control of the symmetry of plume ridge interaction by...

Article

Volume 11

2 July 2010

Q0AC05, doi:10.1029/2009GC002986

ISSN: 1525‐2027

ClickHere

for

FullArticle

Control of the symmetry of plume‐ridge interactionby spreading ridge geometry

O. Shorttle and J. MaclennanDepartment of Earth Sciences, University of Cambridge, Downing Street, Cambridge CB2 3EQ, UK([email protected])

S. M. JonesSchool of Geography, Earth and Environmental Science, University of Birmingham, Birmingham B152TT, UK

[1] The Iceland, Galápagos, and Azores plumes have previously been identified as interacting asymmetri-cally with adjacent spreading centers. We present evidence that the flow fields in these plume heads areradially symmetric, but the geometry of the mid‐ocean ridge systems imparts an asymmetric compositionalstructure on outflowing plume material. First, we quantify the degree of symmetry in geophysical and geo-chemical observables as a function of plume center location. For each plume, we find that bathymetry andcrustal thickness observations can be explained using a single center of symmetry, with these calculatedcenters coinciding with independently inferred plume center locations. The existence of these centers ofsymmetry suggests that the flow fields and temperature structure of the three plume heads are radially sym-metric. However, no centers of symmetry can be found for the incompatible trace element and isotopicobservations. To explain this, we develop a simple kinematic model to predict the effect of mid‐ocean ridgegeometry on the chemical composition of outflowing plume material. The model assumes radially symmetricoutflow from a compositionally heterogeneous plume source, consisting of a depleted mantle component andenriched blebs. These blebs progressively melt out during flow through the melting regions under spreadingcenters. Asymmetry in trace element and isotopic profiles develops when ridges on either side of the plumecenter receive material that has been variably depleted according to the length of flow path under the ridge.This model can successfully explain compositional asymmetry around Iceland and Galápagos in terms of anaxisymmetric plume interacting with an asymmetric ridge system.

Components: 15,700 words, 13 figures, 1 table.

Keywords: Iceland; Galapagos; Azores; plume‐ridge interaction; mantle plumes.

Index Terms: 8121 Tectonophysics: Dynamics: convection currents, and mantle plumes (3614); 1032 Geochemistry: Mid‐oceanic ridge processes (3614); 1009 Geochemistry: Geochemical modeling (3610).

Received 1 December 2009; Revised 7 May 2010; Accepted 17 May 2010; Published 2 July 2010.

Shorttle, O., J. Maclennan, and S. M. Jones (2010), Control of the symmetry of plume‐ridge interaction by spreading ridgegeometry, Geochem. Geophys. Geosyst., 11, Q0AC05, doi:10.1029/2009GC002986.

Theme: Geochemical Heterogeneities in Oceanic Island Basalt and Mid‐ocean RidgeBasalt Sources: Implications for Melting Processes and Mantle DynamicsGuest Editors: C. Beier and P. Asimow

Copyright 2010 by the American Geophysical Union 1 of 27

http://dx.doi.org/10.1029/2009GC002986

1. Introduction

[2] The role of mantle plumes in generating geo-chemical and geophysical anomalies is well docu-mented [Schilling, 1973; Schilling et al., 1982; Yuet al., 1997; Hooft et al., 2006] and has beenlinked to their anomalous temperatures [Courtneyand White, 1986], flow fields [Maclennan et al.,2001] and compositions [Hart, 1971]. On impact-ing the base of the lithosphere plume material isforced to flow laterally, advecting the thermal andcompositional signal of the plume away from theupwelling stalk. When spreading ridges lie closeto plumes, the geochemical and geophysical con-sequences of this plume outflow for ridge magma-tism can be pronounced: oceanic crustal thickness,ridge axial depth, axial morphology and eruptedbasalt geochemistry all show pronounced long‐wavelength (∼1000 km) deviations. It is the along‐ridge distribution of this plume signal that concernsthe current study.

[3] The simplest conceptual model for the inter-action of plumes and ridges has been of axisym-metric plume influence along adjacent spreadingcenters. This model naturally leads to the predic-tion that geochemical and geophysical observablesalong ridges should be the same at a given radialdistance from the plume center. However, it hasbeen suggested that this idealized situation doesnot hold for the Iceland, Galápagos and Azoresplumes, from the observation that the geophysicaland geochemical signatures are not distributedsymmetrically along the ridge axes. Comparingoceanic crustal thickness and bathymetric profilesnorth and south of Iceland (Figure 1a), Hooft et al.[2006] observed a 200–500 m greater elevationand 2–2.5 km thicker crust along the ReykjanesRidge compared with the Kolbeinsey Ridgebeyond 150 km radial distance from the plumecenter. Along‐ridge geochemical profiles, con-structed from analyses of dredged basalts, havealso been identified as differing either side of theplume. In particular, Sr and Pb isotopic composi-tions at the Reykjanes Ridge south of Iceland areenriched in comparison with those at the Kol-beinsey Ridge to the north [Poreda et al., 1986;Mertz et al., 1991; Schilling et al., 1999; Blichert‐Toft et al., 2005]. These studies suggest that theIceland plume is conveying its thermal and com-positional signature to the south more effectivelythan to the north.

[4] Both symmetric and asymmetric plume‐ridgeinteraction has been reported between the Galápagos

plume and the Galápagos Spreading Center(Figure 1c). Schilling et al. [2003] observed sym-metric Pb‐Hf‐Sr‐Nd isotope gradients, but notedthat ridge axis elevation is systematically ∼500 mgreater in the east. On the basis of three componentmixing models, Schilling et al. [2003] calculated agreater dilution of plume material reaching theeastern GSC than that reaching its western half, at anequal radial distance. Christie et al. [2005] locatedifferent points of symmetry between 90°30′W and92°10′W for each of Pb‐Hf‐Sr‐Nd isotope profilesand ridge axial depth. The grouping of these pointsof symmetry into two regions leads Christie et al.[2005] to infer the presence of two primary loca-tions of material transfer from plume to ridge.

[5] A pronounced asymmetry in plume‐ridgeinteraction has been reported around the Azores(Figure 1b). Geophysical studies of the Mid‐Atlantic Ridge (MAR) about the Azores [Goslin andParty, 1999; Maia et al., 2007] have indicated adecline in plume influence on crustal accretion overthe region 43–44°N, much closer to the plumecenter than is observed south. Similarly, fromstudying Nd isotopes which indicate a greatersouthward extent of enriched material, Yu et al.[1997] inferred a plume preferentially dischargingto the southwest.

[6] Various dynamical scenarios have been pro-posed to account for the apparent asymmetry inmany of the observations. Recourse has been madeto tilted plumes [Shen et al., 2002; Yang et al.,2006; Yu et al., 1997], lithospheric damming ofoutflow [Vogt and Johnson, 1975], bulk astheno-spheric flow [Chase, 1979; Mertz et al., 1991],ambient mantle compositional anomalies [Melloet al., 1999] and plume zonation [Murton et al.,2002]. One reason for this proliferation of dynami-cal scenarios is that there is no generally acceptedglobal model for plume‐ridge asymmetry. A sec-ond cause is that a given plume can have someobservables that are symmetrically distributed andothers that are asymmetrically distributed, a resultarising from this study. A globally consistentmodel of plume‐ridge dynamics must simulta-neously explain these two sets of observations.

[7] The purpose of this paper is to explore thenature of plume‐ridge interaction for each of theIceland, Galápagos and Azores plumes. We findthat many reported cases of asymmetry are artifactsof subjective choices of plume center. Quantifyingthe degree of symmetry of any given observable iscritical in order to identify a plume center objec-tively, and we develop a method of quantifying

GeochemistryGeophysicsGeosystems G3G3 SHORTTLE ET AL.: SYMMETRY OF PLUME‐RIDGE INTERACTION 10.1029/2009GC002986

2 of 27

symmetry in section 2. By making an objectivechoice of plume center, observations related toasthenosphere flow and temperature structures (e.g.,bathymetry and crustal thickness) turn out to beconsistent with radially symmetric plumes. How-ever, we find that the asymmetry in incompatibletrace element and isotopic compositions is pro-nounced, and much stronger than any found inbathymetry or crustal thickness. Since geophysicalobservations related to asthenosphere structureappear to bemost consistent with a symmetric plumehead, we search for a model in which the composi-tional asymmetry is imparted by the geometry of theridges and the way they allow plume mantle to be

processed through melting regions. In section 3, wedevelop a simple kinematic model of radially sym-metric plume outflow beneath a spreading ridgegeometry that can be asymmetric about the plumecenter. This model provides a remarkably good fit tothe first‐order features of compositional asymmetryfound on mid‐ocean ridges close to Iceland and theGalápagos Islands.

2. Symmetry of Along‐RidgeObservables

[8] Radially symmetric outflow represents the sim-plest kinematic model of plume dispersal in the

Figure 1. General stereographic projections of (a) Iceland, (b) the Azores, and (c) the Galápagos plume‐ridge systems.In each case the preferred plume center is marked by a white and red bull’s‐eye, and white lines emanating from thispoint represent the flow path of plume material spreading radially symmetrically. Ridge crests are marked as thick greyand white lines, overlain on a 112 km wide blue or red zone, delineating the assumed width of the top of the deep sub-ridge melt region. Themelt regions are colored according to how the profiles were subsequently split on either side of theplume. For each map the area of trial plume centers used in symmetry mapping is outlined in black. In Figure 1a, theReykjanes ridge is colored blue, and the Kolbeinsey ridge is colored red. In Figure 1b, the plume center is locatedbeneath Faial island, and the MAR split is north and south of the plume. In Figure 1c, the Galápagos plume center islocated beneath Fernandina island, and the red and blue profiles denote the east and west GSC, respectively. Bathymetryfrom GEBCO 30 arc sec grid, http://www.gebco.net.


3 of 27

asthenosphere. Symmetric outflow models shouldonly be rejected if systematic and objective obser-vational tests necessitate greater dynamical com-plexity. Failure to perform such tests can give riseto a false inference of asymmetry in along‐ridgeobservables. The application of a symmetric out-flow model also has a dynamical basis. Numericaland analogue models show that a plume head willspread symmetrically from the top of the plumeconduit if the base of the lithosphere is flat and ifrelative motion between the plume head and thelithosphere does not cause significant drag [Ito,2001; Campbell, 2007; Campbell and Griffiths,1990]. This latter requirement is met by the lowviscosity of the asthenosphere, which thermal,geodetic and rheological constraints place as being10–100 times lower than adjacent lithosphericmantle [Buck and Parmentier, 1986;Robinson et al.,1987; Rabinowicz et al., 1990; Hirth and Kohlstedt,1996]. The asthenosphere’s low viscosity makesit unlikely to be strongly coupled to movementof the lithosphere, and hence plume outflow to afirst order might be expected to be symmetric. Thereis also evidence that the upper boundary to theasthenosphere is flat, defined not by the thermallithosphere, but by the dry solidus [Hirth andKohlstedt, 1996]. The origin of this definition ofthe base of the lithosphere is the sensitivity ofperidotite viscosity to its volatile content. Thesevolatile elements are rapidly extracted during thefirst few percent of melting, producing order ofmagnitude increases in mantle viscosity [Hirth andKohlstedt, 1996]. In consequence, ridge axes areunlikely to provide channels for plume material andtransform faults are unlikely to act as barriers toplume outflow, as the dry solidus will not vary withplate age. Taken together, these lines of argument atleast justify a symmetric plume outflow model as astarting point for models of plume‐ridge interaction[Jones et al., 2002; Poore et al., 2009; Rudge et al.,2008].

