Short wavelength lateral variability of lithospheric mantle beneath the

Tectonophysics 650 (2015) 34–52

Contents lists available at ScienceDirect

Tectonophysics journal homepage: www.elsevier.com/locate/tecto

Short wavelength lateral variability of lithospheric mantle beneath the Middle Atlas (Morocco) as recorded by mantle xenoliths Hicham El Messbahi a,b, Jean-Louis Bodinier b, Alain Vauchez b,⁎, Jean-Marie Dautria b, Houssa Ouali a, Carlos J. Garrido c a b c

Equipe Géomatériaux, Université Moulay Ismaïl, Faculté des Sciences, BP 11201 Zitoune, Meknès, Maroc Géosciences Montpellier, Université de Montpellier 2, CNRS, Cc 60, Place Eugène Bataillon, 34095 Montpellier Cedex 05, France Instituto Andaluz de Ciencias de la Tierra (IACT), CSIC and UGR, Avenida de las Palmeras 4, 18100 Armilla, Granada, Spain

a r t i c l e

i n f o

Article history: Received 31 July 2014 Received in revised form 7 November 2014 Accepted 23 November 2014 Available online 2 December 2014 Keywords: Lithospheric mantle Middle Atlas (Morocco) Lithospheric thinning Melt–rock interactions Decompression melting Strain localization

a b s t r a c t The Middle Atlas is a region where xenolith-bearing volcanism roughly coincides with the maximum of lithospheric thinning beneath continental Morocco. It is therefore a key area to study the mechanisms of lithospheric thinning and constrain the component of mantle buoyancy that is required to explain the Moroccan topography. Samples from the two main xenolith localities, the Bou Ibalghatene and Tafraoute maars, have been investigated for their mineralogy, microstructures, crystallographic preferred orientation, and whole-rock and mineral compositions. While Bou Ibalghatene belongs to the main Middle Atlas volcanic field, in the ‘tabular’ Middle Atlas, Tafraoute is situated about 45 km away, on the North Middle Atlas Fault that separates the ‘folded’ Middle Atlas, to the South-East, from the ‘tabular’ Middle Atlas, to the North-West. Both xenolith suites record infiltration of sub-lithospheric melts that are akin to the Middle Atlas volcanism but were differentiated to variable degrees as a result of interactions with lithospheric mantle. However, while the Bou Ibalghatene mantle was densely traversed by high melt fractions, mostly focused in melt conduits, the Tafraoute suite records heterogeneous infiltration of smaller melt fractions that migrated diffusively, by intergranular porous flow. As a consequence the lithospheric mantle beneath Bou Ibalghaten was strongly modified by melt–rock interactions in the Cenozoic whereas the Tafraoute mantle preserves the record of extensional lithospheric thinning, most likely related to Mesozoic rifting. The two xenolith suites illustrate distinct mechanisms of lithospheric thinning: extensional thinning in Tafraoute, where hydrous incongruent melting triggered by decompression probably played a key role in favouring strain localisation, vs. thermal erosion in Bou Ibalghatene, favoured and guided by a dense network of melt conduits. Our results lend support to the suggestion that lithospheric thinning beneath the Atlas mountains results from the combination of different mechanisms and occurred in a piecewise fashion at a short wavelength scale. © 2014 Elsevier B.V. All rights reserved.

1. Introduction At the triple junction between the Atlantic Ocean, the Western Mediterranean and the West African Craton, Morocco is characterized by a rugged topography including the highest reliefs in northern Africa. There is however a consensus to consider that the Moroccan topography is only partly explained by crustal thickening in response to near- to far-field effects of the convergence between African and European plates. The Atlas mountain ranges were mostly structured in the Cenozoic by inversion of Mesozoic continental rifts related to the opening of the Atlantic and the Alpine Tethys (Frizon de Lamotte et al., 2008 and references therein). However, tectonic shortening did not exceed 10–30%, even in the central High Atlas where the elevation

⁎ Corresponding author. Tel.: +33 467143895. E-mail address: alain.vauchez@um2.fr (A. Vauchez).

http://dx.doi.org/10.1016/j.tecto.2014.11.020 0040-1951/© 2014 Elsevier B.V. All rights reserved.

exceeds 4000 m (Beauchamp et al., 1999; Gomez et al., 1998; Teixell et al., 2003, 2009). Moreover, the Anti-Atlas and the eastern Moroccan Meseta, including the ‘tabular’ Middle Atlas, although virtually unaffected by Cenozoic deformation, were uplifted to above 3000 and nearly 2000 m, respectively. A significant component of mantle buoyancy is therefore required to explain the Atlas reliefs, as also supported by evidence for abnormally thin lithospheric mantle based on gravity, heatflux, receiver functions, topography and crustal thickness data (Ayarza et al., 2005; Fullea et al., 2006, 2010; Miller and Becker, 2014; Missenard et al., 2006; Teixell et al., 2005), as well as seismological tomography (Bezada et al., 2014; Palomeras et al., 2014). Further evidence for asthenospheric upwelling arises from recent volcanic activity involving generation of sublithospheric melts at 60–120 km depth (e.g. Duggen et al., 2003, 2009; Frizon de Lamotte et al., 2008). In detail, geophysical studies converge on the existence of a roughly SW–NE trending corridor of thinned lithosphere, 200–500 km wide, where the asthenosphere–lithosphere boundary raises to 90–50 km

H. El Messbahi et al. / Tectonophysics 650 (2015) 34–52

(Bezada et al., 2014; Fullea et al., 2010; Jiménez-Munt et al., 2011; Missenard et al., 2006; Teixell et al., 2005; Zeyen and Fernandez, 1994). The corridor (the ‘Moroccan Hot Line’ of Frizon de Lamotte et al., 2009) extends from the Atlantic margin to the Eastern Rif through the central high-Atlas and the eastern Meseta, and concentrates most of the Moroccan Cenozoic volcanism. Several hypotheses have been put forward regarding the origin of lithospheric thinning beneath Morocco and the geodynamic significance of the Moroccan Hot Line. Debated issues notably addressed the pattern of mantle flow, with models involving upwelling of instable north-African upper mantle (Lustrino and Wilson, 2007), lateral spreading of a Canarian mantle plume (Duggen et al., 2009), toroidal/lateral mantle flow triggered by slab retreat and/or lithosphere delamination in the Alboran back-arc domain (Faccenna and Becker, 2010), or edge-driven convection related to the convergence between Africa and Europe (Kaislaniemi and van Hunen, 2014; Missenard and Cadoux, 2012). The proposed mechanisms of lithospheric thinning vary between two extremes: (1) relational-scale delamination of lithospheric mantle as a result of compressive lithosphere thickening (Duggen et al., 2009) and (2) thermal erosion of lower lithosphere by upwelling upper mantle, possibly reusing Mesozoic rift structures (Berger et al., 2009; Raffone et al., 2009). Between these extremes, to account for the complex topography of the lithosphere–asthenosphere boundary, Bezada et al. (2014) suggested a model involving piecewise lithosphere delamination due to tectonic shortening and loading by high-pressure magmatic segregates, combined with local upwelling of hot mantle and thermal erosion. Mantle xenoliths allow direct observation of lithospheric mantle and provide insights into the mechanisms of lithospheric thinning. Processes such as eclogite loading and thermal erosion can be assessed, and the xenolith studies also provide clues to understand the role of deformation and fluid-rock processes in the convective erosion of the lithosphere. The most conspicuous xenolith localities in Morocco are concentrated in the ‘tabular’ Middle Atlas, where intra-plate Cenozoic volcanism coincides with nearly 2000 m uplifting of the undeformed Mesozoic cover and – roughly – with the maximum of lithospheric thinning beneath continental Morocco (Bezada et al., 2014; Fullea et al., 2010; Miller and Becker, 2014). However, previous xenolith studies focused on a single locality — the Bou Ibalghatene maar, in the central part of the volcanic district (Lenaz et al., 2014; Natali et al., 2013; Raffone et al., 2009; Wittig et al., 2010a,b). There, the mantle xenoliths were affected by extensive reactions with silicate and carbonate melts, a feature which is not observed to a similar degree in the other xenolith localities. The Bou Ibalghatene xenoliths may represent a restricted area of focused melt flow that is not representative of the whole Middle Atlas lithospheric mantle. In addition to representative peridotites from Bou Ibalghatene, analysed for comparison, this study deals with mantle xenoliths from the Tafraoute maar, located 45 km to the North-East of Bou Ibalghatene and distal to the main volcanic district. The maar is situated on the NE–SW North Middle Atlas Fault, a major transpressive fault separating the ‘folded’ Middle Atlas, to the South-East, from the ‘tabular’ Middle Atlas, to the North-West. The samples have been investigated for their mineralogy, microstructures, crystallographic preferred orientation (EBSD-SEM), and bulk rock and mineral compositions (XRF, EPMA and LA-ICP-MS). 2. Geological setting The Middle Atlas volcanic province lies on a Mesozoic limestone plateau that represents the easternmost and most elevated part of the Moroccan Meseta, and is sometimes referred to as the ‘tabular’ Middle Atlas (Fig. 1). The tabular Middle Atlas is separated from the folded Middle Atlas belt to the southeast by a NE-trending, southeast dipping transpressive thrust fault, the ‘North Middle Atlas Fault’. Geophysical imaging suggests that this area coincides with the maximum of lithospheric thinning beneath continental Morocco (Bezada et al., 2014; Fullea et al., 2010). The plateau is crosscut and bounded by NE–

35

SW trending major faults of Variscan age that were reactivated during the Alpine orogeny (Frizon de Lamotte et al., 2008). The highest elevations (nearly 2000 m) and the maximum of volcanic activity are observed on the Azrou–Timahdite plateau, bounded by the TiziN'Tretten Fault to the North-West and the North Middle Atlas Fault to the South-East. The Middle Atlas volcanic province is the largest (~1000 km2) and youngest (16 to 0.6 Ma) volcanic field in Morocco (El Azzouzi et al., 2010). About one hundred well-preserved strombolian cones and maars define a ~ 120 km long, N160–170°E trending alignment (Fig. 1). Each volcano emitted one single lava flow of nephelinite, basanite or alkali basalt (El Azzouzi et al., 1999, 2010). In spite of the relatively large number of emission points, the volume of lava flows is rather small (about 10 km3) and the lack of true differentiated lava excludes the presence of magmatic chambers. Mantle xenoliths may be found in several volcanic centres and lava flows but they are more frequent in two localities: the Bou Ibalghatene and Tafraoute maars. The Bou Ibalghatene maar is located in the centre of the main volcanic field (33°20′12.91″N; 5°03′07.42″W — Fig. 1). The volcanic structure is 3.5 km long and nearly 1.5 km wide. It is composed of two contiguous explosion craters bounded by two anastomosed half-rings of phreatomagmatic tufs. The Bou Ibalghatene mantle xenoliths were previously described by Raffone et al. (2009). We collected a new set of samples, among which 27 peridotites were examined in thin sections and 9 processed with the same methods as the Tafraoute suite, for comparison. The Tafraoute maar is located 45 km to the North-East of Bou Ibalghatene, away from the main volcanic field (33°20′12.91″N; 5°03′ 07.42″W — Fig. 1). It is situated near the Aït Bouzziyane and Tafraoute villages, at the foot of a cliff formed by a slickenside of the North Middle Atlas Fault (Fig. 1). The volcanic edifice is about 2 km long and 500 m wide, oriented NE–SW, parallel to the North Middle Atlas Fault. It is composed of an elliptical explosion crater bounded by a semicircular phreatomagmatic tuff ring. The xenoliths are mostly found in the upper part of the tuf ring, on its SE flank. They are rounded, generally devoid of lava crust and bigger, on average, that in the other Middle Atlas localities (up to 30 cm in diameter). Mantle xenoliths include both peridotites and spinel or garnet pyroxenites, and are associated with garnetand kyanite-bearing granulite xenoliths from the lower crust. 150 mantle xenoliths were collected for this study, among which 40 peridotites and spinel pyroxenites were examined in thin sections and 12 selected for detailed investigations. 3. Petrography and sample selection The Bou Ibalghatene mantle xenoliths show a wide range of modal compositions including spinel lherzolites, harzburgites, dunites, wehrlites, olivine-spinel websterites, and spinel websterites (Fig. 2). According to Raffone et al. (2009), about 65% of the Bou Ibalghatene peridotites also contain secondary amphibole ± phlogopite. In contrast, the Tafraoute suite is dominated by fertile spinel lherzolites (12.5–15% clinopyroxene); harzburgites are subordinate while dunites and wehrlites are virtually absent. The suite also comprises olivine–spinel websterites and garnet clinopyroxenites, and further differs from the Bou Ibalghatene suite by the lack of amphibole and phlogopite, except for very rare exceptions. In both localities, the peridotites are equilibrated in the spinel peridotite facies. Spinel is most often interstitial and aligned along foliation in samples that are foliated. However, several peridotites from Tafraoute contain symplectites of orthopyroxene + clinopyroxene + spinel (Fig. 3a) that likely represent destabilization products of former garnets (Morishita and Arai, 2003; Nicolas et al., 1987; Shimizu et al., 2008). About half of the Tafraoute samples also contain interstitial microgranular aggregates, 0.1 to 1 μm in grain size, forming a network of anastomosed micro-veins (b 100 μm in width) predominantly developed at olivine–olivine grain boundaries. The micro-veins are locally

36


Fig. 1. Sketch map showing the location of the sampling sites (Tafraoute and Bou Ibalghatene maars) and the main geological features of the Middle Atlas volcanic district: unfolded (a) and folded (b) lower Jurassic carbonate deposits; c — major faults; d — lava flows; e — maars, f — strombolian cones (modified from El Azzouzi et al., 2010). Insert: sketch map showing the location of the Middle Atlas volcanic district showing the location of the Middle Atlas volcanic district and the main mountain belts in Morocco (A.A.: Anti-Atlas; H.A.: High Atlas; M.A.: Middle Atlas; M.M.: Moroccan Meseta).

