Lithos 113 (2009) 179–189

Contents lists available at ScienceDirect

Lithos

j ourna l homepage: www.e lsev ie r.com/ locate / l i thos

Mantle wedge asymmetries and geochemical signatures along W- and E–NE-directedsubduction zones

Carlo Doglioni a,⁎, Sonia Tonarini b, Fabrizio Innocenti b,c

a Dipartimento di Scienze della Terra, Università La Sapienza, Rome, Italyb CNR-Istituto di Geoscienze e Georisorse, Pisa, Italyc Dipartimento di Scienze della Terra, Università di Pisa, Pisa, Italy

⁎ Corresponding author.E-mail address: carlo.doglioni@uniroma1.it (C. Dogli

0024-4937/$ – see front matter © 2009 Elsevier B.V. Aldoi:10.1016/j.lithos.2009.01.012

a b s t r a c t

a r t i c l e i n f o

Article history:Received 27 March 2008Accepted 26 January 2009Available online 8 February 2009

Keywords:Mantle wedgeSubduction zoneWestward driftB and Nd isotopes

Subduction zone kinematics predict that, assuming a fixed lower plate, the velocity of the subduction equalsthe velocity of the subduction hinge (Vs=−Vh). In all subduction zones the subduction hinge migrates towardthe lower plate. However, two main types of subduction zones can be distinguished: 1) those where theupper plate converges toward the lower plate slower than the subduction hinge (mostly W-directed), and2) those in which the upper plate converges faster than the subduction hinge (generally E- or NE-directed).Along the first type, there generally is an upward flow of the asthenosphere in the hanging wall of the slab,whereas along the opposite second type, the mantle is pushed down due to the thickening of the lithosphere.The kinematics of W-directed subduction zones predict a much thicker asthenospheric mantle wedge, largervolumes and faster rates of subduction with respect to the opposite slabs. Moreover, the larger volumes oflithospheric recycling, the thicker column of fluids-rich, hotter mantle wedge, all should favour greatervolumes of magmatism per unit time. The opposite, E–NE-directed subduction zones show a thinner, if any,asthenospheric mantle wedge due to a thicker upper plate and shallower slab. Along these settings, themantle wedge, where the percolation of slab-delivered fluids generates melting, mostly involves the coolerlithospheric mantle. The subduction rate is smaller, andesites are generally dominant, and the lithospherethickens, there appears to be a greater contribution to the growth of the continental lithosphere.Another relevant asymmetry that can be inferred is the slab-induced corner flow in the mantle alongW-directed subduction zones, and an upward suction of the mantle along the opposite E- or NNE-directedslabs. The upward suction of the mantle inferred at depth along E–NE-directed subduction zones provides amechanism for syn-subduction alkaline magmatism in the upper plate, with or without contemporaneousrifting in the backarc. Positive δ11B and high 143Nd/144Nd characterizeW-directed subduction zones where athicker and hotter mantle wedge is present in the hanging wall of the slab. However, this observationdisappears where large amounts of crustal rocks are subducted as along the W-directed Apennines sub-duction zone.

© 2009 Elsevier B.V. All rights reserved.

1. Introduction

The mantle wedge (Fig. 1) is the triangular section of the mantleconfined between the top of the slab and the base of the upper plate(e.g., vanKeken, 2003;Wienset al., 2008). It is generallyconsidered tobecomposed of asthenosphere, although some authors also include theentire lithospheric mantle section above the slab. The mantle wedgefilters fluids released by the slab that melt the overlying mantle (Aberset al., 2006), and feed arcmagmatism (Tatsumi et al.,1983; Syracuse andAbers, 2006). Themantlewedge is usually conceived as a relatively “hot”body, where the melting feeding the magmatic arc can take place

oni).

l rights reserved.

(N1200 °C?), boundedby lower temperatures at the inclinedbase (top ofthe slab) and the top (base of the lithosphere?). The transit and locationof melting areas into thewedge have been identified bymagnetotelluricor electrical conductivity studies (Brasse et al., 2002; Brasse, 2005).

The mantle wedge is therefore a crucial area for plate tectonics,where relevant chemical transfer occurs and newmaterial is producedand added to the crust. What happens in the mantle wedge can beinferred from seismic tomography, geochemistry of lavas andxenoliths, plus other indirect information such as gravimetric andgeoelectrical studies. In the Tonga backarc basin, themantlewedge hasbeen seismically illuminated showing a series of well-beddedreflectors, indicating a form of stratified architecture (Zheng et al.,2007). Martinez and Taylor (2002) proposed an eastward flow in themantle wedge to compensate for slab rollback, this flow beingdistorted by the corner flow associated with the subduction. These

Fig. 1. The mantle wedge is the triangular section of mantle in the hanging wall ofsubduction zones. It is considered as the source for the magmatic arc, being percolatedand metasomatized by the fluids delivered by the dehydration of the descending slab.Relative to the upper plate, the subduction hinge can diverge or converge. Thekinematics of the hinge is a good indicator on the mantle wedge geometry. The legendin the figure indicates the range of values of the main parameters.

180 C. Doglioni et al. / Lithos 113 (2009) 179–189

authors suggest that this flow system can explain the geochemicalasymmetries in the backarc spreading and the slab-related volcanicfront. Moreover, shear wave splitting analyses often indicate trench-orthogonal direction in the backarc basin, and trench-parallel directionin the forearc (e.g., Levin et al., 2004). This has been interpretedsignifying that deformation in the mantle causes lattice-preferredorientation (LPO), which in turn affects the directional dependence ofseismic wave velocity (Kneller et al., 2005). Based on shear-wavesplitting analysis, trench-parallel ultra-fast velocities (500 mm/year)have been measured in the Tonga mantle wedge (Conder and Wiens,2007), consistently with the “eastward” mantle flow implicit in thenet-rotation, or “westward”drift of the lithosphere (Gripp andGordon,2002; Scoppola et al., 2006; Doglioni et al., 2007), although the superfast velocities of Tonga would favour the faster net rotation inferred inthe shallow hotspot reference frame (Crespi et al., 2007).

The mantle wedge is considered as a section with higher mantletemperature (anomalies up to 400–600 °C, Koper et al., 1999), rich influids released by the downgoing slab (Billen and Gurnis, 2001; Abers,2005; Grove et al., 2006; Panza et al., 2007a,b; Peccerillo et al., 2008),and marked by low velocity of the seismic waves (Conder and Wiens,2006), and lower viscosity. In the literature, themantlewedge is mostlyundifferentiated, with variations related to the thickness and composi-tion of the upper and lower plates. However, profounddifferences occur,for example when comparing the mantle wedge of the western versusthe eastern Pacific subduction zones (Plank and Langmuir, 1988), orwhen comparing the Apennines (W-directed slab) and the Alps (SE-directed slab), (e.g., Peccerillo, 2005; Panza et al., 2007a,b). Multiplesubduction components even within a single mantle wedge have beenproposed in the arc magmatism of the central and southern Americasubduction zone (Hickey-Vargas et al., 2002; Tonarini et al., 2007).

It has also been noted that backarc spreading must be part of amantle flow associated with the mantle wedge. Therefore, significantdifferences in themantlewedge shouldoccuras a functionofwhetherornot there is active backarc spreading (Ribe, 1989; Conder et al., 2002;Wiens et al., 2008). Since backarc spreading generally forms along W-directed subduction zones (e.g., Doglioni et al., 2007), we contribute inthis article some kinematic and geochemical ideas that support theconcept of an asymmetry in the mantle wedge as a function of thesubductionpolarity. Basedon thedata compilation inWinter (2001), theW-directed subduction-related volcanic arcshave in general lowerK,Na,Al2O3 content, and higher FeO, MgO, CaO/MgO with respect to the

opposite subduction zones. This may be ascribed to the thinner column(if any) of continental crust percolated by melts along W-directedsubduction zones. In fact, even along W-directed zones, K can be veryabundant when continental lithosphere occurs both in the lower and inthe upper plate (e.g., the central Apennines, Peccerillo, 2005).

2. Subduction asymmetry

Subduction zones can be analyzed in terms of a wide range ofparameters, such as convergence rate, topographic and structuralelevation of the related orogen, subsidence rate in the trench or foredeep,erosion rate, metamorphic evolution, magmatism, dip of the forelandmonocline, depth and geometry of the decollement planes that generatethe accretionary prism and the belt of the upper plate, the thickness andcomposition of the upper and lower plates, gravity, magnetic and heatflow anomalies, seismicity and slab dip. Therefore, there is a long list ofparameters, which are relevant to the geometry and evolution of eachparticular subduction zone. However, since the lithosphere has a net-rotation relative to the mantle (the so-called “westward” drift, e.g., LePichon,1968; Bostrom,1971), subduction zones appear to be sensitive tothis polarization, that is not E–W, but along an undulated flow that hasthe pole of rotation displaced about 30° with respect to the Earth'srotation pole (Crespi et al., 2007). Therefore, two main different classescanbedistinguished as a function of the subductionpolarity, i.e., in favouror against the westerly polarized tectonic mainstream that depicts thepredominant direction of plate motion (Doglioni et al., 1999, 2007).

