32. BASALTIC GLASSES FROM THE MARIANA TROUGH1

Patricia Fryer,2 John M. Sinton, and John A. Philpotts, Hawaii Institute of Geophysics,University of Hawaii, Honolulu, Hawaii

ABSTRACT

The tectonic character oceanic crustal generation in the Mariana Trough, an active, extensional, back-arc basin, issimilar in many ways to that at normal mid-ocean ridges. Thus it might be expected that the basalts erupted in the basinwould be like normal mid-ocean ridge basalt. Basaltic glasses from each environment show consistent differences,however, Mariana Trough basaltic glasses have lower total iron and TiO2 and higher A12O3 at a given MgO concentra-tion. The rare earth patterns for Mariana Trough glasses are essentially flat relative to chondrite. Incompatible elementabundances are higher for the trough basalts than for ridge basalts. The volatile content of trough basalts is high (ap-proximately 1% H2O).

These compositional characteristics are also true of basalts from other active marginal basins and suggest that thesources of these lavas differ from those of normal mid-ocean ridge basalts. Part of the reason for these differences maybe the peculiar tectonic environment of active back-arc basins, in particular their location above the subducted oceaniclithospheric plate.

INTRODUCTION

Geologic Setting

The Mariana Trough (Fig. 1) is a young extensionalbasin situated between the volcanically active MarianaIsland arc and the West Mariana Ridge, a remnant arc.Spreading began there approximately 6 Ma (Fryer andHussong, this volume). Patterns of magnetic anomaliesthroughout the basin (Hussong and Fryer, 1980), seis-micity in the central portion (Hussong and Sinton,1979), the morphology of the spreading center (Uyedaand Hussong, 1978), and the increase in seismic veloc-ities with age of crustal layers (Ambos and Hussong,1979) are similar to features of slowly spreading seg-ments of normal mid-ocean ridges. There are a numberof ways in which the Mariana Trough differs tectonic-ally from normal mid-ocean ridges, however. It issituated very close to a convergent plate boundary, andthe changes of spreading direction within the basinnoted by Hussong and Fryer (1980) are probably relatedto the convergence. At 18°N the spreading rate is muchslower (1.65 cm/y., according to Hussong and Fryer[1980]) and the central graben is deeper (4600 m, Fryerand Hussong [this volume]) than encountered at otherspreading ridges. Seismic refraction studies show thatcrust and mantle velocities are lower than at mid-oceanridges (Ambos and Hussong, 1979). Gravity modelssuggest a mass deficiency at depth, indicating mantledensities that are lower than normal (Sager, 1980). It issuspected that the degree of fracturing of the crust, atleast at this latitude, is greater than is usually encoun-tered at mid-ocean ridges (Ambos, 1980; Fryer and Hus-song, this volume). The fractured nature of the crustmay account for the low-magnitude earthquakes (M <

140 145° 150°

Initial Reports of the Deep Sea Drilling Project, Volume 60.Present address: Department of Geological and Geophysical Sciences, Princeton

University, Princeton, New Jersey.

Figure 1. Position of the Mariana arc-basin system in the western Pa-cific. DSDP Sites 453 to 456 were drilled in the Mariana Trough.

3; Hussong and Sinton, 1979) and the high degree of hy-drothermal activity in the region (see site chapters forSites 453 and 456 and Uyeda and Horai, this volume).

These similarities and differences of mid-ocean ridgeand Mariana Trough tectonics raise some basic ques-tions concerning magma genesis in the back-arc basin.

601

P. FRYER, J. M. SINTON, J. A. PHILPOTTS

How do Mariana Trough basalts and other back-arcbasin basalts compare with mid-ocean ridge basalt(MORB)? How is magma composition related to thetectonic setting of the Mariana Trough? Have theserelationships changed with time?

SamplingPrior to 1977, samples were dredged from the central

portion of the Mariana Trough (Fig. 2) on two cruisesby the Scripps Institution of Oceanography. The com-positions of some of these samples are reported in Hartet al. (1972) and Meijer (1976). In preparation for Leg60 of the Deep Sea Drilling Project, detailed surveys ofthe prospective sites were carried out by the Hawaii In-stitute of Geophysics (HIG) during 1976 and 1977.Dredging was done in the vicinity of each proposeddrilling site, but only those stations from the centralportion of the basin yielded fresh basalts. During Leg60, four locations were drilled in the trough (Fig. 2). Ofthese, only two sites (454 and 456) penetrated igneousrocks of the marginal basin crust.

