Dacite Petrogenesis on Mid-Ocean Ridges: Evidence for Oceanic Crustal Melting and Assimilation

Dacite Petrogenesis on Mid-Ocean Ridges:Evidence for Oceanic Crustal Meltingand Assimilation

V. D.WANLESS1,2*, M. R. PERFIT1,W. I. RIDLEY3y AND E. KLEIN4

1DEPARTMENT OF GEOLOGICAL SCIENCES, UNIVERSITY OF FLORIDA, GAINESVILLE, FL 32611, USA2DEPARTMENT OF GEOLOGY AND GEOPHYSICS, WOODS HOLE OCEANOGRAPHIC INSTITUTION, WOODS HOLE,

MA 02540, USA3US GEOLOGICAL SURVEY, DENVER, CO 80225, USA4NICHOLAS SCHOOL OF THE ENVIRONMENT, DUKE UNIVERSITY, DURHAM, NC 27708, USA

RECEIVED NOVEMBER 23, 2009; ACCEPTED AUGUST 26, 2010

Whereas the majority of eruptions at oceanic spreading centers

produce lavas with relatively homogeneous mid-ocean ridge basalt

(MORB) compositions, the formation of tholeiitic andesites and

dacites at mid-ocean ridges (MORs) is a petrological enigma.

Eruptions of MOR high-silica lavas are typically associated with

ridge discontinuities and have produced regionally significant vol-

umes of lava. Andesites and dacites have been observed and sampled

at several locations along the global MOR system; these include pro-

pagating ridge tips at ridge^transform intersections on the Juan de

Fuca Ridge and eastern Gala¤ pagos spreading center, and at the

98N overlapping spreading center on the East Pacific Rise. Despite

the formation of these lavas at various ridges, MOR dacites show

remarkably similar major element trends and incompatible trace

element enrichments, suggesting that similar processes are controlling

their chemistry. Although most geochemical variability in MOR bas-

alts is consistent with low-pressure fractional crystallization of vari-

ous mantle-derived parental melts, our geochemical data for MOR

dacitic glasses suggest that contamination from a seawater-altered

component is important in their petrogenesis. MOR dacites are char-

acterized by elevated U,Th, Zr, and Hf, low Nb andTa concentra-

tions relative to rare earth elements (REE), and Al2O3, K2O, and

Cl concentrations that are higher than expected from low-pressure

fractional crystallization alone. Petrological modeling of MOR

dacites suggests that partial melting and assimilation are both inte-

gral to their petrogenesis. Extensive fractional crystallization of a

MORB parent combined with partial melting and assimilation of

amphibole-bearing altered crust produces a magma with a

geochemical signature similar to a MOR dacite.This supports the

hypothesis that crustal assimilation is an important process in the

formation of highly evolved MOR lavas and may be significant in

the generation of evolved MORB in general. Additionally, these pro-

cesses are likely to be more common in regions of episodic magma

supply and enhanced magma^crust interaction such as at the ends

of ridge segments.

KEY WORDS: assimilation; dacite; fractional crystallization;

mid-ocean ridge; MORB; partial melting

I NTRODUCTIONFast to intermediate oceanic spreading centers typicallyerupt geochemically diverse basaltic lavas (e.g. Klein,2005); however, a much more extensive range of lava com-positions, including ferrobasalts and FeTi basalts as well asrarer high-silica andesites and dacites have been recovered(Perfit et al., 1983; Langmuir et al., 1986; Natland &MacDougall, 1986; Natland et al., 1986; Regelous et al.,1999; Smith et al., 2001). High-silica lavas have erupted onseveral ridges and are commonly associated with specifictectonic settings; these include propagating ridge tips(Christie & Sinton, 1981; Perfit & Fornari, 1983; Fornariet al., 1983), overlapping spreading centers (OSCs)(Christie & Sinton, 1981; Perfit et al., 1983; Sinton et al.,

*Correspondingauthor.Telephone: (352) 392-2231. Fax: (352) 392-9294.E-mail: [email protected] address: National Science Foundation, Ocean DrillingProgram, Arlington,VA 22230, USA

� The Author 2010. Published by Oxford University Press. Allrights reserved. For Permissions, please e-mail: [email protected]

JOURNALOFPETROLOGY VOLUME 51 NUMBER12 PAGES 2377^2410 2010 doi:10.1093/petrology/egq056 at U

niversity of Florida on D

ecember 5, 2010

petrology.oxfordjournals.orgD

ownloaded from

http://petrology.oxfordjournals.org/

1983; Bazin et al., 2001), regions of ridge^hotspot inter-action (Chadwick et al., 2005; Haase et al., 2005), and at10830’N on the East Pacific Rise near the ridge^transformintersection (Regelous et al., 1999). These wide variationsin composition are commonly attributed to low magmasupply and/or cooler crust at ridge segment ends, or thecold edge effect, which promote greater differentiation ofmagmas prior to eruption (Christie & Sinton, 1981; Perfitet al., 1983; Sinton et al., 1983; Perfit & Chadwick, 1998;Rubin & Sinton, 2007).The formation of highly evolved, silicic magmas in

non-ridge settings (e.g. ocean islands, arc volcanoes, andcontinental interiors) has been attributed to various pro-cesses, including crystal fractionation, partial melting ofoverlying crust, and/or assimilation of crustal materialinto an evolving magma chamber (e.g. Bowen, 1928; DePaolo, 1981; Hildreth, 1981). On mid-ocean ridges (MORs),many studies have documented the dominant role crystalfractionation plays in magma differentiation (e.g. Clague& Bunch,1976; Bryan &Moore,1977; Byerly,1980) wherebyextensive crystallization of olivine (Ol), plagioclase (Plag),pyroxene (Px) and Fe^Ti oxides leads to the generation ofhighly evolved melts enriched in SiO2 and depletedin MgO, FeO and TiO2 (e.g. Juster et al., 1989). However,although crystal fractionation is undoubtedly a primaryprocess involved in the differentiation of most MORBmagmas, it may not be the only mechanism involved inthe formation of high-silica MOR andesites and dacites.Partial melting (or anatexis) of basaltic crustal material

may produce evolved compositions, particularly in settingswhere magma^rock interactions are likely, such as the topof an axial magma chamber (e.g. Coogan et al., 2003b;Gillis, 2008). It has been suggested as the origin forhigh-silica lavas erupted on many ocean islands (e.g.Iceland; O’Nions & Gronvold, 1973; Sigurdsson & Sparks,1981; Martin & Sigmarsson, 2007; Galapagos Islands;McBirney, 1993; Socorro Island; Bohrson & Reid, 1997;Bohrson & Reid, 1998; Hawaii; Van Der Zander et al.,2010). This process may also explain the formation ofhigh-silica lavas in back-arc settings, most recently withinthe Lau Spreading Center (e.g. Kent et al., 2002) andManus Basin (Sinton et al., 2003), although the influenceof a subduction zone in these tectonic settings makesmagma genesis much more complicated.Evidence from ophiolites suggests that the top of the

axial magma chamber at MORs is a dynamic boundarywhere magmas may interact with and melt differentlayers of crustal material, including both gabbros andsheeted dikes (Coogan et al., 2003b). Recent experimentalevidence suggests that partial melting of hydrous gabbroicrock similar to that in the lower ocean crust can form sili-cic magmas (Koepke et al., 2004; Kvassnes & Grove, 2008)and may explain the presence of highly evolved plagiogra-nite veins in the ocean crust (Koepke et al., 2004; Brophy,

2009). Other studies indicate that low degrees of dehydra-tion partial melting of altered basalt, similar in compos-ition to dikes of the upper ocean crust, can producedacitic melts (Beard & Lofgren, 1991). These experimentalstudies suggest that oceanic crust will begin to melt at tem-peratures as low as 850^9008C and 510% melting ofthe crust will yield dacitic or tonalitic melts (Beard &Lofgren, 1991; Koepke et al., 2004; Kvassnes & Grove,2008). Kvassnes & Grove (2008) stated that mineral pairs(plagioclaseôlivine and plagioclaseâugite) similar tooceanic gabbros from the lower crust will melt quicklyand easily at temperatures similar to that of primitiveMOR magmas (1220^13308C). All of these studies indicatethat partial melting of ocean crust can produce high-silicamelts at MORs, but the role that this process may play inthe formation of extrusive silicic lavas on the seafloor hasnot yet been assessed.The compositional variability observed in arc and

continental volcanic rocks is commonly ascribed to theassociated processes of assimilation and fractional crystal-lization (AFC; e.g. Bowen, 1928; De Paolo, 1981). Similarprocesses may also occur where thickened oceanic crustleads to magma^crust interaction; for instance, withinIcelandic volcanoes (e.g. Nicholson et al., 1991). On smallerscales, the combined effects of these processes have beenobserved in ophiolites, where sub-axial intrusive magmashave been in contact with and have partially melted theoverlying sheeted dikes (Gillis & Coogan, 2002; Coogan,2003; Gillis, 2008; France et al., 2009). AFC processes havealso been invoked to explain the high Cl concentrationsobserved in some MORBs (Michael & Schilling, 1989;Michael & Cornell, 1998; le Roux et al., 2006). During thisprocess, a magma undergoes crystal fractionation, and theresultant latent heat of crystallization provides the heatneeded to partially melt the surrounding wall-rock. Thesemelts are then assimilated into, and homogenized with,the fractionating magma reservoir. AFC processes canproduce a wide range of rock types depending on the ini-tial composition of the intruding magma, the degree ofcrystal fractionation, the initial wall-rock composition,and the amount of melting and assimilation.High-silica compositions are found throughout the

ocean crust and are commonly observed as intrusive orplutonic material. As mentioned above, plagiogranites area ubiquitous component of the ocean crust and have beenobserved as small intrusions and veins in ophiolites (e.g.Pedersen & Malpas, 1984), drill cores from the ocean crust(Casey, 1997; Dick et al., 2000; Wilson et al., 2006), and asxenoliths in Icelandic lavas (Sigurdsson, 1977). The originof these silicic rocks remains unclear but two main hypoth-eses are (1) partial melting of gabbroic crust (e.g. Koepkeet al., 2004, 2007; Nunnery et al., 2008) and (2) extremecrystal fractionation of tholeiitic basalt magmas(Coleman & Donato, 1979; Beccaluva et al., 1999; Niu

JOURNAL OF PETROLOGY VOLUME 51 NUMBER 12 DECEMBER 2010

2378

at University of F

lorida on Decem

ber 5, 2010petrology.oxfordjournals.org

Dow

nloaded from


et al., 2002). There are many examples of highly evolvedplutonic rocks from slower spreading centers (e.g.Mid-Atlantic Ridge, Aumento, 1969; South West IndianRidge, Dick et al., 2000), which may suggest that AFC orpartial melting processes are occurring on much smallerscales, deeper in the ocean crust.In this study we examine the geochemistry of high-silica

lavas from three MORs, the East Pacific Rise, Juan deFuca Ridge, and Gala¤ pagos Spreading Center (Fig. 1), andshow they have remarkably similar major and trace elem-ent compositions (Fig. 2), suggesting that similar sourcesand processes control their petrogenesis. More specifically,we examine the roles that crystal fractionation, partialmelting, and AFC may have played in the formation ofan exceptional suite of high-silica lavas from the 98NOSC on the East Pacific Rise, and evaluate if these resultsapply generally to the formation of high-silica lavas onother MORs. We focus on the petrogenesis of dacites atthe 98N OSC because it is the most complete and geo-logically well-constrained dataset available; however,descriptions of the geological settings of high-silica lavasin the other environments are important to ascertain

the role the tectono-magmatic setting may play in theirpetrogenesis.

GEOLOGICAL AND TECTONICSETT ING98N East Pacific Rise overlappingspreading centerThe 98N OSC (Fig. 1a) is located on the East Pacific Risebetween the Clipperton and Siqueiros transform faults.It consists of two north^south-trending ridges that overlapby �27 km and partly enclose a large overlap basin(Sempere & Macdonald, 1986). The limbs are separatedby �8 km (Singh et al., 2006).The eastern limb is propagat-ing to the south into older crust at a rate of �42 kmMyr^1 (Carbotte & Macdonald, 1992).The 98N OSC has been the focus of several geophysical

studies (Detrick et al., 1987; Harding et al., 1993; Kentet al., 1993, 2000; Bazin et al., 2001; Dunn et al., 2001; Tonget al., 2002); including the first 3D multi-channel seismicsurvey of a MOR (Kent et al., 2000) and a 3D seismic re-fraction study (Dunn et al., 2001). These studies resulted in

Fig. 1. Bathymetric maps showing the tectonic setting of the MOR dacites discussed in the text (data from GeoMapApp; Carbotte et al., 2004).Boxes show the general locations of high-silica lavas on each ridge. Dacites are commonly associated with the ends of ridge segments, such asoverlapping spreading centers (OSCs) and propagating ridge tips at ridge^transform intersections. (a) 98N OSC on the East Pacific Rise(EPR), (b) propagating ridge tip on the Juan de Fuca Ridge and Axial Seamount and (c) possible OSC near the propagating ridge tip on theGala¤ pagos Spreading Center (GSC).

WANLESS et al. DACITE PETROGENESIS ON MORs

2379

at University of F

lorida on Decem


Dow

nloaded from


the first 3D image of a subsurface magma chamber along aMOR, which showed that a shallow melt lens lies beneathboth limbs of the OSC with an anomalously large meltlens in the interlimb region, north of the overlap basin.This suggests that the region currently has an unusuallyhigh magma supply rate for a ridge segment end (Kentet al., 2000).High-silica andesites and dacites were recovered from

the eastern limb during the MEDUSA2007 cruise(AT15-17) using the ROV JasonII (Wanless et al., 2008;

White et al., 2009). Several high-silica lavas were also re-covered from this area during dredging operations in thelate 1980s (Langmuir et al., 1986). The siliceous lavas areprimarily confined to the northern section of theneo-volcanic zone on the eastern, propagating limb, alongthe eastern edge of the melt lens (Fig. 3). Morphologically,the dacites form large single bulbous to elongate pillowsthat can each be several meters in diameter (Fig. 4). Thepillows are highly striated and have a coarse bread-crustsurface texture. Typically, the pillows are stacked into

Fig. 1. Continued.


2380

at University of F

lorida on Decem


Dow

nloaded from


mounds, which can be several meters high, or construc-tional domes. Dacites largely occur in two regions: as anearly linear pillow mound in the center of the east limbneo-volcanic zone and as large, elongate pillow lavas onthe flanks of the axial graben (Fig. 3). Their axial andnear-axial positions, low sediment cover and unalterednature suggest they are relatively young.

Juan de Fuca Ridge propagating ridgetip and Axial SeamountThe Cleft segment is the southernmost segment of theJuande Fuca Ridge (Fig. 1b). It terminates at 44827’N at aridge^transform intersection, where it intersects and over-laps the Blanco Transform Zone (Embley et al., 1991;Embley & Wilson, 1992; Smith et al., 1994). This intersec-tion is characterized by a series of curved ridges that over-shoot the Blanco Transform Zone onto the older Pacificplate (Stakes et al., 2006).High-silica andesites and dacites form two small con-

structional domes on the Pacific plate, where the axialridge intersects and is believed to propagate past theBlanco Transform Zone into the older ocean crust (�6·3Ma) that was created at the Gorda Ridge (Embley &Wilson, 1992; Stakes et al., 2006). The domes are �20^30m high and 200^500 m in diameter and were sampledusing a rock core and the ROV Tiburon during researchcruises in 2000 and 2002 (Cotsonika, 2006).

