileelement, its mobility has only masked primary variations in a veryfew of the analyzed samples. This conclusion concurs with that ofTaylor et al. (in press), working on boninitic samples cored duringLeg 126. They argued that, because Sr is mostly located withinPlagioclase, it may be less mobile than many of the normally moreimmobile incompatible elements that are located within the muchmore reactive boninitic glass.

Of the Sr isotope data in Table 2, all analyses of basic andintermediate samples were made on leached samples and show nocorrelation with alteration indices such as LOI (loss on ignition).However, the three unleached rhyolite samples all have significantlyhigher initial ratios. All three are strongly altered, exhibiting sig-nificant enrichment in Rb. Moreover, their radiogenic Sr isotoperatios are not matched by the variations in other immobile element orisotope ratios that would be expected if these rocks originated by adifferent petrogenetic process. We therefore infer that these haveundergone extensive interaction with seawater-derived fluids, and thedata from these rocks are not plotted on the isotope diagrams thatfollow. The Sr isotope ratio of the sample from Core 57R-4 is alsoconsidered to be higher than the primary value, even though thissample has been leached. In this case the higher Sr isotope ratio ismatched by a lower Nd isotope ratio and higher ratios of Pb isotopes(particularly 207Pb/204Pb) compared with the otherwise similar low-Caboninite from Core 62R-3. These characteristics indicate that thesample contains a pelagic sediment component. Because this com-ponent was not found in any of the surrounding rocks analyzed, we

246


Table 2 (continued).

206 p b / 204 p b

18.75518.90418.90118.86218.83518.90018.81618.808

18.87318.862

207 p b / 204 p b

15.56615.53515.52815.51615.52415.51915.51515.523

15.52015.526

208 p b / 204 p b

38.73738.45938.44538.39938.39738.40038.36138.403

38.40938.434

( 206 p b / 204 p b ) o

18.71918.85518.87818.83418.81318.84618.77518.788

18.82518.791

( 2 0 7 p b / 2 0 4 p b ) o

15.56415.53315.52715.51515.52315.51615.51315.522

15.51815.523

( 208 p b / 204 p b ) o

38.71438.43538.43638.38138.37838.37538.33538.381

38.38938.400

18.848 15.505 38.365 18.784 15.502 38.311

18.82618.80418.78218.74418.87618.82618.58418.38118.62018.66718.629

15.51315.52915.52515.50915.51415.51715.54015.48515.46215.49715.481

38.30838.38138.34838.28138.29538.31838.41238.17038.10338.26638.174

18.74418.74618.72418.69618.78918.79718.56818.37918.58018.64118.603

15.50915.52615.52215.50715.51015.51515.53915.48515.46015.49615.480

38.27638.36838.32438.26738.27838.30138.40638.16938.09038.25338.158

Table 3. Recalculated (alteration corrected) lead isotope ratios (subscript r) and delta values for Leg 125 samples.3

Hole, Core,Interval (cm)

786B-57R-4, 69-77786B-62R-3, 40-42786B-70R-1, 68-73786B-51R-1, 51-55786B-11R-1, 122-126786B-30R-1, 29-31786B-1R-1, 61-64786B-8R-1,45-47786B-19R-1, 91-94786B^9R-4, 32-37786B-21R-1, 26-29786B-35R-1, 73-76786B-21R-1, 129-131786B-66R-1, 88-90786B-63R-2, 72-74786B-21R-2, 72-76786B-67R-1, 63-65786B^0R-2, 54-57786B-5R-2,70-72786B-17X-CC, 4-6786B-16X-CC, 25-28786A-12X-1, 140-142786B-6X-5, 24-27786A-50X-1, 37-42782A-49X-CC, 31-34782A-45X-1,32-35

Rocktype

LCBLCB

LCBrzAICBICB

ICBrzAICBrzA

AAAADRRR

ICBICB

HCBHCBICBICBBABA

ThAThAThD

Age(Ma)

414141

41(35)41(35)

4141

41(35)4141

41(35)41

41(35)414135353517

35(17)35(17)

17103535

35(17)

206 p b / 204 p b

18.75518.90418.90118.86218.83518.90018.81618.808

18.87318.862

18.848

18.82618.80418.78218.74418.87618.82618.58418.38118.62018.66718.629

207 p b / 204 p b

15.56615.53515.52815.51615.52415.51915.51515.523

15.52015.526

15.505

15.51315.52915.52515.50915.51415.51715.54015.48515.46215.49715.481

208 p b /204 p b

38.73738.45938.44538.39938.39738.40038.36138.403

38.40938.434

38.365

38.30838.38138.34838.28138.29538.31838.41238.17038.10338.26638.174

( 206p b / 204 p b ) r

18.74518.88118.90418.85918.83718.87218.80118.813

18.85118.823

18.815

18.80618.79618.77418.72018.82218.82918.61818.37918.60518.66618.624

( 2 0 7 p b / 2 0 4 p b ) r

15.56615.53415.52815.51615.52415.51815.51415.523

15.51915.524

15.503

15.51215.52915.52515.508 .15.51215.51715.54215.48615.46115.49715.481

(208Pb/204Pb)r

38.73138.45238.45338.39838.39538.39238.35338.399

38.40638.420

38.350

38.30238.37638.34538.27738.29338.31938.42838.17638.10138.26438.166

Δ7/4

4.3-0.4-1.2-2.0-0.9-1.9-1.5-0.7

-1.6-0.7

-2.7

-1.80.0

-0.2-1.3-2.0-1.5

3.20.1

-4.7-1.8-2.9

Δ8/4

44.2-0.2-2.9-3.1-0.7-5.1-0.5

2.7

-1.23.5

-2.5

-6.12.42.11.6

-9.5-7.828.931.2-2.0

6.92.1

1 The assumed metamorphic age is given in parentheses if it differs from the magmatic age. The key to the rock names is given in Table 1.

suspect that the "brown clay" identified in vesicles and microfracturesis pelagic sediment incorporated during eruption on the seafloor. Thissample is therefore omitted from further discussion.

Figure 4B shows the behavior of Nd relative to Zr. This diagramshows a main axis of dispersion for all groups toward increasing Ndand Zr contents, consistent with major phase crystallization. How-ever, some samples are displaced to high Nd contents beyond thelimits of experimental error. This is most apparent in the chondrite-normalized rare earth element (REE) data in Table 1. Samples fromCores 51R-1, 8R-1, and 21R-1 (26-30 cm) exhibit strong LREEenrichment compared with other samples from their magmatic

groups. This enrichment is matched by enrichment in P but not by thevariations in Th, Zr, or Nb that might be expected if it were a primaryfeature. REE mobility was also recognized in Leg 126 cores by Tayloret al. (in press) and attributed to alteration of the boninitic glass. Someof the late dikes form a trend parallel to the principal axis of dispersionbut displaced to higher Nd for a given Zr content. This may be aprimary feature. By contrast, the sample from Core 12X-1 in Hole786A has low REE contents and a positive Ce anomaly, indicative ofloss of REE by leaching in an oxidizing environment.

Despite this evidence for extensive REE mobility, however, theε Nd values of the LREE-enriched samples from the LCB and ADR

247

J. A. PEARCE ET AL.

20 -,

OLGBrzA

10 -

• 786B - LCB

A 786B - ICB

* 786B - ADR° 786B - dikes

786A - ICB786A - BA782-ThAD

Oo

10 -

5 -

0

s •

o

f>

ICB

^ N H C E

LCB

ICBrzA

LCBrzA

D

i

50 60 70 80

wt% SiO,

Figure 3. Classification of samples chosen for isotopic analysis using the projec-tions of Arculus et al. (this volume). Fields on the diagram are: LCB = low-Caboninite; LCBrzA = low-Ca bronzite andesite; ICB = intermediate-Ca boninite;ICBrzA = intermediate-Ca bronzite andesite; HCB = high-Ca boninite; A =andesite; D = dacite; R = rhyolite. For details on the key, see the text.

groups do not differ significantly from those of the unenrichedsamples (6.95, 6.86, and 6.94 for the enriched compared with 7.02,6.96,6.91,6.95, and 6.76 for the unenriched). Any effect of alterationon the ε Nd values thus appears to have been small compared with theprimary magmatic variations. This must indicate either that Nd-enichment is a very localized phenomenon (e.g., from the rim to coreof a clast) or that the fluids involved in the alteration circulatedthrough rock of very similar composition. The inference in either caseis that the εNd values can be treated as close to primary values, evenin the LREE-enriched samples.

The Th-Zr diagram in Figure 4C shows a significant positivecorrelation for all samples from Site 786. Samples from Site 782 havea similar correlation but are displaced to lower Th values for a givenZr content. The magnitude of the scatter is consistent with analyticalerror and pyroxene cumulation, suggesting that Th as well as Zr hasbehaved as an immobile element. This enables the chemically moreappropriate element, Th, to be used instead of Zr for assessing themobility of Pb and U.

