JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 94, NO. B4, PAGES 4497-4514, APRIL 10, 1989

SHOSHONITIC VOLCANISM IN THE NORTHERN MARIANA ARC 2. LARGE-ION LITHOPHILE AND RARE EARTH ELEMENT ABUNDANCES:

EVIDENCE FOR THE SOURCE OF INCOMPATIBLE ELEMENT ENRICHMENTS IN INTRAOCEANIC ARCS

Ping-Nan Lin and Robert J. Stern

Center for Lithospheric Studies, University of Texas at Dallas, Richardson

Sherman H. Bloomer

Department of Geology, Duke University, Durham, North Carolina

Abstract. The Mariana-Volcano-Izu arc system extends 2400 km north to south and is an outstanding example of an intraoceanic magmatic arc. In spite of this, the system is poorly known because most of it is submarine. Volcanism is entirely submarine in the northernmost Mariana arc, between 20040 ' and 24øN. This is the Northern Seamount Province (NSP) and was the focus of a detailed marine geologic and geochemical study, with additional data drawn from adjacent arc segments to the north (Volcano arc) and south (Central Island Province (CIP) of the Mariana arc). Samples from 24 submarine volcanoes and three islands were analyzed for concentrations of K, Rb, Sr, Ba, and the rare earth elements (REE). These data show strong variations along the arc, being relatively depleted in the more mature, tholeiitic and low-K calc-alkaline volcanoes of the Volcano

arc and the Mariana CIP, containing on average 6100 ppm K, 300ppmSr, 200 ppmBa, and6ppmLa. All of the NSP is enriched in large ion lithophile (LIL) and light rare earth elements (REE), particularly the northern half (26,000 ppm K, 700 ppm Sr, 900 ppm Ba, 47 ppm La); these lavas have strong shoshonitic affinities. These enrichments do not result from fractional crystallization of CIP-type melts. The source responsible for these enrichments shares some features in common with Mariana

CIP and Volcano arc sources: K/Rb and K/Ba in particular are similar (-500 and -30, respectively). However, Ba/La, Sr/Nd, and (Ce/Yb) n change drastically, with Ba/La and Sr/Nd decreasing to mantle values with increasing LIL and LREE enrichment. The origin of the LIL and LREE enrichments in the NSP shoshonites does not result from

variations in the behavior or composition of the subducted lithosphere. Melting occurs exclusively within the mantle wedge, and forward modeling of the REE patterns for all Mariana and Volcano arc lavas indicates that melt

generation occurs within the stability field of spinel lherzolite, probably within the upper 40-50 km of the subarc asthenosphere. Lavas from the large volcanoes of the Mariana CIP and Volcano arc result from 10-20%

melting of spinel lherzolite, followed by varying amounts of low-P fractional crystallization. Inferences based on REE forward models that the NSP shoshonites manifest

very low (1%) degrees of partial melting of LIL- and LREE-enriched spinel lherzolite are inconsistent with observed similar concentrations in tholelites and

shoshonites of high field strength cations such as Ti02 and Yb. Some of this inconsistency can be explained as resulting from source or melt mixing, with the NSP shoshonites being derived from a LIL- and LREE-

Copyright 1989 by the American Geophysical Union.

Paper number 7B7162. 0148-0227/89/007B- 7162505.00

enriched source or melt, with Ba/La and La/Yb indistinguishable from that of ocean island basalts (OIB), while Mariana CIP and Volcano arc melts are derived from

a depleted mild-ocean ridge basalt-like mantle that has been recharged with K, Rb, Sr, and Ba by hydrous fluids. These variations are interpreted as reflecting the evolution of the subarc asthenosphere, with a depletion in time resulting from the continuous extraction of basaltic melts.

Introduction

In spite of a plethora of petrologic, geochemical, and isotopic data, a vigorous debate continues regarding the processes responsible for the production of melts beneath convergent margins. Part of the uncertainty stems from the fact that most studies have examined magmatic systems where melts passed through easily fusible continental crust. In these "Andean-type" arcs, contamination of primitive melts by anatectic hybridization is always a concern but cannot easily be corrected for [Leeman, 1983; James, 1982; Davidson et al., 1987]. For this reason, scientists have become increasingly interested in studying the composition and evolution of magmatic arcs built on oceanic crust (intraoceanic arcs), where the opportunity for interaction between melt and crust is greatly reduced [Stern, 1982]. Over the past decade, intraoceanic arc systems such as the western Aleutians, Kuriles, Izu-Volcano-Mariana, Tonga- Kermadec, South Sandwich, and Lesser Antilles have thus been subjected to the full spectrum of petrologic, geochemical and isotopic scrutiny.

Nearly all studies of intraoceanic arcs have dealt with subaerially exposed volcanoes. Studies of subaerial edifices are extremely important, but these deal with only the upper portions of the largest and most evolved volcanoes. The submerged portions comprise 95% or more of the surface area of any intraoceanic magmatic arc, so it is critical that we investigate the composition of submarine arc volcanoes if we want to have a complete understanding of the composition of intraoceanic arcs. Furthermore, the relatively small volumes of submarine volcanoes indicate that they must either be younger or manifest relatively lower degrees of partial melting than the subaerial volcanoes. A better understanding of submarine arc volcanoes thus gives us a much better perspective with which to reconstruct the temporal evolution of arcs as well as the composition of their deeper sources.

The purpose of this paper is to report the results of a recent expedition to look at the petrological evolution of submarine arc volcanism. A central theme of this report is the significance of the discovery of an extensive province of shoshonitic volcanism for our understanding of arc petrogenesis. In the following sections we will present salient features of a recent expedition to the northern Mariana and Volcano arcs before we present concentration data for K, Rb, Sr, Ba, and the rare earth elements (REE). We will use these data to examine the composition of

4497

4498 Lin et al.: Northern Mariana Shoshonites, 2

submarine arc volcanoes, to reconstruct the depth of melting, and to define the trace element characteristics of the source region.

Regional Setting and Previous Work

The active, magmatic arc system formed by the subduction of the Pacific plate beneath the Philippine Sea plate extends south 2400 km from Japan, at 35øN to Guam, near 13øN. In the north it is known as the Izu arc and includes numerous volcanic islands; the southernmost volcanic island occurs at about 29øN. The Volcano arc (VA) extends from Nishino-Shima near 27øN to Minami Iwo Jima at 24øN; this segment is also referred to as the Banin arc [e.g., Simkin et al., 1981]. The northernmost volcanic island of the Mariana arc is Uracas at 20ø40'N; we assign all the submarine volcanoes as far north as Minami Iwo Jim to the Mariana arc as well. This is based

on the observation that an active backarc basin (Mariana Trough) is present behind the Mariana arc but not behind the Volcano arc. Following the geographic distribution of volcanic islands and seamaunts, the Mariana arc has been subdivided into three provinces (Figure 1): the Northern Seamaunt Province (NSP), the Central Island Province (CIP), and the Southern Seamaunt Province (SSP). The portion of the arc discussed in this report extends from the middle of the CIP northward to include all of the NSP and

the Volcano arc (Figure 1). Hussong and Uyeda [1981 ] argued that the active

140øE

28 ø , , •. )( 0 ' o ø

'?• ( •nln Islands) i

,

Kita Iwo Jlma

• Iwo J•ma

iMmam • J•ma

145 ø

(Guam) = Frontal Arc Island

Uracas = Achve Arc Island

Contour Interval=1000 frns

2O ø

A4od/hed from Chase, e_.r_•l, /968

140øE 145 ø I$(

150øE 28øN

25 ø

_

_

_

ZO ø

_

[5 ø

EI3øN Fig. 1. Locator map (modified after Chase et al., [1968]. Location of Figures 2 through 5 is shown by dashed boxes with numbers in lower left. Contour intervals are in

fathoms (1 fathom=l.829 m).

Mariana arc is constructed on young crust of the Mariana back arc basin, just west of the frontal arc ridge. This active arc is defined by well-defined edifices spaced 10-70 km apart [Bloomer et al., 1988]. These are the volcanoes of the magmatic front, the easternmost and predominant locus of magmatic activity in this and other arcs [Sugimura, 1968]. Associated with several volcanoes of the magmatic front are chains of smaller volcanoes extending into the back arc basin along WSW trends [Hussong and Fryer, 1983; Bloomer et al., 1988]. Because these "cross- chain" volcanoes may reflect the different processes of melt generation resulting from their position at a greater height above the Wadati-Benioff Zone relative to the volcanoes of the magmatic front, we do not consider them further here. We concentrate instead on the volcanoes of the magmatic front, which are situated about 100-150 km above the Wadati-Benioff Zone [Katsumata and Sykes, 1969]. The magmas of these volcanoes, by virtue of their similar position above the subducted slab, probably tap source regions of similar composition and thermal state.

Sampling of the volcanoes occurred during November and December, 1985, aboard the University of Washington R/V T. G. Thompson. A total of 82 dredge lowerings occurred during this cruise. The data reported here are based on sampling of 24 submarine edifices from along the magmatic front, only one of which (Fukujin) has been reported on before [Garcia et al., 1979; Wood et al., 1981]. Outlines of the physical volcanology and submarine mineralization have been reported elsewhere [Bloomer et al., 1988; Hein et al., 1987], as is a discussion of the relationship of the enriched magmatism with the tectonic setting and stage of arc evolution [Stern et al., 1988]. Petrography, mineral chemistry, and major and selected trace element chemical compositions are reported in a companion paper [Bloomer et al., this issue]. Also reported here for the first time are large ion lithophile (LIL) and REE data for samples from two islands (Nishino-Shima and Uracas) as well as more complete REE data for lavas from Iwo Jima previously reported by Stern et al., [1984]. For comparative purposes, we draw heavily on published reports of LIL and REE data for four islands in the study area (Pagan, Agrigan, Asuncion, and Iwo Jima [Dixon and Batiza, 1979; Stern, 1979; Chow et al., 1980; White and Patchett, 1984; Stern et al., 1984; Hole et al., 1984]).

Analytical Techniques

Dredge hauls from the 24 submarine volcanoes of the magmatic front were culled and sawed aboard the ship and immediately sorted for the range of lithologic variations. Locations of these dredges may be found elsewhere [Bloomer et al., 1988]. From this suite, 252 thin sections were prepared, examined, and described in order to better assess sample freshness and petragraphic variations. Representative samples which showed a minimum of evidence in thin section for low-temperature alteration were selected for analysis. Completely fresh samples could easily be obtained for all the volcanoes identified as active or dormant, but some slight alteration was always present in those selected from extinct edifices [Bloomer et al., 1988]. Seventy-one samples from 24 volcanoes were analyzed for concentrations of K, Rb, Sr, Ba, and the REE (La, Ce, Nd, Sm, Eu, Gd, Dy, Er, and Yb) using standard isotope dilution and cation exchange techniques. One sample each from the volcanic islands of Uracas and Nishino-Shima was also analyzed. Four samples from Iwo Jima which had previously been analyzed for LIL and most of the REE [Stern et al., 1984] were reanalyzed for La, and in one case (Suribati) for all REE. All dredge samples were slabbed and then cleaned by consecutively boiling in deionized water, rinsing in acetone, reboiling in deionized water, and ultrasonification in distilled water. Samples

Lin et al.: Northern Mariana Shoshonites, 2 4499

were dried overnight and pulverized in a tungsten-carbide ball mill. The chemical procedures are identical to those of Stern and Bibee [1984] except that the determination of La required complete separation of La and Ba. This was accomplished using an additional separation for light REE (LREE) at first using AG 50WxS, 200-400 mesh and eluting with HNO3, and later on by using Teflon beads coated with HDEHP resin and eluting with 0.15N HC1, a procedure modified after that of Richard et al., [1976]. Chondrite normalization for REE patterns is that of Nakamura [1974]. Total processing blanks for the procedure are 60 ng K, 0.1 ng Rb, 2 ng Sr, 7 ng Ba, 3.5 ng Ce, 0.5 ng Nd, 0.3 ng Sm, 0.2 ng Eu, 0.6 ng Gd, 0.3 ng Dy, 0.3 ng Er, and 0.4 ng Yb. Mean concentrations (and reproducibilities) for three to five determinations of USGS standard (BCR-1) during the course of this study follow: 14,191 ppm K (2.3%), 47.8 ppm Rb (0.7%), 336 ppm Sr (1.8%), 668 ppm Ba (0.4%), 24.6 ppm La (2.1%), 53.3 ppm Ce (0.6%), 28.1 ppm Nd (1.4%), 6.44 ppm Sm (1.7%), 1.99 ppm Eu (4.3%), 6.68 ppm Gd (2.1%), 6.34 ppm Dy (2.3%), 3.50 ppm Er (2.2%), and 3.17 ppm Yb (1.1%).