High-resolution bathymetric maps show that there arenumerous other constructional domes in the region butthey have not yet been sampled, although we surmise thatthey are also composed of high-silica lavas. Rare andesiteshave also been recovered within the axis and along thebounding faults of the southern Cleft segment (Stakeset al., 2006).Seismic studies of the southern Juan de Fuca Ridge indi-

cate the presence of an axial magma chamber beneathmost of the Cleft segment (Canales et al., 2005). However,an axial magma chamber reflector is absent south of44838’N where the high-silica lavas were recovered, sug-gesting the presence of small melt volumes resulting fromweak melt supply to the ridge^transform intersection.Zircon thermochronology and U-series data indicate thatthe dacites erupted less than 30 kyr ago (Perfit et al., 2007).The axial segment of the Juan de Fuca Ridge is a

second-order ridge segment that currently overlies theCobb hotspot, which has produced a chain of seamountstrending NW away from the ridge axis (Chadwick et al.,2005). The Juan de Fuca Ridge is migrating NW at a rateof 3·1cm a^1 and has been situated above the Cobb hotspotfor the last�0·2^0·7Myr, creatinga large on-axis seamount,known as Axial Seamount (Karsten &Delaney,1989).Axial Seamount is the largest feature on this segment of

the Juan de Fuca Ridge and has a large summit caldera(Embley et al., 1999) underlain by a large seismically

Fig. 1. Continued.


2381

at University of F

lorida on Decem


Dow

nloaded from


Fig. 2. Comparison of major and trace element compositions in MOR high-silica andesites and dacites from the East Pacific Rise (OSCDacite), Juan de Fuca Ridge (JdF Dacite), Gala¤ pagos Spreading Center (GSC Dacite) and Axial Seamount (Axial Andesite). MOR daciticlavas have similar major and trace element compositions, whereas andesites have more variable compositions that lie between dacites andhighly evolved MORBs. (a) Mantle-normalized diagram showing similarities in trace element compositions between dacites from the threeridges and an andesite from Axial Seamount on the Juan de Fuca Ridge. Average composition for N-MORB from the 98170^108N segment ofthe East Pacific Rise (EPR) is shown for comparison. MOR dacites are characterized by low Nb and Ta and high U, Th, Zr and Hf relativeto elements of similar incompatibilities. (b, c) Major element variation diagrams showing the range of compositions of the MOR dacites com-pared with East Pacific Rise MORB. Gray field shows the range of compositions from41200 analyses of MORB glasses from the segment ofthe East Pacific Rise north of the 98NOSC (from PetDB, Perfit et al., 1994; Sims et al., 2002, 2003; our unpublished data).


2382

at University of F

lorida on Decem


Dow

nloaded from


Fig. 3. Bathymetric map of the 98N OSC showing the locations of samples collected during the MEDUSA cruise in 2007. Samples aredivided into rock types based on silica content (dacite462wt % SiO2; andesite 57^62wt % SiO2; basaltic andesite 52^57wt % SiO2; basalt552wt % SiO2 and FeTi basalt552wt % SiO2 and412wt % FeO). Dacites are primarily found on-axis on the eastern limb of the OSC. The50 m contour intervals are shown.


2383

at University of F

lorida on Decem


Dow

nloaded from


imaged axial magma chamber (West et al., 2001). It has twoprominent rift zones, extending to the north and south,which create bathymetric highs. Extensive sampling of themain edifice shows that it is composed of moderatelyevolved and slightly enriched MORB (Chadwick et al.,2005). The rift zones have linear ridges that appear to ac-commodate extensive diking from the main calderasystem (Chadwick et al., 2005). Rare high-silica andesitessampled by three rock cores (Chadwick et al., 2005) arelocated east of the northern rift zone and may be asso-ciated with dike propagation from the main axial magmachamber into older ridge crust.

Gala¤ pagos Spreading Centerçextinct OSCor propagating ridge tip?High-silica lavas were sampled at the eastern end of theGala¤ pagos Spreading Center at �858W. The area was ex-tensively studied though dredging and Alvin explorationin the early 1980s (Fornari et al., 1983; Perfit & Fornari,

1983; Perfit et al., 1983; Embley et al., 1988). The evolvedlavas erupted within the axial valley and along theaxis-bounding faults of the Gala¤ pagos Spreading Center�20 km east of the ridge^transform intersection with theInca transform fault (Fig. 1c). Bathymetric data reveal twocurved ridges surrounding a depression within this region,which has been interpreted as an old, small, extinct OSCor deviation from axial linearity (Embley et al., 1988;Perfit et al., 1999) that has been rifted away from theneo-volcanic zone. Most of the evolved lavas (�63%SiO2) at the Gala¤ pagos Spreading Center were foundoff-axis along the bounding faults associated with thesouthern portion of the extinct OSC.

PETROGRAPHYThe 98N OSC dacites are glassy and predominantly aphy-ric, with sparse microphenocrysts and very rare, smallphenocrysts of plagioclase and clinopyroxene. The small

Fig. 4. Photographs of MOR high-silica lavas from 98NOSC East Pacific Rise (a, b), Gala¤ pagos Spreading Center (c) andJuan de Fuca Ridge(d). Morphologically, the dacites typically form blocky angular flows and large elongate pillows with roughly corrugated or striated surfaces.


2384

at University of F

lorida on Decem


Dow

nloaded from


phenocrysts of plagioclase commonly have resorbed edgesand sieve textures. Several of the samples contain smallbasaltic xenoliths composed of subophitic plagioclase andclinopyroxene surrounded by dacitic or andesitic glass.Juan de Fuca Ridge dacites typically are microcrystal-

line to nearly aphyric and also contain rare basaltic xeno-liths and xenocrysts. Unlike spatially associated MORB,which have low vesicularity, the dacites are moderatelyvesicular with 10^15% (by volume) elongate vesicles(millimeter-sized to 10 cm long). More crystalline samplescontain microphenocrysts of ferroaugite^hedenbergite andferropigeonite^ferrosilite, and lesser amounts of sodicplagioclase and FeTi oxides set in a glassy matrix withcrystallites (�20^30%) of plagioclase and pyroxene withrare fayalite, quartz and zircon crystals. Clinopyroxenemicrophenocrysts exhibit both normal and reverse zoningwith rims that have the same composition as groundmasspyroxenes. Titanomagnetite is common in most samples,whereas ilmenite is significantly less abundant (Cotsonika,2006).The petrology and mineral chemistry of andesites and

dacites from the Gala¤ pagos Spreading Center have beendescribed in detail by Perfit & Fornari (1983). Many ofthe lavas are nearly aphyric and extremely glassy withmoderate vesicularity (55 vol. %). In the more crystallineandesites rare phenocrysts and microphenocrysts ofaugite^ferroaugite, pigeonite, FeTi oxides and intermediatecomposition plagioclase predominate. Apatite crystalsare present but extremely rare. Similar to the Juan deFuca high-silica lavas, calcic plagioclase and magnesianclinopyroxene xenocrysts are present in the Gala¤ pagosSpreading Center andesites.

ANALYTICAL METHODSGlass chips from the outer rims of 18 dacites collected atthe 98N OSC were analyzed using a JEOL 8900 ElectronMicroprobe for major element concentrations at theUSGS in Denver, Colorado (Table 1). Eight to 10 pointswere analyzed per sample and averages are reported inTable 1. USGS mineral standards were used to calibratethe microprobe and secondary normalizations were doneto account for instrument drift using the JdF-D2 glass‘standard’ (Reynolds, 1995), University of Florida in-housestandard ALV 2392-9 from the East Pacific Rise(Smith et al., 2001) and USGS standard dacite glass GSC(for more detail on methods see Smith et al., 2001). Thebeam diameter during glass analyses was 20 mm, and anaccelerating voltage of 15 keV and a beam current of 20nAwere used. High-precision chlorine and potassium con-centrations were also determined by microprobe on sevenof the dacites using 200 s peak and100 s background count-ing times. St 7820 Sodalite was used as the chlorinestandard.

Small glass fragments (10^50mm) were handpicked,avoiding microphenocrysts and alteration, cleaned, anddissolved for trace element and isotope analyses followingmethods described by Goss et al. (2010). Fourteen dacitesfrom the 98N OSC were analyzed for trace elementconcentrations by inductively coupled plasma mass spec-trometry (ICP-MS) at medium resolution using a high-precision Element2 system at the University of Florida(Table 1). Radiogenic isotope ratios (Pb, Sr, and Nd) weredetermined for 10 dacites by multi-collector ICP-MS usinga Nu-Plasma system at the University of Florida Centerfor Isotope Geoscience (Table 2). A detailed description ofsample preparation, dissolution procedures, standards,and errors has been given by Kamenov et al. (2007) andGoss et al. (2010). External calibration was done to quantifyresults using a combination of in-house basalts (ENDVçEndeavour and ALV 2392-9) and USGS (AGV-1, BIR-1,BHVO-1, BCR-2 and STM-1) rock standards (Kamenovet al., 2007; Goss et al., 2010).

GEOCHEMICAL RESULTSMajor elementsMajor element compositions of the 98N OSC high-silicaandesites and dacites are presented (along with the traceelement abundances) in Table 1. The major element geo-chemistry of the 98N OSC lavas is similar to that of theJuan de Fuca Ridge and Gala¤ pagos Spreading Centerlavas (Fig. 2). Here, only data from the 98N OSC are dis-cussed in detail; however, it is important to note the simi-larity in major and trace element contents and elementaltrends in high-silica lavas from all three ridges. Newanalyses of some of the high-silica samples previouslyanalyzed and discussed by Perfit et al. (1983, 1999) and rep-resentative samples from the Juan de Fuca Ridge are pre-sented as Supplementary Data (http://www.petrology.oxfordjournals.org/). All high-silica samples from the98N OSC appear unweathered with minimal amounts ofFe^Mn oxide coating and are essentially aphyric.The 98N OSC tholeiitic andesites to dacite samples ex-

hibit increasing SiO2 with decreasing FeO, TiO2, andMgO (Figs 2 and 5), with the most differentiated daciteshaving �67wt % SiO2 and51wt % MgO. Al2O3 concen-trations in the dacites (12·9^13·3wt %), however, do notshow a large decrease compared with the OSC basalts(Fig. 5). The dacites have high incompatible major elementconcentrations (K2O40·90wt % and Na2O43·4wt %;Fig. 5) but low P2O5 concentrations (50·26wt %; Fig. 5)compared with basalts. Chlorine concentrations in thedacites range from 0·24 to 0·70wt % compared with50·01 to 0·04wt % in the OSC basalts (Fig. 6).

Trace elementsThe 98N OSC dacites are enriched in incompatible traceelements compared to 98N OSC basalts (Fig. 7); the latter


2385

at University of F

lorida on Decem


Dow

nloaded from

http://www.petrology


Table 1: Major and trace element compositions of dacites from the 98NOSC, East Pacific Rise

Latitude (N): 9·148 9·145 9·145 9·143 9·147 9·146 9·128 9·145 9·137

Longitude (W): 104·198 104·207 104·207 104·207 104·203 104·204 104·211 104·206 104·211

Sample no.: 266-58 265-65 265-64 265-67 266-50 266-53 265-84 265-63 266-47

SiO2 63·0 63·8 64·0 64·1 64·3 64·3 64·4 64·4 64·5

TiO2 1·10 1·26 1·28 1·34 1·07 1·06 1·13 1·29 0·99

Al2O3 13·1 13·2 13·1 13·3 13·2 13·3 13·2 13·3 13·2

FeO 8·43 8·14 8·27 8·49 8·08 8·06 8·18 8·22 7·74

MnO 0·16 0·15 0·16 0·16 0·14 0·14 0·15 0·15 0·14

MgO 1·75 1·34 1·60 1·49 1·27 1·12 1·23 1·29 1·02

CaO 4·34 4·21 4·45 4·41 3·78 3·73 3·92 4·21 3·53

Na2O 3·63 3·84 3·46 3·93 4·23 4·16 3·41 3·71 4·94

K2O 0·96 0·97 0·97 0·95 1·10 1·09 1·19 0·99 1·22

P2O5 0·26 0·22 0·20 0·23 0·24 0·25 0·22 0·21 0·23

Cl 0·24 0·65

Total 96·69 97·17 97·55 98·42 97·35 97·20 96·98 97·76 97·54

Trace elements (ppm)

Li 32 34 32 30 29 30

Sc 17·4 19·9 18·2 14·8 15·2 17·2

V 122 140 121 61 102 121

Cr 12·9 12·3 12·4 3·8 1·7 9·8

Co 14·8 17·3 15·5 12·8 14·0 14·9

Ni 9·0 9·8 9·2 5·8 7·0 8·4

Cu 17·3 19·1 17·6 18·4 20·8 16·7

Zn 110·1 124·1 122·4 109·0 100·4 112·7

Ga 29·7 35·4 28·1 28·8 29·4 30·9

Rb 9·1 10·5 9·6 12·8 12·7 9·5

Sr 76 90 89 86 70 76

Y 132 148 142 157 133 132

Zr 735 842 622 856 968 745

Nb 13·0 14·8 13·0 15·5 14·6 13·2

Cs 0·11 0·12 0·12 0·14 0·13 0·10

Ba 50·5 57·1 52·9 64·6 59·8 49·8

La 23·25 26·31 23·43 28·53 27·27 23·53

Ce 67·48 76·53 67·97 82·17 77·81 68·11

Pr 10·06 11·37 10·05 11·87 11·21 10·16

Nd 46·4 52·8 45·9 53·7 50·9 47·3

Sm 14·09 15·37 14·62 16·25 14·32 13·77

Eu 3·00 3·35 3·12 3·18 2·80 2·99

Gd 16·83 18·13 17·43 19·00 16·58 16·37

Tb 3·19 3·43 3·36 3·65 3·09 3·10

Dy 21·08 22·81 22·85 24·21 20·31 20·53

Ho 4·58 4·88 4·92 5·19 4·33 4·40

Er 13·55 14·65 14·86 15·79 13·16 13·19

Tm 2·19 2·33 2·37 2·50 2·10 2·10

Yb 14·25 15·20 15·66 16·74 13·66 13·54

Lu 2·15 2·31 2·44 2·56 2·10 2·06

Hf 18·69 20·95 17·54 22·28 23·14 18·62

Ta 0·81 1·10 0·92 1·03 1·39 0·99

Pb 3·33 5·05 3·41 3·15 4·89 5·82

Th 1·74 1·82 1·64 2·29 2·28 1·65

U 0·59 0·65 0·64 0·86 0·82 0·59

(continued)


2386

at University of F

lorida on Decem


Dow

nloaded from


Table 1: Continued

Latitude (N): 9·128 9·133 9·116 9·164 9·140 9·155 9·128 9·116 9·154

Longitude (W): 104·210 104·216 104·198 104·205 104·211 104·213 104·208 104·198 104·215