The Pb-Th plot is shown in Figure 4D. The large scatter, well in excessof analytical error, suggests that Pb has been mobilized during alteration.Because Pb is less incompatible than Th, the Hole 786B samples should

follow a positive, shallow trend on this diagram. The equation for thistrend cannot be defined precisely, but is taken from the diagram as Pb= 2 × Th + 1. The late dikes and boninites from Hole 786A and the Site782 tholeiites have been assumed to follow primary trends of Pb = 4 × Thand Pb = 8 × Th respectively. These equations were used in the recalcula-tion of Pb isotope ratios in Table 3. The most noticeable changes in Pbconcentration appear to have taken place near the base of Hole 786B. Thisis particularly true of the low-Ca bronzite-andesite dikes, where significantenrichment has taken place (compare for example the high Pb concentra-tion in the sample from Core 70R-1 with the much lower Pb concentrationin the sample from Core 62R-3 in Table 1). The two basaltic andesites fromHole 786A have the highest Pb/Th ratios. Because they are clasts withinpelagic and volcanogenic sediments that have experienced alteration, it ismost unlikely that their Pb contents are primary. The ratios chosen for thetholeiitic andesites are assumed also to apply to these rocks.

As far as isotope ratios are concerned, there is again little obviousdifference in lead isotope ratios between more and less alteredsamples. For example, the sample from Core 70R-1 (Pb-enriched) iswithin error of that from Core 62R-3. The samples from Cores 8R-1,21R-1, and 51R-1, all of which were enriched in LREEs, have similarPb-isotope ratios to those showing no enrichment. Again, therefore,the Pb isotope ratios are taken as primary, with the exception of thesample from Core 57R-4 as discussed above.

Figure 4E shows the U-Th covariation. A significant number ofsamples have experienced enrichment or depletion in excess ofanalytical error, indicating that U has been highly mobile duringalteration—a predictable observation in view of the known mobilityof U during oxidative alteration. It is, however, possible to define anapproximate primary trend of increasing Th and U. The primary Th/Uratios are taken as 2.0 for the 41 Ma samples from the boninite edifice,and 1.5 for other samples. These ratios were used in the recalculationof Pb isotopes in Table 3.

ISOTOPE SYSTEMATICS

Nd-Sr Isotope Covariations

Figure 5 A shows the distribution of data points from Sites 782 and786 on an ε Nd-ε Sr covariation diagram. For comparison, we have alsoplotted a number of reference points and fields on this and subsequentfigures. The range of oceanic mantle compositions is represented by thefour "end-members" postulated by Zindler and Hart (1986), namelydepleted MORB mantle, high U/Pb mantle (HIMU), and two enrichedmantle components (EM and EMU). The enriched components arerepresented as arrows rather than points because they are vectors ratherthan precise compositions. The depleted MORB mantle is represented byaverage Pacific MORB Mantle (PMM) and Indian MORB Mantle(IMM), and a field for Pacific MORB Mantle has also been drawn usingthe MORB data of White et al. (1987). Subducted sediments are repre-sented by a field of Pacific Pelagic Sediment (PPS) and an average PacificVolcanogenic Sediment (PVS) from the data of Woodhead (1989) andBen Othman et al. (1989). Regional plume compositions are representedby the two closest large regional hotspots of Kerguelen (Dosso andMurthy, 1980) and Hawaii (Staudigel et al., 1984). Note that both hot-spots are included because the Western Pacific Eocene terrane fallsbetween the Indian and Pacific isotopic provinces and could, in theory,have been influenced by the isotopic composition of either province.Data from the Caroline Islands, the nearest seamount chain to theBonin-Mariana system, lie within the Hawaii field (Hart, 1988) andtherefore have not been plotted separately. Bulk Silicate Earth (BSE)has also been plotted. The fields drawn approximate to 90% prob-ability contours and outliers (such as basalts from ridge segmentsadjacent to hotspots in the PMM group) have been omitted.

This diagram shows that, on this scale, the Leg 125 lavas and dikesform a well-defined cluster with a main axis of dispersion subparallel toa line through PMM and PVS (equivalent to the "mantle array"), al-though displaced to higher ε Sr values. The ranges of ε Sr and ε Nd

248


300 π

200 -

Q.Q.

100 -

EQ.Q.

10 -|

8 -

6 -

2 -

0

(a)

1 | • | I | 1 J I | ¥ |

0 20 40 60 80 100 120Zr ppm

error

°8 (b)

20 40 60 80 100 120Zr ppm

6 -

5 -

4 -

EQ. 3 -na.

2 -

1 -

0

error

+

0.0

0.0

" o0° (d)

0.2 0.4Th ppm

0.6 0.8

w.u -

0.5 -

0.4 -

0.3 -

0.2 -

0.1 -

0.0 -

error+

o *

o

•

i

C

$ oB oo

•

. A. %&× & jit, . . .

• <!& ®

1 •

%

(e)

0.2 0.4Th ppm

0.6 0.8

0.6 -

0.4 -

0.2 -

0.0 -

error

+

j?* *

V ,o°

OSftβOo A

ü

• i • i •

1 "

(c)

1 • 1 • 1

Key to symbols

0 20 40 60 80 100 120Zr ppm

Figure 4. Trace element covariation diagrams to highlight the effects ofalteration on the elements related to the Sr, Nd, and Pb isotope systems.

values are broadly comparable to those of Hawaiian lavas, althoughthey have higher ε Sr values and higher εNd values than an averageHawaiian basalt.

The distribution of Leg 125 data is shown in more detail on theexpanded diagram in Figure 5B. It is apparent that the data form severaldistinct groupings. The LCB group has the most radiogenic Sr isotoperatios. The two samples that make up this group, one a boninite, the other

786B - Low-Ca boninite group

786B - Intermediate-Ca boninite group

786B - Andesite-dacite-rhyolite group

786B - Late boninite dikes

786A - Intermediate-Ca boninites

786A - Basaltic andesites

782 - Tholeiitic andesites and dacites


a bronzite andesite, are identical within analytical error. The 786B-ICB and -ADR groups form a single cluster, displaced to lower εSrvalues and slightly higher ε Nd values than the LCB group. The ICB groupdoes have consistently higher εNd values than the ADR group but thedifference is close to the limits of experimental error and cannot beconsidered significant. The rock types that make up these two groups(intermediate-Ca boninites, intermediate-Ca bronzite andesites, andesites,and dacites) can therefore be considered co-genetic on this projection.Note that the ihyolites have εNd values of 6.98 and 7.26, slightly higherthan the andesites and dacites, but have not been plotted on this diagrambecause of the unreliability of the Sr isotope ratios. These values are not,

249

J. A. PEARCE ET AL.

15

10 -

5 -

0

-5 -|

-10

o Leg 125

BSE

EMI

EMM

Kerguelen

-40 -20 0 20 40ESr

60 80

10 π

9 -

1 8

7 -

errorj.T

- % * 4

i •

• 786B-LCB

A 786B-ICB

» 786B-ADR

° 786B-dikes786A-ICB

786A-BA

782-ThAD

-22 -20 -18 -16 -14 -12 -10 -8

Figure 5. εNd-ε Sr diagrams for Leg 125 samples. The upper figure (A)) showsthe distribution of data points relative to the Hawaii and Kerguelen hotspots,Pacific MORB mantle (PMM), Bulk Silicate Earth (BSE), Pacific Pelagic Sedi-ment (PPS), Pacific Volcanogenic Sediment (PVS), and the high-µ(HMU), EMIand EMU mantle end-members. The lower figure (B) is an expansion of part ofthis diagram to highlight variations within the Leg 125 samples.

however, sufficiently different to suggest that the rhyolites had an originother than by fractional crystallization from the andesites and dacites.

The boninitic late dikes from Hole 786B and lavas from Hole 786A(including the basaltic andesite from Core 12X-1) form a distinctgroup subparallel to a PMM-PVS line. Of the dikes, the two inter-mediate-Ca boninites are those with the lowest εNd values. Thehigh-Ca boninites and the lavas from Hole 786A have slightly higherεNd values. It is thus possible that some of the late dikes were feedersto the boninites at the top of the edifice. If the interpretation of theK-Ar ages is correct, this process may have taken place some 6 Maafter the formation of the main edifice. The trend followed by thesesamples can be extrapolated to the LCB group. The orientation of thetrend is typical of that created by simple heterogeneities within themantle array. The tholeiitic lavas from Site 782 and the basalticandesite clast from Core 6X-5 in Hole 786A have the highest εNdvalues and a significantly greater εSr shift from the PMM-PVS line.These characteristics resemble those from many oceanic island arcs,where they are generally interpreted in terms of enrichment of a depletedmantle source by seawater-derived Sr (e.g., Hawkesworth, 1982).

Lead Isotope Covariations

Figures 6A and 6D show the variations in 207Pb/204Pb and 208Pb/204Pbisotope ratios relative to the 206Pb/204Pb ratio in the context of the isotopecomponents and reference compositions plotted in Figure 5A. Thefigures also show the Northern Hemisphere Reference Line (NHRL)of Hart (1984), which is the best-fit line for the Pb isotope compositionsfor MORBs and a group of oceanic islands (Hawaii, Iceland, Azores,Canaries, Cape Verde, Galapagos, NE Pacific seamounts) in the NorthernHemisphere. Because of the problem, discussed earlier, of Pb and Umobility, these diagrams have been plotted both as initial ratios and asrecalculated (alteration-corrected present-day) values. Note that there areslight differences of detail in the two sets of diagrams, but that the overallpatterns are similar. It is apparent on both projections the bulk of theLeg 125 samples plot on or just below the NHRL but are displaced towardslightly more radiogenic compositions than PMM.

