Seismic properties of an asthenospherized lithospheric ... · Seismic properties of an asthenospherized lithospheric mantle: constraints from lattice preferred ... coarse-granular

Seismic properties of an asthenospherized lithospheric mantle:constraints from lattice preferred orientations in peridotite

from the Ronda massif

Alain Vauchez a;*, Carlos J. Garrido b

a ISTEEM, Laboratoire Tectonophysique, Universite de Montpellier 2 and CNRS, Place Euge©ne Bataillon cc49, 34095 Montpellier,France

b Instituto Andaluz de Ciencias de la Tierra, CSIC and Universidad de Granada, Facultad de Ciencias, Campus Fuentenueva s/n,18002 Granada, Spain

Received 29 January 2001; received in revised form 17 May 2001; accepted 12 July 2001

Abstract

Above a mantle plume, the lithosphere is thermally èroded'. It is however not clear whether heating and partialmelting of the lithosphere may erase the mineral lattice preferred orientation (LPO) inherited from previous tectonicevents or if, in the absence of large-scale flow, this fabric is preserved. To evaluate the effect of heating and partialmelting on the seismic properties of the lithospheric mantle, we have measured the LPO and computed the seismicproperties of peridotites from the Ronda massif (Spain). In this massif, a narrow (9 400 m) coarsening front separates aporphyroclastic peridotite domain, interpreted as old lithospheric mantle, from a coarse-granular peridotite domainproduced by annealing and limited partial melting (6 6.5%) of the porphyroclastic peridotites. The olivine LPO in theporphyroclastic peridotites is moderate. The [100] and [001] axes are distributed within the foliation with a maximum of[100] parallel to the lineation, and the [010] axes are concentrated close to the normal to the foliation. The olivine LPOdoes not vary drastically across the coarsening front: the LPO strength decreases slightly and symmetry of the patternprogressively turns more orthorhombic. On the other hand, the strength of the orthopyroxene LPO increases. Theconsistency of olivine LPO translates to similar seismic properties of peridotites in the two domains. Especially, theanisotropy of both compressional and shear waves (P- and S-waves) remains almost unchanged across the entire massif.These results support that heating and partial melting (asthenospherization) of the lithospheric mantle do notnecessarily obliterate the minerals LPO inherited from previous tectonic events. The `structural memory' of thelithosphere may therefore be preserved even in the àsthenospherized' mantle. In a region of asthenosphere^lithosphereinteraction, tomography studies would indicate a largely attenuated lithosphere from the presence of a shallow low-velocity anomaly while S-wave splitting measurement yields delays between arrivals of the fast and slow split wavesrequiring a larger lithosphere thickness. This apparent discrepancy may be resolved considering the existence of a `ghostlithosphere' having lithospheric characteristics regarding anisotropy studies and asthenospheric properties regardingseismic waves velocities. ß 2001 Elsevier Science B.V. All rights reserved.

0012-821X / 01 / $ ^ see front matter ß 2001 Elsevier Science B.V. All rights reserved.PII: S 0 0 1 2 - 8 2 1 X ( 0 1 ) 0 0 4 4 8 - 4

* Corresponding author. Tel. : +33-467-143895; Fax: +33-467-143603.E-mail addresses: [email protected] (A. Vauchez), [email protected] (C.J. Garrido).

EPSL 5952 27-9-01

Earth and Planetary Science Letters 192 (2001) 235^249

www.elsevier.com/locate/epsl

Keywords: asthenosphere; lithosphere; lattice; preferred orientation; seismic methods; anisotropy; mantle; spinel peridotite;partial melting

1. Introduction

During orogenic processes, the plastic deforma-tion of the lithospheric mantle generates a latticepreferred orientation (LPO) of olivine ^ the mainmineral constituent of peridotites ^ that remainsfrozen at the end of the orogeny (e.g. [1,2]). Oli-vine LPO is a ubiquitous feature of peridotitexenoliths brought to the surface by volcanismand of peridotites in orogenic mantle massifsand ophiolites (e.g. [3^5]). Because olivine crystalsare elastically and mechanically anisotropic [6,7],the LPO of olivine in mantle peridotites is re-garded as the main source of seismic and mechan-ical anisotropy of the subcontinental lithosphericmantle (SCLM) [1,2,8^10].

Where seismic shear waves (S-waves) travelthrough an elastically anisotropic medium, theyare split into two orthogonally polarized S-wavestraveling at di¡erent velocities. Measurements ofthe splitting of S-waves provide information onthe anisotropy of the mantle that can be used tounravel the structure and £ow pattern of theupper mantle (e.g. [11,12]). In continental rift do-mains a critical issue for the interpretation of seis-mic anisotropy measurements is whether the pre-existing olivine fabric and related seismic proper-ties of the SCLM are preserved during its thermalerosion and/or melting by upwelling astheno-sphere [13,14]. Although it is clear that mantlexenoliths equilibrated at high temperature havean olivine LPO [2,5], their small size and thelack of structural reference frame make it unclearwhether the observed olivine LPO is coherent atlength-scales relevant to seismic observation.

The coarsening front of the Ronda orogenicmassif (S. Spain) is a narrow (6 400 m) frontseparating a domain of porphyroclastic perido-tites (the spinel tectonite domain) from a domainof seemingly undeformed coarse-granular perido-tites (the coarse-granular domain) [15^17] (Fig. 1).Geochemical, structural and petrological evidenceindicates that the Ronda coarsening front repre-sents a narrow boundary between a partially mol-

ten domain (the coarse-granular peridotites)formed at the expenses of lithospheric peridotites(the spinel tectonite domain) at near-astheno-spheric conditions (v1200³C, V1.5 GPa; [15^19]). The Ronda coarsening front hence o¡ers aunique opportunity for the in situ investigation ofvariations of the LPO of minerals and the intrin-sic seismic properties of peridotites through a fos-sil àsthenospherization' front at length-scales ofseveral kilometers. Here we present an electronback-scattered di¡raction (EBSD) study of theLPO of olivine and pyroxenes in 16 peridotitessampled across the Ronda coarsening front. Thisstudy aims to investigate how coarsening and par-tial melting a¡ected the olivine LPO and the in-trinsic seismic properties of the lithospheric peri-dotites.