Sample no.: 265-85 266-46 265-94 264-09 265-70 265-42 265-83 265-95 265-40

SiO2 65·0 65·0 65·2 65·8 66·3 66·5 67·5 67·5

TiO2 1·06 0·94 0·97 0·89 0·87 0·94 0·76 0·77

Al2O3 13·1 12·9 13·0 13·2 13·2 13·0 13·3 13·1

FeO 7·99 7·17 7·90 7·03 7·17 7·92 6·68 6·47

MnO 0·16 0·15 0·14 0·13 0·14 0·16 0·13 0·12

MgO 1·18 1·41 1·13 1·06 0·80 0·89 0·67 0·94

CaO 3·78 3·71 3·54 3·48 3·23 3·50 2·98 3·01

Na2O 3·67 4·76 4·29 4·24 4·08 3·99 3·88 4·43

K2O 1·22 1·19 1·14 1·21 1·33 1·20 1·37 1·21

P2O5 0·20 0·17 0·23 0·20 0·19 0·21 0·16 0·15

Cl 0·64 0·58 0·70 0·51 0·67

Total 97·41 97·43 97·61 97·78 97·27 98·91 97·37 97·67


Li 31 31 27 34 32 32 31 31

Sc 14·5 13·1 11·9 12·3 13·8 11·0 10·5 12·5

V 73 63 46 45 58 32 52 51

Cr 4·6 1·5 3·8 3·7 3·4 3·4 3·0 4·2

Co 12·3 10·7 10·2 9·9 10·7 8·3 8·5 9·9

Ni 6·6 4·9 5·7 5·0 5·9 4·7 5·4 5·3

Cu 18·8 16·8 15·9 15·9 15·5 14·0 14·7 16·6

Zn 107·6 104·6 88·8 106·4 119·0 103·1 98·3 103·3

Ga 28·1 30·2 28·3 28·8 29·5 29·2 30·4 30·2

Rb 13·8 12·4 12·9 15·0 13·7 15·5 12·5 12·4

Sr 81 68 78 78 83 76 61 73

Y 154 146 151 160 160 159 145 146

Zr 872 1050 824 934 816 922 985 1013

Nb 15·3 16·1 14·9 15·9 16·4 15·6 16·5 16·6

Cs 0·15 0·13 0·13 0·17 0·16 0·17 0·13 0·13

Ba 68·1 60·0 65·7 72·7 70·0 76·4 62·2 59·7

La 29·07 29·10 28·98 30·89 29·03 30·69 29·16 29·47

Ce 82·46 83·93 83·89 88·15 82·95 87·16 83·65 85·00

Pr 11·78 12·15 11·97 12·49 11·98 12·32 12·02 12·44

Nd 52·1 55·2 52·7 55·0 53·5 54·0 54·6 56·6

Sm 16·01 15·74 15·92 16·67 16·69 16·46 15·64 16·89

Eu 3·01 3·02 2·95 3·09 3·39 3·05 2·89 3·32

Gd 18·51 18·17 18·54 19·47 19·69 18·90 17·66 19·68

Tb 3·55 3·39 3·55 3·69 3·76 3·64 3·35 3·67

Dy 23·80 22·36 23·80 25·06 25·37 24·51 22·13 23·89

Ho 5·12 4·77 5·12 5·38 5·47 5·27 4·74 5·20

Er 15·50 14·37 15·66 16·42 16·48 16·18 14·61 15·35

Tm 2·49 2·29 2·52 2·63 2·64 2·62 2·32 2·49

Yb 16·69 14·81 16·84 17·54 17·56 17·54 15·12 16·19

Lu 2·58 2·30 2·61 2·73 2·75 2·72 2·32 2·42

Hf 22·96 24·87 22·68 24·80 22·24 24·75 24·51 24·97

Ta 1·05 1·18 1·04 1·10 1·12 1·08 1·23 1·02

Pb 3·59 5·75 2·76 3·84 3·47 3·80 4·12 3·64

Th 2·43 2·35 2·59 2·68 2·36 2·80 2·36 2·49

U 0·92 0·84 0·98 1·04 0·91 1·05 0·86 0·84


2387

at University of F

lorida on Decem


Dow

nloaded from


have compositions typical of normal, incompatible traceelement-depleted mid-ocean ridge basalts (N-MORB)from the northern East Pacific Rise. For example, Zr andHf concentrations in the dacites range from 622 to1050 ppm and from 18 to 25 ppm, respectively (Table 1).

The dacites also contain high concentrations of Rb, Ba, Uand Th, but relatively low Sr and Eu contents. Comparedwith East Pacific Rise N-MORB the dacites have relativelyflat REE patterns (Fig. 2). On mantle-normalized traceelement diagrams the dacites have positive Zr, Hf, U, and

Table 2: Radiogenic isotopic compositions of lavas from 98NOSC, East Pacific Rise

208Pb/204Pb 2s 207Pb/204Pb 2s 206Pb/204Pb 2s 87Sr/86Sr 2s 143Nd/144Nd* 2s eNd

East Limb basalts

264-04 37·6992 17 15·4758 8 18·2789 9 0·702496 11 0·513163 11 10·2

265-18 37·6766 19 15·4723 7 18·2749 8 0·702487 18 0·513190 13 10·8

265-35 37·6635 16 15·4704 6 18·2496 7 0·702472 12 0·513164 5 10·3

265-43 37·6608 15 15·4674 5 18·2492 6 0·702502 21 0·513154 9 10·1

265-113 37·6949 17 15·4767 6 18·2750 7 0·702494 16 0·513172 5 10·4

266-01 37·6417 20 15·4693 7 18·2348 9 0·702456 15 0·513158 7 10·1

266-33 37·6831 22 15·4739 9 18·2771 9 0·702444 13 0·513160 5 10·2

265-05 37·6873 20 15·4776 6 18·2936 7 0·702430 12 0·513191 8 10·8

East Limb basaltic andesites

265-24 37·6834 17 15·4731 6 18·2644 6 0·702562 13 0·513187 5 10·7

265-56 37·6813 15 15·4736 6 18·2702 6 0·702528 12 0·513187 4 10·7

265-91 37·6827 14 15·4745 5 18·2659 5 0·702532 12 0·513179 4 10·6

265-103 37·6742 21 15·4721 8 18·2608 9 0·702450 12 0·513162 4 10·2

265-109 37·6729 18 15·4714 7 18·2594 7 0·702493 16 0·513145 7 9·9

265-125 37·6717 19 15·4712 7 18·2648 8 0·702458 45 0·513154 6 10·1

264-20 37·6899 15 15·4758 6 18·2694 6 0·702438 11 0·513179 4 10·6

265-49 37·6890 16 15·4745 6 18·2685 6 0·702438 12 0·513196 7 10·9

East Limb andesites

264-14 37·6725 21 15·4724 10 18·2624 11 0·702428 14 0·513153 5 10·0

265-25 37·6614 17 15·4688 7 18·2511 7 0·702544 18 0·513152 5 10·0

265-90 37·6754 15 15·4716 6 18·2617 7 0·702496 11 0·513159 6 10·2

265-100 37·6804 15 15·4735 6 18·2650 6 0·702489 12 0·513179 4 10·6

266-54 37·6877 17 15·4749 7 18·2717 8 0·702463 12 0·513193 5 10·8

East Limb dacites

264-09 37·6764 18 15·4721 7 18·2679 8 0·702536 19 0·513171 8 10·4

265-40 37·6737 21 15·4713 8 18·2655 9 0·702466 11 0·513149 8 10·0

265-42 37·6782 17 15·4719 7 18·2703 7 0·702458 15 0·513140 6 9·8

265-64 37·6822 25 15·4756 10 18·2670 11 0·702576 15 0·513185 7 10·7

265-70 37·6760 20 15·4707 8 18·2663 8 0·702476 24 0·513141 5 9·8

265-83 37·6814 15 15·4728 6 18·2712 7 0·702534 15 0·513147 7 9·9

265-84 37·6887 17 15·4761 7 18·2739 7 0·702507 11 0·513179 7 10·6

265-85 37·6767 21 15·4713 8 18·2674 8 0·702481 19 0·513148 5 9·9

265-95 37·6795 17 15·4746 7 18·2633 8 0·702465 19 0·513165 6 10·3

266-53 37·6780 13 15·4718 5 18·2676 5 0·702471 15 0·513175 10 10·5

The 2s error reflects in-run machine error (precision at the last significant figure). Long-term reproducibility estimatesare: 87Sr/86Sr¼�0·00003, 143Nd/144Nd¼�0·000018, 206Pb/204Pb¼�0·0034 (205 ppm), 207Pb/204Pb¼ 0·0028 (184 ppm),208Pb/204Pb¼�0·0086 (234 ppm).*Unknowns were normalized to an 143Nd/144Nd value of 0·511215� 0·000007 for JNdi-1, which was reported by Tanakaet al. (2000) relative to a La Jolla 143Nd/144Nd value of 0·511858 (Lugmair & Carlson, 1978).Major and trace element data for basalts, FeTi basalts, basaltic andesites and andesites have been presented by Wanless(2010).


2388

at University of F

lorida on Decem


Dow

nloaded from


Th anomalies, and negative Nb and Ta anomalies (Fig. 2).Consequently, the dacites also have slightly lower Nb/Laand higher Zr/Dy ratios compared with OSC basalts(Fig. 8), and Ce/Yb, Th/Nb and Nd/Y ratios increase frombasalt to dacites (Fig. 8). Compatible trace elements arelow in the dacites, with Ni concentrations ranging from9·8 to 4·9. ppm (Fig. 7) and Cr concentrations from 13 to1ppm. Most incompatible trace elements exhibit negativecorrelations with MgO; however, compatible elements

(i.e. Ni and Cr) and Nb/La are positively correlated.U/Nb, Nd/Y Th/Nb and Ce/Yb ratios are negatively corre-lated with MgO in the dacites.

Isotopic dataThe 98N OSC dacites have very limited ranges of Pb, Srand Nd isotopic compositions, which lie within the generalfield of East Pacific Rise MORB (Table 2; Fig. 9). 87Sr/86Srratios range from 0·70246 to 0·70258, with an average of

Fig. 5. Major element variations vs MgO (wt %) in dacites from the 98NOSC on the East Pacific Rise (filled squares). Gray field represents alllavas collected in 2007 from the 98N OSC. Dacites are compared with three low-pressure fractional crystallization trends (calculated usingMELTS; Ghiorso & Sack,1995) using parental compositions (FC-1, FC-2, FC-3) from OSC basalts (see text for modeling parameters and detailsand Table 3 for compositions). Not all dacite major element variations can be explained by fractional crystallization alone (e.g. Al2O3, K2O,and P2O5). Experimental compositions from partial melting of altered basalt (Beard & Lofgren,1991) are shown for comparison (B&L PMelts).


2389

at University of F

lorida on Decem


Dow

nloaded from


0·70250. These values are well within the range ofN-MORB East Pacific Rise lavas from 9 to 108N (Simset al., 2002, 2003; Goss et al., 2010) and similar toN-MORB lavas from the 98N OSC. 143Nd/144Nd ratios arealso similar to East Pacific Rise N-MORB and rangefrom 0·513140 (eNd¼ 9·8) to 0·513196 (eNd¼10·9). Pbisotopes ratios for the 98N OSC dacites are indistinguish-able from those for 98N OSC basalts and other EastPacific Rise lavas, with average 206Pb/204Pb¼18·268,207Pb/204Pb¼15·473, and 208Pb/204Pb¼ 37·679.The Pb iso-tope compositions of the 98N dacites form a tight clusterin the center of the field defined by other 98N lavas (Fig. 9).

PETROGENET IC MODELS FORH IGH- S I L ICA LAVASWe now examine the results of various models of fractionalcrystallization, partial melting and assimilation andcompare the results with the geochemical data describedabove to evaluate their relevance to the formation ofMOR dacites. Specifically, we focus on physically reason-able models that are consistent with the highly

differentiated major element concentrations, high concen-trations of incompatible elements, distinct trace elementpatterns, and N-MORB-like isotopic signatures. Themodels must also be able to explain relatively high U, Th,Zr and Hf, and low Nb andTa as well as the flat REE pat-terns. Additionally, markedly high Cl, K (and high Cl/K),Al2O3, and low P2O5 must be accounted for in successfulpetrogenetic schemes.

Crystal fractionation

Several petrological models, including MELTS thermody-namic modeling (Ghiorso & Sack, 1995), Rayleigh crystalfractionation, crystal^melt segregation, and in situ crystal-lization are investigated here to determine if various pro-cesses of crystal fractionation can account for the majorand trace element compositions of MOR dacites.

Rayleigh fractional crystallization

The program MELTS (Ghiorso & Sack, 1995) providesa useful framework to evaluate if the major element

Fig. 6. Variation diagram showing Cl (wt %) vs MgO (wt %) for OSC lavas. Superimposed are three model liquid lines of descent (calculatedusing MELTS; Ghiorso & Sack, 1995), showing that the maximum amount of Cl enrichment owing to extensive fractional crystallizationcannot produce the high Cl concentrations in the OSC dacites. The dashed rectangle represents the range of compositions that can be producedthrough 1^15% partial melting of altered basalt with 350 ppm Cl (star); 350 ppm Cl is the median value of Cl analyzed in sheeted dikes fromODP Hole 504B, with minimum and maximum values (49 and 650 ppm) shown with an error bar. Partition coefficients for Cl are from Gilliset al. (2003). The range of MgO values for partial melts was taken from experimental partial melts of less than 15% (Beard & Lofgren, 1991).


2390

at University of F

lorida on Decem


Dow

nloaded from


compositions of MOR dacites can be produced by crystalfractionation. Petrological modeling was also carried outusing the program PETROLOG (Danyushevsky, 2001)with similar results for high-silica lavas, although the pro-grams generate somewhat different results for intermediatecompositions. Additionally, the MELTS calculations areconsistent with results from Perfit et al. (1983), which werecalculated using least-squares best-fit equations. Several98N N-MORBs were used as starting parental melt com-positions (Table 3) to determine if a moderately evolved

magma could partially crystallize to produce a dacite.These included a slightly evolved MORB (265-113), a ferro-basalt (265-43), and a FeTi basalt (264-08). Pressures foreach MELTS run were set at 1 kbar to simulate an approx-imate minimum depth of crystallization in the shallowoceanic crust, the oxygen fugacity was set at the quartz^fayalite^magnetite (QFM) buffer, and the H2O concentra-tions varied from 0·2 to 0·35wt % depending on theparent melt composition. Both Perfit & Fornari (1983) andJuster et al. (1989) determined that extensive crystallization

Fig. 7. Trace element variations vs Zr (ppm) in 98NOSC lavas. Superimposed on the diagrams are calculated trends for fractional crystalliza-tion (model 1 and 2 using the Rayleigh fractionation equation), 1^15% partial melting (assuming batch melting), and AFC simulations(EC-AFC; Bohrson & Spera, 2001). (See text for model parameters.) Inflection points in model trend lines represent changes in crystallizingphases.


2391

at University of F

lorida on Decem


Dow

nloaded from


occurred at low pressures and oxygen fugacities slightlyabove QFM. Liquid lines of descent were also calculatedfor higher pressures (up to 5 kbar) to simulate depths ofcrystallization within the nascent layer 3 and the shallowmantle. However, the liquid lines of descent converge on

similar end-member compositions at low MgO and highSiO2 as fractional crystallization proceeds.Liquids with MgO and SiO2 contents similar though

not identical to 98N OSC dacites can be produced by�75^85% crystal fractionation of a ferrobasaltic parent.

Fig. 8. Normalized trace element ratio diagrams showing the range of dacite compositions at the 98N OSC. Fractional crystallization, partialmelting and AFC trends are shown as in Fig. 4. Tick marks on trend lines for partial melts are in 1% increments, whereas fractional crystalliza-tion tick marks represent 10% intervals. Concentrations are mantle-normalized (Sun & McDonough, 1989).


2392

at University of F

lorida on Decem


Dow

nloaded from


Results predict a crystallization sequence of Ol, followedby OlþPlag, OlþPlagþCpx, PlagþCpxþ Sp (titano-magnetite), and in some models, late-stage crystallizationof apatite. No orthopyroxene crystallization is predicted,which is similar to some experimental results (Juster et al.,1989); however, pigeonite is predicted by these experimentsand is observed in some MOR andesites and dacites.The calculations suggest that temperatures of59808C arereached when residual liquids attain compositions similarto dacites. The models are in agreement with anhydrousexperimental results that indicate that residual daciticliquids form after �87% crystallization at temperaturesof �10408C (Juster et al., 1989).For several major elements (TiO2, FeO, and SiO2), com-

positions similar those of to 98NOSC dacites are obtainedthrough crystal fractionation of a MORB magma; the

total amount of crystallization varies slightly dependingon whether the starting composition was a moderatelyevolved basalt, ferrobasalt, or FeTi basalt (Fig. 5; maxi-mum of �85% crystallization). In contrast, calculatedabundances of K2O, P2O5, Al2O3, and Cl do not matchthe dacitic end-member compositions using any of the par-ents; modeled residual liquids have higher P2O5 by factorsof 5^10, lower K2O by factors of 1·5^2·5, lower Al2O3 byfactors of 1·4^1·5 and lower Cl by factors of 10^12 (Figs 5and 6). Although MELTS does not predict apatite satura-tion in andesitic liquids the decreasing P2O5 contents insome andesites and very low values in dacites strongly sug-gest apatite crystallization. Juster et al. (1989) calculatedthat apatite saturation would occur at �0·7wt % P2O5 inGala¤ pagos Spreading Center andesites.