Indeed, subduction zones directed to thewest (Barbados, Apennines,Marianas, Tonga) showanumberof commoncharacteristics, suchas lowtopography and low structural elevation, a deep trench or foredeepwithhigh subsidence rates, generally a steep slab, an accretionary prismmostly composed by the shallow rocks of the lower plate and aconjugate backarc basin. In contrast, subduction zones directed to theeast (e.g., Andes) or north-east (Himalayas, Zagros) exhibit oppositesignatures such as high structural and morphological elevation,generally no backarc basin, shallower trench or foredeep with lowersubsidence rate, deeply rooted thrust planes affecting the whole crustand lithospheric mantle, ultra-high pressure rocks andwide outcrops ofmetamorphic rocks, and dominantly shallower dip of the slab. Basaltsand less evolved lavas are more typical along W-directed subductionzones, whereas andesites are abundant along the opposite E- or NE-directed subduction zones (Andes, Indonesia arc).

All these asymmetries have generally been interpreted as related tothe older age of the subducting lithosphere alongW-directed subductionzones; however, they occur regardless the age and composition of thesubducting slab, being more sensitive to the geographic polarity of thesubduction (Doglioni et al.,1999; Cruciani et al., 2005; Lenci andDoglioni,2007). A study on the consequences of these two end members on themantle wedge has not yet been carried out. The westward drift of thelithosphere should affect the nature and geometry of the mantle wedge,such as the tensional or compressional tectonic regime in the upper plate,the dip of the slab, and the composition and thickness of the upper plate,all parameters that seem chiefly dictated by the subduction polarity.

3. Slab–mantle kinematics

The subduction hinge is a helpful indicator of the kinematics andnature of subduction zones. The behaviour of the subduction hingecan be studied either relative to the upper plate (Fig. 1), or the lowerplate (Fig. 2), or relative to the mantle. It has been shown thatsubduction zones have rates faster or slower than the convergencerate as a function of whether the subduction hinge migrates away ortoward the upper plate (Doglioni et al., 2007). When the subductionhinge moves toward the upper plate a double verging orogen forms,whereas if the subduction moves away from the upper plate, a singleverging, low-elevation prism and a backarc basin form. We presenthere a further simple kinematic analysis of the subduction system

Fig. 2. Kinematics of subduction zones assuming fixed the lower plate L. The upper plateU converges at Vu=80mm/year in both cases. The transient location of the subductionhinge H moves at Vh=100 mm/year and Vh=20 mm/year respectively. The resultingsubduction S is given by Vs=−Vh and therefore is 100mm/year in the upper panel, and20 mm/year in the lower panel. The shortening in the orogen (lower panel), is Vu−Vh.Values are only as an example. S increases when H diverges relative to the upper plate(above), whereas S decreases if H converges relative to the lower plate (below). Theupper example is accompanied by backarc spreading and mantle upwelling, a lowprism, and it is typical of W-directed subduction zones. The lower example is rathercharacterized by lowering of the mantle, double verging and elevated orogen. It formsfrequently along E- to NNE-directed subduction zones. The velocity of the hinge equalsthe velocity of the subduction in both cases. In the upper case the subduction isindependent of the upper plate velocity, whereas in the lower case it is a function of it.These opposite kinematic settings indicate different dynamic origin of the subduction,i.e., slab/mantle interaction for the upper section, and upper/lower plates interaction forthe lower case. The two cases support two end members of the mantle wedge, verythick for the upper case and very thin for the lower case.

181C. Doglioni et al. / Lithos 113 (2009) 179–189

assuming fixed the lower plate L (Fig. 2). Assuming the upper plate Uconverging at a fixed rate (Vu=80mm/year), the subduction hinge Hismoving faster (Vh=−100mm/year) or slower (Vh=−20mm/year)than the upper plate. In both cases the subduction hinge converges orrolls back toward the lower plate. In this reference system, thesubduction rate Vs is equal to −Vh, i.e., the velocity of the subductionhinge H relative to L. In the first case, the subduction rate is notcontrolled by the convergence of the upper plate. In the second case,the subduction rate is instead function of the convergence rate,decreased by the amount of shortening of the upper plate, which iscontrolled by the viscosity of the lithosphere (Doglioni et al., 2006).The two cases show that 1) subduction zones have two differentorigins, 2) the mantle in the hanging wall is either uplifted or pusheddown as a function of the prevailing mechanism, 3) the subductionrate can be either faster or slower than the convergence rate, 4) thedifferent behaviour and origin of the two end members suggest thepassive role of subduction zones in plate tectonics, 5) where thesubduction hingemigrates faster than the upper plate toward the lowerplate, the subduction is controlled only by the slab–mantle flowrelationship (W-directed subduction zones); in case the hinge migratesslower than the upper plate toward the lower plate (E- or NE-directedsubduction zones), the subduction is controlled by the upper–lowerplates convergence rate, plus density, thickness and viscosity of theupper and lower plates.

Another relevant kinematic observation relates to themotion of theslabs in the mantle reference frame. In this reference, the subductionhinge should be fixed relative to the mantle along W-directed sub-duction zones, whereas it should be W- or SW-ward retreating alongE- or NE-directed subduction zones. However, this reference frameremains problematic because some authors consider the opening of

the backarc basin as a reliable indicator of the subduction hinge speed(e.g., Lallemand et al., 2005), while other separate the speed of thebackarc spreading from the subduction hinge velocity as a function ofthe accretion in the prism or the asthenospheric intrusion at thesubduction hinge (Doglioni, 2008). However, assuming the hotspotreference frame as valid (Gripp and Gordon, 2002), regardless if thevolcanism is sourced by the deep or shallow mantle (Cuffaro andDoglioni, 2007), the E- or NE-directed subduction zones are remount-ing in the mantle, i.e., moving toward the direction opposite to that ofthe subduction. In this case the subduction occurs because the upperplate overrides faster the lower plate (Doglioni et al., 2007).

The fastest global plate velocity is along the Tonga subduction zone(Bevis et al., 1995), where the subduction hinge converges relative tothe lower plate at 240 mm/year as a minimum. This high speedimplies that 1) the subduction can reach the base of the upper mantlein about 2.5Ma; 2) the fast slab retreat implies a similar velocity in themantle compensating the volume left by the slab. This is consistentwiththe shear-wave splitting analysis of Conder and Wiens (2007) whopropose a WNW–ESE mantle flow in the backarc, shifting to a directionabout NNE–SSW parallel to the Tonga subduction direction whenapproaching the slab, and having a super fast rate of about 500mm/year.According to the aforementioned kinematics, W-directed subductionzones have the fastest descending slabs, providing i) about three timeslarger volumes of lithospheric recycling into the mantle than theopposite subduction zones and ii) the larger amount of fluids releasedinto the mantle wedge.

Moreover, along W-directed subduction zones, the accretionaryprisms are smaller because the basal decollement plane is travelling atshallow level on top the lower plate (Lenci and Doglioni, 2007) and theupper plate inmostly in extensional regime.Along theopposite E- orNE-directed subduction zones, the basal decollement layers are deeper andaffecting the entire lithosphere. The accretionary prism is rathercomposed by shortening of the upper plate, whereas the lower plate ismorewidely affected during the collision stage. These orogenic systemsprovide minor loss of material due to subduction, but more accretiondue to themore deeply rooted thrust plates. Moreover they exhibit largeplutonic emplacement. All this generates a larger fluxofmaterial for thegrowth of the continental crust and related lithospheric mantle.

In summary, the analysis of the subduction hinge both in the upperand in the lower plate reference frame confirms the existence of twoend members of subduction style and related orogen, regardless if themantle reference frame is accepted or not.

4. Mantle wedge geometry

The source of arc magmatism is mainly within the mantle wedge,triggered by fluids released by the downgoing slab. The estimateddepth of the top of the slab beneath volcanic arcs ranges between 72and 173 km, being the mean depth projection at around 105 km(Syracuse and Abers, 2006). We have seen that the steeper the slaband thinner the upper plate, the thicker is the resulting mantle wedge.The shallower the slab and the thicker the upper plate, the thinner isthe mantle wedge. These two end-members are in general associatedwith subduction zones inwhich the slab hingemigrates away from theupper plate, or toward the upper plate, respectively. In the lower platereference frame, the classification applies to cases where the slabhinge migrates toward the lower plate faster or slower than the upperplate. The two end members typically apply to W-directed and theE- or NE-directed subduction zones.

As noted in the previous section, the W-directed subduction zonesare generally faster because they have the subduction hinge generallymoving awaywith respect to the upper plate, and converging relative tothe lower plate faster than the upper plate. Therefore these subductionzones should supply much larger volumes to mantle recycling than theopposite subduction zones (Fig. 3). The total budget of fluids releasedby the slab should then be also greater. Moreover, the mantle wedge of

Fig. 3. The main differences between orogens are a function of the subduction polarityalong the tectonic mainstream reported in the map above, which appears polarized dueto the relative “eastward” mantle counterflow. The volumes recycled along W-directedsubduction zones are about 2–3 times higher than along the opposite settings due to theaforementioned kinematic constraints. Moreover, the asthenospheric wedge aboveslabs is much thicker along W-directed subduction zones (AW) with respect to the E–NEdirected subductions, if any (AE), modified after Doglioni et al. (2007).