2. The Ronda coarsening front

The Ronda peridotite (S. Spain) is the largest(300 km2) of several orogenic peridotite massifsexposed in the Betic (Ronda massif) and theRift (Beni Boussera massif) orogenic belts in thewesternmost part of the Alpine orogen (Fig. 1).Reviews of the regional geology of the Rondaperidotite and the Alboran area can be found in[16,20].

A ¢rst order feature of the Ronda peridotite isits distinct metamorphic, tectonic and geochemi-cal zoning in km-scale domains [16,21]. Van derWal and Vissers [15,16] mapped three main tecto-no-metamorphic domains in the western part ofthe Ronda massif (Fig. 1):

(1) A spinel tectonite domain formed of in-tensely foliated porphyroclastic spinel peridotites.This domain represents an old lithospheric man-tle, isolated from the convective mantle at ca. 1.36Ga [22].

(2) A coarse-granular domain composed ofgranular spinel peridotites and subordinate spinelpyroxenite layers, exposed in the central part ofthe massif. This domain is characterized by

EPSL 5952 27-9-01

A. Vauchez, C.J. Garrido / Earth and Planetary Science Letters 192 (2001) 235^249236

coarse-granular peridotites displaying completelyannealed microstructures.

(3) A plagioclase tectonite domain composed ofporphyroclastic plagioclase peridotites usually as-sociated with km-scale shear zones in the southernand eastern parts of the Ronda massif. The pres-ence of crustal bodies embedded in the plagioclasetectonites [15,23] indicates that the developmentof this tectonic domain was related to the em-placement of the peridotites into the crust [15].

The transition from the spinel tectonite domainto the coarse-granular peridotite domain occursover a relatively narrow front (`the coarseningfront') marked by a sharp increase in the grainsize of peridotite and pyroxenite minerals andby the development of a typical annealed micro-

structure. Detailed mapping indicates that thecoarsening front extends eastward over a distanceof at least 20 km [16,24]. Van der Wal [24]stressed several lines of ¢eld evidence suggestingthat coarse-granular peridotites developed afterspinel tectonites through grain growth: (1) defor-mation in the spinel tectonites domain occurred ata signi¢cantly lower temperature 6 1000³C thanthe development of the coarse-granular domain(v1250³C); (2) the transition from spinel tecton-ites to coarse-granular peridotites is not abrupt,and the microstructure of peridotites at the coars-ening front is intermediate between porphyroclas-tic and coarse-granular; (3) the tectonic fabric inthe spinel tectonites and the coarse-granular do-mains are parallel (Fig. 1). Indeed, although the

Fig. 1. Upper left: map of Ronda peridotite showing the boundaries of the metamorphic facies simpli¢ed from Obata [21](PLAG = plagioclase lherzolite facies). Right: map of the western part of the Ronda peridotite (modi¢ed after Van der Wal [16]and Lenoir et al. [19]) showing the extension of the di¡erent tectono-metamorphic domains, the `coarsening front', and the loca-tion of the samples selected for the present study.

EPSL 5952 27-9-01

A. Vauchez, C.J. Garrido / Earth and Planetary Science Letters 192 (2001) 235^249 237

macroscopic foliation of spinel tectonites dimsgradually due to grain coarsening, the composi-tional layering as well as the spinel trails thatunderline the lineation in spinel tectonites are pre-served in the coarse-granular domain; (4) thecoarse-granular domain displays totally recrystal-lized melting residues of former garnet pyroxenite

dykes (spinel websterites) that still preserve afolded structure typical of the spinel tectonitesdomain; and (5) no evidence of deformation as-sociated with the propagation of the coarseningfront across the lithospheric mantle was foundeither in the ¢eld or in thin sections.

Geochemical data support that the coarse-gran-ular peridotites do not represent a `true' astheno-spheric mantle but rather derived from the spineltectonites, through re-heating, partial melting andinteraction with pervasive basaltic melt [17,18]. Atthe coarsening front, the chemical composition ofpyroxenites [18] and peridotites [17,19] variessharply. Peridotites at either side of the front dis-play a strong bimodal distribution of YbN (Fig.2A), (Ce/Sm)N and (Sm/Yb)N (Fig. 2B). The var-iations of YbN of lherzolites (Fig. 2A) across thecoarsening front may be explained considering thecoarse-granular peridotites as residues of 2.5^6.5% partial melting from peridotites with a com-position similar to the spinel tectonites [19]. Le-noir et al. [19] have also shown that the spineltectonites ahead of the coarsening front displayevidence of melt consuming reactions that over-print the porphyroclastic fabric developed previ-ously under lower temperature conditions. Thesereactions involve precipitation of undeformed cli-

6Fig. 2. Variations of rare-earth element ratios (A,B) and mi-crostructural (C^F) parameters in Ronda peridotites as afunction of the distance from the coarsening front (circles =lherzolites; triangles = harzburgites; open symbols = porphyro-clastic peridotites; dotted symbols = transitional peridotites;¢lled symbols = coarse-granular peridotites). (A,B) Variationof the whole-rock content of YbN normalized to primitivemantle (normalizing values after [46]) in lherzolites and harz-burgites, and (Sm/Yb)N ratio in lherzolites (data from[17,19]). (C^F) Variation of the mean surface area per vol-ume of olivineôlivine interfaces (Sol

v , mm31), the mean inter-cept length of olivine (Lol

3 , mm), and the maximum chordparallel to the foliation of orthopyroxene (Lopx

m , mm) and cli-nopyroxene (Lcpx

m , mm). Measurements have been performedfor the 16 samples selected for this study (samples labeled inC). Additional samples studied by Van der Wal and Bodinier[17] have been integrated in this ¢gure; Sol

v and Lol3 for those

samples were estimated from the raw data of intercept count-ing provided by the authors. The distance from the coarsen-ing front was calculated relative to the front as mapped byLenoir et al. [19] (see Fig. 1). See text for further details onthe computation of microstructural parameters.

EPSL 5952 27-9-01


nopyroxene+spinel aggregates that were subse-quently overprinted by pervasive grain coarsen-ing.