Fig. 9. Radiogenic isotope compositions of 98N OSC lavas. (a) Pb-isotope diagram showing that the 98N OSC dacites have 208Pb/204Pb and206Pb/204Pb ratios similar to OSC N-MORB basalts and the northern East Pacific Rise N-MORB (gray field; data from Sims et al., 2002,2003; Goss et al., 2010). (b) Nd and Sr isotope data show that 98N OSC lavas are similar to East Pacific Rise and OSC N-MORB. Black barrepresents calculated Sr enrichment (0·0001) during assimilation of altered crust (see text for details). (SeeTable 2 for data.)


2393

at University of F

lorida on Decem


Dow

nloaded from


Additional tests of Rayleigh fractionation include model-ing the behavior of trace elements using as input para-meters the degree of crystallization and mineral modalproportions determined from the MELTS modeling.Basaltic partition coefficients are used in the trace elementmodeling up to �57% SiO2, and andesitic partition coeffi-cients are used for 457% SiO2 (Table 4). The Rayleighfractionation equation [Cl/Co¼F(D ^ 1)] is used to simulatea continuously evolving magma chamber in whichphenocrysts are immediately separated from the liquid(Cl¼ concentration of element in the liquid;Co¼ concentration of the element in the parent; F¼%melt; and D¼ bulk partition coefficient). The startingcomposition for these calculations was a ferrobasalt fromthe 98NOSC (265-43).As shown in Fig. 7, the trace element concentrations

observed in the dacites cannot be reproduced usingRayleigh fractionation (Fig. 7) with the constraintsimposed by major element variations. This model repro-duces some dacite incompatible element compositions butdoes not reproduce the observed enrichments in most ofthe incompatible elements. Although the most incompati-ble elements (Rb, Ba, Th, U) show the greatest differencebetween the observed and calculated compositions, evenless incompatible elements (Nb, Zr, Y, Hf) require490%crystal fractionation. For instance, maximum calculatedZr and Nb concentrations are 705 ppm and 13 ppm,respectively, compared with an average of 870 and 15 ppmin the dacites. In addition, the UN/NbN is predicted bymodeling to be 51, whereas the measured dacite valuesare 41 and the modeling does not reproduce the highZrN/DyN and CeN/YbN and low NbN/LaN ratios in the98N OSC lavas (Fig. 8). The middle to heavy REE(MREE to HREE) concentrations (i.e. Nd, Sm, Eu, Dy,Yb, and Lu) can be generated only by490% crystal frac-tionation. Regardless, such extreme degrees of crystalliza-tion are inconsistent with major element modelcalculations.In summary, the calculated liquid lines of descent do not

provide a good fit to the observed major and minor com-positions of the high-silica lavas, and trace elementmodels parameterized from the MELTS calculations donot reproduce measured trace element abundances ortrace element ratios. Thus, we conclude that extensivelow-pressure crystal fractionation is unlikely to be the solemechanism to explain the genesis of the high-silica lavasat the 98NOSC.

Crystal^melt segregation model

Bachmann & Bergantz (2004) suggested that intermediateliquids (andesites and dacites) will separate from crystals(via filter pressing) when a magma has undergone440^50 vol. % crystallization. The segregated melt, which is

Table 3: Starting compositions for geochemical modeling

Sample: 264-08 (FC-3) 265-43 (FC-1) 265-113 (FC-2)

SiO2 50·1 50·5 51·9

TiO2 2·68 1·92 2·17

Al2O3 12·7 13·9 13·4

FeO 14·1 11·6 12·8

MnO 0·26 0·21 0·23

MgO 5·69 6·98 5·93

CaO 9·58 11·14 9·48

Na2O 3·27 2·86 3·28

K2O 0·21 0·13 0·22

P2O5 0·28 0·19 0·28

Cl 0·07 0·01 0·02

Total 99·24 99·67 99·72


Li 10 8 11

Sc 42·1 42·3 38·0

V 450 347 323

Cr 18·0 108·6 32·1

Co 43·5 41·4 39·0

Ni 33·8 54·3 33·7

Cu 60·4 58·8 50·8

Zn 116·1 93·7 103·2

Ga 21·3 18·3 21·0

Rb 2·1 1·1 2·2

Sr 126 120 111

Y 58·9 43·8 61·0

Zr 180 126 229

Nb 5·5 3·1 5·7

Cs 0·03 0·01 0·03

Ba 17·1 8·4 15·6

La 6·81 4·54 7·63

Ce 20·51 14·16 23·69

Pr 3·32 2·34 3·92

Nd 17·31 12·56 20·09

Sm 5·99 4·42 6·68

Eu 1·94 1·48 1·99

Gd 7·74 5·78 8·52

Tb 1·45 1·08 1·58

Dy 9·69 7·22 10·13

Ho 2·05 1·52 2·20

Er 5·94 4·39 6·26

Tm 0·91 0·67 0·96

Yb 6·01 4·43 6·17

Lu 0·93 0·68 0·95

Hf 4·83 3·44 5·72

Ta 0·37 0·21 0·36

Tm 0·61 0·39 0·95

Gd 0·34 0·18 0·40

Pb 0·13 0·08 0·15


2394

at University of F

lorida on Decem


Dow

nloaded from


more evolved than the original melt, will once again crys-tallize until it reaches 40^50% phenocrysts, when againthe new evolved melt will separate from the pheno-crysts. Although this segregation model applies strictlyto systems of intermediate composition (Bachmann &Bergantz, 2004), here we evaluate whether a basalticmagma can evolve geochemically, through a series of seg-regation events, to form compositions similar to the MORdacites.To simulate these conditions a 98N OSC ferrobasalt

was allowed to undergo equilibrium crystallization to anandesitic composition, using MELTS thermodynamic cal-culations (Ghiorso & Sack, 1995) and starting conditionsdescribed for Rayleigh fractionation models. At this ande-sitic composition, the liquid separates from the phenocrysts(Bachmann & Bergantz, 2004), creating a new parentmelt. This parent composition becomes the new startingconcentration (an andesite) for the next run, which subse-quently crystallizes 50% by volume. This process wasrepeated until MgO and SiO2 concentrations similar tothose of the 98N OSC dacites were obtained (Fig. 10). This

occurred after three segregation events or 87·5wt % crys-tallization; however, as was noted during the MELTS cal-culations, several dacitic major and minor elementconcentrations (i.e. Al2O3, K2O and P2O5) could not bereproduced. In addition, the calculated incompatibletrace element abundances were also lower than thoseobserved in the MOR dacites (Fig. 10).

In situ crystallization calculations

A different approach to magma crystallization is in situ

crystallization, where phenocrysts do not separate fromthe interstitial melt until a small remaining volume ofliquid is pressed from the crystallizing mush and mixedwith the main body of melt (e.g. Langmuir, 1989;Reynolds & Langmuir, 1997; Pollock et al., 2005). This pro-cess assumes that crystallization occurs along a tempera-ture gradient within a solidification front (or boundarylayer) and interstitial melt evolves independently from themain melt body. The mixing of interstitial melt back intothe main magma body gradually causes bulk increases

Table 4: Partition coefficients used in Rayleigh fractional crystallization, partial melting and AFC calculations

Element Andesite partition coefficients Basalt partition coefficients

Olivine Cpx Plag Apatite Ilmenite Amphibole Olivine Cpx Plag Ilmenite

Rb 0·01 0·02 0·025 0·001 0·034 0·04 0·0003 0·0004 0·056 0·034

Ba 0·01 0·02 0·155 0·12 0·00034 0·1 0·00001 0·0003 1·45 0·00034

Th 0·01 0·01 0·19 1·28 0·00055 0·15 7·00E-06 0·0021 0·13 0·00055

U 0·005* 0·0075* 0·34 1·4 0·0082 0·008 9·00E-06 0·001 0·051 0·0082

Nb 0·00017 0·005 0·033 0·0011 2 0·28 0·00005 0·0089 0·045 2

Ta 0·00002 0·014 0·11 0·003 1·7 0·27 0·00012* 0·013 0·066 1·7

La 0·00006 0·062 0·082 11·4 0·000029 0·027 0·0002 0·054 0·13 0·000029

Ce 0·00006 0·116 0·072 12·9 0·000054 0·0293 7·00E-05 0·086 0·11 0·000054

Sr 0·00217 0·08 2·7 4·3 0·00027* 0·28 0·00004 0·091 1·4 0·00027*

Nd 0·00015 0·33 0·045 32·8 0·00048 0·0325 0·0003 0·19 0·066 0·00048

Zr 0·00450 0·14 0·0009 0·042 0·29 0·26 0·001 0·26 0·048 0·29

Hf 0·00370 0·21 0·017* 0·014 0·38 0·43 0·0029 0·33 0·051 0·38

Sm 0·00044 0·41 0·033 16·1 0·00059 0·024 0·0009 0·27 0·054 0·00059

Eu 0·00056 0·57 0·55 25·5 0·009 0·0498 0·0005 0·43 0·65 0·009

Dy 0·00250 0·94 0·034 34·8 0·01 0·0136 0·0027 0·44 0·024 0·01

Y 0·00380 0·9 0·01 7·1 0·0045 0·0196 0·0082 0·47 0·013 0·0045

Yb 0·05600 0·63 0·014 15·4 0·17 0·102 0·024 0·43 0·0079 0·17

Lu 0·03* 0·605 0·039 3·92 0·084 0·6 0·016 0·56 0·06 0·084

Primary Zanetti Klein et al., Dunn & Prowatke & Zack & Bottazzi Halliday Halliday Dunn & Zack &

reference et al., 2004 2000 Sen, 1994 Klemme, 2006 Brumm, 1998 et al., 1999 et al., 1995 et al., 1995 Sen, 1994 Brumm, 1998

Secondary Gill, Gill, Rollinson, Fujimaki, Rollinson, Haase Rollinson, Rollinson, Rollinson, Rollinson,

reference 1979 1979 1993 1986 1993 et al., 2005 1993 1993 1993 1993

*Values interpolated using nearest neighbors. Values for Lu are averages of Y and Yb. Italicized values indicate secondaryreference.


2395

at University of F

lorida on Decem


Dow

nloaded from


in incompatible element abundances and changes in traceelement ratios (Langmuir, 1989). This process will causean increase in highly incompatible elements comparedwith Rayleigh crystal fractionation, because these elementsare continually returned to the residual magma body(Langmuir, 1989).In situ crystallization was evaluated following Reynolds

& Langmuir (1997), with a starting composition of a 98NOSC ferrobasalt (265-43) using the same partition coeffi-cients as for Rayleigh fractional crystallization. Crystalliz-ing phases include Ol, Plag, Cpx, and eventually,Fe-oxides. In the modeled system, the boundary layer isalways 5% of the liquid magma chamber volume and theboundary layer crystallizes until 35% interstitial liquid

remains (Reynolds & Langmuir, 1997). All of the residualliquid mixes back into the magma chamber during eachiteration of boundary layer crystallization. As the magmachamber continues to crystallize by this mechanism, theboundary layer moves inward leaving restite crystalsbehind and the volume of the magma body decreases.Theoretically, this process will continue to modify themagma chamber liquid composition until an infinitesi-mally small amount of melt is left.After 85% in situ crystallization (Fig. 10) calculated

major and incompatible trace element concentrationsdo not match those measured in the MOR dacites. Thisprocess can only account for Zr concentrations in the meltthat are less than three times the original concentration,

Fig. 10. Elemental variation diagrams showing the 98NOSC dacites vs calculated liquid lines of descent produced during two alternative typesof crystallization. Crosses show the evolution of a melt during in situ crystallization (Langmuir, 1989). Circles show the liquid line of descent ofa magma that undergoes 50% fractional crystallization and is then separated (via filter pressing) from the phenocrysts (melt-segregationmodel). The resulting magma undergoes two additional steps of 50% crystallization and melt segregation, following the model of Bachman &Bergantz (2004). Inflection points in models represent changes in crystallizing phases, particularly Fe-oxides.


2396

at University of F

lorida on Decem


Dow

nloaded from


reaching values of 320 ppm. Even495% in situ crystalliza-tion cannot reproduce the enriched trace element signa-tures of the MOR dacites. Although in situ crystallizationdoes enrich the residual melt in incompatible elementscompared with Rayleigh crystal fractionation, the failureto reproduce the observed enrichments in MOR dacitessuggests that the latter cannot result from this processeither.

Partial melting (anatexis)Experimental studies (Beard & Lofgren, 1991; Koepkeet al., 2004) suggest that510% partial melting of alteredoceanic crust can produce melt with major elementconcentrations consistent with MOR dacite compositions.Of particular note are the high SiO2, Al2O3 and K2Oand low FeO,TiO2, and P2O5 concentrations produced bypartial melting of amphibolite-facies or greenschist-faciesminerals because these are also the chemical characteris-tics of MOR dacites. Basaltic rocks are known to undergopartial or complete alteration and recrystallization at theridge axis as a result of pervasive high-temperature hydro-thermal alteration (Alt et al., 1986; Gillis & Roberts, 1999).Such altered rocks are an attractive starting compositionfor anatexis because their solidus temperatures are muchlower than those of fresh MORB.Altered oceanic crust can have a wide range of trace ele-

ment concentrations, depending on the degree of alteration

(Alt et al., 1986). To better evaluate the composition of thewall-rock involved in the formation of dacites on the 98NOSC, we use the batch melting equation, Cl¼Co/[Dbulk

(1 ^F)þF] and andesitic partition coefficients (Table 4)to solve for a range of possible wall-rock compositions(Co) and then compare these results with compositions offresh and altered MORB. Trace element patterns gener-ated from the calculations suggest that altered basaltprovides a better fit than fresh MORB as a source (wall-rock) composition for the 98N OSC dacites (Fig. 11).Consequently, we model partial melting of an alteredMORB (Nakamura et al., 2007) to determine if this processcould produce the geochemical characteristics observed inthe 98NOSC lavas (Fig. 11).Altered basaltic wall-rock is melted using different

modal mineralogies, one with amphibole (Haase et al.,2005) and one without (Koepke et al., 2004). Calculatedtrace element concentrations resulting from 1^15% partialmelting of oceanic crust are shown in Figs 7, 8, and 12.Partial melting in the absence of amphibole (19% Ol,30% Cpx, 50% Plag, 1% Ilm) can reproduce some, butnot all, of the trace element enrichments observed in the98N OSC dacites (Fig. 12a). In particular, the HREE inthe 98N OSC dacites are higher than the calculated abun-dances. The concentrations derived from partial meltingof altered crust with amphibole [20% Cpx, 25% Opx,49% Plag, 5% Amph, 1% Fe-oxide; based on modal

Fig. 11. Partial melting model showing a range of possible sources (gray lines) that could produce the 98NOSC dacites (bold line) from1^15%partial melting. Sheeted dikes and the upper parts the gabbroic layer may be composed of a range of compositions depending on the composi-tion of the starting material and degree of alteration. Possible source compositions were calculated using the batch melting equation and solvingfor the initial composition (Co). Three possible wall-rock compositions are superimposed on the calculated parental range. Altered basalt 1(squares; Nakamura et al., 2007) provides the best match to the calculated source composition and is used as the source rock in subsequent par-tial melting and AFC models.


2397

at University of F

lorida on Decem


Dow

nloaded from


proportions from Haase et al. (2005)] are much closer tothe concentrations of incompatible elements in the 98NOSC dacites, suggesting that amphibole is an importantcomponent in the source rock (Fig. 12b).The best-fit model relies on 510% partial melting

of amphibole-bearing altered oceanic crust to produceincompatible trace element compositions comparable withthose in the 98NOSC dacites. In particular, the importantcharacteristics of this model are melts with positive Zrand Hf anomalies, negative Nb and Ta anomalies onmantle-normalized diagrams (Fig. 12), relatively high Uand Th concentrations, and high UN/NbN and CeN/YbNratios (Fig. 8).Melting of altered basalt can also produce elevated Cl

concentrations similar to those observed in the MOR

dacites (Fig. 6). Although Cl partition coefficients arepoorly constrained, we use estimates to model Cl partition-ing during melting (Gillis et al., 2003). Using the medianCl concentration in altered basalts in Ocean DrillingProgram (ODP) Hole 504B (350 ppm) as a starting com-position and modal proportions described above, we calcu-late a range of possible Cl enrichments for 1^15% meltingto be from 0·2wt % to41·0wt %. This spans the rangeof Cl concentrations observed in MOR dacites (Fig. 6).