[9] Despite the strong case for symmetric outflow,many studies choose plume centers which renderthe observables asymmetrically distributed. If theunderlying data really is symmetrically distributed,why have inappropriate plume centers been chosenin many studies? Partly the answer is historicalinertia. Once one study has chosen a plume centerlocation, there will be a tendency for it to be quotedand reused without determining if it is appropriateto newer data sets. In addition, many plume centersare constrained by relation to surface features. Forexample, the Iceland plume is often said to be

centered on the Grimsvötn caldera. While volca-noes act as a useful proxy for plume locations,magma transport in the crust and mantle also playsan important role in determining the locus ofmagma accumulation. This makes volcano loca-tions an imperfect indicator of the distribution ofplume material. Beyond using surface features,there are methods both geophysical, for examplemantle tomography [Hooft et al., 2003], and geo-chemical, such as locus of high 3He/4He over anocean island [Kurz and Geist, 1999], for constrain-ing a plume’s location. All approaches however, arefundamentally limited by their resolution, leavingconsiderable flexibility in assigning plume centersand thereby allowing for possible misidentificationof asymmetry. A systematic approach is required,combining all available data sets to select a plumecenter best able to create symmetry in along‐ridgeobservables and match independent plume centerestimates. Thus, any asymmetry that remains whena preferred plume center is chosen can be attributedto a genuine asymmetry in the plume‐ridge system.

2.1. Quantifying Symmetry

[10] Given the simple starting hypothesis of sym-metric radial plume outflow, a convenient methodfor determining the geometry of interaction inplume‐ridge systems is to plot along‐ridge geo-chemical and geophysical observables as a functionof radial distance from the plume center. Thus, anyasymmetry in plume interaction with ridges will bemanifest as deviations between the profiles. How-ever, making such plots requires a priori knowl-edge of the plume center’s location. Selection of aninappropriate position for the putative plume centercould result in the false appearance of asymmetry,a scenario illustrated in Figure 2. In practice thespatial distribution of plume material in the mantleis only inferred from geochemical and geophysicalobservations, so our choice of a reference frame inwhich to view the data is key. We begin by quan-tifying symmetry about trial plume centers, withthe aim of identifying points that render theobservables symmetrically distributed.

[11] The raw data used for this study are presentedin Figure 3. In order to assess symmetry, these datapoints must be interpolated onto regular distanceintervals from a trial plume center. Raw geo-chemical data are first smoothed using a travelingGaussian filter and then the points linearly inter-polated between to provide the required samplespacing. It is useful in comparing misfit betweenobservables, for the profiles’ geochemical or geo-


4 of 27

physical parameters to be normalized. This isachieved by converting to nondimensional z scores(zi = [xi − m]/s), where the estimated mean of thepopulation (m) is subtracted from an individualobservation (xi) and the result is divided by thepopulation’s estimated standard deviation (s). Misfitcalculated from these profiles is then expressed as zscore root‐mean‐square misfit (Z‐RMS); details ofthis calculation are in Appendix A. Topographicprofiles are produced from a low‐pass filtered(wavelength 85 km) version of the GEBCO_08global elevation and bathymetry gridded data set(version 20090202, http://www.gebco.net). Furtherdetails regarding the processing of the profiles arediscussed in detail in Appendix B.

[12] Z‐RMS misfits were calculated using datawithin a spatial window chosen to exclude datawith no clear plume signal. The correspondingridge lengths are shown in Figures 1 and 3. In thecase of Iceland the result is a truncation of datanorth and south at a radial distance of ∼500 kmfrom Iceland’s center, where a Jan Mayen signaturebecomes dominant in mid‐ocean ridge basalt(MORB) chemistry. The large increase in thechemical variability of on‐land data from Icelandmakes spatial averaging difficult, so we limit geo-chemical profiles to the submarine ridge sections.Data along around the Galápagos plume is includedas far east and west along the GSC as there is datacoverage. For the Azores hot spot, an anomalous

mantle regime to the north of the KurchatovFracture Zone, previously associated with densegarnet rich mantle [Mello et al., 1999], limits thelength of profile being matched.

[13] We use a grid search method to find the centersof symmetry for each observable in each plume‐ridge system. The algorithm is as follows.

1. Select a point within one of the boxedregions in Figure 1 as a trial plume center.

2. Replot the data set of interest as a func-tion of distance from the trial center.

3. Calculate misfits between data at commondistances either side of the trial center at 1 kmincrements.

4. Calculate Z‐RMS misfits for the wholeprofile.

5. Complete the misfit grid by repeating steps1–4 for all trial plume centers within the boxedregion, at intervals of 0.01 degree. Hence identifythe loci of centers of symmetry for this observable.

6. Repeat steps 1–5 for the remainingobservables.

7. Compare the centers of symmetry for all theobservables with plume centers inferred from inde-pendent data sets not restricted to the mid‐oceanridge.

Figure 2. These diagrams demonstrate the importance of careful plume center selection. Points A and B mark twoputative plume centers, overlain on top of a synthetic Gaussian plume swell (for zero age crust), the center of whichunderlies point B. Dotted lines delineate the profile of an adjacent spreading center, which has its axial depth affectedby the plume as a function of radial distance from the plume center. Non‐plume‐influenced ridge has a depth, z, of 0.In the two plots, the along‐ridge profiles for depth are recorded as a function of radial distance from each of theputative plume centers. Red and blue lines denote the two sections of ridge depth profile, split at the ridge’s closestapproach to the points A or B. When an inappropriate plume center is selected, as is the case for point A, the depthprofiles appear asymmetric, with the right‐hand length of ridge (blue) being systematically shallower than the left(red). This is rectified by plotting the ridge profiles about point B, which corresponds to the true plume center.


5 of 27

Figure 3. Maps and plots of the unprocessed geophysical and geochemical data. Ridge names are marked on forreference. Grey bars mark the latitude or longitude of ocean islands. (a) Map of the distribution of samples northand south of Iceland. Associated graphs show three of the six geochemical and geophysical observables consideredfor Iceland in this study, plotted against latitude. Data from Schilling et al. [1983], Mertz et al. [1991], Mertz andHaase [1997], Murton et al. [2002], Jakobsson et al. [2008], Thirlwall et al. [2004], Schilling et al. [1999], Deveyet al. [1994], Blichert‐Toft et al. [2005], and Peate et al. [2009]. (b) Map of samples along the Eastern and WesternGSC. The raw Galápagos data comprising the profiles used in this study, elevation, �Nd, and 87Sr/86Sr are plottedalongside. Data from Schilling et al. [2003] and Ingle et al. [2010]. (c) MAR around the Azores, sample map and rawelevation and isotopic profiles. Data from Yu et al. [1997], Debaille et al. [2000], and Dosso et al. [1999].


6 of 27

2.2. Searching for Centers of Symmetry

[14] A center of symmetry is characterized byhaving a low Z‐RMS misfit. Perfect symmetry ofan observable about a plume center would returna Z‐RMS misfit score of zero, like the syntheticprofile B in Figure 2. However, all observablesrecord analytical uncertainty as well as natural var-iability unrelated to plume dynamics. This uncer-tainty means real data sets will never exhibit perfectsymmetry. We therefore adopt an approach of con-sidering relative symmetry between observables. Thepresence and position of the centers of symmetryfound can then be combined with volcanological,seismic tomographic and geochemical constraintson plume center location to decide if a particularobservable is symmetric.

[15] Results of searching for centers of symmetryof the Iceland plume are presented in Figure 4 interms of Z‐RMS misfit. Common to most of themaps is a region of high symmetry (low Z‐RMSmisfit) oriented southeast‐northwest. This featureis a function of the lack of constraint that a singletwo dimensional geochemical or geophysical pro-file can place on the center of a three dimensionaldistribution of plume material in the mantle. Sig-nificant though is the location of the region ofhigh‐symmetry trial plume centers: For crustalthickness and elevation the high‐symmetry regionextends into central Iceland, close to where theplume center is presumed to lie from mantletomographic and previous crustal thickness studies[Shen et al., 2002; Darbyshire et al., 2000]. For Na8,symmetry is generally poor across the searched area,and the lowest values of Z‐RMS misfit are displaced∼50 km to the southwest of the elevation and crustalthickness high‐symmetry regions. For isotopic pro-files the most favorable symmetry center locationslie further southwest still, at ∼100 km from the ele-vation high‐symmetry region. Zr/Y in contrastshows no region of low misfit (Figure 4). From thealong‐ridge profile of Zr/Y in Figure 4, it is clearthat this absence of a high‐symmetry zones is aresult of the fundamentally asymmetric nature ofthe profiles attempting to be matched: significantdepletion of Zr/Y along the Kolbeinsey ridgemakes its profile unmatchable with the enrichmenttrend seen along the Reykjanes ridge. However, forthe other geochemical observables, Figure 4 de-monstrates that the profiles are sufficiently similar inshape that a center of symmetry can be found,although plume centers placed too far to the south-west begin to appear unrealistic given what is knownof Iceland’s geodynamics.

[16] The process of identifying centers of sym-metry has been repeated for plume centers in theGalápagos and Azores regions and the results arepresented in Figures 5 and 6, respectively. As withthe maps for Iceland, a region of trial plume centerswith low Z‐RMS misfit extends away from theridges roughly perpendicular to their strike. For theGalápagos (Figure 5) the high‐symmetry regionfrom elevation extends directly over the presumedplume location beneath Fernandina Island [Schillinget al., 2003]. High‐symmetry fits to the isotopeprofiles however, requires trial plume centers dis-placed to the west from this point, away from theactive intraplate volcanism. Thus, while the influ-ence of the plume swell on ridge depth appearsaxisymmetric about a physically reasonable hot spotcenter, the isotopic composition of erupted basaltsrecords an asymmetry in plume‐ridge interaction.

[17] A similar scenario exists with the Azoresplume, although in this instance the data coveragealong the southern ridge profile is poor, due to thecombined effects of an along‐ridge sampling gapand a fracture zone offsetting the spreading centeraway from the Azores. The result is to leave thesouthern profile unconstrained over a distance rangefor which the northern profile has a more contin-uous distribution of samples. Given also that theAzores profiles are limited in distance to the radialextent of the plume’s influence on geochemistry,the distance over which profiles are well constrainedis ultimately very short, ∼30 km. With these caveatsin mind, the displacement of the isotopic high‐symmetry regions to south of the Azores should betreated with caution. The elevation profile however,which is the most continuous, does show a regionof trial plume centers with high symmetry elongatein the direction of Faial Island. Therefore, despitepoor constraints on the geochemical symmetry, thebathymetric expression of the plume’s presence isconsistent with it being centered under the Azoresplatform and interacting in a radially symmetricfashion with adjacent ridges.

2.3. Choosing Plume Centers

[18] The maps of trial plume center symmetry inFigures 4–6 are next used to choose a preferredplume center location. Greatest weight is placed onobservables that most directly reflect the physicalplume such as spreading ridge crustal thickness andaxial elevation, which both depend directly onasthenosphere temperature. Given these observ-ables also showed centers of symmetry systemati-cally offset from the geochemical observables, the


7 of 27

geophysical and geochemical sets of observationsare grouped separately in considering their centersof symmetry. These data sets were then comparedwith the results from global seismic tomographyand 3He studies, in order to improve constraints onthe plume location perpendicular to the spreading

axis. These combined observations can be used totest simple radially symmetric models of plumeoutflow.