In summary, coherent textural and geochemicalvariations are interpreted as evidence for develop-ment of the coarse-granular domain through fastand transient re-heating up to asthenospheric con-ditions (v1200³C at 1.5 GPa) and partial meltingof a deformed lithosphere. The Ronda coarseningfront is then considered as a fossil melting frontseparating an old SCLM domain (the spinel tec-tonite domain) from its melting residues (thecoarse-granular domain).

3. Sampling and methods

We have focused our study on the western partof the Ronda massif where the coarsening frontand the coarse-granular peridotites are particu-larly well exposed (Fig. 1). In this area, we haveselected 15 samples of peridotites encompassingthe spinel tectonite domain (¢ve samples), thecoarsening front (one sample) and the coarse-granular domain (nine samples) at di¡erent dis-tances from the coarsening front (Figs. 1 and 2).In addition, to explore large-scale variations inthe fabric of spinel tectonites, we have selectedone additional sample from the easternmost partof this domain (sample L37; location not shownin Fig. 1). We avoid sampling close to the shearzone at the NW boundary of the massif becausethe geodynamic meaning of this area is still con-troversial.

The crystallographic orientations of individualgrains of olivine, orthopyroxene and clinopyrox-ene in the selected samples were determined byindexation of EBSD patterns (e.g. [25]). The dif-fraction pattern is generated in a JEOL JSM 5600scanning electron microscope (Universite deMontpellier 2) by the interaction of a vertical elec-tron beam with the grain in an ultra-polished thinsection tilted 70³ relative to the horizontal plane.The pattern is then projected onto a homemadephosphor screen and recorded by a low-light,high-resolution CCD camera. Automatic index-ing of the di¡raction pattern using theCHANNEL+0 software (HKL Technology) re-

trieves the mineralogical nature and the orienta-tion of the grain. The orientation measurementswere conducted grain-per-grain in thin sectionscut perpendicular to the foliation and parallel tothe lineation. Pole ¢gures were computed fromseveral hundreds of grain measurements for oli-vine, and between 60 and 150 grain measurementsfor orthopyroxene and clinopyroxene.

In the same thin sections, we have estimated themean surface area per volume of olivineôlivineinterfaces (Sol

v ; mm31) and the mean interceptlength of olivine (Lol

3 ; mm) using the stereologicalequations Sol

v = 2PolL and Lol

3 = 2Volv /Sol

v [26,27],where Pol

L is the number of olivineôlivine inter-sections with a sampling grid divided by the totallength of the grid, and Vol

v is the volume fractionof olivine. Counting of intersections of olivineôlivine boundaries with a circular grid probewas performed on digital images of the thin sec-tions taken in crossed-polarized light at di¡erentpolarizer orientations. These microstructural pa-rameters were chosen because Sol

v is very sensitiveto grain growth processes and Lol

3 provides a re-liable three-dimensional (3D) measure of themean grain size.

As small grains of pyroxene are hard to recog-nize in digital images, instead of the mean grainsize, we measured the mean maximum chord (Lm,parallel to the sample foliation; Fig. 2E,F) of thethree larger sections of crystals of orthopyroxeneand clinopyroxene in each thin section.

4. Results

4.1. Microstructural parameters

The Solv of porphyroclastic peridotites is rather

variable ranging from 2 to 8 mm31 (ca. 4 mm31

on average; Fig. 2C). Grain growth across thecoarsening front is marked by a factor two de-crease of the Sol

v (1.5 mm31 on average; Fig.2C) of the coarse-granular peridotites relative toporphyroclastic peridotites. The transitional peri-dotites show intermediate Sol

v values. The Lol3 of

porphyroclastic peridotites is 0.4 mm on averageand sharply increases at the coarsening front byup to a factor ¢ve (Fig. 2D). In detail, the Lol

3

EPSL 5952 27-9-01


Fig. 3. Olivine pole ¢gures showing the orientation of the [100], [010] and [001] axes in the 16 samples from the spinel tectonitesand transitional domains (A) and from the coarse-granular domain (B). Equal area projection, lower hemisphere, and contoursat one multiple of uniform distribution intervals. X is the direction de¢ned by the mineral lineation (assumed to parallel the £owdirection), Z is the pole to foliation (assumed to parallel the £ow plane) and Y is the direction normal to the lineation in the fo-liation plane (normal to the ¢gure). Labels on the left side of each set of stereonets are: top: the sample number (Lz = lherzolite;Hz = harzburgite); middle: the distance of the sample from the coarsening front (in m); bottom: the number of grain measure-ments performed in the sample (n). Numbers on the bottom right of stereonets are the maximum density for the given crystallo-graphic axis.

EPSL 5952 27-9-01


increase in the coarse-granular domain is largerfor harzburgites than for lherzolites (Fig. 2D),likely indicating the pinning e¡ect of pyroxeneson olivine growth.

The Lopxm of harzburgites and lherzolites in-

creases at the coarsening front. It reaches a max-imum around 350 m behind the front, and thendecreases toward the center of the coarse-granulardomain (Fig. 2E). Grain coarsening at the front isalso accompanied by a change in aspect ratio oforthopyroxenes. Although less evident, the Lcpx

mseems to vary similarly across the coarsening front(Fig. 2F).

4.2. Olivine LPO

According to Bunge's classi¢cation [28], theporphyroclastic peridotites display olivine LPOfabrics with a [010]-¢ber pattern (Fig. 3A). Thispattern is characterized by clustering of olivine[010] axes close to the normal to the foliation(the Z structural axis) and a stronger concentra-tion of [010] relative to the other crystallographicaxes (see Fig. 3 caption for a de¢nition of the X,Y and Z structural axes). The olivine [100] axesare distributed within or close to the foliation,forming a girdle that contains a maximum con-centration close to the lineation (the X structuralaxis). Although more scattered, the [001] axes dis-play a similar distribution. This fabric patternprobably results from the simultaneous activationof the (010)[001] and (010)[100] slip systems ofolivine at moderate temperature [4] or from atranspressional deformation regime [29].