Results

Analytical results for 77 samples from the CIP and NSP of the Mariana arc and the Volcano arc are listed in

Table 1. The locations of the edifices that were sampled and analyzed are shown in Figures 2-5, denoted by the presence of REE patterns for each edifice from which rocks were analyzed. Also shown for purposes of comparison are REE patterns reported elsewhere for three CIP subaerial volcanoes: Pagan, Agrigan, and Asuncion [Dixon and Batiza, 1979; Hole et al., 1984; White and Patchett, 1984].

An extremely large variation is found in the data, especially for concentrations of K, Rb, Sr, Ba, and the LREE. The ranges observed are so large as to render meaningless any attempt to calculate mean concentrations of these elements in the study area as a whole. For this reason, we prefer to look at variations as a function of latitude which, since the arc trends about N-S, approximates position along the arc. Unless explicitly noted otherwise in the following discussion, active, dormant, and extinct edifices are discussed together. The geographic subdivision of the Mariana arc [Bloomer et al., 1988] is followed here, with the NSP being further subdivided into northern and southern portions as a result of the distinctly enriched composition of magmas in the northern NSP. This boundary is set between the seamounts of Nikko and Ko-Hiyoshi. Because Iwo Jima shares the distinctive enrichments of the northern NSP, it is included here as part of the northern NSP in spite of the fact that geographically it belongs to the Volcano arc. As the following discussion shows, there is little difference in the composition of magmas from the CIP and VA, and these overlap with the composition of magmas from the southern NSP.

Figure 6 shows the range in concentration of some of the most diagnostic elements (K, Sr, Ba, La, and Yb) as a function of latitude. Large variations are seen for all elements except Yb. Mean contents (+1 standard deviation) of K for the studied portion of the CIP (6100+3300 ppm) and the Volcano arc north of Iwo Jima (6600+3200 ppm) are similar to averages for the subaerial Mariana arc given by Chow et al., [1980]: K--6,896 ppm. No significant difference is observed when only submarine CIP volcanoes are considered. In contrast, lavas from the NSP are slightly more potassic than either the CIP or the Volcano arc north of Iwo Jima and are especially potassic in the far north. Compared with mean potassium contents in the CIP and the Volcano arc north of Iwo Jima, the southern NSP contains higher mean K (9,400+3800 ppm), although there

is some overlap among S-NSP, CIP, and VA, while the northern NSP is distinctly more enriched (26,000+9400 ppm K). Low-P fractionation of basaltic magmas may be responsible for some of the observed variations, especially between the CIP, VA, and S-NSP, but cannot account for the much larger enrichments found for the N-NSP.

Strontium contents show along-arc variations that are similar to but less extreme than potassium. Mean Sr contents of 288+48 ppm for the CIP north of 18øN are indistinguishable from a mean of 328+225 for the Volcano arc north of Iwo Jima; this is identical to the mean Sr concentration for subaerial Mariana arc lavas of 304 ppm listed by Chow et al., [1980]. Mean Sr contents in the southern NSP are about 50% greater, 424+139 ppm, while the northern NSP is enriched twofold, with a mean of 674+242 ppm. The range of Sr contents in the S-NSP shows considerable overlap with those of the CIP and VA, while the N-NSP shows much less overlap.

The along-arc variation in barium mimics that of potassium. The CIP north of 18øN averages 196+66 ppm Ba, indistinguishable from a mean of 215+89 ppm for the Volcano arc north of Iwo Jima; these means overlap with the mean of 215 ppm Ba calculated previously for the subaerial Mariana arc [Chow et al., 1980]. Ba is enriched about twofold in the southern NSP (mean 399+127 ppm) and about fourfold in the northern NSP (mean 856+200 ppm).

The variation in lanthanum is similar to but more

extreme than potassium or barium. Means for the studied portion of the CIP (5.6+3.1 ppm La) and the Volcano arc north of Iwo Jima (6.3+4.9 ppm La) are indistinguishable. Mean La content for the southern NSP is more than twice

as high (14.0+7.2 ppm La), while that of the northern NSP is almost an order of magnitude greater (46.6+12.7 ppm). In contrast, there is no significant along-arc variation of Yb contents, with means of 2.3+0.6, 2.5+0.8, 2.9+0.9, and 2.6+1.0 ppm calculated for the CIP, southern NSP, northern NSP, and Volcano arc north of Iwo Jima, respectively. These are identical to the mean of 2.5+_0.8 for the system as a whole.

The large variations in LIL and LREE cannot simply be the result of low-pressure fractional crystallization. This conclusion is based on the lack of systematic covariation of indices of fractionation and LIL and LREE

contents. Figure 7 is a plot of SiO 2 versus Ce for the samples studied here, along with observed and predicted fractionation trajectories of CIP-type melts. All low- pressure fractionation trajectories, both observed and modeled, result in relatively modest increases in the LREE over the silica range observed within the study area. A doubling or at most a tripling of Ce contents may be expected during the fractional crystallization of basaltic melts to produce andesites. The much higher contents of Ce (and all other LREE and LIL) observed in mafic rocks from Eifuku, Fukutoku, the Hiyoshi complex, and Iwo Jima therefore cannot result from low-pressure fractional crystallization of CIP-type basaltic melts. High-pressure fractional crystallization models are unlikely to be responsible because the enriched lavas of the NSP and Iwo Jima are not unusually depleted in MgO and compatible trace elements compared with CIP lavas [Bloomer et al., this issue]. We conclude instead that the enrichments of LIL and LREE in NSP and Iwo Jima lavas are

predominantly manifesting the characteristics of primary melts, although some fractional crystallization may have occurred subsequently to produce more siliceous magmas.

Incompatible element ratios K/Rb, K/Ba, Ba/La, and to a lesser degree Sr/Nd and (Ce/Yb) n manifest the composition of the melt source. To examine compositional variations along the arc, these are plotted in Figure 8 as a function of latitude in a fashion similar to that of Figure 6. Relatively minor variations in K/Rb are observed. The

4500 Lin et al.: Northern Mariana Shoshonites, 2

TABLE 1. Trace Element Analyses of Igneous Rocks from Mariana Seamounts

Central Island Province

Poyo Seamount (E) 19.16 ø

Element, ppm DS- 1 D8-3 D9-6

Cheref Seamount (E) 19.4 ø

D9-7 D9-20 D10-1-2 D10-2-6

K

Rb

Sr

Ba

La

Ce

Nd

Sm

Eu

Gd

Dy Er

Yb

K/Rb K/Ba Ba/La Sm/Nd Sr/Nd

(La/Yb) n (Ce/Yb) n Eu/Eu* Ce/Ce*

sio 2

1792

2.7

237

43.1

1.89

4.61

4.11

1.39

0.58

1.88

2.12

1.29

1.18

664

42

22.8

0.34

57.7

1.07

1.00

1.10

0.87

2028 63!4

3.8 14.7 244 284

48.1 190

1.92 6.7

4.87 15.22

4.21 10.6

1.42 2.85 0.60 1.02

1.93 3.37

2.16 3.45

1.37 2.07

1.28 1.97

534 430

42 33

25.0 28.4

0.34 0.27

58.0 26.8

1.00 2.27

0.97 1.96

1.11 1.01

0.90 0.95

48.46 48.22 53.54

4991 4862 9883 5604 11.6 11.1 24.8 10.4

245 306 284 228

169 128 267 191 4.4 5.4 10.4 5.0

11.00 12.70 22.4 12.7

8.4 8.9 12.8 10.2 2.55 2.47 3.02 3.16

0.94 0.89 1.03 1.15

3.28 3.19 3.20 4.24 3.83 3.48 3.29 4.6 2.55 1.89 2.09 2.98

2.27 1.71 2.24 2.74

430 438 399 539

30 38 37 29

38.4 23.7 25.7 38.2 0.30 0.28 0.24 0.31

29.2 34.4 22.2 22.4

1.30 2.11 3.10 1.22 1.23 1.89 2.54 1.18

1.00 0.97 1.01 0.97

0.95 0.96 1.00 0.94

55.35 51.30 63.98 56.32

Central Island Province

Cheref Seamount (E) 19.4 ø

Element, ppm D10-2-7 D10-2-11 Dll-ll D13-2

Supply Reef (D) 20.13 ø

D13-11 D14-1 D14-15

K

Rb

Sr

Ba

La

Ce

Nd

Sm

Eu

Gd

Dy Er

Yb

K/Rb K/Ba Ba/La Sm/Nd Sr/Nd

(La/Yb) n (Ce/Yb) n Eu/Eu* Ce/Ce*

5328 4740 4128 13.2 13.3 8.4

229 268 260

181 115 192 4.32 3.41 4.73

11.80 7.94 10.78 8.9 7.3 7.6 2.64 2.18 2.17 0.95 0.82 0.79 3.41 2.96 2.68 3.71 3.08 2.99 2.33 1.84 1.86 2.30 1.72 1.84

404 356 491 29 41 22 41.9 34.0 40.9

0.30 0.30 0.29 25.7 36.7 34.2

1.25 1.33 1.72 1.30 1.17 1.49 0.97 0.99 1.01 1.00 0.84 0.94

5611

10.8 237

276

5.4

13.26 10.1

3.08

1.12

4.16

5.0

3.11

3.00

520

20

51.1

0.30

23.5

1.20

1.16

0.96

0.95

3161 3730 6.3 6.7

255 215

155 186

3.64 2.82 9.27 8.86 7.1 7.7

2.20 2.77

0.85 0.97

2.96 3.53

3.46 4.22 2.25 2.74

2.11 2.63

502 557 20 20

42.6 66.0

0.31 0.36

35.9 27.9

1.15 0.72

1.12 0.86

1.02 0.95

0.96 0.99

4028

7.4

225

195

3.32

9.11

7.7

2.44

0.96

3.53

4.22

2.78

2.51

544

21

58.7

0.32

29.2

0.88

0.92

1.01

0.94

SiO 2 55.65 49.50 52.68 59.30 53.97 - 55.76

Lin et al.: Northern Mariana Shoshonites, 2 4501

TABLE 1. (continued)

Central Island Province

Ahyi Seamount (D) 20.40 ø

Element, ppm D15-3-2 D15-3-3 D16-3

Makhahnas

Seamount (D) 20.25 ø

D18-9 D18-11

NW Uracas

Uracas Seamount

(A) (E) 20.53 ø 20.58 ø

UR-23 D19-3-5

K

Rb

Sr

Ba

La

Ce

Nd

Sm

Eu

Gd

Dy Er

Yb

K/Rb K/Ba Ba/La Sm/Nd Sr/Nd

(La/Yb) n (Ce/Yb) n Eu/Eu* Ce/Ce*

SiO 2

5840 4917 11.2 9.4

299 254 232 215

4.52 3.5 11.17 9.13

8.0 7.1 2.34 2.02 0.86 0.78 3.12 2.86 3.57 3.46 2.23 2.16 2.17 2.17

521 523 25 23 51.6 61.4

0.29 0.28

37.4 35.8 1.39 1.08 1.39 1.07 0.98 0.96 0.98 0.96

3511

6.5

302

99 3.58 8.78

6.0

1.77

0.66

2.09

2.45

1.47

1.41

540

36

27.7

0.30

50.3

1.70

1.58

1.05

0.99

52.78 52.95 47.79

9257 10907 5902 3624 23.5 25.0 12.5 7.3

369 359 396 265 253 293 272 155

11.0 13.4 7.9 2.54 23.3 25.9 16.61 6.70 12.3 13.8 10.3 5.1

2.89 3.23 2.87 1.65 1.00 1.08 1.03 0.72 3.22 3.54 3.49 2.48 3.48 3.91 3.92 2.96 2.13 2.45 2.43 1.85 2.19 2.45 2.33 1.85

