Ti-V plots and the petrogenesis of modern and … - My Articles...Earth and Planetary Science...

18
Earth and Planetary Science Letters, 59 (1982) 101 - 118 101 Elsevier Scientific Publishing Company, Amsterdam - Printed in The Netherlands [31 Ti-V plots and the petrogenesis of modern and ophiolitic lavas John W. Shervais * Department of Geological Sciences, University of California, Santa Barbara, CA 93106 (U.S.A.) Received September 1, 1981 Revised version received February 8, 1982 Plots of Ti vs. V for many modern volcanic rock associations are diagnostic of tectonic setting and can be used to determine possible tectonic settings of ophiolites as well. The basis of this plot is the variation in the crystal/liquid partition coefficients for vanadium, which range with increasing oxygen fugacity from > 1 to << 1. Since the partition coefficients for Ti are almost always << 1, the depletion of V relative to Ti is a function of the fo2 of the magma and its source, the degree of partial melting, and subsequent fractional crystallization. Volcanic rocks from modern island arcs have Ti/V ratios of ~20, except for calc-alkaline volcanics which show the effects of magnetite fractionation. MORB and continental flood basalts have Ti/V ratios of about 20-50 and alkaline rocks have Ti/V generally >50. Back-arc basin basalts may have either arc-like or MORB-like Ti/V ratios, and sample suites from single back-arc basins may have Ti/V ratios ranging from 10 to 50. This range in Ti/V ratios in samples from a restricted geographical area may be diagnostic of the back-arc setting. The Ti/V plot is applied here to published data on ophiolites from a variety of postulated settings and in general supports the conclusions of previous investigators. Ophiolites from the western Mediterranean (Corsica, northern Apennines) and the "lower" Karmoy volcanics have Ti/V trends similar to MORB; the "upper" Karmo'y volcanics have alkaline Ti/V ratios. Lavas and tonalites in the Papuan ultramafic belt, the high-Mg andesites of Cape Vogel, and the upper pillow lavas at Troodos all have Ti/V ratios < 20, consistent with formation in an island arc setting. More specific evaluation of the tectonic setting of these and other ophiolites requires application of detailed geologic and petrologic data as well as geochemistry. The Ti/V discrimination diagram, however, is a potentially powerful adjunct to these techniques. 1. Introduction Ophiolites are commonly considered to repre- sent fragments of oceanic crust formed at spread- ing ridges, and emplaced tectonically in orogenic zones. Since only the uppermost parts of in situ oceanic crust are exposed or sampled by drilling, most models of ocean crust formation draw heavily on comparisons with ophiolite complexes. It has become clear, however, that ophiolites may form in a variety of tectonic settings, e.g., back-arc basins, island arcs, ocean islands and intra-arc rifts, as well as at mid-ocean ridges. Each of these * Present address: Department of Geology, University of California, Davis, CA 95616, U.S.A. settings will have profound significance in recon- structing the geologic history of the region in which the ophiolite is found, and also severely constrains the extent to which models of ocean crust formation may be based on specific ophiolite complexes. Because ophiolites are commonly in tectonic contact with adjacent terranes, and because low- temperature alteration and metamorphism are ubiquitous in ophiolite complexes, structural rela- tionships and major element geochemistry are not definitive of origin in most cases. Trace elements that are relatively immobile during metamorphism have been used successfully as discriminants of ophiolite petrogenesis [1,2] and that approach will be explored further in this paper. Data presented 0012-821 X/82/0000-0000/$02.75 ©1982 Elsevier Scientific Publishing Company

Transcript of Ti-V plots and the petrogenesis of modern and … - My Articles...Earth and Planetary Science...

Earth and Planetary Science Letters, 59 (1982) 101 - 118 101 Elsevier Scientific Publishing Company, Amsterdam - Printed in The Netherlands

[31

Ti-V plots and the petrogenesis of modern and ophiolitic lavas

John W. Shervais * Department of Geological Sciences, University of California, Santa Barbara, CA 93106 (U.S.A.)

Received September 1, 1981 Revised version received February 8, 1982

Plots of Ti vs. V for many modern volcanic rock associations are diagnostic of tectonic setting and can be used to determine possible tectonic settings of ophiolites as well. The basis of this plot is the variation in the crystal / l iquid partition coefficients for vanadium, which range with increasing oxygen fugacity from > 1 to << 1. Since the partition coefficients for Ti are almost always << 1, the depletion of V relative to Ti is a function of the fo2 of the magma and its source, the degree of partial melting, and subsequent fractional crystallization. Volcanic rocks from modern island arcs have T i / V ratios of ~20 , except for calc-alkaline volcanics which show the effects of magnetite fractionation. MORB and continental flood basalts have T i / V ratios of about 20-50 and alkaline rocks have T i / V generally >50. Back-arc basin basalts may have either arc-like or MORB-like T i /V ratios, and sample suites from single back-arc basins may have T i / V ratios ranging from 10 to 50. This range in T i / V ratios in samples from a restricted geographical area may be diagnostic of the back-arc setting. The T i / V plot is applied here to published data on ophiolites from a variety of postulated settings and in general supports the conclusions of previous investigators. Ophiolites from the western Mediterranean (Corsica, northern Apennines) and the "lower" Karmoy volcanics have T i / V trends similar to MORB; the "upper" Karmo'y volcanics have alkaline T i / V ratios. Lavas and tonalites in the Papuan ultramafic belt, the high-Mg andesites of Cape Vogel, and the upper pillow lavas at Troodos all have T i / V ratios < 20, consistent with formation in an island arc setting. More specific evaluation of the tectonic setting of these and other ophiolites requires application of detailed geologic and petrologic data as well as geochemistry. The T i / V discrimination diagram, however, is a potentially powerful adjunct to these techniques.

1. Introduction

Ophiolites are commonly considered to repre- sent fragments of oceanic crust formed at spread- ing ridges, and emplaced tectonically in orogenic zones. Since only the uppermost parts of in situ oceanic crust are exposed or sampled by drilling, most models of ocean crust formation draw heavily on comparisons with ophiolite complexes. It has become clear, however, that ophiolites may form in a variety of tectonic settings, e.g., back-arc basins, island arcs, ocean islands and intra-arc rifts, as well as at mid-ocean ridges. Each of these

* Present address: Department of Geology, University of California, Davis, CA 95616, U.S.A.

settings will have profound significance in recon- structing the geologic history of the region in which the ophiolite is found, and also severely constrains the extent to which models of ocean crust formation may be based on specific ophiolite complexes.

Because ophiolites are commonly in tectonic contact with adjacent terranes, and because low- temperature alteration and metamorphism are ubiquitous in ophiolite complexes, structural rela- tionships and major element geochemistry are not definitive of origin in most cases. Trace elements that are relatively immobile during metamorphism have been used successfully as discriminants of ophiolite petrogenesis [1,2] and that approach will be explored further in this paper. Data presented

0012-821 X/82 /0000-0000 /$02 .75 ©1982 Elsevier Scientific Publishing Company

102

here show that plots of Ti vs. V from many modern volcanic rock associations are diagnostic of tectonic setting and may by used to dis- criminate among possible tectonic settings of ophiolite volcanic rocks as well. This is because the crystal/liquid partition coefficients for V vary from > 1 to << 1 with increasing oxygen fugacity, mak- ing it a sensitive indicator of fo2 conditions. The depletion of V relative to the incompatible element Ti is a measure of this variation and thus the relative fo2 of the magma and its source. Recent data show that the volatile content (and thus fo, ) of magmas erupted in oceanic settings increases in the order: MORB < ocean islands < back-arc basins<is land arcs [3-5]. However, as we shall see, the actual T i / V relationships are somewhat complex, for reasons explored below.

Titanium and vanadium are present in basaltic and intermediate rocks in abundances that greatly exceed their detection limit in routine XRF analy- sis [6], making accurate determination reasonably easy. The problem of Ti-K~ interference on V-K,~ may be dealt with using interference standards, as discussed by Nisbet et ai. [6].

2. Trace element behavior

2.1 Theory

The use of immobile trace elements as discrimi- nants of ophiolite petrogenesis was pioneered by Pearce and Cann [1,2,7,8] and has since been applied by many others. Trace element discrimi- nant diagrams have been largely empirical until recently when papers by Saunders et al. [9], Pearce and Norry [10], and Pearce [8] have attempted to place them on a firmer theoretical basis. Saunders et al. [9] have proposed a division of mag- matophilic elements into two groups (excluding REE) based on the ionic character of the elements. Elements with large ionic radii, low charges and high radius/charge ratios are called "low-field strength" (LFS) cations. These cations, which are equivalent to the "LIL" cation group of Schilling [11], have been shown to be mobile during low- temperature alteration (e.g. [ 12,13]). This mobility renders the "LFS" (LIL) group of questionable

value in evaluating older, ophiolite-related volcanics. Elements having small ionic radii and low radius/charge ratios are called "high-field strength" (HFS) cations; they tend to be strongly incompatible, having very small bulk partition coefficients in most situations and are considered immobile during low-temperature alteration [9]. Elements whose cations are in this group include Zr, Hf, Ti, P, Nb and Ta; however, only Zr, Ti and P occur in basaltic rocks in concentration high enough to be determined accurately by XRF anal- ysis. Many of the transition metals (Ni, Co, Sc, Ti, V, Cr) are also immobile during alteration and metamorphism, but with the important exceptions of Ti (a HFS cation) and V, they are compatible elements which are moderately to strongly parti- tioned into refractory residual phases during par- tial melting and into early-liquidus mafic phases during crystallization. The geochemical behavior of Ti, which occurs terrestrially in only one va- lence state, is relatively well known and is a com- monly used discriminator (e.g. [2]). The remainder of this discussion will focus on the geochemical behavior of vanadium under differing conditions.