Despite the abrupt increase of the olivine grainsize, the overall LPO fabric of the transitional andcoarse-granular peridotites is comparable to thatof the spinel tectonites in magnitude and symme-try (Fig. 3B). In detail, however, there is a slightchange in the symmetry of the LPO farther intothe coarse-granular domain where the olivine fab-ric becomes orthorhombic (e.g. samples 33, 6, 2,41; Fig. 3B). The orthorhombic fabric is charac-terized by olivine [100] axes clustered close to theX-axis.

The strength of the LPO of a mineral can beevaluated quantitatively using the integral of theorientation distribution function over the volume

of the aggregate, the J-index [28]. The magnitudeof the J-index depends on the strain of the sampleand ranges from one for a random fabric to in-¢nity for a single crystal [28,30]. The olivine J-index (Jol) of the spinel tectonites varies fromca. 3.5 to ca. 8 (Fig. 4a). These relatively lowvalues of Jol are due to the dispersion of the oli-vine [100] and [001] axes within the foliation plane(Fig. 3). The slightly higher Jol of two porphyro-clastic harzburgites (samples 28 and 12; Fig. 3A)is due to a better concentration of olivine [010]axes and a relatively narrower olivine [100] girdlethat probably re£ects a higher ¢nite strain.

Compared with their spinel tectonite counter-parts, the Jol of the coarse-granular harzburgitesdecreases whereas the Jol of coarse-granular lherz-olites increases slightly (Fig. 4a). This evolutionresults in a relative homogenization of the olivinefabric strength of coarse-granular peridotites.

4.3. Pyroxenes LPO

The LPO of orthopyroxene (Fig. 5) is poorly

Fig. 4. Variation of the J-index of olivine (Jol ; a) and ortho-pyroxene (Jopx ; b) as a function of the distance from thecoarsening front. Labels on symbols in (b) are the number ofgrain measurements used to compute the J-index in eachsample. Symbols as in Fig. 2.

EPSL 5952 27-9-01


developed in harzburgites and lherzolites of thespinel tectonite domain. Conversely, the transi-tional and the coarse-granular peridotites displaymarked orthopyroxene LPOs with a concentra-

tion of [001] axes around the lineation, and of[100] axes close to the pole of the foliation. Thispattern is consistent with activation of the(100)[001] slip system of orthopyroxene.

Overall, there is no clear correlation betweenthe fabric strength of olivine and orthopyroxene(Fig. 4). The lowest orthopyroxene J-indexes(Jopx = 3^5) are found in the spinel tectonite sam-ples farther from the coarsening front (Fig. 4b).From these samples, the Jopx of the spinel tecton-ites increases steadily toward the recrystallizationfront and further into the coarse-granular domainup to around 10. Because there is no correlationbetween the number of measurements and theJopx (Fig. 4b), it is clear that this trend is notdue to a measurement bias, but to strengtheningof the orthopyroxene crystallographic fabric to-ward the coarse-granular domain. The increaseof the Jopx in the coarse-granular peridotites far-ther from the front is related to an enhanced clus-tering of orthopyroxene [001] and [100] axes rela-tive to other coarse-granular peridotites andspinel tectonites (Fig. 5).

In the studied samples, clinopyroxenes (notshown) do not display a well developed LPOeven for the samples in which the orientation ofa relatively large number of grains (s 100) wasmeasured.

4.4. Intrinsic seismic properties of peridotites

The intrinsic seismic properties (i.e. those duethe LPO of minerals) at room pressure and tem-perature of the Ronda peridotites are shown inFig. 6. They have been computed following theprocedure described by Mainprice [31], using themodal composition of the sample, the LPOs andsingle crystal elastic properties of olivine, ortho-

Fig. 5. Orthopyroxene LPO for ¢ve peridotites representativeof the various structural domains de¢ned in the Ronda mas-sif and for which the number of grain measurements waslarge enough to retrieve reliable LPO diagrams. Samples L37and 12 are in the spinel tectonite domain, L26 at the coars-ening front and 33 and 41 are in the coarse-granular domain.Same notations as Fig. 3.

CFig. 6. Seismic properties computed from the LPO of olivine, orthopyroxene and clinopyroxene for the 16 samples from thespinel tectonites and transitional domains (A) and from the coarse-granular domain (B). Data are presented in an equal area low-er hemisphere projection. Column 1 (Vp): 3D distribution of the P-wave velocity; the numbers right and left of the diagram arerespectively the maximum Vp (shown as squares on the plots) and the Vp azimuthal anisotropy. Column 2 (AVs): 3D distribu-tion of the polarization anisotropy of S-waves owing to S-wave splitting. AVs is calculated as 100U(Vs13Vs2)/Vs1. Maximum val-ue of AVs is given on the bottom right of each plot. Column 3 (Vs1-PP): polarization plane of the fast split S-wave (S1) as afunction of the orientation of the incoming wave relative to the structural framework (X, Y, Z ; see Fig. 3 caption) of the sample.Each small segment on the ¢gure represents the trace of the polarization plane on the point at which S1 penetrates the hemi-sphere.

EPSL 5952 27-9-01


EPSL 5952 27-9-01


and clinopyroxenes. We used Voigt^Reuss^Hillaveraging that gives an arithmetic average of theupper and lower bounds [32] and is regarded asthe most reliable [33]. Moreover, in this paper thecomputed seismic velocities and anisotropies aremerely used to compare the properties of por-phyroclastic and coarse-granular peridotites, andthe choice of the averaging procedure has no sig-ni¢cant e¡ect. This procedure allows the predic-tion of the 3D distribution of the velocity of P-waves (Vp) and S-waves (Vs) in the sample, theazimuthal anisotropy of P-waves and the polar-ization anisotropy of S-waves. Comparison ofmeasured and calculated velocities and anisotro-pies (e.g. [34,35]) shows that computation fromLPO of mantle peridotites usually leads to slightlyoverestimated body waves velocity and anisotro-py. Computation overlooks the e¡ect of grainboundaries, mineral defects, and the presence ofminor constituents.