Many of the samples collected in the dredging opera-tions and in subsequent DSDP drilling are fragments ofbasaltic pillow lavas. They have glassy rims (0.1-1 cmthick), spherulitic to variolitic margins (up to 4 cmthick), and fine-grained hypocrystalline interiors. Therocks from dredge hauls 13 and 14 are sparsely por-phyritic, containing skeletal microphenocrysts of pla-gioclase and olivine, which generally occcur in smallcrystal clumps. Spinel inclusions occur in olivine andPlagioclase phenocrysts in some of the dredge 13 and 14samples. A large number of aphyric and porphyriticpillow fragments were retrieved in dredge haul 3. In ad-dition to skeletal olivine and Plagioclase in the dredge 3pillow fragments, there are also some rounded and em-bayed Plagioclase and pyroxene phenocrysts. No spinelwas observed in any of these rocks. The groundmass ofthe rocks from all three dredge locations varies fromhyaline to hypocrystalline. In most, microlitic to verysmall skeletal laths of Plagioclase are the dominantgroundmass phase. Many of the rocks contain glob-ulites and margarites of pyroxene(?). Minute grains ofolivine occur in the groundmass of some of the rocksfrom dredge 13. Most of the rocks chosen for this studyare very fresh. Some of them contain a small amount ofsecondary clays occurring as vesicle linings or vein fill-ings. The vesicularity of these rocks is generally greaterthan 10% by volume. The thin sections described inTable 1 were cut from the outer glassy portions of thesamples, where vesicle development is lowest.

The DSDP samples are aphyric to sparsely por-phyritic with Plagioclase and olivine phenocrysts. Al-though the glass samples from the cores were too smallto permit thin sectioning, descriptions of the lithologicunits from which these samples were taken are includedin the petrology sections of site chapters 454 and 456.

Small chips from the glassy rims of 30 of the freshestpillow fragments retrieved in HIG dredge hauls 3, 13,and 14 were chosen for analysis. Only seven pillow frag-ments containing fresh glassy basalt were recovered

from DSDP sites in the Mariana Trough. All of thesewere analyzed. In addition to these, a number ofsamples of basaltic glass from several other marginalbasins were analyzed—one from the Lau Basin, twofrom the Parece Vela Basin, and five from the ScotiaSea. The Lau Basin, located between the Lau and TongaKermadec Ridges, shows evidence of currently active ex-tension but appears to have a complicated pattern ofspreading (Hawkins, 1976a). Both the Parece Vela Ba-sin and the Scotia Sea show normal mid-ocean ridgespreading patterns (Langseth and Mrozowski, in press;Barker, 1970). The Parece Vela Basin, bounded on thewest by the Palau-Kyushu Ridge and on the east by theWest Mariana Ridge, was actively spreading fromOligocene to Miocene time (Scott et al., in press). TheScotia Sea is a currently active marginal basin similar inmany respects to the Mariana Trough (Barker, 1970;Saunders et al., in press). Comparison of MarianaTrough basalts with basalts from these other marginalbasins allows us to examine the consistency of composi-tions among marginal basin lavas.

ANALYTICAL PROCEDURESAlthough numerous investigators have studied the

composition of MORB (Engel and Engel, 1964; Miya-shiro et al., 1970; Byerly et al., 1975; Melson et al.,1976), the rocks which were studied vary widely indegree of crystallinity and in character and extent ofalteration. Furthermore, the techniques used to analyzethese rocks differ considerably. Most commonly the ba-salts collected in the ocean ridge provinces are frag-ments of pillow lava with quenched, glassy rims repre-senting the liquid composition of the lavas at the timethey were erupted. If only fresh portions of the glassyrims are chosen for analysis, the problems of variationin degree of crystallinity and alteration among samplescan be avoided. Thousands of basaltic glasses from mid-ocean ridge provinces have been analyzed for major ele-ment composition by microprobe at the Smithsonian In-stitution (Melson et al., 1976). The technique is de-scribed in Byerly et al. (1975). The back-arc basinglasses chosen for the present study were analyzed bymicroprobe (Table 2) at the Smithsonian Institution us-ing the same technique as for MORB analyses. Thus theSmithsonian glass file provides a large body of data withwhich to compare the major element composition of ourback-arc basin samples.

Analysis of selected samples for certain rare earth(Ce, Nd, Sm, Eu, Gd, Dy, Er, Yb, Lu) and other minorand trace element (Li, K, Rb, Sr, Ba) compositions(Table 3) were performed by mass spectrometry at theHawaii Institute of Geophysics, using stable isotopedilution methods. This technique is described in Web-ster (1960) and Schnetzler et al. (1967) as modified bySchuhmann et al. (1980).

CHEMICAL ANALYSESBasalts from normal mid-ocean ridge segments

(N-MORB)—that is, those not influenced by the pecu-liar chemistry of melting anomalies ("mantle plumes"

602

17° 45'

17°30'

Pagan

143°45'E 144C 144° 15' 144° 30' 144°45' 1 4 5 C

145°15' 145°30' 145° 45' 146C

Figure 2. Bathymetry map of the Mariana Trough at about 18°N showing locations of HIG dredge sites (open squares) and DSDP Sites 453 to 456 (filled circles). Open triangles and opencircles are locations of SIO dredges (Hart et al., 1972). Glasses analyzed are from dredge sites 3, 13, and 14 and DSDP Site 454.

>>rHoO

>inCΛmCΛ

P. FRYER, J. M. SINTON, J. A. PHILPOTTS

Table 1. Petrologic description of basalt samples dredged from theMariana Trough.a

Sample0

MO3O2M0310MO338M0370MO373M0379MO38OM03100M03101MO31O2M1300M1316Ml 330M l 349M1355M1371M1372M l 380M1381M l 396M1417M1419M1441Ml 463M14104M14106

PI.