Fig. 2. Modal compositions of the studied mantle xenoliths from Tafraoute (solid circles) and Bou Ibalghatene (open squares). The Bou Ibalghatene samples studied by Raffone et al. (2009) and Wittig et al. (2010a) are shown for comparison (grey crosses).

developed at some angle to the peridotite foliation and may also cut large porphyroclasts. The micro-granular aggregates are also present at clinopyroxene–spinel grain boundaries where clinopyroxene shows a spongy texture and spinel a chromite rim. These features suggest that the xenoliths were belatedly percolated by small melt fractions that migrated through grain boundaries and reacted with clinopyroxene and spinel. Micro-granular interstitial aggregates are also observed in the Bou Ibalghatene xenoliths. However the latter show evidence for an origin by recrystallization of a glass phase formed by decompression melting of amphibole, most likely during transport of the xenoliths by host lavas. The Bou Ibalghatene samples selected for this study comprise four spinel lherzolites, one amphibole-rich peridotite, two harzburgites, and two wehrlites (Table 1, Fig. 2). Most samples contain a few percent amphibole. Exceptions include three amphibole-free lherzolites (samples Ib15N, Ib10, and Ib21N) and the amphibole-rich peridotite sample Ib4 that contains ~ 14% amphibole. In addition to amphibole, the wehrlite sample IB14 contains some flakes of phlogopite. The samples from Tafraoute include nine spinel lherzolites, two harzburgites, and a composite xenolith showing a 3 cm-thick layer of olivine–spinel websterite hosted by an orthopyroxene-poor (7.5%) lherzolite (samples Taf21a and b, respectively — Table 1, Fig. 2). The websterite layer is parallel to the foliation in the host peridotite and shows a similar microstructure, intermediate between coarse-equant and porphyroclastic (see below). The peculiar modal composition of


37

Fig. 3. Microphotographs illustrating the main textural and micro-structural features of the Tafraoute (a–f) and Bou Ibalghatene (g–h) mantle xenoliths: (a) symplectitic spinel–pyroxenes clusters crystallized from original garnets in sample Taf21b (plane-polarized light); (b–f) typical microstructures in the Tafraout peridotites (cross-polarized light): (b) coarse equant (Taf21b), (c) porphyroclastic (Taf38), (d) granular (TAf51), (e) and (f): mylonitic (Taf19 and Taf42, respectively); (g–h) sample Ib14 from Bou Ibalghatene, an amphibole-bearing wehrlite characterized by extensive textural annealing (plane- and cross-polarized light, respectively).

the host lherzolite is unique among our Tafraoute sampling and is somewhat reminiscent of the Bou Ibalghatene wehrlites (Fig. 2). This sample is also one of the two selected peridotites from Tafraoute containing clusters of pyroxene and vermicular spinel; the other one is the harzburgite sample Taf37. The Tafraoute xenoliths are generally amphibole-free, the only noticeable exception being the fine-grained (mylonitic) lherzolite Taf42 that contains minute amounts of amphibole of very small size (b100 μm) and with euhedral needle shape, associated with calcite and Ni-rich Fe–sulphides. 4. Microstructures The studied xenoliths display a variety of microstructures, among which three types are predominant and observed both in the Bou

Ibalghatene and Tafraoute suites (coarse-equant, porphyroclastic and granular — Fig. 3b,c,d). Mylonitic microstructures, characterized by a finer grain size, are specific of the Tafraoute suite (Fig. 3e,f). Most xenoliths show evidence of annealing, a feature that tends to be more conspicuous in Bou Ibalghatene (Fig. 3g, h). Coarse-equant xenoliths (Fig. 3b) are dominated by large olivine and pyroxene crystals, up to 5 mm in diameter; smaller grains (neoblasts) are rare or virtually absent. Olivine usually does not display any substructure or scarcely contains spaced subgrain boundaries accommodating rather large misorientations. Grain boundaries are predominantly regular, locally indented or cuspate at olivine–olivine boundaries. The microstructural characteristics of olivine suggest that annealing favoured recovery and limited grain-boundary migration. Orthopyroxene is often irregular in shape, with embayments filled with large olivine crystals. In the

38


Table 1 Rock types, microstructures and modal compositions of the mantle xenoliths selected for this study. Lhz = lherzolite, Hz = harzburgite, Wb = websterite, Per = peridotite, Wh = wehrlite; Ol = olivine, Opx = orthopyroxene, Cpx = clinopyroxene, Sp = spinel, and Amp = amphibole. Sample

Rock type

Tafraoute maar Taf 12 Taf 2 Taf 36 Taf 38 Taf 42 Taf 50 Taf 19 Taf 43 Taf 37 Taf 51 Taf 21 b Taf 21 a

Lhz Lhz Lhz Lhz Lhz Lhz Lhz Lhz Hz Hz Lhz Wb

Bou Ibalghatene maar Ib 10 Ib15 N Ib 21 N Ib 15 Ib 4 Ib 12 N Ib 13 N Ib 1 Ib 14

Lhz Lhz Lhz Lhz Amp-rich Per Hz Hz Wh Wh

Microstructure

Modal compositions (vol.%) Ol

Opx

Cpx

Sp

Amp

Coarse equant Porphyroclastic Granular Porphyroclastic Mylonitic Granular Mylonitic Granular Porphyroclastic Granular Coarse equant/porphyroclastic Porphyroclastic

64.7 64.5 56.8 69.7 61.5 54.2 55.5 62.1 68 76.6 78.8 28.4

18.8 20 24.6 20.4 23.4 28.1 21.2 21.5 27 17.8 7.5 36.4

14.6 13.6 14.8 8.6 12.6 15.1 20.5 12.9 3 4.8 12.5 32.4

1.7 1.9 3.8 1.3 2.5 2.6 2.7 3.5 2 0.8 1.2 2.8

0 0 0 0 0 0 0 0 0 0 0 0

Granular Granular Granular Porphyroclastic Coarse equant Porphyroclastic Porphyroclastic Granular Coarse equant

60 57.3 57.6

24.8 25.8 27

13.4 14.6 13.3

1.8 2.2 2

0 0 0

59.3 64.6 60.6 75.2 69.1

21 28 27.2 3.4 0

5.4 2.8 7.7 18.1 27.2

0.1 0.1 2.6 0 2.3

14.2 4.4 1.9 3.3 1.3

Tafraoute suite, the pyroxene + spinel clusters are mostly observed in peridotites characterized by coarse-equant microstructures. Porphyroclastic xenoliths (Fig. 3c) contain two populations of olivine crystals with different grain-sizes: large porphyroclasts (3–5 mm) of olivine and pyroxenes embedded in a matrix of smaller recrystallized crystals. The relative proportion of porphyroclasts and neoblasts varies continuously, suggesting that the porphyroclastic microstructure is transitional between coarse-equant and granular microstructures. The small grains are usually equant and up to 600–700 μm in diameter; they are polyhedral and their boundaries form frequent 120° triple junctions. They are free of any substructure and many of them contain small crystals of spinel. Olivine porphyroclasts usually display contorted, sometimes cuspate boundaries when in contact with the matrix, and contain close subgrain boundaries with a frequent fan-shaped disposition. In addition, some porphyroclasts display smaller subgrains similar in size to the neoblasts, suggesting dynamic recrystallization through subgrains rotation. In many cases, however, new grains occur inside porphyroclasts that do not display any substructure, or independently of the subgrain boundaries, suggesting nucleation and grain growth as recrystallization mechanism. Orthopyroxene in porphyroclastic samples has irregular boundaries, with embayments filled with olivine neoblasts. Granular xenoliths (Fig. 3d) are mostly composed of crystal grains around 0.5–1 mm in diameter, similar to the neoblasts in porphyroclastic samples. The crystals are substructure free; they tend to be polygonal and form many 120° triple junctions. Several granular samples contain remnants of porphyroclasts (up to 2 mm) showing close subgrain boundaries in olivine, as well as former large crystals divided in several fragments isolated within aggregates of smaller, recrystallized crystals. Orthopyroxene ranges from a few hundreds of μm to several mm in size. The largest orthopyroxene crystals are often contorted and show large embayments filled with olivine neoblasts. Superimposed to the porphyroclastic and granular microstructures, several samples from the Tafraoute maar display very fine-grained aggregates, usually with a ‘cloudy’ appearance, dominantly composed of small olivine crystals (~10–50 μm) and interstitial spinel (Fig. 4a). The fine-grained domains are elongated and form anastomosed networks. Locally, they also form larger pockets between original grains. The proportion of fine-grained aggregates is variable; when they are predominant the microstructure is transitional with fine-grained, ‘mylonitic’

microstructures. Olivine crystals in the fine-grained aggregates are systematically polygonal and some of them have grown subidiomorphic (Fig. 4b,c). Olivine porphyroclasts in contact with the fine-grained domains frequently display fluid inclusions and neoblasts near the rims, even when the original grains are substructure free. More seldom, the finely recrystallized aggregates crosscut olivine crystals and isolate fragments of the original grains. At the contact with fine-grained domains, pyroxenes display many deeply penetrating embayments with ‘fjordlike’ or ‘den-like’ shapes — sometimes reaching the crystal core. The embayments are filled with fine-grained olivine aggregates or, in some cases, with acicular olivine crystals (Fig. 4d,e). Small neighbouring orthopyroxene grains separated by fine-grained olivine sometimes have similar crystallographic orientations, suggesting that they derive from a single, larger grain. Fine-grained, or mylonitic xenoliths (Fig. 3e,f) are peridotites in which the development of the fine-grained olivine aggregates dominates the microstructure. Olivine is 20–100 μm in diameter, equant to slightly elongated and systematically polygonal with well-developed crystallographic faces. Many subidiomorphic to idiomorphic olivine crystals are dispersed in the matrix and the largest ones display sharp crystallographic faces, sometimes set into pre-existing larger olivine crystals at the expense of which they have probably grown (Fig. 3h). This finegrained matrix may wrap around aggregates of larger (up to 500 μm), polygonal olivine crystals very similar to the recrystallized olivine grains of the porphyroclastic and granular samples (Fig. 3a). It contains large amounts of small ortho- and clinopyroxenes (b100 μm), the proportion of which is inversely proportional to olivine grain size. Some samples (e.g., Taf19 and Taf41) still display a few porphyroclasts of olivine and pyroxene, and are thus intermediate between porphyroclastic and mylonitic microstructures. The pyroxene porphyroclasts are associated with fine-grained olivine crystallized either in embayments and contain needle-like crystals or ribbon-like aggregates of olivine or olivine + orthopyroxene neoblasts crosscutting mineral grains. 5. Analytical methods Crystal preferred orientations (CPO) of olivine were measured by indexing Electron Backscattered Diffraction (EBSD) patterns produced using the JEOL-5600 Scanning Electron Microscope equipped


39

Fig. 4. Microphotographs illustrating typical microstructures of the fine-grained, “mylonitic” xenoliths from Tafraoute. (a) large orthopyroxene (Opx) crystal associated with secondary olivine (Ol2) occurring in embayments or as “needle-like” monocrystal (sample Taf19, cross-polarized light). (b) and (c) sub-idiomorphic crystals of secondary olivine grown at the expense of primary olivine (Ol1); the new crystals display well-defined crystallographic faces and frequently contain minute, rounded grains of spinel (Sp) (sample Taf19, cross-polarized light). (d) Secondary olivine crystals grown at the expense of primary olivine showing concave grain boundaries; the arrow shows a pocket of small limpid olivine crystals, several of which display a polyhedral shape; the microphotograph also shows (upper left) acicular secondary olivines crystallized within an orthopyroxene porphyroclast (sample Taf41, cross-polarized light). (e) EBSD map illustrating the foliation observed in the most “mylonitic” microstructure (sample Taf 42), wherein small orthopyroxene (yellow) and clinopyroxene (blue) crystals are dispersed within layers of fine-grained olivine (green). The negative correlation between pyroxene proportion and olivine grain size suggests pinning of olivine boundaries by pyroxenes.

with an Oxford-HKL Instruments EBSD system available at Geosciences Montpellier (France). Diffraction pattern acquisition was performed with an acceleration voltage of 17 kV and a working distance of ~ 25 mm. Processing and indexation of Kikuchi bands, and postacquisition processing of crystallographic orientation measurements, have been performed using the Channel 5 software (Oxford-HKL Instruments). For each sample, we obtained crystallographic orientation maps covering most of the thin section with a regular grid step ranging from 25 to 60 μm depending on the size of the finest grains in the sample. Raw indexation rate varies from 70 to 86%. Post-acquisition data

processing allowed us to further increase the indexation rate by (1) filling the non-indexed pixels that have up to 8 identical neighbours with the same orientation, (2) repeating this operation using 7, 6, and 5 identical neighbours, (3) identifying the grains, i.e., continuous domains characterized by an internal misorientation b 15°, and (4) within each olivine crystal, searching and correcting for systematic indexing errors due to the olivine hexagonal pseudosymmetry, which results in almost similar diffraction patterns for orientations differing by a rotation of 60° around [100]. At each step the resulting orientation maps were checked to avoid over extrapolation of the data. The strength of the preferred

40


Table 2 Mineral major-element compositions and equilibrium temperatures in Tafraoute and Bou Ibalghatene mantle xenoliths. Abbreviations for microstructures: CE = coarse-equant, G = granular, P = porphyroclastic, My = mylonitic. N stands for the number of analyses in a given sample and the bracketed values represent standard deviations. Fo = % forsterite content, Mg# = 100 ∗ Mg / (Mg + Fe2+), Cr# = 100 ∗ Cr / (Cr + Al + Fe3+). Wo = wollastonite, En = enstatite, Fs = ferrosilite. Pyroxene equilibrium temperatures calculated using the Wells (1977) (T°W), and Brey and Köhler (1990)-(T°Bk) calibrations. c = crystal cores, r = crystal rims, P = porphyroclasts, n = neoblasts; * samples lacking Cpx porphyroclasts. Taf 12

Taf 36

Taf 43

Taf 51

Taf 2

Taf 19

Taf 37

Taf 38

Taf 42

Taf 50

Taf 21 b

Taf 21 a

Lhz (CE)

Lhz (G)

Lhz (G)

Hz (G)

Lhz (P)

Lhz (My)

Hz (P)

Lhz (P)

Lhz (My)

Lhz (G/P)

Lhz (CE/P)

Wb (P)

Olivine N 6 4 3 Fo 89.56 (0.01) 89.55 (0.04) 89.2

4 89.44 (0.06)

7 88.86 (0.07)

10 89.01 (0.24)

5 5 6 3 89.43 (0.27) 89.65 (0.09) 89.03 (0.33) 89.1

4 89.98 (0.05)

3 90

Orthopyroxene N 28 %Wo 1.46 (0.25) %En 88.36 (0.30) %Fs 10.18 (0.09) Mg# 89.87 (0.11) Al2O3 4.5 (0.3)