As the olivine LPO controls the intrinsic seis-mic properties of peridotites, it is not surprisingthat seismic properties do not vary signi¢cantlyacross the coarsening front of the Ronda massif.Vp ranges from 8.0 to 8.8 km s31 and the P-waveazimuthal anisotropy from 5.5 to 10% (Fig. 6; Vp

panel). In samples that have an olivine LPO dis-playing an orthorhombic symmetry (e.g. samples2 and 41), the slowest and the fastest P-wavespropagate respectively normal to the foliationand parallel to the lineation (Fig. 6B). The peri-dotites with an olivine [010]-¢ber texture (e.g.samples 27, 12, 19; Fig. 6A) display a broad do-main of high Vp for rays propagating within orclose to the foliation.

The S-waves polarization anisotropy (Fig. 6,panel Vs1-PP) is manifested by the splitting ofan incoming polarized S-wave into two quasi S-waves traveling at di¡erent velocities in the samedirection but polarized in two orthogonal planes(e.g. [9]). In all samples, the highest birefringence(i.e. the di¡erence between the velocity of the fastand slow S-waves) is found in S-waves propagat-ing close to the foliation and at high angle to thelineation. S-waves propagating between 45³ and90³ to the foliation plane display the smallest ani-sotropy. The fast S-wave is systematically polar-ized in a plane containing its propagation direc-

tion and the lineation. Olivine [010]-¢ber texturegenerates a broad girdle of high birefringence par-allel to the foliation plane. In samples with anolivine LPO symmetry intermediate between theorthorhombic and [010]-¢ber texture (e.g. samples28, L26 and L27), the high birefringence girdlecontains a point maximum at high angle to thelineation.

5. Discussion

Our results indicate that during coarsening andpartial melting associated with the propagation ofthe Ronda coarsening front the pre-existing oli-vine LPO of the spinel tectonites was largely pre-served (Figs. 3 and 4a), and that of orthopyroxeneeven enhanced (Figs. 4b and 5). Using the meanJol of lherzolites and harzburgites, and their rela-tive proportion in each tectonic domain, we esti-mate that the volume averaged Jol of spinel tec-tonite domain (Jol = 4.3) is equivalent to that ofcoarse-granular peridotites (Jol = 4.7). Neithercoarsening nor variation of the harzburgite/lherz-olite ratio during the asthenospherization of thelithospheric domain resulted in signi¢cant modi¢-cation of the olivine LPO. We hence concludethat, if not coeval with a new episode of deforma-tion, annealing and partial melting of lithosphericperidotites do not erase or modify signi¢cantly thepre-existing olivine LPO fabric. This result is validat kilometric length-scales relevant to seismic ob-servation, and up to 6.5% of partial melting andT = 1230³C corresponding to the conditions in-ferred for the asthenospherized lithosphere ofthe Ronda massif [19].

5.1. Preservation of lithospheric LPOs in themolten domain

Simultaneous geochemical and microstructuralvariations across the coarsening front (Fig. 2)strongly suggest that annealing-driven coarseningwas somehow enhanced during the onset of melt-ing [17,19]. At the coarsening front, there is aclear decoupling between microstructural evidencefor grain growth (Fig. 2C^F) and the evolution ofthe mineral LPO of peridotites.

EPSL 5952 27-9-01


For olivine our results indicate that graingrowth and partial melting homogenize the initialvariations in LPO strength (Fig. 4) and slightlymodify the symmetry of the LPO pattern (Fig.3). The modi¢cation of the olivine LPO symmetryin the coarse-granular peridotite suggests thatgrain growth was enhanced in grains with [100]axes close to the lineation, at the expenses ofgrains with their [100] axes oriented close to theY-axis (i.e. in the foliation but at high angle to thelineation). Because these latter grains probablydeformed by slip on the `hard' (010)[001] system(e.g. [36]), they may have a higher dislocationdensity and hence a higher intrinsic internal en-ergy that may favor their consumption duringannealing/melting processes.

Despite the steady strengthening of the ortho-pyroxene LPO toward the coarse-granular do-main, observations indicate that it was producedby di¡erent mechanisms at each side of the coars-ening front. The Jopx of spinel tectonites is nega-tively correlated with Sol

v (Fig. 7), whereas forcoarse-granular peridotites it is negatively corre-lated with Lopx

m . The negative correlation with Solv

may indicate that enhancement of orthopyroxeneLPO is connected with the onset of grain growthprocesses just behind the front (Fig. 2). In thecoarse-granular domain, the negative correlationof Lopx

m with Jopx and the decreasing (Sm/Yb)N oflherzolites farther away from the front suggestthat the strengthening of the orthopyroxeneLPO was somehow related to higher extents ofpartial melting.

The e¡ect of annealing on the LPO of rock-forming minerals is still poorly known. Mercierand Nicolas [37] have already suggested that`coarse-tabular' olivine in xenoliths, which usuallydisplay a quite strong LPO, may result from par-tial melting and annealing of peridotites previ-ously deformed at lower temperature. More re-cently, Heillbronner and Tullis [38] haveexperimentally shown that annealing of quartzitedoes not erase the pre-existing LPO of quartz,although the size and shape of crystal are signi¢-cantly modi¢ed. Park et al. [39] have obtainedsimilar results for octachloropropane (C3Cl8)and norcamphor (C7H10O) that are good analogsof crustal rocks.

5.2. Seismic signature of an asthenospherizationfront

We have shown that despite the slight variationin the symmetry of fabric patterns across thecoarsening front, the intrinsic seismic propertiesderived from the di¡erent peridotite fabrics arerather similar in both magnitude and symmetry(Fig. 6). To predict the e¡ect of asthenospheriza-tion of the lithosphere, one needs however to es-timate the seismic properties at the P^T condi-tions of development of the melting front and toassess how they were a¡ected by the presence ofmelt in the coarse-granular domain.

To estimate the average seismic properties ofthe molten domain, we have computed the mean

Fig. 7. Plots of the mean surface area per volume of olivineôlivine interfaces (Sol

v , mm31), the maximum chord parallelto the foliation of orthopyroxene (Lopx

m , mm) versus J-indexof orthopyroxene (Jopx). Symbols as in Fig. 2. The labels onsymbols refer to the sample number.