182 C. Doglioni et al. / Lithos 113 (2009) 179–189

W-directed subduction should be hotter because it involves a thickersection of asthenosphere that is generally assumed N1300 °C. This canexplain why it has lower P and S seismic velocity and very low Q factorwith respect to the E- or NNE-directed subduction zones (Fig. 4).

W-directed subduction zones show peculiar gravimetric signa-tures in which there is a much larger positive anomaly in the upperplate with respect to the opposite E- or NE-directed subductionsettings. Free-air gravity anomaly is in average about 75 mgal in thehanging wall of W-directed subduction zones, whereas it is about50mgal along the E- or NE-directed subduction zones (Harabaglia andDoglioni, 1998). This can be explained by the shallower presence of the

Fig. 4. Comparison between the Vp tomography of the Tonga and Andean subduction zones.and Wiens, 2006) with respect to the opposite Andean setting (Heit, 2005).

asthenosphere in the hanging wall of W-directed slabs, and is alsosupported by the higher heat flow. For example, the heat flow (Hurtiget al., 1992) in the Tyrrhenian Sea is much higher (N200–250 mW)than in the Aegean rift (b100–120 mW), or in the western Pacificbackarc basins than in the Andean Cordillera (Pollack and Chapman,1977). Along the Chilean Andes, the top of the slab beneath thevolcanic arc is between 90 and 110 km (England et al., 2004; Syracuseand Abers, 2006). However, the thickness of the continental litho-sphere along the same belt is estimated between 100 and 125 km(Artemieva and Mooney, 2001; Gung et al., 2003). These two valuesprevent the existence of an appreciable thickness of themantle wedgeof asthenospheric origin along this section of the Cordillera. Thereforethe so-called mantle wedge will be primarily composed by the upperplate continental lithospheric mantle, possibly with an averagetemperature lower than 1300 °C (Fig. 5).

There are few transitional cases between the two end members, i.e.,those inwhich a rifting occurs in the hangingwall of E- or NNE-directedsubduction zones (e.g., theAegean, Andaman, andBasinandRange). Thefirst two are syn-subduction rifts, whereas the third is post-subduction.However there are profound differences between backarc spreadingassociated with W-directed or to E–NE-directed subduction zones (e.g.,Doglioni,1995). The first type occurswhen the retreat of the lithosphereleaves an empty volume, which is replaced by the asthenosphere. Theupper plate extension along the opposite E- or NE-directed subductionzones has another origin and cannot be compared to the W-directedsettings. The Aegean and Andaman rifts have for example a number ofdifferences from the classic backarc setting. The rift is concentrated onlyin limited areas in the hangingwall of the subduction,whereas along theW-directed subduction zones the backarc rift is ubiquitous in the upperplate. The rift (e.g., Aegean, Innocenti et al., 2005) forms where thehanging wall lithosphere is split into two independently advancingplates, like the Andaman rift located at the transition between theslower Himalayas convergence (34–38 mm/year) and the fasterIndonesian subduction zone arc (around 60–64 mm/year). Moreover,this type of rift has a slower rate of spreading with respect to theopposite setting. The W-directed subduction rift-related backarc ispervasively distributed throughout the whole upper plate, and is fast(usually N0.6 km/Ma subsidence rate, N10 km/Ma spreading rate),arriving at oceanization in few million years. It may develop with orwithout convergence between the upper and lower plates, and it cansimply be explained as being related to slab retreat relative to the upperplate. The retreat occurs regardless its origin, i.e., the slab pull or theeastward mantle flow implicit in the westward drift of the lithosphere.

Summing up the differences in polarity of the subduction withrespect to the undulate tectonic mainstream of plate motion (Doglioni

Note the much slower velocities in the mantle wedge of the Tonga subduction (Conder

Fig. 5. The four cases represent the possible settings ofW-directed and E- or NE-directed subduction zones as a function of the composition of the lower plate (oceanic or continental).The magmatism volume should be controlled by the slab dehydration, the asthenospheric wedge thickness and the subduction rate. The asthenospheric wedge thickness increaseswith the slab dip and decreases with the upper plate thickness. The thickestmantle wedge of asthenosphere is along theW-oceanic case, where the slab is steeper and the upper plateis young oceanic lithosphere. A slightly thinner asthenospheric wedge occurs in W-continental, where it is expected a shallower melting of the lower plate. The steep slab along theW-cases is controlled by the negative buoyancy (if any) and the advancing mantle flow. Along the E-oceanic, widespread volcanism forms; the upper plate along the E-oceanic canalso be oceanic (usually older than the lower plate). The thinnest asthenosphere should occur along the E- or NE-continental example, where in fact there is the lowest amount ofvolcanism. The shallow slab dip of the E-cases is controlled by low negative buoyancy (if any), and the sustaining mantle flow. AOC, altered oceanic crust.

183C. Doglioni et al. / Lithos 113 (2009) 179–189

et al., 1999; Crespi et al., 2007) and the “westward” drift of thelithosphere relative to the mantle (Scoppola et al., 2006), the mantlewedge varies as a function of the polarity of the subduction, plus thethickness and composition of the upper and lower plates (Fig. 5).

The upward flow of the mantle in the hanging wall of a subductionzone is controlled by the amount of rifting in the upper plate. However,the backarc width and its spreading rate are determined by few con-comitant parameters, i.e., the migration rate of the subduction hingeaway from the upper plate, minus the amount of accretionary prismgrowth and/or the mantle intrusion at the subduction hinge (Doglioni,2008).

Therefore, backarc spreading forms in two settings: 1) alongW-directed subduction zones where the basin opens as the astheno-sphere replaces the retreated lithosphere, and 2) along the E–NE-directed subduction inwhich the upper lithosphere is split into two sub-plates that have a differential advancement velocity relative to the lowerplate. AlongW-directed subduction zones, the hinge diverges relative tothe upper plate, and the backarc spreading is given by the rate of hingeretreat, minus the volume of the accretionary prism, or, in case of scarceor no accretion,minus the volume of the asthenospheric intrusion at thesubduction hinge. Since the volume of the accretionary prism is

proportional to the depth of the decollement plane, the backarc riftingis inversely proportional to the depth of the decollement (Doglioni et al.,2007).

Fig. 5 shows four simplified cartoons of the possible settings ofW-directed and E- or NE-directed subduction zones as a functionof the composition of the lower plate. The volume of magmatismshould be controlled by slab dehydration, asthenospheric wedgethickness and subduction rate. The asthenospheric wedge thicknessdecreases with the shallowing of slab dip and with the increase of theupper plate thickness. The thickest wedge of asthenosphere is alongthe W-directed subduction zones where the slab tends to be verysteep, and the lower and upper plates are oceanic. The steep slab alongthe W-cases is controlled by the negative buoyancy (if any), and theadvancing mantle flow. Along the E-directed subduction zones, wherethe lower plate is oceanic, and the convergence rate is high (N3 cm/year?) widespread volcanism occurs. The thinnest asthenosphereshould be present along the E- or NE-continental example, where infact there is the lowest amount of volcanism (e.g., Himalayas). Theshallow slab dip of the E-cases is controlled by low negative buoyancy(if any), and the E–NE-ward relative mantle flow, which should rathersustain the slab (Fig. 5).

Fig. 6. Schematic representation showing the 143Nd/144Nd and δ11B evolution of hydrousfluids; isotope variations associated to crustal contamination are also shown. AOC =altered oceanic crust.

184 C. Doglioni et al. / Lithos 113 (2009) 179–189

5. Chemical fluxing in subduction zones: boron isotope evidence

The differences between W-ward and E–NE-ward directed sub-duction zones is likely to be apparent in the geochemistry of theassociated magmas. Plank and Langmuir (1988) identified two endmembers of mantle wedge, thick and thin. They suggested the twoend members represent a short and a long melting column, respec-tively, plus a number of differing geochemical signatures such as theNa6 content, increasing with the short column. Metallogenesis alsoappears to be controlled by subduction style (Mitchell and Garson,1981). Porphyry copper deposits are concentrated in collisionalsettings and Chilean type subduction zones. Mariana type subductionis instead characterized by Kuroko or similar volcanogenic sulphidedeposits (Nishiwaki and Uyeda, 1983).