6 0.65 (0.03) 89.16 (0.30) 10.27 (0.30) 89.96 (0.29) 3.18 (0.73)

8 0.67 (0.11) 88.97 (0.16) 10.37 (0.09) 89.78 (0.13) 3.34 (0.48)

8 0.65 (0.09) 89.60 (0.14) 9.78 (0.13) 90.4 (0.17) 2.28 (0.34)

16 1.17 (0.06) 88.38 (0.10) 10.45 (0.48) 89.64 (0.07) 4.03 (0.14)

21 1.28 (0.86) 88.32 (1.07) 10.43 (0.25) 89.64 (0.33) 3.8 (0.61)

4 0.84 (0.02) 88.60 (0.01) 10.56 (0.01) 89.58 (0.01) 3.47 (0.17)

8 1.64 (0.51) 88.42 (0.41) 9.94 (0.15) 90.1 (0.10) 4.15 (0.23)

10 0.89 (0.03) 88.45 (0.14) 10.66 (0.15) 89.45 (0.14) 3.8 (0.40)

9 0.82 (0.52) 88.61 (0.60) 10.55 (0.22) 89.54 (0.24) 3.71 (0.87)

8 1.24 (0.06) 89.27 (0.04) 9.50 (0.03) 90.58 (0.03) 3.4 (0.08)

11 1.19 (0.07) 89.18 (0.21) 9.63 (0.18) 90.47 (0.18) 4.27 (0.60)

Clinopyroxene N 16 %Wo 45.6 (0.51) %En 48.08 (0.3) %Fs 6.32 (0.24) Mg# 88.64 (0.3) Al2O3 6.69 (0.21) Na2O 1.55 (0.01) TiO2 0.44 (0.06) Cr2O3 0.8 (0.03)

7 48.59 (1.12) 45.68 (1.07) 5.73 (0.27) 89.08 (0.52) 6.28 (0.86) 1.95 (0.28) 0.56 (0.18) 0.7 (0.20)

6 49.50 (0.28) 45.67 (0.27) 4.83 (0.07) 90.7 (0.14) 6.80 (0.37) 1.85 (0.12) 0.63 (0.02) 0.75 (0.07)

6 48.76 (0.36) 46.00 (0.46) 4.24 (0.14) 91.96 (0.30) 4.04 (0.75) 1.19 (0.14) 0.27 (0.03) 1.04 (0.23)

11 47 (0.24) 47.46 (0.26) 5.54 (0.07) 89.79 (0.11) 7.09 (0.19) 1.83 (0.03) 0.76 (0.04) 0.72 (0.05)

11 48.26 (0.37) 46.27 (0.36) 5.46 (0.12) 89.71 (0.21) 6.79 (0.23) 1.88 (0.09) 0.64 (0.04) 0.83 (0.05)

4 45.78 (0.18) 48.3 (0.07) 5.92 (0.10) 89.31 (0.14) 6.03 (0.03) 1.07 (0.01) 0.39 (0.23) 0.71 (0.01)

11 46.38 (0.77) 47.99 (0.68) 5.63 (0.14) 89.79 (0.17) 6.17 (0.16) 1.37 (0.31) 0.52 (0.04) 0.74 (0.14)

10 48.25 (0.35) 45.95 (0.50) 5.8 (0.17) 89.08 (0.37) 6.59 (0.75) 1.85 (0.19) 0.6 (0.10) 0.72 (0.11)

6 47.52 (2.77) 46.71 (2.75) 5.77 (0.22) 89.23 (0.64) 6.85 (0.23) 1.98 (0.13) 0.66 (0.05) 0.66 (0.06)

6 47.25 (0.20) 47.93 (0.20) 4.82 (0.05) 91.14 (0.03) 4.29 (0.03) 1.08 (0.03) 0.16 (0.01) 0.77 (0.04)

13 46.91 (0.33) 48.02 (0.21) 5.07 (0.18) 90.69 (0.29) 6.01 (0.85) 1.02 (0.02) 0.22 (0.02) 0.65 (0.19)

Spinel N 14 4 4 Cr# 12.13 (0.21) 9.56 (0.09) 9.87 (0.16) Mg# 76.38 (0.26) 75.71 (0.57) 74.48 (0.11)

4 26.06 (0.80) 70.43 (1.86)

5 8.67 (0.16) 76.40 (0.42)

18 11.13 (0.41) 76.32 (0.37)

2 24.2 74.3

4 6 2 11.27 (0.89) 10.16 (0.10) 9.1 76.90 (0.82) 75.74 (0.82) 76.4

3 20.5 76.4

3 16.4 77.2

Equilibrium temperature T°W c 974 (15) 800 (15) r 996 (7) 807 (32)

831 (10) 818 (17)

826 (3) 826 (12)

P c 944 (10) P r 954 (11) n

969 (56) 827 (2) 890 (7)

994 (3) 973 (7)

947 (16) 971 (9)

855 (11) 871 (21) 854 (5)

1100 820 830

921 (8) 925

921 925

T°Bk c r

814 (19) 796 (26)

776 (6) 775 (10)

P c 983 (14) P r 997 (11) n

999 (59) 930 (2) 911 (9)

1017 (5) 1020 (6)

970 (20) 999 (12)

870 (19) 875 (33) 853 (10)

1150 810 830

908 (11) 914 (3)

908 914

1008 (17) 1035 (14)

778 (22) 782 (53)

Ib 10

Ib 15 N

Ib 21 N

Ib 4

Ib 1

Ib 14

Amphibole-free

Ib 12 N

Ib 13 N

Amphibole-bearing

Lhz (G)

Lhz (G)

Lhz (G)

Per (CE)

Wh (G)

Wh (CE)

Hz (P)

Hz (P)

19 89.55 (0.18)

4 89.49 (0.17)

7 89.45 (0.04)

9 89.9 (1.11)

2 89.8

9 86.19 (0.06)

2 90.6

5 90.60 (1.10)

Orthopyroxene N 9 %Wo 0.73 (0.02) %En 89.17 (0.06) %Fs 10.10 (0.07) Mg# 90.03 (0.07) Al2O3 3.16 (0.21)

6 1.48 (0.19) 88.09 (1.13) 10.30 (0.11) 89.64 (0.20) 4.44 (0.81)

5 0.83 (0.03) 88.81 (0.19) 10.38 (0.06) 89.75 (0.22) 3.69 (0.57)

6 1.52 (0.04) 87.91 (0.05) 10.56 (0.03) 89.54 (0.04) 4.13 (0.03)

6 1.49 (0.05) 89.46 (0.08) 9.06 (0.02) 91.04 (0.07) 2.89 (0.07)

7 1.40 (0.08) 88.58 (0.09) 10.05 (0.05) 90.06 (0.10) 3.88 (0.17)

Clinopyroxene N 9 %Wo 48.64 (0.27) %En 45.74 (0.37) %Fs 5.62 (0.15) Mg# 89.28 (0.31) Al2O3 6.03 (0.71) Na2O 1.30 (0.16) TiO2 0.55 (0.14) Cr2O3 0.69 (0.02)

7 48.15 (0.44) 46.23 (0.46) 5.62 (0.16) 89.40 (0.30) 6.20 (0.71) 1.48 (0.17) 0.50 (0.14) 0.60 (0.08)

4 48.10 (0.24) 46.26 (0.43) 5.64 (0.23) 89.39 (0.48) 5.59 (0.68) 1.06 (0.08) 0.37 (0.08) 0.69 (0.04)

5 45.15 (0.11) 48.28 (0.06) 6.57 (0.07) 88.37 (0.08) 5.85 (0.03) 1.54 (0.03) 0.18 (0.01) 0.81 (0.02)

6 45.31 (0.34) 49.40 (0.27) 5.29 (0.09) 90.63 (0.11) 4.12 (0.14) 1.53 (0.06) 0.03 (0.01) 1.42 (0.08)

9 45.40 (0.33) 48.62 (0.91) 5.98 (0.70) 89.36 (0.12) 5.16 (0.10) 1.31 (0.04) 0.32 (0.03) 0.81 (0.07)

Spinel N Cr# Mg#

4 8.28 (0.22) 77.52 (1.30)

4 10.37 (1.37) 75.46 (0.25)

Olivine N Fo

8 9.84 (0.43) 76.10 (0.82)

4 47.10 (2.10) 46.91 (3.24) 5.99 (1.15) 88.91 (2.7) 3.92 (1.23) 1.34 (0.49) 0.63 (0.91) 0.93 (0.53)

2 30.4 69.46

8 44.83 (0.40) 46.81 (0.47) 8.37 (0.10) 85.23 (0.27) 6.95 (0.37) 1.21 (0.06) 0.98 (0.11) 0.71 (0.04)

3 17.5 76.5


41

Table 2 (continued) Ib 10

Ib 15 N

Ib 21 N

Ib 4

Ib 1

Ib 14

Ib 12 N

Amphibole-free Lhz (G)

Lhz (G)

Ib 13 N

Amphibole-bearing Lhz (G)

Amphibole N K2O Na2O TiO Mg#

Per (CE) 9 0.50 (0.01) 3.67 (0.05) 0.80 (0.01) 86.97 (0.2)

Wh (G)

Wh (CE)

Hz (P)

Hz (P)

11 2.07 (0.02) 2.43 (0.04) 3.28 (0.05) 82.61 (0.2)

4 1.56 (0.01) 3.09 (0.05) 0.20 (0.01) 89.3 (0.2)

2 0.41 3.57 0.78 87.8

Equilibrium temperature T°W c 879 (7) r 877 (6)

877 (6) 887 (26)

891 (15) 887 (1)

976 (11) 996 (8)

Pc Pr n

* * 972 (13)

987 (5) 988 (1) 971

T°Bk c r

874 (6) 881 (40)

881 (16) 872 (10)

1008 (17) 1042 (12)

Pc Pr n

* * 999 (17)

1017 (6) 1018 (1) 991

884 (14) 865 (7)

orientation was determined using the J-index, which is the volumeaveraged integral of the squared orientation densities (Bunge, 1982). Pole figures have been plotted using average Euler angles for each grain instead of individual measurements to avoid over-representation of large crystals in the CPO. Data processing to generate and rotate pole figures and compute the J-index was performed using D. Mainprice's software package (ftp://saphir.gm.univ-montp2.fr/mainprice//CareWare_ Unicef_Programs/). Major elements in minerals (Table 2 and supplementary data) were determined by electron microprobe at ‘Microsonde Sud’ facility (Montpellier, France), using a CAMECA-SX100 equipped with five wavelength-dispersive spectrometers. Operating conditions comprised an acceleration voltage of 20 kV and a 10 nA beam current. Major and several minor elements in bulk rocks (Table 3) were analysed by wide-angle X-ray fluorescence (WDXRF) at ‘Instituto Andaluz de Ciencias de la Tierra’ (Granada, Spain), using a sequential spectrometer Bruker S4 Pioneer. After crushing and grinding in an agate mortar, the samples were fused with di-lithium tetraborate. Element concentrations were measured on glass beads, using reference geological standards and the recommended values of Govindaraju (1994). Trace elements in clinopyroxene were analysed by Laser Ablation (LA) ICP-MS at Geosciences Montpellier (France), using a GeoLas Q + Excimer (Compex 102) laser system operating in the deep UV (193 nm). Ablations were performed on hand-picked mineral grains (Table 4), in a pure He atmosphere (≈0.6 l.min−1), using beam diameters ranging between 30 and 120 μm, with an energy density c.a. 15.10−3 J·cm−2. The laser ablation platform was linked to an extended range (XR-) Element 2 ICP-MS operated in low-resolution mode at 1350 W. The ICP-MS was daily tuned to maximum sensitivity while keeping oxide production to its minimum level (ThO/Th ≤ 1%). The NIST612 glass (Pearce et al., 1997) was used as external standard and CaO content determined by electron probe was used as an internal standard. Data were processed using the GLITTER package (Griffin et al., 2008). The precision and accuracy of the LA-ICP-MS analyses were assessed from the results obtained for the BIR-1G glass (Jochum et al., 2005, 2006). 6. Crystallographic preferred orientations The crystallographic preferred orientation (CPO) of olivine displays a large variety of symmetry and fabric strength (Fig. 5). Three main CPO symmetries are represented: (1) orthorhombic, with point concentrations for [100], [010] and [001], (2) [100]-fiber or -axial with a point maximum for [100] and a dispersion in a girdle normal to the [100] maximum for the two other axis, and (3) [010]-fiber or -axial with a point maximum for [010] and a dispersion in a girdle normal to the [010] point concentration for the two other axis. Indeed, CPO classified in one of the axial-

symmetry groups are always intermediate between orthorhombic and one of the axial symmetry. The dispersion in a girdle is dominant, but there is always a well-defined point concentration within the girdle. No significant variation is observed between the Bou Ibalghatene and Tafraoute suites. CPO symmetry and microstructures do not show any clear correlation either, except for the fact that the [100]-axial symmetry is observed only in 3 samples with a granular microstructure and in one sample showing a transitional microstructure between granular and mylonitic. Apart from these samples, most xenoliths show either orthorhombic or [010]-fiber CPO. The variations in fabric strength for the coarse-grained, porphyroclastic and mylonitic xenoliths are correlated with the average grain-size and mostly reflect the proportion of fine grains in the rock (Fig. 6). The coarse-equant samples are characterized by the highest Jvalues while the fine-grained, mylonitic samples show the lowest values, with the porphyroclastic xenoliths yielding intermediate values. The granular xenoliths form a slightly distinct group characterized by lower J-index than the porphyroclastic xenoliths with similar average grain size. 7. Mineral compositions The mineral compositions obtained for this study are on overall consistent with the compositions reported by Raffone et al. (2009) for the Bou Ibalghatene xenoliths and show variations related to bulk rock compositions as well as more subtle differences between the Tafraoute and Bou Ibalghatene suites (Table 2). Olivine composition tends to be less variable in Tafraoute (88.9– 90.0% Fo) than in Bou Ibalghatene (86.2–90.6 Fo). In the Bou Ibalghatene suite, the lowest Fo value (86.2) is specific of the amphibole- and phlogopite-bearing wehrlite Ib14 whereas the highest values (Fo N 90) are observed in the harzburgite samples Ib13N and Ib12N. Orthopyroxene shows Mg# values (Mg# = 100 × Mg / (Mg + Fe) in a.p.f.u.) in the range 89.5–91.0 and Al2O3 contents varying between 2.3% and 4.5%, with no significant difference between the Bou Ibalghatene and Tafraoute suites. Al2O3 in orthopyroxene is positively correlated with Al2O3 in bulk rocks (Table 3), i.e., with peridotite fertility. The lowest Al2O3 contents in orthopyroxene, in particular, are observed in the most refractory (Al-poor) harzburgites, both in Bou Ibalghatene (sample Ib12N, with 2.89% Al2O3 in orthopyroxene and 1.65% in bulk rock) and in Tafraoute (sample Taf51, with 2.27% Al2O3 and 1.27% in bulk rock). Clinopyroxene is more variable in composition than olivine and orthopyroxene, with, notably, Mg# varying between 85.2 and 92.0, and TiO2 between 0.03 and 0.98%. In part, the Mg# range reflects a significant difference between the Tafraoute and Bou Ibalghatene suites, the Tafraoute xenoliths being distinguished by higher Mg# values in