The Leg 125 samples, annotated according to lithostratigraphic group,are plotted on expanded scales on Figures 6B, C, E, and F. These plotsshow that the various magmatic groups are systematically displaced alongthe NHRL. The tholeiitic lavas from Site 782 have the least radiogeniccompositions, followed by the late dikes and 786A lavas, then theintermediate-Ca boninites and bronzite andesites and andesite-dacitelavas, and finally the low-Ca boninites and bronzite andesites which havethe most radiogenic compositions. These diagrams thus confirm thegroupings made on the basis of Figure 5B. The consistency of the data,in particular the similarities in isotope ratios between variably-alteredsamples from the same lithostratigraphic group, suggest that alterationeffects are small compared with petrogenetic variations, although theymay contribute to the scatter. Apart from the tholeiites from Site 782, thePb isotope variations are not far in excess of that expected from the 2 s.d.reproducibility of the SRM981 analysis as denoted by the error bars inFigure 6. However, these standard data are a compilation of valuesproduced by several analysts. The constant relationships between lithol-ogy and Pb isotope composition suggest that the analyses reported hereare considerably more reproducible.

This dispersion along the NHRL has several possible explana-tions: that it is accidental; that it represents an isochron; or that itrepresents two-component mixing. The slope on the 207Pb/204Pb-206Pb/204Pb diagram corresponds to an apparent age of 1.65 Ga, too greatan age for the Pb isotope variation to have developed as an isochron withinoceanic lithosphere. The fact that the various magmatic groups plot in aconsistent order on both these and the εNd-ε Sr diagram indicates thatthe trend is unlikely to be accidental and is most likely to be explainedby mixing of a less and a more radiogenic component. Since mixingis linear on Figure 6, two likely components are PMM and a componentalong the NHRL, possibly close to PVS. The latter does not imply thatPacific volcanogenic sediment was necessarily itself involved. Thebulk of Pacific volcanogenic sediments are of Cretaceous age andderived from oceanic, intraplate volcanoes. The PVS composition cantherefore also be considered a Pacific hotspot composition. Note thatthe data are consistent with the involvement of Pacific, rather thanIndian or South Pacific, mantle. The latter plot above the NHRL on the208pb/204pb. 206pb/204pb d i a g r a m

The lead isotope plots also demonstrate the absence of a signifi-cant pelagic sediment component in most of the Leg 125 samples asthis would cause a displacement to high 208pb/204pb a n d 207pb/204pb

ratios at a given 206Pb/204Pb ratio. This is true of the tholeiitic lavasfrom Site 782 as well as the boninites from Site 786. It is also acommon feature of Eocene lavas from the Mariana forearc (Hickey-Vargas, 1989; Stern et al., 1991). It is an enigma that these boninites,apparently erupted above a subduction zone, do not exhibit thiscommon characteristic of subduction-related lavas. Pacific vol-canogenic sediments do, however, lie along the NHRL and could bethe source of a subduction component. The two notable exceptionsfrom this study are the samples from Cores 12X-1 and 6X-5, which

250


19 20Pb206/204

B15.56 -,

A 15.52 -

cBCM

15.48 -

15.44

18.2 18.4 18.6 18.8 19.0

Pb206/204)o

15.56 π

15.52 -

CM

15.48 -

15.44

18.3 18.5 18.7

Pb206/204

18.9 19.1

Figure 6. Diagrams showing Pb isotope variations in Leg 125 samples. A, B,and C show 2 0 7Pb/2 0 4Pb vs. 2 0 6Pb/2 0 4Pb covariations. A shows the distributionof data points on the scale of global variations. B and C are an expansion ofpart of this diagram to highlight variations within the Leg 125 samples. D, E,and F show 2 0 8Pb/2 0 4Pb vs. 2 0 6Pb/2 0 4Pb covariations. D shows the distributionof data points on the scale of global variations. E and F are an expansion ofpart of this diagram to highlight variations within the Leg 125 samples. Fieldsand terms are identified in the caption to Figure 5 except NHRL (NorthernHemisphere Reference Line). The ratios with subscript "o" are initial ratios,those with subscript "r" are present-day ratios recalculated to take alterationinto account.

S

SJQ.

40 -

39 -

38 -

37 -

Kerguelen,

E M I • "\

HawaiMp''

/

EMM

/<m •

ùf -PPS

PVS

o

V

Leg

<HIMU

125

17 18 19 20Pb206/204

21 22

38.5 •η

38.4 -

S 38.3 •CMCO

§ 38.2 -Q.

38.1

38.0

o ::

18.3 18.5 18.7

Pb206/204)o

18.9

38.5 -,

38.4 -

cf 38.3 •

38.2 -

38.1

38.0

error

18.3


18.5 18.7 18.9

Pb206/204

19.1

are displaced toward PPS. These are the youngest samples, and thisobservation could therefore be taken to indicate a recent input ofsubducted pelagic lead into the mantle source. However, both areclasts within sediment containing a pelagic component and haveexperienced extensive alteration and probable Pb enrichment. Al-though lavas and dikes within the edifice may also have experiencedPb enrichment, the source of the lead in these samples would havebeen surrounding lavas of similar isotopic composition. By contrast,the lead in the two anomalous samples most probably would have

251

J. A. PEARCE ET AL.

ΦT3

DU

40 -

20 "

o-

-20 -

-40 -

-60 "

HIMU-*- »- DUPAL

° Leg 1251 "•" 1 r

-120 -80 -40 0 40

Pb208/20480 120

Figure 7. Pb isotopic composition of Eocene-Oligocene Leg 125 samplesplotted as a function of latitude, assuming values of 5° N for Eocene (41 Ma)and 10° N for Oligocene (35 Ma) groups. The shaded area is that of suboceanicmantle unrelated to subduction as defined by Hart (1988).

been derived from the pore waters of the enclosing pelagic sediment.The isotope variations themselves do not, however, enable subductedlead to be distinguished from the effects of alteration.

Apart from the two samples with the pelagic sediment lead signa-ture, the samples have Δ7/4 and £&IA values (a measure of thedeviation from the NHRL) of abour -5 to +3 and -10 to +7 respec-tively (see Table 3). Reconstruction of the global setting of theIzu-Bonin forearc from Eocene to Recent (see references in Pearce etal., this volume) shows that the Leg 125 sites would have lain at theequator or slightly further north in the Eocene. The isotope data thuslie well within the range of delta values for ocean island basalts ofthis latitude, as compiled by Hart (1984) (Fig. 7).

The plot of W ^ P b against W ^ P b has been shown separate-ly in Figure 8. On the scale of Figure 8 A, they form a trend at an acuteangle to the NHRL toward an enriched composition in the region ofPPS. On this projection, the samples containing an apparent sedimentcomponent do not separate because the sediment component and the otherenriched component are co-linear.

On the detailed scale of Figure 8B, the Leg 125 samples can beseen to form two distinct subparallel trends. The lower trend, dis-placed to a higher 208Pb/204Pb ratio, is followed by the 41 Ma lavasand dikes from the main boninite edifice in Hole 786B. The othertrend is followed by the late dikes from Hole 786B, lavas fromHole 786A, and forearc tholeiites from Hole 782. As in Figure 6, thetrends on this diagram can be viewed as accidental, time-related, orthe product of component mixing. If the two distinct trends areaccidental, the likely cause is fractionation in the mass spectrometer.The error of 0.05%/amu is not large in absolute terms but becomessignificant because of the small variations in 207Pb/204Pb ratios be-tween the magma types. The error bar is oriented in the directions ofthe two parallel trends, indicating that variations within each of thetwo groups may be explained in part by analytical error, but that thevariations between the two groups are likely to be real. Note that thevariations are unlikely to be caused by systematic errors because thetwo groups of samples were analyzed together. If the differencebetween the two trends is to be interpreted as time-related, this cannotbe caused by variations in Th/Pb within a single source because theolder (41 Ma) samples have lower Th/Pb ratios yet higher 208Pb/204Pbratios. In order to obtain the observed trends, it is necessary topostulate that the source of the edifice lavas did have a lower Th/Pbratio, but that this has been masked by recent mixing and partialmelting events. Mixing is also, therefore, a necessary part of thishypothesis. Differences in U/Pb ratios could generate the difference

Leg125

38 39 40Pb208/204

41

15.56 π

15.52 -\

15.48 -I

15.44

38.0 38.1 38.2 38.3

Pb208/204

38.4 38.5

Figure 8. Diagrams showing 2 0 8Pb/2 0 4Pb vs. 2 0 7Pb/2 0 4Pb covariations. A showsthe distribution of data points on the scale of global variations. B is anexpansion of part of this diagram to highlight variations within the Leg 125samples. Fields and terms are identified in the captions to Figures 5 and 6.

in 207Pb/204Pb ratios in about 120 Ma (for µ= 16.5), but this shouldthen also be apparent in projections using the 206Pb/204Pb ratio. Inorder to explain the two trends by mixing, one component would needto have a composition of about 38.1,15.46, and the other would haveto comprise a mixture of PPS and PVS. This is not, however, consis-tent with the plot of 208Pb/204Pb against 206Pb/204Pb, which shows nocomponent in the direction of PPS.

The preferred explanation is therefore that the subparallel trends arelargely the result of fractionation in the mass spectrometer, but that thepresence of two trends does imply two boninite groups (one containing41 Ma edifice rocks from Hole 786B, the other containing late dikes fromHole 786B and lavas from Hole 786A) as seen on the εNd-ε Sr diagramin Figure 5. Without the effects of fractionation, these would have plottedalong the NHRL with the 41 Ma edifice lavas being the more radiogenic.The plot would then have resembled that of ^Pb/^Pb against206Pb/^04Pb. By the same argument, fractionation has increased the scatterbelow the NHRL on the diagram of ^Pb/^Pb against 206Pb/2CWPb.Overall, therefore, the Leg 125 samples are remarkable for their coin-cidence with the NHRL in all projections.