[19] Plume centers generating high symmetry inobservables are identified based upon the combinedand averaged symmetry maps of the geochemical

Figure 4. Mapped Z‐RMS misfit (symmetry) of trial plume centers for chemical and geophysical profiles along theReykjanes and Kolbeinsey ridges, with corresponding plots of the maximum symmetry solution. Underlain beneaththe contours is the outline of southeast Iceland, and the on‐land extension of the MAR is drawn as a thick black andwhite line. The plume center location which minimizes misfit between along‐ridge profiles is marked by a yellowtriangle, and our preferred plume center is marked as a white and red bull’s‐eye. Graphs display the profiles whichwere compared north and south of Iceland, plotted about the maximum symmetry trial plume center. The nondimen-sional standardized z score is given along the y axis in addition to conventional units; note the differing x axis scales.Details of the misfit calculation procedure can be found in section 2.1.


8 of 27

and geophysical observables. The result is the grey(geophysical misfit) and pale yellow (geochemicalmisfit) regions marked on in Figure 7, which cor-respond to the low Z‐RMS misfit regions of theseaveraged maps. Figure 7 also shows the independentconstraints on plume center locations. For eachplume‐ridge system studied, the geophysical pro-file high‐symmetry regions correspond closely tothese independently constrained plume locations.The geochemical centers of symmetry however,are consistently offset from global seismic and 3Heconstraints.

[20] Plume center positions estimated from crustalthickness maps and mantle tomographic images are

displayed for Iceland in Figure 7a. Our preferredplume center, that obtains symmetry in both thecrustal thickness and elevation profiles, lies at17.4°W 63.95°N. This is ∼60 km south of thecluster of plume centers estimated from crust andshallow mantle seismic constraints, but within errorof the estimates from deeper mantle structure. Astudy of the arrival times of P to S transition zoneconversions beneath the Galápagos hot spot, recov-ered a plume center location of 91.7 ± 0.8°W 0.7 ±0.8°S [Hooft et al., 2003]; from our symmetry map-ping we place the plume center ∼40 km northeastat 91.45°W 0.40°S beneath Fernandina island, butwithin the locus of the Hooft et al. [2003] estimate.

Figure 4. (continued)


9 of 27

For the Azores, Yang et al. [2006] have produced aP wave velocity model that centers the hot spot at38.5°N 28.5°W. We use this constraint to place ourpreferred plume center at the eastern extent of thehigh‐symmetry region drawn in Figure 7b, beneathFaial island. For each plume the preferred hot spotlocations are marked in Figure 7 as red and whitebull’s‐eyes.

2.4. Symmetric or Asymmetric PlumeOutflow?

[21] Preferred plume center locations have beenselected, incorporating both the results from sym-metry mapping and the independent plume centerestimates. It is now possible to replot the data aboutthese plume centers and assess the symmetrypresent in each observable. Interpretation of theseplots places important constraints on the dynamicsof plume outflow.

[22] Table 1 provides a summary of the RMS andZ‐RMS misfits for profiles taken about each plumecenter, including both minimum misfit plume cen-ters and those we selected as our preferred plumecenters. The latter have had standard deviationscalculated for the Z‐RMS misfit of elevation. Thiscalculation was performed to test the sensitivity ofthe Z‐RMS misfit to the low resolution and dis-continuous nature of sample collection representedin the geochemical profiles. The bathymetry datawas repeatedly randomly resampled at a resolutionof 10–20 samples per profile, the Z‐RMS misfitrecalculated and the standard deviation then deter-mined from the population of Z‐RMS misfit valuesthus obtained.

[23] Figure 8 demonstrates that for Iceland, althoughthe preferred plume location is able to generatesymmetric elevation and crustal thickness profiles,the isotopic and incompatible trace element profiles

Figure 5. Trial plume center symmetry maps and along‐ridge geochemical profiles around the Galápagos Islands.The Galápagos Spreading Center is drawn as the thick black and white line. Other details as in Figure 4.


10 of 27

created are asymmetric. Table 1 records the Z‐RMSmisfit for elevation and crustal thickness to be lessthan 0.6, while the Z‐RMS misfit for the chemicalobservables is at least twice this and many timesoutside the standard deviation of the elevationZ‐RMS estimate. When plotted about our favoredplume center the northern profile is systematicallymore depleted than the southern (Figure 8). Themajor element trends show more scatter, likely aresult of their sensitivity to crustal processes, whichare poorly deconvolved from mantle derived sig-nals by the simplistic linear regression correctionapplied. Despite this, Na8 records depletion of theKolbeinsey ridge with respect to the Reykjanesridge, albeit over a shorter distance than the traceelements and isotopes, 300–410 km comparedwith 300–550 km for 87Sr/86Sr.

[24] For the Galápagos plume, Sr and Nd isotopicanalyses have been considered alongside ridgeaxial elevation (Figure 9). The Z‐RMS misfit forelevation (Table 1) is overall low at 0.41, butresults partly from a slightly shallower eastern GSC

compared to the western GSC, as noted by Schillinget al. [2003]. In particular, for 150 km along itslength the eastern GSC is an average 140 m shal-lower than the western GSC at equivalent dis-tances; however this feature does not persist beyond450 km from the plume center. The isotopic data,when plotted as a function of radial distance fromthe same point as the bathymetry, shows a markedand consistent offset to more depleted values alongthe eastern GSC. This asymmetry in �Nd averagesa Z‐RMS misfit of 0.94 and extends from approxi-mately 200–800 km from the hot spot center, theactual offset varying along its length. Comparingthe Z‐RMS misfits between data sets indicatesthat the geochemical observables are much moreasymmetric than the geophysical observables, withZ‐RMSmisfits more than twice that seen in the ridgeelevation.

[25] As already discussed, the less continuous natureof the Azores geochemical data leaves symmetryharder to constrain. The elevation data shows amoderate Z‐RMS misfit of 0.64 (Table 1) but with

Figure 6. Trial plume center symmetry maps and along‐ridge profiles around the Azores islands. The Mid‐AtlanticRidge is drawn as the thick black and white line. Other details as in Figure 4.


11 of 27

only ∼75 km of profile fitted; this is not repre-sentative of the longer‐wavelength swell. Visualinspection of Figure 10 suggests a rough corre-spondence between bathymetry further from theAzores, although the segmented nature of the MARin this region again makes trends in offset difficultto follow. There is a single cluster of isotopic datain Figure 10 between 100–150 km south alongthe MAR, which falls within the distance range ofthe misfit calculations. The Z‐RMS misfit from thiscluster with respect to the data north along the ridgeis high at 1.75, which exposes a shortcoming of themethod; a single smoothed line has been fitted tothe data and the dispersion of sample compositionsabout the mean not fed into the misfit calculations.While this is suitable for the Galápagos and off-shore Iceland, where generally the variability of data

at a given distance is low and the offset betweenprofiles high, in the Azores there is a large variabilityin the data. The result is an apparent discrepancybetween chemistry north and south of the Azores,but which is lessened when the overlap of data isconsidered. Therefore, it remains uncertain whetherthe north and south profiles are different over theregions for which there is evidence that the plume isthe dominant geodynamic feature.

[26] The ability of the preferred plume centers tomatch elevation and crustal thickness along‐ridgewith a low Z‐RMS misfit, while simultaneouslyfalling close to seismological constraints on plumelocation, indicates that dispersal of Iceland andGalápagos plume material is symmetric. That thesesame plume centers are not centers of symmetry forthe incompatible element and isotope ratio profiles

Figure 7. Maps marking independent estimates of plume center locations, compared with our preferred plume centerpositions for (a) Iceland, (b) the Azores, and (c) the Galápagos. In each map, bathymetry is contoured at 1 km inter-vals, and the areas of trial plume center grid searches are included as black rectangles. The averaged high‐symmetryregions from the symmetry mapping calculations are marked in grey (geophysical observables) and pale yellow (geo-chemical observables). When the results of tomographic studies are used, the approximate center of any low‐velocityanomaly imaged is taken as representing the axis of the plume. For constraints from transition zone thickness, thecenter of the region of thinned transition zone is taken as representing the axis of the plume at the base of the uppermantle. In 3He studies, the locus of lavas displaying the maximum 3He/4He is taken to be the center of the plume.Shen et al. [2002] used crustal thickness as a proxy for the location of the plume center in the shallow mantle. Assuch, points representing estimates of maximum crustal thickness have been included. In Figure 7b, the ridge per-pendicular extension of the region of maximum along‐ridge crustal thickness, as determined by Detrick et al. [1995],is drawn in light blue. In Figure 7c, the dashed green line marks the region of thinned transition zone imaged by Hooftet al. [2003].


12 of 27

suggests a decoupling between geochemical andgeophysical tracers of plume dispersal.

3. Kinematic Modeling

[27] The result of section 2 indicates that asymmetryin geochemical observables must be superimposedupon an essentially symmetrically outflowing plumehead. Therefore, the lithospheric damming of out-flow [Vogt and Johnson, 1975], bulk asthenosphericflow [Chase, 1979; Mertz et al., 1991] and tiltedplume [Shen et al., 2002; Yang et al., 2006; Yu et al.,1997] models cannot account for the asymmetry inthe geochemical profiles. All of these processeswould affect the advection of the plume’s thermalsignal, and be recognizable as asymmetry in ridgeelevation and crustal thickness, an asymmetry we donot observe in either of the Iceland or Galápagosplume‐ridge systems. Another possibility is that theplumes are compositionally zoned [e.g.,Fitton et al.,1997; Murton et al., 2002; Geist et al., 1988;Hoernle et al., 2000], and this zonation in the plumeconduit is being advected to the ridges [Thirlwallet al., 2004]. Alternatively, an asymmetry intrinsicto plume‐ridge systems, which could influence thegeochemistry of erupted basalts, is the distributionof mid‐ocean ridges around the plume center. In thissection we demonstrate with a simple kinematicmodel how ridge locations can give rise to asym-metric geochemical profiles, from a symmetricallyoutflowing plume. The key process is the partialmelting of plume material as it flows under spread-ing ridges. First, the dynamics of the scenario are

described, followed by a description of the kine-matic model.

3.1. Dynamical Basis of Model

[28] Depending on the geometry of the plume‐ridgesystem, there is the potential for different plume outflow paths to have traveled through variablelengths of melt region at a given distance from theplume center. This feature is evident from Figure 1,in which radially symmetric flow paths from eachplume experience quite different lengths of sub-ridge flow. This outflowing plume material willinteract with the spreading centers it passes under.Corner flow beneath ridges creates regions of pas-sive upwelling within the shallow asthenosphere,as mass moves in response to plate divergence.Hot plume material, flowing out beneath the high‐viscosity lid of the anhydrous melt region, willdecompress as material in the layer above upwells.Thus, different outflow paths cause plume materialto have decompressed by varying amounts at agiven radial distance from plume center.