Lhz (My)

43.07 1.36 9.07 0.13 43.32 2.79 0.11 0.01 0.12 0.01 0 90.44 4 5 43 1830 116 2571 36 43 20 45.23 3.77 9.30 0.15 36.46 4.43 0.26 0.05 0.29 0.05 0 88.6 17 120 77 2585 109 2110 18 52 38 45.57 3.62 8.93 0.14 38.16 3.13 0.23 0.01 0.2 0.01 0 89.43 7 14 77 2525 105 2022 15 53 30 44.62 3.41 9.36 0.14 39.19 2.81 0.21 0.01 0.22 0.02 0 89.24 10 15 69 2428 107 2024 10 55 34

Lhz (G/P) Lhz (G) Lhz (My)

44.69 3.13 9.04 0.13 39.89 2.78 0.15 0 0.18 0.01 0 89.73 6 2 67 2388 107 2167 18 52 18

44.97 3.16 9.26 0.14 39.22 2.8 0.2 0 0.23 0.01 0 89.35 8 4 71 2005 107 2054 11 50 29

Lhz (CE)

44.72 3.1 8.88 0.13 39.64 3.1 0.21 0.01 0.19 0.01 0 89.84 8 7 44 1838 116 2581 36 44 15 44.22 2.62 9.3 0.14 41.02 2.37 0.13 0.01 0.18 0.01 0 89.73 6 2 60 2002 112 2234 10 51 27

Lhz (P) Hz (P)

44.3 2.09 9.57 0.14 42.69 0.8 0.17 0.07 0.14 0.02 0 89.83 5 24 44 1916 111 2235 5 59 46 44.02 1.27 9.46 0.14 43.92 1.07 0.01 0.01 0.1 0.01 0 90.2 3 2 40 2659 118 2408 12 51 29

Hz (G) Per (CE)

44.74 3.36 9.13 0.15 38.89 2.72 0.39 0.05 0.21 0.03 0.32 89.4 16 111 82 2297 103 1991 11 58 66 44.91 3.45 9.03 0.14 38.50 3.32 0.14 0.02 0.20 0.01 0.27 89.41 10 392 85 2901 153 2694 24 86 69

Lhz (G) Lhz (G)

44.41 3.44 8.79 0.14 38.56 2.94 0.14 0.02 0.2 0.02 1.36 89.68 7 27 75 2598 105 2060 27 56 94 44.22 3.31 9.35 0.15 39.07 2.93 0.15 0.16 0.18 0.02 0.44 89.22 8 78 70 2493 109 2099 14 64 90

Lhz (P) Wh (CE)

41.95 3.03 12.01 0.2 35.87 5.46 0.4 0.12 0.49 0.04 0.43 85.54 41 98 93 2064 104 1775 20 90 40 44.95 2.93 8.97 0.14 38.76 3.00 0.06 0.07 0.17 0.02 0.91 89.53 8 415 70 2754 147 2655 25 83 118

Lhz (G) Hz (P)

Ib 15 N Ib 10 Ib 15 Ib 14 Ib 21 N Ib13 N

43.70 2.35 11.64 0.18 39.44 1.83 0.12 0.08 0.18 0.05 0.4 87.03 14 208 61 3085 112 2146 5 106 133 41.34 1.68 10.12 0.18 40.51 3.95 0.17 0.03 0.4 0.12 1.51 88.8 40 208 66 2111 113 2167 38 81 138

Wh (G) Hz (P)

46.13 1.65 8.31 0.14 42.18 1.33 0.10 0.04 0.09 0.03 0 90.95 7 38 45 3052 106 2203 5 60 53 SiO2 % Al2O3 Fe2O3 MnO MgO CaO Na2O K2O TiO2 P2O5 LOI Mg# Zr (ppm) Sr V Cr Co Ni Cu Zn Ba

Lhz (G)

Lhz (P)

45.03 3.45 8.74 0.13 38.95 3.17 0.3 0.01 0.2 0.01 0 89.82 7 36 75 2403 105 2013 22 54 48

Taf 19 Taf 50 Taf 36 Taf 42 Taf 2 Taf 43 Taf 12 Taf 38 Taf 51

Taf 37

Tafraoute maar

Ib1 Ib12 N

Ib 4 Bou-Ibalghatene maar

Table 3 Whole-rock major and trace-element analyses run by XRF for Tafraoute and Bou Ibalghatene mantle xenoliths. LOI = lost of ignition; Mg# = 100 ∗ Mg / (Mg + Fe2+) with total iron recasted as Fe2+.

Lhz (CE/P)


Taf 21 b

42

clinopyroxene (88.6–92.0) compared with Bou Ibalghatene (85.2– 90.6). The two localities largely overlap with regard to the TiO2 content of clinopyroxene, but the Bou Ibalghatene xenoliths are more variable. The lowest value (0.03) is observed in the most refractory harzburgite sample Ib12N whereas the highest value (0.98) is specific of the amphibole- and phlogopite-bearing wehrlite Ib14. Spinel shows variable Cr# values (=100 × Cr / (Cr + Al) in a.p.f.u.) between 8.3 and 30.4 with no significant difference between Bou Ibalghatene and Tafraoute. Both localities show a bimodal distribution of Cr# in spinel, with low values (b12.1) in fertile, amphibole-free lherzolites and a higher range (N17.5) in harzburgites, in the Opxpoor lherzolite Taf21b from Tafraoute, and in the Bou Ibalghatene wehrlites. Finally, as previously noted by Raffone et al. (2009), the Bou Ibalghatene amphiboles show a compositional range from pargasite to kaersutite, with Mg# values in the range 82.6–87.8.

8. Equilibrium temperatures Equilibrium temperatures were calculated using the geothermometers of Wells (1977) and Brey and Köhler (1990), based on the two-pyroxene solvus. The two methods provided consistent results within error ranges (Table 2). The Bou Ibalghatene and Tafraoute temperature estimates (865–1040 °C and 780–1150 °C, respectively) largely overlap but the temperature range of the Tafraoute suite is wider, compared with Bou Ibalghatene. Most importantly, relationships between temperature estimates and mineralogical or microstructural rock types indicate that the two suites have registered markedly distinct thermal evolutions. Our temperature estimates for the Bou Ibalghatene xenoliths are comparable to those reported by Raffone et al. (830–1030 °C). In detail, however, our values fall into two different sub-ranges that are clearly related to rock types: a relatively low-temperature group (870– 890 °C) recorded by the amphibole-free lherzolites (samples Ib10, Ib15N and Ib21N) and a higher-temperature group (970–1040 °C) characteristic of the amphibole-bearing peridotites, which includes an amphibole-rich peridotite (sample Ib4) and two amphibole harzburgites (samples Ib13N and Ib12N). No significant difference is observed between porphyroclasts and neoblasts. As discussed in detail by Kourim et al. (2014) for mantle xenoliths from the Hoggar swell (Algeria), the existence of a correlation between equilibrium temperatures and modal metasomatism in the Bou Ibalghatene suite is more likely explained by small-scale, advective heating of wall-rock peridotites rather than by a deeper origin of the higher-temperature, amphibole-bearing xenoliths. The latter alternative would imply that the lithospheric mantle beneath Bou Ibalghatene is modally and chemically stratified, being mostly composed of harzburgites, wehrlites and amphibole-rich peridotites at depth. This scenario which is very unlikely since the geochemical characteristics of these rocks are more readily explained by small-scale processes involving channelled porous flow and wall–rock reactions (Kourim et al., 2014, in press; Raffone et al., 2009). The Tafraoute temperature estimates also define two sub-ranges but, in contrast with Bou Ibalghatene, the difference between the two groups is clearly correlated with textural types and/or grain size. The higher temperatures (910–1150 °C) are recorded by coarse equant peridotites (samples Taf12 and Taf21b), porphyroclast grains in porphyroclastic xenoliths (samples Taf2, Taf38, and Taf37), as well as by the cores of relictual porphyroclasts in granular and mylonitic peridotites (samples Taf50 and Taf19, respectively). Conversely, the low temperature estimates (780–910 °C) are yielded by the recrystallized neoblasts of the granular and mylonitic peridotites (samples Taf19, Taf50, Taf36, Taf42, Taf 43, and Taf51). The observed temperature systematics indicates that the Bou Ibalghatene mantle xenoliths have undergone heating associated with amphibole-forming metasomatism (Raffone et al., 2009), from relatively low-temperature primary conditions (b900 °C). In contrast, the


43

Table 4 Trace-element content measured by LA-ICP-MS on hand-picked clinopyroxenes from Tafraoute and Bou Ibalghatene mantle xenoliths. Mean values and standard deviations are also reported for 6 analyses of the BIR-1G glass. Bou Ibalghatene maar

Rb (ppm) Sr Zr Nb Ba La Ce Pr Nd Sm Eu Ti Gd Tb Dy Ho Er Tm Yb Lu Hf Ta Th U

Ib 12 N

Ib 1

Ib13N

Ib21N

Ib 14

Ib 15

Ib 10

Ib 15N

Ib 4

Hz (P)

Wh (G)

Hz (P)

Lhz (G)

Wh (CE)

Lhz (P)

Lhz (G)

Lhz (G)

Per (CE)

0.004 430 35.4 2.63 0.05 17.39 57.67 8.83 41.05 8.44 2.50 228 6.99 0.90 4.74 0.80 1.94 0.25 1.50 0.21 0.37 0.529 0.750 0.149

0.009 923 31.0 0.42 0.16 35.79 111.72 14.76 56.64 9.11 2.71 311 7.64 0.94 5.24 0.99 2.59 0.38 2.44 0.37 0.91 0.044 1.081 0.232

0.000 230 30.2 0.78 0.05 13.44 40.06 5.89 26.35 5.11 1.53 952 4.55 0.65 3.91 0.75 2.04 0.28 1.77 0.27 0.86 0.061 0.738 0.125

0.139 342 25.2 0.02 5.07 33.41 58.21 5.21 15.61 2.02 0.66 2225 2.58 0.38 2.66 0.56 1.61 0.25 1.56 0.24 1.12 0.0005 4.627 3.677

0.035 260 66.4 0.70 1.35 6.03 20.58 3.71 20.06 5.34 1.78 2638 4.67 0.68 3.76 0.66 1.60 0.22 1.26 0.18 1.71 0.225 0.335 0.106

0.187 277 20.7 1.10 0.10 24.40 50.52 5.96 21.35 3.56 1.17 1168 3.38 0.49 3.00 0.60 1.63 0.24 1.50 0.23 0.70 0.077 3.010 0.690

0.064 179 27.7 0.14 5.87 47.83 28.79 1.37 4.52 1.64 0.69 3904 2.55 0.49 3.54 0.78 2.22 0.33 2.10 0.32 1.10 0.003 2.749 1.440

0.019 104 26.4 0.01 3.90 10.13 9.98 0.86 3.98 1.57 0.65 3603 2.17 0.46 3.45 0.75 2.20 0.33 2.09 0.31 1.03 0.0004 1.378 0.500

0.004 363 54.0 1.67 0.07 23.31 57.87 7.10 26.90 4.74 1.48 961 4.51 0.64 3.93 0.77 2.03 0.29 1.83 0.27 1.22 0.411 2.197 0.379

Tafraoute maar

Rb (ppm) Sr Zr Nb Ba La Ce Pr Nd Sm Eu Ti Gd Tb Dy Ho Er Tm Yb Lu Hf Ta Th U

Glass material

TAF 51

Taf 37

Taf 38

Taf 12

Taf 43

Taf 2

Taf 42

Taf 36

TAF 50

Taf19

Taf 21 b

Taf 21 a

BIR-1G

Hz (G)

Hz (P)

Lhz (P)

Lhz (CE)

Lhz (G)

Lhz (P)

Lhz (My)

Lhz (G)

Lhz (G/P)

Lhz (My)

Lhz (CE/P)

Wb (P)

n=6

0.003 62 19.8 0.17 0.06 2.97 4.93 0.83 4.59 1.29 0.45 1643 1.49 0.28 2.03 0.45 1.34 0.21 1.40 0.21 0.54 0.035 0.323 0.241

0.029 85 19.9 0.43 2.00 10.65 19.80 1.60 5.38 1.69 0.70 4272 2.58 0.50 3.73 0.82 2.32 0.35 2.23 0.34 0.96 0.014 0.999 0.727

0.008 42 29.9 0.02 0.11 0.51 2.81 0.61 3.83 1.60 0.68 3482 2.33 0.46 3.32 0.72 2.03 0.31 1.95 0.29 1.17 0.001 0.002 0.007

0.005 277 27.4 0.28 0.08 21.91 43.78 4.44 14.99 2.49 0.87 2329 2.73 0.45 3.11 0.66 1.86 0.27 1.74 0.26 0.96 0.007 2.677 0.563

0.008 34 21.1 0.03 0.01 0.32 1.89 0.51 3.61 1.61 0.68 3913 2.22 0.48 3.48 0.77 2.18 0.32 2.05 0.31 0.95

0.006 49 36.4 0.02 0.15 0.64 3.34 0.73 4.51 1.86 0.78 4191 2.49 0.52 3.59 0.78 2.19 0.32 2.06 0.31 1.31 0.001 0.001 0.001

0.047 98 37.5 0.16 4.03 3.37 7.90 1.09 5.25 1.68 0.68 2636 2.25 0.45 3.20 0.69 1.98 0.29 1.86 0.29 1.29 0.005 0.154 0.094