Sr-Pb and Nd-Pb Isotope Covariations

The relationships between εSr, εNd, and the ^Pb/^Pb ratio areshown in Figure 9. The ε Nd- 206Pb/204Pb diagram shows the data plottingin a consistent order between PMM and PVS, indicating that almost allthe data plot on a single two-component mixing trend between PMM andPVS in ε Nd-Pb isotope space. The two basaltic andesite clasts from

252


10 -

5 -

0 -

-5 -

10 -

IMM ×m

Hawaii f

Kerguelen

( j\|§§/

EM If

0MM

as PVS

TEMII

I I I PPS1 1 1

Leg125

HIMU

-i 1 1

17

10 -,

9 -

18

PMM

19 20Pb206/204

21 22

7 -error ×>,

PPS

118.3 18.4 18.5 18.6 18.7 18.8 18.9 19.0

Pb206/204

80

60 -

40 -

COvu

PPS, Leg 125

EMIli

Kerguelen

PVS

HIMU

17 19 20Pb206/204

21 22

"1

-14 -

-16 -

-18 -

-20 -

-22 -

error

iPPS

° oo

o

4

<

18.3 18.4 18.5 18.6 18.7 18.8 18.9 19.0

Pb206/204

Figure 9. Diagrams showing ε Sr and εNd against 2 0 6Pb/2 0 4Pb. A and B showεNd vs 2 0 6Pb/2 0 4Pb covariations. A shows the distribution of data points onthe scale of global variations. B is an expansion of part of this diagram tohighlight variations within the Leg 125 samples. The right-hand diagrams showε Sr vs. 2 0 6Pb/2 0 4Pb covariations. C shows the distribution of data points on thescale of global variations. D is an expansion of part of this diagram to highlightvariations within the Leg 125 samples. Fields and terms are identified in thecaptions to Figures 5 and 6.

Hole 786A plot slightly off this trend because of alteration or a subductedpelagic sediment component, as discussed earlier. Because εNd behavescoherently with the lead isotopes, the ε Sr-206Pb/2O4Pb diagram resemblesthe ε Nd-ε Sr diagram of Figure 5. Most samples form a trend parallel to aline joining PMM and PVS, but displaced to higher εSr values. Thetholeiites and the younger basaltic andesite plot off the trend to still higherε Sr values. These tholeiites did, however, plot on the two-component Ndand Pb isotope mixing line. Thus, although Nd and Pb isotopes showcoherent behavior, it is possible that Sr isotopes provide evidence for anadditional, possibly seawater-derived, component in the mantle source.

We can therefore conclude from the isotope diagrams that mostof the Eocene-Oligocene boninites plot on simple two-componentmixing trends between Pacific MORB mantle (PMM) and PacificVolcanogenic Sediment (PVS). The tholeiites from Site 782 plotclosest to PMM, the boninitic edifice in Hole 786B plots closestto PVS, and the late dikes in Hole 786B together with the lavasfrom Site 786Aoccupy an intermediate position. A shift to higher ε Srvalues may indicate the involvement of an additional, possibly seawater-derived, component. The shift of the two basaltic andesite clasts towardpelagic sediment lead compositions may indicate a further component ormay be the result of alteration by sedimentary pore fluids. These infer-


ences, and the relationships between the boninites and the tholeiites,are now investigated further on isotope-trace element diagrams.

ISOTOPE-TRACE ELEMENT COVARIATIONS

The previous section has shown that the Nd and Pb isotopes behavecoherently in the Leg 125 samples and therefore that only one isotope isneeded to represent the variation from the depleted to the enrichedcomponent, ε Nd has been selected because of its higher variability relativeto analytical precision and is plotted against a series of trace element ratiosin Figure 10. The elements are plotted as ratios to reduce the effects ofpartial melting, fractional crystallization, and alteration. Ideally, Nd wouldbe used as the denominator because this would create linear mixing lineson the diagrams. However, the proven mobility of the REEs precludes thisoption, as Figure 4 demonstrated. Zr is therefore used instead as thedenominator, having been shown to have been immobile and to haveexhibited a petrogenetic behavior closer to that of Nd than the otherimmobile elements (Ti, Th, and Nb). The data plotted have been screenedto exclude ratios on samples known to have been strongly affected byalteration (for Pb, Y, Sr, and REEs) or fractionation (for Sr and Ti).

The five diagrams in Figure 10, based on the Ti/Zr, Y/Zr, Nb/Zr,Pb/Zr, and Th/Zr ratios, confirm the evidence from isotope covaria-tion diagrams that the magmatic groups from Leg 125 can beexplained in terms of simple two-component mixing, at least to a firstapproximation. For this, they satisfy the criteria of (1) formingcurvilinear trends within the limits of experimental error, partialmelting and fractional crystallization, and (2) following a consistentorder regardless of the isotope or trace element ratio used. The

253

J. A. PEARCE ET AL.

100 -

80 -

S 60 -

40 -

20 -

0

0.6 -,

0.5

0.4 .

0.3

0.2 -

0.1 •

0.0

0.04 ,

0.03 J

0.02 -

0.01 -

PMM

(a)

10

PMM

(b)

5 6 7 86Nd

9 10

(c)

4 5 6 7 8 9 10

0.00

Figure 10. Plots of (A) Ti/Zr, (B) Y/Zr, (C) Nb/Zr, (D) Pb/Zr, and (E) Th/Zragainst εNd. PMM = Pacific MORB mantle. Fields shown are approximatepossible ranges for inferred components Aand B. Curved lines are mixing linesbetween these components assuming Nd/Zr = 0.12 in component A and 0.06in component B.

isotopically-depleted and -enriched end-members have been desig-nated "A" and "B" respectively. The curvature of the mixing linesdepends on the Nd/Zr ratios of these two end members. On the basisof the εNd-Nd/Zr diagram in Figure 11, the mixing curves are basedon ratios of 0.12 in component A and 0.06 in component B.

Component A is characterized by high ε Nd. There are no con-straints on the precise isotopic composition of this component, so aPMM value of+10 is assumed. The trace element ratios of componentA are then read off the mixing curves at ε Nd = 10. However, with thepossible exception of Ti/Zr, none of the trace element ratios cor-respond to the PMM composition, even if the εNd value of com-ponent A is taken from the lower (ca 8) or higher (ca 12) extremes of

(d)

PMM

0.016 -

0.012 -

0.008 -

0.004 -

0.000

PMM

Key to symbols

• 786B - Low-Ca boninite group

* 786B - Intermediate-Ca boninite group

ss 786B - Andesite-dacite-rhyolite group

° 786B - Late boninite dikes

786A - Intermediate-Ca boninites

786A - Basaltic andesites

782 - Tholeiitic andesites and dacites


the PMM range. This observation suggests that component A has theisotopic composition of PMM, but a different trace element composi-tion. Elements more compatible than Zr (Y and to a lesser extentTi—Pearce and Norry, 1979) are enriched in component A relative to Zr,whereas elements that are more incompatible (Nb, Th, and Pb) are depletedrelative to Zr. These characteristics suggest that component A representssuboceanic lithosphere which has lost a melt fraction during its previoushistory of upwelling at a mid-ocean ridge. An origin in another setting ofoceanic volcarasm cannot be ruled out, although the low contents ofincompatible elements probably rule out an origin at an oceanic island,while the low contents of Th and Pb do not suggest an origin in an islandarc. Such an interpretation is consistent with most models for boninitegenesis, which require a depleted, lithospheric rather than fertile, asthenos-pheric mantle source (see papers in Crawford, 1989). Note that, if com-ponent A is suboceanic lithosphere, it would be expected to vary incomposition with depth, becoming more depleted toward the Moho. This

254


q>o

0.14 π

0.12 -

0.10 -

0.08 -

0.06

PMM

10

0.14 η

0.12 -

0.10 -

§ 0.08 -

0.06 -

0.04 -

0.02

10

0.05 -|

0.04 -

N 0 0 3 •

ε0.02 -

0.01 -

0.00

error

PMM

10

Figure 11. Plots of (A) Ce/Zr, (B) Nd/Zr, (C) Sm/Zr, and (D) Yb/Zr againstε Nd. PMM = Pacific MORB mantle, fields shown are approximate possibleranges for inferred components A and B. Curved lines are mixing lines betweenthese components assuming Nd/Zr = 0.12 in component A and 0.06 in com-ponent B. The key to the symbols is that given in Figure 10.

is also a common consideration in boninite models (e.g., Stern et al.,1991). It is therefore quite probable that component A will vary in itstrace element ratio according to which part of the lithosphere wasmelted. Isotopically it may also be variable, as even single ridgesegments are known to exhibit small isotopic variations (e.g., Whiteet al., 1987). The component has therefore been denoted by an ellipserather than a point in Figure 10, and mixing lines have been drawn, wherepossible, for both fertile and refractory end-members of component A.