Understanding the relations between slab outputs (e.g., aqueousfluid, silicate melt, supercritic fluid; see Abers, 2005; Kessel et al.,2005; Hermann et al., 2006), their ascending into the mantle wedgeand the physical controls of the “subduction factory” are complicatedby the large number of parameters involved (see above). For example,the addition of slab-derived fluids and possibly melts to the mantlewedge explains many geochemical characteristics of arc magmas.However, the geochemical variability of subduction related magmas(regionally, and sometimes within single volcanoes) suggests acomplex interplay between lithologic heterogeneities of subductedmaterials, metamorphic reactions, and element mobility duringdevolatilization and concomitant mass transfer. Hydrothermal fluids,derived frommetamorphic dehydration reactions may account for themobilization of FME (Fluid-Mobile Elements) into arc magma sourceregions (Noll et al., 1996). In cross-arc transects, the decline of FME/fluid immobile element ratios from the volcanic arc towards the back-arc regions, was interpreted as due to the increase of the slab depthand thus to the increment of the metamorphic grade in subductionassociations (Ryan and Langmuir,1993; Ryan et al., 1996). On the otherhand, based on a compilation of published data and detailed studies ofoceanicmagmatic andmantle rocks, Scambelluri and Philippot (2001)and Chalot-Prat et al. (2003) have shown that metamorphism up tothe eclogite facies is isochemical and that, contrary to what it isusually assumed, the fluids are not extracted during HP-LT meta-morphism, even within the eclogite facies, up to 13 kbar. Bebout(2007) suggests that H2O-rich fluids are the dominant fluids phasereleased by dehydration of sediments and basalts in forearc regions,whereas supercritical fluids and hydrous silicate melts appear to bemore important at higher P and T beneath the volcanic fronts andacross the arcs.

Table 1Boron isotopic variations in the major geochemical reservoirs.

Material δ11B‰ Reference

Range

MORB −10.5 −1 Roy-Barman et al., 1998−7 −1 Chaussidon and Marty, 1995−7.6 −1.8 LeRoux et al., 2005

OIB −15 −1 Chaussidon and Marty, 1995−14 −6 Gurenko and Chaussidon, 1997−6 0 Tanaka and Nakamura, 2005−3 −1 Tonarini et al., 2005−7.6 −2 Turner et al., 2007

Continental crust −13 −7 Chaussidon and Albarède, 1992Kasemann et al., 2000

Altered oceanic crust −4 17 Spivack and Edmont, 1987Smith et al., 1995

Pelagic sediments −15 5 Ishikawa and Nakamura, 1993Metasediments −10 −5 You et al., 1995Forearc serpentinites 5 21 Benton et al., 2001Serpentinites 5 15 Bonatti et al., 1984

Boschi et al., 2007

Boron and B isotopes are, perhaps, the most successful geochem-ical tracers of hydrous fluids migrating from slab into the overlyingmantle wedge. Boron is released from downgoing slabs earlier andmore efficiently than other elements (e.g., Rb, K, Ba) and its abundancesuggests a close relationship to H2O-rich fluids. It is high soluble inaqueous fluids, thus it is easily decoupled from less soluble elements(REE, Rare Earth Elements, and HFSE, High Field Strength Elements)via separation of fluid phases (Leeman and Sisson, 1996 and referencetherein). Boron isotopes exhibit a large range in isotopic compositions(δ11B, permil variation with respect to the 11B/10B ratio of NIST-SRM951 standard, varies between −30 to +50‰, Palmer and Swihart,1996). Table 1 summarizes B isotope characteristics of major geochem-ical reservoirs at subduction zones. The δ11B shows large variability inthe mantle products (with MORB, mid-ocean ridge basalts in the range−1 to−10.5‰ and OIB, ocean island basalts, between−1 and−15‰;the continental crust has an average δ11B of about−10.5‰, Chaussidonand Albarède, 1992; Kasemann et al., 2000). Altered oceanic crust(AOC) is enriched in seawater derived B and its isotopic composition isbetween +17 and −4‰ (Smith et al., 1995). In marine sediments B ispresent in two components: a desorbable one, (δ11B=+15‰, Spivackand Edmont, 1987) largely lost at relatively low temperature (You et al.,1995); and a structurally bonded one (from +2 to −15‰). Anotherimportant boron reservoir is represented by serpentinized perido-tites of the oceanic lithosphere (δ11B≈+5/+15‰, Boschi et al.,2007), where B is hosted by serpentine. Serpentinites, generated inthe forearc position and sampled in the Mariana forearc, are char-acterized by an average of B and δ11B values of 27 ppm and +14‰,respectively (Benton et al., 2001).

δ11B exhibits large cross-arc and inter-arc variations in volcanic arcs,like the other fluid mobile elements. Redistribution of boron andfractionation of its isotope species is temperature dependent, henceshallow dehydration of the slab yields fluids with isotopically heavyboron, since 11B is preferentially partitioned in the fluid phase, leavingthe residual slab progressively more 11B-depleted (e.g., Benton et al.,2001). Nevertheless, studies on trace elementmobility stress the impacton the boron budget of somemetamorphicminerals such as phengite inthe subducting slabs (Marschall et al., 2007). B is very efficientlyabsorbed by phengite-free assemblage, whereas its concentration influids released from phengite-bearing rocks is significantly lower.Moreover, the δ11B offluids releasedduring thefirst steps of dehydrationare positive, while fluids generated at greater depth, as slab dehydrationproceeds, rapidly reach negative values (Marschall et al., 2007).

The B isotope composition appears to be relatively insensitive tocrustal contamination processes during magma ascent. Studies on

Fig. 7. Worldwide distribution of 143Nd/144Nd vs. δ11B in volcanic lavas from the mainsubduction zones. W-directed subduction zones have more positive δ11B and higher143Nd/144Nd, apart the Apennines subduction zone, which is strongly contaminated bycrustal rocks. Data sources are reported in Table 2.

185C. Doglioni et al. / Lithos 113 (2009) 179–189

continental arc-volcanism (in the Central Andes, Western Anatoliaand Phlegrean Volcanic District, Rosner et al., 2003; Tonarini et al.,2004, 2005) where crustal contamination is well documented,indicate a small variation in Sr and Nd isotope compositions,suggesting a relatively constant degree of contamination. By contrast,δ11B shows a wide range of values that is plausibly linked to slabcomponents, suggesting that B can be used as tracer for subductionprocesses not only in oceanic settings but also in continental envi-ronments. Also, interaction of calc-alkaline magmas with lowercontinental crust is unlikely to modify their δ11B because boron is

Table 2Mean and extreme B and Nd isotope compositions in arc lavas.

143Nd/144Nd

Rock types Mean SD

Western-directedPacific arcsKamchatka C-A basalts to dacites 0.51308 0.00003Kurili C-A basalts to dacites 0.51306 0.00005Izu C-A basalts to rhyolites; HMg andesites 0.51306 0.00004Marianna C-A basalts to dacites 0.51299 0.00002Halmahera C-A basalts to dacites 0.51291 0.00010

Mediterranean arcsAeolian C-A basalts to dacites 0.51267 0.00011Phlegrean Volcanic District C-A ktrachybasalts to latites 0.51254 0.00001North Apennines C-A rhyolites 0.51211 0.00010

Atlantic arcsSSI LK tholeiites to dacites 0.51302 0.00004Martiniquea C-A basalts to dacites 0.51287 0.00012

Eastern-directedPacific arcsCascades LK tholeiites-HKCA basalts 0.51291 0.00005El Salvador, Central America C-A basalts to dacites 0.51299 0.00002Andes C-A basaltic andesites to dacites 0.51242 0.00003

Mediterranean arcsWestern Anatolia C-A bas-andesites to latites 0.51244 0.00004

1: Ishikawa et al., 2001 and reference therein; 2: Ishikawa and Tera, 1997 and reference therand reference therein; 5: Palmer, 1991; 6: Tonarini et al., 2001; 7: D'Antonio et al., 2007, 8:10: Smith et al., 1997 and reference therein; 11: Leeman et al., 2004; 12: Tonarini et al., 200SD = standard deviation; n = number of samples.

a Exclused strongly crustal contaminated samples.

systematically mobilized during prograde metamorphism, thus gen-erally the lower crust appears to be significantly depleted in B(Leeman et al., 1992).

Summing up, geochemical behaviour of B is related to fluid phasesso that it is decoupled with respect to elements barely mobile in thefluid but sensitive to crustal contamination like Nd (Fig. 6). Generallyspeaking, we may expect that in warm subducting slabs such as in theE-directed slabs, the higher heat produces shallow release of aqueousfluids rich in 11B and high 143Nd/144Nd, whereas at greater depth(beneath the volcanic arcs), it is likely that hydrous silicate melts and/or supercritical liquids (Kessel et al., 2005) with δ11B negative and low143Nd/144Nd ratio, are released (Fig. 6). On the other hand, in coldersubducting slabs, the lower heat allows deeper slab dehydration andprobably large volatile flux below the arcs (Fig. 6), although the lowertemperature favours larger shear heating. In these cases, the δ11B forarc magma suites are consistent with progressive loss of B andspecifically 11B (low δ11B) whereas Nd again maintains almost thesame isotopic compositions of its source (altered oceanic crust andlithosphere). One caveat is that the forearc regions (metasomatizedby shallow aqueous fluids strongly enriched in 11B) may be involvedin magma genesis. Such forearc materials, dragged downward byconvective mantle flow may release large volume of water (as well asdissolved FME) to the mantle beneath the volcanic arcs (90–130 kmdepths; Hattori and Guillot, 2003) giving to the arc magmas a heavyboron signature (Straub and Layne, 2002; Tonarini et al., 2007).Leeman et al. (2002) suggest that boron enrichment and isotopesignature in arc magmas generally show strong correlations withphysical parameters such as subduction zone geometry, convergencerate, slab dip, slab thermal state and lithosphere thickness, but theyare poorly correlated with the estimates of sediment flux given by Reaand Ruff (1996).