394 436 472 496 37 37 22 23 23.0 21.9 34.4 61.0

0.23 0.23 0.28 0.32 30.0 26.0 38.4 52.0

3.36 3.66 2.27 0.92 2.71 2.69 1.81 0.92 1.00 0.988 1.00 1.09 1.02 0.96 0.95 0.98

56.25 57.72 53.80 51.79

Central

Island

Province

NW Uracas

(œ) 20.58 ø

Element, ppm D19-3-10

Chamorro

Seamount

(œ) 20.84 ø

D23-3

Northern Seamount Province

Southern Diakoku Seamount (D) 21.02 ø

D25-1 D25-8 D25-18 D26-3-1 D26-4-5

K 3435 10784 Rb 6.3 21.4

Sr 265 307 Ba 160 447 La 2.78 7.7 Ce 7.06 16.30 Nd 5.4 8.6 Sm 1.74 2.09 Eu 0.68 0.74 Gd 2.35 2.38

Dy 2.74 2.66 Er 1.79 1.76 Yb 1.70 1.90

K/Rb 545 K/Ba 21 Ba/La 57.6 Sm/Nd 0.32 Sr/Nd 49.1

(La/Yb) n 1.09 (Ce/Yb) n 1.06 Eu/Eu* 1.03 Ce/Ce* 0.96

SiO 2 52.42

504 24

58.1 0.24

35.7

2.71

2.18

1.02

1.02

65.81

12088 5771 8523 7838 7930 23.5 10.9 17.0 15.3 15.8

363 296 304 371 362 388 242 332 357 348

10.4 5.2 8.0 7.9 7.9 23.9 13.59 19.2 18.7 18.8 16.4 10.2 12.9 13.0 12.8 4.54 2.90 3.40 3.77 3.72 1.45 1.02 1.17 1.26 1.24 5.51 3.61 4.15 4.50 4.54 5.9 4.08 5.2 5.1 5.1 3.73 2.64 3.20 3.26 3.31 3.55 2.60 3.23 3.00 3.25

514 529 501 512 502 31 24 26 22 23 37.3 46.5 41.5 45.2 43.9

0.28 0.28 0.26 0.29 0.29 22.1 29.0 23.6 28.5 28.3

1.96 1.34 1.66 1.76 1.63 1.71 1.33 1.51 1.59 1.47 0.89 0.97 0.96 0.94 0.93 0.96 0.98 0.99 0.97 0.98

58.19 54.33 56.53 61.79 60.03

4502 Lin et al.: Northern Mariana Shoshonites, 2

TABLE 1. (continued)

Element, ppm

S. Daikoku

Seamount (D) 21.02 ø

D26-4-6

Northern Seamount Province

Daikoku Seamount (D) 21.32 ø

D29-1-1 D29-1-2 D29-2-2 D29-2-3

Eifuku Seamount (D) 21.40 ø

D30-6 D30-7

K

Rb

Sr

Ba

La Ce Nd

Sm

Eu

Gd

Dy Er

Yb

K/Rb K/Ba Ba/La Sm/Nd Sr/Nd

(La/Yb) n (Ce/Yb) n Eu/Eu* Ce/Ce*

SiO 2

7882

15.7

355

364 8.4

19.9

13.5

3.79 1.27

4.24

5.3

3.40 3.32

15214 14490 16525 16314 8839 9476 38.8 38.5 42.6 42.3 21.1 22.3

374 370 379 382 653 665 530 526 560 564 372 421

22.3 22.2 24.2 23.9 17.5 ! 9.2 44.1 44.0 48.4 47.2 35.7 37.3

22.3 22.2 23.9 24.0 18.2 18.2 4.98 4.98 5.38 5.29 3.86 3.87 1.62 1.45 1.56 1.53 1.26 1.25

4.8 5.0 5.3 5.4 3.7 3.8 5.0 5.1 5.3 5.2 3.33 3.23

3.05 3.00 3.34 3.27 1.83 1.86 2.99 2.90 2.96 3.16 1.74 1.67

502 392 376 388 386

22 29 28 30 29 43.3 23.8 23.7 23.1 23.6

0.28 0.22 0.22 0.23 0.22

26.3 16.8 16.7 15.9 15.9 1.69 4.99 5.12 5.47 5.06 1.52 3.75 3.86 4.16 3.80

0.97 1.00 0.89 0.89 0.87 0.98 0.99 0.99 1.00 0.99

58.95 59.07 59.91 63.26 62.50

419 425

24 23 21.3 21.9

0.21 0.21

35.9 36.5 6.73 7.69

5.22 5.68 1.01 0.99

1.01 0.99

52.53 50.50

Element, ppm D30-8

Northern Seamount Province

Eifuku Seamount (D) 21.40 ø

D31-1-4 D31-1-6 D31-2-2

Kasuga Seamount (E)

21.77 ø

D33-3-1

Fukujin Seamount (A)

21.89 ø

D34-1-2 D34-2-2

K

Rb

Sr Ba

La

Ce Nd

Sm Eu

Gd

Dy Er

Yb

K/Rb K/Ba Ba/La Sm/Nd Sr/Nd

(La/Yb) n (Ce/Yb) n Eu/Eu* Ce/Ce*

10912

26.0 772

467

24.4

45.7

22.8

4.66

1.45

4.17

3.49

1.87

1.67

7672 7172 18.2 16.9

562 513

349 313 15.6 14.9

31.7 32.5 16.4 15.6

3.52 3.47

1.18 1.16

3.43 3.45

3.09 3.11 1.71 1.66

1.64 !.61

10630

25.2

706

448

21.4

42.7

21.5

4.50

1.33

4.12

3.79 2.22 1.90

420 422 424 422 23 22 23 24 !9.1 22.4 21.0 20.9

0.20 0.21 0.22 0.21 33.9 34.3 32.9 32.8

9.77 6.36 6.19 7.53 6.96 4.92 5.13 5.72 0.99 1.03 1.02 0.93 0.96 1.00 1.07 1.00

3371 6.3

241

237 3.51 8.86

6.1

1.90

0.68

2.58

3.10 1.95 1.95

535 14

67.5

0.31

39.5

1.20

1.16

0.94 1.01

8549 9801 23.6 27.0

482 451

360 358 13.9 15.7

28.5 30.7 15.1 15.8

3.53 3.71 1.27 1.25

3.63 3.95 3.58 3.86 2.14 2.26

2.04 2.16

362 363 24 27

25.9 22.8

0.23 0.23 31.9 28.5

4.56 4.86

3.55 3.61 1.08 1.00

1.00 0.98

SiO 2 51.53 49.63 49.05 52.54 56.77 52.69 53.11

Lin et al.: Northern Mariana Shoshonites, 2 4503

TABLE 1. (continued)

Northern Seamount Province

Fukuyama Soyo Ichiyo Fukujin Seamount Seamount Seamount

Seamount (A) (E) (E) (E) 21.89 ø 22.37 ø 22.47 ø 22.93 ø

Nikko

Seamount

(A) 23.09 ø

Element, ppm D35-1-2 D35-2-1 D38-2 D41-52 D44-1-5 D45-11 D46-1-1

K

Rb Sr

Ba

La

Ce Nd

Sm

Eu

Gd

Dy Er

Yb

K/Rb K/Ba Ba/La Sm/Nd Sr/Nd

(La/Yb) n (Ce/Yb) n Eu/Eu* Ce/Ce

SiO 2

9968 11234 2443 4019 28.3 30.2 4.5 8.4

394 405 266 309

527 555 168 268 16.4 18.3 2.28 4.43 32.4 35.5 6.14 10.04 18.0 18.4 4.5 7.1

4.11 4.26 1.43 2.19 1.49 1.29 0.55 0.78 4.22 4.70 1.95 2.97 4.41 4.7 2.37 3.55 2.55 2.78 1.47 2.20 2.51 2.68 1.43 2.14

10472

22.1

410

428 12.3

27.2

15.4 3.88

1.19

4.28 4.5

2.63

2.61

352 372 543 479 474 19 20 15 15 24 32.1 30.3 73.7 60.5 34.8

0.23 0.23 0.32 0.31 0.25 21.9 22.0 59.1 43.5 26.6

4.37 4.57 1.07 1.38 3.15 3.28 3.37 1.09 1.19 2.65 1.09 0.90 1.02 0.96 0.89 0.96 0.97 1.01 0.93 1.02

55.40 - 52.50

13123 3013 28.5 6.3

298 457 683 !65

23.1 6.1

48.1 13.86

26.6 8.0 6.4 2.03 1.76 0.83 6.6 2.38 7.3 2.57 4.9 1.52 4.77 1.43

461 478

19 18 29.6 27

0.24 0.25 11.2 57.1

3.24 2.85 2.56 2.50

0.82 1.16 0.99 1.03

53.21 61.7! 45.90

Element, ppm

Ko-Hiyoshi Seamount (D)

23.37 ø

D47-1-1

Northern Seamount Province

S. Hiyoshi Seamount (A) 23.51 ø

D48-1-2 D49-PB D49-1-2 D49-2-1 D49-3-2

C. Hiyoshi Seamount (D)

23.60 ø

D51-3

K

Rb

Sr

Ba

La

Ce Nd

Sm

Eu

Gd

Dy Er

Yb

15602 35.8

1063 791

5!

97.7 44.4

8.1

2.4

6.3 5.1

2.74

2.27

23426 16392 16969 28597 19403 69.5 47.2 48.7 86.5 56.6

491 666 610 507 589 833 642 667 990 767

47.8 33.9 34.8 57 45.0 90.0 69.0 61.4 !07.5 86.1 40.2 32.! 28.7 47.2 34.9

7.9 6.2 5.60 8.7 6.6

2.22 1.74 1.59 2.25 1.73 6.9 5.5 5.0 7.7 5.9

6.4 4.9 4.6 6.8 5.1 3.79 2.85 2.68 3.98 2.82 3.75 2.68 2.30 3.87 2.72

K/Rb 436 337 347 348 331 343 K/Ba 20 28 26 25 29 25 Ba/La 16 17.4 18.9 19.2 17 17.0 Sm/Nd 0.18 0.20 0.19 0.20 0.18 0.19 Sr/Nd 23.9 12.2 20.7 21.3 10.7 16.9

(La/Yb) n 15.02 8.52 8.46 10.12 9.85 11.06 (Ce/Yb) n 10.95 6.10 6.55 6.79 7.06 8.05 Eu/Eu* 0.99 0.90 0.90 0.90 0.83 0.83 Ce/Ce* 1.00 1.00 1.04 0.94 1.00 1.04

11236

23.0

932 649

32.2

63.0

30.4

6.1

! .86 5.5

4.5 2.4!