[29] If decompression causes the solidus of plumelithologies to be intersected, then the ensuing frac-tional melting will alter the composition of theremaining plume material. Coupled with the var-iable lengths of ridge intersection experienced bydifferent outflow paths, this is a means by whichthe plume head can develop an asymmetric com-positional structure. This occurs without the needfor long‐wavelength compositional zonation in theplume conduit on the 100 km scale. As discussed

Table 1. RMS and z Score RMS Misfits of the Maximum Symmetry Solutions From Misfit Mapping and From the ProfilesGenerated by Plotting About Our Preferred Plume Centersa

Observable

Maximum Symmetry Preferred Plume Center

RMS Z‐RMS RMS Z‐RMS

IcelandElevation 0.15 km 0.31 0.20 km 0.38 (±0.02)Crustal thickness 1.79 km 0.32 2.4 km 0.5287Sr/86Sr 4.8 (×10−5) 0.53 15 (×10−5) 1.30�Nd 0.38 0.60 1.21 1.35Zr/Y 0.32 1.50 0.66 1.80Na8 0.21 0.86 0.27 1.16

GalápagosElevation 0.11 km 0.38 0.12 km 0.41 (±0.08)87Sr/86Sr 7.4 (×10−5) 0.41 18 (×10−5) 0.97�Nd 0.66 0.61 0.91 0.94

AzoresElevation 0.06 km 0.29 0.09 km 0.64 (±0.12)87Sr/86Sr 4.9 (×10−5) 0.41 26 (×10−5) 1.63�Nd 0.50 0.38 2.4 1.68

aZ‐RMS, z score root‐mean‐square. The error on the preferred plume center elevation Z‐RMS misfits is one standard deviation.


13 of 27

previously, the rheological step that defines thetop plume channel is likely the anhydrous peri-dotite solidus [Hirth and Kohlstedt, 1996].Therefore, plume material will only melt belowthis depth if it is more fusible than typical depletedperidotite [Hirschmann et al., 1999]. However, thereis compelling evidence to suggest that the mantleis compositionally heterogeneous on short lengthscales, consisting of enriched components within adepleted matrix [Zindler et al., 1979; Allégre andTurcotte, 1986; Stracke et al., 2003; Sobolev et al.,2005; Kokfelt et al., 2006; Maclennan, 2008].These enriched portions of the mantle are likelyto be lithologically distinct eclogitic or pyroxeniticblebs set within a more depleted peridotite matrix.The results of experimental petrology indicate thatthese assemblages have a lower solidus tempera-

ture than peridotite [e.g., Yaxley and Green, 2000;Kogiso et al., 1998; Hirschmann et al., 2003;Dasgupta et al., 2006; Yaxley and Sobolev, 2007],and as such begin melting at depths below the topof the plume channel. We thus envisage decompres-sion of the plume material beneath spreading centersto generate small degree melts from the enrichedblebs, while the depleted peridotite matrix remainsunmelted. As the first few percent of melting willpreferentially strip from the residue all the mostincompatible elements, the bulk incompatible ele-ment concentration and isotopic composition of out-flowing plume material will become progressivelyweighted to the depleted matrix. In this way, along‐ridge gradients in geochemical proxies for plumedispersal can be generated, and asymmetry devel-oped as a function of flow path.

Figure 8. Geochemical and geophysical data plotted as a function of radial distance from the preferred plume center forIceland. Fracture zones are drawn as grey bars, in rough proportion to the distance range they cover and labeled the samecolor as the ridge segment they offset. SFZ, Spar Fracture Zone. The chemical data were Gaussian filtered at 85 kmwavelength to produce running means and represent the smoothed profiles used in calculating misfit. Representativeanalytical error bars for the geochemical data are included for reference. While crustal thickness and elevation arebroadly symmetric off‐land, the trace element ratios and isotopes delineate a more depleted Kolbeinsey ridge. Asym-metry in Na8 is less clear, with asymmetry present for just the first 75 km following the Tjornes Fracture Zone (TFZ).


14 of 27

Figure 9. Profiles plotted about the preferred plume center for the Galápagos. Details as in Figure 8. Ridge segmentsare split here as in Figure 1c. Isotopically, the eastern GSC shows a more depleted signature than the western sectionsof the ridge, with systematically higher �Nd and lower 87Sr/86Sr for most of its length. In elevation, ridge profiles arebroadly symmetric east and west of the plume center, although slightly shallower depths (140 m) are found on theeastern segment west of the 95.5W transform. IFZ, Inca Fracture Zone; EFZ, Ecuador Fracture Zone; PFZ, PanamaFracture Zone.

Figure 10. Profiles plotted about the preferred plume center for the Azores. Details as in Figure 8. Ridge segmentsare split here as in Figure 1b. The filtered mean lines (thin blue and red lines) for the Azores are drawn only as far asdata was fitted north and south of the plume center, the limit being the transition in chemistry north of the KirchovFracture Zone (KFZ) to increasingly enriched signatures, presumably associated with the transition to a separate (non‐Azores influenced) mantle regime. A sampling gap between 38.0 and 38.4°N makes it difficult to constrain the near‐plume symmetry, but for the data that are present the chemistry appears symmetric within the natural variability andanalytical uncertainty. PFZ, Pico Fracture Zone.


15 of 27

[30] The model is generated in two steps. First, asimple distance calculator is run to determine thedistance plume material flows through deep sub-ridge melt zones: these are the parts of the meltregion where solid material has a significant hori-zontal component to its velocity and melt genera-tion is predominantly from enriched fusible blebs.Above this depth, in the shallow melt region, thevelocity field is dominated by corner flow and meltproduction is mostly from the depleted plumematrix. The distances calculated are then combinedwith simple models of melt extraction in order totrack the depletion of a 1D column of plumematerial as it passes beneath spreading centers.

3.2. Melt Region Traversal Distance

[31] The initial stage of the modeling is to deter-mine the distance traveled through the deep meltregions by material reaching a ridge. A kinematicmodel is used in which a plume source flowsradially outward from plume centers, intersectingridge segments and their melt regions as it does so.The width of the deep melt region is a poorlyconstrained parameter in these calculations. Con-trolling factors are likely to be the ridge angle, itselfat least partly a function of mantle viscosity,spreading rate, diffusivity and melt‐solid densitycontrast [Spiegelman and McKenzie, 1987], and thedepth at which plume material is spreading. Beneatha spreading center the latter parameter is assumedto be ∼60–160 km, governed by the onset ofdehydration melting, which provides a viscositybarrier to any shallower buoyant upwelling [Hirthand Kohlstedt, 1996; Ito et al., 1999; Hall andKincaid, 2003]. The ridge angle however isunknown, thus given these uncertainties, a simpli-fying assumption is made of a constant 112 kmacross‐axis width for all model runs (see Figure 1).For the purposes of trying to understand howasymmetry is generated, the absolute width is lesssignificant than relative differences between meltregion traversal distances. When calculating thedistance through a deep melt region, the plume flowis reduced to a single dimension. Therefore, despitethe plume material having some vertical thicknessand depth range, which might be expected to feedinto the width of the melt region for the enrichedblebs, traversal distance is calculated only at asingle depth of flow.

[32] Maps illustrating the plume‐ridge systems foreach of the Iceland, Galápagos and Azores hotspots are presented in Figure 1. From these diagramsit is possible to qualitatively assess the asymmetry

present in ridge geometry in each system, andtherefore the effect this may have on trace elementand isotopic symmetry. For Iceland, with the plumeplaced at the south eastern edge of the island, flowlines reaching the Kolbeinsey ridge must first tra-verse the base of the Northern Volcanic Zone,while plume material is fed to the Reykjanes ridgeobliquely and without prior passage under a majorspreading center (Figure 1a). Along the GSC the91°W transform, ∼2° north of the plume center,steps the eastern GSC south toward the GalápagosIslands. Simply from visual inspection of Figure 1c,it is clear the result of this transform is to causematerial to enter the base of the eastern GSC meltregion more obliquely than it does the western GSC.Therefore, the deep melt region path length ofmaterial destined for the eastern GSC is increased. Incontrast to the previous two settings, where a pro-nounced asymmetry is visible, the Azores systemappears essentially symmetric (Figure 1b).

[33] The calculations of deep melt region distancetraversed for flow lines reaching ridges around theIceland, Galápagos and Azores plumes, are pre-sented in Figure 11. For Iceland (Figure 11a) thenorth‐south asymmetry identified from the map(Figure 1a), is manifest as a maximum difference inmelt region distance traveled of ∼390 km. Plumematerial reaching the Kolbeinsey ridge has system-atically traveled a greater distance beneath meltingregions than that reaching the Reykjanes ridge at anequivalent distance. This pattern is repeated inFigure 11e around the Galápagos plume, in whichthe eastern GSC is receiving plume material that hastraveled obliquely along its length. Consistent withour previous inference of symmetry around theAzores, this system shows matching distances ofmelt region traversal north and south of the plumecenter. We have excluded the ultraslow spreadingTerceira rift [Vogt and Jung, 2004; Beier et al.,2008] from consideration in these calculations,because its spreading rate is less than a tenth that ofthe adjacent MAR and thus its affect on astheno-spheric velocity gradients is minimal.

[34] The significant result of these simple distancecalculations is that material traveling along axi-symmetric flow paths from a plume can haveexperienced quite different degrees of interactionwith spreading centers at the same radial distance.

3.3. Plume Material Preconditioningby Subridge Flow

[35] Having established the distance that plumematerial is traveling through the subridge mantle,


16 of 27

prior to reaching spreading centers, it is possible toextend the calculation to tracking the material’scomposition. Pearce [2005] defines and modelsmantle preconditioning for a variety of geodynamicscenarios, but primarily examines the geochemicalconsequences in isotope‐trace element ratio space.Here, the modeling emphasis is placed on the needto relate spatial variables (upwelling rate, ridgelocation and plume position) to the source traceelement chemistry. A simpler petrological modelthan that of Pearce [2005] is thus considered, inwhich the low degrees of melting modeled (5%maximum) allows for the melting reaction to beassumed constant. The model set up is illustratedin Figure 12.

[36] Plume material is modeled to contain a fusibleenriched component present as a short‐wavelengthheterogeneity. These enriched blebs are progres-

sively depleted by melting, up to a point wheremost of their incompatible element load will havebeen extracted. A fictive element, Y, with the prop-erties of a light REE, is used to track the effect ofpartial melting on a 100 km thick column of thisplume material. A fictive element is used becauseof the model’s simplicity, which makes predictionsof actual source chemistry inappropriate. Our fictivetracer is given partition coefficients of 0.0005 inolivine and 0.033 in clinopyroxene and garnet.Mineral‐melt partition coefficients are kept constantduring melting. Initial mineral modal abundancesare taken to be that of an enriched assemblage,45:34.5:20.5 clinopyroxene:olivine:garnet, meltingin the ratio 55:20:25. Partial melting of these blebsis calculated using Shaw [1970, equation 1]:

Cf ¼ C0

1� Fð Þ 1� PF

D

� �1=P

; ð1Þ

Figure 11. The cumulative deep melt region distance traveled by plume material reaching a given point on the ridge,plotted as a function of radial distance from plume center, for (a) Iceland, (c) the Galápagos, and (e) the Azores. Thedark grey triangular area in each plot indicates the >1:1 region, and vertical light grey bars denote fracture zones. ForIceland and the Galápagos, material flowing north and east, respectively (red lines), traverses a greater melt regiondistance than that traveling south or west (blue lines); however, for the Azores no significant differences in meltregion traversal distance occur north or south of the plume center. The Nd isotopic profiles around (b) Iceland,(d) Galápagos, and (f) Azores for comparison with the model result.


17 of 27

where Cf is the source composition followingmelting by some fraction F, C0 is the initial sourcecomposition and P and D are the bulk reactioncoefficient and partition coefficient, respectively.