128271586

454233

12012101Tr

Phenocrysts

Ol.

251212264

2233221Tr12611633

Px.

Tr

Sp.

Tr

TrTrTrTr

11

Groundmass

PI.

10131531520335115243157310810

53

105

Ol.

4

5

23Tr1

Px.

10101831820124

123105105510108

31

Tr5

Ores

Tr

Tr

Tr

Tr1

Tr

Tr1Tr

Glass

4767548545678677719757826676616884657274648289796480

Alt.

30

451

2

23

Tr2

Tr31Tr

3

Ves.

1435105238273128315187310731510

Note: PI. = Plagioclase; Ol. = olivine; Px. = pyroxene; Sp. = spinel; Ores =opaque minerals; Alt. = secondary minerals (mainly brown and green clays);Ves. = vesicles.

aVolume percent estimated from thin section.b Samples M03, M13, and M14 are from dredges 3, 13, and 14 (Fig. 2).

or "hot spots") or propagating rifts—have a distinctiveset of major element compositional traits (see Table 2).These basalts have low values of K2O (0.1-0.5 wt.%),TiO2 (1.0-2.0 wt.%), total iron (6.5-12.0 wt.%), andP2O5 (0.08-0.23 wt.%) but high values of CaO(10.5-12.5 wt.<tt>) and A12O3 (15.0-18.0 wt.%) (Engeland Engel, 1964; Engel et al., 1965; Aumento, 1968;Miyashiro et al., 1970; Kay et al., 1970; Melson et al.,1976). Major element composition of the back-arc basin(BAB) basaltic glasses for our study do show thesecharacteristics, but the BAB glasses also have a distinc-tive set of compositional traits and show certain differ-ences from those of N-MORB glasses. A comparison ofliquid trends on MgO variation diagrams (Fig. 3A-C)for N-MORB versus BAB glasses shows displacement ofthe trends for total iron (FeO1) (2.5 wt.% < MORB),TiO2 (0.5 wt.<7o < MORB), A12O3 (2.0 wt.Vo >MORB). When the BAB glasses are compared withMORB glasses affected by proximity to mantle plume orhot spot sources (P-MORB), the FeO1 and A12O3 trendsremain distinct, whereas that of TiO2 does not (Fig.3C).

Some trace element abundances also help to distin-guish MORB and the BAB glasses. N-MORB showslight rare earth element (REE) depletion, and P-MORBshows light REE enrichment with respect to chondriticabundances (Schilling, 1975b; Sun et al., 1979). REEpatterns for the Mariana Trough glasses are essentiallyflat with respect to chondrite and fall near the upperlimit for MORB (Fig. 4). Plots of REE for whole-rocksamples from the Scotia Sea show patterns similar to theMariana Trough glasses (Tarney et al., 1977). REE pat-terns of BAB are more similar to those of MORB transi-

tional between P- and N-MORB (T-MORB), except thatBAB lack the slightly concave upward pattern of thelight REE that is characteristic of T-MORB (e.g., Whiteand Bryan, 1977; Sun et al., 1979).

Incompatible element abundances are generally higherfor the Mariana Trough glasses, and ratios of K/Ba,Ce/Ba, Sr/Ba, and K/Rb are lower than N-MORB butsimilar to T- and P-MORB (Table 3).

DISCUSSIONThis chapter presents the results of initial investiga-

tions of the basaltic glasses gathered in site surveys andduring Leg 60 drilling in the Mariana Trough. Detailedinterpretation of these data will be presented elsewhere.At this stage of our investigations it is important to notethe major and distinctive characteristics of the MarianaTrough glass compositions and to consider whetherthese compositions imply tectonic control of magmagenesis in the active marginal basin environment.

Previous investigations of BAB basalts have led tothe conclusion that, with regard to major element com-position, there is little difference between MORB andBAB basalts (Hart et al., 1972; Hawkins, 1974, 1976a,1976b, 1977; Hawkins and Batiza, 1975; Gill, 1976;Tarney et al., 1977; Saunders et al., in press; Matteyet al., in press). But trace element analysis shows thatin many cases the BAB lavas are intermediate betweenMORB and basalt of the island arc tholeiitic series (Gill,1976; Saunders et al., in press; Tarney et al., 1977; Fry-er, 1980; Wood et al., this volume). Most of these inves-tigations deal exclusively with whole-rock analyses. Thedata here demonstrate that major element compositionsof BAB basaltic glasses are distinctive. The trends forBAB and MORB glasses in MgO variation diagrams(Fig. 3) indicate control by mineral fractionation, pri-marily olivine, for the lavas erupted in these environ-ments.