2.2 Vanadium

Vanadium like chromium, differs from the other trace transition metals (Ni, Co, Sc, Ti), in having three common valence states under terrestrial con- ditions that exhibit strongly contrasting geochemi- cal behavior. Table 1 compares the ionic charge, radius, and radius/charge ratio of vanadium to the incompatible HFS element titanium, and to the compatible transition metals, chromium, scandium, nickel and cobalt. Also shown for com- parison are aluminum and the mafic major ele- ments magnesium and iron. All of these elements have ionic radii in the range 0.6-0.8 A. The mafic major elements and the compatible trace transition metals have similar charges (+2 , +3) and rela- tively large radius/charge ratios (1>0.21). These similarities account for the strongly compatible character of these trace elements in the presence of refractory or fractionating mafic minerals, where they occur in octahedrally coordinated sites. In contrast, titanium, with its high charge and small radius/charge ratio, is a HFS cation [9] which has

103

T A B L E 1

Ionic characterist ics of vanadium, t i tanium and related species in octahedral coordinat ion , listed in order increasing rad ius / ' charge

rat io [61]

V 5~ V 4+ Ti 4~ A13~ Cr 3+ V 3+ Fe 3- Sc 3~ Ni 2+ Mg 2+ Co 2 ' Fe 2 '

Ionic radius (A) 0.62 0.67 0.69 0.61 0.70 0.72 0.73 0.83 0.77 0.80 0.83 0.86

R a d i u s / c h a r g e 0.12 0.17 0.17 0.20 0.23 0.24 0.24 0.28 0.39 0.40 0.42 0.43

very small crystal/liquid partition coefficients for most minerals, with the important exception of magnetite solid solutions [14-17]. Reduced vanadium (V 3+) has ionic characteristics similar to the compatible trace transition metals and com- monly substitutes for other trivalent cations in spinel and pyroxene. The more oxidized species (V 4+, V 5+) are HFS cations with high charges and low radius/ charge ratios (~<0.17), similar to titanium (Table 1).

Experimental determinations of the crystal/melt partitioning of vanadium by Lindstrom [18] has shown the D~ 1/1 for pyroxenes (both high and low calcium) and magnetite vary by over two orders of

magnitude as a function of oxygen fugacity. At low oxygen fugacities (Fe-Wti buffer, fo~ = 10 ~ to 10 12 at about 1200°C) V 3+ is the dominant species and partition coefficients are much greater than one; at high oxygen fugacities ( fo , = 10 4.5) V5 ÷ is the dominant species and partition coeffi- cients are much less than one (Table 2). Varying proportions of V 3+ , V 4+ and V 5+ are expected to exist under intermediate ,/o2 conditions. Table2 summarizes the experimentally determined parti- tion coefficients of Lindstrom [18] and some em- pirically derived partition coefficients from litera- ture.

T A B L E 2

C r y s t a l / l i q u i d part i t ion coeff ic ients

Augi te Low-Ca p y r o x e n e Ol iv ine Plagioclase Magnet i te Hornb lende

Titanium 1.0 (1100°C) ~ 0.6 b

0.5 (1220°C) " 0.3 b

0 .2-0 .8 e 0.14--0.5 e

Vanadium

5 ( logfo2 = -- 12) a 2.5 c

1 ( log fo2 = -- 10) ~ 0.5 c 0.05 (log fo2 = --6) a 0.025 c

0.024 <0 .0 5 ~ 9 16 e 4.3 6.8 f

(l 112-1134°C) a

0.03 --0 .0 g 0 . 1 1 ( I o g f o > - 4 . 2 ) a 6 .3-13 f

(1112-1134°C; 23 ( l o g f o = - 1 0 ) "

Q F M buffer) a 67 (Io_g f % = 13) ~' 23 -190 d

5 -1 0 0 e

~' L inds t rom [18].

b For same T as augite, using D~ px/cpx --0 .6 [18].

For same f o 2 as augite, using O~ xp/cpx =0 .5 [14].

d Ewart et al. [14]: basalt to andesite. e L e e m a n et al. [16,17]: ferrobasalt to ferrolatite. t Luhr and Carmichae l [26].

g Assumed,

104

2.3 Partial melting and fractional crystallization

The experiments of Lindstrom [18] confirm the predictions from ionic characteristics that vana- dium is a sensitive indicator of oxygen fugacity conditions during both partial melting and frac- tional crystallization. Fig. 1 shows partial melting curves based on a fixed bulk partition coefficient for titanium, and on bulk partition coefficients for vanadium which vary as a function of oxygen fugacity. The bulk partition coefficient for titanium (DT~=0.15) was calculated from the data in Table 2 and assumed refractory mantle mineralo- gies ranging from spinel lherzolite to harzburgite (see Appendix A for a more detailed discussion of the parameters and methods used in the construc- tion of Fig. 1). The bu]k partition coefficient for vanadium, /~v, for the same refractory mineralo- gies varies from ~ 0.15 (fo: ~ 10-7) to about 1.15 (fo:-~ 10-12) which spans the range of naturally- occurring oxygen fugacities in most magmatic sys- tems from the reducing conditions prevalent in the source of MORB to the more oxidizing conditions found in many convergent margin settings [3,4,19,20,25]. Curves are drawn for Dv=0.15 , 0.45, 0.75, 1.0 and 1.15. The curves have been calculated from the Berthelot-Nernst equilibrium melting equation assuming the bulk earth Ti and V abundances of Ganapathy and Anders [21]. These abundances are slightly higher than chondritic but have about the same Ti /V ratio (bulk earth Ti /V = 10; chondritic T i /V = 8.4-10 [22]) and are con- venient for illustration. Also shown are trend lines of constant initial T i /V for 20% and 30% partial melts. It is readily apparent (Fig. 1) that the Ti /V ratios of "primary" mantle melts increase with (a) decreasing fraction of melt produced and (b) in- creasing/ffv. Primary melts produced by 20-30% partial melting under relatively reducing condi- tions like MORB (fo~ = 10-It to 10 12; /~v ~> 0.6 ) will have initial T i /V ratios of about 20-50. Simi- lar melts produced under more oxidizing condi- tions (such as the mantle wedge overlying a de- volatizing, subducting oceanic slab) should have /ffv ~< 0.5 and initial T i /V ratios of around 10-20. Subsequent silicate fractionation in the latter case (high fo~) will parallel or sub-parallel the constant T i /V trend lines as long as D--vi and /)v remain

350

3OO

25O

V 2OO ppm

150

I00

50

T i /V= I 5 T i /V=IO /

t T i /V= 17.5

A~C ,

/.~ i/ / ~]lv= 2 0 / / T I / V = 25

i / t / t , / . .~5

' , ' ,;o ,T,,v:35

50 j / / 10 / / , 1" z.z 3__0--/ T i /M= 48

, .,

.~ EO O,,o., / ." ¢~ ' H / 10 ,/[~V= r5 / -

Earth . . ~ 504.0 39 29 / 19 . / 5V=l.0 /

M t - % ~ ~ ~ ~ i

s ~ Mt -D ~

T i p p m / 1 0 0 0

Fig. 1. T i /V plot showing partial melting and fractional crys- tallization trends. Melting curves (solid) are based on "bulk earth" Ti and V abundances (solid star; [2]), bulk equilibrium melting, /~Ti=0.15 and /fly=0.15, 0.45, 0.75, 1.0, and 1.15. Bulk partition coefficients are chosen to span the range of values calculated assuming spinel lherzolite and harzburgite refractory mineralogy and fo2--10 -7 to 10 12, using the crystal/liquid partition coefficients from Table 2 (see Appen- dix A for a detailed discussion of these calculations). Numbers adjacent to the partial melting curves equal % of melt. Initial T i /V ratios and their trends (dashed lines) for 20% and 30% melts are shown for three of the curves; also shown is the trend for T i / V = 100. Open s tar= C3 meteorite Ti and V abundances [22]. Fractionation under high fo2 conditions parallels or sub- parallels the constant T i /V ratio trends. Fractionation under lower fo2 conditions shown by the heavy solid line for the assemblage olivine-+plagioclase and for olivine-gabbro; un- derlined numbers represent % crystallization. Mt-C (six-point s tar)= 10% magnetite fractionation calculated assuming initial Ti=4600 ppm, V=302 ppm (Ti /V = 15) and O~ii t ---9, D~ t = 17 [16,17]. Mt-D (heavy arrows)=magnetite fractionation trend deduced from the trend of calc-alkaline series andesites and dacites. H = 10% hornblende fractionation assuming initial Ti = 4600 ppm, V=302 ppm, and oh~ =4.9, Ohv b =8.8 [26].

similar and less than one. This will be true of most fractionating assemblages under oxidizing condi- tions until the onset of magnetite or hornblende fractionation, which will strongly deplete both Ti and V [16,17,23,24-26].