B and Nd isotope data from arc lavas are plotted in Fig. 7; the meanand the extreme isotope values in W- and E-directed subductionzones are reported in Table 2. Most of arc lavas from Pacific andAtlantic W-directed subduction zone are characterized by high Nd

δ11B‰

n Max value Min value Mean SD n Max value Min value Reference

6 0.51312 0.51305 1.5 2.7 12 5.6 −3.7 127 0.51315 0.51299 3.1 2.6 25 5.9 −3.8 29 0.51312 0.51299 4.9 2.3 9 7.3 1.2 3

14 0.51301 0.51296 5.0 0.7 19 6.2 4 414 0.51305 0.51269 0.6 2.1 8 3.5 −2.3 5

22 0.51285 0.51243 −2.4 2.4 22 2.3 −6.1 622 0.51273 0.51245 −7.6 1.6 23 −3.6 −10.6 74 0.51225 0.51205 −6.3 1.7 4 −4.3 −6 8

15 0.51304 0.51297 15.1 1.6 15 17.6 12 98 0.51302 0.51667 1.0 3.0 7 5.8 −3.2 10

9 0.51300 0.51286 −6.3 1.7 10 −0.4 −9.8 1117 0.51301 0.51295 3.5 2.6 17 6.29 −2.7 1217 0.51251 0.51225 −0.4 2.1 17 4.1 −7.2 13

10 0.51249 0.51237 −8.5 5.0 10 −0.1 −14.6 14

ein; 3: Ishikawa and Nakamura, 1994 and reference therein; 4: Ishikawa and Tera, 1999Tonarini et al., 2003 and reference therein; 9: Tonarini and Leeman, unpublished data;7; 13: Rosner et al., 2003; 14: Tonarini et al., 2005.

186 C. Doglioni et al. / Lithos 113 (2009) 179–189

isotope compositions whereas the δ11B values show awide range from+18‰ (SSI, Tonarini et al., 2006) to −2.3 (Halmahera arc, Palmer,1991). Other negative values are found in back-arc lavas from Kuriliand Kamtchacka (Ishikawa and Tera,1997,1999; Ishikawa et al., 2001).However, the negative δ11B values are few and the average δ11B ispositive (see Table 2). E-directed Pacific subduction zones arecharacterized by products with a wide range in both Nd and B isotopecompositions. In particular, the higher δ11B values are found in the ElSalvador (maximum value 6.3‰, Tonarini et al., 2007), whereasnegative values are common in the Cascades (minimumvalues−9.8‰,Leeman et al., 2004); values lower than −5‰ were also found inCentral Andes and Southern Andes (Rosner et al., 2003; Tonarini et al.,2006). The W-directed Mediterranean arcs show Nd and B isotoperatios significantly lower with respect to their Pacific counterparts,suggesting involvement of sediments and crustal materials (Fig. 7).The two isotope ratios show large variations and are positively cor-related. The western Anatolia is the only NE-directed subduction zoneof the Mediterranean area for which δ11B values are published;δ11B values are as low as −15‰ whereas the Nd isotope data is al-most constant. The W-directed subduction zone of the southernand northern Apennines is characterized by negative δ11B (Table 2);Nd isotope ratios clearly support the presence of crustal componentsas expected for the subduction of the Adriatic continental litho-sphere and/or acquired through crustal contamination during magmaupraise.

Comparing the δ11B against 87Sr/86Sr (not shown), the scenario issimilar to that illustrated in Fig. 7, however some differences can bepointed out. Sr is a relativelymobile element in the aqueous fluid, thusa weak positive correlation between δ11B and 87Sr/86Sr is sometimesobserved in arc lavas (e.g. El Salvador, Tonarini et al., 2007). Whencrustal materials are involved in arc magma genesis, a negativecorrelation between δ11B and Sr isotope ratios is observed, as forexample in W-directed Mediterranean arcs (D'Antonio et al., 2007).

6. Discussion

Independent of the westward drift, there are at least two basicreasons why the E- or NE-directed subduction zones (e.g., centralAmerica, Andes, Alps, Dinarides, Zagros, Himalayas, Indonesia) have athinner mantle wedge: 1) they commonly have a thick continentalupper plate (e.g., Panza, 1980; Panza et al., 1982; Artemieva andMooney, 2001; Panza et al., 2003; Manea et al., 2004), and 2) the slabis on average less inclined (e.g., Cruciani et al., 2005). In the Andes theslab depth beneath the volcanic arc can be even thinner than the

Fig. 8. The subduction zones disturb or deviate the general “eastward” flow of the mantle reslab should rather generate an upward suction flow from the underlying mantle. Such suctioOIB-type magmatism (e.g. Patagonia back-arc basalts, Bruni et al., 2008). The fluids releassubduction) decrease the viscosity at the top of the asthenosphere, speeding up the upperaway from the lower plate along W-directed subduction zones, facilitating the backarc sprewhere the westward increase of the upper plate velocity rather favours the convergence betworogen. BABB: back-arc basin basalts; IAT: island-arc tholeiites; CA, SHO: calc-alkaline and shhinge; L, lower plate; U, upper plate. The arrows of L and H refer to fixed U, not to the man

thickness of the upper plate continental lithosphere. This suggestseither a very thin asthenosphere between the slab and the upperplate, or even the absence of the asthenosphere, being the mantlewedge composed only by lithospheric mantle. Moreover, from thekinematics of subduction zones and the behaviour of the slab hinge,the E- or NE-directed slabs have slower sinking velocity than theopposite W-directed subduction zones. Tatsumi and Eggins (1995)have shown a correlation between convergence rate and volumes ofmagmatism along subduction zones. The larger volumes of subduc-tion predicted along W-directed slabs should favour the formationof a greater amount of arc-related magma, and generation of largevolumes of oceanic crust in the backarc setting for a number ofreasons: 1) the larger subduction rate should generate also moreabundant slab dehydration, lowering the melting temperature; 2) thethicker and hotter (asthenospheric) mantle wedge should have athicker column of potential melting; 3) faster slab enteringmeans alsolarger shear heating.

Subduction zones trigger a corner flow in the host mantle (e.g.,Turcotte and Schubert, 2002). However, the aforementioned asym-metric kinematics of subduction zones would predict that thismechanism might be valid only for W-directed subduction, whereasalong the opposite settings, the mantle should be flowing upward, orsucked by the “westward” motion of the lithosphere relative to themantle (Fig. 8). In this reconstruction, at a global scale the mantlewould be flowing eastward relatively to the lithosphere, generating afirst order flow. Subduction and rift zones are then a second orderturbulence disturbing the main flow. Along W-directed subductionzones, slab retreat is compensated by the asthenosphere in thebackarc, but it determines a downgoing corner flow in the hostmantle.Conversely, along E- or NE-directed subduction zones, in general thereis no void to fill in the backarc setting. However, the slab is“remounting” relative to the mantle, and a suction flow from belowis expected (Fig. 8). In fact, relative to the mantle, the lower plate ismovingwestwardout of themantle, in the direction opposed to the dipof the slab. The subduction occurs because the upper plate is movingwestward faster than the lower plate. It is noteworthy that a number ofsubduction zones are characterized by alkaline magmatism in theforeland of the retrobelt or within the orogen itself (e.g., MineralnieVodi in the northern Caucasus; Euganei Hills, in the Southern Alps;Patagonia; Aegean Sea, etc.). This magmatism may be related to anupward suction of the mantle due to the opposite slab motion, asillustrated in Fig. 8. In this model, the high velocity body recorded bytomography along E-directed subduction zones (e.g., see the Andean“slab” of Fig. 4) could be interpreted not as subducted lithosphere (in

lative to the lithosphere. W-directed slabs produce a corner flow, whereas the oppositen could trigger “fertile”mantle from below, and its decompression may locally generateed by both W- and E-directed slabs (e.g., the white lenses in the hanging wall of theplate. The fluids are sheared by the lithospheric decoupling and determine a migrationading. It triggers an opposite behaviour along the E- or NE-directed subduction zoneseen the upper and lower plates, i.e., determining a double verging Andean-Alpine type

oshonitic series; OIB-type: basalts with ocean island or intraplate affinity. H, subductiontle.

187C. Doglioni et al. / Lithos 113 (2009) 179–189

fact it ismostly aseismic), but as deepermantle upraised by the suctionmechanism. The deeper, more viscous and rigid mantle has a morecompacted crystallographic structure and higher seismic velocities.