[37] As the plume layer passes beneath spreadingcenters it is allowed to upwell. Using a simplifiedapproximation of ridge driven corner flow, upwell-ing in the center of the deep melt region is takento be 2

3Vx, where Vx is the half spreading rate,decreasing linearly to zero at 56 km from thespreading axis. The Euler poles for calculation ofplate spreading rates are taken from DeMets et al.[1994]. In order to determine how far materialupwells, a horizontal velocity also needs to beassigned to the outflowing material. Here we simplytake the velocity for the whole radially outflowingplume layer (u) to be that of the mean across streamvelocity for Poiseuille flow, using the equation fromRudge et al. [2008, Appendix A], u = q/2pr,where r is radial distance from plume center and qthe mean area flux (q = Q/2h, Q = volume fluxand h = channel width, 100 km). Velocity is thusexpected to decrease with distance from plumecenter, causing melt region traversals to take longerand greater depletion and thinning of the plumelayer to occur. Given the model’s aims are limited

to describing relative differences in melting his-tory of axisymmetric plume outflow, the effect ofdifferences between the plumes is a secondary con-cern and thus q is taken to be fixed for all plumesat 1.3 × 106 km2/Ma.

[38] Combining the melting and outflow kinematicsenables the composition of enriched material withinthe plume layer to be tracked. Each increment ofupwelling is translated into a fraction of melt gen-eration (F), by taking traversal of the 100 km layerheight to represent a total 5% melting and linearlyrelating smaller steps to this. After every steppedmelting phase, the modal mineralogy and bulkpartition coefficients are recalculated prior to melt-ing progressing further. By this method an initial100 km thick plume layer is generated at the top ofthe plume conduit, with a depletion gradient fromthe least melted at the bottom (as it has traveled onlya short vertical distance across the plume channel)to the most depleted at the top. This meltingabove the stalk of the plume is analogous to intra-plate melting, where ocean island basalts are gen-erated from the deep melting of plume material. Theinitial melting event is followed by further partialmelting as the outflowing layer passes beneathridges, with the top becoming entrained into the

Figure 12. Illustration of the model setup for melting calculations. Plume material flows laterally outward in thechannel between the base of the anhydrous melt region and the base of the hydrous melt region. Initial melting ofthe plume material occurs as decompression melting in the top of the plume conduit, prior to lateral outflow (indicatedby 1). On flowing under ridges, material from the top of the channel is entrained into the shallow melt region, at a rateproportional to the vertical upwelling velocity in that area (vz) (indicated by 2). Material must then decompress overthe width of the plume channel to replace that lost from the top; doing so causes partial melting of the enriched het-erogeneities within the plume material. This melting progressively depletes that material remaining beneath the anhy-drous solidus in incompatible elements, rendering outflowing plume material more depleted the greater the distance ithas traveled through deep subridge melt regions.


18 of 27

melting regions according to the amount of pas-sive upwelling in the layer above. This subridgeupwelling effectively drives the depletion of theenriched plume column. A material at the base isupwelled to replace that lost above, in doing so itmelts and loses some of its trace element inven-tory. On reaching a particular segment of ridge,the plume channel will thus have been thinnedand the bulk chemistry driven to more depletedcompositions. The material flowing in below theplume layer to replace that lost at its top is depletedambient upper mantle. This last point reflects thenature of the model we propose, in which the plumematerial represents a long‐wavelength composi-tional anomaly in the mantle, consisting of smallheterogeneities.

[39] The model described above has been run foreach of the Iceland, Galápagos and Azores plume‐ridge systems, taking into account their varyingalong‐ridge spreading rates and ridge geometries.

The effect of the incremental fractional meltingfrom subridge flow is recorded by the concentra-tion of the fictive element Y, which is presented asan average over the column (Figure 13). We takethe depletion of Y as a proxy for both the incom-patible trace element and isotopic depletion of meltextracted. These can be expected to be related in asystem when melts from different source compo-sitions are mixing to give the average isotopic andtrace element composition of the eruptive products.In this case the isotopic composition of the eruptiveproduct is dependent upon the isotopic compositionof the two melts being mixed, and importantly, onthe concentration of the element whose isotopes areunder consideration. Thus, as the concentration ofY in the enriched blebs decreases from deep low‐degree melting, it produces melts that also containlower concentrations of Y. The average composi-tion of a mixed enriched bleb melt and matrix melttherefore become weighted more toward the depletedisotopic composition of the matrix.

Figure 13. (a, c, and e) Results of the kinematic modeling of source depletion during radial plume outflow. A single100 km thick column of mantle is tracked from under the plume center, where an initial melting event occurs, topoints along the ridge. Source depletion is tracked by the fictive element Y, which has been assigned properties sim-ilar to that of a LREE. Further details of the modeling can be found in section 3.1. For both Iceland and the Galápagos(Figures 13a and 13c), the plume source becomes variably depleted with distance. The Azores, however (Figure 13e),is essentially symmetric in source depletion about the plume center. Gaussian smoothed along‐ridge Sr isotopicprofiles for comparison with model results from (b) Iceland, (d) Galápagos, and (f) Azores.


19 of 27

[40] Both Iceland and the Galápagos (Figures 13aand 13c) gradually develop a relative depletion ofthe material outflowing in one direction comparedwith the other, indicated by low concentrations ofY. This difference in depletion is fundamentally theresult of plume material having upwelled furtherand having undergone more melting, before reach-ing a given point along the Kolbeinsey ridge oreastern GSC, than the material flowing south to theReykjanes ridge or to the western GSC. Moreupwelling is experienced by material spendinglonger flowing under ridges, but also from verticalvelocities under ridges being higher, which in thismodel is the result of an increased half spreadingrate. For the Galápagos, spreading rate and meltregion traversal distance (Figure 11c) are actingsympathetically to deplete plume material out-flowing beneath the eastern GSC, the along‐ridgegradient in plate spreading velocity being an increasetoward the east, so that by 900 km from the plumecenter the eastern GSC is spreading >20 mm yr−1

faster than the western GSC [DeMets et al., 1994].However around Iceland, the along‐ridge gradientof plate spreading velocities is a slight northwarddecrease [DeMets et al., 1994]. Despite this drop inspreading rate, the much greater melt region dis-tance traveled by north flowing material countersthe upwelling velocity effect and the north‐southdepletion offset develops. The Azores (Figure 13e),as expected from the distance calculation(Figure 11e), shows little difference in depletionabout the plume center for the first ∼350 km.Beyond this distance, the northern MAR profiledevelops depletion with respect to the southernsegment, however this is after the point at whichthe geochemical profiles can be compared for theAzores.

4. Discussion

[41] Plots of along‐ridge observables as a functionof radial distance from preferred plume centersindicate that for trace elements, isotopic ratios and toa lesser extent major elements, significant asym-metry in along‐ridge profiles is present. In contrast,the signal of plume swell at the ridge axis wasshown to be essentially radially uniform at a givendistance and misfit between the profiles, whenpresent, was generally not systematic along ridges.There is thus the appearance of a decouplingbetween the geophysical and geochemical compo-nents of plume‐ridge interaction. In an attempt tounderstand these observations a simple kinematicmodel was developed, which used the observation

that melt region distance traversed by plume outflowis likely to be asymmetric about plume centers, as afunction of the plume‐ridge geometry. The con-centration of a fictive element Y, was then trackedin the enriched portion of an outflowing mantlecolumn, to explain in a very simple sense how thevariable depletion of plume material at a givendistance could be generated.

[42] The model included a number of simplifyingassumptions in order to track the depletion ofmaterial spreading out from the plume. These canbe broadly grouped into assumptions regardingmantle velocity fields, about the dynamics of melt-ing and of the composition, and compositionalstructure, of mantle plumes. It is useful to studythese assumptions to determine which of thepossible scenarios for plume‐ridge interaction areconsistent with the processes envisaged in ourmodel and with the observed asymmetries aroundplume centers.

4.1. Mantle Velocity Fields

[43] The velocity structure of the convecting inte-rior of the earth is extremely difficult to measurethrough direct or indirect observation; this thereforeplaces an emphasis on numerical modeling to pro-vide constraints. However, the fluid mechanicalproperties of the mantle, which are vital for accuratesimulation of plume dynamics, are also subject touncertainties at the order of magnitude level. Withinthe confines of the physical or kinematical relationsexpressed in a model’s governing equations, thereis thus a range of possible plume‐ridge interactionsas a function of input parameters. While the modelstherefore define a population of valid scenarios,observations of plume‐ridge interaction on theearth must identify those that apply.

[44] Modeling of plume‐ridge systems has beenundertaken by Ribe et al. [1995], Ribe [1996], Itoet al. [1999], Ribe and Delattre [1998], Hall andKincaid [2003], and Ruedas et al. [2004], and aclear result from these studies is the importance ofdehydration melting in controlling the nature ofplume outflow. Hall and Kincaid [2003] find, inparticular, that dehydration melting at the top of theconduit of an upwelling plume generates a plug ofviscous material, flattening the rheological bound-ary layer and forcing plume material to flow outhorizontally. In our model we have considered thishorizontal flow to be axisymmetric about the plumecenter, interacting with the ridges only by gaining avertical upwelling velocity and having its uppersections entrained into the shallow melt region.


20 of 27

Previous researchers, however, have suggested thatflow can occur that is channelized under the ridgeaxis [White et al., 1995; Albers and Christensen,2001]. Given the tendency of dehydration meltingreactions to flatten the basal topography of therheological lithosphere, the approximation of radialoutflow seems justified and is consistent withindependent constraints [White and Lovell, 1997;Jones et al., 2002; Poore et al., 2009]. Our con-ceptual model of asymmetry generation does notrequire radial outflow though and it would be pos-sible for material flowing along the ridge axis toexperience gradual extraction of the plume compo-nent in a similar manner to that proposed for radiallyspreading material. In this case the magnitude ofasymmetry would be set by the relative proximityof ridges to the plume center, but in generalchannelized outflow would promote more sym-metric along‐ridge geochemical distributions.

[45] A second question regarding the flow ofplume material in the mantle is whether the tilt ofplumes, coupled with plate shear, can cause thepreferential advection of plume material in a par-ticular direction. As noted previously, tilt of theIceland, Galápagos and Azores plumes have allbeen used to explain asymmetry in along ridgeobservables [Shen et al., 2002; Schilling et al.,2003; Yang et al., 2006]. Our tests for symmetricplume centers allows for these conclusions to bereexamined. The key result from the misfit mappingis that when the geochemical and bathymetric pro-files are plotted about a preferred plume center,bathymetry is symmetric (low Z‐RMS misfit), butisotope and trace element profiles retain a strongasymmetry (Figures 8–10 and Table 1). Thisindicates the operation of a process to which theincompatible elements of the plume source aresensitive, but which essentially leaves the advectionof the plume’s thermal signal unchanged. Prefer-ential flow of plume material in the direction ofplume tilt and/or in the direction of plate shearwould not produce these observations; one wouldinstead expect an asymmetry of this kind to con-currently influence basalt trace element chemistryand elevation at the ridge axis. The fact that weobserve symmetry in bathymetry is thus consistentwith a mainly radially uniform horizontal advec-tion of the plume’s thermal signature.

[46] We propose that the process which acts onthis background of axisymmetric flow to generateasymmetry in the incompatible element and iso-topic composition of ridge basalts, is deep partialmelting of enriched components within the plumesource during its transit beneath spreading centers.