The effects of magnetite fractionation are shown in Fig. 1, as calculated from magnetite/liquid par- tition coefficients (10% magnetite removal) and as deduced from the trend of calc-alkaline andesites and dacites. These paths are virtually identical and support the origin of these dacites by 10-15% magnetite fractionation. Hornblende/l iquid parti- tion coefficients are about half those of magnetite, with Dhv b / l~ 1.5--2.5*DThi h/l [26]. This means that hornblende and magnetite will have the same rela- tive effect on Ti and V concentrations, but that approximately twice as much hornblende as mag- netite must separate (Fig. 1). Recent studies of arc volcanism have shown that hornblende is in gen- eral a late crystalling phase, and is less important than magnetite in creating calc-alkaline trends (e.g. [25]). In any case, samples subject to either or both magneti te /hornblende fractionation will have negative trends on plots of Ti (or V) vs. SiO 2, Zr, or FeO*/MgO [23]; these samples will have anomalously high T i / V ratios and should not be used to discriminate tectonic setting on Ti-V plots.

In case of low-fo 2 conditions (MORB-type mag- mas) fractionation dominated by olivine and plagioclase will follow the constant initial T i /V trend lines because of their small crystal/liquid partition coefficients for Ti and V under any con- ditions. Clinopyroxene or spinel fractionation will cause curved fractionation trends that cross the constant T i / V ratio lines at low angles. Fig. 1 shows a hypothetical fractionation path of the assemblages ol, ol + pig, ol + pig + cpx (the crys- tallization sequence of most MORB) starting from a 30% partial melt with an initial T i / V ratio of 20. It should be noted that because /7 v is always I>/)'ri, fractionation cannot drive magmas to lower T i / V ratios than those inherited from the parental melt. This means that even if high fo~ conditions arise at some stage during the evolution of a mid-ocean ridge magma system (e.g., near a frac- ture zone), silicate fractionation will evolve mag- mas that have T i / V ratios the same as or greater than primary MORB.

105

An important implication of the chondritic or near chondritic trends in the T i / V ratios of island arc tholeiites is that their low Ti (and by analogy other HFS cation) abundances cannot be due to retention of Ti-rich minor phases (e.g., ilmenite, rutile) in the refractory residue unless that phase also retains V in chondritic proportion to Ti, or unless V is retained by other refractory phases in just the right modal proportions to maintain chondritic T i /V ratios. Neither of these possibili- ties seems likely. The low HFS cation contents of arc tholeiites are thus more probably due to greater degrees of partial melting, followed by extensive closed system fractional crystallization (e.g. [10]). Similar arguments may also apply to calc-alkaline magmas.

2.4. Stability of Ti and V during alteration and metamorphism

Since the use of trace element plots to investi- gate ophiolite petrogenesis involves the application of criteria based on more-or-less fresh samples to ancient rocks which may have undergone several episodes of alteration and metamorphism, it is necessary to evaluate briefly the effect of these processes on Ti-V distributions. These processes include low-temperature sea floor weathering ("halmyrosis"), greenschist facies hydrothermal al- teration and, possibly, subsequent regional meta- morphism. The effect of seawater/basalt interac- tion at low and high temperatures has been studied both experimentally (e.g. [27,28, and references therein]) and on natural samples from dredge hauls and DSDP cores (e.g. [28,29]). These studies have shown that Ti and V are stable over a wide range of temperatures and water / rock ratios, even in samples which have been converted entirely to chlor i te+ quartz [28]. Absolute concentrations may decrease due to dilution by hydration or by the precipitation of anhydrite or calcite, and may increase by passive accumulation as more mobil elements are leached (e.g., Si, Ca). The largest changes are observed during palagonitization of glassy pillow rims, where Ti and V abundances may drop by as much as 50% compared to fresh cores [29]. Alteration of the crystalline interiors of massive flows and pillows may either increase or

106

decrease Ti and V by 20% or more [29]. In most cases, Ti and V show coherent behavior during alteration, so that the T i / V ratio remains about the same in the altered rock as in the fresh rock. The data plotted in Fig. 2 (next section) include many altered samples, although relatively fresh rocks dominate. The altered samples plot in the same tectonic setting, suggesting that seawater al- teration effects on Ti and V are negligible.

Ophiolites may also undergo regional metamor- phism at intermediate to high grades either during or after emplacement. Recent studies of trace ele- ment mobility during amphibolite and granulite facies metamorphism suggest that Ti and V are relatively immobile under these conditions [30,31]. In summary, Ti and V appear to be stable over a wide range of temperatures during both seafloor and regional metamorphism. The glassy rims of pillows are most susceptible to changes in T i / V during alteration. However, this problem can be overcome by judicious sampling in the field.

3. Ti /V plots of modern volcanic rocks

There are four basic tectonic settings in which thick sections of basaltic rocks may accumulate: ocean basins, island arcs, back-arc basins and con- tinental interiors. Each of these settings may be further subdivided on a geochemical or geological basis (e.g., ridge basalts vs. oceanic islands). Alka- lic rocks may occur in any of these settings. Fig. 2 A - D presents the Ti and V abundances of sam- ples from a number of specific localities or regions representative of these settings and their subdivi- sions. These data have been compiled from the sources listed in Appendix B.

3.1 MORB

MORB are confined almost entirely to T i /V ratios between 20 and 50 regardless of whether they are "normal" or "enriched" in LREE and LIL (e.g. [32]; Fig. 2A). This is consistent with the model presented in the preceding section: 20-30% partial melting of a source with chondritic T i / V = 10, and with the difference ~[/~Ti - - / )v ] = --0.6 to --1.0 (Fig. 1). The crystallization sequence ob-

served in MORB (ol -+ sp, ol + plg, ol + plg + cpx) is also consistent with the hypothetical fractiona- tion path depicted in Fig. 1 for MORB: increasing Ti and V along the ratio inherited from the primary melt by olivine-+plagioclase fractionation, fol- lowed by a curved trajectory across the constant ratio lines during olivine-gabbro fractionation. There is a tendency for "enriched" MORB to have higher absolute abundances of both Ti and V than "normal" MORB, but both types span the same range in T i / V ratios. There is also a pronounced tendency toward regional variation: basalts from the Pacific and Indian Oceans extend to higher Ti and V abundances than those from the Atlantic ocean, and are seldom as depleted (Fig. 2A). The MORB field is overlapped in part by the field of back-arc basin basalts; this will be discussed fur- ther below.

3.2 Flood basalts

Tholeiitic flood basalts of the Columbia River plateau are similar to MORB, being derived by large degrees of partial melting (20-30%) under conditons of low volatile fugacities [33]. The Col- umbia River basalts have the same range in T i / V ratios as MORB, but tend to have higher absolute abundances (Fig. 2D). The contrasts in flow mor- phology and associated sediments between con- tinental flood basalts and MORB makes it un- likely that these settings will be confused, despite their similarity in T i / V ratios.