Component B is characterized by low ε Nd. Its isotopic composi-tion cannot be determined directly, but the Ti/Zr and Y/Zr plots placeconstraints on the possible range of values. It is apparent from theseplots that component B has low Ti/Zr and Y/Zr ratios and that anyreasonable extrapolation of the mixing trend would intersect thehorizontal axis at a value of εNd between about 3 and 5. The precisevalue will depend on the curvature of the mixing trend and the exactcomposition of component A. The εNd value of component B couldlie anywhere between this intercept and the samples with lowest εNd(the LCB group), i.e., between about 4 and 6.5. We have assumed amean value of +5 in Figure 10 but have presented the component asan ellipse with its long axis along the mixing curve. It is then apparentthat component B has Th/Zr and Pb/Zr ratios significantly greater thanPMM, a Nb/Zr ratio less than PMM (but greater than component A),and Ti/Zr and Y/Zr ratios significantly less than PMM.

The REEs also help to characterize component B, despite theirmobility during alteration which has reduced the effective sample sizeand precluded the use of the most mobile element, La. Plots of fourREE/Zr ratios against Nd are presented in Figure 11. These are also

0.08 -,

0.06 -

0.04 -

0.02 -

0.00

error

10


consistent with a two-component mixing model. The Ce/Zr ratio isunchanged within the large limits of alteration and experimental error,giving values of 0.8-1.2 for components A and B. Nd, Sm, and Yb allhave lower ratios to Zr in component B compared with component A.The Sm/Zr ratio is particularly significant. Sm and Zr are generallyregarded as exhibiting similar petrogenetic behavior, and are often placedadjacent to each other in multi-element patterns (Wood, 1979; Pearce,1983; Thompson et al., 1983). The fact that component A has a similarSm/Zr ratio to PMM in Figure 11 is consistent with the similarity inpetrogenetic behavior of these two elements. Several authors have notedthat boninites commonly have positive Zr anomalies on these patterns(Sun and Nesbitt, 1978; Cameron et al, 1983; Hickey-Vargas, 1989;Murton et al., this volume). It is apparent from this diagram that com-ponent B, if a global feature, could be the cause of these anomalies. Thenature of component B is not, however, immediately obvious and it istherefore discussed in more detail in the next section.

A further point to emerge from these figures is the spread of traceelement ratios across the mixing trends. This may in part be causedby experimental error. However, it is also a predictable consequenceof mixing between a variable component A (from more fertile to morerefractory lithosphere) and a single component B. A given massfraction of component B will have a much greater effect on thecomposition of depleted mantle lithosphere than on the compositionof more fertile mantle lithosphere. The fact that the low-Ca boninitegroup tends to plot with slightly lower εNd values coupled withslightly higher ratios of Y/Zr and lower ratios of Nb/Zr and Th/Zrcompared to the intermediate-Ca boninite group, may therefore sug-gest derivation from the more refractory lithosphere. Note, however,that partial melting and alteration will also contribute to these varia-tions. These ideas are developed further in the light of other lines ofevidence in the synthesis chapter (Pearce et al., this volume).

255

J. A. PEARCE ET AL.

N

C/)

10 η

8 -

6 -

4 -

2 -

0PMMD

6 7 8 9 10

1.0 i

/Sr

>—N

0

0

0

.8 -

.6 -

.4 -

0.2 -

PMM

Figure 12. Plots of (A) Sr/Zr against εNd and (B) Zr/Sr against ε Sr showingthe additional component, C. Mixing lines assume Zr/Sr = 0.85 in componentA, 0.4 in component B, and 0 in component C. The details otherwise are thosegiven in Figures 10 and 11.

All the elements so far plotted can be interpreted in terms of a simpletwo-component mixing model. Sr is a notable exception, as shown inFigure 12A. The diagram Sr/Zr vs. εNd gives a mixing trend for theboninites but intersects the ε Nd = 10 line at very high Sr/Zr ratios, of atleast 10. The tholeiites may follow a similar trend but intersect the εNd= 10 line at a value of about 5. By contrast, PMM has a Sr/Zr ratio ofabout 1. Moreover, this ratio should be even lower in suboceanic litho-sphere because Sr is more incompatible than Zr during MORB genesis.The simplest explanation is that a third component, carrying Sr but notthe REE, Th, Pb, Ti, Zr, Y, or Nb, has increased the Sr/Zr ratio. The mostobvious source of this component is an aqueous phase derived from asubduction zone. This component, termed C, has been depicted on thediagram as an arrow to very high Sr/Zr ratios.

An equivalent diagram of ε Sr against Zr/Sr ratio has also beenplotted in Figure 12B to define this component more clearly. Thecompositional ranges for components A and B have been constrainedby Figure 5 and Figure 12A. Figure 12B then constrains componentC to very low Zr/Sr ratios but also to low ε Sr values of between about-25 and -20. Thus, if this component were a fluid derived fromsubducted crust, it could not have been displaced far from the originalcrustal composition. In other words, the strontium released contains

a large crustal component but only a small seawater component. Thisis explicable if the subducted crust were young, as it would then carrylittle sediment and have experienced little oxidative alteration. It isalso explicable if the fluid were derived from serpentinite dehydra-tion, as Sr would then have been extracted primarily from the rela-tively unaltered crustal rocks.

Figure 12 also shows that the tholeiites from Hole 782 and thebasaltic andesite from Hole 786A have changed their order on themixing trend when ε Sr is used as the horizontal axis. On mixing linesbased on εNd and Pb isotope ratios, they plot closest to PMM. Onmixing lines based on εSr values, they are intermediate between thetwo boninite groups (the 41 Ma boninitic edifice, and the late dikesfrom Hole 786B plus the lavas from Hole 786A). The geometry ofFigure 12B requires that the order can only change if component C ismore radiogenic in the tholeiites than in the boninites. The fact thatthe tholeiites lie off the boninite trend in Figure 12A further suggeststhat the tholeiites differ from the boninites in their interaction withthe high-Sr component.

It may be particularly significant that the mixing trend lies in thedirection of component B in the diagrams on Figure 12. This would bethe expected trend if component C were added to the mantle lithosphere(A) before component B. If component B were added to the mantlelithosphere before component A, an apparent trend could result by mixingexactly the right proportions of component C to a variable mixture ofcomponents A and B. This option cannot be ruled out at this stage,although it must be regarded as having a lower probability.

The average compositions of components A-C can be read off theplots in Figures 10-12 as element/Zr ratios. These are presented inpattern form in Figure 13 as the element/Zr ratio in the componentnormalized to the element/Zr ratio in PMM. Note that there are noconstraints on the absolute Zr concentrations (and hence the absoluteconcentrations of other elements) in these components. PMM itselfforms a straight line on these plots, its position depending on these Zrconcentrations. PMM reference lines have been drawn for Zr contentsof 11 ppm in PMM, 4 ppm in component A, and 100 ppm incomponent B. The pattern for component A is comparable to thatexpected for suboceanic lithosphere that has lost a small melt fraction,as discussed above. The pattern for component C is comparable tothat expected for a slab dehydration component carrying only themost mobile large-ion lithophile (LIL) elements. The pattern forcomponent B is characterized by high contents of Pb, Th and Sr,moderate contents of Zr, La, and Ce, lower contents of Ce-Sm, andlow contents of M-HREE, Y, and Ti. Component B is quite distinctfrom most components conventionally considered to be involved inmagma genesis and is investigated further below.

Figure 14 compares the trends followed by Leg 125 samples onthree of the diagrams with those followed by the Mariana arc (Lin etal., 1989, 1990; Bloomer et al., 1989), Mariana Trough (Stern et al.,1990; Volpe et al., 1987,1990; Hawkins et al., 1990) and Sumisu Rift(Hochstaedter et al., 1990). It is apparent that the present-dayMariana-Bonin arc-basin system is characterized by Ti/Zr and Sm/Zrratios that remain constant or increase as the compositions becomemore radiogenic, rather than decrease as in the Leg 125 samples. Onthe plot of Sr/Zr against ε Sr, the rocks from the extant arc basin systemform a simple two-component mixing line between PMM and aradiogenic component with a high Sr/Zr ratio. The nature of the com-ponents in the active Mariana arc-basin system has been discussed insome detail by the authors cited above. They consider three components:a MORB-manüe; a subduction component that has metasomatized thisMORB mantle; and an ocean island basalt (OIB) type mantle that mightrepresent "trapped" lithosphere. Clearly, there is no similarity betweenthese components involved in present arc-magmatism in the Izu-Boninregion and the components involved in Eocene forearc magma genesisat Sites 782 or 786.

The nature of component B can also be investigated by comparingthe same isotope-trace element plots for the Leg 125 lavas with thosefor Pacific hotspots and ridges (Fig. 15). Data plotted are taken from

256


10

.1

. PMM

Component A

/ £Nd =

' ESr= c

Pb6/4 =

c.+10

.-30

= c. 18.5

1 ~ T•

100 -

50 -

0 -

α

error

i

5 3 0 " < ^

ti ° °

. o

a

* *

Th Pb Nb La Ce Sr Nd Sm Zr Ti Y Yb8 10 12

10 i

Component B

É Nd = +4 to +6

B *Sr=-10to-6

Pb6/4 = 19.0 to 19.3- Q

PMM.1 ' i i i i i i i i i i : T = F

Th Pb Nb La Ce Sr Nd Sm Zr Ti Y Yb

100

1 0 -ü

Component C

€Sr=-25to-15

Figure 13. Trace element patterns and isotopic composition for the threecomponents inferred for the Leg 125 boninites and tholeiites. A*, B* and C*refer to the ratios of elements to Zr in the three components. PMM* refers tothe ratios of elements to Zr in Pacific MORB Mantle (PMM). Compositionsof components A and B are taken from Figures 10 and 11.