The fluids released from the slab into the overlying mantle shouldtrigger a decrease of viscosity in the hanging wall of the subduction, atthe bottom of the upper plate. The asthenosphere top is the maindecollement surface of the lithosphere, and the decrease of theviscosity can accelerate the relative decoupling. Therefore the upperplate increases its velocity moving away from the lower plate alongthe W-directed subduction zones. This facilitates the formation of thebackarc spreading (Fig. 8). Along the E- or NE-directed subductionzones, the upper plate is converging faster with respect to the lowerplate, facilitating the generation of double verging orogens such as theAndes or Himalayas.

7. Conclusions

The kinematics and the resulting geometries along subductionzones would indicate that: 1) the mantle wedge is different as afunction of the subduction polarity, being thicker and mostly astheno-spheric in the hanging wall of W-directed slabs; conversely, it isthinner and mostly composed by lithospheric mantle along the E- orNE-directed subduction zones; 2) the subduction rate is dependent onthe slab–mantle interaction along W-directed subduction zones,independent of the movement of the upper plate; conversely, thesubduction rate is controlled by the far field velocity of plates and thevelocity of the hinge along E- or NE-directed subduction zones; thisobservation indicates a different origin of the opposite subductionsystems as a function of the geographic polarity (Fig. 2). Moreover itsupports a passive behaviour of the slab, rather than it being theprimary force in driving plate tectonics. This information contains agenetic and geographic asymmetry in the origin of subduction zones,regardless the mantle (hotspot) reference frame. The B and Nd isotopesseem to confirm the asymmetry of subduction zones (Fig. 7). Hotterand thicker asthenosphere in the hanging wall of W-directedsubduction zones is generally accompanied by positive and higherδ11B and 143Nd/144Nd, except where there is a significant crustalcontribution (e.g., Apennines).

We suggest that mantle wedge thickness, composition and tem-perature are affected by the asymmetries imposed by the westwarddrift of the lithosphere (Fig. 3) and the consequent differences amongsubduction zones (Fig. 5). This flow can be interpreted as the firstorder flowbetween lithosphere andmantle, whereas in this kinematicinterpretation, the subduction and rift zones are secondary turbu-lences (Fig. 8). While it has been widely accepted a corner flow inmantle around subduction zones, it is here proposed that thismechanism occurs dominantly along W-directed subduction systems,whereas it is the opposite along the E- or NE-directed subductionzones (Fig. 8), where plates move W-ward or SW-ward relative to themantle, in a direction opposite to that of the subduction (Cuffaro andDoglioni, 2007). If this is true, the motion of the slab should generatean upward flow of the mantle beneath the subduction zone, suckingup the mantle and depressurising it (Fig. 8). This mechanism upraisesthemantle from deeper levels, which has faster seismic velocities withrespect to themantle at shallower depths. This could explain the ghostof a slab beneath E–NE-directed subduction zones even without thepresence of a real slab, and the absence of continuous seismicity.

This could in turn facilitate partial melting and alkaline magma-tism in “backarc” settings, with orwithout extensional tectonics, alongthe retrobelt of orogens associated with E- or NE-directed subductionzones. Release of fluids from the slab to the asthenosphere speeds upthe decoupling at the base of the upper plate. However, this generatesopposite tectonic consequences in the relationship between upperand lower plates along subduction zones. In fact, due to the W-warddrift of the lithosphere, it facilitates the widening between the upperand the lower plates along W-directed subduction zones, whereas it

promotes the convergence along the opposite E–NE-directed subduc-tion systems (Fig. 8).

Acknowledgements

The day after this article was accepted, Fabrizio Innocenti passedaway. Fabrizio has been a precious mentor and a close friend of usduring the years. We are very grateful to him for the endless numberof constructive discussions together. We will miss his widespreadculture, his fine sense of humour, his generosity, and his enthusiasmfor life, science and company. Many thanks to Yildirim Dilek forinviting us to contribute to this volume, and to Andrew Kerr for hiseditorial and linguistic help. Thanks also to Françoise Chalot-Prat andan anonymous referee for the critical reading and criticisms.Discussions with Samuele Agostini, Enrico Bonatti, Eugenio Carminati,Marco Cuffaro, Giuliano Panza and Federica Riguzzi were very fruitful.This research was supported by the Sapienza University.

References

Abers, G.A., 2005. Seismic low-velocity layer at the top of subducting slabs beneathvolcanic arcs: observations, predictions, and systematics. Physics of the Earth andPlanetary Interiors 149, 7–29. doi:10.1016/j.pepi.2004.10.002.

Abers, G.A., van Keken, P.E., Kneller, E.A., Ferris, A., Stachnik, J.C., 2006. The thermalstructure of subduction zones constrained by seismic imaging: implications for slabdehydration and wedge flow. Earth and Planetary Science Letters 241, 387–397.

Artemieva, I.M., Mooney, W.D., 2001. Thermal structure and evolution of Precambrianlithosphere: a global study. Journal of Geophysical Research 106, 16,387–16,414.

Bebout, G.E., 2007. Metamorphic chemical geodynamics of subduction zones. Earth andPlanetary Science Letters 260, 373–393.

Benton, L.D., Ryan, J.G., Tera, F., 2001. Boron isotope systematics of slab fluids as inferredfrom a serpentine seamount, Mariana forearc. Earth and Planetary Science Letters187, 273–282.

Bevis, M., Taylor, F.W., Schutz, B.E., Recy, J., Isacks, B.L., Helu, S., Singh, R., Kendrick, E.,Stowell, J., Taylor, B., Calmant, S., 1995. Geodetic observations of very rapidconvergence and back-arc extension at the Tonga arc. Nature 374, 249–251.

Billen, M.I., Gurnis, M., 2001. A low viscosity wedge in subduction zones. Earth andPlanetary Science Letters 193, 227–236.

Bonatti, E., Lawrence, J.R., Morandi, N., 1984. Serpentinization of oceanic peridotites:temperature dependance of mineralogy and boron content. Earth and PlanetaryScience Letters 70, 88–94.

Boschi, C., Dini, A., Früh-Green, G., Kelley, D.S., 2007. Isotopic and element exchange duringserpentinization andmetasomatism at the Atlantis Massif (MAR 30°N): insights fromB and Sr isotope data. Geochimica et Cosmochimica Acta 72, 1801–1823.

Bostrom, R.C., 1971. Westward displacement of the lithosphere. Nature 234, 536–538.Brasse, H., 2005. The mantle wedge in the Bolivian orocline in the view of deep

electromagnetic soundings. ISAG, International Symposium on Andean Geodynamics,Barcelona.

Brasse, H., Lezaeta, P., Rath, V., Schwalenberg, K., Soyer, W., Haak, V., 2002. The BolivianAltiplano conductivity anomaly. Journal of Geophysical Research 107 (B5), 2096.doi:10.1029/2001JB000391.

Bruni, S., D'Orazio, M., Haller, J.M., Innocenti, F., Manetti, P., Pécskay, Z., Tonarini, S.,2008. Time-evolution of magma sources in a continental back-arc settino: theCenozoic basalts from Sierra de san Bernardo (Patagonia, Chubut, Argentina).Geological Magazine. 145, 714–732. doi:10.1071/S0016756808004949.

Chalot-Prat, F., Ganne, J., Lombard, A., 2003. No significant element transfer from theoceanic plate to the mantle wedge during subduction and exhumation of the Tethyslithosphere (Western Alps). Lithos 69, 69–103.

Chaussidon, M., Albarède, F., 1992. Secular boron isotope variations in the continentalcrust: an ion microprobe study. Earth and Planetary Science Letters 108, 229–241.

Chaussidon, M., Marty, B., 1995. Primitive Boron isotope composition of the mantle.Science 269, 383–386.

Conder, J.A., Wiens, D.A., 2006. Seismic structure beneath the Tonga arc and Lau back-arc basin determined from joint Vp, Vp/Vs tomography. Geochemistry, Geophysics,and Geosystems 7, Q03018.

Conder, J.A., Wiens, D.A., 2007. Rapid mantle flow beneath the Tonga volcanic arc. Earthand Planetary Science Letters 264, 299–307.

Conder, J.A., Wiens, D.A., Morris, J., 2002. On the decompression melting structure atvolcanic arcs and back-arc spreading centers. Geophysical Research Letters 29, 15.doi:10.1029/2002GL015390.

Crespi, M., Cuffaro, M., Doglioni, C., Giannone, F., Riguzzi, F., 2007. Space geodesyvalidation of the global lithospheric flow. Geophysical Journal International 168,491–506. doi:10.1111/j.1365-246X.2006.03226.x.

Cruciani, C., Carminati, E., Doglioni, C., 2005. Slab dip vs. lithosphere age: no directfunction. Earth Planetary Science Letters 238, 298–310.

Cuffaro, M., Doglioni, C., 2007. In: Foulger, G., Jurdy, D. (Eds.), Global Kinematics in theDeep vs Shallow Hotspot Reference Frames. The Origins of Melting Anomalies:Plates, Plumes, and Planetary Processes, vol. 430, pp. 359–374.