3. 3 Alkali basalts

Tholeiitic and alkalic basalts from both conti- nents and oceans plot in distinct fields with a minimum of overlap, separated by a T i / V ratio of about 50 (Fig. 2A, B; [34,35]). Since both tholeiitic and alkalic basalts form relatively deep in the mantle under low fo2 conditions, the difference 6[/~Ti--Ov] should be about the same in both cases. The consistent difference in T i / V ratios between tholeiitic and alkalic basalts may reflect either (a) primary differences in the T i / V ratios of their respective sources, (b) retention of contrast- ing refractory assemblages during partial melting of their sources, (c) derivation of alkali basalts

107

600 -

500 -

400 -

v

wm

300 -

0 ATLANTIC OCEAN (47) 0 I’.‘\ PACIFIC OCEAN (47)

0 JURASSIC ATLANTIC (14) : 0: PACIFIC SEAMOUNTS (20)

fi; INDIAN OCEAN (23) \/ * RED SEA AXIS (4)

.*

MORB

0 2 4 6 8 IO 12 14 16 I8

Ti ppm/ 1000

Fig. 2. Ti/V plots of modern volcanic rocks. showing fields for specific regions and trend lines of constant TI/V ratios= IO. 20. 50 and 100 for reference. Data sources are listed in Appendix B. A. MORB: Atlantic Ocean ( )I ~61); Pacific Occ~n (PI ~47); Indian

Ocean (II = 23) and the Red Sea axis basalta (II =4). Fields drawn by cyc to include the mqority of samples for each given region.

from smaller amounts of partial melting, or (d) varying degrees (e.g. [ 11,281).

formation of alkalic magmas by partial melting Alkali basalts in ocean basins may occur at under high CO, activities, which will reduce the spreading centers, but are most common off-axis

activities of both oxygen and water. It is not as oceanic islands and seamounts. Off-axis sills

possible to distinguish between these possibilities which intrude sediment in the Shikoku Basin

with Ti/V data, and all may be operative in (DSDP Leg 38, Hole 444; [36]) are either tholeiitic

108

V

ppm

600

500

4 0 0

300

200

I00

10

BACK-ARC BASIN BASALTS ( n = 6 6 ) • Lau Basin (22)

West Mariana Basin (8) 2 0 • Shikoku Basin (Basement) (26)

• Shikoku Basin (Sills) (10)

OCEAN ISLAND & ALKALIC BASALTS (77) Hawaii [] Tholeiites (30) O Alkalic Basalts (33) East Pacific Rise

Transitional (4) • ~ Alkalic Basalts (10)

/ / " . . "..-° ° ~ : ° ° \ ;~ o o \

• / ;2 "...- ~,,~°~..-°°o-~oooo ol • i , ,, • ~ - 7 " . ~ . " , o o o ~ Oo /

/ • I • • • / I o.11-- - ~ \ 7 t , v v

' / " . . . ; / ~ ' o ~ S ~

B

0 2 4 6 8 I0 12 14 16 18 20 22 24

Ti ppm/lO00 Fig. 2 (continued). B. Alkali basalts (n =43) and transitional tholeiites (n =34) from ocean islands and off-axis volcanic rocks are shown by open symbols; back-arc basin basalts (n =66) from the western Pacific are shown by closed symbols.

( T i / V = 30-50) or transitional to alkaline (T i /V = 35-70). Transi t ional basalts f rom small seamounts near the East Pacific Rise [37] have T i / V ratios of 40-50, while Hawaiian tholeiites [38] range from 42 to 60 and Hawaiian alkali basalts [39] range from 50 to 110 (Fig. 2B). This suggests that seamounts which form off-axis may

begin as tholeiitic basalts similar to "enriched" MORB and evolve toward more alkalic T i / V ratios and compositions [40].

3.4 Island arc lavas

Volcanic rocks from island arc-related settings may be divided into at least three suites [41]:

109

V

ppm

6 0 0

500

4 0 0

300

200

IOO

ARC T H O L E I I T E S •

N - - 1 6 0 10, 20

Z~

NEW HEBRIDES

Picrite-Ankaramite Suite

~ ' Tholeiites <57% SiO 2

Tholeiites >57%SIO 2

SO. SANDWICH

• < 56% SiO 2

>56% SiO 2 MARIANAS SEAMOUNTS

• <60% SiO 2

O >60% SiO 2

MARIANAS FORE-ARC

' ~ Basalts

' ~ Boninites

TONGA

• Whole Rock "~ <60% SiO It'! Matrix J 17 > 60% SiO 2

50

100

C

0 2 4 6 8 I0 12 14 16 18

Ti p p m l I 0 0 0

Fig. 2(continued). C. Island arc tholeiite series: basic rocks not affected by magnetite fractionation shown in solid symbols (n = 105); intermediate rocks which have been affected by magnetite fractionation shown by open symbols (n = 30); boninites from the Mariana fore-arc (n --6) shown by open five-point stars; ankaramites from the New Hebrides (n = 19) shown by solid six-point stars.

island arc tholeiite (Fig. 2C), calc-alkaline (Fig. 2D) and shoshonite (Fig. 2D). Recent work in many intra-oceanics arcs has shown that all grada- tions exist between tholeiitic and calc-alkaline suites [25]. These terms will be retained here, how-

ever, because they are representative of the gross geochemical variations possible within arc terrains and because they exhibit contrasting behavior with respect to Ti and V. The arc tholeiite and calc-al- kaline series are each represented by both basic

110

6 0 0

500

4OO

V

ppm

300

200

IOO

CALC-ALKALINE SUITE

SHOSHONITES

COLUMBIA RIVER FLOOD BASALTS (22)

,o/ ,o / e. O

/ O

O C

/'0 o o /

/ ' 0

/ ° i/ CALC-ALKALINE SUITE

~A SiO2 <58% >58% - - St. Kitts (18) • o

-- 7 A 100 Lau Ridge (15) • A / / ° / a ~ New Britain ( 1 2 ) . [] [ / .[3o ~ J Bagana (33) ,~ e~ - -

7.,,7D~ O / Fiji (8) "~ o

2 4 6 8 I0 12 14 16 18

Ti p p m / 1 0 0 0

Fig. 2 (continued). D. Calc-alkaline series volcanic rocks: solid and plus symbols have SiO 2 <58% (n =55); open symbols have SiO 2/>58% (n =22). Shoshonite series (n =8) shown by solid stars. Columbia River tholeiitic flood basalts (n -24) shown by circled stars. Note the effects of magnetite fractionation in the calc-alkaline series rocks, causing trends toward increasing Ti/V with increasing fractionation regardless of silica content. Magnetite fractionation can be detected by the use of supplementary plots, as described in the text.

( S I O 2 < 5 8 - 6 0 % ) and more fract ionated inter-

media te rocks (SiO 2 > 58-60%). Shoshonites are

all relatively basic. Other, less c o m m o n rock suites

which may occur in an island arc setting are the

bonini te suite [42,43] and the picr i te-ankaramite

association [44,45].

Arc tholeiite series. Arc tholeiites with silica < 60% plot almost exclusively on or near chondri t ic trends

and have T i / V ratios ~< 20 (Fig. 2C). There is

minor overlap with M O R B ratios (20-27) when V

is less than 350 ppm; however, suites of related

samples with a range of Ti and V abundances will

display the distinctive "chondritic" trend that clearly distinguishes arc tholeiites from MORB. The more evolved rocks of the arc tholeiite suite (SiO 2 > 58-60%) show the effects of titanomagne- tite fractionation: sudden drastic reduction in Ti and V abundances and increasing T i /V ratios. The effects of magnetite fractionation are dis- cussed in more detail below,

Calc-alkaline series. In contrast to the arc tholeiite series, calc-alkaline series eruptives have negative slopes on plots of Ti (or V) vs. Zr, FeO*/MgO, or SiO 2 regardless of their silica content, which im- plies magnetite fractionation throughout their evolution [23]. As a result, calc-alkaline volcanic rocks define trends on Ti-V diagrams (Fig. 2D) that are parallel to magnetite fractionation paths (Fig. 1) calculated from empirically-derived mag- neti te/l iquid partition coefficients [16,17]. The V abundances of both tholeiitic and calc-alkaline dacites are consistent with 1 0 - 1 5 % magnetite fractionation from basalts containing 300-350 ppm V (i.e., the same as the most V-rich calc-alkaline basalts shown in Fig. 2D). Magnetite fractionation causes T i / V ratios to increase, so that calc-al- kaline rocks have Ti//V~> 15 (Fig. 2D). The in- fluence of magnetite control can be easily ascertained by the use of supplementary plots of Ti and V against an independent fractionation index such as Zr, F e O , / M g O or SiO 2 [23]. Rock suites with negative slopes throughout their evolu- tion are calc-alkaline with obvious island arc affin- ities and the T i /V plot is not applicable or needed. Rock suites with positive slopes at low Zr (or SiO 2) and negative slopes at high Zr (or SiO 2) are tholeiitic, and samples from the positive slope may be plotted on the T i /V diagram to distinguish the nature of their original tectonic setting.

Shoshonites. Shoshonitic basalts from Fiji [46] have T i / V ratios and abundaces similar to arc tholei- ires, although T i / V may be as low as 4.5 (Fig. 2D). Vanadium generally decreases with increasing silica, but Ti varies irregularly so that the effect of magnetite fractionation is unclear. There is, how- ever, a rough positive correlation between in creas- ing T i /V and K 2 0 / N a 2 0 in the Fiji shoshonites

111

(Fig. 3). This coupled with the very low T i /V ratios ( ~ 1/2 chondritic) in samples with the lowest K 2 0 / N a 2 0 ratios, suggests that Ti and K 20 were selectively added to a source region that had been previously melted to create the low T i /V ratios, or that a previously unmelted phase rich in TiO 2 and K 2 0 (e.g., Ti-phlogopite) began to participate dur- ing progressive melting. Models similar to the latter have been proposed by Jakeg and White [41]. Shoshonites are clearly distinguished from intra- oceanic and continental alkali basalts by their arc-like T i / V ratios.