EPSL 5952 27-9-01


LPO of four coarse-granular peridotites (samples30, 2, 33 and 35; see Fig. 8 caption for furtherdetails). The olivine LPO of this `mean coarse-granular peridotite' is characterized by an ortho-rhombic symmetry, with a relatively weak maxi-mum density of [100], [010] and [001] axes respec-tively parallel to the lineation, normal to thefoliation and normal to the lineation in the folia-tion (Fig. 8a). The orthopyroxene displays a weakLPO with a preferential concentration of [001]close to the X-axis and [100] close to the Z-axis.The inferred seismic properties of the mean sam-ple at the P^T conditions of melting of the litho-spheric domain (1250³C and 1.5 GPa; [19]) areshown in Fig. 8b (see ¢gure caption for furtherdetails on the calculations). When compared tothe seismic properties of the mean sample at 1.5GPa and 750³C (not shown), corresponding to a`normal continental geotherm', the seismic aniso-tropy predicted for both P^T conditions is rathersimilar. The anisotropy of S-waves remains mod-erate (5.6% at 750³C and 5.8% at 1250³C), and itis highest for rays propagating in the foliation

plane in a direction close to the Y-axis (Fig. 8b).The fast propagation direction of P-waves is par-allel to the X-axis, and the azimuthal anisotropyat 750³C and 1250³C is 7.6 and 7.7%, respectively.An excess temperature of 500³C at 50^55 kmwould generate a Vs and Vp anomaly of about33%, quite comparable with the anomalies de-tected in tomography models in domains a¡ectedby thermal erosion of the lithosphere.

Because the instantaneous melt fraction andmelt topology during partial melting are un-known, the e¡ect of the presence of melt on theseismic anisotropy of the coarse-granular perido-tites cannot be assessed quantitatively. Experi-mental studies in partially molten peridotiteswith an initial olivine LPO have demonstratedthat melt tends to be preferentially distributedalong the (010) planes of olivine [40,41]. It ishence conceivable that the preservation of the in-herited olivine LPO in the partially molten do-main led to an anisotropic distribution of melt.As olivine (010) planes in all studied Ronda sam-ples tend to parallel the foliation plane, the melt

Fig. 8. (a) Olivine LPO of a `mean coarse-granular peridotite', which integrates 1690 grain measurements from four samplesfrom the coarse-granular domain (35, 33, 2, and 30; see map on the right for sample location). (b) Estimated seismic propertiesof the `mean coarse-granular peridotite' (1690 olivines, 365 orthopyroxenes and 287 clinopyroxenes) at 1250³C and 1.5 GPa. Forthese calculations, we have used the temperature and pressure derivatives of single crystal elastic constants (Isaak et al. [47] andAbramson et al. [48] for olivine; Chai et al. [49] for orthopyroxene and Collins et al. [50] for clinopyroxene). Notations as inFig. 7. (c) Simpli¢ed map of the western part of the Ronda peridotite showing the olivine stretching lineation and the peridotitefoliation in this area after Darot [23].

EPSL 5952 27-9-01


pockets were likely preferentially oriented parallelto the pre-existing foliation. It has been shownthat the presence of pancake-shaped melt pocketsaligned along the foliation plane increases the ani-sotropy of P- and S-waves without modifying sig-ni¢cantly the orientation of the anisotropy pat-tern relative to the structural framework [42,43].Considering that the averaged ìnstantaneous'melt fraction is less than the melt fraction inferredfrom geochemistry considerations (i.e. 6 6.5% inthe coarse-granular domain; [19]), we concludethat the seismic anisotropy of the partially moltencoarse-granular peridotites was equivalent or evenhigher than that estimated for the melt-absentmean sample. On the other hand, the presenceof a few percents of melts likely decreased signi¢-cantly Vp and Vs in the partially molten peridotitedomain, leading to even larger velocity anomalies(33 to 35%) than those calculated for the melt-absent mean sample.

In the studied area, the structural pattern of theRonda peridotite is rather homogeneous andcharacterized by steeply dipping foliations (60^90³) and subhorizontal lineations (Fig. 8c). Con-sidering this structural pattern and the seismicproperties of the `mean sample', we can predictthe seismic signature of the àsthenospherized'lithospheric domain. As the spinel tectonite do-main, the molten domain would be characterizedby S-wave splitting with fast S-waves polarizedparallel to the structural trend of the massif(Fig. 8c). The delay time between the arrivals ofthe fast and slow split S-waves would be propor-tional to the birefringence and thickness of boththe lithospheric mantle and the portion of moltenlithosphere where the inherited fabric is preserved.Likely, the seismic anisotropy would be enhancedby the presence of melt pockets aligned along theinherited foliation. Altogether, this would lead tosplitting measurements suggesting an anisotropiclayer thicker than the thermal lithospheric mantle.On the other hand, P- and S-wave travel-timetomography would detect a clear negative anom-aly (s 3%) in this intermediate domain, suggest-ing a thin lithosphere in contrast with the thick-ness of anisotropic lithospheric material necessaryto account for the delay time between the fast andslow split S-waves.

6. Implications and conclusions

In several regions of active asthenospheric up-welling, for instance the Yellowstone hot spot [44]and the French Massif Central [45], observationof negative velocity anomalies in tomographymodels suggests a shallow lithosphereâstheno-sphere boundary. On the other hand, seismic ani-sotropy measurements in the same areas yield de-lays between the arrivals of the fast and slow splitS-wave that require a much thicker `seismic litho-sphere' than assumed from tomography models.This paradox might be reconciled in the light ofthe preservation of the pre-existing LPO duringannealing and limited melting as suggested bythe seismic properties of the Ronda peridotites.A plausible scenario in such geodynamic contextis sketched in Fig. 9. We suggest the presence of atransitional layer, a `ghost lithosphere' located in-between the `true' lithosphere and the `true' asthe-nosphere. This `ghost lithosphere' combines litho-spheric and asthenospheric seismic properties. P-and S-wave tomography would detect a negativevelocity anomaly suggesting a shallow astheno-sphere (Fig. 9). S-wave splitting measurements,on the other hand, would probe the `ghost litho-sphere' and detect delay times suggesting a muchthicker anisotropic layer than the thermal `litho-sphere' inferred from seismic tomography. Thethickness of this `ghost lithosphere' remains un-certain but it is likely to extend as far as the initial

Fig. 9. Cartoon illustrating the concept of the seismic `ghostlithosphere' formed during thermal erosion and melting ofthe subcontinental lithosphere by a mantle plume. The por-tion of lithosphere immediately above the plume head wouldbe imaged as having a seismic anisotropy similar to the `nor-mal' lithosphere, and low Vp and Vs similar to those of theunderlying asthenospheric mantle.