6.643 184 15.4 7.17 49.25 9.34 33.37 4.87 21.10 4.41 1.47 4217 4.28 0.71 4.62 0.95 2.62 0.39 2.49 0.36 0.48 0.047 0.747 0.110

0.188 167 23.3 0.19 11.72 14.96 16.08 1.16 4.66 1.50 0.61 3979 2.21 0.43 3.11 0.69 1.98 0.29 2.06 0.29 0.886 0.002 0.562 0.657

0.222 85 27.3 0.18 6.05 3.90 5.99 0.79 4.40 1.71 0.74 4575 2.22 0.43 3.09 0.67 1.88 0.26 1.89 0.25 0.84 0.014 0.091 0.086

0.007 44 11.3 0.05 0.08 0.60 2.44 0.45 2.38 0.74 0.34 1314 0.95 0.21 1.48 0.33 0.94 0.14 0.87 0.13 0.44 0.008 0.034 0.026

0.004 51 13.6 0.08 0.10 1.44 4.14 0.64 3.06 0.89 0.39 1349 1.15 0.23 1.72 0.38 1.08 0.15 0.98 0.15 0.53 0.009 0.084 0.030

0.210 108 12.65 0.52 6.70 0.65 2.10 0.40 2.51 1.11 0.54 6307 1.74 0.36 2.71 0.60 1.77 0.27 1.73 0.27 0.57 0.041 0.033 0.021

0.006 0.053

Tafraoute suite records cooling from higher-temperature primary conditions (N900 °C). 9. Whole-rock major-element compositions As a whole, the studied mantle xenoliths encompass almost the whole range of major-element compositions reported for off-craton mantle xenoliths (Pearson et al., 2003). When the wehrlites from Bou Ibalghatene and the olivine websterite Taf21a from Tafraoute are disregarded, the two suites of xenoliths show similar ranges of majorelement compositions, with Al2O3 varying mostly between 1 and 4%, MgO between 36 and 44%, FeO between 7.7 and 11%, and CaO between 0.8 and 4.4% (Table 3). Mg# varies between 87 and 91, but the lowest

1SD

0.012 1 0.14 0.01 0.15 0.01 0.02 0.00 0.03 0.04 0.01 130 0.07 0.01 0.03 0.01 0.02 0.01 0.03 0.00 0.01 0.002 0.002 0.001

and highest values are yielded by two harzburgites from Bou Ibalghatene (samples Ib13N and Ib12N, respectively). The lherzolites show a narrower variation range, from 88.6 to 90.2. The compositional arrays of Middle Atlas mantle xenoliths on the whole-rock Al2O3 covariation diagram (Fig. 7) closely resemble those of mantle xenoliths from the Hoggar swell, southern Algeria (Beccaluva et al., 2007; Dautria, 1988; Dupuy et al., 1986; Kourim et al., 2014, in press), including the anomalously high CaO, FeO, and/or TiO2 contents observed in some samples. MgO is negatively correlated with Al2O3 whereas CaO and TiO2 are positively correlated, in spite of some scatter. The scatter is more important in the Bou Ibalghatene suite where it reflects the selective enrichment of CaO, FeO, and/or TiO2 in wehrlites, as well as FeO in the harzburgite Ib13N. In Tafraoute,

44


Fig. 5. Crystallographic preferred orientations (CPO) of olivine in Tafraoute and Bou Ibalghatene peridotites. The selected Tafraoute samples are representative of the coarse-equant, porphyroclastic, granular and fine-grained microstructures. The two samples from Bou Ibalghatene are representative of the coarse-equant and granular microstructures typical of this locality. The fine-grained samples show a weak but well defined CPO characterized by an [010]-fiber symmetry. The other samples from Tafraoute and those from Bou Ibalghatene display an orthorhombic symmetry with a tendency towards an [010]-fiber symmetry in some samples. Lower hemisphere equal area stereographic projections, contours at multiples of uniform distribution.

the scatter is limited to CaO, which is depleted relative to Al2O3 in the harzburgite sample Taf57 (Al2O3/CaO = 0.38) and enriched in the wall–rock lherzolite Taf21b (Al2O3/CaO = 2.05). In term of Al2O3/CaO ratio, the orthopyroxene-poor lherzolite Taf21b resembles the Bou Ibalghatene wehrlites (Al2O3/CaO = 1.8–2.35). Other samples have Al2O3/CaO ratios comparable to the majority of mantle rocks, with slightly higher values in harzburgites (1.2–1.3) compared with lherzolites (0.85–1.2). The olivine websterite sample Taf21a (not

Fig. 6. Fabric strength index (J) vs. mean grain size for the Tafraoute xenoliths. J shows a clear tendency to decrease with mean grain size, from the coarse-equant to the finegrained, mylonitic microstructure. CPO weakening in porphyroclastic and granular peridotites likely results from dynamic recrystallization. J values b 2 are characteristic of the fine-grained samples in which the deformation microstructure is mostly governed by dissolution–crystallization processes ascribed to incongruent melting.

shown on Fig. 7) is distinguished by a more fertile composition typical of mantle pyroxenites (Al2O3 = 10.1%; MgO = 26.6%). However, this sample shows a relatively high Mg# value (89.4) and low TiO2 content (0.27%) suggesting chemical equilibrium with host peridotite. No relationship is observed between major-element variations and microstructures or equilibrium temperatures. 10. Clinopyroxene trace-element compositions On chondrite-normalized diagrams (Fig. 8a–d), Rare Earth Element (REE) analyses of hand-picked clinopyroxene grains show a variation from light and middle REE- (LREE- and MREE-) enriched compositions distinguished by a hump-shaped pattern for the La–Sm segment (group 1 — Fig. 8a) to LREE-depleted compositions with a flat heavy REE (HREE) segment from Eu to Lu (group 4 — Fig. 8d), through a range of signatures characterized by selective enrichment of LREE and MREE (group 2 — Fig. 8b), or only LREE (group 3 — Fig. 8c). Such a spectrum of chondrite-normalized REE patterns has been observed in several suites of spinel peridotites worldwide, both in clinopyroxene and whole rocks (see, e.g., Kourim et al., 2014, in press, for the Hoggar swell, southern Algeria). It has been ascribed by Navon and Stolper (1987) to the chromatographic fractionation of REE as a result of melt percolation through mantle peridotites. The four sample groups distinguished on Fig. 8 on the basis of REE patterns fall in four distinct arrays on the (La/Sm)N vs. (Sm/Yb)N diagram (Fig. 9a) that compares the fractionation of LREE (La) and HREE (Yb) from MREE (Sm): - group 1 is characterized by high (Sm/Yb)N values (≥ 2) but nearly chondritic (La/Sm)N ratio;


45

Fig. 7. MgO, FeO, CaO, and TiO2 vs. Al2O3 covariation diagram for the Tafraoute and Bou Ibalghatene mantle xenoliths, compared with samples from the Hoggar volcanic swell (shaded area — Beccaluva et al., 2007; Dautria, 1988; Dupuy et al., 1986; Kourim et al., 2014). Solid symbols = Tafraoute; empty symbols = Bou Ibalghatene; circles = lherzolites, squares = harzburgites, triangles = wehrlites, stars = data from Wittig et al. (2010a) for the Bou Ibalghatene mantle xenoliths.

- group 2 has lower (Sm/Yb)N but higher (La/Sm)N ratio compared with group 1; it is intermediate between group 1 and the LREEenriched apex of group 3 array; - group 3 is characterized is nearly-chondritic (Sm/Yb)N but a wide range of supra-chondritic (La/Sm)N ratios, up to ~ 20; - group 4 also shows nearly-chondritic (Sm/Yb)N values, like group 2, but differs by sub-chondritic (La/Sm)N values, down to ~0.1.

Significant differences are however observed between the Tafraoute and Bou Ibalghatene suites. The clinopyroxenes from Tafraoute show much lower LREE contents, on average, compared with Bou Ibalghatene clinopyroxenes (Fig. 9b). In detail, nearly half of the Tafraoute samples show LREE-depleted clinopyroxene compositions (group 4) whereas these compositions are absent from the Bou Ibalghatene suite (Figs. 8d and 9a). Conversely, LREE- and MREE-enriched clinopyroxenes (groups 1 and 2) are subordinate in Tafraoute whereas they are predominant in Bou Ibalghatene (Figs. 8a,b and 9a). Furthermore, in a given group the clinopyroxenes from Tafraoute tend to be less enriched in LREE than those from Bou Ibalghatene, a difference which is especially apparent in group 3 where the two suites show clearly distinct (La/Sm)N ratios (Figs. 8c and 9a). Finally, the LREEand MREE-enriched clinopyroxenes from Tafraoute (groups 1 and 2) are distinguished from their counterparts from Bou Ibalghatene by less fractionated MREE and HREE, a feature which is reflected in lower (Sm/Yb)N values (b2.5 — Fig. 9a). A clear relationship is observed in the Bou Ibalghatene suite between REE distribution in clinopyroxene and rock types since the most LREE- and MREE-enriched compositions (group 1) are specific of harzburgites (samples Ib12 and Ib13 — Table 1) and wehrlites (samples Ib1 and Ib14 — Table 1). These two rock types were interpreted by Raffone et al. (2009) as melt–rock reaction products related to infiltration of sub-lithospheric melts in the lithosphere. Such relationship is not observed in the Tafraoute suite, where one of the two harzburgites

(sample Taf37) falls in group 1 but the other one (samples Taf51) belongs to group 3. However, the Tafraoute xenoliths show a clear relationship between LREE enrichment in clinopyroxene and the presence of interstitial micro-veins of micro-aggregates: the latters are absent from the LREE-depleted samples whereas they are present in the others, with no exception. The olivine websterite Taf21a and its host peridotite Taf21b belong to group 4. They are however distinguished from the other samples in this group by lower MREE and HREE contents and flatter REE patterns. Except for a slight difference in LREE contents, the pyroxenite and its host peridotite show almost identical REE patterns in clinopyroxene, which confirms that they are in chemical equilibrium. The Bou Ibalghatene xenoliths also show some relationship between REE distribution and equilibrium temperatures since all peridotites with LREEand MREE-enriched clinopyroxenes (group 1) belong to the highertemperature group recognized in this suite. Such relationship is not observed in Tafraoute and the REE distribution shows no relationship with microstructures in any of the xenolith suites. Fig. 8e–h displays the distribution of other incompatible trace elements in clinopyroxene grains, normalized to Primitive Mantle values. These elements show systematic variations in relation with the different types of chondrite-normalized REE patterns, as well as more subtle differences between the Bou Ibalghatene and Tafraoute suites. Among the High-Field Strength Elements (HFSE: Nb, Ta, Zr, Hf, and Ti), Zr, Hf, and Ti show negative anomalies that are conspicuous in MREEenriched clinopyroxenes (group 1, and group 2 to a lesser degree — Fig. 8e,f) but hardly significant in the others (groups 3 and 4 — Fig. 8g,h). Nb and Ta also show negative anomalies relative to neighbouring elements, but these anomalies are much more pronounced in clinopyroxenes that are selectively enriched in LREE relative to MREE (groups 2 and 3 — Fig. 8f,g). These clinopyroxenes are also more enriched in Th and U. Similar to LREE, Th and U are on overall much more enriched in clinopyroxenes from Bou Ibalghatene, compared with Tafraoute (Fig. 9c). As previously observed in xenolith suites

46


Fig. 8. Chondrite-normalized REE (a, b, c, d) and primitive mantle-normalized trace-element (e, f, g, h) patterns for clinopyroxene from the Tafraoute and Bou-Ibalghatene mantle xenoliths. Groups 1–4 are distinguished on the basis of REE distributions as defined in the text. Symbols as in fig. 7, except for the solid diamonds that represent the websterite sample Taf21a from Tafraoute. Normalizing values after McDonough and Sun (1995).

(e.g., Alard et al., 2011, and references herein), U may be selectively enriched relative to Th in LREE-depleted clinopyroxenes (Fig. 8h). Except for subtle negative anomalies in group 1 and a positive anomaly in one of group 3 clinopyroxenes, Sr is generally well interpolated between Pr and Nd. Rb and Ba are strongly depleted relative to other highly incompatible elements. A few exceptions are observed among the Tafraoute xenoliths, which notably include the two samples that belong to group 1 (Fig. 8c). 11. Discussion Our study reveals that the mantle xenoliths brought to the surface by the Bou Ibalghatene and Tafraoute maars differ in several respects. The Bou Ibalghatene suite shows a wide range of modal compositions including refractory peridotites and wehrlites formed by

melt–rock reactions (Raffone et al., 2009), as well as modallymetasomatized peridotites (e.g., the amphibole-rich peridotite sample Ib4 — Table 1). The xenoliths also bear a strong metasomatic imprint in their geochemistry: several samples are selectively enriched in Fe and/or Ti, and all of them are enriched in LREE, Th, U, and other incompatible trace elements. These features were already noted in previous studies and ascribed to infiltration of sub-lithospheric alkaline partial melts in response to Cenozoic mantle upwelling (Raffone et al., 2009; Natali et al., 2013). Our results further indicate that the metasomatism was associated with heating of lithospheric peridotites. Except for the LREE- and MREE-enriched harzburgite sample Taf37 that is somewhat akin to the Bou Ibalghatene harzburgites — although less enriched in REE (Fig. 8a), the Tafraoute xenoliths do not include reacted rock types comparable to the Bou Ibalghatene suite of refractory peridotites and wehrlites (Raffone et al., 2009). The orthopyroxenepoor, wall–rock lherzolite Taf21b shows some resemblance with the


47

Atlas lithospheric mantle, most likely during Mesozoic rifting. Then, we compare the Bou Ibalghatene and Tafraoute suites in term of metasomatic imprint to evaluate to which degree and which extent the Middle Atlas lithospheric mantle was modified by sub-lithospheric melt during the Cenozoic.