Boninites. Boninites are petrographically and geo- chemically distinctive rocks which have been found only in the fore-arc region of modern oceanic arcs [42,43]. They have low Ti//V ratios of about 10 or less, but with lower Ti and V abundances than most arc tholeiites (Ti ~< 2500 ppm; V ~< 200 ppm).

2O

Ti v

I0

0 1 I I I I I I I

.4 .6 .8 1.0 1.2

K20/Na20

Fig. 3. T i /V vs. K20/Na20 for the Fiji shoshonites [46]. The positive correlation between these ratios (and the lack of corre- lation between TiO 2 and SiO2) implies either (a) that TiO 2 and K 20 were selectively added to the source region or (b) that a phase rich in TiO 2 and K 2 0 (like Ti-phlogopite) began to participate during progressive partial melting.

112

This is consistent with their proposed origin by hydrous melting of a source previously depleted by arc tholeiite formation [42,43]. Meijer [43] has described more fractionated rocks from the Marianas which appear to be derived from a boninite parent. These "boninite series" rocks would be expected to have Ti and V abundances somewhat higher than true boninites, but still de- pleted relative to equivalent arc tholeiites.

Ankaramites. The picrite-ankaramite association has been documented in several island arcs, most notably at New Georgia in the Solomons arc [44], and throughout the New Hebrides arc (e.g. [45]). These olivine and clinopyroxene-rich lavas are sub-alkaline to mildly alkaline and appear to be related to extensional tectonics within the Solomons and New Hebrides arcs. Ankaramites from the New Hebrides arc have T i / V ratios ranging from 7 to 27, but most are between 10 and 20 (Fig. 2C). Ankaramites have the same range in Ti and V concentrations, and the same "chondritic" trend on the Ti-V plot, as the island arc tholeiites with which they are commonly associated.

3.5 Back-arc basins

Basaltic rocks from back-arc basins [9, 47] de- fine a field on the T i / V plot that overlaps both MORB and island arc volcanic rocks, although MORB-like abundances are most common (Fig. 2B). This tendency to show overlapping or transi- tional geochemical characteristics has been noted by many investigators for other geochemical data (e.g. [9,47]). It appears that the conditions of melt- ing and fractionation vary between the two ex- tremes of oxygen fugacity in a back-arc basin setting. However, back-arc basin basalts tend to show less enrichment in Ti and V than is seen in either MORB or island arc basalts. This may be related to slow, diffuse spreading centers in back- arc basins, which lack well-developed magma chambers [47].

3.6. Summary

It is clear from the foregoing discussion and from Fig. 2 that T i / V ratios alone are not suffi-

cient to define the tectonic setting of any given volcanic rock in most cases. More important are the trends defined by suites of related samples that exhibit a range of Ti and V abundances. The T i / V plot is also poorly constrained (a) after the onset of magnetite (-+ hornblende) fractionation, and (b) for volcanic rocks of the calc-alkaline suite. The first difficulty may be circumvented by first plot- ting ei ther /or both Ti, V against an independent fractionation index, e.g., Zr, FeO*/MgO or SiO2, and omitting those data points that indicate mag- netite control. This will also reveal whether the rocks are calc-alkaline or tholeiitic [23]. The diffi- culties with calc-alkaline rocks are not serious, since calc-alkaline series volcanic rocks in the geo- logic record are seldom mistaken for ophiolites, and they are easily distinguished by their negative slopes on Ti vs. Zr or SiO 2 plots. Shoshonites are more problematical, but are generally spacially associated with calc-alkaline rocks, and have arc- related T i / V and Zr abundances (e.g. [46]). Island arc tholeiites, boninites and ankaramites, however, may form "ophiolite" assemblages. The T i / V plot promises to be useful in distinguishing these from MORB ophiolites and ocean islands. Distinguish- ing back-arc basin related ophiolites will require careful study of local and regional geology as well as geochemistry.

4. Application of Ti/V plots to ophiolite voicanics

In testing the applicability of a new discrimi- nant diagram to older ophiolite terrains, it is de- sirable to apply it first to familiar, well-studied examples whose tectonic settings are reasonably well-constrained by previous geochemical and geo- logic investigations. Unfortunately, vanadium is not commonly determined (or reported) for many of these (e.g., Samail, Bay of Islands). Published examples for which vanadium analyses are availa- ble include: Corsica [48]; the northern Apennines [49]; Troodos [50] (ranges and means only re- ported by [48]); the K a r m ~ Complex in Norway [51]; and Papua-New Guinea [52,53]. These data are presented in Fig. 4.

Western Mediterranean. The ophiolites of Corsica and the northern Apennines have volcanic rocks

113

V

ppm

500

400

300

200

IOO

v v

10 / $ / 3 0

II OTL / I i t;" . . . . . . . . . - ,

I ! l / WESTERN MEDETERRANEAN

• • /" • i • N. Apennines < • Corsica

I KARMOY COMPLEX [] Ophiolite Volcanies A Off-Axis Volcanics

/ TROODOS OPHIOLITE PAPUA Means Ranges v PUB Volcanics

(~ ~ TU: Upper Volcanics O PUB Tonalites • ~ 2 qJ TL: Lower L . . . . • Cape Vogel HMA

0 2 4 6 8 I0 12 14 16 18 20 22 24 26

Ti p p m / 1 0 0 0

Fig. 4. Ti-V plots of ophiolites. Ophiolite volcanic rocks from Corsica [48], the northern Apennines [49] and the Karme~y Complex in Norway [51] have Ti /V ratios between about 30 and 55, except for rare alkaline basalts in the western Mediterranean. The strong linear trends in these suites imply fractionation dominated by olivine + plagioclasc (see Fig. I). These ratios and trends arc consistent with their origin as MORB (see text). Alkaline basahs which are intercalated with sediments that overlie the Karme~y ophiolitc have r i / v > 50, consistent with their proposed origin as an oceanic island [51]. V analyses for Troodos are unpublished [50], but the means and ranges reported by Beccaluva et al. [48] are shown here: T L - Troodos lower pillow lavas: T U - Troodos upper pillow lavas: large circles respective means. The Troodos upper pillow lavas and the lavas and tonalites of the Papuan uhramafic belt [52] and the high-Mg andesites of Cape Vogel [53J, all have T i / V < 20, which implies their formation near a convergent plate boundary (e.g., within an island arc, or in a small back-arc basin). Ranges reported for the Troodos lower pillow lavas [50] are too large for definite interpretation, but the low mean (-- 1 I) implies origin in a setting similar to the Troodos upper lavas.

w i t h T i / V r a t io s b e t w e e n 30 a n d 50, w i t h r a r e

e x a m p l e s of a l k a l i n e b a s a l t w i t h T i / V > 5 0

(F ig . 4). T h e s e T i / V r a t i o s a n d t r e n d s a re c o n -

s i s t en t w i th a n o r i g i n as M O R B , a n i n t e r p r e t a t i o n

s u p p o r t e d b y t he p r i m a r y i n v e s t i g a t o r s [8,48,49,

54,55]. P e a r c e [8] ha s a l so n o t e d t he t e n d e n c y

t o w a r d a l k a l i n i t y in t he se r o c k s w h i c h he a t t r i -

b u t e s to t h e i r o r ig in in s m a l l o c e a n b a s i n s a l o n g a

r i f t e d c o n t i n e n t a l m a r g i n .

Troodos. T r o o d o s , in the e a s t e r n M e d i t e r r a n e a n ,

h a s b e e n a c o n t r o v e r s i a l s u b j e c t s ince M i y a s h i r o

[55] f i rs t s u g g e s t e d t h a t it was r e l a t e d to i s l a n d a rc

v o l c a n i s m . S u b s e q u e n t d a t a a p p e a r to s u p p o r t h is

114

contention (e.g. [7,8,42,56]), especially in regard to the "Upper Pillow Lavas". The mean and ranges of Ti and V from the "Upper" and "Lower" volcanic sequences are shown in Fig. 4 [48,49]. The means for both lava groups plot on the chondritic trend of Ti//V ~ 10, however, the ranges for the lower volcanic sequence are too large to draw any definite conclusions. The ranges of Ti and V from the upper pillow lavas are restricted to ratios close to 10, suggesting that these rocks at least were formed in a high volatile environment such as an island arc or intra-arc basin. This conclusion is consistent with Pearce [7,8], Cameron et al. [42], and Dick et al. [56], but at odds with older inter- pretations [57,58].

KarrraCy. The Caledonian Karmery Complex in Norway consists of two greenstone units: a lower unit of ophiolitic lavas (Visne Group) and an upper unit of alkaline basalts (Vikingstad Green- stone), which is intercalated with 700 m of hemi- pelagic sediments [51]. The Visne and Vikingstad units have been interpreted by Sturt et al. [51] to represent ocean crust and oceanic island rocks, respectively. This conclusion is supported by their Ti and V abundances, which show MORB and alkaline characteristics, respectively (Fig. 4). Note the distinct linearity sub-parallel to constant T i / V ratios exhibited by the K a r n ~ y lava suites, and by the western Mediterranean ophiolite lavas, which suggest that fractionation of these lavas was dominated by olivine and plagioclase.