EPSL 5952 27-9-01


lithospheric fabric is not totally erased by the on-set of £ow in the new asthenosphere (either large-scale deformation or small convection cells).

Acknowledgements

Xavier Lenoir and Jean-Louis Bodinier haveprovided the samples used in this study and Da-vid Mainprice the programs for the computationof seismic properties. This work has greatly bene-¢ted from discussions with Andrea Tommasi, Da-vid Mainprice, Jean-Louis Bodinier and AdolpheNicolas. Financial support for C.J.G. came froma return fellowship of the Spanish MEC and aMarie-Curie Fellowship of the European Com-munity (MCFI-2000-00639). Thanks are due toD. Mainprice and G. Barruol for their involve-ment in the development of a SEM-EBSD systemat the Laboratoire de Tectonophysique ^ Mont-pellier, France. This development was made pos-sible thanks to the funding by the INSU-CNRS,the University of Montpellier II, the Ìnstitut desSciences de la Terre de l'Eau et de l'Espace deMontpellier' and the National Science Founda-tion (project #EAR-9526840 Ànatomy of an Ar-chean craton').[AC]

References

[1] A. Vauchez, A. Nicolas, Mountain building strike-parallelmotion and mantle anisotropy, Tectonophysics 185 (1991)183^201.

[2] A. Nicolas, N.I. Christensen, Formation of anisotropy inupper mantle peridotites ^ A review, in: K. Fuchs, C.Froidevaux (Eds.), Composition, Structure and Dynamicsof the LithosphereÂsthenosphere System 16, Am. Geo-phys. Union, Washington, DC, 1987, pp. 111^123.

[3] N.I. Christensen, Fabric, seismic anisotropy and tectonichistory of the twin sister dunite, Washington, Geol. Soc.Am. Bull. 82 (1971) 1681^1694.

[4] A. Nicolas, J.P. Poirier, Crystalline Plasticity and SolidState Flow in Metamorphic Rocks, Wiley, London, 1976.

[5] W. BenIsmail, D. Mainprice, An olivine fabric database:an overview of upper mantle fabrics and seismic anisotro-py, Tectonophysics 296 (1998) 145^158.

[6] Q. Bai, S.J. Mackwell, D.L. Kohlstedt, High-temperaturecreep of olivine single crystals. 1. Mechanical results forbu¡ered samples, J. Geophys. Res. 96 (1991) 2441^2463.

[7] M. Kumazawa, O.L. Anderson, Elastic moduli, pressure

derivatives, and temperature derivatives of single-crystalolivine and single-crystal forsterite, J. Geophys. Res. 74(1969) 5961^5972.

[8] N.I. Christensen, The magnitude, symmetry and origin ofupper mantle anisotropy based on fabric analysis of ultra-ma¢c tectonites, Geophys. J. R. Astron. Soc. 76 (1984)89^112.

[9] V. Babuska, M. Cara, Seismic Anisotropy in the Earth,Kluwer, Dordrecht, 1992.

[10] A. Tommasi, A. Vauchez, Continental rifting parallel toancient collisional belts: An e¡ect of the mechanical ani-sotropy of the lithospheric mantle, Earth Planet. Sci. Lett.185 (2001) 199^210.

[11] P.G. Silver, Seismic anisotropy beneath the continents:Probing the depths of geology, Annu. Rev. Earth Planet.Sci. 24 (1996) 385^432.

[12] G. Barruol, P.G. Silver, A. Vauchez, Seismic anisotropyin the eastern United States: Deep structure of a complexcontinental plate, J. Geophys. Res. Solid Earth 102 (1997)8329^8348.

[13] S. Gao, P.M. Davis, H. Liu, P.D. Slack, W. Rigor, Y.A.Zorin, V.V. Mordvinova, V.M. Kozhevnikov, N.A. Lo-gatchev, SKS splitting beneath continental rift zones,J. Geophys. Res. 102 (1997) 22781^22797.

[14] A. Vauchez, G. Barruol, A. Nicolas, Comment on `SKSsplitting beneath rift zones', J. Geophys. Res. 104 (1999)10787^10790.

[15] D. Van der Wal, R.L.M. Vissers, Uplift and emplacementof upper mantle rocks in the western Mediterranean,Geology 21 (1993) 1119^1121.

[16] D. Van der Wal, R.L.M. Vissers, Structural petrology ofthe Ronda Peridotite, SW Spain: deformation history,J. Petrol. 37 (1996) 23^43.

[17] D. Van der Wal, J.L. Bodinier, Origin of the recrystalli-sation front in the Ronda peridotite by km-scale pervasiveporous melt £ow, Contrib. Mineral. Petrol. 122 (1996)387^405.

[18] C.J. Garrido, J.-L. Bodinier, Diversity of ma¢c rocks inthe Ronda peridotite: evidence for pervasive melt/rockreaction during heating of subcontinental lithospheric byupwelling asthenosphere, J. Petrol. 40 (1999) 729^754.

[19] X. Lenoir, C.J. Garrido, J.-L. Bodinier, J.-M. Dautria, F.Gervilla, The recrystallization front of the Ronda perido-tite: thermal erosion and melting of the subcontinentallithospheric mantle beneath the Alboran basin, J. Petrol.(2001), in press.

[20] J.M. Tubia, J. Cuevas, I.J.I. Gil, Sequential developmentof the metamorphic aureole beneath the Ronda perido-tites and its bearing on the tectonic evolution of the BeticCordillera, Tectonophysics 279 (1997) 227^252.

[21] M. Obata, The Ronda peridotite: garnet-, spinel-, andplagioclase-lherzolite facies and the P^T trajectories of ahigh-temperature mantle intrusion, J. Petrol. 21 (1980)533^572.