D'Antonio, M., Tonarini, S., Arienzo, I., Civetta, L., Di Renzo, V., 2007. In: Beccaluva, L.,Bianchini, G., Wilson, M. (Eds.), Components and Processes in the Magma Genesis

188 C. Doglioni et al. / Lithos 113 (2009) 179–189

of the Phlegrean Volcanic District, Southern Italy. Cenozoic Volcanism in theMediterranean Area, vol. 418. Geological Society of America, pp. 203–220.

Doglioni, C., 1995. Geological remarks on the relationships between extension andconvergent geodynamic settings. Tectonophysics 252 (1–4), 253–267.

Doglioni, C., 2008. Comment on “The potential influence of subduction zone polarityon overriding plate deformation, trench migration and slab dip angle” by W.P.Schellart. Tectonophysics 463, 208–213. doi:10.1016/j.tecto.2008.02.012.

Doglioni, C., Harabaglia, P., Merlini, S., Mongelli, F., Peccerillo, A., Piromallo, C., 1999.Orogens and slabs vs theirdirection of subduction. Earth Science Reviews 45,167–208.

Doglioni, C., Carminati, E., Cuffaro, M., 2006. Simple kinematics of subduction zones.International Geology Review 48 (6), 479–493.

Doglioni, C., Carminati, E., Cuffaro, M., Scrocca, D., 2007. Subduction kinematics anddynamic constraints. Earth Science Reviews, 83, pp. 125–175. doi:10.1016/j.earscirev.2007.04.001.

England, P., Engdahl, R., Thatcher, W., 2004. Systematic variation in the depths of slabsbeneath arc volcanoes. Geophysical Journal International 156, 377–408.

Gung, Y., Panning, M., Romanowicz, B., 2003. Global anisotropy and the thickness ofcontinents. Nature 422, 707–711.

Gurenko, A.A., Chaussidon, M., 1997. Boron concentrations and isotopic composition ofthe Icelandic mantle: evidence from glass inclusions in olivine. Chemical Geology135, 21–34.

Gripp, A.E., Gordon, R.G., 2002. Young tracks of hotspots and current plate velocities.Geophysical Journal International 150, 321–361.

Grove, T.L., Chatterjee, N., Parman, S.W., Médard, E., 2006. The influence of H2O onmantle wedge melting. Earth and Planetary Science Letters 249 (1–2), 74–89.doi:10.1016/j.epsl.2006.06.043.

Harabaglia, P., Doglioni, C., 1998. Topography and gravity across subduction zones.Geophysical Research Letters 25 (5), 703–706.

Hattori, K.H., Guillot, S., 2003. Volcanic fronts form as consequence of serpentinitedehydration in the forearc mantle wedge. Geology 31, 525–528.

Heit, B. 2005. Teleseismic tomographic images of the Central Andes at 21°S and 25.5°S.FU Berlin, Digitale Dissertation, http://www.diss.fu-berlin.de/2005/319/.

Hermann, J., Spandler, C., Hack, A., Korsakov, A.V., 2006. Aqueous fluids and hydrousmelts in high-pressure and ultra-high pressure rocks: implications for elementtransfer in subduction zones. Lithos 92, 399–417.

Hickey-Vargas, R., Sun, M., López-Escobar, L., Moreno-Roa, H., Reagan, M.K., Morris, J.D.,Ryan, J.G., 2002. Multiple subduction components in the mantle wedge: evidencefrom eruptive centers in the Central Southern volcanic zone, Chile. Geology 30,199–202.

Hurtig, E., Cermak, V., Haenel, R., Zui, V.,1992. Geothermal Atlas of Europe. GeoForschungsZentrum Potsdam. Publication 1.

Innocenti, F., Agostini, S., Di Vincenzo, G., Doglioni, C., Manetti, P., Savascin, Y., Tonarini,S., 2005. Neogene and Quaternary volcanism in Western Anatolia: magma sourcesand geodynamic evolution. Marine Geology 221, 397–421.

Ishikawa, T., Nakamura, E., 1993. Boron isotope systematics of marine sediments. Earthand Planetary Science Letters 152, 123–138.

Ishikawa, T., Nakamura, E., 1994. Origin of the slab component in arc lavas from across-arc variation of B and Pb. Nature 370, 205–208.

Ishikawa, T., Tera, F., 1997. Source composition and distribution of the fluid in the Kurilemantle wedge: constraints from across-arc variations of B/Nb and B isotopes. Earthand Planetary Science Letters 152, 123–138.

Ishikawa, T., Tera, F., 1999. Two isotopically distinct fluid components involved in theMariana arc: evidence from Nb/B ratios and B, Sr, Nd and Pb isotope systematics.Geology 27, 83–86.

Ishikawa, T., Tera, F., Nakazawa, T., 2001. Boron isotope and trace element systematics ofthe three volcanic zones in the Kamchacta arc. Geochimica et Cosmochimica Acta65, 4523–4537.

Kasemann, S., Erzinger, J., Franz, G., 2000. Boron recycling in the continental crust of thecentral Andes from the Paleozoic to Mesozoic, NW Argentina. Contributions toMineralogy and Petrology 140, 328–343.

Kessel, R., Schmidt,M.W.,Ulmer, P., Pettke, T., 2005. Trace element signature of subduction-zone fluids,melts and supercritical liquids at 120–180 kmdepth. Nature 437, 724–727.

Kneller, E.A., van Keken, P.E., Karato, S.-I., Park, J., 2005. B-type olivine fabric in themantle wedge: insights from high-resolution non-Newtonian subduction zonemodels. Earth and Planetary Science Letters 237, 781–797.

Koper, K.D., Wiens, D.A., Dorman, L., Hildebrand, J., Webb, S., 1999. Constraints on theorigin of slab and mantle wedge anomalies in Tonga from the ratio of S to Pvelocities. Journal of Geophysical Research 104 (7), 15,089–15,104.

Lallemand, S., Heuret, A., Boutelier, D., 2005. On the relationships between slab dip, back-arc stress, upper plate absolute motion and crustal nature in subduction zones.Geochemistry, Geophysics, andGeosystems 6, Q09006. doi:10.1029/2005GC000917.

Leeman,W.P., Sisson, V.B.,1996. Geochemistry of boron and its implications for crustal andmantle processes. In: Grew, E.S., Anovitz, L.M. (Eds.), Boron:Mineralogy, PetrologyandGeochemistry. Reviews in Mineralogy & Geochemistry, vol. 33, pp. 709–744.

Leeman,W.P., Sisson, V.B., Reid, M.R., 1992. Boron geochemistry of lower crust: evidencefrom granulite terrenes and deep crustal xenoliths. Geochimica et CosmochimicaActa 56, 775–788.

Leeman, W.P., Huang, S., Tonarini, S., 2002. Comparative thermal models for subductionzones: evaluation of slab contributions to volcanic arc magmas. Geochimica etCosmochimica Acta 66, A444.

Leeman, W.P., Tonarini, S., Chan, L.H., Borg, L.E., 2004. Boron and lithium isotopevariations in a hot subduction zone—the southern Washington Cascades. ChemicalGeology 212, 101–124.

Lenci, F., Doglioni, C., 2007. On some geometric prism asymmetries. In: Thrust belts andForeland Basins: From Fold Kinematics to Hydrocarbon Systems. In: Lacombe, O.,Lavé, J., Roure, F., Verges, J. (Eds.), Frontiers in Earth Sciences. Springer, pp. 41–60.

Le Pichon, X., 1968. Sea-floor spreading and continental drift. Journal of GeophysicalResearch 73 (12), 3661–3697.

LeRoux, P.J., Dixon, J., Shrey, S.B., Hauri, E.H., 2005. Boron isotope composition of SouthAtlantic MORB and mantle sources. Geochimica et Cosmochimica Acta 69, A94.

Levin, V., Droznin, D., Park, J., Gordeev, E., 2004. Detailed mapping of seismic anisotropywith local shear waves in southeastern Kamchatka. Geophysical Journal Interna-tional 158, 1009–1023.

Manea, V.C., Manea, M., Kostoglodov, V., Currie, C.A., Sewell, G., 2004. Thermal structure,coupling and metamorphism in the Mexican subduction zone beneath Guerrero.Geophysical Journal International 158, 775–784.

Marschall, H.R., Altherr, R., Rüpke, L., 2007. Squeezing out the slab-modelling the releaseof Li, Be and B during progressive high-pressure metamorphism. Chemical Geology239, 323–335.

Martinez, F., Taylor, B., 2002. Mantle wedge control on back-arc crustal accretion. Nature416, 417–420. doi:10.1038/416417a.

Mitchell, A.H.G., Garson, M.S., 1981. Mineral Deposits and Global Tectonic Settings.Academic Press, London.

Nishiwaki, C., Uyeda, S., 1983. Accretion tectonics and metallogenesis. In: Hashimoto,M., Uyeda, S. (Eds.), Accretion Tectonics in the Circum–Pacific Regions, pp. 349–355.