Papua-New Guinea. The high-magnesium andesites (HMA) of Cape Vogel are similar to both komati- ites and boninites in their low abundances of incompatible elements and high abundances of MgO, Cr and Ni [53]. These rocks have very low absolute abundances of Ti ( < 2600 ppm) and V (~< 211 ppm) and low to very low T i / V ratios, ranging from 4.3 to 16.7 (9.3 average) [53]. These rations and abundance levels are consistent with their origin by relatively large percentages of fu- sion of a previously depleted source, as suggested by Jenner [53] for the Cape Vogel rocks, and by others for boninites from Bonin-Marianas arc [42,43]. This refractory source was enriched in LREE shortly before the second melting event

[53]. Jenner [53] claims that this component has an oceanic island or continental affinity, and is not similar to that found in island arcs. We have seen, however, that alkalic rocks and transitional tholei- ites from both ocean basins and continents have high T i / V ratios (/> 35 about). Arc-related alkalic rocks (shoshonites) have low T i / V ratios similar to arc tholeiites. This suggests that the Cape Vogel high Mg-andesites are related to some phase of island arc volcanism, and that by inference the LREE component added may have been derived from a subduction related source.

The Papuan ultramafic belt (PUB) comprises a full ophiolite sequence which has been intruded by Eocene tonalities [52]. The tonalities have T i / V ratios of 10-20 and very low abundances ( T i < 2000 ppm; V < 150 ppm), consistent with their proposed origin as early island arc magmas related to the HMA of Cape Vogel [52]. Ophiolite basalts of the PUB have T i / V ratios of 15-18 and V abundances of 450-500 ppm. There is no overlap between MORB and arc tholeiites at these abun- dance levels, and the island arc affinities of these basalts are clearly demonstrated. Jaques and Chappell [52] inferred that the PUB volcanics were most similar to MORB, but noted many dissimi- larities from MORB as well, such as low Ti, Zr and H R E E at high FeO*/MgO, and the presence of cumulate orthopyroxene in the plutonic section. Geophysical evidence suggests that the PUB formed in a marginal basin [59,60]. If so, it must have been a small basin close to the subduction zone since the T i / V ratios imply hydrous melting of the mantle source. The high V abundances and the presence of cumulate orthopyroxene are more suggestive of arc tholeiite magmas than of most "back-arc" basin basalts.

5. Conclusions

The data presented here show that plots of Ti vs. V clearly distinguish arc related tholeiites, MORB and alkali basalts. The T i / V ratio of indi- vidual samples is not as diagnostic as the trends defined by suites of related samples that exhibit a range in Ti and V abundances. The theoretical basis of this plot can be modelled assuming a

chondr i t i c mant le (for Ti and V), reasonable re- f ractory minera l assemblages, and bulk par t i t ion coefficients for V which vary as a funct ion of oxygen fugacity. The results of this model l ing are consis tent with what is known abou t the physical condi t ions under which each m a g m a series is gen- e ra ted and with exper imenta l de te rmina t ions of c r y s t a l / l i q u i d par t i t ion coefficients for vanad ium and t i tanium. Calc-a lka l ine magmas m a y be gener- a ted under condi t ions s imilar to arc tholeiites, but they are d o m i n a t e d by magnet i te f rac t iona t ion th roughout their evolut ion causing var iable T i / V rat ios which over lap M O R B and alkali basal t rat ios at low abundance levels. This makes it necessary to use supp lemen ta ry plots of V a n d / o r Ti vs. Zr,

F e O * / M g O o r S i O 2 in order to dis t inguish the ca lc-a lkal ine series (which is def ini te ly arc-related) , and to e l iminate tholei i te samples which are d o m i n a t e d by magnet i te f rac t ionat ion. Shoshoni tes have T i / V rat ios s imilar to arc tholeiites, which serves to dis t inguish them from other basal ts . Al- though V vs. SiO 2 plots imply magnet i te control , the source of shoshoni tes appears to have been enr iched in Ti, so that Ti vs. SiO 2 plots are incon- clusive. The only cr i ter ia by which "back-a rc" bas in basal ts may be recognized on this plot are var iable T i / V rat ios in a suite of samples from a single geographic area, and poss ib ly by a lack of en r ichment in ei ther Ti or V, relat ive to the other m a g m a series.

The use of V as a d i sc r iminant offers ad- vantages over plots based on less a b u n d a n t ele- ments (e.g., Ta, La, Nb, Ce, etc.) because it is easily analyzed by rout ine X R F techniques (e.g. [6]), having concent ra t ions in basal t ic rocks (50 - 600 ppm) which great ly exceed the detec t ion limit ( ~ 3 ppm). App l i ca t i on of this technique to ophio l i te volcanic rocks yields in te rpre ta t ions which are consis tent with conclusions based on o ther geochemical techniques and on recent petro- logic and s t ructura l invest igat ions.

The p rob l ems associa ted with t rans i t ional tectonic sett ings such as back-a rc basins or incipi- ent ocean islands, however, require the app l ica t ion of careful geologic, pe t ro logic and s t ructural inves- t igat ions in add i t ion to various geochemical meth- ods to e lucidate the origin of ophiol i tes formed in these settings.

115

Acknowledgements

This work is an outgrowth of research on the trace e lement character is t ics of Cord i l le ran ophio- lites begun when the au thor was a N A T O post- doc tora l fellow at the Swiss Federa l Polytechnic Ins t i tu te (ETH) in Ztirich under the sponsorsh ip of Prof. V. Trommsdor f f . I wish to thank V. Diet r ich (ETH-Zt i r ich) for my in t roduc t ion to trace e lement analysis and s t imula t ing discussions on ophio l i te origins; and D.L. K imbrough (Univers i ty of Cal i forn ia at Santa Barbara) for discussion and cri t ical review of an earl ier version of the manuscr ip t . C.A. Hopson , D.J. L inds t rom and two o ther reviewers con t r ibu ted valuable comment s which improved the manuscr ip t .

Appendix A

The bulk partition coefficients ( 4 ) for Ti and V used to construct Fig. 1 were calculated from the relationship:

D, = a , D7 / ' + f l , Di ~/' + 3 , D , a/' + . . . + t~.,D,* '

where a=weight proportion of phase a, such that a + fl + 8 + ... + ~ = 1.0, and D, '~/I =partition coefficient for element i between phase a and the liquid. Two refractory mineralogies were assumed: spinel lherzolite (65% ol, 20% opx, 12% cpx, 3% Cr-spinel) and spinel harzburgite (70% ol, 24% opx, 3% cpx, 3% Cr-spinel). These should span the range in the refractory source regions of arcs and MORB, but probably not that of alkali basalts. Spinel/liquid partition coefficients for Ti (-- 1.2) and V (26.5) were estimated from an unpublished spinel-whole rock pair (an arc-related ankaramite, corrected for phenocryst con- tent). Other partition coefficients are from Lindstrom [18]; most of these values are summarized in Table 2. /gTi was calculated from the high-temperature values in Table 2: lherzo- lite Dvi 20.17; harzburgite /~yi ~0.14; an "average"~0.15 is assumed since residual harzburgite is more likely for moderate to large degrees of melting. Dv was calculated from the data below for both spinel lherzolite and spinel harzburgite at three nominal oxygen fugacities, and are also listed below:

l o g [ o - 8 10 12

D~, ~ 0.03 0.03 0.04 D~'.P ~/I 0.05 0.50 2.5(1 D~P ~/I 0.10 I.(X) 5.(X1 D ~ 6.50 6.50 6.5(1 D v (harzburgite) 0.23 0.37 0.97 /5 v (lherzolite) 0.24 0.44 1.32

/~v D-rl (----0.151 is assumed to be the limiting case for high fo: conditions. Due to uncertainties in (a) the experimental

116

data, (b) the estimated spinel/liquid partition coefficients, (c) the assumed refractory assemblages, and (d) the actual values of fo2 during partial melting in the upper mantle, these calcu- lated values were used as guidelines in constructing Fig. 1, rather than exact values. The D- v values used to construct Fig. 1 (Dv =0.15, 0.45, 0.75, 1.0, 1.15) were chosen to span the range of calculated values but with the melting curves sufficiently spaced to be viewed clearly. Melting curves were calculated from the batch melting equation:

' -- ( I - - ~ ) F ] C] : C , / [ D i +

where C] = concentration of element i in the liquid; C, ' = initial concentration of i in the source; Di =bulk partition coefficient for i in the refractory residue, and F = % partial melt. Strictly speaking, this equation applies to modal melting and thus represents a grossly oversimplified model. However, in view of the other uncertainties involved, the use of more complex models is unwarranted and for a first approximation, unneces- sary. Fractionation paths are based on Rayleigh fractionation:

Ci I : COl(O, - I)

where C] = concentration of element i in the liquid, C, ° = the initial concentration of i in the magma, and f = fraction of melt remaining.