This process would have a minimal effect on theoutflow of plume material. First, it would requireonly a small supply of latent heat from the overallthermal reservoir advected by plume material, themelt fractions modeled being at most 5%. Second,this low degree of melting is anticipated to occurbefore material intersects the anhydrous peridotitesolidus, the point at which the greatest increasesin viscosity are predicted [Hirth and Kohlstedt,1996]. Finally, although the model includes thin-ning of the sheet of outflowing plume material asit is incorporated into the shallow subridge meltregions, only the uppermost several kilometers arelost this way, which would have a minimal iso-static effect. The anticipated absence of stronggeophysical tracers of this partial melting eventare why geochemical and geophysical profiles canbecome decoupled along ridges adjacent to aplume.

[47] Fracture zones have also been suggested aspotentially interacting with plume outflow, dam-ming the spread of buoyant asthenosphere as itmeets a step‐like change in lithospheric thickness[Vogt and Johnson, 1975]. The likely importanceof this process seems minimal however, given thepreviously discussed control of melting on the rhe-ological lithosphere’s basal topography. The resultsof our model runs are sensitive to fracture zonesthough, and Figures 11 and 13 show that rapidchanges in the deep melt region distance traversedand depletion of Y are present when ridge stepsoccur. This is a product of the simplicity of themodel, which considers melt regions as discretetriangular zones of upwelling. With a more realisticcorner flow model, these discontinuities in meltregion distance profiles and Y would becomesmoothed, as the velocity field at depth is madecontinuous.

4.2. Plume Composition

[48] There remain fundamental questions sur-rounding the composition and compositional struc-ture of mantle plumes. Elevated incompatible traceelement concentrations over regions of mid‐oceanridge have for a long time been taken as evidencefor plumes supplying enriched material to the uppermantle [Schilling, 1973]. However, the super-position of compositional anomalies in the plumewith its high upwelling velocity and elevated tem-perature, means that basalt chemistry records aconvolution of source and dynamical signals[Maclennan et al., 2001; Ito and Mahoney, 2005].A further uncertainty lies in the mapping of the


21 of 27

compositional variation from the mantle into vol-canic systems in the crust, during which informa-tion on the spatial distribution of heterogeneitiesis lost [Rubin et al., 2009]. An important implicationof this study is that ridge geometry’s influence inprocessing of mantle geochemical signals shouldalso be included in future refinements of modelsthat seek to use MORB compositions to extract thewavelength of mantle heterogeneity [Agranier et al.,2005; Meyzen et al., 2007].

[49] Despite these hindrances to observing plumecompositional structure, it is possible to make pre-dictions about what the source material must belike in order for certain processes to be operating.In the case of our model, the development ofasymmetry is dependent upon the presence of anenriched fusible component within the source.This component progressively melts out as theplume material is decompressed in the deep partof subridge melt regions, allowing for the devel-opment of plume material that is variably depletedin its enriched component as a function of deepmelt region distance traversed. The model used tocalculate the depletion of Y specifically required along‐wavelength heterogeneity in the mantle, asso-ciated with the upwelling plume material. This isbecause lateral plume outflow was constrained tooccur within a channel beneath the base of anhy-drous melt regions. Had the mantle upwelling tocompensate for the loss of plume material to theshallow melt regions been of the same compositionand compositional structure (enriched blebs in amore depleted matrix) as the plume material, thenthere would have been no net change in sourcereaching the melt regions with distance from plumecenter. In such a scenario neither asymmetry nor adecrease in plume signature would have devel-oped with distance from the plume center. Thusour model required ambient mantle to consist of amore depleted bulk composition to match theobservations.

[50] However, the assumption that plumes carry anenrichment that is either not present, or presentonly at much lower proportions, in the ambientmantle, does not have to hold. Outflow of theplume material in a channel below the anhydrousmelt region, as in our model, is only one of mul-tiple possible dynamical scenarios. If this constraintis relaxed, more complex situations can be exploredinvolving the effects of temperature, compositionand flow field, which could produce asymmetrywithout a long‐wavelength compositional anomaly.One example would be allowing for some lateralflux of plume material within the anhydrous part

of the melting region. In this event, ambientmantle material subjected to plume like flow fields,could create along‐ridge incompatible element andisotopic profiles similar to those predicted for long‐wavelength compositional anomalies. The require-ment is, again, a compositionally heterogeneoussource, with fusible components and a depletedmatrix. Thus, the results of this study do not provideevidence for or against long‐wavelength composi-tional variation in the mantle. Our model is simplyone possible method for generating asymmetry,having used the assumption that plumes representmaterial of a different bulk composition to ambientmantle.

[51] In the model of Murton et al. [2002], sheathsof transition zone and upper mantle material arewrapped around the Iceland plume during its ascentthrough the upper mantle. The effect is envisagedto be a plume with radially symmetric composi-tional zonation, mapped into MAR basalts as analong‐ridge gradient in chemistry. We model theplume as heterogeneous, but in contrast to theMurton et al. [2002] model, with the enrichedcomponent uniformly dispersed. Although plumesmay have internal compositional zonation, we donot consider it necessary in the case of either theIceland or the Galápagos hot spots; both asymmetryin observables about the plume center and alongridge gradients in geochemical profiles can beexplained simply by the progressive depletion ofoutflowing plume material.

4.3. Melting Dynamics

[52] The feasibility of our model is dependent uponthe greater fusibility of enriched heterogeneitiescompared with more depleted material, whichenables there to be a decoupling in melting his-tories between components in the plume source.The matrix material, modeled as undergoing nomelting in the deeper part of the melt region, isimagined to have lower water contents and a morerefractory mineralogy, only beginning to melt whenthe anhydrous solidus is intersected. There is goodreason to suspect that water content at even ppmlevel and a modal mineralogy richer in clinopyr-oxene and/or garnet, can produce mantle compo-nents that intersect their solidus at greater pressuresthan anhydrous peridotite. From both thermody-namic models [Hirschmann et al., 1999] and exper-imental constraints [Hirth and Kohlstedt, 1996;Pertermann and Hirschmann, 2003] it is likelythat water can lower the depth of solidus intersectionby 20–90 km and a pyroxenitic source composition


22 of 27

by 35–50 km, depending upon the particular thermaland compositional regime studied. Nichols et al.[2002] have identified the Iceland plume as poten-tially having up to 920 ppm of water in its source.Were this volatile enrichment to be spatially coin-cident with the enrichment of other incompatibleelements, then these enriched blebs would be par-ticularly prone to subanhydrous solidus melt extrac-tion. It therefore seems reasonable to suppose thatpartial melting of the enriched component withinthe mantle plume could be occurring at depths atwhich matrix peridotite remains subsolidus.

[53] An important assumption regarding the low‐degree melting modeled, is that the melt generatedis able to be extracted to the surface. Given thelikely small overall melt fractions produced at thisdepth [Hirschmann et al., 1999], significant reten-tion of the melt phase until major melting beginsat the anhydrous solidus would dampen the devel-opment of depletion, possibly masking the effectaltogether. However, from constraints placed onretained melt fractions from uranium series dis-equilibria, it is likely that separation of melt fromsource occurs once melting has exceeded only afew tenths of a percent [McKenzie, 2000]. It there-fore seems plausible for the small melt fractionswe model to be rapidly lost to the surface and theplume isotopic and incompatible element signa-ture with them.

5. Conclusions

[54] Long‐wavelength swell and isotopic profilesalong the mid‐ocean ridges adjacent to the Azores,Galápagos and Iceland hot spots have been assessedfor their symmetry about an array of potentialplume centers. It was found that the bathymetricprofiles could be largely reconciled with symmetricplume outflow with little systematic depth anomalybetween ridge segments. However, around Icelandand the Galápagos, isotopic profiles (and in thecase of Iceland the incompatible element profiles)are fundamentally asymmetric. The Kolbeinseyridge and Galápagos Spreading Center east of the91° West Transform, both showed a more rapiddecline back toMORB like "Nd and 87Sr/86Sr valuesthan corresponding ridge segments on the other sideof the plume. In the region where comparisons aremeaningful and given the scatter of the data, theMid‐Atlantic Ridge around the Azores records moresymmetric plume‐ridge interaction.

[55] In order to understand these observations akinematic model was developed examining the role

of plume‐ridge geometries in imprinting differingsource characteristics on the plume material reach-ing ridges. The key assumptions of the model arethat plume outflow is radial and that the plumematerial consists of an enriched phase embedded ina more depleted matrix. The greater fusibility ofthe enriched material leads to its melting at ahigher pressure than the matrix, allowing for adecoupling in the melting histories of the twocomponents. Thus, by partial melting of the morefusible heterogeneities during flow under spreadingcenters, the isotopic composition of the plumematerial is progressively weighted to that of theremaining matrix. In consequence ridge sourcematerial is depleted as a function of melt regiondistance traversed and distributed about the plumecenter symmetrically or asymmetrically accordingto the particular plume‐ridge geometry.

[56] The result of applying this model to theIcelandic and Galápagos plume‐ridge systems isan asymmetry in source depletion consistent withtheir observed along‐ridge geochemical profiles; thegreater melt region distance traveled by materialflowing north in Iceland’s case and east from theGalápagos, results in it being more heavily depleted,which is expected to produce a more depletederupted basalt chemistry along these ridge segments.For the Azores, source depletion is predicted to besimilar north and south of the plume, although thelimited data available render this poorly constrained.After the initial decay in Azores plume influence,the transition to increasingly enriched lavas northof the Azores cannot be explained by the modeland thus another geodynamic forcing is required.However, for the Iceland and Galápagos plumes itis possible to predict an asymmetry in the senseobserved, without need for plume zonation, tilt orlithospheric damming of outflow. In fact, the strongsymmetry present in bathymetric and crustal thick-ness profiles indicates that plume dispersal is essen-tially radially symmetric. Geochemical asymmetrycan be generated solely by the variable degree ofpartial melting of fusible heterogeneities in theplume source.

Appendix A: Calculating z ScoreRoot‐Mean‐Square Misfit

[57] The z score root‐mean‐square misfit (Z‐RMS)is calculated according the equation, Z‐RMS =(Pdf

x¼d0(px − qx)

2/N)0.5, where px and qx are valuesof the normalized geophysical or geochemicalobservable along each profile at a distance x from


23 of 27

the plume center and N is the total number of pointsalong the profile considered. The starting distanced0, is the common distance of closest approach ofeach profile to the plume center and df the com-mon furthest distance along each profile. Whenprofiles are different lengths the distance of thelonger is reduced to the same range as the shorter.

Appendix B: Handling of Geophysicaland Geochemical Data

[58] Elevation: Axial profiles of elevation werepicked from a low‐pass filtered version of theGEBCO_08 global elevation and bathymetrygridded data set (version 20090202, http://www.gebco.net). The filtering was applied using thesame technique as Canales et al. [2002], with an85 km cutoff wavelength such that the pickedtopography reflected long‐wavelength swell andnot variations in ridgemorphology. For Iceland, datawas included south along the Northern VolcanicZone to 64.75°N, this southern limit placed due to atransition to the propagating Eastern Volcanic Zone.

[59] Crustal thickness: Estimates of crustal thick-ness have only been included for the Icelandplume‐ridge system. Data from Darbyshire et al.[2000] was used for the on‐land sections of ridge,Hooft et al. [2006] data for the Kolbeinsey ridgeand Poore [2008] for estimates of crustal thicknessalong the Reykjanes ridge. The resolution at whichcrustal thickness is mapped being already low, thisdata was neither smoothed nor filtered.