11.1. Deformation assisted by lithospheric incipient melting in Tafraoute mantle xenoliths: a record of lithospheric thinning during Mesozoic rifting? The variety of microstructures displayed by the Tafraoute xenoliths (Fig. 3) record different stages of deformation most probably associated with decompression and lithospheric thinning. Porphyroclastic samples likely derive from coarse-equant peridotites through recrystallization. The presence of close sub-boundaries with a frequent fan-like disposition in porphyroclasts of partially recrystallized samples, in contrast with the spaced sub-boundaries observed in coarse-equant xenoliths, suggests that they were deformed under relatively high stress conditions. Recrystallization occurred both through subgrains rotation and via germination and growth of new grains, with a prevalence of the latter mechanism in the most recrystallized samples. New grains in those samples are polygonal and tend to form a foam-like microstructure with many 120°-triple junctions. Granular xenoliths likely represent the most evolved stage of this evolution. They are characterized by equilibrated microstructures, with most olivine crystals polygonal and free of any substructures, and contain only rare remnants of porphyroclasts. Germination and growth of polygonal new grains require efficient diffusion, a condition which is hardly consistent with deformation under high stress conditions. These observations may be reconciled by assuming deformation associated with decompression. As shown in Fig. 10, this process will result in an increase of the T/Tm ratio (Tm = melting temperature), particularly in peridotites containing a small amount of water (~0.1% — Inoue, 1994, and references herein). At T/Tm approaching 1, new grains germination and growth become active and may predominate locally. In this scenario, no heating is needed to account for the different olivine microstructures.

Fig. 9. Covariation of the chondrite-normalized La/Sm and Sm/Yb ratios (a) and plots of the sum of chondrite-normalized LREE (La, Ce, and Pr) and (Th + U) abundances vs. La/ Sm ratio in clinopyroxene from the Tafraoute and Bou-Ibalghatene mantle xenoliths. Symbols as in Fig. 8. Normalizing values after McDonough and Sun (1995).

Bou Ibalghatene wehrlites with regard to its modal composition (Fig. 2), but this sample is depleted in LREE (group 4 on Fig. 8d). Most importantly, amphibole is virtually absent from the Tafraoute suite and nearly half of the selected samples are LREE-depleted while the others are on overall much less enriched in LREE and Th–U than their counterparts from Bou Ibalghatene (Fig. 9). Hence, the Tafraoute xenoliths show only limited imprint of mantle metasomatism and the metasomatic agent involved was only mildly enriched in incompatible trace elements. In addition, the Tafraoute suite shows no trace of the heating event recorded by the Bou Ibalghatene xenoliths. Instead, the samples record cooling from relatively high-temperature conditions and recrystallization of garnet peridotites into the spinel stability field. Finally, the Tafraoute suite is also distinguished by the presence of peculiar, fine-grained peridotites (mylonites) showing textural evidence for orthopyroxene dissolution associated with olivine precipitation. In the following we first discuss the significance of the Tafraoute xenoliths and propose that the suite chiefly records thinning of the Middle

Fig. 10. P–T diagram showing the inferred primary and secondary equilibrium conditions inferred for the Tafraoute (blue) and Bou Ibalghatene (orange) mantle xenoliths. The dotted curves represent the spinel–garnet transition after O'Neill (1981). Dry and hydrous (0.05 and 0.1 wt.% H2O) solidi after Ringwood (1975), and amphibole–lherzolite solidus after Green et al. (2010). Dry solidus and H2O-saturated solidus are after Ringwood (1975). Geotherms for continental lithospheres of 65 or 115 km thick after Turcotte and Schubert (1982).

48


Microstructural evidence for orthopyroxene dissolution and olivine precipitation is observed in all fine-grained xenoliths (mylonites) and some porphyroclastic and granular samples (Fig. 4). These samples contain significantly smaller polygonal olivine, either pervasive (in mylonites) or confined in elongated domains (in porphyroclastic xenoliths). Olivine in fine-grained microstructures is associated with corroded orthopyroxene grains containing acicular olivine crystals and vein-like aggregates of olivine neoblasts. Some vein-like aggregates crosscutting throughout orthopyroxene porphyroclasts probably represent fractures sealed by olivine crystallized from infiltrated melt. In most cases, however, the aggregates were more likely formed as a result of olivine crystallization coupled with orthopyroxene dissolution. The fine-grained microstructures also contain abundant idiomorphic crystals of olivine grown at the expense of pre-existing olivine crystals (Fig. 4b,c). Growth of polyhedral olivine crystals with perfectly developed crystallographic faces in solid rocks usually occurs under high T/Tm conditions, in the presence of a fluid/melt phase (e.g., Boullier and Nicolas, 1975; Drury and Van Roermund, 1989; Vauchez et al., 2005). Incongruent melting of orthopyroxene, as suggested by the observed dissolution–crystallization microstructures, is consistent with experimentally determined melting reactions for the stability field of spinel (e.g., Inoue, 1994) and the highest temperatures yielded by the Tafraoute mantle xenoliths (T N 1000 °C — Fig. 10). Hence the Tafraoute mantle likely passed through wet solidus conditions during decompression and underwent incipient melting involving incongruent orthopyroxene dissolution. It is worth noting that no relationship is observed between melting microstructures and major-element chemistry that would indicate that the samples underwent significant melt extraction. The mylonite sample Taf19 is even distinguished by the most fertile composition among the analysed samples, with an anomalously high CaO content (4.43%) relative to Al2O3 (3.77%) suggesting that it was somewhat fertilized. These features are suggestive of a low melting degree and limited melt movement through deforming peridotites. Most olivine CPOs measured in Tafraoute xenoliths display either orthorhombic or [010]-fiber symmetry. The fine-grained, mylonitic samples yield a weak CPO, yet showing a clear [010]-fiber symmetry. At medium to high temperature and lithospheric pressure, the [010]-fiber symmetry results from transpressional deformation (e.g., Tommasi et al., 1999). However several studies have stressed that such CPO symmetry may also develop through compaction of molten peridotites (Higgie and Tommasi, 2012, 2014), even when the deformation accommodates normal faulting (Frets et al., 2014). The Tafraoute olivine CPOs also show a clear correlation between grainsize and fabric strength (Fig. 6): the finest the grain size, the weakest the preferred orientation. Weakening of the fabric strength with decreasing grain size is common in mantle rocks and generally ascribed to activation of grain boundary sliding, although subgrains rotation associated with dynamic recrystallization may also play a role. In the Tafraoute fine-grained peridotites, a major contribution to the CPOs dispersion comes however from the randomly oriented, fine-granular secondary olivine crystallized during incongruent melting (e.g., Tommasi et al., 2004). In addition, the presence of a melt fraction in the finely recrystallized microstructures probably contributed to the CPOs dispersion by favouring diffusion and rigid rotation of original crystals on wet grain boundaries. In several peridotite massifs where deformation occurred in the presence of melt, weakening of the olivine CPO and grain-size reduction are tightly related to melt–rock reactions (e.g., Dijkstra et al., 2002; Kruckenberg et al., 2013; Le Roux et al., 2008; Soustelle et al., 2009, 2010). Such evolution was also observed in experimentally deformed olivine-melt aggregates (e.g., Holtzman et al., 2003). It is frequently associated with strain localization, enhanced by feedback relationships: heterogeneous distribution of melt fractions favours strain localization and the resulting grain-size reduction will in turn enhance melt focusing in fine-grained domains and further strain localization.

When combined together, the microstructural, textural and petrologic characteristics of the Tafraoute xenoliths suggest that the lithospheric mantle beneath this locality has undergone an evolution marked by nearly isothermal decompression from the stability field of garnet, accompanied by incongruent melting of orthopyroxene and followed by cooling in the stability field of spinel (Fig. 10). In addition to the inferred primary and final conditions, the P–T trajectory proposed in Fig. 10 for the Tafraoute lithospheric mantle is constrained by studies in orogenic peridotites providing evidence that decompression associated with extensional or transtensional lithospheric thinning follows a nearly isothermal trajectory, before final cooling (e.g., Fabriès et al., 1998; Frets et al., 2014). Primary equilibrium pressure is poorly constrained; the fact that pyroxene–spinel symplectites are present only in some xenoliths suggests an origin at moderate depth, in the transition field between spinel and garnet peridotites (~ 2–3 GPa pressure — O'Neill, 1981). Given the range of primary temperature conditions recorded by the Tafraoute xenoliths (910–1150 °C) such a depth is consistent with a continental geotherm for a lithospheric thickness of ~115 km (Turcotte and Schubert, 1982). The observed olivine CPOs are typical of deformation under medium to high temperature (N 900 °C) and lithospheric pressure conditions supporting that the low temperatures recorded by the fine-grained Tafraoute samples (780–910 °C in the stability field of spinel peridotites) represent a post-deformation cooling stage. These final conditions roughly coincide with a geotherm for a 65 km-thick, thinned continental lithosphere (Fig. 10). Cooling down these final conditions had to be fast enough to allow the preservation of the very fine-grained microstructures observed in the Tafraoute xenoliths. Considering the location of the Tafraoute maar at the tectonic front of the ‘folded’ Middle Atlas, it seems logical to ascribe the extensional thinning of lithospheric mantle recorded by the Tafraoute mantle xenoliths to Mesozoic rifting. Inversion of Mesozoic continental rift in the Cenozoic led to building of the Middle Atlas orogen (Frizon de Lamotte, 2008). 11.2. Modification of Middle Atlas mantle lithosphere by sub-lithospheric melts in the Cenozoic (‘mantle metasomatism’): to which degree and which extent? Similar to mantle xenoliths from the Hoggar volcanic swell (Kourim et al., 2014), the Bou Ibalghatene peridotites illustrate a wide range of melt–rock interactions including: - near-solidus melt–rock reactions that generated a suite of harzburgites and wehrlites showing both refractory and enriched geochemical characters, such as high Cr# in spinel (Table 2) and LREE- and MREE-enrichment in clinopyroxene (Fig. 8a). These rocks record reactions at high temperature and high melt/rock ratio in porous flow channels or in wall rocks of melt conduits (Kourim et al., 2014; Raffone et al., 2009). - Fe–Ti metasomatism characterized by selective Fe ± Ti enrichment, essentially observed in the harzburgite–wehrlite suite (Fig. 7). This form of metasomatism has for long been ascribed to interaction of wall–rock peridotites with basaltic melts flowing into vein conduits (Bodinier et al., 1990; Menzies et al., 1987). - modal metasomatism, characterized by amphibole (± phlogopite) precipitation from percolating melt, a feature observed in 65% of the Bou Ibalghatene xenoliths according to Raffone et al. (2009), and in 5 of our selected samples, out of 9. - cryptic metasomatism, distinguished by selective enrichment of highly incompatible trace elements in peridotites that otherwise lack any imprint of melt–rock interactions. Cryptically metasomatised samples are often characterized by extreme fractionation of incompatible trace elements, a feature ascribed to chromatographic fractionation associated with percolation of small volume melts (Bodinier et al., 1990; Navon and Stolper, 1987). This is typically the case in the


49

Bou Ibalghatene xenoliths where the highest La/Sm ratios in clinopyroxene are observed in amphibole-free lherzolites (Figs. 8c and 9a). As discussed in previous works (Bedini et al., 1997; Bodinier et al., 2004; Kourim et al., 2014), these different forms of melt–rock interactions may record successive stages of differentiation of a melt infiltrated in lithospheric peridotites and percolating down a conductive thermal gradient. Variation of minerals saturation and volatiles content in melt may explain the different mineralogical reactions while chromatographic fractionation accounts for the whole spectrum of REE patterns. Combined with chromatography, gradual solidification of the melt through reactions at decreasing melt mass, also referred to as ‘percolative fractional crystallization’ (Harte et al., 1993), accounts for extreme enrichment of the most incompatible elements in evolved melt. This mechanism may explain the REE composition of Bou Ibalghatene equilibrium melts, calculated from clinopyroxene compositions and experimental Cpx/melt partition coefficients (Figs. 11 and 12). According to the chromatographic theory, group-1 equilibrium melts represent the most primitive melt compositions; they are nevertheless clearly distinguished from Middle Atlas basalts by more enriched LREE and MREE contents (Fig. 11), indicating that they have undergone some differentiation through reactions at decreasing melt mass. Being strongly enriched in LREE (Fig. 12a) and other highly incompatible elements such as Th (Fig. 12b), equilibrium melts calculated for groups 2 and 3 Bou Ibalghatene clinopyroxenes represent even more evolved melt compositions. In this scheme, the negative HFSE anomalies may result from the lower solubility of these elements in volatile-rich evolved melts (e.g., Ryerson and Watson, 1987; Tropper and Manning, 2005), possibly coupled with the precipitation of HFSE-bearing microphases (Bedini and Bodinier, 1999; Bodinier et al., 1996). The Bou Ibalghatene mantle xenoliths also record heating associated with melt–rock reactions and modal metasomatism. This relationship is strongly reminiscent of the correlation between equilibrium temperatures and metasomatic enrichments observed in the Hoggar xenoliths (Kourim et al., 2014). It lends support to the hypothesis that melt– rock reactions involving modal changes and/or modal metasomatism occurred in high-permeability channels and/or wall rocks of vein conduits whereas cryptic metasomatism represents more diffuse melt infiltration in host peridotites (Bodinier et al., 2004; Kourim et al., 2014; Raffone et al., 2009). In this scheme, advective heat transport along melt conduits accounts for the discrepancy of equilibrium temperatures between the two types of xenoliths. It is worth noting that the existence of small-scale temperature heterogeneities implies that thermal equilibrium was not achieved in the lithospheric mantle sampled by the Bou Ibalghatene xenoliths. This lends support to the assumption that

Fig. 11. Chondrite-normalized REE abundances in melts in equilibrium with group-1 clinopyroxenes (see text for definition) compared with Cenozoic alkali basalts from the Middle Atlas (shaded area, data from El Azzouzi et al., 2010). Equilibrium melts were calculated from Cpx LA-ICP-MS analyses and published Cpx/melt partition coefficients (Hart and Dunn, 1993; Kelemen et al., 1993). Symbols as in fig. 8. Normalizing values after McDonough and Sun (1995).