2.24

489 17

20.2 0.20

30.7 9.61

7.15

0.97

1.00

SiO 2 53.06 53.10 50.11 50.18 54.94 46.93 51.84

4504 Lin et al.: Northern Mariana Shoshonites, 2

TABLE 1. (continued)

Element, ppm

Northern Seamount Province

Central Hiyoshi Seamount (D) 23.60 ø

D51-g(a) D51-g(b) D52-1-1 D52-3-1

North Hiyoshi Seamount (D) 23.79 ø

D53-1-2 D53-1-3 D54-1-1

K

Rb

Sr

Ba

La

Ce Nd Sm

Eu Gd

Dy Er

Yb

K/Rb K/Ba Ba/La Sm/Nd Sr/Nd

(La/Yb) n (Ce/Yb) n Eu/Eu* Ce/Ce*

SiO 2

32104 36562 21119 18607 82.0 85.2 65.1 54.4

378 369 1015 873 1087 1107 821 745

64 64 42.5 38.3 125 128.4 77.6 70.0

57.9 59.5 34.7 30.6 10.6 11.0 6.5 5.67

2.5 2.6 2.02 1.80 7.8 7.5 5.3 4.9

6.6 6.8 4.2 3.96 3.83 3.90 2.18 2.12 3.79 3.87 2.03 2.05

392 429 324 342 30 33 26 25 17 17 19.3 19.5

0.18 0.18 0.19 0.19 6.5 6.2 29.3 28.5

11.30 11.06 14.00 12.49 8.39 8.44 9.72 8.68 0.81 0.83 1.02 1.02 1.01 1.03 0.97 0.98

34400 34948 11675

107.9 106.4 27.3 920 916 1047

816 832 707 59 54 31.8

103.2 117.9 63.1

41.5 41.8 31.0 7.3 7.2 6.2

2.00 1.99 1.95 5.7 5.9 5.2

4.5 4.4 4.03 2.38 2.33 2.01 2.37 2.31 1.77

319 329 428

42 42 17

14 15 22.2

0.18 0.17 0.20 22.2 21.9 33.8

16.65 15.63 12.01 11.07 12.98 9.07

0.92 0.91 1.03 0.97 1.18 1.00

50.25 51.86 55.80 55.39 46.16

Element, ppm

Northern

Seamount Volcano Arc

Province

Fukutoku

Seamount (E) Iwo Jima Volcano (A) 24.04 ø 24.78 ø

S. Kita

Iwo Jima (E) 25.10 ø

D55-1-2 D57-6 Trachyandesite 5-2 7-2 Suribati D73-4-2

K

Rb

Sr

Ba

La

Ce Nd Sm

Eu

Gd

Dy Er

Yb

K/Rb K/Ba Ba/La Sm/Nd Sr/Nd

(La/Yb) n (Ce/Yb) n Eu/Eu* Ce/Ce*

21177 10042 35510 29200 32150 34480 10369 78.9 18.0 66.4 68.2 69.2 61.9 18.8

848 796 444 800 561 459 751

601 588 1123 1044 1029 1133 363

27.5 23.7 56 54 59 56 16.4

55.2 47.6 123.7 113.2 124.9 103.9 31.9 26.2 22.0 60.8 55.0 63.0 56.2 17.2

5.21 4.35 11.2 10.3 11.5 10.2 3.91 1.67 1.32 2.8 2.7 2.7 2.7 1.32

4.71 3.82 6.9 8.2 8.6 8.0 3.70 3.60 3.31 6.9 7.2 6.9 6.5 3.4

1.88 1.80 4.06 3.80 3.85 3.58 1.98

1.70 1.64 4.08 3.90 3.78 3.37 1.85

268 558 535 428 465 557 552 35 17 32 28 31 30 29 21.9 24.8 20 19 17 20 22.1

0.20 0.20 0.18 0.19 0.18 0.18 0.23 32.4 36.2 7.3 14.5 8.9 8.2 43.7 10.82 9.66 9.2 9.3 10.4 11.1 5.93

8.26 7.39 7.71 7.38 8.41 7.84 4.38 1.02 0.97 0.91 0.87 0.80 0.89 1.05 1.02 1.03 1.07 1.04 1.03 0.93 0.96

SiO 2 48.83 48.04 59.6 56.2 - - 53.67

Lin et al.: Northern Mariana Shoshonites, 2 4505

TABLE 1. (continued)

Element, ppm

S. Kita

Iwo Jima (E) 25.10 ø

D73-4-4

Volcano Arc

Kaitoku (A) Nishino Shima (A) 26.07 ø 27.20 ø

D75-4 D77-1 D79-1 D79-6 D80-1 1973/74 flow

K 3878

Rb 5.9

Sr 614 Ba 170

La 5.6

Ce 12.83

Nd 8.5

Sm 2.22

Eu 0.83

Gd 2.36

Dy 2.42 Er 1.38

Yb 1.29

9327 8888 3355 3473 3473 9656 16.5 15.5 6.8 5.7 6.5 20.1

254 256 174 177 166 228 277 258 122 124 139 268

7.5 7.0 1.71 1.69 2.08 8.4 18.3 16.8 5.60 6.67 6.58 21.5 15.1 14.2 5.43 5.77 6.25 15.7

4.50 4.32 1.92 2.00 2.20 4.64 1.44 1.34 0.71 0.75 0.78 1.46 5.8 5.2 2.85 2.88 3.14 5.4 6.3 6.0 3.47 3.52 4.07 6.3 3.90 3.84 2.14 2.21 2.52 3.82 3.61 3.51 1.75 2.21 2.46 3.79

K/Rb 657 565 573 493 609 534 480 K/Ba 23 34 34 27 28 25 36 Ba/La 30.4 36.9 37.0 71.3 73.4 66.8 31.9 Sm/Nd 0.26 0.30 0.30 0.35 0.35 0.35 0.30 Sr/Nd 72.2 16.8 18.0 32.0 30.7 26.6 14.5 (La/Yb) n 2.90 1.39 1.33 0.65 0.51 0.57 1.48 (Ce/Yb)n 2.53 1.29 1.22 0.81 0.77 0.68 1.44 Eu/Eu* 1.10 0.87 0.87 0.94 0.96 0.91 0.89 Ce/Ce* 0.97 0.90 0.89 0.94 1.08 0.94 0.99

SiO 2 47.39 57.64 55.55 53.23 51.21 52.02 58.50

Codes in parentheses correspond to state of activity of volcano or seamount: A, active; D, dormant; and E, extinct.

studied portion of the CIP has a mean K/Rb of 469+84, similar to the mean of 493 determined for subaerial

Mariana volcanoes by Chow et al., [1980]. This is only slightly higher than the mean K/Rb of 447+61 observed for southern NSP volcanoes and 406+84 for the northern NSP and Iwo Jima. A somewhat higher K/Rb (558+58) is observed for the Volcano arc north of Iwo Jima.

K/Ba variations along the arc are also subdued. The studied portion of the CIP has a mean K/Ba (31+9) identical to that calculated for subaerial Mariana volcanoes

(31.2 [Chow et al., 1980]) and indistinguishable from that of the Volcano arc north of Iwo Jima (30+5). The southern NSP has a lower mean K/Ba (23+4), while the northern NSP and Iwo Jima have mean K/Ba (28+7) indistinguishable from that of the CIP or the Volcano arc north of Iwo Jima. A mild, long wavelength variation is suggested, with high K/Ba observed around Agrigan in the CIP and again in the Volcano arc, with slightly lower K/Ba in the NSP (Figure 8).

Marked variations in Ba/La are observed along the arc system. In the studied portion of the CIP, mean Ba/La of 37+13 is observed. With the CIP, there is a suggestion of a long wavelength variation, with a minimum of about 24 observed for Agrigan and Poyo and a regularly increasing ratio as far north as Supply Reef (Figure 8). The marked decrease of Ba/La at Makhahnas seamount may be related to its position slightly to the rear of the volcanic front. Mean Ba/La of 35+15 is calculated for the southern NSP, but extreme variations are noted. For example, Eifuku and Daikoku have Ba/La of 19 to 24 but are flanked by

volcanoes with Ba/La of 37 to 45 (south Daikoku) and 68 (Kasuga). A similar situation is found for Fukujin with Ba/La of 23-32 flanked by Kasuga and Fukuyama seamounts (Ba/La = 74). We note that at least within the southern NSP, the active and dormant edifices have lower Ba/La compared with the extinct edifices (Figure 8). In contrast to the heterogeneity of the southern NSP, Ba/La for the northern NSP and Iwo Jima is relatively constant and low (19+3). Ba/La for the Volcano arc north of Iwo Jima averages 46+21, but with a monotonic increase from 22-31 at south Kita Iwo Jima to 32-73 at Nishino Shima

(Figure 8). Strictly speaking, Sr/Nd is not an incompatible

element ratio because low-pressure fractionation, dominated by plagioclase removal, affects Sr concentration to a much greater extent than Nd. Thus fractionated lavas will have a lower Sr/Nd than mafic precursors. Sr and Nd do behave similarly, as incompatible elements, during partial melting of eclogite or spinel-or garnet-lherzolite. Thus Sr/Nd ratios for the least fractionated lavas are good indications of the source region Sr/Nd. Mean Sr/Nd for the studied portion of the CIP is 34+11, similar to that of the Volcano arc north of Iwo Jim (32+19). Mean Sr/Nd for the southern NSP is similar as well, 30+11, while mean Sr/Nd is significantly lower in the northern NSP, 18+10. It is worth noting that the lavas of the northern NSP with low Sr/Nd are dominated by relatively unfractionated mafic lavas [Bloomer et al., this issue], so that the low Sr/Nd for this region may reflect closely source region compositions.

4506 Lin et al.: Northern Mariana Shoshonites, 2

100 CHEREF SMT. "--........

/ g,%--: DII (I 17-2.79) ill i • i '1 i / i t I I I I I

ioof POYO SMT. D8 -- (o 97- i oo)

ß ASU.•?N (A)

""•ilCHEREF SMT. • (E)

.•POYO SMT. ..-' (E)

'-xk._,• , '

// //

'""""""'-., '"., •3• • "'-.

i00I ASUNCION ,o

.... I (i oo-i 81) lOOt AGRIGAN

,/ .........

iOOf NORTH PAGAN

18øN 146øE

Io0[ SOUTH PAGAN ,/ .........

Fig. 2. Dredge locations and REE patterns, CIP. Location of this figure is shown in Figure 1. Chondrite-normalized REE patterns for Pagan, Agrigan, and Asuncion are from Dixon and Batiza [1979], Hole et al. [1984], and White and Patchett [1984]. Data in parentheses in lower right of each REE plot are the range of (Ce/Yb) n for each edifice sampled. Letters in parentheses correspond to state of activity of the sampled edifice (A, active; D, dormant; and E, extinct), as inferred by Bloomer et al. [1988]. Contours are in thousands of fathoms (or in multiples of 1829 m). Lines of latitude and longitude are at 30' intervals.