The observed depletion in total iron and TiO2 and en-richment in A12O3 may be related to mineralogy of thesource. For example, melting of a Plagioclase peridotiteat low pressure could produce enrichment in A12O3 andconsequent relative depletion in iron and TiO2. It seemsmore likely that the major element composition of theBAB glasses reflects the extent to which the volatile con-tent of the lavas controls the order and composition ofphases crystallizing during fractionation of the magmas.Glass samples from the Mariana Trough and the ScotiaSea have high contents of H2O, CO2, F, and Cl (Garciaet al., 1979). Under hydrous conditions the field ofstability of olivine is greatly expanded, Plagioclase crys-tallization is inhibited, and the relative amounts of crys-tallization of pyroxene and amphibole vary (Hamilton,1964; Green and Ringwood, 1968). These effects mayexplain the relative depletion in iron and TiO2 and theenrichment in A12O3 characteristic of BAB lavas.

We are currently analyzing glass samples from theParece Vela and Lau basins for volatile content. If highvolatile content of BAB lavas is a consistent trait, thepresence of volatiles may be explained in a number ofways: (1) Alteration of submarine lavas, particularly inregions with high hydrothermal activity, would increase

604

BASALTIC GLASSES

Table 2. Results of microprobe analysis of basaltic glasses from the Mariana Trough, LauBasin, Parece Vela Basin, and the Scotian Sea.

Samplea

M0302M0310M0311MO338MO338C

M036OMO37OMO373MO379MO38OM03100M03101M03102M1300M1316Ml 330Ml 349M1355M1371Ml 372M1380M1381M1396M1410M1417M1419M1441M1463M14104M14106M454-4-CCM454-8-1M454-11-1M454-14-1M454-14-1M454-16-1M454-19-1SS20.13SS20.31SS20.35SS20.43SS23.4L203-6-3P54-8-1P149-17-3MORB-1MORB-2

SiO2

48.8551.7952.5351.2451.5951.7750.8851.7649.4451.4750.6351.9050.4451.0451.5051.2751.7251.6451.4351.4050.5450.6751.2750.5149.3850.6851.2250.1550.8349.9650.2450.1750.9252.6952.7952.3451.0851.7052.1651.6951.5851.3451.9950.1148.9049.3450.47

A12O3

17.0415.9516.0116.1015.9815.5516.0315.9217.1215.2315.3115.5515.5617.1516.2215.9616.2616.37;i6.4016.1917.1317.0817.2817.5917.8917.2016.3215.9817.1916.9216.5616.8817.0817.0116.8316.7816.5816.2316.1916.3616.3915.9916.1817.3616.5817.0415.31

FeO1

9.818.999.588.949.119.368.428.829.799.619.819.319.307.598.518.438.688.558.688.567.717.667.728.018.687.899.179.767.747.837.737.557.487.867.897.928.328.148.127.998.228.917.529.299.369.97

10.42

MgO

8.106.355.166.166.176.746.656.397.496.406.256.796.577.546.676.716.787.426.837.007.687.937.367.015.776.906.946.897.217.268.097.897.946.887.186.897.306.997.047.097.086.697.148.067.597.197.46

CaO

10.7210.609.14

10.3910.5611.0410.6810.5511.0110.5610.5211.3210.4311.8411.2511.2111.2111.3011.2111.2711.7311.8411.8011.6512.0611.4411.2011.8411.5911.5212.1912.4612.7211.1010.9911.0611.8511.1411.2111.2411.2111.3612.6411.2611.7611.7211.48

Na2O

2.983.253.733.363.303.072.963.272.843.273.353.013.402.702.862.863.003.093.133.003.002.932.812.673.612.663.113.492.592.652.422.352.442.672.672.542.653.113.073.093.103.382.072.942.432.732.64

K2O

0.200.330.410.380.340.250.280.260.250.260.310.240.360.170.200.200.180.120.200.210.190.180.200.260.230.270.190.180.200.270.190.190.180.260.240.260.150.250.260.230.250.350.280.110.220.160.16

TiO2

1.181.531.751.531.541.631.311.521.151.741.751.56I Al1.061.351.371.371.301.371.391.131.091.131.151.461.141.391.591.131.061.010.900.891.191.151.171.131.371.421.341.371.540.700.990.961.491.58

P2O5

0.150.180.210.180.160.160.160.160.120.170.200.190.170.120.160.150.150.130.130.160.140.130.130.130.170.140.160.180.160.130.120.140.110.120.130.150.130.200.170.160.180.200.160.110.100.160.13

Total

99.0398.9798.5298.2898.7599.5797.3798.6599.2198.7198.1399.8797.7099.0998.5698.1699.3599.9299.3899.1899.2599.5199.7098.9899.2598.3299.70

100.0698.6497.6098.5598.5399.7699.7899.8799.1199.1999.1399.6499.1999.4699.7698.68

100.2397.6099.8099.65

Sourceb

MTMTMTMTMTMTMTMTMTMTMTMTMTMTMTMTMTMTMTMTMTMTMTMTMTMTMTMTMTMTMTMTMTMTMTMTMTSSSSSSSSSSLB

PVBPVB

AM

Note: Analyst, Timothy O'Hearn.a Samples M03, M13, and M14 are from dredges 3, 13, and 14, respectively (Fig. 2). M454 samples are from

DSDP Site 454.b MT = Mariana Trough basaltic glass (this study); LB = Lau Basin basaltic glass (this study); PVB =

Parece Vela Basin basaltic glass (this study); SS = Scotia Sea basaltic glass (this study); A = average mid-ocean ridge basalt (Aumçnto, 1968); M = average of basaltic glass from normal sections of the East PacificRise and the Mid-Atlantic Ridge; total: 64 samples (Melson et al., 1976).

c Duplicate analysis.