0.12 -

0.10 -

0.08 -

0.06 -

0.04 -

0.02 -

0.00

a(3 ö

O

•

Mariana arc

Mariana TroughSumisu Rift

Leg 125

error

10 12

Figure 14. Plots of Ti/Zr and Sm/Zr against ε ND showing the contrast betweenthe Leg 125 boninites and tholeiites and subduction-related lavas from thepresent-day Mariana/Volcano arcs, Mariana Trough and Sumisu Rift. See textfor data sources. Compositions of components A and B are taken from Figures10 and 11.

East Pacific Rise off-axis seamounts (Zindler et al., 1984), Hawaii(Chen and Frey, 1985; Staudigel et al., 1984), and the Caroline islands(Dixon et al., 1984). All these data plot on trends of constant orincreasing Ti/Zr and Sm/Zr ratio, again showing no evidence for acomponent of the type seen in the boninites. Thus, although paperson boninites frequently refer to "OIB components," there is noevidence that such a component was involved in the genesis of theLeg 125 boninites.

One key question in understanding component B is how to frac-tionate both the Ti/Zr and Sm/Zr ratios. This question is discussed inmore detail elsewhere (Pearce et al., this volume). The mainpossibilities are partial melting, a subduction component, and anasthenospheric (OIB) component. If these ratios are to be fractionatedduring partial melting, it is most unlikely that this could beaccomplished by minor phases such as zircon and REE-phases, bothof which are soluble in basic melts at mantle temperatures (e.g.,Watson and Harrison, 1984). The only phase that might do this isamphibole, which would retain Ti and Sm relative to Zr. In a partialmelting model the low-Ca boninites and intermediate-Ca boniniteswould have to retain the most amphibole in their residues to explaintheir high Zr/Sm ratios. This scenario is not consistent with the

257

J. A. PEARCE ET AL.

scarcity of amphibole in the residual peridotites recovered from theforearc (Ishii and Robinson, this volume), nor with the fact that traceelement fractionation is accompanied by isotopic variations.

In the models requiring subduction zone enrichment, melting ratherthan dehydration must be invoked to mobilize Zr and residual am-phibole must be present to fractionate Zr from the M-HREE (residualzircon would fractionate these elements, but deplete, rather than enrich,Zr in the melt). Melting of hydrated oceanic crust in amphibolite faciesis the simplest explanation. Many tonalitic and trondhjemitic in-trusions, and andesite-rhyolite lavas, most notably in the Archaean(e.g., De Wit et al., 1987; Drummond and Defant, in press) but also inophiolites (Pedersen and Malpas, 1984) and young volcanic arcs(Defant and Drummond, in press) have been interpreted as partial meltsof amphibolite. They typically have Zr/Y ratios of 20 or more andZr/Sm ratios of 60 or more, which would explain the trends in Figure10. Moreover, amphibolite melting has been observed in metamorphicsoles beneath ophiolitic slabs (e.g., Pearce, 1989) and subduction zonemetamorphic terranes (Sorensen, 1988). Note, however, that com-ponent B is more radiogenic than PMM. This model would not,therefore be consistent with fusion of normal oceanic crust, but wouldfit the fusion of volcanogenic sediments or of transitional oceanic crust.

The third option, metasomatism by an asthenospheric melt, wasproposed by Hickey-Vargas (1989), who was one of the first to stressthe importance of high Zr/Sm ratios in boninites. Although Figure 15demonstrated that this option fails to explain the unusual fractionationof Zr from Sm, an asthenospheric melt which had equilibrated withamphibole could achieve the required fractionation. One possiblescenario is during ridge subduction, where asthenospheric fluidscould penetrate the "slab window" and interact with a mantle wedgecontaining amphibole-bearing peridotite. In this case, the chroma-tographic effect, applied to the mantle wedge by Navon and Stolper(1987) and to the genesis of boninites in particular by Stern et al.(1991), may have a significant effect on isotope as well as traceelement ratios. These scenarios are depicted in schematic form inFigure 16 and a more detailed discussion is carried out in the synthesischapter (Pearce et al., this volume). It must be emphasized, however,that these and other options require further examination before anyrigorous tectonic model can be developed.

REGIONAL VARIATIONS IN ISOTOPECOMPOSITION

The boninites of Leg 125 are the latest in a series of Eocene-Oligocene terranes from the Izu-Bonin-Mariana region that have beenanalyzed isotopically. These terranes comprise:

1. The West Philippine Basin (Meijer, 1975)2. Chichijima (Hickey and Frey, 1982; Cameron et al, 1983;

Dobson and Tilton, 1989, Karpenko et al., 1985)3. Guam (Meijer, 1976; Hickey-Vargas and Reagan, 1987)4. Saipan (Meijer, 1983)5. DSDPLeg 60 (Mariana forearc) (Hickey-Vargas, 1989)6. The Mariana inner trench wall (Stern et al., 1991)The interpretation of these data, and their implications for the

understanding of the Eocene-Oligocene crustal accretion in this partof the Western Pacific, has already been covered in depth by the aboveauthors. The range of magma types recovered by Leg 125 does,however, greatly extend the forearc data base so that further discus-sion is worthwhile. Our aim in this section is to examine the regionalimplications of the Leg 125 isotope data.

Lead isotope plots are generally regarded as most informative.Figure 17 shows a plot of 208Pb/204Pb isotope ratios relative to the206Pb/2O4Pb for Leg 125 samples in the context of the regional data.Figure 17A addresses the question of the origin of the suboceaniclithosphere in the Izu-Bonin region (component A). It is generallyheld (e.g., Meijer, 1980; Hickey-Vargas, 1989) that subduction beganat a transform margin to the present West Philippine Basin, and thatWest Philippine Basin suboceanic lithosphere, hydrated and further

140 -

120 -

100 -

80 -

60 -

40 -

20 -

0

o o

o , t>

<P + +o + l . H

++

error

+

10 12

ECΛ

0.04 -

0.03 -

0.02 -

0.01 -

0.00 -

° So>*

error :

+ %

• i •

,;.;; O

1

O

×

x „

< M

Hawaii (H)

Hawaii (L)

Caroline Is.

EPR smts.

Leg 125—j !

10 12

Figure 15. Plots of Ti/Zr and Sm/Zr against ε Nd showing the contrast betweenthe Leg 125 boninites and ridge and hotspot-related lavas from the East Pacific

Rise seamounts, Hawaii (L= Loihi; H = Haleakala) and Caroline islands. See

text for data sources. Compositions of components A and B are taken from

Figures 10 and 11.

metasomatized from an incipient subduction zone, is the source ofmost Eocene and Oligocene boninites from the region. However,Figure 17 A shows that the West Philippine Basin gives a trend parallelto the NHRL, but displaced to higher 208Pb/204Pb ratios. Moreover, Srisotope data from dredged and drilled basement samples from thePhilippine Sea (Matsuda, 1985) give values in excess of MORB (onthe order of 0.7038 ±4), again suggesting that the Philippine Sealithosphere is unlikely to be related to PMM and hence is unlikely torepresent component A in the genesis of Leg 125 samples. MarianaTrough data, plotted for comparison, form a trend between the Philip-pine Sea and the Leg 125 samples. They also cannot obviously berelated to the genesis of the Eocene-Oligocene lavas and dikes. Wecan therefore infer that the Pb in the Leg 125 samples was probablyderived from Pacific lithosphere located east of the West PhilippineBasin in the Eocene and Oligocene.

Figure 17B compares Leg 125 data with data from other WesternPacific boninites of comparable age. On this diagram, the DSDP Leg 60samples, boninites from the inner trench wall and the Facpi group fromGuam most resemble the Eocene-Oligocene samples from Leg 125. Allplot close to the NHRL with a similar range of lead isotope ratios. Thetholeiitic andesite and dacite from Site 782 and the tholeiites from DSDPLeg 60 also plot on or near the NHRL but at lower lead isotope ratios.The tholeiitic suites from the Alutom group on Guam and andesites-rhyolites on Saipan have still lower radiogenic lead isotope ratios. Thisdistribution of Eocene-Oligocene data along the NHRL, indicating an

258


1. Creation of zoned, depletedoceanic lithospheric mantle(component A) at a mid-oceanridge.

2. Dehydration of young,subducted lithosphere torelease an aqueous fluid(component C)

39.0 -j

S 38.5 -

38.0 -

37.5

PVS

PPS

PPMo Philippine Sea

Mariana Trough

* Leg 125

18.0 18.5 19.0 19.5Pb206/204

20.0

*~•N

B13. Partial melting of subductedcrust to give a tonalitic melt(component B1) or chromatographyfractionation of asthenosphericmelt (component B*). Amphibole isinvoked in both cases tofractionate the Sm/Zr and Ti/Zrratios.

Figure 16. Schematic representation of a possible tectonic explanation for theisotopic components in Leg 125 lavas. Note that other scenarios involvinghot-hot subduction are also possible. For a more detailed discussion of thetectonic options, see Pearce et al. (this volume).

absence of subducted sediment lead in the mantle source, was notedby Hickey-Vargas (1989) who concluded that the subducted sedimentsignature was absent from pre-Miocene lavas from the Marianaarc-trench region. The exception is the lavas from Chichijima, whichplot off the NHRL toward pelagic sediment compositions. TheChichijima lead iso-tope distribution was discussed by Dobson andTüton (1989) who concluded that several percent of subducted sedi-ment are required to explain the 208Pb/204Pb displacement.