A p p e n d i x B

Ash, R.P., Carney, J.N. and MacFarlane, A., 1980. Geology of the northern Banks islands. Reg. Rep. Geol. Surv. New Hebrides, 52 pp.

Baker, P.E., 1968. Petrology of Mt. Misery Volcano, St. Kitts, West Indies. Lithos, I : 124-150.

Baker, P.E., 1978 The South Sandwich Islands, III. Petrology of the volcanic rocks. Br. Antart. Surv., Sci. Rep., 93: 1-26.

Bryan, W.B., Frey, F.A. and Thompson, G., 1977 Oldest Atlantic seafloor. Contrib. Mineral. Petrol., 64: 223-242.

Carney, J.N. and MacFarlane, A., 1979. Geology of Tanna, Aneityum, Futuna and Aniwa. Reg. Rep. Geol. Surv. New Hebrides, 71 pp.

Clague, D.A. and Beeson, M.H., 1980. Trace element geochem- istry of the East Molokai volcanic series, Hawaii. Am. J. Sci., 280-A: 820-844.

Coleman, R.G., Fleck, R.J., Hedge, C.E. and Ghent, E.D., 1977. The volcanic rocks of southwestern Saudi Arabia and the opening of the Red Sea. In: Red Sea Research 1970- 1975. Mineral Resour. Bull. 22: DI-D30.

Colley, H. and Ash, R.P., 1971. The Geology of Erromango. Reg. Rep. Geol. Surv. New Hebrides, 111 pp.

Dietrich, V.J., Emmerman, R., Oberhanski, R. and Puchelt, H., 1978. Geochemistry of basaltic and gabbroic rocks from the West Mariana basin and the Mariana Trench. Earth Planet. Sci. Lett., 39: 127-144.

Dixon, T.H. and Stern, R.J., 1982. Volcanic seamounts in the southern Marianas island arc. Geol. Soc. Am. Bull. (in press).

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

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

Fleet, A.J., Henderson, P. and Kempe, D.R.C., 1976. Rare earth element and related chemistry of some drilled south- ern Indian Ocean basalts and volcanogenic sediments. J. Geophys. Res., 81: 4257-4268.

Frey, F.A., Bryan, W.B. and Thompson, G., 1974. Atlantic Ocean floor: geochemistry and petrology of basalts from Legs 2 and 3 of the Deep Sea Drilling Project. J. Geophys. Res., 79: 5507-5527.

Gill, J.B., 1970. Geochemistry of Viti Leva, Fiji, and its evolu- tion as an island arc. Contrib. Mineral. Petrol., 27: 179-203.

Gorton, M.P., 1977. The geochemistry and origin of Quaternary volcanism in the New Hebrides. Geochim. Cosmochim. Acta, 41: 1257-1270.

Hawkins, J.W., 1976. Petrology and geochemistry of basaltic rocks of the Lau Basin. Earth Planet. Sci. Lett., 28: 283-297.

Hubbard, N., 1967. Some trace elements in Hawaiian lavas. Ph. D. Thesis, University of Hawaii, 123 pp.

Johnson, J.R., 1979. Transitional basalts and tholeiites from the East Pacific Rise, 9°N. J. Geophys. Res., 84: 1635-1651.

Kempe, D.R.C., 1974. The petrology of the basalts, Leg 26. In: Initial Reports DSDP, 26: 465-503.

Kempe, D.R.C., 1976. Petrological studies on DSDP Leg 34 basalts: Nazca plate, eastern Pacific Ocean. In: Initial Re- ports DSDP, 34: 189-213.

Langmuir, C.H., Bender, J.F., Bence, A.E., Hanson, G.N. and Taylor, S.R., 1977. Petrogenesis of basalts from the FA- MOUS area: Mid-Atlantic Ridge. Earth Planet. Sci. Lett., 36: 133-156.

Leeman, W.P., Budahn, J.R., Gerlach, D.C., Smith, D.R. and Powell, B.N., 1980. Origin of Hawaiian tholeiites: trace element constraints. Am. J. Sci., 280-A: 794-819.

Lowder, G.G. and Carmichael, I.S.E., 1970. The volcanoes and caldera of Talasea, New Britain: geology and petrology. Geol. Soc. Am. Bull., 81: 17-38.

MacDonald, G.A. and Katsura, T., 1964. Chemical composi- tion of Hawaiian lavas. J. Petrol., 5: 82-133.

McDougall, I., 1976. Geochemistry and origin of basalt of the Columbia River Group, Oregon and Washington. Geol. Soc. Am. Bull., 87: 777-792.

Muir, I.D. and Tilley, C.E., 1964. Basalts from the northern part of the rift zone of the Mid-Atlantic Ridge. J. Petrol., 5: 409-434.

Nisterenko, G.V., 1980. Petrochemistry and geochemistry of basalts in the Shikoku Basin and Daito Basin, Philippine Sea. In: Initial Reports DSDP, 58: 791-804.

Sun, S.S. and Nesbitt, R.W., 1978. Geochemical regularities and genetic significance of ophiolitic basahs. Geology, 6: 689-693.

Sun, S.S., Nesbitt, R.W. and Sharaskin, A.Y., 1979. Geochemi- cal characteristics of mid-ocean ridge basalts. Earth Planet, Sci. Lett., 44: 119-138.

117

Thompson, G., Bryan, W.B., Frey, F.A., Dickey, J.S. and Suen, C.J., 1976. Petrology and chemistry of basalts from DSDP Leg 34, Nazca plate. In: Initial Reports DSDP, 34: 215-226.

Wood, D.A., Tarney, J., Varet, J., Saunders, A.D., Bougault, H., Joron, J.L., Trenil, M. and Cann, J.R., 1979. Geochem- istry of basalts drilled in the North Atlantic by IPOD Leg 49: implications for mantle heterogeneity. Earth Planet. Sci. Lett., 42: 77-97.

References

1 J.A. Pearce and J.R. Cann, Ophiolite origin investigated by discriminant analysis using Ti, Zr and Y, Earth Planet Sci. Lett. 12 (1971) 339-349.

2 J.A. Pearce and J.R. Cann, Tectonic setting of basic volcanic rocks determined using trace element analyses, Earth Planet. Sci. Lett. 19 (1973) 290-300.

3 J.B. Gill, Composit ion and age of Lau Basin and Ridge volcanic rocks: implications for evolution of an interarc basin and remnant arc, Geol. Soc. Am. Bull. 87 (1976) 1384-1395.

4 M.O. Garcia, D.W. Muenow and C.D. Byers, Volatiles in spreading center lavas: mid-ocean ridges vs. back-arc basins. Abstr., Am. Geophys. Union, Chapman Conf. on Oceanic Lithosphere ( 1981 ).

5 P. Fryer, J.M. Sinton and J.A. Philpotts, Basalts from active back-arc basins, a comparison with MORB, Abstr., Am. Geophys. Union, Chapman Conf. on Oceanic Lithosphere (1981).

6 E.G. Nisbet, V.J. Dietrich and A. Esenwein, Routine trace element determination in silicate minerals and rocks by X-ray fluorescence, Fortschr. Mineral. 57 (1979) 264-279.

7 J.A. Pearce, Basalt geochemis t~ used to investigate past tectonic environments on Cyprus, Tectonophysics 25 (1975) 41-67.

8 J.A. Pearce, Geochemical evidence for the genesis and eruptive setting of lavas from Tethyan Ophiolites, Proc. Int. Ophiolite Symp., Cyprus (1979) 261 272.

9 A.D. Saunders, J. Tarney, N.G. Marsh and D.A. Wood, Ophiolites as ocean crust or marginal basin crust: a geo- chemical approach, Proc. Int. Ophiolite Symp., Cyprus (1979) 193-204.

10 J.A. Pearce and M.G. N o r ~ , Petrogenetic implications of Ti, Zr, Y, and Nb variations in volcanic rocks. Contrib. Mineral. Pctrol. 69 (1979) 33-47.

I 1 J.G. Schilling, Iceland mantle plume: geochemical evidence along the Reykjanes Ridge, Nature 242 (1973) 565 571.

12 D.A. Wood, I.L. Gibson and R.N. Thompson, Elemental mobility during zeolite facies metamorphism of Tertiary basalts of eastern Iceland, Contrib. Mineral. Petrol. 55 11976) 24t-254.

13 J.N. Ludden and G. Thompson, An evaluation of the behaviour of the rare-earth elements during weathering of sea-floor basalt, Earth Planet. Sci. Lett. 43 (1970) 85-92.

14 A. Ewart, W.B. Bryan and J.B. Gill, Mineralogy and geo-

chemistry of the younger volcanic islands of Tonga, S.W. Pacific, J. Petrol 14 (1973) 429-465.

15 J.J. Dudas, M.E. Harward and R.A. Schmitt, Identification of dacitic tephra by activation analysis of their primary mineral phenocrysts, Quaternary Res. 3 (1973) 307-315.