The chondrite-normalized ratio of Ce and Yb, denoted

(Ce/Yb) n, is also not strictly speaking an incompatible element ratio. This ratio may reflect the importance of residual (and cumulate?) garnet and, to a lesser degree, clinopyroxene and hornblende, in the fractionation or partial fusion history of melts. An understanding of the role of especially garnet is critical for understanding the evolution of arc melts [e.g., Apted, 1981; Brophy and Marsh, 1986], and for this reason, the variation of this ratio along the arc is considered to be important (Figure 8). This ratio will not change appreciably during the low- pressure, amphibole-free fractional crystallization experienced by Mariana arc melts [Baker, 1987]. For example, in the fractionation of 51% SiO 2 basaltic melts to yield 60% SiO 2 andesitc on Agrigan, (Ce/Yb) n actually decreased from 2.1 to 1.9 [Dixon and Batiza, 1979]. Therefore the (Ce/Yb) n of Mariana and Volcano arc melts are good approximations to those of parental melts. As noted by earlier workers, REE patterns for the subaerial Mariana arc are approximately flat or are very slightly LREE enriched or depleted [Dixon and Batiza, 1979; Hole et al., 1984; White and Patchett, 1984]. Our results for the portion of the CIP studied here substantiate these conclusions, with mean (Ce/Yb) --1 5+0 5 This is indistinguishable from that of the Volcano arc north of Iwo Jima, with a mean of 1.6+1.3. A trend of decreasing LREE enrichment towards the north is responsible for the large standard deviation of the Volcano arc. A somewhat higher (Ce/Yb) n is observed for the southern NSP (3.2+1.7), with active and dormant edifices generally erupting more LREE-enriched lavas than extinct edifices

(Figure 8). Strong LREE enrichments are observed for the lavas of the northern NSP and Iwo Jima, with a mean

(Ce/Yb) n of 8.4+1.7. Complete REE patterns for volcanic rocks from the

study area are shown in Figures 2-5. In addition to the LREE enrichments noted above, several other aspects are noteworthy. Europium anomalies (Eu/Eu* in Table 1) are observed in the lavas and may either be positive or negative. Basalts and basaltic andesites from Poyo seamount (DS) and south Kita Iwo Jima (D73) manifest a 5-10% positive anomaly, as do some samples from Ahyi, NW Uracas, Fukujin, and Nikko. Some portion of this effect may be due to plagioclase accumulation; samples with Eu/Eu*=l.05-1.1 contains 12-20% plagioclase phenocrysts. Samples DS-1 and 3 contain only 12-15% plagioclase phenocrysts and have Eu/Eu*=l.10-1.11. These plagioclases may be in equilibrium with the melt and, even if not, cannot contain the excess Eu necessary to generate the positive Eu anomaly of these rocks. Slight negative curopium anomalies are common, particularly in the andesitic rocks of the southern NSP and the Volcano arc

but also in the mafic lavas from the Hiyoshi seamounts. Low-pressure fractional crystallization involving plagioclase is responsible for the negative Eu anomalies of the andesitic lavas, but a source effect is probably responsible for the negative curopium anomaly of the Hiyoshi volcanoes.

Negative cerium anomalies have been documented from subaerial Mariana volcanoes [Dixon and Batiza, 1979; Hole et al., 1984; White and Patchett, 1984], and are also observed for some of the samples analyzed here. Strong

Fig. 3. Dredge locations and REE patterns, northern CIP and southernmost NSP. Location of this figure is shown in Figure 1. Letters in parentheses correspond to state of activity of the sampled edifice (A, active; D, dormant; and E, extinct), as inferred by Bloomer et al. [1988]. Data in parentheses in lower right of each REE plot are the range

of (Ce/Yb)n_ for each edifice sampled. Contours are in thousands of fathoms (or in multiples of 1829 m). Lines of latitude and longitude are at 30' intervals.

Lin et al.: Northern Mariana Shoshonites, 2 4507

D44 -- [ D41 •

•ICHIYO SMT. ,,,'/ " iii III i II / ?// IOI-

/• l/ • ii l l i I l l l lll

• • / ', ,,' SOYOSMT' ( • ,,

SMT. FUKUYAMA . \•7•\ \

(E) 143øE

lOOt ..., FUKUJIN SMT. I FUKUJIN ,o '""" L ....... /

FUKUYAMA SMT.

22øN I I I

,.•½LKASUGA ()SMT. '"" (E)

144øE

Fig. 4. Dredge locations and REE patterns, southern NSP. Location of this figure is shown in Figure 1. Letters in parentheses correspond to state of activity of the sampled edifice (A, active; D, dormant; and E, extinct), as inferred by Bloomer et al. [1988]. Data in parentheses in lower right of each REE plot are the range of (Ce/Yb) n for each edifice sampled. Contours are in thousands of fathoms (or in multiples of 1829 m). Lines of latitude and longitude are at 30' intervals.

anomalies (Ce/Ce*<0.9) are found for samples from Poyo, Cheref, Iwo Jima, and Kaitoku. Samples with more modest cerium anomalies (Ce/Ce*=0.9-0.95) were found on Cheref, Supply Reef, Uracas, Fukujin, Soyo, south Hiyoshi, and Nishino Shima. Thus, to a first approximation, negative cerium anomalies are especially common in the CIP and Volcano arc and appear to be less common in the NSP. These are generally less negative than those observed for CIP islands (Agrigan 0.81-0.92; Asuncion 0.84-0.88; and Pagan 0.91-0.92 [Hole et al., 1984; White and Patcherr, 1984]).

Discussion

The LIL and REE data presented here, along with the major and other trace element data presented in the companion paper [Bloomer et al., this issue] indicate that the submarine volcanoes of the Mariana NSP and Iwo Jima comprise a distinctive province of LIL- and LREE- enriched magmatism. The rocks of the NSP have strong shoshonitic affinities as defined by Morrison [1980], especially those of the northern NSP and Iwo Jima. Rocks of the southern NSP consist of medium-K calc-alkaline basalts and andesires, as defined by Gill [1981]. These lavas are distinctly enriched in LIL and LREE elements relative to those along the magmatic front of the Mariana

CIP and Volcano arc as well as compared with volcanic rocks from other intraoceanic arcs at a similar stage in their evolution such as the Tonga-Kermadec arc JEwart et al., 1973; Ewart and Hawkesworth, 1987], the Izu arc [Masuda and Aoki, 1978], and the Aleutian are [Kay et al., 1982]. Shoshonitic lavas, while common in evolved Andean-type convergent margins, are especially uncommon in intraoceanic arcs and are rarely observed along the magmatic front [DeLong et al., 1975].

Recognition of an extensive province of shoshonitic magmatism in the northern NSP and Iwo Jima is important for three main reasons, all of which are important potential lines of investigation. First, since the shoshonitic volcanoes are largely submarine, relatively small, and presumably young, we may be seeing the initial stages in submarine are volcanism, a stage heretofore inferred to be dominated by depleted tholeiites [Jakes and White, 1972]. Second, since the location of the shoshonitic province is very close to the proposed location of a propagating rift tip, we may have the opportunity here to investigate the relationship between back arc basin rift propagation and the composition of melts in the are. Finally, since understanding the cause of LIL and LREE enrichments in the source region of are melts is widely acknowledged to

D75

D77 -- (i 22-129)

iOOf S KITA IWO JIMA (2 53-4 38)

I--

Ioo[ FU KUTOKU SMT I0 f (7,39- 8 26)

D51-- D52----

(7 15-972)

Iø f D48• D49-- --

(6 10-706)

I00• NISHINO SHIMA

...... ,_ot ' / DBO----

,I 1,9•3,F•O•WT 7, I0,6•-: 14),

KAITOKU ioof" (•)

13 Iø f ( 7 38- 8 41) ß KITA IWO JIMA ,'

• (E) /

/////// I00

O JIMA Iø f (A)

(9 07-12 98)

i/• i , , , i i J , • i , J 1

NAMI !,•O JIMA(E) FUKU '•OKU SMT.(E)

H HIYOSHI SMT.(D)

HIYOSHI KNOLL(D) SOUTH HIYOSHI SMT. (A)

SMT. (D)

•NIKKO SMT. iOE /' 142OE

iOOf NIKKO SMT (2 50-2 56)

Fig. 5. Dredge locations and REE patterns, northern NSP and Volcano are. Location of this figure is shown in Figure 1. Letters in parentheses correspond to state of activity of the sampled edifice (A, active; D, dormant; and E, extinct), as inferred by Bloomer et al. [1988]. Some data for Iwo Jima are from Stern et al. [1984]. Data in parentheses in lower right of each REE plot are the range of (Ce/Yb) n for each edifice sampled. Contours are in thousands of fathoms (or in multiples of 1829 m). Lines of latitude and longitude are at 1 ø intervals.

4508 Lin et al.: Northern Mariana Shoshonites, 2

K

(ppm)

Sr

(ppm)

BQ

(ppm)

LQ

(ppm)

Yb

(ppm)

36000

24000

12000

0

960

480

0

1200

600

o

64

32

< cIP

6100 4- 3300

N -NSP i S - NSP •< + •T • v A

IWO JIMA I (! (•)26%00 (?'-'I

/

4244- 139 288 4- 48

(12) (i)

--- -. • (4) .• ('f__ T (6) (,, (• 1-

328 __. 225 (2)

•(a) (4)

196 -I- 66

o (?) 2.3 -I- 0.6

4 PA UR [ • I (4) /// • ___ (2 ,,1',

L AG •) AS j (2)

01 18 20

399 4- 127

(2)

(4) (4) (

(5) 856 -- 200

(i)T L (3)

46.6

215_+89

(2) (4)

6.34-4.9

'"'NS

\(2) (• (2) (4)

5.6 -I- 3.1 14.0 -t- 7.2

(4) (2) ,I,,,j, (4)

2.5'1' 0.8 2 9 + 0.9 (5) ' -- (3)

(6L (4) """•t• (4) I ,,'F ,,,I

22 24

(4) (2)

/

I• / / 2.6+ 1.0

// i i i

26 28

LATITUDE(øN)

Fig. 6. Variations in K, Sr, Ba, La, and Yb along the Mariana and Volcano arcs. Numbers in parentheses indicate the number of analyses involved in determining the mean (solid dot) and range (error bars) for each volcano. Large circles denote extinct edifices. Means +1 standard deviation are shown for the studied portion of the CIP, southern NSP (S-NSP), northern NSP (N-NSP) plus Iwo Jima and Volcano arc north of Iwo Jima (VA). Location of major volcanic islands are also shown: PA,Pagan; AG,Agrigan; AS,Asuncion; UR,Uracas; IJ,Iwo Jima; and NS,Nishino Shima. Data sources in addition to those presented here are Stern [1979], Dixon and Batiza [1979], Chow et al. [1980], Yuasa and Tamaki [1982], Notsu et al. [1983], Hole et al. [1984], and White and Patchett [1984].

be a key for deciphering convergent margin melt generation [Perfit et al., 1980; Morris and Hart, 1983; Arculus and Johnson, 1981], the unusually large enrichments observed in the shoshonitic rocks of the

northern NSP and Iwo Jima give us a new opportunity to examine the source of these enrichments. The first two of

the above potential lines of investigation are examined

elsewhere [Stern et al., 1984, 1988], and we focus here on the third prospect.

In the following discussion, we attempt to further elucidate the nature of LIL and LREE enrichments in the

source region of arc melts, with an emphasis on the significance of the NSP-Iwo Jima shoshonites. The Mariana-Volcano arc system is a particularly suitable one

Lin et al.: Northern Mariana Shoshonites, 2 4509

CIP Islands, Pagan and North Cheref N•kko Other Submarine Mar•ana Volcanoes and Iwo Jima

Volcano Arc, North of Iwo J•ma AG R I G A•N. F_R ACT I 0 N.•AT I O.__.•N Observed (D•xon & Bat•za, 1979)

---- Frachonol Crystalhzot•on

0 45 615

Fig. 7. Ce-SiO 2 variation diagram for volcanic rocks from the Mariana-Volcano are study area. Fractionation paths observed for samples from Agrigan are plotted [Dixon and Batiza, 1979], along with a model trajectory calculated using an assumed parent (45% SiO 2, 10 ppm Ce), the major element fractionation model of Stern [1979], and distribution coefficients of Irving [1978]. Note the relatively modest increase of Ce during fractional crystallization. Also shown are inferred fractionation paths for lavas from the submarine volcanoes of Cheref and

Nikko. In contrast, the very high Ce contents of Eifuku and especially Fukutoku, the Hiyoshi complex, and Iwo Jima cannot be related by low-pressure fractional crystallization of CIP-type parental magmas.

in which to conduct this examination, not only because it is built on relatively refractory oceanic crust, but also because there exists a consensus that the melts of this arc are generated by melting in the mantle wedge. The principal disagreement concerns how large a contribution is made by the subducted slab, with some investigators interpreting the data to indicate 1% or less [Meijer, 1976; Stern and Ito, 1983; Ito and Stern, 1986] and others seeing 1-2% or so [White and Patchett, 1984; Hole eta!., 1984]. Woodhead and Fraser [1985] argue that up to 70% of the lead in subaerial Mariana arc volcanoes may be derived from subducted sediments. We note, however, that this conclusion is based on consideration of mid-ocean ridge basalt (MORB)-type sources as the only potential mantle reservoir. If the much larger range in Pb-isotopic composition represented by ocean island basalt (OIB)-type sources is considered, no sedimentary contribution is required. Despite these points of disagreement, a consensus exists that Marlaria arc magmagenesis overwhelmingly manifests mantle-derived melts that were later subjected to varying degrees of low-pressure fractional crystallization within crustal magma chambers.