[60] Geochemical profiles: For the purposes ofcalculating Z‐RMS misfit between two profileseither side of a plume center, the scattered raw dataneeds to be represented by a single “mean” line.Equally, short‐wavelength variability in the chemi-cal signal needs to be removed in order to expose thelonger‐wavelength signature of plume influence. Tomeet these requirements, raw chemical data wassmoothed by a Gaussian filter of 85 km width(standard deviation 14.2 km) and it was this profilethat was interpolated for misfit mapping. Althoughthe techniques differ, this wavelength of smoothingis in accordance with the wavelength used to filterbathymetric data, thus profiles of geochemical andgeophysical observables preserve spatial variabilityon similar scales.

[61] The major elements have been corrected in anattempt to compensate for the effects of low‐pressure fractionation according to the scheme ofKlein and Langmuir [1987] and are presented asX8, where “8” indicates a shift back to the con-

centration of X at 8 wt % MgO; however, ratherthan use a fixed regression line, a local correctiongradient was calculated from all available data,giving the equation Na8 = Na2O + 0.27(MgO − 8).Only lavas with MgO concentration between 5 and8.5 wt % MgO have been included for correction.Incompatible trace element ratios can be moderatelysensitive to fractional crystallization, although onlyin the most evolved melts, therefore lavas withMgO <6 wt % have been excluded from theprofiles.

Acknowledgments

[62] Constructive reviews from Vincent Salters, ChristophBeier, and two anonymous reviewers are gratefully acknowl-edged, as are comments on an early draft from Sarah Nixon.O.S. was supported by NERC grant NE/H2449/4.

References

Agranier, A., J. Blichert‐Toft, D. Graham, V. Debaille,P. Schiano, and F. Albarède (2005), The spectra of isotopicheterogeneities along the Mid‐Atlantic Ridge, Earth Planet.Sci. Lett., 238, 96–109, doi:0.1016/j.epsl.2005.07.011.

Albers, M., and U. R. Christensen (2001), Channeling ofplume flow beneath mid‐coean ridges, Earth Planet. Sci.Lett., 187, 207–220.

Allégre, C. J., and D. L. Turcotte (1986), Implications of atwo‐component marble‐cake mantle, Chem. Geol., 323,123–127.

Allen, R. M., et al. (2002), Plume‐driven plumbing and crustalformation in Iceland, J. Geophys. Res., 107(B8), 2163,doi:10.1029/2001JB000584.

Beier, C., K. M. Haase, W. Abouchami, M.‐S. Krienitz, andF. Hauff (2008), Magma genesis by rifting of oceanic lith-osphere above anomalous mantle: Terceira Rift, Azores,Geochem. Geophys. Geosyst., 9, Q12013, doi:10.1029/2008GC002112.

Blichert‐Toft, J., A. Agranier, M. Andres, R. Kingsley, J.‐G.Schilling, and F. Albarède (2005), Geochemical segmentationof the Mid‐Atlantic Ridge north of Iceland and ridge‐hot spotinteraction in the North Atlantic, Geochem. Geophys.Geosyst., 6, Q01E19, doi:10.1029/2004GC000788.

Buck, W. R., and E. M. Parmentier (1986), Convectionbeneath young oceanic lithosphere: Implications for thermalstructure and gravity, J. Geophys. Res., 91(B2), 1961–1974.

Campbell, I. H. (2007), Testing the plume theory, Chem.Geol., 241, 153–176.

Campbell, I. H., and R. W. Griffiths (1990), Implications ofmantle plume structure for the evolution of flood basalts,Earth Planet. Sci. Lett., 99, 79–93.

Canales, J. P., G. Ito, R. S. Detrick, and J. Sinton (2002),Crustal thickness along the western Galápagos SpreadingCentre and the compensation of the Galápagos hotspot swell,Earth Planet. Sci. Lett., 203, 311–327.

Chase, C. G. (1979), Asthenospheric counterflow: A kinematicmodel, Geophys. J. R. Astron. Soc., 56, 1–18.

Christie, D. M., R. Werner, F. Hauff, K. Hoernle, and B. B.Hanan (2005), Morphological and geochemical variations


24 of 27

along the eastern Galápagos Spreading Center, Geochem.Geophys. Geosyst., 6, Q01006, doi:10.1029/2004GC000714.

Courtney, R. C., and R. S. White (1986), Anomalous heat‐flow and geoid across the Cape‐Verde Rise—Evidencefor dynamic support from a thermal plume in the mantle,Geophys. J. R. Astron. Soc., 87, 815–867.

Darbyshire, F. A., R. S. White, and K. F. Priestley (2000),Structure of the crust and uppermost mantle of Iceland froma combined seismic and gravity study, Earth Planet. Sci.Lett., 181, 409–428.

Dasgupta, R., M. M. Hirschmann, and K. Stalker (2006),Immiscible transition form carbonate‐rich to silicate‐richmelts in the 3 GPa melting interval of eclogite + CO2 andgenesis of silica‐undersaturated ocean island lavas, J. Petrol.,57(4), 647–671.

Debaille, V., J. Blichert‐Toft, A. Agranier, R. Doucelance,P. Schiano, and F. Albarède (2000), Structure of the crustand uppermost mantle of Iceland from a combined seismicand gravity study, Earth Planet. Sci. Lett., 181, 409–428,doi:10.1016/j.epsl.2005.11.004.

DeMets, C., R. G. Gordon, D. F. Argus, and S. Stein (1994),Effect of recent revisions to the geomagnetic reversal timescale on estimates of current plate motions, Geophys. Res.Lett., 21(20), 2191–2194.

Detrick, R. S., H. D. Needham, and V. Renard (1995), Gravityanomalies and crustal thickness variations along the Mid‐Atlantic Ridge between 33°N and 40°N, J. Geophys. Res.,100(B3), 3767–3787.

Devey, C. W., C. D. Garbeshonberg, P. Stoffers, C. Chauvel,and D. F. Mertz (1994), Geochemical effects of dynamicmelting beneath ridges: Reconciling major and trace‐elementvariations in Kolbeinsey (and global) mid‐ocean ridge basalt,J. Geophys. Res., 99(B5), 9077–9095.

Dosso, L., H. Bougault, C. Langmuir, C. Bollinger, O. Bonnier,and J. Etoubleau (1999), The age and distribution of mantleheterogeneity along theMid‐Atlantic Ridge (31–41°N),EarthPlanet. Sci. Lett., 170, 269–286.

Fitton, G. J., A. D. Saunders, M. J. Norry, B. S. Hardarson,and R. N. Taylor (1997), Thermal and chemical structureof the Iceland plume, Earth Planet. Sci. Lett., 153, 197–208.

Geist, D. J., W. M. White, and A. R. McBirney (1988), Plume‐asthenosphere mixing beneath the Galápagos archipelago,Nature, 333, 657–600.

Goslin, J., and T. S. Party (1999), Extent of Azores plumeinfluence on the Mid‐Atlantic Ridge north of the hotspot,Geology, 27, 991–994, doi:10.1130/0091-7613(1999)027<0991:EOAPIO>2.3.CO;2.

Hall, P. S., and C. Kincaid (2003), Melting, dehydration, andthe dynamics of off‐axis plume‐ridge interaction, Geochem.Geophys. Geosyst., 4(9), 8510, doi:10.1029/2003GC000567.

Hart, S. R. (1971), K, Rb, Cs, Sr and Ba contents and Sr isotoperatios of ocean floor basalts, Philos. Trans. R. Soc. London,Ser. A, 268(1192), 573–587.

Hirschmann, M. M., P. D. Asimow, M. S. Ghiorso, and E. M.Stolper (1999), Calculation of peridotite partial melting fromthermodynamic models of mineral and melts. III. Controlson isobaric melt production and the effect of water on meltproduction, J. Petrol., 40(5), 831–851.

Hirschmann, M. M., T. Kogiso, M. B. Baker, and E. M. Stolper(2003), Alkalic magmas generated by partial melting of garnetpyroxenite, Geology, 31, 481–484, doi:10.1130/0091-7613(2003)031<0481.

Hirth, G., and D. L. Kohlstedt (1996), Water in the oceanic uppermantle: Implications for rheology, melting and the evolution ofthe lithosphere, Earth Planet. Sci. Lett., 144, 93–108.

Hoernle, K., R. Werner, J. P. Morgan, D. Garbe‐Schönberg,J. Bryce, and J. Mrazek (2000), Existence of complete spa-tial zonation in the Galápagos plume for at least 14 m.y.,Geology, 28, 435–438.

Hooft, E. E. E., D. R. Toomey, and S. C. Solomon (2003),Anomalously thin transition zone beneath the Galápagos hot-spot, Earth Planet. Sci. Lett., 215, 55–64, doi:10.1016/S0012-821X(03)00517-X.

Hooft, E. E. E., B. Brandsdttir, R. Mjelde, H. Shimamura, andY. Murai (2006), Asymmetric plume‐ridge interactionaround Iceland: The Kolbeinsey Ridge Iceland SeismicExperiment, Geochem. Geophys. Geosyst., 7, Q05015,doi:10.1029/2005GC001123.

Ingle, S., G. Ito, J. J. Mahoney, W. Chazey III, J. Sinton,M. Rotella, and D. M. Christie (2010), Mechanisms ofgeochemical and geophysical variations along the westernGalápagos Spreading Center, Geochem. Geophys. Geosyst.,11, Q04003, doi:10.1029/2009GC002694.

Ito, G. (2001), Reykjanes ‘V’‐shaped ridges originatingfrom a pulsing and dehydrating mantle plume, Nature, 411,681–684.

Ito, G., and J. H. Mahoney (2005), Flow and melting of a het-erogeneous mantle: 1. Method and importance to the geo-chemistry of ocean island and mid‐ocean ridge basalts,Earth Planet. Sci. Lett., 230, 29–46, doi:10.1016/j.epsl.2004.10.035.

Ito, G., Y. Shen, G. Hirth, and C. J. Wolfe (1999), Mantleflow, melting, and dehydration of the Iceland mantle plume,Earth Planet. Sci. Lett., 165, 81–96.

Jakobsson, S., K. Jónasson, and I. Sigurdsson (2008), Thethree igneous rock series of Iceland, Joekull, 58, 117–138.

Jean‐Baptiste, P., P. Allard, R. Coutinho, T. Ferreira, E. Fourré,G. Queiroz, and J. L. Gaspar (2009), Helium isotopes inhydrothermal volcanic fluids of the Azores archipelago,Earth Planet. Sci. Lett., 281, 70–80.

Jones, S. M., N. White, and J. Maclennan (2002), V‐shapedridges around Iceland: Implications for spatial and temporalpatterns of mantle convection, Geochem. Geophys. Geosyst.,3(10), 1059, doi:10.1029/2002GC000361.

Kaban, M. K., O. G. Flóvenz, and G. Pálmason (2002), Natureof the crust‐mantle transition zone and the thermal state ofthe upper mantle beneath Iceland from gravity modelling,Geophys. J. Int., 149, 281–299.

Klein, E. M., and C. H. Langmuir (1987), Global correlationsof ocean ridge basalt chemistry with axial depth and crustalthickness, J. Geophys. Res., 92(B8), 8089–8115.

Kogiso, T., K. Hirose, and E. Takahashi (1998), Melting experi-ments on homogeneous mixtures of peridotite and basalt:Application to the genesis of ocean island basalts, EarthPlanet. Sci. Lett., 162, 45–61.