16 W.B. Leeman, M.S. Ma, A.V. Murali and R.A. Schmitt, Empirical estimation of magneti te/ l iquid distribution coef- ficients for some transition elements, Contrib. Mineral. Petrol. 65 (1978) 269-272.

17 W.B. Leeman, M.S. Ma, A.V. Murali and R.A. Schmitt, A correction, Contrib. Mineral. Petrol. 66 (1978) 429.

18 D.J. Lindstrom, Experimental study of the partitioning of the transition metals between clinopyroxene and coexisting silicate liquids, Ph.D. Dissertation, University of Oregon (1976) 188 pp. (unpublished).

19 R.F. Fudali, Oxygen fugacities of basaltic and andesitic magmas, Geochim. Cosmochim. Acta 29 (1965) 1063-1075.

20 I.S.E. Carmichael, The mineralogy of Thingmuli, a Tertiary' volcano in eastern Iceland, Am. J. Sci. 276 (1967) 309-329.

21 R. Ganapathy and E. Anders, Bulk compositions of the moon and earth, estimated from meteorites, Proc. Fifth Lunar Sci. Conf., Suppl. 5, Geochim. Cosmochim. Acta 2 (1974) 1181-1206

22 B. Mason, Handbook of Elemental Abundances in Meteorites (Gordon and Breach, New York, (N.Y. 1971) 555 pp.

23 A. Miyashiro and F. Shido, Tholeiitic and calc-alkalic series in relation to the behaviors of titanium, vanadium, chro- mium and nickel, Am. J. Sci. 275 (1975) 265-277.

24 S.R. Taylor, M. Kaye, A.J.R. White, A.R. Duncan and A. Ewart, Genetic significance of Co, Cr, Ni, Sc and V content of andesites, Geochim. Cosmochim. Acta 33 (1969) 275-286.

25 J.B. Gill, Orogenic Andesites and Plate Tectonics (Springer-Verlag, Berlin, 1981 ) 390 pp.

26 J.F. Luhr and I.S.E. Carmichael. The Cotina volcanic com- plex, Mexico, Contrib. Mineral. Petrol. 71 (1980) 343-372.

27 M.J. Mottl, H.D. Holland and R.F. Corr, Chemical ex- change during hydrothermal alteration of basalt by seawater, II. Experimental results for Fe, Mn, and sulfur species, Geochim. Cosmochim. Acta 43 (t979) 869-884.

28 M.J. Mottl, Metabasalts, axial hot springs and the structure of hydrothermal systems at mid-ocean ridges, Geol. Soc. Am. Bull. (in review).

29 G. Thompson, A geochemical study of the low-temperature interaction of sea-water and oceanic igneous rocks. Trans. Am. Geophys. Union 54 (1973) 1015 1019.

3(1 C. Nicollet and D.R. Andrianbololona, Distribution of transition elements in crustal metabasic igneous rocks, Chem. Geol. 28 (1980) 79-90.

31 B.L. Weaver and J. Tarney, Chemical changes during dyke metamorphism in high-grade basement terrains, Nature 289 ( 1981 ) 47-49.

32 W.B. Bryan, G. Thompson, F.A. Frey and J.S. Dickey, Inferred geologic settings and differentiation in basalts from the Deep Sea Drilling Project, J. Geophys. Res. 81 (1976) 4285 -4304.

118

33 I. McDougall, Geochemistry and origin of basalt of the Columbia River Group, Oregon and Washington, Geol. Soc. Am. Bull. 87 (1976) 777-792.

34 S.Y. Wass, Geochemistry and origin of xenolith-bearing and related alkali basaltic rocks from the southern high- lands, New South Wales, Australia, Am. J. Sci. 280-A (1980) 639-666.

35 C.H. Langmuir, J.F. Bender, A.E. Bence, G.N. Hanson and S.R. Taylor, Petrogenesis of basalts from the FAMOUS area: Mid-Atlantic Ridge, Earth Planet. Sci. Lett. 36 (1977) 133-156.

36 G.V. Nisterenko, Petrochemistry and geochemistry of basalts in the Shikoku Basin and Daito Basin, Philippine Sea, in: Initial Reports DSDP 58 (1980) 791-804.

37 J.R. Johnson, Transitional basalts and tholeiites from the East Pacific Rise, 9°N, J. Geophys. Res. 84 (1979) 1635- 1651.

38 W.P. Leeman, J.R. Budahn, D.C. Gerlach, D.R. Smith and B.N. Powell, Origin of Hawaiian tholeiites: trace element constraints Am. J. Sci. 280-A (1980) 794-819.

39 D.A. Clague and M.H. Beeson, Trace element geochemistry of the East Molokai volcanic series, Hawaii, Am. J. Sci. 280-A (1980) 820-844.

40 R. Batiza, Age, volume, compositional and spatial relations of small isolated oceanic central volcanoes, Mar. Geol. 24 (1977) 169-183.

41 P. Jake~ and A.J.R. White, Major and trace element abun- dances in volcanic rocks of orogenic areas, Geol. Soc. Am. Bull. 83 (1972) 29-40.

42 W.E. Cameron, E.G. Nisbet and V.J. Dietrich, Petrographic dissimilarities between ophiolitic and ocean floor basalts, Proc. Int. Ophiolite Syrup., Cyprus (1979) 182-192.

43 A. Meijer, Primitive arc volcanism and a boninite series: examples from the western Pacific island arcs, in: The Tectonic Evolution of Southeast Asian Seas and Islands, Hayer, ed., Am. Geophys. Union, Geophys. Monogr. (1980).

44 R.L. Stanton and J.D. Bell, Volcanic and associated rocks of the New Georgia Group, British Solomon Islands Pro- tectorate, Overseas Geol. Mineral. Resour. l0 (1969) 113- 145.

45 A.J. Warden, Evolution of Aoloa caldera volcano, New Hebrides, Bull. Volcanol. 34 (1970) 107-140.

46 J.B. Gill, Geochemistry of Viti Leva, Fiji, and its evolution as an island arc, Contrib. Mineral. Petrol. 27 (1970) 179-203.

47 J.W. Hawkins, Petrology of back-arc basins and island arcs:

their possible role in the origin of ophiolites, Proc. Int. Ophiolite Syrup., Cyprus (1979) 244-254.

48 L. Beccaluva, D. Ohnenstetter, M. Ohnenstetter and G. Venturelli, The trace element geochemistry of Corsican ophiolites, Contrib. Mineral. Petrol. 64 (1977) I 1-3 I.

49 G. Ferrara, F. Innocenti, C.A. Ricci and G. Serri, Ocean- floor affinity of basalts from North Apennine ophiolites: geochemical evidence, Chem. Geol. 17 (1976) 101-111.

50 A. Desmet, Contribution a l'6tude de la Croflte Oceanique Mesozoique de Medeterann6e Orientale: les pillows lavas due Troodos (Chypre), Thbse de Specialit+, Universit6 du Nancy.

51 B.A. Sturt, A. Thon and H. Fumes, The geology and preliminary geochemistry of the Karm~y ophiolite, Norway, Proc. Int. Ophiolite Symp., Cyprus (1979) 538-554.

52 A.L. Jaques and B.W. Chappell, Petrology and trace ele- ment geochemistry of the Papuan ultramafic belt, Contrib. Mineral. Petrol. 75 (1981) 55-70.

53 G.A. Jenner, Geochemistry of high-Mg andesites from Cape Vogel, Papua-New Guinea, Chem. Geol. 33 ( 1981 ) 307-332.

54 G. Venturelli, R.S. Thorpe and P.J. Potts, Rare-earth and trace element characteristics of ophiolitic metabasalts from the Alpine-Apennine belt, Earth Planet. Sci. Lett. 53 (1981) 109-123.

55 A. Miyashiro, The Troodos ophiolite complex was probably formed in an island arc, Earth Planet. Sci. Lett. 19 (1973) 218-244.

56 H.J.B. Dick, R.L. Fisher and T. Bullen, Mineralogy and composition of abyssal versus alpine-type peridotites (in press).

57 I.G. Gass and J.D. Smewing, Intrusion, extrusion and meta- morphism at constructive margins: evidence from the Troodos Massif, Cyprus, Nature (London) 242 (1973) 26-29.

58 J.D. Smewing, K.O. Simonian and I.G. Gass, Metabasalts from the Troodos Massif: genetic implication deduced from petrography and trace element geochemistry, Contrib. Mineral. Petrol. 51 (1975)49-61.

59 D.E. Karig, Remnant arcs, Geol. Soc. Am. Bull. 83 (1972) 1057-1068.

60 D.M. Finlayson, B.J. Drummond, C.D. Collins and J.B. Connelly, Crustal structures in the region of the Papuan ultramafic belt, Phys. Earth Planet. Int. 14 (1977) 13-29.

61 E.J.W. Whittaker and R. Muntas, Ionic radii for use in geochemistry, Geochim. Cosmochim. Acta 34 (1970) 945- 956.