The next question is at what depth melting occurs in the mantle. The unique thermal structure of the subarc mantle is an important constraint. The downgoing lithosphere is cold and is responsible for an inverted thermal gradient in the lower part of the mantle wedge. This is due not only to thermal inertia of the cold slab but also to endothermic dehydration reactions [Anderson et al., 1978]. Low stress in the Wadati-Benioff Zone below the Marianas indicates negligible frictional heating resulting in a subduction zone geotherm that is "...probably the coldest yet proposed for this planet" [Bird, 1978]. Melting of the subducted slab beneath the Marianas is thus theoretically impossible. Convective overturn of the mantle wedge is

required for melting of even damp peridotitc, and this will occur in the portion of the asthenosphere defined by the base of the subarc lithosphere and the downgoing slab [Anderson et al., 1980].

The conclusion that Mariana-Volcano arc melts are

generated in the upper portion of the subarc asthenosphere is of limited utility due to uncertainties about the thickness of the lithosphere beneath the Mariana arc. This lithosphere should nevertheless have a very wide range in thickness, from the entire mantle wedge (50-100 km) beneath the cold forearc and frontal arc region to very thin (6-10 km) beneath the active spreading center of the Mariana Trough [Bibee et al., 1980]. The Mariana active arc is located just to the back arc side of the fundamental break between back arc and frontal arc crust, so a lithospheric thickness appropriate to 0-5 Ma oceanic crust thickened by arc processes is likely. On this basis, we estimate a lithospheric thickness of 35 km for the arc volcanoes at 18øN built on 5 Ma back arc crust [Parker and Oldenburgh, 1973; Kay and Kay, 1985]. Mantle at this depth would be near the pressure of the plagioclase- to spinel-peridotite transition [Jaques and Green, 1980], so that arc melts in the CIP should not generally be generated by melting of plagioclase peridotitc. The pressure above which spinel peridotitc transforms into garnet peridotitc has a very strong temperature dependence, but in the synthetic system, CaO-MgO-A1203-SiO 2 occurs on the solidus at 22 kbar [Gasparik, 1984], corresponding to a depth of 70-75 km. This depth estimate may be a minimum because MacGregor [1970] noted that increasil•g proportions of trivalent cations such as Cr, A1, and Fe +• increase the pressure stability of spinel peridotitc at the expense of garnet peridotitc.

Partitioning of the REE between melt and mantle is especially sensitive to the presence or absence of garnet in the residue and so may be used to constrain the depth and degree of fusion in the mantle wedge. Forward modeling studies of REE patterns expected from partial melting of garnet- and spinel-lherzolite were thus undertaken and the results shown in Figures 9 and 10. Low degrees of partial melting of garnet peridotitc result in liquids with strong LREE enrichment and a progressive decrease of the heavy REE (HREE). Low degrees of melting of spinel peridotitc also show strong LREE enrichment but differ from garnet peridotitc melts in having much flatter REE patterns from Dy to Yb (Figure 9a). Comparison of Mariana-Volcano REE patterns with the modeling results (Figure 9b) shows that while lavas from the northern NSP are greatly enriched in LREE relative to those of the CIP, they have identical and flat HREE patterns over the diagnostic range Dy to Yb.

The analytical data can be compared with the modeling studies in a slightly different manner, as shown in Figure 10. This plots (La/Yb) n versus chondrite- normalized La(Lan) and emphasizes the very different trajectories expected for melting of garnet- and spinel- lherzolite. Clearly, the origin of melts following high degrees (>15%) of melting cannot be distinguished. Furthermore, the effects of extended fractional crystallization cannot easily be distinguished on the diagram from decreasing degrees of melting spinel lherzolite. This complicates interpretation of the diagram but is a relatively minor effect. For example, the most fractionated sample obtained from the CIP islands is an Agrigan andesitc that has La -50 and (La/Yb) -2.5; this n n andesitc doubtless results from 75% fractional

crystallization of mafic melts [Stern, 1979]. Such extensive fractionation is uncommon in the CIP, and even including this sample, the range of CIP melts is limited, generally indicating 10-20% melting of spinel lherzolite. Another example, from Nikko seamount (fractionation of D46-1-1 to D45-11), shows strong fractionation and is consistent

•4510 Lin et al' Northern Mariana Shoshonites, 2

600

K/Rb 3OO

K/Ba

Ba/La

Sr/Nd

o

60

, N-NSP S- NSP >1<•--- + --•

I IWOJIMA • I

1 (i;) )

v A

(2) j (2)

469+ 84 J (4) (4) ([)'"' I J - - I 447 + - 61 406+84 I

• I i

284' 7 J 30 +- 5

(Ce/Yb) n

8O

:4) (8) -'"

....

(4)

6I PA AG AS L I / ,• (8) / •. ...... •.__ (7) (

ol 18 2O

(3) (5)'I' (•) (6)• (2) (4)

._

I o

) (•) 19+_ ' $ I 46"- 21 ,(6) / • / ' ' .... ,, • 11 ", (I) I (.• ..... ' ....

I

•4 • ii J 5o • ii J 18 ß lo 52 • 19

30 •" (•)" "(')(,)(,).,,(,),• (,).•.).,.'• • (,,•)• .... o I I

I

12 •.5• o.5 $.2• •.7 (•, 8.4• •.7 I •.6• i.$ / • / •' z (4)

, ,,,• '• I (6)

• • /

I

i i i

(4)

"...%; ............. •

NS

(2) (4)

.......

22 24 26

i

LATITUDE (øN)

Fig. 8. Incompatible element ratios plotted as a function along the Mariana-Volcano arc system. Means +1 standard deviation, symbols, and data sources are in Figure 6.

with the expected trend of fractional crystallization. Similar conclusions are obtained for the Volcano arc

samples.

The (La/Yb) n versus La., behavior of much of the NSP and Iwo Jima is quite different from that of the CIP and VA. Analyses of Fukujin, Eifuku, Daikoku, Fukutoku, the Hiyoshi complex, and Iwo Jima define a trend parallel to the CIP; this trend could be interpreted as manifesting low degrees of melting of spinel lherzolite. An alternate explanation, that the NSP-Iwo Jima melts manifest fractional crystallization of parental melts generated upon melting garnet lherzolite, is not preferred because of the HREE arguments developed for Figure 9. Another explanation is that the NSP-Iwo Jima melts are generated by large degrees of melting of a strongly LREE- enriched source. This is consistent with the observation that NSP-Iwo Jima samples define a diffuse trend that apparently extends back toward a source that is distinctly more enriched in LREE than that inferred to have produced CIP or VA melts.

Another argument can be used to evaluate the degree of partial melting required to generate the NSP-Iwo Jima shoshonites. Although it appears from Figure 10 that these represent significantly lower degrees of partial melting than the CIP (<!% versus 10-20%), this conclusion is difficult to reconcile with the relatively similar abundances of HFSC such as TiO 2 [Bloomer et al., this issue] and Yb in NSP and CIP lavas. Recent experimental results indicate that Ti-bearing accessory phases are unlikely to be stable during conditions of melting in the mantle wedge [Green and Pearson, 1986; Ryerson and Watson, 1987]. If NSP lavas represent an order-of-magnitude lower degree of partial melting of LREE-enriched spinel lherzolite, TiO 2 and Yb should be greatly elevated over CIP abundances, although these concentrations may be buffered to some extent by low-pressure fractionation of clinopyroxene and titanomagnetite. The fact that these concentrations are similar in least fractionated samples supports a conclusion that similar degrees of partial melting are involved [Bloomer et al., this issue]. Similar

Lin et al.: Northern Mariana Shoshonites, 2 qS1 1

ioo

GARNET LHERZOUTE SPINE L•LHE RZ•O LI•TE

[ 01 '1 .509 I-- 18 ) I En 217 j 4i I

-:"•::'• HIYOSHI, FUKUTOKU, IWO alMA F•I SPINEL LHERZOLITE MELTING •l• GARNET LHERZOLITE MELTING

B

Lo Ce Nd Sm Eu Gd Dy Er Yb •o C•e •d Sg E'u •d I•y I•r Y•b

Fig. 9. (a) Models of batch melting of garnet lherzolite (solid line) and spinel lherzolite (dashed line). Mineralogy (x) of garnet lherzolite is that of PHN 1161 [Nixon and Boyd, 1973]; melt fractions (P) are from Harrison [1979]. Note that P is negative for olivine because the incongruent melting of orthopyroxene yields olivine plus liquid. Mineralogy of spinel lherzolite is after Carter [1970]; melt fractions are those of D.C. Presnall (personal communication, 1987). Numbers on modeled REE patterns refer to percent melting of garnet lherzolite (left) or spinel lherzolite (right). A mantle REE composition of 3x chondritic is that of Kay and Gast [1973]. Note that melting of garnet lherzolite exhausts garnet at 17.5% melting; in the case of spinel lherzolite, cpx is exhausted before spinel, at 34% melting. Distribution coefficients are taken from Irving [1978], Nicholls and Harris [1980], and Lambert and Simmons [1987]. (b) Range of modeled REE patterns compared with the observed range of REE patterns from the Mariana CIP and Mariana NSP plus Iwo Jima.

LIL and LREE enrichments in some OIB have been argued to possibly result from relatively high degrees of partial melting of LIL- and LREE-enriched sources [Sun and Hanson, 1975; Clague and Frey, 1982].

Further constraints on the nature of Mariana-Volcano

arc source regions can be obtained from examining Ba-La- Yb systematics. Figure 11 plots Ba/La versus (La/Yb)n; the data fall along a hyperbolic curve. Due to exceedingly low partition coefficients for both elements, Ba/La should not vary significantly during differing degrees of fractional crystallization of mafic melts or partial melting of either spinel- or garnet-lherzolite. (La/Yb) n may vary with differing degrees of fractionation or partial melting, so crystal-liquid equilibrium during or after melting of a homogeneous source can be responsible only for the horizontal part of the trend. Metasomatism of mantle periodotite by Ba-rich, REE-poor fluids will result in source modifications that can generate melts with high Ba/La but not high (La/Yb) n, and so can only be responsible for the vertical part of the trend. Source metasomatism by fluids with Ba>La>Yb such as that observed in the experiments of Tatsumi et al. [1986] will lead to a diagonal trend normal to that observed.

The simplest explanation for the data distribution is that it represents mixing between two end-members (mantle sources and/or melts), as shown in Figure 11. End-member 1 has a very high Ba/La (-80) and low (La/Yb) n (-0.5); these are the signatures of depleted mantle (or derivative melts) that has been metasomatized by Ba-rich, REE-poor fluids. End-member 2 has a low Ba/La (*.15) and high (La/Yb) n (-20); these are similar to that found in western Pacific "hot spot" or OIB such as the

Caroline Islands. End-member 1 dominates in the source

region of the CIP and the Volcano arc, while end-member 2 dominates the source of NSP-Iwo Jima shoshonites.