[22] L. Reisberg, J.-P. Lorand, Longevity of sub-continentalmantle lithosphere from osmium isotope systematics inorogenic peridotite massifs, Nature 376 (1995) 159^162.

EPSL 5952 27-9-01


[23] M. Darot, Methodes d'analyse structurale et cinematique.Application a© l'etude du massif ultrabasique de la SierraBermeja, The©se de 3e©me Cycle, Universite Nantes, 1973.

[24] D. Van der Wal, Deformation processes in mantle peri-dotites, Geol. Ultraiectina 102 (1993) 1^180.

[25] G.E. Lloyd, N.H. Schmidt, D. Mainprice, N.O. Olesen,R.D. Law, M. Casey, Textural determination via SEMelectron channeling, Textures Microstruct. 14^18 (1991)213^218.

[26] R.T. DeHo¡, Stereology and metallurgy, Metal Forum 5(1982) 4^12.

[27] H.E. Exner, Quantitative characterization of microstruc-tural geometry of interfaces, in: J.A. Parks, A.G. Evans(Eds.), Ceramics Microstructure '86. Role of Interfaces,Materials Science Research 21, Plenum Press, NewYork, 1986, pp. 73^86.

[28] H.J. Bunge, Texture Analysis in Material Sciences, Buttle-worth, London, 1982.

[29] A. Tommasi, B. Tiko¡, A. Vauchez, Upper mantle tec-tonics: Three-dimensional deformation, olivine crystallo-graphic fabrics and seismic properties, Earth Planet. Sci.Lett. 168 (1999) 173^186.

[30] D. Mainprice, P.G. Silver, Interpretation of SKS-wavesusing samples from the subcontinental lithosphere, Phys.Earth Planet. Int. 78 (1993) 257^280.

[31] D. Mainprice, A FORTRAN program to calculate seis-mic anisotropy from the lattice preferred orientation ofminerals, Comp. Geosci. 16 (1990) 385^393.

[32] R. Hill, The elastic behavior of a crystalline aggregate,Proc. R. Soc. Lond. A65 (1952) 349^354.

[33] D. Mainprice, M. Humbert, Methods of calculating pet-rophysical properties from lattice preferred orientationdata, Surv. Geophys. 15 (1994) 575^592.

[34] G. Barruol, H. Kern, P and S waves velocities and shearwave splitting in the lower crustal/upper mantle transition(Ivrea Zone). Experimental and calculated data, Phys.Earth Planet. Int. 95 (1996) 175^194.

[35] H. Kern, L. Burlini, I.V. Ashchepkov, Fabric-related seis-mic anisotropy in upper-mantle xenoliths: evidence frommeasurements and calculations, Phys. Earth Planet. Int.95 (1996) 195^209.

[36] W.B. Durham, G. Goetze, Plastic £ow of oriented singlecrystals of olivine. 1. Mechanical data, J. Geophys. Res.82 (1977) 5737^5753.

[37] J.-C. Mercier, A. Nicolas, Textures and fabrics of uppermantle peridotites as illustrated by xenoliths from basalts,J. Petrol. 16 (1975) 454^487.

[38] R. Heillbronner, J. Tullis, E¡ect of static annealing onmicrostructures and CPO of quartzites deformed in axialcompression and shear, in: Deformaton Mechanisms,Rheology and Tectonics, Universiteit Utrecht, Utrecht,2001, p. 68.

[39] Y. Park, J.-H. Ree, S. Kim, Lattice preferred orientationin deformed-then-annealed material: observations fromexperimental and natural polycrystalline aggregates, Int.J. Earth Sci. 90 (2001) 127^135.

[40] H.S. Wa¡, U.H. Faul, E¡ects of crystalline anisotropy on£uid distribution in ultrama¢c partial melts, J. Geophys.Res. 97 (1992) 9003^9014.

[41] M.J. Daines, D.L. Kohlstedt, In£uence of deformation onmelt topology in peridotites, J. Geophys. Res. 102 (1997)10257^10271.

[42] D. Mainprice, Modeling the anisotropic seismic propertiesof partially molten rocks found at mid-ocean ridges, Tec-tonophysics 279 (1997) 161^179.

[43] A. Vauchez, A. Tommasi, G. Barruol, J. Maumus, Uppermantle deformation and seismic anisotropy in continentalrifts, Phys. Chem. Earth 25 (2000) 111^117.

[44] D. Schutt, D. Humphreys, K. Dueker, Anisotropy of theYellowstone hot spot wake, Eastern Snake river plain,Idaho, Pure Appl. Geophys. 151 (1998) 443^462.

[45] M. Granet, A. Glahns, U. Achauer, Anisotropic measure-ments in the Rhinegraben area and the French MassifCentral, Pure Appl. Geophys. 151 (1998) 333^364.

[46] S.-S. Sun, W.F. McDonough, Chemical and isotopic sys-tematics of oceanic basalts: implications for mantle com-position and processes, in: A.D. Saunders, M.J. Norry(Eds.), Magmatism in the Ocean Basins 42, GeologicalSociety Special Publication, 1989, pp. 313^345.

[47] D.G. Isaak, O.L. Anderson, R.E. Cohen, The relationshipbetween shear and compressional velocities at high pres-sures: reconciliation of seismic tomography and mineralphysics, Geophys. Res. Lett. 19 (1992) 741^744.

[48] E.H. Abramson, M. Brown, L.J. Slutsky, J. Zaug, Theelastic constants of San Carlos olivine up to 17 GPa,J. Geophys. Res. 102 (1997) 12252^12263.

[49] M. Chai, J.M. Brown, L.J. Slutsky, The elastic constantsof an aluminous orthopyroxene to 12.5 GPa, J. Geophys.Res. 102 (1997) 14779^14785.

[50] M.D. Collins, J.M. Brown, Elasticity of an upper mantleclinopyroxene, Phys. Chem. Minerals 26 (1998) 7^13.

EPSL 5952 27-9-01


Documents

Seismic properties of an asthenospherized lithospheric ... · Seismic properties of an asthenospherized lithospheric mantle: constraints from lattice preferred ... coarse-granular