Assimilation^fractional crystallizationThe energy constrained assimilation^fractional crystalli-zation (EC-AFC) formulation of Bohrson & Spera (2001)is used to assess the role that these associated processesplay in the formation of MOR dacites. The amount of

Fig. 12. Mantle-normalized trace element diagrams showing the results of 1^15% partial melting (PM) of an altered basalt (see Fig. 11) usingtwo modal mineralogies. Partial melts were calculated using the batch melting equation. (See text for further details.) (a) Partial melting resultsusing a non-amphibole-bearing gabbro assemblage (19% Ol, 30% Cpx, 50% Plag,1% Ilm). (b) Partial melting results using an amphibole-bear-ing gabbro (20% Cpx, 25% Opx, 49% Plag, 5% Amph, 1% Fe-oxide; Haase et al., 2005). The amphibole-bearing gabbro provides the best fitto the 98N OSC dacites.


2398

at University of F

lorida on Decem


Dow

nloaded from


crystallization required to produce enough heat to melt thesurrounding crust is calculated and, in turn, this producesa specific mass of melt of a specific composition. Severalphysical parameters are required as inputs to EC-AFC cal-culations (Table 5). These include the liquidus temperatureof the parental magma (�12008C based on results ofMELTSmodeling of OSC lavas), the temperature and soli-dus of the wall-rock, and the temperature of equilibriumbetween the wall-rock and the magma. The initialmagma composition is assumed to be N-MORB and theassimilant is amphibole-bearing altered oceanic crustwith the modal composition described above.The wall-rock may span a range of temperatures (from

8008C to 408C) depending on the age of the ocean crust(Maclennan, 2008). Higher initial wall-rock temperaturesallow crustal melting to begin earlier in the evolution ofthe magma reservoir because less additional heat isrequired to raise the wall-rock above its solidus tempera-ture; that is, at the onset of assimilation, the amount offractional crystallization the magma has undergone islower. This lowers the overall amount of incompatibletrace element enrichment in the resulting magma becausethe fractionating magma is less chemically evolved at theonset of assimilation and, despite the high trace elementconcentrations in the anatectic melt, the overall enrich-ment is less than that of a highly fractionated magma(Fig. 13). For example, crust with an initial temperatureof 8008C will begin melting after 50^55% magma crystal-lization and the resultant magma has a maximum of�25 ppm La, which is too low to produce the dacitic com-positions (Fig. 13). Antithetically, lowering the wall-rocktemperature decreases the total mass of wall-rock assimi-lated, while increasing the amount of crystallizationneeded to initiate melting. Consequently, this causes anincrease in the overall incompatible element concentration

possible in the melt.Thus, crust with an initial temperatureof 508C requires 485% magma crystallization to beginmelting, but results in concentrations of �35 ppm La inthe magma (Fig. 13), in agreement with the abundancesobserved in the dacites. Based on these competing pro-cesses, the best-fit wall-rock temperature to generate the98NOSC dacites is between 650 and 7208C.The local solidus, as described by Bohrson & Spera

(2001), is the solidus of the assimilant, in this case amphibo-litized basalt. Several experimental studies have examineddehydration melting of altered oceanic crust and amphibo-lites (Hacker, 1990; Rapp et al., 1991; Wolf & Wyllie, 1994;Johannes & Koepke, 2001) but few studies were performedunder conditions comparable with those expected atMORs (Beard & Lofgren, 1991). These experiments deter-mined that the solidus temperatures of altered basalts arebetween 8508C and 9008C (Beard & Lofgren, 1991). Gillis& Coogan (2002) discussed the effects of melting alteredcrust at the roof of an axial magma chamber and suggesteda solidus temperature of 8758C. Ti-in-zircon thermometry(905�348C at a TiO2 activity of 0·32�0·02 estimatedfrom coexisting Fe^Ti oxides) on phenocrysts in the Juande Fuca Ridge dacites is broadly consistent with zirconsaturation thermometry (average of 824�158C) andFe^Ti oxide temperatures (�8308C; A. Schmitt, personalcommunication, 2009). Based on these combined results,8758C was used as the local solidus temperature for AFCcalculations.The final thermal input parameter is the temperature of

equilibrium, which is defined as the final equilibrium tem-perature of the magma and wall-rock. Generally, this tem-perature should correspond to the temperature of theerupted lava. Although the temperature of the erupted98N OSC dacites is uncertain, the temperatures of otherdacitic magmas have been estimated. The temperature ofcrystallization of a Gala¤ pagos Spreading Center andesiticmagma was calculated to be as low as �910^9408C basedon coexisting titanomagnetite and ilmenite grains (Perfitet al., 1983), and experimental partial melts resulting frommelting of oceanic gabbros showed that dacitic meltsformed at temperatures of �9008C (Koepke et al., 2004).Consequently, we use 9008C as the input equilibrium tem-perature. The partition coefficients are the same as thoseused for partial melting and fractional crystallizationmodels. Basaltic partition coefficients are used for the frac-tionating magma, whereas andesitic bulk Kd values wereused for the assimilant in the absence of a comprehensivedataset of dacite Kd values.Results of EC-AFC calculations suggest that combina-

tions of 73^85% crystal fractionation of a parental basalticmagma and assimilation of 5^20% by mass of partiallymelted wall-rock produces melts that have trace elementcompositions consistent with those of the 98NOSC dacites.In the best-fit model, melting and assimilation begins

Table 5: AFC modeling parameters

Parameter Abbreviation Value Units

Magma liquidus temperature tlm 1200 8C

Magma temperature tmo 1200 8C

Assimilant liquidus temperature tla 1100 8C

Country rock temperature tao 720 8C

Solidus temperature ts 875 8C

Magma specific heat cpm 1484 J kg–1 K–1

Assimilant specific heat cpa 1388 J kg–1 K–1

Crystal enthalpy Hcry 396000 J kg–1

Fusion enthalpy Hfus 354000 J kg–1

Equilibration temperature Teq 900 8C


2399

at University of F

lorida on Decem


Dow

nloaded from


after 68% crystallization and a further 5^17% crystalliza-tion occurs as the wall-rock melt is assimilated.EC-AFC trace element calculations suggest that many of

the incompatible trace element concentrations and ratiosobserved in the 98NOSC dacites can be explained throughthis combination of processes (Figs. 7, 8, 14). Of particularimportance are negative Nb and Ta anomalies (relative to

La), an increase in Zr and Hf concentrations (relative toHREE), relatively flat mantle-normalized HREE patterns,and ratios of light REE (LREE) to HREE and MREE toHREE that are similar to those observed in MOR dacites(Fig. 14). For example, Zr concentrations in the AFCmodels are 852 ppm and Nb concentrations are 16 ppmcompared with average 98N OSC dacite concentrations of

Fig. 13. Diagram showing the calculated effects of varying wall-rock temperature on incompatible trace element compositions (La) duringAFC. (a) Higher wall-rock temperatures cause earlier onset of assimilation compared with lower initial temperatures. Higher initial tempera-tures produce lower overall incompatible element abundances compared with lower wall-rock temperatures because the amount of fractionalcrystallization the magma has undergone is lower at the onset of assimilation. The average La concentration of the 98N OSC dacites is�28 ppm, which can be produced by assimilating partial melts of a wall-rock at starting temperatures of 650^7208C. (b) Plot showing theratio of assimilation to crystallization (Ma*/Mc) for various temperatures of wall-rock. Lower ratios produce higher incompatible trace elementconcentrations in the melt.


2400

at University of F

lorida on Decem


Dow

nloaded from


870 and 15, respectively. Although the overall fit of themodel data to the observed data is encouraging, the calcu-lated model values for Ba, Th, U and Hf are slightlyunder-enriched (Fig. 14). However, it should be pointedout that some of the input parameters to these calculations,such as the actual degree of alteration and hence composi-tion of the crustal assimilant and the temperature of thewall-rocks, are not well constrained and are very probablynot constant.

DISCUSSIONPetrogenesis of high-silica lavasExtreme crystal fractionation, partial melting of crustalmaterial, and/or AFC processes have been proposed asexplanations for the formation of highly silicic composi-tions in continental interiors, arc and ocean island settings,but only a few studies have focused on the formation ofhigh-silica lavas at MORs (Byerly & Melson, 1976; Perfitet al., 1983; Juster et al., 1989; Haase et al., 2005). The petro-genetic calculations presented above demonstrate thatcrystal fractionation alone is not a viable mechanism forthe formation of high-silica MOR lavas, despite using arange of starting compositions and several end-membermodels (Figs 5, 7 and 8). Instead, the results emphasize

the importance of partial melting and assimilation ofaltered material in the formation of dacites on MORs.

Geochemical evidence of partial melting

Will crustal anatexis create geochemical signatures similarto those observed in MOR dacites? Based on elemental sys-tematics (e.g. Cl, U/Nb; Figs 6 and 7) and the partial melt-ing calculations presented, it appears that partial melts ofaltered oceanic crust may be involved in the generation ofthe MOR dacites. Partial melts of hydrothermally alteredcrust produce distinct signatures compared with those ofunaltered oceanic crust, as a result of changes in mineral-ogy and bulk composition during hydrothermal circula-tion. Hydrothermal circulation in layer 2B or the top oflayer 3 may cause alteration to greenschist or amphiboliteassemblages, where Ca-rich plagioclase is replaced bysodic plagioclase and pyroxenes develop rims or over-growths of amphibole (Alt et al., 1986; Coogan et al.,2003b). The degree to which this occurs depends on tem-perature, water/rock ratios and fluid chemistry. To explainthe geochemical signatures in the MOR dacites, our melt-ing assemblage must include amphibole.Melting of amphibole-bearing assemblages, a common

component in altered layer 2B (Alt et al., 1986; Coogan,2003; Coogan et al., 2003b), can explain many of the

Fig. 14. Mantle-normalized trace element diagram showing results of the best-fit AFC model (fine black lines). This model requires a total of73^85% fractional crystallization in combination with 5^20% assimilation of partially melted wall-rock to produce trace element compositionssimilar to the 98N OSC dacites (bold line). Fractional crystallization is the dominant process until 68% of the magma has crystallized. This isfollowed by 5^20% assimilation of partial melts in conjunction with an additional 5^17% crystallization.


2401

at University of F

lorida on Decem


Dow

nloaded from


major element concentrations of the MOR dacites, includ-ing the anomalously high Al2O3 concentrations observedin the oceanic dacites (Fig. 5). Dehydration partial meltingexperiments, where water exists only as hydrous phaseswithin the rock, provide a better fit to MOR dacite com-positions than hydrous partial melting results (Koepkeet al., 2004). Dehydration melting experiments produce arange of Al2O3 concentrations (Beard & Lofgren, 1991)that are similar to or higher than in MOR dacites (Fig. 5).Comparatively high Na2O concentrations in the dacitesmay result from the melting of albitic plagioclase.Elevated Na2O concentrations are not observed in allthe experimental results of Beard & Lofgren (1991) butappear to be a function of the degree of albitization of thestarting material. Variable P2O5 concentrations are alsoobserved in the experimental melts, suggesting that P2O5

contents are very low in the source rock or that apatite isa residual phase in the melting residue. Similar conclusionscan be applied to the MOR dacites, which have low phos-phorus contents (Fig. 5); however, P2O5 low concentrationsare probably influenced by apatite crystallization duringAFC processes (see below). Additionally, low FeO andTiO2 suggest that Fe-oxides are not a primary meltingcomponent in the source rock. This is consistent with ourproposed source rock composed of olivine, plagioclase,cpx, amphibole,�Fe-oxides.High Cl concentrations in MOR dacites also support the

role of partial melting of rocks altered by seawater-derivedfluids (Michael & Schilling, 1989; Michael & Cornell,1998; Coogan et al., 2003b; Gillis et al., 2003). Although Clbehaves incompatibly during crystallization many MORlavas show over-enrichments compared with other ele-ments with similar compatibilities. For example, after�85% fractional crystallization of a MORB parent (with0·01wt % Cl), there is less than a 10-fold enrichment inCl, resulting in concentrations of �0·07wt %, comparedwith an order of magnitude more (�0·7wt % Cl) in thedacites. Analyses of altered basalt from sheeted dikes indrill holes show that Cl concentrations span a range from49 to 650 ppm (Sparks, 1995). Partial melting (1^15%) ofan amphibole-bearing wall-rock (with 350 ppm Cl) resultsin anatectic melts with 0·3^0·9wt % Clçcovering therange observed in dacites (Fig. 6).Hydrothermal alteration and metamorphism are known

to cause increased concentration of some trace elements,including U, Th, Rb, and Ba, as well as Cl (e.g. Alt &Teagle, 2003). Observed positive anomalies of some highlyincompatible elements (e.g. U and Th) relative to otherincompatible elements with similar distribution coefficientsare consistent with partial melting of hydrothermallyaltered ocean crust (Fig. 12). Partial melting may alsoexplain some of the anomalies in the high field strengthelements whose concentrations are not affected by altera-tion or metamorphism but can be fractionated as a result

of mineralogical effects. For example, the relatively highZr (highly incompatible during melting) concentrationsin the high-silica lavas are a consequence of partial meltingof altered crustal material (Fig. 12). Additionally, the melt-ing and AFC models discussed above point to the impor-tance of amphibole in the melting assemblage to explainthe HREE (compare Fig. 12a and 12b).

The need for crystallization, assimilation and altered crustin dacite petrogenesis

We propose that the partial melting and assimilation ofoceanic crust plays a significant role in the formationhigh-silica MOR lavas; however, we stress that most ofthe heat required to melt the wall-rock is a consequence ofextensive fractional crystallization. Coogan et al. (2003b)showed that the latent heat of crystallization from the for-mation of a 4 km thick gabbro sequence provides enoughenergy to heat �1·3 km of overlying crust from 450 to11508C, which promotes partial melting. It is the assimila-tion of these partial melts into a fractionally crystallizingmagma reservoir that produces the highly evolved meltswith enriched incompatible trace element signatures (e.g.De Paolo, 1981; Be¤ dard et al., 2000).Major element compositions of MOR dacites often lie

between those of experimental partial melts of alteredbasalt (Beard & Lofgren, 1991) and liquids produced bymoderate to large extents of crystal fractionation (Fig. 5).The major element compositions of magmas produced byAFC may, therefore, lie between those of partial melts ofaltered oceanic crust and fractionated basaltic magmas.This is particularly apparent in Al2O3 and may explainwhy very few evolved lavas at the 98NOSC (including fer-robasalts and basaltic andesites) lie on the calculatedliquid lines of descent (Fig. 5).Results from EC-AFC calculations (Bohrson & Spera,

2001) confirm that assimilation of anatectic melts into aresidual fractionated magma can explain a wide range oftrace element concentrations in MOR dacites (Figs 7, 8and 14). The best-fit model for 98N OSC dacite composi-tions requires significant crystal fractionation (73^85wt%) of a ferrobasalt parental magma in combination with10^25% (by mass) anatectic melt, which provides the addi-tional incompatible element enrichments observed. Ourmodels also indicate that to explain the enrichments inRb, Ba, Th and U concentrations relative to other highlyincompatible elements present in the MOR dacites assimi-lation of low-degree partial melts of hydrothermallyaltered oceanic basalt is required. It is important to notethat the extreme Cl enrichments in MOR dacites requirea seawater component that can be derived byAFC processand that Cl over-enrichments observed in many MORBshave been explained by small amounts of assimilation ofeither hydrothermally altered ocean crust or Cl-richbrines stored in the crust (Michael & Schilling, 1989;


2402

at University of F

lorida on Decem


Dow

nloaded from


Michael & Cornell, 1998; Perfit et al., 1999; Coogan et al.,2003a; le Roux et al., 2006).Minor phases, such as FeTi oxides, zircon, and apatite,

can affect both major and trace element concentrations inthe MOR dacites and can potentially provide insight intothe roles of fractional crystallization, partial melting andassimilation. Relatively low Nb and Ta anomalies and lowFeO and TiO2 concentrations may be a consequence ofboth the removal of phenocrysts during late-stage crystalfractionation and residual iron^titanium oxides in the par-tially melted wall-rock. Elevated Zr and Hf concentrationsresult from little to no zircon crystallization in the fractio-nating magma and/or no residual zircon in the meltingassemblage. Apatite crystallization and/or low phosphoruscontents in the source rock can account for the low P2O5

concentrations observed in the MOR dacites. This is sup-ported by experimental results, which suggest that apatitesaturation should occur by 0·7wt % P2O5 (Juster et al.,1989).