End-member 2 is most simply identified as an OIB-type mantle (or derivative melts), similar to that interpreted for some arc sources on the basis of K/Rb, K/Ba, and isotopic arguments [Stern and Itc, 1983; Morris and Hart, 1983; Itc and Stern, 1986]. These arguments have suffered because most arc lavas have high ratios of LIL and HFSC, including high Ba/La, and thus are dominated by what we identify here as end-member 1. The recognition of OIB- type Ba/La in the shoshonitic lavas of the NSP and Iwo Jima is a strong argument that OIB-type sources do participate in the generation of some arc melts. The fact that the OIB source is most clearly identified in those samples showing the greatest enrichment in LIL and LREE further suggests that this source is extremely important for controlling these enrichments in arcs.

The origin of end-member I is more enigmatic. This source is relatively depleted, as demonstrated by the flat to light-depleted REE patterns, and by the fact that samples that plot close to this end-member have less Ba (and K, Rb, Sr, and REE) than those plotting close to end-member 2. A source region similar to that of MORB or Mariana back arc basin basalts (BABB [Hawkins and Melchior, 1985]) is appropriate for the inferred (La/Yb)n. This source is greatly enriched in Ba relative to La, and this must reflect an enrichment of Ba to the source.

Metasomatism of a MORB-type mantle by hydrous fluids with high Ba/La derived from the dehydrating subducted crust has been advocated for the origin of the high Ba/La characteristic of arc lavas [Perfit et al., 1980; Kay, 1984]. Although we know all too little about Ba/La of pelagic sediments and altered crust of the western Pacific or about

REE- SYSTEMATICS 6

DURING BATCH MELTING INSERT A OF PERIDOTITE, 30 MARIANA-VOLCANO ARC

4 (Lo/Yb)n 17 • x x x .. F=I ,,. /'x

,.. ,..•, / x .•:i::.::.:•:' '"•-'--'••' •"•• (Lø/Yb)n .e, •, - *

• •x• 3 •.•• x

• f x FUKUTOKU HIYOSHI / -• x ,wo'•,•, ' x•

_ 5• • x •x x

.•...•.•-•..•• • - •ont•e Source '1 / '--••-.'.• 3 FRACTIONAL • - CIP Islonds

•:"•' • I , c"•s•*•'z*• 'ø"> .... 0 ..... 0 50 I00 150 200

LOn

Fig. 10. Plot of (La/Yb) n versus La n for lavas from the Mariana-Volcano arc. The "mantle source" and trajectories expected for melting of garnet lherzolite and spinel lherzolite are taken from models developed from Figure 9a; F refers to the percent of partial melting. The box in the lower left of the figure is shown enlarged as insert A. Samples from CIP islands shown in Figures 2-3 define the stippled field. All samples studied here define the field outlined by the dash-dot line at its low end. Samples from NSP submarine volcanoes Daikoku, Eifuku, Fukujin, and Fukutoku-Hiyoshi-Iwo Jima are outlined as discrete fields. A trend calculated for low-pressure fractional crystallization evolution of melts is also shown.

q512 Lin et al.: Northern Mariana Shoshonites, 2

80-

Ba/La'

20-

ENDMEMSEI•iI• 1 Ba - La -Yb SYST E MATI CS,

Bo/Lo =80 x (Lø/Yb)n --0.5 MARIANA -VOLCANO ARC

-•xx tMETASOMATISM BY •x x JHYDROUS FLUIDS,/

• J / METASOMATISM

Z:, // •MARIANA •ISLANDS •:.:.•.• Decreasing degree of

'- •'•:• '•.•:J FUKUJIN

x •:??:. '• X • E I FUK / • • FU K U TOKU, H I YO SHI,

MORB

o • ,'o ,• ;o (Lo/¾b).

Fig. 11. Plot of Ba/La versus (La/Yb) n for samples from the Mariana-Volcano arc with fields outlined as for Figure 10. Field expected for a MORB-type mantle is shown as cross-hatched box in the lower left. Samples from a wester,n Pacific hot spot chain, Ponape in the Carolina Islands, are also shown [Dixon et al. 1984]. Trajectories expected for physical processes are shown schematically in the center of the figure. Metasomatism by a fluid with an exceedingly high Ba/La and Ba/Yb would move source compositions vertically, while metasomatism involving a fluid where Ba>La>Yb would move compositions to the upper right; decreasing degrees of partial melting would move melt compositions horizontally to the right. None of these processes can account for the hyperbolic trend which is best matched by mixing between two end-members. The mixing is labeled with fraction of end-member 1.

E'NDME'M•'E'R.• 2

Be/Lo * I 5

(Lo/Yb) n = 20

the partitioning of Ba and La into hydrous fluids, this explanation is a plausible one for the origin of end- member 1. An alternate explanation, that the high Ba/La component was stripped from an OIB-type mantle by hydrous fluids, is also plausible and is favored by the similarity of K/Rb and K/Ba for both end-members.

A final constraint on the origin of the LIL and LREE enrichment in the Mariana-Volcano arc comes from consideration of the temporal evolution of the arc. Shoshonitic volcanism of the NSP and Iwo Jima has been interpreted as the earliest stages in the formation of a new magmatic arc following the northward propagation of the Mariana back arc basin extensional regime [Stern et al., 1988]. In this interpretation, the large volcanoes of the CIP and Volcano arc represent older and more mature stages in arc evolution. The observation that end-member 1 is predominant in the source region of the more mature portion of the arc and that end-member 2 is predominant in the source region of the less mature portion of the arc suggests a temporal progression in the composition of the mantle wedge. Early arc melts tap a relatively undepleted OIB-type mantle (that nonetheless contains low contents of high field strength cations (HFSC), while volcanoes in the older portion of the arc tap a depleted mantle (that nonetheless has been continuously recharged with more soluble cations K, Rb, and Ba by metasomatic fluids). The evolution of arc sources thus may simply reflect the progressive depletion of an OIB-like subarc mantle by the continuous extraction of melts coupled with an ongoing recharge of the source by alkali-rich hydr0us fluids.

An alternative explanation of the Ba/La-(La/Yb) n data is that this represents the result of melting a "plum- pudding" mantle, where small OIB "plums" are set in a much vaster matrix of LIL-metasomatized depleted mantle. In this case, small degrees of melting would preferentially

tap end-member 2, which larger degrees of melting result in melts dominated by end-member 1. The implication of this is that it is a low degree of partial melting that causes the shoshonitic character of the northern NSP. We do not

at present favor this argument, largely because of the Ti and Yb arguments outlined above and elsewhere [Stern et al., 1988; Bloomer et al., this issue] and because historical records do not support a lesser volume erupted from the shoshonitic province relative to the CIP and VA [Bloomer et al., 1988]. We do appreciate, however, that the issue of whether the shoshonites reflect a source effect or a degree- of-melting effect is far from being resolved.

Conclusions

The active and dormant volcanoes at the northern end

of the Mariana arc as far north as Iwo Jima represent a province of LIL- and LREE-enriched magmatism that is perhaps unique among intraoceanic arcs. This major shoshonitic province is developed along the magmatic front of the arc and provides an excellent opportunity to examine the source of these enrichments. The Mariana shoshonites contrast with low-K calc-alkaline and tholeiitic

melts of the larger volcanoes of the Mariana and Volcano arc is being enriched 4x in K, Rb, and Ba, 2x in Sr, and 10x in La, but containing identical Yb. The shoshonites and more depleted lavas nevertheless have similar source characteristics as indicated by K/Rb and K/Ba. Marked variations of Ba/La, Sr/Nd, and (Ce/Yb) n are observed along the arc, with Ba/La and Sr/Nd decreasing and (Ce/Yb) n increasing as the lavas become increasingly enriched. These enrichments cannot be the result of

differing degrees of fractional crystallization or melting of a homogeneous source but must represent participation of two or more different sources.

Melt generation in the Mariana-Volcano arc results from melting of the subarc asthenosphere, and modeling of the REE indicates this occurs in the spinel-peridotite facies. Melt generation thus occurs at pressures of 22 kbar or less, probably in the top 40 km of the mantle. Compelling evidence for mixing sources is demonstrated by an observed trend of decreasing Ba/La with increasing (La/Yb) n. The shoshonites are derived from a source that is extremely similar to that responsible for OIB, except for strong depletions of HFSC. The lavas of the Volcano arc and the CIP of the Marianas are derived from a much

more depleted mantle that has been reenriched by alkali- rich hydrous fluids. The fact that the shoshonites are restricted to that portion of the arc that is reforming following back arc rifting while the more depleted lavas are found in the older portions of the arc indicates that the development of shoshonites is restricted in this setting to the earliest stages of arc evolution. The LIL- and LREE- enriched mantle beneath the arc will inevitably evolve into a depleted source as arc melts continue to be extracted.

Acknowledgments. We thank the officers and crew of the R/V Thomas G. Thompson for assistance in sample collection. We are grateful to I. Barnes, H. Brooks, J. Hein, E. Ito, J. Morris, K. Nakamura, Y. Nakamura, C. Proctor, R. Smith, and T. ¾allier for help during the cruise and for sharing ideas with us afterward. We thank S. Box, S. DeLong, A. Ewart, J. Morris, and K. Nakamura for critical reviews. This work was supported by NSF grants OCE-8415699 and OCE-8415739. The University of Texas at Dallas Programs in Geosciences Contribution 586.

References

Anderson, R. N., S. E. DeLong, and W. M. Schwarz, Thermal model for the subduction with dehydration in the downgoing slab, J. Geol., 86, 731-739, 1978.

Lin et al.: Northern Mariana Shoshonites, 2 •4513

Anderson, R. N., S. E. DeLong, and W. M. Schwarz, Dehydration, asthenospheric convection, and seismicity in subduction zones, J. Geol., 88, 445-451, 1980.

Apted, M. J., Rare earth element systematics of hydrous liquids from partial melting of basaltic eclogite--A re- evaluation, Earth Planet. Sci. Lett., 52, 177-182, 1981.

Arculus, R. J., and R. W. Johnson, Island-arc magma sources: A geochemical assessment of the roles of slab-derived components and crustal contamination, Geochem. J., 15, 109-133, 1981.

Baker, D. R., Depths and water content of magma chambers in the Aleutian and Mariana island arcs, Geology, 15, 496-499, 1987.

Bibee, L. D.,, G. G. Shor, Jr., and R. S. Lu, Inter-arc spreading in the Mariana Trough, Mar. Geol.. 35, 183- 197, 1980.

Bird, P., Stress and temperature in subduction shear zones: Tonga and Mariana, Geophys. J. R. Astron. Soc. 55, 411-434, 1978.

Bloomer, S. H., R. J. Stern, and N. C. Smoot, Physical volcanology of the submarine Mariana and Volcano Arcs, Bull. Volcanol., in press, 1988.

Bloomer, S. H., R. J. Stern, E. Fisk, and C. H. Geschwind, Shoshonitic volcanism in the Northern Mariana arc, 1, Mineralogic and major and trace element characteristics, J. Geophys. Res., this issue.

Brophy, J. G., and B4 D. Marsh, On the origin of high- alumina arc basalt and the mechanics of melt

extraction, J. Petrol., 27, 763-789, 1986. Carter, J. C., Mineralogy and chemistry of the Earth's

upper mantle based on the partial fusion - partial crystallization model, Bull. Geol. Soc. Am., 81, 2021- 2034, 1970.

Chase, T. E., H. W. Menard, and J. Mammerickx, Bathymetry of the North Pacific, Chart 6, Tech. Rep. Ser. TR-11, Inst. of Mar. Resour., 1968.

Chow, T. J., R. J. Stern, and T. H. Dixon, Absolute and relative abundances of K, Rb, Sr, and Ba in circum- Pacific island-arc magmas, with special reference to the Marianas, Chem. Geol. 28, 111-121, 1980.

Clague, D. A., and F. A. Frey, Petrology and trace element geochemistry of the Honolulu Volcanics, Oahu: Implications for the oceanic mantle below Hawaii, J_ Petrol., 23, 447-504, 1982.