Table 3. Rare-earth element and large ion lithophile element abun-dances for selected basaltic glass samples from the MarianaTrough.

Samplea

CeNdSmEuGdDyErYbLuBaKRbSrLi

M0379

11.28.362.731.08

4.723.283.180.47

38.41729

2.34147

5.16

MO31O2

16.112.33.821.38

5.603.483.270.50

75.22338

3.93182

7.01

M1300

10.58.642.661.033.834.242.512.400.35

31.01758

2.48176

5.20

M1316

11.510.33.381.254.495.223.273.090.45

35.11993

3.00161

6.05

M1417

15.010.93.191.25

4.632.902.920.42

66.01778

1.37212

5.95

M1419

11.69.522.911.08

4.642.942.670.40

48.42350

3.52166

5.89

a See Note b, Table 1.

the volatile content of the lavas. However, the sampleschosen for analysis in this study were fresh glasses, thusthe problem of alteration does not apply. (2) Loggingdata (density) from Site 454 in the Mariana Trough indi-cate that some flows may be interbedded with sedimentlayers (see site chapter for Site 454, this volume). It hasbeen suggested that the environment at this site wouldbe suitable for sediment assimilation into the lavaserupted there (Wood et al., this volume). It is possiblethat the high water content of the glasses is a result ofassimilation of sediment into the lavas. It seems un-likely, however, that the high water content also notedin inclusions in Plagioclase phenocrysts (xenocrysts?) byGarcia et al. (1979) in the same samples derived fromthe incorporation of sediment into the lava. (3) Finally,the high volatile content of the BAB lavas may be re-lated to the tectonic environment in which they formed.Each of the active back-arc basins from which we have

605

P. FRYER, J. M. SINTON, J. A. PHILPOTTS

115 —

10

s1 _ _ _ _ _ - —

A

- '

- - - - ~"

i

— " ~ . . . . / • • • ' '

i

MgO

MgO

Figure 3. MgO variation diagrams for A12O3 (A), FeO1 (B), and TiO2

(C) for fields of basaltic glasses from back-arc basins: MarianaTrough, Parece Vela Basin, Lau Basin, and the Scotia Sea (solidline), normal segments of the Mid-Atlantic Ridge (dotted line),East Pacific Rise (dashed line), and the FAMOUS area (dashedand dotted line).

analyzed data or are analyzing glass samples for volatilecontent is associated with a convergent plate boundary.Presumably, active spreading in the Parece Vela Basinwas associated with subduction of Pacific oceanic litho-sphere before the opening of the Mariana Trough (Scottet al., in press). Volatile constituents of the subsidingoceanic plate are likely to permeate the overlying mantleto some extent. The Mariana Trough is a narrow basinand the subduction zone lies close to the spreading cen-ter. Thus volatiles in the basaltic glasses from that basinmay more readily be assumed to originate in the sub-ducted slab. In other back-arc basins, such as the Scotia

Sea, the Benioff zone is much farther from the spread-ing center (see, for example, Barker, 1970). It is possiblethat convection patterns set up in the mantle above thesubsiding plate (see Andrews and Sleep, 1973; Toksözand Bird, 1977) could account for the lateral transportof volatiles necessary to enrich the regions below thespreading centers of such basins.

Analysis of existing trace element data collected sofar impose several possible constraints on the natureand previous melting history of BAB lavas. TheMariana Trough glasses do not show the light REE de-pletion characteristic of normal MORB (Fig. 4). Pat-terns of REE for Mariana Trough glass and Scotia Seawhole-rock analyses are essentially flat relative to chon-drites. and lie near the upper limit of the field of MORBat 15 to 20 times chondritic abundances.

If the sources of the lavas erupted in these back-arcbasins had not been subject to previous episodes ofmelting they should have retained a primordial "chon-dritic" pattern of REE abundances. Basaltic liquids de-rived at high degrees of melting of such an undepletedsource would be slightly enriched in light REE (Phil-potts et al., 1972; Schilling, 1975b; Yoder, 1976). Rela-tive enrichment in light rare earths is characteristic of"plume" magmas (Schilling, 1973a, 1973b, 1975a) withan undepleted source region different from and presum-ably deeper than the source region for normal MORB.The Philippine Sea plate is composed of normal MORB,depleted in light REE (Mattey et al., in press; Scott etal., in press). Thus the mantle source of those lavas hasundergone at least one episode of melting. In order toproduce magmas with chondritic REE patterns in amarginal basin formed adjacent to the Philippine Sea, apreviously undepleted source, possibly material fromdeeper in the mantle, might be required.