Fig. 12. Plots of the chondrite-normalized La/Sm and Th/La ratios vs. Sm/Yb for melts in equilibrium with clinopyroxenes from the Tafraoute and Bou Ibalghatene mantle xenoliths, compared with the Cenozoic alkali basalts from the Middle Atlas (shaded area, data from El Azzouzi et al., 2010). Symbols as in Fig. 8. Normalizing values after McDonough and Sun (1995).

the metasomatism occurred belatedly and was therefore related to the Cenozoic igneous activity. In contrast with the Bou Ibalghatene suite that was deeply modified by melt–rock interactions, the Tafraoute xenoliths show only limited imprint of mantle metasomatism. The olivine websterite Taf21a is chemically equilibrated with its LREE-depleted host peridotite. It is typical of the type 1b mantle xenoliths defined by Frey and Prinz (1978) and unlikely related to Cenozoic alkaline metasomatism. Moreover, this sample shows the same microstructural features as the Tafraoute peridotites and was therefore present in the Tafraoute mantle during the lithospheric thinning ascribed to Mesozoic rifting. Cenozoic metasomatism in the Tafraoute suite is thus restricted to about half of our selected samples showing moderate LREE ± MREE enrichment in clinopyroxene. These enrichments are clearly associated with occurrences of intergranular micro-aggregates forming a pervasive network at olivine–olivine grain boundaries. The development of this network is obviously late in the xenoliths evolution, indicating that the lithospheric mantle beneath Tafraoute was belatedly (and heterogeneously) percolated by small melt fractions related to Cenozoic igneous activity. Equilibrium melts calculated for group-1 Tafraoute clinopyroxenes show LREE contents in the range of the Middle Atlas alkaline basalts (Fig. 11). They are however distinguished by a relatively flat HREE segment compared with basalts and Bou Ibalghatene equilibrium melts, as illustrated by their lower Sm/Yb ratio in Fig. 12. Similar to the Bou Ibalghatene equilibrium melts, the Tafraoute metasomatic agents may derive from a silicate melt comparable to the Middle Atlas volcanism. However, the melt was buffered by the host peridotites for mildly incompatible trace elements such as HREE, which implies that percolation occurred at low melt/rock ratio. Similarly, major-element and thermal buffering of small melt fractions may explain that the Tafraoute suite lacks the mineralogical reactions and heating observed in the Bou Ibalghatene xenoliths. In turn, the lack of mineralogical reactions accounts for the absence of LREE and Th–U enrichment in percolating equilibrium melts relative to the basalts.

50


12. Summary and concluding remarks The Bou Ibalghatene and Tafraoute maars are the two main xenolith localities in the Middle Atlas – and in Morocco – and provide samples from the lithospheric mantle that were brought to the surface recently (b mid-Quaternary). The two localities show very similar volcanic structures but differ in their locations: while the Bou Ibalghatene maar belongs to the main Middle Atlas volcanic field, the Tafraoute maar is situated about 45 km away, on the North Middle Atlas fault that bounds the ‘folded’ Middle Atlas belt to the northwest (Fig. 1). Both xenolith suites record infiltration of sub-lithospheric melts that are akin to the Middle Atlas volcanism but were differentiated to variable degrees as a result of interactions with lithospheric mantle. However, while the Bou Ibalghatene mantle was densely traversed by high melt fractions, mostly focused in high-permeability channels or melt conduits, the Tafraoute suite records heterogeneous infiltration of smaller melt fractions that migrated diffusively, by intergranular porous flow. As a consequence the lithospheric mantle beneath Bou Ibalghatene was strongly modified by melt–rock interactions in the Cenozoic whereas the Tafraoute mantle preserves the record of extensional lithospheric thinning, most likely related to Mesozoic rifting. Among the processes involved in lithospheric thinning, our observations reveal that hydrous incongruent melting triggered by decompression probably played a key role in favouring strain localisation. The question that arises is whether the upper lithosphere sampled by the Bou Ibalghatene xenoliths was comparable to the Tafraoute mantle before the onset of metasomatism and also recorded Mesozoic thinning. The issue behind is to understand whether the present day abnormally thin lithosphere beneath the tabular Middle Atlas is mainly inherited from Mesozoic rifting, as recorded by the Tafraoute suite, or was partly, or dominantly, contributed by subsequent processes such as delamination or thermo-mechanical erosion related to Cenozoic asthenospheric upwelling. A key feature is that the Bou Ibalghatene xenoliths have preserved microstructural evidence of a high-temperature deformation comparable to the one registered by the Tafraoute xenoliths but show no trace of coeval high equilibrium temperatures in their mineral compositions. Instead, the three amphibole-free peridotites (considered to record ambient temperature, away from melt conduits) yield a narrow range of relatively low equilibrium temperatures (870–890 °C). This might be considered as evidence that the deformation observed in the Bou Ibalghatene xenoliths is unrelated to Mesozoic rifting and record an older event, followed by complete mineral equilibration at lower temperature conditions. However, two subtle but probably significant differences between Bou Ibalghatene and Tafraoute xenoliths should be noted: first, the ambient lithospheric temperature inferred from the amphibole-free Bou Ibalghatene xenoliths is somewhat higher than the final temperature conditions recorded by the Tafraoute xenoliths (Fig. 10) and, second, in spite of their overall resemblance with the Tafraoute deformation microstructures, those from Bou Ibalghatene tend to be more readily overprinted by static recrystallization, even in the amphibole-free samples. This may indicate that the ‘low-temperature’ primary conditions registered by the amphibolefree peridotites from Bou Ibalghatene actually record conductive heating of the lithosphere from somewhat lower-temperature conditions. Conductive heating was possibly favoured by advective heat transport through a dense network of melt conduits, as suggested by Kourim et al. (2014, in press) for the Hoggar. This event would have reset previous temperature conditions preserved in mineral compositions before the onset of Cenozoic mantle upwelling and igneous activity. Together with the strong metasomatic imprint registered by the Bou Ibalghatene mantle xenoliths suggesting that the maar is located above a zone of focused melt circulation, the proposed thermal evolution is consistent with the existence of local and narrow asthenospheric upwelling beneath the Azrou–Timahdite volcanic field. It may also be inferred that thermal erosion favoured and guided by dyking and/or high-permeability channels played a significant role in lithospheric

thinning. Taken as a whole, our finding suggest the existence of smallscale instabilities, a few km across, involving feedback relationships between (1) mantle upwelling into rifted lithosphere, (2) decompression melting and (3) thermal erosion of lower lithosphere enhanced by melt infiltration and advective heat transport. The Bou Ibalghatene and Tafraoute xenolith suites both record lithospheric thinning, but illustrate markedly distinct mechanisms that were possibly superimposed in Bou Ibalghatene. Our results lend support to the suggestion based on seismic tomography that lithospheric thinning beneath the Atlas mountains results from the combination of different mechanisms and occurred in piecewise fashion at a short wavelength scale (Bezada et al., 2014; Kaislaniemi and van Hunen). Acknowledgements This study was performed as part of a collaborative multidisciplinary research project on mantle rocks from Morocco involving the Faculty of Sciences of Meknes (University Moulay Ismail, Morocco), the Faculty of Sciences of Tetouan (University Abdelmalek Esaâdi, Morocco), Geosciences Montpellier (CNRS & University of Montpellier, France), and the “Instituto Andalus de Ciencias de la Tierra” (IACT, CSIC & UGR, Spain). Funding for research and/or participants mobility was provided by CNRS (INSU and DERCI, France), CNRST (Morocco), the French Ministry of Foreign Affairs (MAE), through the CNRS-CNRST bilateral cooperation project #24499 (2010–2012), the MAE cooperation project #042/STU/09 in the frame of the “Volubilis Hubert Curien” programme (2010–2013), three INSU research grants (“Actions Coordonnées” 2008, and SYSTER 2010 and 2011), and the FP7-PEOPLE-2013-IRSES project MEDYNA funded under Grant Agreement PIRSES-GA-2013-612572 (WP3 — Deep structures and mantle processes), and the International Lithosphere Program CC4-MEDYNA. C.J.G. acknowledges funding from the Spanish “Ministerio de Economía y Competitividad” (CGL2010-14848), and Junta de Andalucía (RNM131 and 2009RNM4495). This research has benefited from EU Cohesion Policy funding from the European Regional Development Fund (ERDF) and the European Social Fund (ESF) in support of innovation and research projects and infrastructures. This work greatly benefited from the expertise of Christophe Nevado and Doriane Delmas in the preparation of thin sections. Bernard Boyer and Claude Merlet are thanked for their assistance during electron probe microanalyses at the “Microsonde Sud” facility, and Fabrice Barou and Olivier Bruguier for their help during CPO determinations at the EBSD-SEM system and during LA-ICP-MS trace-element analyses, respectively. Appendix A. Supplementary data Supplementary data to this article (major elements in minerals analyzed by electron microprobe) can be found online at http://dx.doi.org/ 10.1016/j.tecto.2014.11.020. References Alard, O., Lorand, J.P., Reisberg, L., Bodinier, J.L., Dautria, J.M., O'Reilly, S.Y., 2011. Volatilerich metasomatism in Montferrier xenoliths (southern France): consequences for chalcophile and highly siderophile elements abundances in the sub-continental mantle. J. Petrol. 52, 2009–2045. Ayarza, P., Alvarez-Lobato, F., Teixell, A., Arboleya, M., Teson, E., Julivert, M., Charroud, M., 2005. Crustal structure under the central Hign Atlas Mountains (Morocco) from geological and gravity data. Tectonophysics 400, 67–84. Beauchamp, W., Allmendinger, R.W., Barazangi, M., Demnati, A., Alji, M.E., Dahmani, M., 1999. Inversion tectonics and the evolution oh the High Atlas Mountains, Morocco, based on a geological–geophysical transect. Tectonics 18, 163–184. Beccaluva, L., Azzouni-Sekkal, A., Benhallou, A., Bianchini, G., Ellam, R.M., Marzola, M., Siena, F., Stuart, F.M., 2007. Intracratonic asthenosphere upwelling and lithosphere rejuvenation beneath the Hoggar swell (Algeria): evidence from HIMU metasomatised lherzolite mantle xenoliths. Earth Planet. Sci. Lett. 260, 482–494. Bedini, R.M., Bodinier, J.L., 1999. Distribution of incompatible trace elements between the constituents of spinel peridotite xenoliths: ICP-MS data from the East African Rift. Geochim. Cosmochim. Acta 63, 3883–3900.

H. El Messbahi et al. / Tectonophysics 650 (2015) 34–52 Bedini, R.M., Bodinier, J.L., Dautria, J.M., Morten, L., 1997. Evolution of LILE-enriched small melt fractions in the lithospheric mantle: a case study from the East African Rift. Earth Planet. Sci. Lett. 153, 67–83. Berger, J., Liégeois, J.P., Ennih, N., Bonin, B., 2009. Flow of Canary mantle plume material through a subcontinental lithospheric corridor beneath Africa to the Mediterranean: comment. Geology 38, e202. Bezada, M.J., Humphreys, E.D., Davila, J.M., Carbonell, R., Harnafi, M., Palomeras, I., Levander, A., 2014. Piecewise delamination of Moroccan lithosphere from beneath the Atlas Mountains. Geochem. Geophys. Geosyst. 15, 975–985. Bodinier, J.L., Vasseur, G., Vernières, J., Dupuy, C., Fabriès, J., 1990. Mechanisms of mantle metasomatism: geochemical evidence from the Lherz orogenic peridotite. J. Petrol. 31, 597–628. Bodinier, J.L., Merlet, C., Bedini, R.M., Simien, F., Remaïdi, M., Garrido, C.J., 1996. Distribution of niobium, tantalum, and other highly incompatible trace elements in the lithospheric mantle: the spinel paradox. Geochim. Cosmochim. Acta 60, 545–550. Bodinier, J.L., Menzies, M.A., Shimizu, N., Frey, F.A., McPherson, E., 2004. Silicate, hydrous and carbonate metasomatism at Lherz, France: contemporaneous derivatives of silicate melt–harzburgite reaction. J. Petrol. 45, 299–320. Boullier, A.M., Nicolas, A., 1975. Classification of textures and fabrics of peridotite xenoliths from South African kimberlites. Phys. Chem. Earth 9, 467–475. Brey, G.P., Köhler, T., 1990. Geothermobarometry in four-phase lherzolites II. New thermobarometers, and practical assessment of existing thermobarometers. J. Petrol. 31, 1353–1378. Bunge, H.J., 1982. Texture Analysis in Materials Sciences. Buttleworth, London. Dautria, J.M., 1988. Relations entre les hétérogénéites du manteau supérieur et le magmatisme en domaine continental distensif. (Thèse d'Etat). Exemple des basaltes alcalins du Hoggar (Sahara central, Algérie) et de leurs enclaves. Université de Montpellier 2, p. 421. Dijkstra, A.H., Drury, M.R., Vissers, R.L.M., Newman, J., 2002. On the role of melt–rock reaction in mantle shear zone formation in the Othris peridotite massif (Greece). J. Struct. Geol. 24, 1431–1450. Drury, M.R., Van Roermund, H.L.M., 1989. Fluid assisted recrystatlization in upper mantle peridotite xenoliths from kimberlites. J. Petrol. 30, 133–152. Duggen, S., Hoernle, K., Bogaard, P.V.D., Rüpke, L., Phipps Morgan, J., 2003. Deep roots of the Messinian salinity crisis. Nature 422, 602–606. Duggen, S., Hoernle, K., Hauff, F., Klugel, A., Bouabdellah, M., Thirlwall, M., 2009. Flow of Canary mantle plume material through a subcontinental lithospheric corridor beneath Africa to the Mediterranean. Geology 37, 283–286. Dupuy, C., Dostal, J., Dautria, J.M., Girod, M., 1986. Geochemistry of spinel peridotite inclusions in basalts from Hoggar (Algeria). J. Afr. Earth Sci. 5, 209–215. El Azzouzi, M., Bernard-Griffiths, J., Bellon, H., Maury, R.C., Piqué, A., Fourcade, S., Cotten, J., Hernandez, J., 1999. Evolution des sources du volcanisme marocain au cours du Néogène. C. R. Acad. Sci. Paris 329, 95–102. El Azzouzi, M., Maury, R.C., Bellon, H., Youbi, N., Cotten, J., Kharbouch, F., 2010. Petrology and K–Ar chronology of the Neogene–Quaternary Middle Atlas basaltic province, Morocco. Bull. Soc. Géol. Fr. 181, 243–257. Fabriès, J., Lorand, J.P., Bodinier, J.L., 1998. Petrogenetic evolution of orogenic lherzolite massifs in the central and western Pyrenees. Tectonophysics 292, 145–167. Faccenna, C., Becker, T.W., 2010. Shaping mobile belt from small scale convection. Nature 465, 602–605. Frets, E.C., Tommasi, A., Garrido, C.J., Vauchez, A., Mainprice, D., Targuisti, K., Amri, I., 2014. The Beni Bousera peridotite (Rif Belt, Morocco): an oblique-slip low-angle shear zone thinning the subcontinental mantle lithosphere. J. Petrol. 55, 283–313. Frey, F.A., Prinz, M., 1978. Ultramafic inclusions from San Carlos Arizona: petrologic and geochemical data bearing on their petrogenesis. Earth Planet. Sci. Lett. 38, 129–176. Frizon de Lamotte, D., Zizi, M., Missenard, Y., Hafid, M., Azzouzi, M.E., Maury, R.C., Charrière, A., Taki, Z., Benammi, M., Michard, A., 2008. In: Michard, A., Saddiqi, O., Chalouan, A., Frizon de Lamotte, D. (Eds.), Continental Evolution: The Geology of Morocco. Springer, Berlin Heidelberg, pp. 133–202. Frizon de Lamotte, D., Leturmy, P., Missenard, Y., Khomsi, S., Ruiz, G., Saddiqi, O., Guillocheau, F., Michard, A., 2009. Mesozoic and Cenozoic vertical movements in the Atlas system (Algeria, Morocco, Tunisia): an overview. Tectonophysics 475, 9–28. Fullea, J., Fernandez, M., Zeyen, H., 2006. Lithospheric structure in the Atlantic–Mediterranean transition zone (southern Spain, northern Morocco): a simple approach from regional elevation and geoid data. In: Frizon de Lamotte, D., Saddiqi, O., Michard, A. (Eds.), Some Recent Developments on the Maghreb Geodynamics. Comptes Rendus Geoscience 338, pp. 140–151. Fullea, J., Fernandez, M., Afonso, J.C., Verges, J., Zeyen, H., 2010. The structure and evolution of the lithosphere–asthenosphere boundary beneath the Atlantic– Mediterranean transition Region. Lithos 120, 74–95. Gomez, F., Allmendinger, R., Barazangi, M., Er-Raji, A., Dahmani, M., 1998. Crustal shortening and vertical strain partitioning in the Middle Atlas Mountains of Morocco. Tectonics 17, 520–533. Govindaraju, K., 1994. 1994 complication of working values and sample descriptions for 383 geostandards. Geostand. Newslett. 18, 1–55. Green, D.H., Hibberson, W.O., Kovacs, I., Rosenthal, A., 2010. Water and its influence on the lithosphere–asthenosphere boundary. Nature 467, 448–451. Griffin, W.L., Powell, W.J., Pearson, N.J., O'Reilly, S.Y., 2008. GLITTER: data reduction software for laser ablation ICP-MS. In: Sylvester, P. (Ed.), Laser Ablation-ICP-MS in the Earth Sciences. Mineralogical Association of Canada Short Course Series 40, pp. 204–207. Hart, S.R., Dunn, T., 1993. Experimental Cpxx/melt partitioning of 24 trace elements. Contrib. Mineral. Petrol. 113, 1–8. Harte, B., Hunter, R.H., Kinny, P.D., 1993. Melt geometry, movement and crystallization, in relation to mantle dykes, veins and metasomatism. Philos. Trans. R. Soc. Lond. Ser. A 342, 1–21. Higgie, K., Tommasi, A., 2012. Feedbacks between deformation and melt distribution in the crust–mantle transition zone of the Oman ophiolite. Earth Planet. Sci. Lett. 359–360, 61–72.