Isotopic signature of assimilation

Given that the 98N OSC dacites have radiogenic isotopecompositions similar to those in spatially related basalts(Fig. 9) what effects might assimilation, particularly ofaltered crust, have on derivative melts? AFC processescan change radiogenic isotopic ratios if the assimilant hasrelatively high concentrations of the element in questionand significantly different isotopic ratios from the originalmagma reservoir (e.g. Taylor, 1980; De Paolo, 1981). In gen-eral, however, the effect on isotopic compositions is lessdramatic in AFC processes compared with partial meltingbecause AFC processes create mixtures of altered andfresh material, whereas melting alone will retain the isoto-pic signature of the altered crust.Assimilation of altered oceanic crust may increase Sr

isotope ratios depending on the amount of assimilationand extents of fluid^rock interaction (e.g. Alt & Teagle,2003). Altered oceanic crust can have a range of Sr concen-trations (from less than to greater than typical MORBcompositions) depending on the type or degree of altera-tion (Alt & Teagle, 2003). Assuming 75% fractional crys-tallization (which will not change the isotopic ratios) andassimilation of 10 mass % wall-rock, we can calculate theisotopic composition of the resulting melt using a ratio of2·5:1 (fractionated melt to anatectic melt). Using reason-able values for the isotopic ratios and Sr concentrations ofaltered sheeted dike lavas (0·7028; average of basal dikesfrom Pito Deep; Hess Deep and Hole 504B; Barker et al.,2008) and initial MORB parental magma (0·7025 and100 ppm, a typical altered East Pacific Rise concentration),mass-balance calculations indicate that assimilation ofaltered crust will cause an increase of just �0·0001 in the87Sr/86Sr of the final magma (Fig. 9). EC-AFC modelingof Sr isotopes produces similar results but requires lesscrystal fractionation (73%) and a ratio of fractionated

melt to anatectic melt of �3·5:1 to produce the most radio-genic 87Sr/86Sr signatures observed in the MOR dacites(0·70258). Therefore, this process has the potential toslightly affect the Sr isotope ratios in the dacitic magma,but will not result in ratios as elevated those generateddirectly from partial melting of altered oceanic crust(�0·7028). Additionally, slightly elevated Sr isotope ratiosin high-silica lavas from the Gala¤ pagos Spreading Centerare consistent with AFC processes (Perfit et al., 1999). Incomparison, Nd isotopes are unaffected during fractionalcrystallization and are relatively immobile during hydro-thermal alteration (Michard & Albare' de, 1986; Delacouret al., 2008). Nd isotopes from the OSC dacites are similarto those of basalts from the region.The 98N OSC dacites form a tight cluster in Pb isotopic

composition compared with 98N OSC basalts (Fig. 9). Pbisotopes are not significantly affected by hydrothermalalteration provided sediments (which are not abundant inthis environment) are not involved in the alteration pro-cess (e.g. Perfit et al., 1999). We suggest that the similarityin isotopic ratios in the dacites compared with the basaltsrepresents an overall averaging of isotopic values frombasalts in the region as a result of melting and assimilatinga range of MORB compositions at the base of the sheeteddike layer.Relatively low oxygen isotope ratios observed in MOR

dacites from the Gala¤ pagos Spreading Center (Perfitet al., 1999) and the 98N OSC (Wanless et al., in prepara-tion) also support these conclusions. Fresh MORBs willhave mantle oxygen isotope values (�5·5); however, sea-water alteration (seawater d18O¼ 0) will decrease thisratio (Gillis et al., 2001), whereas fractional crystallizationof Fe-oxides, and to a lesser extent olivine and pyroxene,will cause an increase (Matsuhisa et al., 1973). Therefore,partial melting and assimilation of altered basalt shouldproduce melts with lower d18O values than predicted byfractional crystallization calculations (Muehlenbach &Clayton, 1972). MOR dacites have oxygen isotope ratiossimilar to MORB values (Perfit et al., 1999; Perfit et al.,2007; Wanless et al., in preparation), suggesting that frac-tional crystallization alone cannot explain the formationof dacites on MORs. Taylor (1968) suggested that to a firstapproximation, the effect of assimilation on oxygen iso-topes can be determined using mass balance. Assumingan evolved magma has a d18O value of 6·8 (largely owingto fractionation of silicates and iron oxides) and an assimi-lant has a d18O value of 3·5 (owing to seawater alteration),the resultant oxygen isotope ratio of the AFC magmawould be �6. This value is similar to those observed inthe MOR dacites and is less than predicted by fractionalcrystallization alone.

AFC processes and tectonic settingThe remarkable geochemical similarity of dacites eruptedat the three MORs discussed here indicates that similar


2403

at University of F

lorida on Decem


Dow

nloaded from


processes are controlling their petrogenesis (Fig. 2) and wepropose that these processes are linked to tectono-mag-matic setting. Andesites and dacites have erupted on sev-eral ridges, often in regions of propagation, such aspropagating ridge tips and OSCs (Christie & Sinton, 1981;Perfit et al., 1983; Sinton et al., 1983, 1991). Similar high-silica lavas have also been found along the PacificÂntarctic Rise (Haase et al., 2005), at Axial Seamount onthe Juan de Fuca Ridge (Chadwick et al., 2005) adjacent toa large axial magma chamber, at a ridge^transform inter-section at the Juan de Fuca Ridge (Cotsonika, 2006), at anextinct OSC on the Galapagos Spreading Center (e.g.Perfit et al., 1983), and at the end of first- and second-orderridge segments on the Northern East Pacific Rise(�8837’N and 10830’N; Langmuir et al., 1986). Collectively,most of these lavas erupted on intermediate to fast spread-ing ridges, in settings where magma reservoirs have thepotential to undergo extensive fractional crystallizationand interact with colder, and variably altered crust.A key result of this study is that the geochemical signa-

tures of MOR dacites require assimilation of partial meltsof hydrothermally altered crust into an extremely frac-tionated magma (where the latent heat of crystallizationprovides the heat needed to melt the oceanic crust). Theextensive amounts of fractional crystallization requiredsuggest episodic or sporadic magma supply to magmareservoirs, which may not be characteristic of more‘steady-state’ ridge environments. These requirements aremet at the ends of ridge segments, where magma reservoirsmay have a sporadic magma supply. In these regions,magmas are considered to be fed intermittently to theridge tip through dike propagation from a more robustcentral region (Christie & Sinton, 1981). Between dikingevents the ridge tip magma supply is cut off, allowing forincreased extents of crystal fractionation and interactionof the melt with older, altered crust. This may increase thelikelihood of eruption of high-silica lavas through AFCprocesses. This is not to suggest that AFC processes do notoccur in ‘steady-state’ ridge environments but that thehigh-silica melts may not be preserved or erupted in theseregions (see the section on ‘Effects of assimilation on typi-cal MORB’).

Model for formation of MOR dacitesBased on the similarity of composition of high-silica lavasfrom three MORs, petrological modeling calculationsapplied to dacites at the 98N OSC, and published experi-mental results, we suggest that MOR dacites form underspecific conditions that include: (1) a tectono-magmatic set-ting in which magma injection is episodic, allowing forextensive crystal fractionation; (2) the presence of alteredcrust, which facilitates the geochemical enrichmentsobserved in the MOR dacites. Based on these two require-ments, the tectonic setting and available geophysical

information, we propose the following model for daciteformation on MORs (Fig. 15).

(1) Basaltic magma is injected by down-rift lateral dikepropagation. Axial magma reservoirs are formed.

(2) Magma supply to the region is cut off or reduced,allowing for extensive fractional crystallization of themagma reservoir.

(3) Extensive fractional crystallization and the conse-quent release of latent heat of crystallization initiallyheats, then partially melts the surrounding alteredwall-rock, which might be layer 2b dikes or high-level altered gabbroic rocks.

(4) Anatectic melts are assimilated into the fractionatingmagma body. High-silica AFC melts may vary incomposition depending on crustal temperature,extent of fractional crystallization, amount of anatec-tic melt, and efficiency of assimilation.

This model may account for the formation of highlyevolved magmas at OSCs, propagating ridge tips, ridge^transform intersections and along dikes associated withthe down-rift volcanism on Axial Seamount.This situationmay not be unique to traditional MORs as evidenced byan analogous situation at Krafla volcano in Iceland,where the imaged melt lens is thought to be primarilycomposed of iron-rich basalt but high-silica lavas are asso-ciated with the edges of the caldera rim, where increasedmagma^rock interactions may be likely (Nicholsonet al., 1991).

Relationship of melt lens to dacites at 98NDacitic lavas at 98NOSC erupted on-axis, over the easternedge of a large, seismically imaged melt lens (Kent et al.,2000; Dunn et al., 2001). Despite the eruption of young,fresh high-silica lavas in the neo-volcanic zone, the under-lying melt lens is not assumed to be dacitic in composition.The relatively evolved composition of the ferrobasaltscurrently overlying the main body of the wide axialmagma chamber (Fig. 3) suggests that it has undergone amoderate amount of crystallization (to ferrobasalt).Mixing of the ferrobasaltic and dacitic magmas beneaththe east limb may explain the wide range of compositionserupting at the 98NOSC.The presence of a large, seismically imaged melt lens at

98N OSC does not contradict the episodic magma supplyrequirement for dacite formation. Instead, it may allowand enhance AFC processes in the region. AFC modelingsuggests that extensive fractional crystallization is requiredto produce dacitic compositions, which suggests low orepisodic magma supply. Somewhat antithetically, the 98Nregion has an anomalously large axial magma chamber,suggesting that the current melt lens may be only indir-ectly related to the dacites. Interaction of the eastern edgeof the large melt lens with high-silica lavas may increase


2404

at University of F

lorida on Decem


Dow

nloaded from


Fig. 15. Schematic illustration showing a possible scenario for dacite formation at MORs. (a) Injection of basaltic magma into a ridge segmentend through dike propagation. (b) Magma supply rates diminish at segment end, abandoning pockets of magma and allowing for extensivefractional crystallization. The latent heat of crystallization begins to heat up and partially melt the hydrothermally altered wall-rock.(c) Partial melts of wall-rock are assimilated or mixed into the evolving magma chamber, resulting in dacitic magmas.


2405

at University of F

lorida on Decem


Dow

nloaded from


the volume of dacites erupted at the OSC by acting as amobilizer for the more viscous magmas. This is supportedby geochemical evidence of small-scale local mixing ofhighly evolved compositions with the moderately evolvedbasaltic melt lens beneath the east limb ridge axis, whichmay account for the diversity of compositions eruptedat the OSC (Fig. 15).

Effects of assimilation on typicalMORB compositionsThe formation of dacite compositions on MORs requiresassimilation of anatectic melts into residual fractionatedmagmas; however, AFC processes may also explainslightly elevated incompatible element abundancesobserved in MORB lavas from some sections of MORs(Bryan &Moore,1977).The geochemical signatures of ana-tectic melts may be subtle in less evolved magmas; how-ever, elevated Cl, Al2O3, and K2O are common (Michael& Schilling, 1989; Michael & Cornell, 1998; Perfit et al.,1999; Coogan et al., 2003a; le Roux et al., 2006). At moremagmatically active ridge sections, the wall-rock mayhave a higher initial temperature, which would allow formelting and assimilation to begin earlier in the evolu-tion of the magma body and require less fractional crystal-lization to occur prior to melting (Fig. 13). For example, at8008C assimilation begins after only 53% fractional crys-tallization compared with 68% crystallization for 7208Ccrust. Assimilation of anatectic melts into a magma reser-voir that has undergone less crystallization produces lessevolved compositions and therefore, cannot produceMOR dacites. It does, however, increase the incompatibleelement abundances in the melt phase and may explainthe commonly noted anomalous incompatible elementenrichments, low FeO and elevated Al2O3 concentrationsin some ‘normal’ MORB lavas.

CONCLUSIONSThe majority of eruptions at spreading centers producebasalts with relatively limited chemical variability; how-ever, high-silica lavas have been sampled at several ridges.Eruptions of andesites and dacites are typically associatedwith ridge discontinuities and produce significant volumesof lava at a local scale. Limited amounts of these lavashave been sampled at the southern terminus of the Juande Fuca Ridge, along the eastern Gala¤ pagos spreadingcenter, as well as at the axis at 8837’N and off-axis at�10830’N on the East Pacific Rise. We have documentedmore voluminous eruptions of high-silica lavas includinghighly evolved dacite on the propagating eastern limb ofthe 98N overlapping spreading center (OSC) on the EastPacific Rise. Collectively, the dacites from these variousMOR environments appear to represent a common end-member composition that shows similar major element

trends and incompatible trace element enrichments, sug-gesting similar processes controlled their petrogenesis.The formation of highly evolved lavas on MORs

requires a combination of partial melting, assimilationand crystal fractionation. The highly enriched incompati-ble trace element signatures cannot be produced throughcrystal fractionation alone and appear to require a meltcomponent derived from partial melting of altered oceaniccrust. EC-AFC modeling suggests that significant amounts(475%) of crystallization of a MORB parent magma andmodest amounts (5^20%) of assimilation of hydrother-mally altered oceanic crust can produce geochemical sig-natures consistent with the dacite compositions. The AFCprocess explains the trace element abundances in high-silica lavas and accounts for several major and minor ele-ment concentrations (i.e. Al2O3, K2O and Cl).An important constraint provided by AFC calculations

is the temperature of the assimilant.Varying the wall-rocktemperature can change the amount of crystallization orassimilation that occurs and consequently the overallenrichment in incompatible trace element concentrations.The formation of dacites at the 98N OSC requires thetemperature of the surrounding crust to be 650^7208C,which in turn requires the latent heat of crystallizationfrom 468% fractional crystallization of ferrobasalticmagma before melting can begin. Although this amountof crystallization is unlikely in regions of high or constantmagma supply, the surrounding wall-rock in typicalridge settings may be much warmer than at the ends ofridge segments, allowing for assimilation at much loweramounts of crystallization. At wall-rock temperatures of8008C, our calculations indicate that assimilation beginsafter �53wt % fractional crystallization. This suggeststhat although conditions are not appropriate for the petro-genesis of dacites at typical ridge settings, assimilationof crustal material may be common but geochemicallycryptic.The formation of high-silica lavas on MORs appears

to require a unique tectono-magmatic setting, where episo-dic magma supply allows for extensive crystal fractiona-tion, partial melting and assimilation. These conditionsare met in regions of ridge propagation, such as OSCsand propagating ridge tips, where diking allows for episo-dic injection of magma into older, altered ocean crust.Here, magma undergoes extreme crystallization withoutrepeated replenishment, creating enough latent heatof crystallization to melt and assimilate surroundingwall-rock.

FUNDINGThis work was supported by the RIDGE2000 programof the National Science Foundation (grant numberOCE-0527075 to M.R.P. and OCE-0526120 to E.M.K.).


2406

at University of F

lorida on Decem


Dow

nloaded from


SUPPLEMENTARY DATASupplementary data for this paper are available at Journalof Petrology online.