Kokfelt, T., K. Hoernle, F. Hauff, J. Fiebig, R. Werner, andD. Garbe‐Schönberg (2006), Combined trace element andPb‐Nd‐Sr‐O isotope evidence for recycled oceanic crust(upper and lower) in the Iceland mantle plume, J. Petrol.,47(9), 1705–1749.

Kurz, M. D., and D. Geist (1999), Dynamics of the Galápagoshotspot from helium isotope geochemistry,Geochim. Cosmo-chim. Acta, 63(23–24), 4139–4156.

Li, A., and R. S. Detrick (2006), Seismic structure of IcelandfromRayleigh wave inversions and geodynamic implications,Earth Planet. Sci. Lett., 191, 295–310.

Maclennan, J. (2008), Lead isotope variability in olivine‐hosted melt inclusions from Iceland, Geochim. Cosmochim.Acta, 72, 4159–4176.


25 of 27

Maclennan, J., D. McKenzie, and K. Grönvold (2001), Plume‐driven upwelling under central Iceland, Earth Planet. Sci.Lett., 194, 67–82.

Maia, M., J. Goslin, and P. Gente (2007), Evolution of theaccretion processes along the Mid‐Atlantic Ridge north ofthe Azores since 5.5 Ma: An insight into the interactionsbetween the ridge and the plume, Geochem. Geophys.Geosyst., 8, Q03013, doi:10.1029/2006GC001318.

McKenzie, D. (2000), Constraints on melt generation andtransport rates from U‐series activity ratios, Chem. Geol.,162, 81–94.

Mello, S. L. M., J. R. Cann, and C. M. R. Fowler (1999),Anomalous mantle at 45°NMid‐Atlantic Ridge?, J. Geophys.Res., 104(B12), 29,335–29,349.

Mertz, D. F., and K. M. Haase (1997), The radiogenic isotopecomposition of the high‐latitude North Atlantic mantle,Geology, 25, 411–414.

Mertz, D. F., C. W. Devey, W. Todt, P. Stoffers, and A. W.Hofmann (1991), Sr‐Nd‐Pb isotope evidence against plumeasthenosphere mixing north of Iceland, Earth Planet. Sci.Lett., 107, 243–255.

Meyzen, C. M., J. Blichert‐Toft, J. N. Ludden, E. Humler,C. Mével, and F. Albarède (2007), Isotopic portrayal of theEarth’s upper mantle flow field, Nature, 447, 1069–1074.

Murton, B. J., R. N. Taylor, and M. N. Thirlwall (2002),Plume‐ridge interaction: A geochemical perspective fromthe Reykjanes Ridge, J. Petrol., 43(11), 1987–2012.

Nichols, A. R. L., M. R. Carroll, and A. Höskuldsson (2002),Is the Iceland hot spot also wet? Evidence from the watercontents of undegassed submarine and subglacial pillowbasalts, Earth Planet. Sci. Lett., 202, 77–87.

Pearce, J. A. (2005), Mantle preconditioning by melt extractionduring flow: Theory and petrogenetic implications, J. Petrol.,46, 973–997, doi:10.1093/petrology/egi007.

Peate, D. W., J. A. Baker, S. P. Jakobsson, T. E. Waight,A. J. R. Kent, N. V. Grassineau, and A. C. Skovgaard (2009),Historic magmatism on the Reykjanes Peninsula, Iceland: Asnap‐shot of melt generation at a ridge segment, Contrib.Mineral. Petrol., 157(3), 359–382, doi:10.1007/s00410-008-0339-4.

Pertermann, M., and M. M. Hirschmann (2003), Anhydrouspartial melting experiments on MORB‐like eclogite: Phaserelations, phase compositions and mineral‐melt partitioningof major elements at 2–3 GPa, J. Petrol., 44(12), 2173–2201.

Poore, H. (2008), Neogene epeirogeny and the Iceland plume,Ph.D. thesis, Univ. of Cambridge, Cambridge, U. K.

Poore, H. R., N. J. White, and S. M. Jones (2009), A Neogenechronology of Iceland plume activity from V‐shaped ridges,Earth Planet. Sci. Lett., 283, 1–13, doi:10.1016/j.epsl.2009.02.028.

Poreda, R., J. G. Schilling, and H. Craig (1986), Helium andhydrogen isotopes in ocean‐ridge basalts north and southof Iceland, Earth Planet. Sci. Lett., 78, 1–17.

Rabinowicz,M.,G.Ceuleneer,M.Monnereau, andC.Rosemberg(1990), Three dimensional models of mantle flow across alow‐viscosity zone: Implications for hotspot dynamics, EarthPlanet. Sci. Lett., 99, 170–184.

Ribe, N.M. (1996), The dynamics of plume‐ridge interaction: 2.Off‐ridge plumes, J. Geophys. Res., 101(7), 16,195–16,204.

Ribe, N. M., and W. L. Delattre (1998), The dynamics ofplume ridge interaction—III. The effects of ridge migration,Geophys. J. Int., 133, 511–518.

Ribe, N. M., U. R. Christensen, and J. Theiing (1995), Thedynamics of plume‐ridge interaction, 1. Ridge‐centredplumes, Earth Planet. Sci. Lett., 134, 155–168.

Robinson, E. M., B. Parsons, and S. F. Daly (1987), The effectof a shallow low viscosity zone on the apparent compensationof mid‐plate swells, Earth Planet. Sci. Lett., 82, 335–348.

Rubin, K. H., J. M. Sinton, J. Maclennan, and E. Hellebrand(2009), Magmatic filtering of mantle compositions at mid‐ocean‐r idge volcanoes, Nat. Geosci. , 2 , 321–328,doi:10.1038/NGEO504.

Rudge, J. F., M. E. S. Champion, N. White, D. McKenzie, andB. Lovell (2008), A plume model of transient diachronousuplift at the Earth’s surface, Earth Planet. Sci. Lett., 267,146–160, doi:10.1016/j.epsl.2007.11.040.

Ruedas, T., H. Schmeling, G. Marquart, A. Kreutzmann, andA. Junge (2004), Temperature and melting of a ridge‐centredplume with application to Iceland. Part I: Dynamics and crustproduction, Geophys. J. Int., 158, 729–743, doi:10.1111/j.1365-246X.2004.02311.x.

Schilling, J. G. (1973), Iceland mantle plume: Geochemicalstudy of Rykjanes Ridge, Nature, 242, 565–571.

Schilling, J.‐G., R. H. Kingsley, and J. D. Devine (1982),Galápagos hot spot‐spreading center system: 1. Spatialpetrological and geochemical variations (83°W–111°W),J. Geophys. Res., 87(B7), 5593–5610.

Schilling, J. G., M. Zajac, R. Evans, T. Johnston, W. White,J. D. Devine, and R. Kingsley (1983), Petrology and geo-chemical variations along the Mid‐Atlantic Ridge from29°N to 73°N, Am. J. Sci., 283, 510–586.

Schilling, J. G., R. Kingsley, D. Fontignie, R. Poreda, andS. Xue (1999), Dispersion of the Jan Mayen and Icelandmantle plumes in the Arctic: A He‐Pb‐Nd‐Sr isotope tracerstudy of basalts from the Kolbeinsey, Mohns, and Knipovichridges, J. Geophys. Res., 104(B5), 10,543–10,569.

Schilling, J.‐G., D. Fontignie, J. Blichert‐Toft, R. Kingsley,and U. Tomza (2003), Pb‐Hf‐Nd‐Sr isotope variations alongthe Galápagos Spreading Center (101°–83°W): Constraintson the dispersal of the Galápagos mantle plume, Geochem.Geophys. Geosyst., 4(10), 8512, doi:10.1029/2002GC000495.

Shaw, D. M. (1970), Trace eement fractionation during ana-texis, Geochim. Cosmochim. Acta, 34, 237–243.

Shen, Y., et al. (2002), Seismic evidence for a tilted mantleplume and north‐south mantle fow beneath Iceland, EarthPlanet. Sci. Lett., 197, 261–272.

Sobolev, A. V., A. W. Hofmann, S. V. Sobolev, andI. K. Nikogosian (2005), An olivine‐free mantle source ofHawaiian shield basalts, Nature, 434, 590–597.

Spiegelman, M., and D. McKenzie (1987), Simple 2D modelsof melt extraction at mid‐ocean ridges and island arcs, EarthPlanet. Sci. Lett., 83, 137–152.

Stracke, A., A. Zindler, V. J. M. Salters, D. McKenzie,J. Blichert‐Toft, F. Albarède, and K. Grönvold (2003), Theis-tareykir revisited, Geochem. Geophys. Geosyst., 4(2), 8507,doi:10.1029/2001GC000201.

Thirlwall, M. F., M. A. M. Gee, R. N. Taylor, and B. J. Murton(2004), Mantle components in Iceland and adjacent ridgesinvestigated using double‐spike Pb isotope ratios, Geochim.Cosmochim. Acta, 68(2), 361–386.

Villagómez, D. R., D. R. Toomey, E. E. E. Hooft, andS. C. Solomon (2007), Upper mantle structure beneath theGalápagos Archipelago from surface wave tomography,J. Geophys. Res., 112, B07303, doi:10.1029/2006JB004672.

Vogt, P. R., and G. L. Johnson (1975), Transform faults andlongitudinal flow below the Midoceanic Ridge, J. Geophys.Res., 80(11), 1399–1428.

Vogt, P. R., and W. Y. Jung (2004), The Terceira Rift as hyper‐slow, hotspot‐dominated oblique spreading axis: A comparison


26 of 27

with other slow‐spreading plate boundaries, Earth Planet. Sci.Lett., 218, 77–90, doi:10.1016/S0012-821X(03)00627-7.

White, N., and B. Lovell (1997), Measuring the pulse of aplume with the sedimentary record, Nature, 387, 888–891.

White, R. S., J. W. Bown, and J. R. Smallwood (1995), Thetemperature of the Iceland plume and origin of outward‐propagating V‐shaped ridges, J. Geol. Soc. London, 152,1039–1045.

Wolfe, C. J., I. T. Bjarnason, J. C. VanDecar, and S. C. Solomon(1997), Seismic structure of the Iceland mantle plume,Nature, 383, 245–247.

Yang, T., Y. Shen, S. V. der Lee, S. C. Solomon, and S. H.Hung (2006), Upper mantle structure beneath the Azoreshotspot from finite‐frequency seismic tomography, EarthPlanet. Sci. Lett., 250, 11–26, doi:10.1016/j.epsl.2006.07.031.

Yaxley, G. M., and D. H. Green (2000), Reactions betweeneclogite and peridotite: Mantle refertilisation by subductionof oceanic crust, Contrib. Mineral. Petrol., 139, 326–338.

Yaxley, G. M., and A. V. Sobolev (2007), High‐pressure par-tial melting of gabbro and its role in the Hawaiian magmasource, Contrib. Mineral. Petrol., 154, 371–383, doi:10.1007/s00410-007-0198-4.

Yu, D., D. Fontignie, and J. G. Schilling (1997), Mantleplume‐rdige interactions in the central North Atlantic: ANd isotope study of Mid‐Atlantic Ridge basalts from 30°Nto 50°N, Earth Planet. Sci. Lett., 146, 259–272.

Zindler, A., S. R. Hart, and F. A. Frey (1979), Nd and Sr iso-tope ratios and rare earth element abundances in ReykjanesPeninsula basalts: Evidence for mantle heterogeneity beneathIceland, Earth Planet. Sci. Lett., 45, 249–262.


27 of 27

control of the symmetry of plume ridge interaction by...

Documents