Davidson, J.P., M. A. Dungan, K. M. Ferguson, and M. T. Colucci, Crust-magma interactions and the evolution of arc magmas: The San Pedro-Pellado volcanic complex, southern Chilean Andes, Geology, 1.•5, 443-446, 1987.

DeLong, S. E., F. N. Hodges, and R. J. Arculus, Ultramafic and mafic inclusions, Kanaga Island, Alaska, and the occurrence of alkaline rocks in island arcs, J. Geol., 83, 721-736, 1975.

Dixon, T. H., and R. Batiza, Petrology and chemistry of Recent lavas in the northern Marianas: Implications for the origin of island arc basalts, Contrib. Mineral. Petrol.. 70, 167-181, 1979.

Dixon, T. H., R. Batiza, K. Futa, and D. Martin, Petrochemistry, age, and isotopic composition of alkali basalts from Ponape Island, western Pacific, Chem. Geol., 43, 1-28, 1984.

Ewart, A., and C. J. Hawkesworth, The Pleistocene-Recent Tonga-Kermadec arc lavas: Interpretation of new isotopic and rare earth data in terms of a depleted mantle source model, J. Petrol., 28, 495-530, 1987.

Ewart, A., W. B. Bryan, and J. B. Gill, Mineralogy and geochemistry of the younger volcanic islands of Tonga, J. Petrol., 14, 429-465, 1973.

Garcia, M. O., N. W. K. Liu, D. W. Muenow, Volatiles in submarine volcanic rocks from the Mariana Island arc

and trough, Geochim. Cosmochim. Acta, 43, 305-312, 1979.

Gasparik, T., Two-pyroxene thermobarometry with new experimental data in the system CaO-MgO-A1203- SiO2, Contrib. Mineral. Petrol., 87, 87-97, 1984.

Gill, J. B., Orogenic Andesites and Plate Tectonics, 390 pp., Springer-Verlag, New York, 1981.

Green, T. H., and N.J. Pearson, Ti-rich accessory phase saturation in hydrous mafic-felsic compositions at high P, T., Chem. Geol., 54, 185-201, 1986.

Harrison, W. J., Partitioning of REE between garnet peridotite minerals and coexisting melts during partial melting, Year Book Carnegie Inst. Washington, 78, 562-568, 1979.

Hawkins, J. W. and J. T. Melchior, Petrology of Mariana Trough and Lau Basin basalts, J. Geoph¾s. Res., 90, 11,431-11,468, 1985.

Hein, J. R., C. L. Fleishman, L. A. Morgenson, S. H. Bloomer, and R. J. Stern, Submarine ferromanganese deposits from the Mariana and Volcano volcanic arcs, western Pacific, U.S. Geol. Surv. Open File Rep., 87- 281, 9 pp., 1987.

Hole, M. J., A.D. Saunders, G. F. Marriner, and J. Tarney, Subduction of pelagic sediments: Implications for the origin of Ce-anomalous basalts from the Mariana Islands, J. Ge01. Soc. London, 141, 453-472, 1984.

Hussong, D. M., and P. Fryer, Back-arc seamounts and the SeaMARC II seafloor mapping system, Eos. Trans. AGU, 64, 627-632, 1983.

Hussong, D. M., and S. Uyeda, Tectonic processes and the history of the Mariana arc: A synthesis of the results of Deep Sea Drilling Project leg 60, Initial Rep. Deep Sea Drill. Pro i., 60, 909-929, 1981.

Irving, A. J., A review of experimental studies of crystal/liquid partitioning, Geochim. Cosmochim. Acta, 42, 743-770, 1978.

Ito, E., and R. J. Stern, Oxygen- and strontium-isotopic investigations of subduction zone volcanism: The case of the Volcano arc and the Marianas Island Arc, Earth Planet. Sci. Lett.. 76, 312-320, 1986.

Jakes, P., and A. J. R. White, Major and trace element abundances in volcanic rocks of orogenic areas, Bull. Geol. Soc. Am. 83, 29-40, 1972.

James, D. D., A combined O, Sr, Nd, and Pb isotopic and trace-element study of crustal contamination in central Andean lavas, 1, Local geochemical variations, Earth Planet. Sci. Lett. 57, 47-62, 1982.

Jaques, A. L., and D. H. Green, Anhydrous melting of periodotite at 0-15 Kb pressure and the genesis of tholeiitic basalts, Contrib. Mineral. Petrol.. 73, 287- 310, 1980.

Katsumata, M., and L. R. Sykes, Seismicity and tectonics of the western Pacific: Izu-Mariana-Caroline and

Ryukyu-Taiwan regions, J. Geophys. Res., 74, 5923- 5948, 1969.

Kay, R. W., Elemental abundances relevant to identification of magma sources, Philos. Trans. R. Soc. London, A310, 535-547, 1984.

Kay, R. W., and P. W. Gast, The Rare Earth content and origin of alkali-rich basalts, J. Geol., 81, 653-682, 1973.

Kay, S. M., and R. W. Kay, Role of crystal cumulates and oceanic crust in the formation of the lower half of the Aleutian Arc, Geology, 13, 461-464, 1985.

Kay, S. M., R. W. Kay, and G. P. Citron, Tectonic controls on tholeiitic and calc-alkaline magmatism in the Aleutian Arc, J. Geoph¾s. Res., 87, 4051-4072, 1982.

Lambert, D. D., and E. C. Simmons, Magma evolution in the Stillwater Complex, Montana, 1, Rare-Earth element evidence for the formation of the Ultramafic Series, Am. J. Sci., 287, 1-32, 1987.

Leeman, W. P., The influence of crustal structure on

•4514 Lin et al.: Northern Mariana Shoshonites, 2

compositions of subduction-related magmas, J. Volcanol. Geotherm. Res., 18, 561-588, 1983.

MacGregor, I.D., The effect of CaO, Cr20 3, Fe20 3 and A120 3 on the stability of spinel and garnet peridotires, Phys. Earth Planet. Inter.. 3, 372-377, 1970.

Masuda, Y., and K.-I. Aoki, Two types of island are tholeiite in Japan, Earth Planet. Sci. Lett., 39, 298- 302, 1978.

Meijer, A., Pb and Sr isotopic data bearing on the origin of volcanic rocks from the Mariana island-arc system, Bull. Geol. Soc. Am., 87, 1358-1369, 1976.

Morris, J. D., and S. R. Hart, Isotopic and incompatible element constraints on the genesis of island are volcanics from Cold Bay and Amak Island, Aleutians, and implications for mantle structure, Geochim. Cosmochim. Acta, 47, 2015-2030, 1983.

Morrison, G. W., Characteristics and tectonic setting of the shoshonite rock association,, Lithos. 13, 97-108, 1980.

Nakamura, N., Determination of REE, Ba, Fe, Mg, Na, and K in carbonaceous and ordinary chondrites, Geochem. Cosmochim. Acta, 38, 757-775, 1974.

Nicholls, I. A., and K. L. Harris, Experimental rare earth element partition coefficients for garnet, clinopyroxene and amphibole coexisting with andesitic and basaltic liquids, Geochim. Cosmochim. Acta, 44, 287-308, 1980.

Nixon, P. H., and F. R. Boyd, Petrogenesis of the granular and sheared ultrabasic nodule suite in kimberlites, in Lesotho Kimberlites, P. H. Nixon, pp. 48-56, edited by Lesotho National Development Corp., Lesotho, 1973.

Notsu, K., N. Isshiki, and M. Hirano, Comprehensive strontium isotope study of Quaternary volcanic rocks from the Izu-Ogasawara are, Geochem. J., 17, 289- 302, 1983.

Parker, R. L., and D. W. Oldenburgh, Thermal structure of oceanic ridges, Nature, 242, 137-139, 1973.

Perfit, M. R., D. A. Gust, A. E. Bence, R. J. Arculus, and S. R. Taylor, Chemical characteristics of island-arc basalts: Implications for mantle sources, Chem. Geol.. 30, 227-256, 1980.

Richard, P., N. Shimizu, and C. J. Allegre, 143Nd/146Nd, a natural tracer: An application to oceanic basalts, Earth Plant. Sci. Lett., 31, 269-278, 1976.

Ryerson, F. J., and E. B. Watson, Rutile saturation in magmas: Implications for Nb-Ta-Ti depletion in island-arc basalts, Earth Planet. Sci. Lett., 86, 225- 239, 1987.

Simkin, T., L. Siebert, L. McClelland, D. Bridge, C. Newhall, and J. H. Latter, Volcanoes of the World, 232 pp., Hutchinson Ross, Stroudsburg, Pa., 1981.

Stern, R. J., On the origin of andesitc in the northern Mariana Island Are: Implications from Agrigan, Contrib. Mineral, Petrol., 68, 207-219, 1979.

Stern, R. J., Strontium isotopes from tireurn-Pacific intra- oceanic island arcs and marginal basins: Regional variations and implications for magmagenesis, Bull. Geol. Soc. Am., 93, 477-486, 1982.

Stern, R. J., and L. D. Bibee, Esmeralda Bank: Geochemistry of an active submarine volcano in the Mariana island are, Contrib. Mineral. Petrol,, 86, 159- 169, 1984.

Stern, R. J., and E. Ito, Trace-element and isotopic constraints on the source of magmas in the active Volcano and Mariana Island Arcs, western Pacific, J_. Volcanol. Geotherm. Res., 18, 461-482, 1983.

Stern, R. J., N. C. Smoot, and M. Rubin, Unzipping of the Volcano are, Japan, Tectonoph¾sics, 102, 153-174, 1984.

Stern, R. J., S. H. Bloomer, P.-N. Lin, E. Ito, and J. Morris, Shoshonitic magmas in nascent arcs: New evidence from submarine volcanoes in the northern

Marianas, Geology. 16, 426-430, 1988. Sugimura, A., Spatial relations of basaltic magmas in island

arcs, in Basalts, Vol. 2, edited by H. H. Hess and A. Poldervaart, pp. 537-571, Interscience, New York, 1968.

Sun, S.S., and G. N. Hanson, Origin of Ross Island basanitoids and limitations upon the heterogeneity of mantle sources of alkali basalts and nephelinites, Contrib. Mineral. Petrol., 52, 77-106, 1975.

Tatsumi, Y., D. L. Hamilton, and R. W. Nesbitt, Chemical characteristics of fluid phase released from a subdueted lithosphere and origin of are magmas: Evidence from high-pressure experiments and natural rocks, J. Volcanol. Geotherm. Res., 29, 293-309, 1986.

White, W. M., and J. Patcherr, Hf-Nd-Sr isotopes and incompatible element abundances in island arcs: Implications for magma origins and crust-mantle evolution, Earth Planet. Sci. Lett., 67, 167-185, 1984.

Wood, D. A., N. G. Marsh, J. Tarney, J.-L. Joron, P. Fryer, and M. Treuil, Geochemistry of igneous rocks recovered from a transcot across the Mariana Trough, Are, Fore-Arc, and Trench, sites 453 through 461, Deep Sea Drilling Project leg 60, Initial Rep. Deep Sea Drill. Proi., 60, 611-645, 1981.

Woodhead, J. D., and D. G. Fraser, Pb, Sr, and 10Be isotopic studies of volcanic rocks from the northern Mariana islands: Implications for magma genesis and crustal recycling in the western Pacific, Geochem. C0smochim. Acta. 49, 1925-1930, 1985.

Yuasa, M., and K. Tamaki, Basalt from Minami-Iwojima Island, Volcano Islands (in Japanese), Bull. Geol. Surv. Jpn., 33, 531-540, 1982.

S. H. Bloomer, Department of Geology, Duke University, Box 6729, College Station, Durham, NC 27708.

P.-N. Lin and R. J. Stern, Center for Lithospheric Studies, The University of Texas at Dallas, Box 830688, Richardson, TX 75083.

(Received October 20, 1987; revised April 12, 1988;

accepted April 26, 1988.)