ACKNOWLEDGEMENTSWe thank the Captain, officers and crew of the R.V. Atlantisfor all their help during cruise AT15-17, theMEDUSA2007 Science Party (including S.White, K. VonDamm, D. Fornari, A. Soule, S. Carmichael, K. Sims,A. Zaino, A. Fundis, J. Mason, J. O’Brien, C. Waters,F. Mansfield, K. Neely, J. Laliberte, E. Goehring andL. Preston) for their diligence in collecting data and sam-ples for this study. We thank the Jason II shipboard andshore-based operations group for their assistance in collect-ing these data, and HMRG for processing all DSL-120side-scan and bathymetry data collected during thiscruise. Discussions with S. White and A. Goss are grate-fully acknowledged, and contributed to this research.Thanks go to G. Kamenov and the UF Center for IsotopeGeoscience for laboratory assistance. We thankW. Bohrson for both the editorial handling of this manu-script and extensive reviews. L. Coogan, J. Maclennan,and F. Meade are thanked for their thorough and insight-ful reviews.

REFERENCESAlt, J., Honnorez, J., Laverne, C. & Emmerman, R. (1986).

Hydrothermal alteration of a 1km section through the upperoceanic crust, Deep Sea Drilling Project Hole 504B: mineralogy,chemistry and evolution of seawater^basalt interations. Journal ofGeophysical Research 91, 10309^10335.

Alt, J. C. & Teagle, D. A. H. (2003). Hydrothermal alteration of upperoceanic crust formed at a fast-spreading ridge: mineral, chemical,and isotopic evidence from ODP Site 801. Chemical Geology 201,191^211.

Aumento, F. (1969). Diorites from the mid-Atlantic ridge at 458N.Science 165, 1112^1113.

Bachmann, O. & Bergantz, G.W. (2004). On the origin of crystal-poorrhyolites: extracted from batholithic crystal mushes. Journal of

Petrology 45, 1565^1582.Barker, A. K., Coogan, L. A. & Gillis, K. M. (2008). Strontium iso-

tope constraints on fluid flow in the sheeted dike complex of fastspreading crust: Pervasive fluid flow at Pito Deep. Geochemistry,

Geophysics, Geosystems 9, 1019.Bazin, S., Harding, A. J., Kent, G. M., Orcutt, J. A., Tong, C. H.,

Pye, J. W., Singh, S. C., Barton, P. J., Sinha, M. C., White, R. S.,Hobbs, R.W. & Van Avendonk, H. J. A. (2001). Three-dimensionalshallow crustal emplacement at the 9803’N overlapping spreadingcenter on the East Pacific Rise: Correlations between magnetiza-tion and tomographic images. Journal of Geophysical Research 106,16101^16117.

Beard, J. S. & Lofgren, G. E. (1991). Dehydration melting and water-saturated melting of basaltic and andesitic greenstones and am-phibolites at 1, 3, and 6·9 kb. Journal of Petrology 32, 365^401.

Beccaluva, L., Chinchilla-Chaves, A. L., Coltorti, M., Giunta, G.,Siena, F. & Vaccaro, C. (1999). Petrological and structural signifi-cance of the Santa Elena^Nicoya ophiolitic complex in Costa

Rica and geodynamic implications. Contributions to Mineralogy and

Petrology 11, 1091^1107.Be¤ dard, J. H., He¤ bert, R., Berclaz, A. & Varfalvy,V. (2000). Syntexis

and the genesis of lower oceanic crust. In: Dilek, Y.,Moores, E. M., Elthon, D. & Nicolas, A. (eds) Ophiolites and

Oceanic Crust: New Insights from Field Studies and the Ocean Drilling

Program. Geological Society of America, Special Papers 349, 105^119.Bohrson, W. A. & Reid, M. R. (1997). Genesis of silicic peralkaline

volcanic rocks in an ocean island setting by crustal melting andopen-system processes: Socorro Island, Mexico. Journal of Petrology38, 1137^1166.

Bohrson,W. A. & Reid, M. R. (1998). Genesis of evolved ocean islandmagmas by deep- and shallow-level basement recycling, SocorroIsland, Mexico: constraints fromTh and other isotope signatures.Journal of Petrology 39, 995^1008.

Bohrson, W. A. & Spera, F. (2001). Energy-constrained open-systemmagmatic processes ii: application of energy constrained assimila-tion^fractional crystallization (EC-AFC) model to magmaticsystems. Journal of Petrology 42, 1019^1041.

Bowen, N. L. (1928). The Evolution of Igneous Rocks. Princeton, NJ:Princeton University Press.

Brophy, J. G. (2009). La^SiO2 andYb^SiO2 systematics in mid-oceanridge magmas: implications for the origin of oceanic plagiogranite.Contributions to Mineralogy and Petrology 158, 99^111.

Bryan,W. B. & Moore, J. G. (1977). Compositional variations of youngbasalts in the Mid-Atlantic Ridge rift valley near 36849’N.Geological Society of America Bulletin 88, 556^570.

Byerly, G. R. (1980). The nature of differentiation trends in somevolcanic rocks from the Galapagos Spreading Center. Journal ofGeophysics 85, 3797^3810.

Byerly, G. R. & Melson, W. G. (1976). Rhyodacites, andesites,ferro-basalts and ocean tholeiites from the Galapagos spreadingcenter. Earth and Planetary Science Letters 30, 215^221.

Canales, P. J., Detrick, R. S., Carbotte, S. M., Kent, G. M.,Diebold, J. B., Harding, A. J., Babcock, J., Nedimovie, M. R. &van Ark, E. (2005). Upper crustal structure and axial topographyat intermediate spreading ridges: Seismic constraints from thesouthern Juan de Fuca Ridge. Journal of Geophysical Research 110,1^27.

Carbotte, S. M., Arko, R., Chayes, D. N., Haxby, W., Lehnert, K.,O’Hara, S., Ryan, W. B. F., Weissel, R. A., Shipley, T.,Gahagan, L., Johnson, K. T. M. & Shank, T. M. (2004). Newintegrated data management system for Ridge2000 andMARGINS research. EOS, Transactions, American Geophysical Union

85(51), 553.Carbotte, S. M. & Macdonald, K. C. (1992). East Pacific Rise

88^10830’N: Evolution of ridge segments and discontinuities fromSeaMARC II and three-dimensional magnetic studies. Journal ofGeophysical Research 97, 6959^6982.

Casey, J. F. (ed.) (1997). Comparison of major- and trace-element geochemis-try of abyssal peridotites and mafic plutonic rocks with basalts from the

MARK region of the Mid-Atlantic Ridge. College Station: OceanDrilling Program.

Chadwick, J., Perfit, M. R., Ridley, W. I., Jonasson, I. R.,Kamenov, G., Chadwick,W.W. J., Embley, R.W., le Roux, P. J. &Smith, M. C. (2005). Magmatic effects of the Cobb hot spot on theJuan de Fuca Ridge. Journal of Geophysical Research 110, 1^16.

Christie, D. M. & Sinton, J. M. (1981). Evolution of abyssal lavas alongpropagating segments of the Galapagos Spreading Center. Earthand Planetary Science Letters 56, 321.

Clague, D. A. & Bunch, T. E. (1976). Formation of ferrobasalt atEast Pacific mid-ocean spreading centers. Journal of Geophysical

Research 81, 4247^4256.


2407

at University of F

lorida on Decem


Dow

nloaded from


Coleman, R. G. & Donato, M. M. (1979). Oceanic Plagiogranite Revisited.Amsterdam: Elsevier.

Coogan, L. A. (2003). Contaminating the lower crust in the Omanophiolite. Geology 31, 1065^1068.

Coogan, L. A., Banks, G. J., Gillis, K. M., MacLeod, C. J. &Pearce, J. A. (2003a). Hidden melting signatures recorded in theTroodos ophiolite plutonic suite: evidence for widespread gener-ation of depleted melts and intra-crustal melt aggregation.Contributions to Mineralogy and Petrology 144, 484^505.

Coogan, L. A., Mitchell, N. C. & O’Hara, M. J. (2003b). Roofassimilation at fast spreading ridges: An investigation combininggeochemical and field evidence. Journal of Geophysical Research 108,1^14.

Cotsonika, L. (2006). Petrogenesis of Andesites and Dacites from the Southern

Juan De Fuca Ridge. Masters Thesis, Gainesville, FL: Departmentof Geological Sciences, University of Florida, 188 p.

Danyushevsky, L. V. (2001). The effect of small amounts of H2O oncrystallization of mid-ocean ridge and backarc basin magmas.Journal of Volcanology and Geothermal Research 110, 265^280.

De Paolo, D. J. (1981). Trace element and isotopic effects of combinedwallrock assimilation and fractional crystallization. Earth and

Planetary Science Letters 53, 189^202.Delacour, A., Fruh-Green, G. L., Frank, M., Gutjahr, M. & Kelley, D.

S. (2008). Sr- and Nd-isotope geochemistry of the Atlantis Massif(308N, MAR): Implications for fluid fluxes and lithosphericheterogeneity. Chemical Geology 254, 19^35.

Detrick, R. S., Buhl, P.,Vera, E., Mutter, J., Orcutt, J. A., Madsen, J.& Brocher, T. (1987). Multichannel seismic imaging of a crustalmagma chamber along the East Pacific Rise. Nature 326, 35^41.

Dick, J. B., Natland, J. H., Alt, J., Bach, W., Bideau, D., Gee, J. S.,Haggas, S., Hertogen, J. G. H., Hirth, G., Holm, P. M.,Ildefonse, B., Iturrino, G. J., John, B. E., Kelley, D. S.,Kikawa, E., Kingdon, A., le Roux, P. J., Maeda, J., Meyer, P. S.,Miller, D. J., Naslund, H. R., Niu, Y., Robinson, P. T., Snow, J.,Stephen, R. A., Trimby, P. W., Worm, H. U. & Yoshinobu, A.(2000). A long in situ section of the lower ocean crust: results ofODP Leg 176 drilling at the Southwest Indian Ridge. Earth and

Planetary Science Letters 179, 31^51.Dunn, R. A.,Toomey, D. R., Detrick, R. S. & Wilcock,W. S. D. (2001).

Continuous mantle melt supply beneath an overlapping spreadingcenter on the East Pacific Rise. Science 291, 1955^1958.

Embley, R. W., Chadwick, W. W. J., Clague, D. A. & Stakes, D. S.(1999). 1998 eruption of Axial Volcano: Multibeam anomalies andseafloor observations. Geophysical Research Letters 26, 3425^3428.

Embley, R.W., Chadwick,W.W. J., Perfit, M. R. & Baker, E.T. (1991).Geology of the northern Cleft segment, Juan de Fuca Ridge:Recent lava flows, sea-floor spreading, and the formation of mega-plumes. Geology 19, 771^775.

Embley, R. W., Jonasson, I. R., Perfit, M. R., Franklin, J. M.,Tivey, M. A., Malahoff, A., Smith, M. F. & Francis, T. J. G.(1988). Submersible investigations of an extinct hydrothermalsystem on the Galapagos Ridge: Sulfide mound, stockwork zone,and differentiated lavas. Canadian Mineralogy 26, 1^35.

Embley, R. W. & Wilson, D. (1992). Morphology of the BlancoTransform Fault zone^NE Pacific: Implications for its tectonic evo-lution. Marine Geophysical Researches 14, 25^45.

Fornari, D. J., Perfit, M. R., Malahoff, A. & Embley, R. W. (1983).Geochemical studies of abyssal lavas recovered by DSRV Alvin

from the Eastern Galapagos Rift, Inca Transform, and EcuadorRift 1. Major element variation in natural glasses and spatial distri-bution of lavas. Journal of Geophysical Research 88, 10159^10529.

France, L., Ildefonse, B. & Koepke, J. (2009). Interactions betweenmagma and hydrothermal system in Oman ophiolite and in

IODP Hole 1256D: Fossilization of a dynamic melt lens at fastspreading ridges. Geochemistry, Geophysics, Geosystems 10, 1^30.

Ghiorso, M. S. & Sack, R. O. (1995). Chemical mass transfer inmagmatic processes. IV. A revised and internally consistentthermodynamic model for the interpolation and extrapolation ofliquid^solid equilibria in magmatic systems at elevated tempera-tures and pressures. Contributions to Mineralogy and Petrology 119,197^212.

Gillis, K. M. (2008). The roof of an axial magma chamber: A hornfel-sic heat exchanger. Geology 36, 299^302.

Gillis, K. M. & Coogan, L. A. (2002). Anatectic migmatites fromthe roof of an ocean ridge magma chamber. Journal of Petrology 43,2075^2095.

Gillis, K. M., Coogan, L. A. & Chaussidon, M. (2003).Volatile elem-ent (B, Cl, F) behaviour in the roof of an axial magma chamberfrom the East Pacific Rise. Earth and Planetary Science Letters 213,447^462.

Gillis, K. M., Muehlenbach, K., Stewart, M., Gleeson, T. &Karson, J. A. (2001). Fluid flow patterns in fast spreading EastPacific Rise crust exposed at Hess Deep. Journal of Geophysical

Research 106, 26311^26329.Gillis, K. M. & Roberts, M. D. (1999). Cracking at the magma^hydro-

thermal transition: Evidence from the Troodos ophiolite, Cyprus.Earth and Planetary Science Letters 169, 227^244.

Goss, A., Perfit, M., Ridley, W. I., Rubin, K. H., Kamenov, G.,Soule, A. S., Fundis, A. & Fornari, D. J. (2010). Geochemistry oflavas from the 2005^2006 eruption at the East Pacific Rise,9846’N^9856’N: Implications for ridge crest plumbing and decadalchanges in magma chamber compositions. Geochemistry, Geophysics,Geosystems 11, 1^35.

Haase, K. M., Stroncik, N. A., He¤ kinian, R. & Stoffers, P. (2005).Nb-depleted andesites from the PacificÂntarctic Rise as analogsfor early continental crust. Geology 33, 921^924.

Hacker, B. R. (1990). Amphibolite-facies to granulite-facies reactionsin experimentally deformed, unpowered amphibolite. American

Mineralogist 75, 1349^1361.Harding, A. J., Kent, A. J. R. & Orcutt, J. A. (1993). A multichannel

seismic investigation of upper crustal structure at 98N on theEast Pacific Rise: Implications for crustal accretion. Journal of

Geophysical Research 98, 13925^13944.Hildreth,W. (1981). Gradients in silicic magma chambers: implications

for lithospheric magmatism. Journal of Geophysical Research 86,10153^10192.

Johannes, W. & Koepke, J. (2001). Incomplete reaction of plagioclasein experimental dehydration melting of amphibolite. Australian

Journal of Earth Science 48, 581^590.Juster, T. C., Grove, T. L. & Perfit, M. R. (1989). Experimental con-

straints on the generation of FeTi basalts, andesites, and rhyodacitesat the Galapagos Spreading Center, 858W and 958W. Journal ofGeophysical Research 94, 9251^9274.

Kamenov, G., Saunders, A. D. & Hames,W. E. (2007). Mafic magmasas sources for gold in Middle-Miocene epithermal deposits ofNorthern Great Basin, USA: evidence from Pb isotopic compos-itions of native gold. Economic Geology 102, 1191^1195.

Karsten, & Delaney, J. R. (1989). Hot spot^ridge crest convergence inthe northeast Pacific. Journal of Geophysical Research 94, 700^712.

Kent, A. J. R., Harding, A. J. & Orcutt, J. A. (1993). Distribution ofmagma beneath the East Pacific Rise near 9803’N overlappingspreading center from forward modeling of CDP data. Journal ofGeophysical Research 98, 13971^13995.

Kent, A. J. R., Peate, D. W., Newman, S., Stolper, E. M. &Pearce, J. A. (2002). Chlorine in submarine glasses from the Lau


2408

at University of F

lorida on Decem


Dow

nloaded from


Basin: seawater contamination and constraints on the compositionof slab-derived fluids. Earth and Planetary Science Letters 202, 361^377.