Chondritic REE patterns can also occur in lavas pro-duced from a very small degree of partial melting of asource that has experienced a previous episode of melt-ing. The light REE are more strongly partitioned into aliquid in equilibrium with the common rock formingminerals (except for Plagioclase) than are the heavyREE (Schnetzler and Philpotts, 1970). The concentra-tion of a given REE in a liquid may be calculatedaccording to the relationship presented by Schilling(1975b, Appendix I):

CJ

j = n

J=I

j = n

j = i

where/ = rare earth element,j = phase,

CL = concentration of rare earth element inliquid,

Cp = concentration of rare earth element inparent material,

606

BASALTIC GLASSES

M1417

Figure 4. Chondrite normalized values of rare earth elements for six basaltic glasses from the MarianaTrough (circles) compared with the range of values for N-type MORB (squares).

Y = weight fraction of liquid (degree ofmelting),

K = partition coefficient between residual phaseand liquid,

E = eutectic proportion in weight percent ofphase j , and

X = fraction of residual phase remaining inequilibrium with liquid in weight percent.

From this relationship it can be seen that a very smallamount of liquid in equilibrium with a light-REE-depleted source could segregate to produce a magmawith a chondritic pattern. Even slightly greater amountsof melting would alter the REE signature considerably,however. It is unlikely that the marginal basin basaltsrepresent such very small degrees of partial melting,since overall abundances fall within the field ofN-MORB, and generation of MORB requires approxi-mately 20 to 30%-partial melting (Schilling, 1975b).

The patterns of REE abundances in rocks are influ-enced by the mineralogy of the source as well as by itsmelting history. The solid-liquid partition coefficients(CS/CL) of heavy REE in pyroxene and garnet aregreater than those in the light REE, olivine shows littledifference, and Plagioclase is the only common rock-forming mineral in which CS/CL is greater for the lightREE than for the heavy REE (Schnetzler and Philpotts,1970). If Plagioclase were an important component inthe source of BAB, there should be a noticeable positiveEu anomaly in MBB lavas, since Eu is strongly parti-tioned into Plagioclase. No such anomaly is present inour data.

Another indication of the character of the sourcederives from the incompatible element ratios of theMariana Trough glasses (Table 3). These indicate thatthe source for BAB is indeed different from that ofN-MORB. Ba concentration in the BAB samples ishigher than that of MORB, and ratios of K/Ba, Ce/Ba,Sr/Ba, and K/Rb are lower than for N-MORB. Rb isdepleted with respect to Sr in BAB lavas, indicating thatthe source for BAB may have experienced loss of melt insome previous event. Isotope ratios 87Sr/86Sr valuesfrom Mariana Trough basalts range from 0.7026 to0.7029 (Hart et al., 1972; Meijer, 1976; Melchior andHawkins, 1980), also implying a previous depletionevent. The higher concentrations of Sr, Ba, Rb, andlight REE in BAB relative to N-MORB probably reflectsa later enrichment of the BAB source region with theseelements.

It is obvious that interpretation of the present data isinconclusive regarding history of the source material forBAB magmas. One method which has proven useful fordetermining previous melting history of mantle sourceregions is Nd isotope studies (DePaolo and Wasserburg,1976; O'Nions et al., 1977). We have selected samplesrepresentative of the range of rare earth values amongthe Mariana Trough glasses for Nd isotope analysis.These analyses will be completed shortly.

CONCLUSIONSThe data presented here, although preliminary, do

permit several conclusions concerning the compositionand generation of basalts in active back-arc basins: (1)Fresh basaltic glasses from all the active back-arc basinsstudied are similar in major and trace element composi-

607

P. FRYER, J. M. SINTON, J. A. PHILPOTTS

tions. (2) They differ from MORB in having higherA12O3 and lower FeO1 for a given MgO concentration.Trace element abundances in the BAB lavas are morelike P-MORB than N-MORB in having relatively higherconcentrations of light REE and incompatible elementsand lower ratios of K/Ba, Ce/Ba, Sr/Ba, and K/Rb. (3)With respect to Sr, K, Ba, and Rb concentrations, thesources for basalts from active back-arc basins are morelike those of mantle plumes or hot spots than normalridge segments. The sources for BAB are not as enrichedin light REE as those of plume basalts, however prob-ably the enriched lithophile elements Sr, K, Ba, and Rband light REE concentrations reflect enrichment of thesource regions of BAB with these elements. (4) Thesource of the volatile and enriched lithophile elements isproblematic, but we feel that hydrous fluids or vaporsderived from the subsiding oceanic lithospheric platemay enrich the source regions of BAB magmas withthese elements.

ACKNOWLEDGMENTS

We would like to express our thanks to the shipboard personnel onthe Kana Keoki and Glomar Challenger for their assistance in procur-ing the samples from the Mariana Trough. Andrew D. Saunders pro-vided the glass samples from the Scotia Sea. We thank TimothyO'Hearn, who performed the microprobe analyses of the glasses at theSmithsonian Institution. We are grateful to Gerard J. Fryer for hisassistance in the computer programming required for the analysis andinterpretation of the data and to James Natland, Charles Langmuir,and Robert Stern for their helpful comments on this paper.

This work was supported by the National Science Foundationunder Grant No. OCE78-09034. The chapter is Hawaii Institute ofGeophysics Contribution No. 1149.