Figure 18 compares the Leg 125 samples with the other WesternPacific Eocene-Oligocene terranes on the basis of three of the isotope-trace element ratio diagrams modeled in this paper. The additional pelagicsediment component in the Chichijima samples is clearly shown by thetrend to low εNd values. This component, labelled "D," could representsediment subducted during the Eocene or have been a long-lived charac-teristic of the lithosphere in that region. The other terranes would plotwithin a triangle made up of components A, B, and C as defined byLeg 125 data. However, the proportions and precise compositions ofthese components cannot be ascertained without further work throughoutthe Western Pacific boninite province.

CONCLUSIONS

Summary of Interpretations Based on IsotopeCovariations

1. The Eocene-Oligocene samples all plot between depletedPacific MORB mantle (PMM) and Pacific Volcanogenic Sediment

CDoçgCOoCM

39.0 -

38.5 -

38.0 -

37.5

PVS

125

Chichijima

Saipan

Guam(Facpi)

Guam(Alutom)

DS DP Leg 60

Mariana Trench

18.0 18.5 19.0Pb206/204

19.5 20.0

Figure 17. Pb isotope diagram showing the Leg 125 Eocene-Oligocene lavasin the context of (A) regional ridge basalts and (B) regional boninites andrelated rocks. PPS = Pacific Pelagic Sediment; PVS = Pacific VolcanogenicSediment; PPM = Pacific MORB Mantle. See the text for data sources.

(PVS) (i.e., along a Northern Hemisphere mantle array) in Pb-Ndisotope space. There is no evidence of subducted pelagic sediment inthe sources of these samples.

2. These samples also plot parallel to a line from PMM to PVS inSr-Pb and Sr-Nd isotope space, but displaced to higher ε Sr values.

3. Isotope ratios indicate possible petrogenic links between (1) thelow-Ca boninites and low-Ca bronzite andesites from the base ofHole 786B; (2) the intermediate-Ca boninites, intermediate-Ca bron-zite andesites, andesites, dacites, and rhyolites from Hole 786B; (3)the late boninite dikes from Hole 786B and the boninite clasts fromHole 786A; (4) the tholeiitic andesites, dacites, and rhyolites fromSite 782. The basaltic andesite clast from Core 6X-5 in Hole 786Ahas a distinctly different composition.

3. The young basaltic andesites from Hole 786A plot on trendstoward subducted sediments on Pb isotope diagrams, but this is aslikely to be the result of interaction with sedimentary pore fluids as itis to have been caused by the presence of any subducted pelagicsediment in their source regions.

4. Within the Eocene-Oligocene samples, there is a systematicvariation along the Northern Hemisphere mantle array, with thelow-Ca boninite group showing most displacement, followed by themedium-Ca boninite group, then the high-Ca boninite group, and thetholeiites from Site 782 plotting closest to PMM.

259

J. A. PEARCE ET AL.

140 -

120 -

100 -

80 -

60 -

40 -

20 -

0 -

GuamÅ Chichijima

a DSDP Leg 60

+ Mariana Trench

ODP Leg 125

A,

yh A

m

à

I F-2 10

0.05 -,

0.04 -

_ 0.03 -NE

w 0.02 -

0.01 -

0.00

E3A

-2 2 4 6(Nd

10 -|

8 -

6 -

W) 4 _

2 -

8 10

-2 10

Figure 18. Plots of Ti/Zr, Sm/Zr, and Sr/Zr against εNd for Leg 125 samplescompared with regional boninites and related rocks. See Figures 10-12 fordetails of terms used. See text for data sources.

Summary of Interpretations Based on Isotope-TraceElement Diagrams

1. Much of the observed variation can be explained by the mixingof three components (A-C).

2. Component A has an isotopic composition close to PMM, butits trace element composition is characterized by high ratios of Yb/Zrand Y/Zr and low ratios of Nb/Zr, Th/Zr and Pb/Zr. It is interpreted

as suboceanic lithosphere that has lost a melt fraction during its earlierhistory at a mid-ocean ridge. This component is likely to have a variabletrace element composition according to the degree of depletion of thelithosphere, and hence the depth at which melting takes place.

3. Component B has an isotopic composition closer to PVS withε Sr of -10 to -6, εNd of 6 to 4, and 206Pb/204Pb of 19.0 to 19.3. It ischaracterized by low ratios of Ti/Zr, Y/Zr, and Sm/Zr and high ratiosof Th/Zr. These ratios are quite different from any known subductionor asthenospheric component presently involved in magma genesisin the Western Pacific. It can be explained in terms of a melt that hasfractionated the above ratios by equilibration with amphibole, but itis not clear at this stage whether this took place by fusion of thesubducted slab or by chromatographic fractionation of normalasthenospheric melts.

4. Component C appears only to affect Sr (and perhaps other alkaliand alkaline earth elements) and ε Sr. It has an ε Sr value of about -25to -15. It is tentatively interpreted as an aqueous fluid derived bydehydration of subducted oceanic crust. The mixing trends suggestthat component C may be added to the suboceanic lithosphere (com-ponent A) before component B.

Summary of Interpretations Based on RegionalVariations

1. The samples from Leg 125 resemble most other Eocene-Oligoceneboninitic terranes in the Western Pacific in having no significant pelagicsediment component. Only lavas and dikes from Chichijima in thesubaerial part of the Izu-Bonin outer-arc high exhibit a clear trend towardPacific pelagic sediment on a Pb isotope diagram.

2. Of all the analyzed samples from the Mariana and Izu-Boninforearcs, the samples from Leg 125 most resemble isotopically theboninitic Eocene Facpi Unit on Guam and boninites and tholeiitesfrom DSDP Leg 60 drill-sites in the Mariana forearc, all of which plotalong the NHRL toward radiogenic Pb isotope ratios.

3. It is possible that the Mariana and Izu-Ogasawara forearcs wereformed by a highly variable (in space and time) petrogenetic systeminvolving, in varying degrees, suboceanic lithosphere, aqueous sub-duction fluids, a slab (amphibolite)-melt subduction component orasthenospheric melt component that had equilibrated with amphibole,and a pelagic sediment-derived subduction component. More work isrequired to test this assertion and to explain in detail the temporal andspatial variability within the forearc terrane.

A C K N O W L E D G M E N T S

J.A.P. and B.J.M. thank the Natural Environment Research Council(U.K.) for an ODP Special Topics Research Grant (GR3/416). J.A.P. isgrateful to Bob Jackson (University of Durham Industrial ResearchLaboratories) for assistance with the ICP-MS and Ceri Jenkins forassistance in the RHBNC isotope laboratory. The RHBNC isotopelaboratory is a London University intercollegiate facility. R.J.A. andS.v.d.L. thank USSAC for logistic and analytical costs. We all greatlyappreciate the shipboard contributions of Patty Fryer, Laura Stokking,and the other shipboard scientists, marine technicians, and crew of theJOIDES Resolution. Pamela Kempton and Chris Hawkesworth providedhelpful reviews of an earlier version of the manuscript.

REFERENCES

Ben Othman, D., White, W. M, and Patchett, J., 1989. The geochemistry ofmarine sediments, island arc magma genesis and crust-mantle recycling.Earth Planet. Sci. Lett., 94:1-21.

Bloomer, S. H., Stern, R. J., Fisk, E., and Geschwind, C. H., 1989. Shoshoniticvolcanism in the northern Mariana Arc. 1. Mineralogic and major and traceelement characteristics. /. Geophys. Res., 94:4469-4496.

Cameron, W. E., McCulloch, M. T., and Walker, D. A., 1983. Boninitepetrogenesis: chemical and Nd-Sr isotopic constraints. Earth Planet. Sci.Lett., 65:75-89.

260


Chen, C.Y., and Frey, F. A., 1985. Trace element and isotopic geochemistry oflavas from Haleakala volcano, East Maui, Hawaii: implications for theorigin of Hawaiian basalts. J. Geophys. Res., 90:8743-8768.

Crawford, A. J. (Ed.), 1989. Boninites and Related Rocks: London (UnwinHyman).

Defant, M. J., and Drummond, M. S., in press. Subducted lithosphere-derivedandesitic and dacitic rocks in young volcanic arc settings. Nature.

De Wit, M. J., Armstrong, R., Hart, R. J., and Wilson, A. H., 1987. Felsicigneous rocks within the 3.3- to 3.5-Ga Barberton greenstone belt: highcrustal level equivalents of the surrounding tonalite-trondhjemite terrane,emplaced during thrusting. Tectonics, 6:529-549.

Dixon, T. H., Batiza, R., Futa, K., and Martin, D., 1984. Petrochemistry, ageand isotopic composition of alkali basalts from Ponape Island, WesternPacific. Chem. Geoi, 43:1-28.

Dobson, P. F., andTilton, G. R., 1989. Th, U and Pb systematics of boninite seriesvolcanic rocks from Chichijima, Bonin Islands, Japan. In Crawford, A. J.(Ed.), Boninites and Related Rocks: London (Unwin Hyman), 396-415.

Dosso, L., and Murthy, V. R., 1980. A Nd isotopic study of the Kerguelenislands: inferences on enriched oceanic mantle sources. Earth Planet. Sci.Lett., 48:268-276.