Noll, P.D., Newsom, H.E., Leeman, W.P., Ryan, J.G., 1996. The role of hydrothermal fluidsin the production of subduction zone magmatism: evidence from siderophile andchalcophile trace elements and boron. Geochimica et Cosmochimica Acta 60,587–611.

Palmer, M.P., 1991. Boron isotope systematics of Halmahera arc (Indonesia) lavas:evidence for involvement of subducted slab. Geology 19, 215–217.

Palmer, M.P., Swihart, G.H., 1996. Boron isotope geochemistry: an overview. In: Grew, E.S.,Anovitz, L.M. (Eds.), Boron: Mineralogy, Petrology and Geochemistry. Reviews inMineralogy & Geochemistry, vol. 33, pp. 709–744.

Panza, G.F., 1980. Evolution of the Earth's lithosphere. In: Davies, P.A., Runcorn, S.K. (Eds.),Mechanisms of Continental Drift and Plate Tectonics. Academic Press, pp. 75–87.

Panza, G.F., Mueller, S., Calcagnile, G., Knopoff, L., 1982. Delineation of the north centralItalian upper mantle anomaly. Nature 296, 238–239.

Panza, G.F., Pontevivo, A., Chimera, G., Raykova, R., Aoudia, A., 2003. The Lithosphere–Asthenosphere: Italy and surroundings. Episodes 26 (3), 169–174.

Panza, G.F., Peccerillo, A., Aoudia, A., Farina, B., 2007a. Geophysical and petrologicalmodelling of the structure and composition of the crust and upper mantle incomplex geodynamic settings: the Tyrrhenian Sea and surroundings. Earth ScienceReviews 80, 1–46.

Panza,G., Raykova,R.B., Carminati, E., Doglioni, C., 2007b.Uppermantleflow in thewesternMediterranean. Earth and Planetary Science Letters 257, 200–214. doi:10.1016/j.epsl.2007.02.03.

Peccerillo, A., 2005. Plio-Quaternary volcanism in Italy. Petrology, Geochemistry,Geodynamics. Springer, Heidelberg. 365 pp.

Peccerillo, A., Panza, G.F., Aoudia, A., Frezzotti, M.L., 2008. Relationships betweenmagmatism and lithosphere–asthenosphere structure in the western Mediterra-nean and implication for geodynamics. Rendiconti Lincei 19, 291–309. doi:10.1007/s12210-008-0020-x.

Plank, T., Langmuir, C.H., 1988. An evaluation of the global variations in the major elementchemistry of arc basalts. Earth and Planetary Science Letters 90 (4), 349–370.

Pollack, H.N., Chapman, D.S., 1977. On the regional variation of heat flow, geotherms,and lithospheric thickness. Tectonophysics 38, 279–296.

Rea, D.K., Ruff, L.J., 1996. Composition and mass flux of sediment entering the world'ssubduction zones: implications for global sediment budgets, great earthquakes, andvolcanism. Earth and Planetary Science Letters 140, 1–12.

Ribe, N.M., 1989. Mantle flow induced by back arc spreading. Geophysical JournalInternational 98, 85–91.

Rosner, M., Erzinger, J., Franz, G., Trumbull, R.B., 2003. Slab-derived boron isotopesignatures in arc volcanic rocks from the Central Andes and evidence for boronisotopic fractionation during progressive slab dehydration. Geochemistry, Geophy-sics, and Geosystems 4 (8). doi:10.1029/2002GC000438.

Roy-Barman, M., Wassemburg, G.J., Papanastassiou, D.A., Chaussidon, M., 1998. Osmiumisotopic compositions and Re–Os concentrations in sulfite globules from basalticglasses. Earth and Planetary Science Letters 154, 331–347.

Ryan, J.G., Langmuir, C.H., 1993. The systematics of boron abundances in young volcanicrocks. Geochemica et Cosmochemica Acta 57, 1489–1498.

Ryan, J.G., Morris, J., Bebout, G., Leeman, W.P., 1996. In: Bebout, G.E. (Ed.), DescribingChemical Fluxes in Subduction Zones: Insights From “Depth-Profiling” Studies of Arcand Forearc Rocks. Monograph. American Geophysical Union, Washington, p. 96.

Scambelluri, M., Philippot, P., 2001. Deep fluids in subduction zones. Lithos 55, 213–227.Scoppola, B., Boccaletti, D., Bevis, M., Carminati, E., Doglioni, C., 2006. The westward

drift of the lithosphere: a rotational drag? Bulletin of the Geological Society ofAmerica 118 (1/2), 199–209. doi:10.1130/B25734.1.

Smith, H.J., Spivack, A.J., Staudigel, H., Hart, S.R., 1995. The boron isotopic composition ofaltered oceanic crust. Chemical Geology 126, 119–135.

Smith, H.J., Leeman, W.P., Davidson, J., Spivack, A.J., 1997. The B isotopic composition ofarc lavas from Martinique, Lesser Antilles. Earth and Planetary Science Letters 146,303–314.

Spivack, A.J., Edmont, J.M., 1987. Boron isotope exchange between seawater and theoceanic crust. Geochimica et Cosmochimica Acta 51, 1033–1044.

Straub, S.M., Layne, G.D., 2002. The systematics of boron isotopes in Izu arc frontvolcanic rocks. Earth and Planetary Science Letters 198, 25–39.

Syracuse, E.M., Abers, G.A., 2006. Global compilation of variations in slab depth beneatharc volcanoes and implications. Geochemistry, Geophysics, and Geosystems 7 (5),Q05017. doi:10.1029/2005GC001045.

Tanaka, R., Nakamura, E., 2005. Boron isotopic constrains on the source of Hawaiianshield lavas. Geochimica et Cosmochimica Acta 69, 3385–3399.

189C. Doglioni et al. / Lithos 113 (2009) 179–189

Tatsumi, Y., Eggins, S., 1995. Subduction Zone Magmatism. Blackwell Science, Cam-bridge, U.K. 211 pp.

Tatsumi, Y., Sakuyama, M., Fukuyama, H., Kushiro, I., 1983. Generation arc basaltmagmas and thermal structure of the mantle wedge in subduction zones. Journal ofGeophysical Research 88, 5815–5825.

Tonarini, S., Leeman, W.P., Ferrara, G., 2001. Boron isotopic variations in lava flows of theAeolian volcanic arc, Italy. Journal of Volcanology and Geothermal Research 110,155–170.

Tonarini, S., Forte, C., Petrini, R., Ferara, G., 2003. Melt/biotite 11B/10B isotopicfractionation and the boron local environment in the structure of volcanic glasses.Geochimica et Cosmochimica Acta 67, 1863–1873.

Tonarini, S., Leeman,W.P., Civetta, L., D'Antonio, D., Ferrara, G., Necco, A., 2004. B/Nb andd11B systematics in the Phlegrean Volcanic District, Italy. Journal of Volcanologyand Geothermal Research 133, 123–139.

Tonarini, S., Agostini, S., Innocenti, F., Manetti, P., 2005. d11B as tracer of slabdehydration and mantle evolution in western Anatolia Cenozoiv magmatism. TerraNova 17, 259–264.

Tonarini, S., Leeman, W.P., Innocenti, F., D'Orazio, M., Leat, P.T., 2006. The boron transferfrom the slab to the mantle wedge: a combined study in South Sandwich island arcand Southern Volcanic Zone in the Chile Andes. Eos Transactions AGU 87 (36) Jt.Assem. Suppl. Abstract, V33A-06.

Tonarini, S., Agostini, S., Doglioni, C., Innocenti, F., Manetti, P., 2007. Evidence forserpentinite fluid in convergent margin systems: the example of El Salvador(Central America) arc lavas. Geochemistry, Geophysics, and Geosystems 8.doi:10.1029/2006GC001508.

Turcotte, D.L., Schubert, G., 2002. Geodynamics. Cambridge University Press. 456 pp.Turner, S., Tonarini, S., Bindeman, I., Leeman, W.P., Schaefer, B., 2007. Boron and oxygen

isotope evidence for recycling of subducted components over the past 2.5 Gyr.Nature 447. doi:10.1038/nature05898.

van Keken, P.E., 2003. The structure and dynamics of the mantle wedge. Earth andPlanetary Science Letters 215, 323–338.

Wiens, D.A., Conder, J.A., Faul, U.H., 2008. The seismic structure and dynamics of themantle wedge. Annual Review of Earth and Planetary Sciences 36, 421–455.doi:10.1146/annurev.earth.33.092203.122633.

Winter, J., 2001. An Introduction to Igneous and Metamorphic Petrology. Prentice Hall,New Jersey, p. 697.

You, C.F., Chan, L.H., Spivack, A.J., Gieskes, J.M., 1995. Lithium, boron, and their isotopesin sediment and pore waters of Ocean Drilling Program Site 808, Nankai Trough:implications for fluid expulsion in accretionary prism. Geology 23, 37–40.

Zheng, Y., Lay, T., Flanagan, M.P., Williams, Q., 2007. Pervasive seismic wave reflectivityand metasomatism of the Tonga mantle wedge. Science 11, 316, 5826, 855–859,doi:10.1126/science.1138074.