Kent, G. M., Singh, S. C., Harding, A. J., Sinha, M. C., Orcutt, J. A.,Barton, P. J., White, R. S., Bazin, S., Hobbs, R.W., Tong, C. H. &Pye, J.W. (2000). Evidence from three-dimensional seismic reflect-ivity images for enhanced melt supply beneath mid-ocean ridgediscontinuities. Nature 406, 614^618.

Klein, E. M. (ed.) (2005). Geochemistry of the Igneous Oceanic Crust.Oxford: Elsevier^Pergamon.

Koepke, J., Berndt, J., Feig, S. T. & Holtz, F. (2007). The formation ofSiO2-rich melts within the deep oceanic crust by hydrous partialmelting of gabbros. Contributions to Mineralogy and Petrology 153,67^84.

Koepke, J., Feig, S. T., Snow, J. & Freise, M. (2004). Petrogenesisof oceanic plagiogranites by partial melting of gabbros: an experi-mental study. Contributions to Mineralogy and Petrology 146, 414^432.

Kvassnes, A. J. S. & Grove, T. L. (2008). How partial melts of maficlower crust affect ascending magmas. Contributions to Mineralogy and

Petrology 156, 49^71.Langmuir, C. H. (1989). Geochemical consequences of in situ crystal-

lization. Nature 340, 199^205.Langmuir, C. H., Bender, J. F. & Batiza, R. (1986). Petrological and

tectonic segmentation of the East Pacific Rise, 5830’^14830’N.Nature 322, 422^429.

le Roux, P. J., Shirey, S. B., Hauri, E. H., Perfit, M. R. & Bender, J. F.(2006). The effects of variable sources, processes and contaminantson the composition of northern EPR MORB (8^108N and12^148N): Evidence from volatiles (H2O, CO2, S) and halogens(F, Cl). Earth and Planetary Science Letters 251, 209^231.

Maclennan, J. (2008). The supply of heat to mid-ocean ridges bycrystallization and cooling of mantle melts. In: Lowell, R. P.,Perfit, M. R., Seewald, J. & Metaxas, A. (eds) Magma to Microbe:

Modeling Hydrothermal Processes at Oceanic Spreading Ridges. Geophysical

Monograph, American Geophysical Union 178, 45^73.Martin, E. & Sigmarsson, O. (2007). Crustal thermal state and origin

of silicic magma in Iceland: the case of Torfajokull, Ljosufjoll andSnaefellsjokull volcanoes. Contributions to Mineralogy and Petrology

153, 593^605.McBirney, A. R. (1993). Differentiated rocks of the Galapagos hotspot.

In: Prichard, H. M. (ed.) Magmatic Processes and Plate Tectonics.

Geological Society, London, Special Publications 76, 61^69.Michael, P. & Cornell, W. C. (1998). Influence of spreading rate and

magma supply on crystallization and assimilation beneathmid-ocean ridges: evidence from chlorine and major element chem-istry of mid-ocean ridge basalts. Journal of Geophysical Research 103,18325^18356.

Michael, P. & Schilling, J. G. (1989). Chlorine in mid-ocean ridgemagmas: Evidence for assimilation of seawater-influencedcomponents. Geochimica Cosmochemica Acta 53, 3131^3143.

Michard, A. & Albare' de, F. (1986). The REE content of some hydro-thermal Fuids. Chemical Geology 55, 51^60.

Muehlenbach, K. & Clayton, R. N. (1972). Oxygen isotope studies offresh and weathered submarine basalts. Canadian Journal of EarthScience 2, 172^184.

Nakamura, K., Kato,Y., Tamaki, K. & Ishii, T. (2007). Geochemistryof hydrothermally altered basaltic rocks from the SouthwestIndian Ridge near the Rodriquez Triple Junction. Marine Geology

239, 125^141.Natland, J. H., Langmuir, C. H., Bender, J. F., Batiza, R. &

Hopson, C. (1986). Petrologic systematics in the vicinity of the 9 de-grees N nontransform offset, East Pacific Rise. EOS, Transactions,American Geophysical Union 67, 1254.

Natland, J. H. & MacDougall, J. D. (1986). Parental abyssal tholeiitesand alkali basalts at the East Pacific Rise near 9 degrees N andthe Siqueiros fracture zone. EOS, Transactions, American Geophysical

Union 67, 410^411.Nicholson, H., Condomines, M., Fitton, J. G., Fallick, A. E.,

Gronvold, K. & Rogers, G. (1991). Geochemical and isotopic evi-dence for crustal assimilation beneath Krafla, Iceland. Journal ofPetrology 32, 1005^1020.

Niu,Y., Gilmore, T., Mackie, S., Greig, A. & Bach,W. (2002). Mineral

chemistry, whole-rock compositions, and petrogenesis of Leg 176 gabbros: data

and discussion. College Station: Ocean Drilling Program.Nunnery, A., Klein, E. M., Perfit, M., White, R. S., Mason, J. L. &

Zaino, A. J. (2008). Correlation of seafloor surface features andunderlying melt bodies at the 98N overlapping spreading center,East Pacific Rise. EOS, Transactions, American Geophysical Union

89(53), Fall Meeting Supplement, Abstract V53D-03.O’Nions, R. K. & Gronvold, K. (1973). Petrogenetic relationships of

acid and basic rocks in Iceland: Sr-isotopes and rare-earth elementsin late and postglacial volcanics. Earth and Planetary Science Letters

19, 397^409.Pedersen, R. B. & Malpas, J. (1984). The origin of oceanic plagiogra-

nites from the Karmoy ophiolite, western Norway. Contributions toMineralogy and Petrology 88, 36^52.

Perfit, M. R. & Chadwick,W.W. J. (1998). Magmatism at mid-oceanridges: constraints from volcanological and geochemicalinvestigations. In: Buck, W. R., Delaney, P. T. J., Karson, J. A. &Lagabrielle, Y. (eds) Faulting and Magmatism at Mid-Ocean

Ridges. Geophysical Monograph, American Geophysical Union 106, 59^115.Perfit, M. R. & Fornari, D. J. (1983). Geochemical studies of abyssal

lavas recovered by DSRV Alvin from Eastern Galapagos Rift,Inca Transform, and Ecuador Rift 2. Phase chemistry and

crystallization history. Journal of Geophysical Research 88,10530^10550.

Perfit, M. R., Fornari, D. J., Malahoff, A. & Embley, R. W. (1983).Geochemical studies of abyssal lavas recovered by DSRV Alvinfrom eastern Galapagos Rift, IncaTransform, and Ecuador Rift 3.Trace element abundances and petrogenesis. Journal of GeophysicalResearch 88, 10551^10572.

Perfit, M. R., Ridley, W. I. & Jonasson, I. R. (1999). Geologic,petrologic and geochemical relationships between magmatismand massive sulfide mineralization along the eastern GalapagosSpreading Center. Reviews in Economic Geology 8, 75^99.

Perfit, M. R., Schmitt, A. K., Ridley,W. I., Rubin, K. H. & Valley, J.(2007). Petrogenesis of dacites from the southern Juan de FucaRidge. Goldschmidt Conference Abstracts, Geochimica et Cosmochimica

Acta 72, A736.Pollock, M. A., Klein, E. M., Karson, J. A. & Tivey, M. A. (2005).

Temporal and spatial variability in the composition of lavasexposed along the Western Blanco Transform Fault. Geochemistry,Geophysics, Geosystems 6, 1^17.

Rapp, R. P., Watson, E. B. & Miller, C. F. (1991). Partial melting ofamphibolite/eclogite and the origin of Archean trondhjemites andtonalites. Precambrian Research 51, 1^25.

Regelous, M., Niu, Y., Went, J. I., Batiza, R., Greig, A. &Collerson, K. D. (1999). Variations in geochemistry of magmatismon the East Pacific Rise at 10830’N since 800 ka. Earth and

Planetary Science Letters 168, 45^63.Reynolds, J. R. (1995). Segment-scale Systematics of Mid-ocean Ridge

Magmatism and Geochemistry. PhD Thesis, Palisades, NY:Columbia University, 483 p.

Reynolds, J. R. & Langmuir, C. H. (1997). Petrological systematics ofthe Mid-Atlantic Ridge south of Kane: Implications for oceancrust formation. Journal of Geophysical Research 102, 14915^14946.


2409

at University of F

lorida on Decem


Dow

nloaded from


Rubin, K. H. & Sinton, J. M. (2007). Inferences on mid-ocean ridgethermal and magmatic structure from MORB compositions. Earthand Planetary Science Letters 260, 257^276.

Sempere, J.-C. & Macdonald, K. C. (1986). Deep-tow studies of theoverlapping spreading centers at 9803’N on the East Pacific Rise.Tectonics 5, 881^900.

Sigurdsson, H. (1977). Generation of Icelandic rhyolites by melting ofplagiogranites in the oceanic layer. Nature 269, 25^28.

Sigurdsson, H. & Sparks, R. S. J. (1981). Petrology of rhyolitic andmixed magma ejecta from the 1875 eruption of Askja, Iceland.Journal of Petrology 22, 41^84.

Sims, K. W. W., Blichert-Toft, J., Fornari, D. J., Perfit, M. R.,Goldstein, S. J., Johnson, P., De Paolo, D. J. & Michael, P. (2003).Abberant youth: Chemical and isotopic constraints on the youngoff-axis lavas of the East Pacific Rise. Geochemistry, Geophysics,

Geosystems 4, 8621.Sims, K. W. W., Goldstein, S. J., Blichert-Toft, J., Perfit, M. R.,

Kelemen, P. B., Fornari, D. J., Michael, P., Murrell, M. T.,Hart, S. R., DePaolo, D. J., Layne, G., Ball, L., Jull, M. &Bender, J. F. (2002). Chemical and isotopic constraints on the gener-ation and transport of magma beneath the East Pacific Rise.Geochimica et Cosmochimica Acta 66, 3481^3504.

Singh, S. C., Harding, A. J., Kent, G. M., Sinha, M. C., Combier,V.,Bazin, S., Tong, C. H., Pye, J. W., Barton, P. J., Hobbs, R. W.,White, R. S. & Orcutt, J. A. (2006). Seismic reflection images ofthe Moho underlying melt sills at the East Pacific Rise. Nature

442, 1^4.Sinton, J. M., Ford, L. L., Chappell, B. & McCulloch, M. T. (2003).

Magma genesis and mantle heterogeneity in the Manus back-arcbasin, Papua New Guinea. Journal of Petrology 44, 159^195.

Sinton, J. M., Smaglik, S. M. & Mahoney, J. J. (1991). Magmatic pro-cesses at superfast spreading mid-ocean ridges: glass compositionalvariations along the East Pacific Rise 138^238S. Journal of

Geophysical Research 96, 6133^6155.Sinton, J. M.,Wilson, D., Christie, D. M., Hey, R. N. & Delaney, J. R.

(1983). Petrological consequences of rift propagation on oceanicspreading ridges. Earth and Planetary Science Letters 62, 193^207.

Smith, M. C., Perfit, M. R., Fornari, D. J., Ridley,W. I., Edwards, M.H., Kurras, G. J. & Von Damm, K. L. (2001). Magmatic processesand segmentation at a fast spreading mid-ocean ridge: Detailed in-vestigation of an axial discontinuity on the East Pacific Rise crestat 9837’N. Geochemistry, Geophysics, Geosystems 2, 1^32.

Smith, M. C., Perfit, M. R. & Jonasson, I. R. (1994). Petrology andgeochemistry of basalts from the Southern Juan de Fuca Ridge:Controls on the spatial and temporal evolution of mid-ocean ridgebasalts. Journal of Geophysical Research 99, 4787^4812.

Sparks, J. W. (1995). Geochemistry of the Lower Sheeted DikeComplex, Hole 504B, Leg 140. In: Keir, B., Graham, A. G. &Alt, J. C. et al. (eds) Proceedings of the Ocean Drilling Program,

Scientific Results, 137/140. College Station, TX: Ocean DrillingProgram, pp. 81^97.

Stakes, D. S., Perfit, M. R., Tivey, M. A., Caress, D.W., Ramirez, T.M. & Maher, N. (2006). The Cleft revealed: Geologic, magnetic,

and morphologic evidence for construction of upper oceanic crustalong the southern Juan de Fuca Ridge. Geochemistry, Geophysics,

Geosystems 7, 1^32.Sun, S. & McDonough,W. F. (1989). Chemical and isotopic systema-

tics of oceanic basalts: implications for mantle composition andprocesses. In: Saunders, A. D. & Norry, M. J. (eds) Magmatism in

the Ocean Basins. Geological Society, London, Special Publications 42,313^345.

Taylor, H. P. J. (1980). The effects of assimilation of country rocks bymagmas on 18O/16O and 87Sr/86Sr systematics in igneous rocks.Earth and Planetary Science Letters 47, 243.

Tong, C. H., Pye, J. W., Barton, P. J., White, R. S., Sinha, M. C.,Singh, S. C., Hobbs, R.W., Bazin, S., Harding, A. J., Kent, G. M.& Orcutt, J. A. (2002). Asymmetric melt sills and uppercrustal construction beneath overlapping ridge segments:Implications for the development of melt sills and ridge crests.Geology 30, 83^86.

Van Der Zander, I., Sinton, J. M. & Mahoney, J. J. (2010). Lateshield-stage silicic magmatism atWai’anae volcano: evidence for hy-drous crustal melting in Hawaiian volcanoes. Journal of Petrology

51, 671^701.Wanless,V. D. (2010). Geology and petrogenesis of lavas from an over-

lapping spreading center: 98N East Pacific Rise, PhD thesis,University of Florida, Gainesville, 178 pp.

Wanless, D., Perfit, M., Ridley,W. I. & Klein, E. M. (2008). Origin ofdacites on mid-ocean ridges. EOS,Transactions, American Geophysical

Union 89(53), Fall Meeting Supplement, Abstract V41B-2084.West, M., Menke,W.,Tolstoy, M.,Webb, S. & Sohn, R. (2001). Magma

storage beneath Axial volcano on the Juan de Fuca mid-oceanridge. Nature 413, 833^836.

White, S. M., Mason, J. L., Macdonald, K. C., Perfit, M. R.,Wanless, V. D. & Klein, E. M. (2009). Significance of widespreadlow effusion rate eruptions over the past two million years for deliv-ery of magma to the overlapping spreading centers at 98N EastPacific Rise. Earth and Planetary Science Letters 280, 175^184.

Wilson, D. S., Teagle, D. A. H., Alt, J., Banerjee, N. R., Umino, S.,Miyashita, S., Acton, G. D., Anma, R., Barr, S. R., Belghoul, A.,Carlut, J., Christie, D. M., Coggon, R. M., Cooper, K. M.,Cordier, C., Crispini, L., Durand, S. R., Einaudi, F., Galli, L.,Gao, Y., Geldmacher, J., Gilbert, L. A., Hayman, N.W., Herrero-Bervera, E., Hirano, N., Holter, S., Ingle, S., Jiang, S.,Kalberkamp, U., Kerneklian, M., Koepke, J., Laverne, C.,Vasquez, H. L. L., Maclennan, J., Morgan, S., Neo, N.,Nichols, H. J., Park, S.-H., Reichow, M. K., Sakuyama, T.,Sano, T., Sandwell, R., Scheibner, B., Smith-Duque, C. E.,Swift, S. A., Tartarottie, P., Tikku, A. A., Tominaga, M.,Veloso, E. A., Yamasaki, T., Yamazaki, S. & Ziegler, C. (2006).Drilling to gabbro in intact ocean crust. Science 312, 1016^1020.

Wolf, M. B. & Wyllie, P. J. (1994). Dehydration-melting of amphiboliteat 10 kbar: the effects of temperature and time. Contributions to

Mineralogy and Petrology 115, 369^383.


2410

at University of F

lorida on Decem


Dow

nloaded from


Dacite Petrogenesis on Mid-Ocean Ridges: Evidence for Oceanic Crustal Melting and Assimilation

Documents