REFERENCES

Ambos, E. L., 1980. Crustal structure of the Mariana Trough fromseismic refraction data [M.S. thesis]. University of Hawaii.

Ambos, E. L., and Hussong, D. M., 1979. Crustal structure of theMariana Trough. EOS, 60(18):389. (Abstract)

Andrews, D. J., and Sleep, N. H., 1973. Numerical modelling of tec-tonic flow behind island arcs. Geophys. J. Roy. Asir. Soc,38:237-251.

Aumento, F., 1968. The Mid-Atlantic Ridge near 45 °N. II. Basaltsfrom the area of Confederation Peak. Can. J. Earth Sci.,5(1):1-21.

Barker, P. F., 1970. Plate tectonics of the Scotia Sea region. Nature(London), 228:1293-1297.

Byerly, G. R., Melson, W. G., Nelen, J. A., et al., 1975. Abyssal ba-saltic glasses as indicators of magma compositions. In Mason, B.(Ed.), Mineral Sciences Investigations, Smithsonian Contributionsto the Earth Sciences, 19:22-30.

DePaolo, D. J., and Wasserburg, G. J., 1976. Nd isotopic variationsand petrogenetic models. Geophys. Res. Lett., 3:249.

Engel, A. E. J., and Engel, C. G., 1964. Composition of basalts fromthe Mid-Atlantic Ridge. Science, 144:1331-1333.

Engel, A. E. J., Engel, C. G., and Havens, R. G., 1965. Chemicalcharacteristics of oceanic basalts and the upper mantle. Geol. Soc.Am. Bull., 76:719-734.

Fryer, P., 1980. Composition and petrogenesis of basalt in the Mari-ana Trough, an active back-arc basin. 26th Intl. Geol. Congr.,1:44. (Abstract)

Garcia, M. O., Liu, N. W. K., and Muenow, D. W., 1979. Volatilesin submarine volcanic rocks from the Mariana island arc andtrough. Geochim. Cosmochim. Acta, 43:305-312.

Gill, J. B., 1976. Composition and ages of Lau Basin and Ridge vol-canic rocks: Implications for evolution of inter-arc and remnantarc. Geol. Soc. Am. Bull., 87:1384-1395.

Green, T. H., and Ringwood, A. E., 1968. Genesis of the calc-alka-line igneous rock suite. Contrib. Mineral. Petrol., 18:105-162.

Hamilton, W., 1964. Origin of high alumina basalt, andesite, and da-cite magmas. Science, 146:635-637.

Hart, S. R., Glassley, W. E., and Karig, D. E., 1972. Basalts and sea-floor spreading behind the Mariana Island arc. Earth Planet. Sci.Lett., 15:12.

Hawkins, J. W., 1974. Geology of the Lau Basin, a marginal sea be-hind the Tonga Arc. In Burke, C , and Drake, C. (Eds.), Geologyof Continental Margins: Berlin (Springer-Verlag), pp. 505-520.

, 1976a. Petrology and geochemistry of basaltic rocks of theLau Basin. Earth Planet. Sci. Lett., 218:283-297.

, 1976b. Petrology of the Lau Basin basalts: Oceanic crustof a marginal basin. EOS, 57:410. (Abstract)

., 1977. Petrology and geochemical characteristics of mar-ginal basin basalt. In Talwani, M., and Pitman, W. C. (Eds.),Island Arcs, Deep Sea Trenches, and Back-Arc Basins: Washing-ton (Am. Geophys. Union), pp. 355-365.

Hawkins, J. W., and Batiza, R., 1975. Tholeiitic basalt from an activespreading center on the North Fiji Plateau 15°3O'S, 173°30'E.EOS, 56:1078. (Abstract)

Hussong, D. M., and Fryer, P., 1980. Tectonic evolution of the mar-ginal basins behind the Mariana Arc. Geol. Soc. Am. Abstractswith Programs, 12:1112. (Abstract)

Hussong, D. M., and Sinton, J. B., 1979. Low-level seismicity in theMariana Trough. EOS, 60(46):877. (Abstract)

Kay, R., Hubbard, N. J., and Gast, P. W., 1970. Chemical charac-teristics and origin of oceanic ridge volcanic rocks. J. Geophys.Res., 75:1585-1613.

Langseth, M. G., and Mrozowski, C , in press. Geophysical surveysfor Leg 59 sites. In Kroenke, L., Scott, R., et al., Init. Repts.DSDP, 59: Washington (U.S. Govt. Printing Office), 487-502.

Mattey, D., Marsh, B., and Tarney, J., in press. The geochemistry,mineralogy, and petrology of basalts from the West Philippine andParece Vela basins and from the Palau-Kyushu and West Marianaridges, Deep Sea Drilling Project Leg 59. In Kroenke, L., Scott,R., et al., Init. Repts. DSDP, 59: Washington (U.S. Govt. Print-ing Office), 753-800.

Meijer, A., 1976. Pb and Sr isotope data bearing on the origin of vol-canic rocks from the Mariana island arc system. Geol. Soc. Am.Bull., 87(9): 1359-1369.