Drummond, M. S., and Defant, M. J., in press. A model for trondhjemite-tonalite-dacite genesis and crustal growth via slab melting: Archaean tomodern comparisons. J. Geophys. Res.

Fryer, P., Pearce, J. A., Stokking, L. B., et al., 1990. Proc. ODP, Init. Repts.,125: College Station, TX (Ocean Drilling Program).

Hart, S. R., 1984. A large-scale isotope anomaly in the Southern hemispheremantle. Nature, 309:753-757.

, 1988. Heterogeneous mantle domains: signatures, genesis andmixing chronologies. Earth Planet. Sci. Lett., 90:273-296.

Hawkesworth, C. J., 1982. Isotope characteristics of magmas erupted alongdestructive plate margins. In Thorpe, R. S. (Ed.), Andesites: New York(Wiley), 549-571.

Hawkesworth, C. J., and Norry, M. J. (Eds.), 1983. Continental Basalts andMantle Xenoliths: Nantwich (Shiva).

Hawkins, J. W, Lonsdale, P. F, Macdougall, J. D., and Volpe, A. M., 1990.Petrology of the axial ridge of the Mariana Trough backarc spreadingcenter. Earth Planet. Sci. Lett., 100:226-250.

Hickey, R. L., and Frey, F. A., 1982. Geochemical characteristics of boniniteseries volcanics: implications for their source. Geochim. Cosmochim.Acta, 46:2099-2115.

Hickey-Vargas, R. L., 1989. Boninites and tholeiites from DSDP Site 458,Mariana forearc. In Crawford, A. J. (Ed.), Boninites and Related Rocks:London (Unwin Hyman), 339-356.

Hickey-Vargas, R. L., and Reagan, M. K., 1987. Temporal variations of isotopeand rare earth element abundances from Guam: implications for theevolution of the Mariana arc. Contrib. Mineral. Petrol, 97:497-508.

Hochstaedter, A. G., Gill, J. B., and Morris, J. D., 1990. Volcanism in the SumisuRift, IL Subduction and non-subduction components. Earth Planet. Sci. Lett,100:195-209.

Karpenko, S. F, Sharaskin, A. Y, Balashov, Y. A., Lyalikov, A. V, andSpiridonov, V. G., 1985. Isotopic and geochemical criteria for the origin ofboninites. Geochem. Int., 22:1-12.

Lin, P. N., Stern, R. J., and Bloomer, S. H., 1989. Shoshonitic volcanism in thenorthern Mariana arc, 2. Large-ion lithophile and rare earth elementabundances: evidence for the source of incompatible element enrichmentsin intraoceanic arcs. /. Geophys. Res., 94:4497-4514.

Lin, P. N., Stern, R. J., Morris, J., and Bloomer, S. H., 1990. Nd- and Sr-isotopiccompositions of lavas from the northern Mariana and southern volcanoarcs: implications for the origin of island arc melts. Contrib. Mineral.Petrol., 105:381-392.

Meijer, A., 1975. Pb and Sr isotopic studies of igneous rocks cored during Leg31 of the Deep Sea Drilling Project. In Karig, D. E., Ingle, J. C, Jr., et al.,Init. Repts. DSDP, 31: Washington (U.S. Govt. Printing Office), 601-605.

, 1976. Pb and Sr isotopic data bearing on the origin of volcanic rocksfrom the Mariana arc system. Geol. Soc. Am. Bull., 87:1358-1369.

, 1980. Primitive arc volcanism and a boninitic series: examplesfrom the Western Pacific island arcs. In Hayes, D. E. (Ed.), The Tectonicand Geological Evolution of Southeast Asian Seas and Islands. Am.Geophys. Union, Geophys. Monogr. Sen, 23:269-282.

-, 1983. The origin of low-K rhyolites from the Mariana frontal arc.Contrib. Mineral. Petrol., 83:45-51.

Matsuda, J., 1985. Sr isotopic studies of rocks from the Philippine Sea andsome implication for mantle material. In Shiki, T. (Ed.), Geology of theNorthern Philippine Sea: Tokyo (Tokai Univ. Press), 63-78.

Navon, O., and Stolper, E., 1987. Geochemical consequences of melt percola-tion: the upper mantle as a chromatographic column. /. Geol., 95:285-307.

Pearce, J. A., 1983. Role of the sub-continental lithosphere in magma genesis atactive continental margins. In Hawkesworth, C. J., and Norry, M. J. (Eds.),Continental Basalts and Mantle Xenoliths: Nantwich (Shiva), 230-249.

, 1989. HT/P metamorphism and granite genesis beneath ophiolitethrust sheets. Ofioliti, 14:195-211.

Pearce, J. A., and Norry, M. J., 1979. Petrogenetic implications of Ti, Zr, Y andNb variations in volcanic rocks. Contrib. Mineral. Petrol., 69:33-47.

Pedersen, R. B., and Malpas, J., 1984. The origin of oceanic plagiogranitesfrom the Karm0y yophiolite, Western Norway. Contrib. Mineral. Petrol.,88:36-52.

Sorensen, S. S., 1988. Petrology of amphibolite-facies mafic and ultramaficrocks from the Catalina Schist, Southern California: metasomatism andmigmatization in a subduction zone metamorphic setting. /. Metamorph.Geol., 6:405-435.

Staudigel, H., Zindler, A., Hart, S. R., Leslie, T., Chen, C.-Y, and Clague, D.,1984. The isotope systematics of a juvenile intraplate volcano: Pb, Nd, andSr isotope ratios of basalts from Loihi Seamount, Hawaii. Earth Planet.Sci. Lett., 69:13-29.

Stern, R. J., Lin, P.-N., Morris, J. D., Jackson, M. C, Fryer, P., Bloomer, S. H.,and Ito, E., 1990. Enriched back-arc basin basalts from the northernMariana Trough: implications for the magmatic evolution of back-arcbasins. Earth Planet. Sci. Lett., 100:210-225.

Stern, R. J., Morris, J., Bloomer, S. H., and Hawkins, J. W., 1991. The sourceof the subduction component in convergent margin magmas: trace elementand radiogenic isotope evidence from Eocene boninites, Mariana forearc.Geochim. Cosmochim. Acta, 55:1467—1481.

Sun and Nesbitt, 1978. Geochemical regularities and genetic significance ofophiolitic basalts. Geology, 6:689-693.

Taylor, R. N., Lapierre, H., Vidal, P., Nesbitt, R. W., and Croudace, I. W, inpress. Igneous geochemistry and petrogenesis of the Izu-Bonin forearcbasin. In Taylor, B., Fujioka, K., et al., Proc. ODP, Sci. Results, 126:College Station, TX (Ocean Drilling program).

Thirlwall, M. F., 1991a. High-precision multicollector isotope analysis of lowlevels of Nd as oxide. Chem. Geol., 14:13-22.

, 1991b. Long term reproducibility of multicollector Sr and Ndisotope ratios. Chem. Geol., 14:85-104.

Thompson, R. N., Morrison, M. A., Dickin, A. P., and Hendry, G. L., 1983.Continental flood basalts ... Arachnids rule OK? In Hawkesworth, C. J., andNorry, M. J. (Eds.), Continental Basalts and Mantle Xenoliths: Nantwich(Shiva), 158-185.

Volpe, A. M., Macdougall, J. D., and Hawkins, J. W, 1987. Mariana Troughbasalts (MTB): trace element and Sr-Nd isotopic evidence for mixing betweenMORB-like and Arc-like melts. Earth Planet. Sci. Lett., 82:241-254.

Volpe, A. M., Macdougall, J. D., Lugmair, G. W., Hawkins, J. W, and Lonsdale,P., 1990. Fine-scale isotopic variation in Mariana Trough basalts: evidencefor heterogeneity and a recycled component in back-arc mantle. Earth Planet.Sci. Lett, 100:251-264.

Watson, E.B., and Harrison, T. M., 1984. Accessory minerals and the geochemicalevolution of crustal magmatic systems: a summary and prospectus of ex-perimental approaches. Phys. Earth Planet. Int., 35:19-30.

White, W. M., Hofmann, A. W, and Puchelt, H., 1987. Isotope geochemistryof Pacific mid-ocean ridge basalt. /. Geophys. Res., 92:4881-4893.

Wood, D. A., 1979. A variably-veined suboceanic upper mantle—geneticsignificance for mid-ocean ridge basalts from geochemical evidence.Geology, 7:499-503.

Woodhead, J. D., 1989. Geochemistry of the Mariana arc (Western Pacific):source composition and processes. Chem. Geol., 76:1-24.

Zindler, A., and Hart, S., 1986. Chemical geodynamics. Ann. Rev. EarthPlanet. Sci., 14:493-571.

Zindler, A., Staudigel, H., and Batiza, R., 1984. Isotope and trace elementgeochemistry of young Pacific seamounts: implications for the scale ofupper mantle heterogeneity. Earth Planet. Sci. Lett., 70:175-195.

Date of initial receipt: 1 October 1990Date of acceptance: 28 October 1991Ms 125B-134

261

13. ISOTOPIC EVIDENCE FOR THE ORIGIN OF BONINITES AND ... › ... › VOLUME › CHAPTERS ›...

Documents

Transcript of 13. ISOTOPIC EVIDENCE FOR THE ORIGIN OF BONINITES AND ... › ... › VOLUME › CHAPTERS ›...