Melchior, J., and Hawkins, J., 1980. Petrology of Mariana back-arcbasin basalt. EOS, 61(46): 1143.

Melson, W. G., Valuer, T. L., Wright, T. L., et al., 1976. Chemicaldiversity of abyssal volcanic glass erupted along the Pacific, At-lantic and Indian Ocean sea-floor spreading centers. In Sutton,G. H., Manghnani, M. H., and Moberly, R. (Eds.), The Geophys-ics of the Pacific Ocean Basin and its Margins: Washington (Am.Geophys. Union), Mon. 19:351-367.

Miyashiro, A., Shido, F., and Ewing, M., 1970. Crystallization anddifferentiation in abyssal tholeiites and gabbros from mid-oceanridges. Earth Planet. Sci. Lett., 7:361-365.

O'Nions, R. K., Hamilton, P. J., and Evensen, N. M., 1977. Varia-tions in 143Nd/144Nd and 87Sr/86Sr ratios in oceanic basalts. EarthPlanet. Sci. Lett., 34:13-22.

Philpotts, J. A., Schnetzler, C. C , and Thomas, H. H., 1972. Petro-genetic implications of some new geochemical data on eclogite andultra basic inclusions. Geochim. Cosmochim. Acta, 36:1131-1166.

Sager, W. W., 1980. Structure of the Mariana Arc inferred fromgravity and seismic data. J. Geophys. Res., 85:5382-5388.

Saunders, A. D., Tarney, J., Weaver, S. D., et al., in press. Scotia Seafloor: Geochemistry of basalts from the Drake Passage and SouthSandwich spreading centers. In Craddock, C. (Ed.), AntarcticGeoscience: Madison (University of Wisconsin Press).

Schilling, J.-G., 1973a. Iceland mantle plume: Geochemical evidencealong Reykjanes Ridge. Nature (London), 242(540O):565-571.

, 1973b. Afar mantle plume: Rare-earth evidence. Nature(London) Phys. Sci., 142(114):2-5.

, 1975a. Azores mantle blob: Rare-earth evidence. EarthPlanet. Sci. Lett., 25:103-115.

., 1975b. Rare earth variations across "Normal Segments"of the Reykjanes Ridge, 6O°-53°N, Mid-Atlantic Ridge, 29°S, andEast Pacific Rise, 2°-19°S, and evidence on the composition ofthe underlying low-velocity layer. J. Geophys. Res., 80(11):1459-1473.

Schnetzler, C. C , and Philpotts, J. A., 1970. Partition coefficients ofrare-earth elements between igneous matrix material and rock

608

BASALTIC GLASSES

forming mineral phenocrysts—II. Gβochim. Cosmochim. Acta,34:331-340.

Schnetzler, C. C, Thomas, H. H., and Philpotts, J. A., 1967. Deter-mination of rare-earth elements in rocks amd minerals by massspectrometric, stable isotope dilution technique. Anal. Chem.,39:1888-1890.

Schuhmann, S., Philpotts, J. A., and Fryer, P., 1980. Suppression ofmonoxide interference in surface ionization mass spectrometry ofrare-earth elements. Anal. Chem., 52:214-216.

Scott, R., Kroenke, L. W., Zakariadze, G., et al., in press. Evolutionof the South Philippine Sea: Deep Sea Drilling Project Leg 59Results. In Kroenke, L., Scott, R., et al., Init. Repts. DSDP, 59:Washington (U.S. Govt. Printing Office), 803-816.

Sun, S. S., Nesbitt, R. W., and Sharaskin, A. Ya., 1979. Chemicalcharacteristics of mid-ocean ridge basalts. Earth Planet. Sci. Lett.,44:119-139.

Tarney, J., Saunders, A. D., and Weaver, S. D,, 1977. Geochemistryof volcanic rocks from the island arcs and marginal basins of the

Scotia Sea region. In Talwani, M., and Pitman, W. C. (Eds.),Island Arcs, Deep Sea Trenches and Back-Arc Basins: Washing-ton, (Am. Geophys. Union), pp. 367-377.

Toksoz, M. N., and Bird, P., 1977. Formation and evolution of mar-ginal basins and continental plateaus. In Talwani, M., and Pit-man, W. C. (Eds.), Island Arcs, Deep Sea Trenches and Back-ArcBasins: Washington (Am. Geophys. Union), pp. 379-393.

Uyeda, S., and Hussong, D. M, 1978. General results of DSDP Leg60, the eastern portion of the South Philippine Sea transect along18°N. EOS, 59(12): 1179. (Abstract)

Webster, R. K., 1960. Mass spectrometric isotope dilution analysis.In Smales, A. A., and Wager, L. R. (Eds.), Methods in Geochem-istry: New York (Interscience), pp. 202-246.

White, W. M., and Bryan, W B., 1977. Sr-isotopes, K, Rb, Cs, Sr,Ba, and rare-earth geochemistry of basalts from the FAMOUSarea. Geol. Soc. Am. Bull., 88(4):570-586.

Yoder, H. S., 1976. Generation of Basaltic Magma: Washington(National Academy of Sciences), p. 155.

609