51

Higgie, K., Tommasi, A., 2014. Deformation in a partially molten mantle: constraints from plagioclase lherzolites from Lanzo, western Alps. Tectonophysics 615–616, 167–181. Holtzman, B.K., Groebner, N.J., Zimmerman, M.E., Ginsberg, S.B., Kohlstedt, D.L., 2003. Stress-driven melt segregation in partially molten rocks. Geochem. Geophys. Geosyst. 4, 8607. http://dx.doi.org/10.1029/2001GC000258. Inoue, T., 1994. Effect of water on melting phase relations and melt composition in the system Mg2SiO4–MgSiO3–H2O up to 15 GPa. Phys. Earth Planet. Inter. 85, 237–263. Jiménez-Munt, I., Fernandez, M., Vergés, J., Garcia-Castellanos, D., Fullea, J., PérezGussinyé, M., Afonso, J.C., 2011. Decoupled crust–mantle accommodation of Africa– Eurasia convergence in the NW Moroccan margin. J. Geophys. Res. Solid Earth 116 (8). http://dx.doi.org/10.1029/2010JB008105. Jochum, K.P., Willbold, M., Raczek, I., Stoll, B., Herwig, K., 2005. Chemical characterisation of the USGS reference glasses GSA-1G, GSC-1G, GSD-1G, GSE-1G, BCR-2G, BHVO-2G and BIR-1G using EPMA, ID-TIMS, ID-ICPMS and LA-ICPMS. Geostand. Geoanal. Res. 29, 285–302. Jochum, K.P., Stoll, B., Herwig, K., Willbold, M., Hofmann, A.W., Aarburg, S., Amini, M., Abouchami, W., Raczek, I., Stracke, A., Alard, O., Bouman, C., Becker, S., Dücking, M., Brätz, H., Klemd, R., De Bruin, D., Canil, D., Cornell, D., De Hoog, C.-J., Dalpé, C., Danyushevsky, L., Gao, Y., Snow, J.E., Groschopf, N., Günther, D., Latkoczy, C., Guillong, M., Hauri, E., Höfer, H.E., Lahaye, Y., Jacob, D.E., Kasemann, S.A., Kent, A.J.R., Ludwig, T., Zack, T., Mason, P.R.D., Meixner, A., Rosner, M., Misawa, K., Nash, B.P., Pfänder, J., Premo, W.R., Sun, W.D., Tiepolo, M., Vannucci, R., Vennemann, T., Wayne, D., Woodhead, J.D., 2006. MPI-DING reference glasses for in-situ microanalysis: new reference values for element concentrations and isotope ratios. Geochem. Geophys. Geosyst. 7, QO2005. http://dx.doi.org/10.1029/2005GC001060. Kaislaniemi, L., van Hunen, J., 2014. Dynamics of lithospheric thinning and mantle melting by edge-driven convection: application to Moroccan Atlas mountains. Geochem. Geophys. Geosyst. 15, 3175–3189. Kelemen, P.B., Shimizu, N., Dunn, T., 1993. Relative depletion of niobium in some arc magmas and the continental crust: partitioning of K, Nb, La and Ce during melt/ rock reaction in the upper mantle. Earth Planet. Sci. Lett. 120, 111–134. Kourim, F., Bodinier, J.L., Alard, O., Bendaoud, A., Vauchez, A., Dautria, J.M., 2014. Nature and evolution of the lithospheric mantle beneath the Hoggar swell (Algeria): a record from mantle xenoliths. J. Petrol. 55, 2249–2280. Kourim, F., Vauchez, A., Bodinier, J.L., Alard, A., Bendaoud, A., 2014. Subcontinental lithosphere reactivation beneath the Hoggar swell (Algeria): localized deformation, melt channelling and heat advection. Tectonophysics (in press). Kruckenberg, S.C., Tikoff, B., Toy, V.G., Newman, J., Young, L.I., 2013. Strain localization associated with channelized melt migration in upper mantle lithosphere: insights from the Twin Sisters ultramafic complex, Washington, USA. J. Struct. Geol. 50, 133–147. Le Roux, V., Tommasi, A., Vauchez, A., 2008. Feedback processes between percolation and deformation in a frozen lithosphere–asthenosphere boundary. Earth Planet. Sci. Lett. 274, 401–413. Lenaz, D., Youbi, N., De Min, A., Boumehdi, M.A., Ben Abbou, M., 2014. Low intra-crytalline closure temperatures of Cr-bearing spinels from the mantle xenoliths of the Middle Atlas Neogene–Quaternary Volcanic Field (Morocco): a mineralogical evidence of cooler mantle beneath the West African Craton. Am. Mineral. 99, 267–275. Lustrino, M., Wilson, M., 2007. The circum-Mediterranean anorogenic Cenozoic igneous province. Earth Sci. Rev. 81, 1–65. McDonough, W.F., Sun, S.S., 1995. The composition of the Earth. Chem. Geol. 120, 223–253. Menzies, M.A., Rogers, N., Tindle, A., Hawkesworth, C.J., 1987. Metasomatic enrichment processes in lithospheric peridotites, an effect of asthenosphere–lithosphere interaction. In: Menzies, M.A., Hawkesworth, C.J. (Eds.), Mantle Metasomatism. Academic Press, London, pp. 313–361. Miller, M.S., Becker, T.W., 2014. Reactivated lithospheric-scale discontinuities localize dynamic uplift of the Moroccan Atlas Mountains. Geology 42, 35–38. Missenard, Y., Cadoux, A., 2012. Can Moroccan Atlas lithospheric thinning and volcanism be induced by edge-driven convection? Terra Nova 24, 27–33. Missenard, Y., Zeyen, H., Frizon de Lamotte, D., Leturmy, P., Petit, C., Sébrier, M., Saddiqi, O., 2006. Crustal versus asthenospheric origin of relief oh the Atlas Mountains of Morocco. J. Geophys. Res. Solid Earth 111 (B3). http://dx.doi.org/ 10.1029/2005JB0033708. Morishita, T., Arai, S., 2003. Evolution of spinel–pyroxene symplectite in spinel–lherzolites from the Horoman Complex, Japan. Contrib. Mineral. Petrol. 144, 509–522. Natali, C., Beccaluva, L., Bianchini, G., Ellam, R.M., Siena, F., Stuart, F.M., 2013. Carbonated alkali-silicate metasomatism in the North Africa lithosphere: evidence from Middle Atlas spinel–lherzolites, Morocco. J. S. Am. Earth Sci. 41, 113–121. Navon, O., Stolper, E., 1987. Geochemical consequences of melt percolation — the upper mantle as a chromatographic column. J. Geol. 95, 285–307. Nicolas, A., Lucazeau, F., Bayer, R., 1987. Peridotites xenoliths in Massif central Basalts, France: textural and geophysical evidence for asthenosheric diapirism. In: Nixon, P.H. (Ed.), Mantle Xenoliths. John Wiley and Sons, Publ, Chischester, pp. 563–574. O'Neill, H.S.C., 1981. The transition between spinel lherzolite and garnet lherzolite, and its use as a geobarometer. Contrib. Mineral. Petrol. 77, 185–194. Palomeras, I., Thurner, S., Levander, A., Lui, K., Villasenor, A., Carbonell, R., Harnafi, M., 2014. Finite-frequency Rayleigh wave tomography of the western Mediterranean: mapping its lithospheric structure. Geochem. Geophys. Geosyst. 15, 140–160. Pearce, N.J.G., Perkins, W.T., Westgate, J.A., Gorton, M.P., Jackson, E.E., Neal, C.R., Chenery, S.P., 1997. A compilation of new and published major and trace element data for NIST SRM 610 and NIST SRM 612 glass reference materials. Geostand. Newslett. 21, 115–144. Pearson, D.G., Canil, D., Shirey, S.B., 2003. Mantle samples included in volcanic rocks: xenoliths and diamonds. In: Carlson, R.W. (Ed.), The Mantle and Core. (Treatise on Geochemistry). Elsevier, Amsterdam, pp. 171–275.

52


Raffone, N., Chazot, G., Pin, C., Vannucci, R., Zanetti, A., 2009. Metasomatism in the lithospheric mantle beneath Middle Atlas (Morocco) and the origin of Fe- and Mg-rich wehrlites. J. Petrol. 50, 197–249. Ringwood, A.E., 1975. Composition and Petrology of the Earth's Mantle, Chap. 5. McGrawHill, New York. Ryerson, F.J., Watson, E.B., 1987. Rutile saturation in magmas — implications for Ti–Nb–Ta depletion in Island-Arc Basalts. Earth Planet. Sci. Lett. 86, 225–239. Shimizu, Y., Arai, S., Morishita, T., Ishida, Y., 2008. Origin and significance of spinel– pyroxene symplectite in lherzolite xenoliths from Tallante, SE Spain. Mineral. Petrol. 94, 27–43. Soustelle, V., Tommasi, A., Bodinier, J.L., Garrido, C.J., Vauchez, A., 2009. Deformation and reactive melt transport in the mantle lithosphere above a large-scale partial melting domain: the Ronda Peridotite Massif, Southern Spain. J. Petrol. 50, 1235–1266. Soustelle, V., Tommasi, A., Demouchy, S., Ionov, D.A., 2010. Deformation and fluid–rock interaction in the supra-subduction mantle: microstructures and water contents in peridotite xenoliths from the Avacha Volcano, Kamchatka. J. Petrol. 51, 363–394. Teixell, A., Arboleya, M.L., Julivert, M., Charroud, M., 2003. Tectonic shortening and topography in the central High Atlas (Morocco). Tectonics 22, 1051. Teixell, A., Ayarza, P., Zeyen, H., Fernandez, M., Arboleya, M., 2005. Effects of mantle upwelling in a compressional setting: the Atlas Moutains of Morocco. Terra Nova 17, 456–461. Teixell, A., Bertotti, G., Frizon de Lamotte, D., Charroud, M., 2009. The geology of vertical movements of the lithosphere: an overview. Tectonophysics 475, 1–8. Tommasi, A., Tikoff, B., Vauchez, A., 1999. Upper mantle tectonics: three-dimensional deformation, olivine crystallographic fabrics and seismic properties. Earth Planet. Sci. Lett. 168, 173–186.

Tommasi, A., Godard, M., Coromina, G., Dautria, J.M., Barsczus, H., 2004. Seismic anisotropy and compositionally-induced velocity anomalies in the lithosphere above mantle plumes: a petrological and microstructural study of mantle xenoliths from French Polynesia. Earth Planet. Sci. Lett. 27, 539–556. Tropper, P., Manning, C.E., 2005. Very low solubility of rutile in H2O at high pressure and temperature, and its implications for Ti mobility in subduction zones. Am. Mineral. 90, 502–505. Turcotte, D.L., Schubert, G., 1982. Geodynamics. Wiley, New York (450 pp.). Vauchez, A., Dineur, F., Rudnick, R., 2005. Microstructure, texture and seismic anisotropy of the lithospheric mantle above a mantle plume: insights from the Labait volcano xenoliths (Tanzania). Earth Planet. Sci. Lett. 232, 295–314. Wells, P.R.A., 1977. Pyroxene thermometry in simple and complex systems. Contrib. Mineral. Petrol. 62, 129–139. Wittig, N., Pearson, D.G., Baker, J.A., Duggen, S., Hoernle, K., 2010a. A major element, PGE and Re–Os isotope study of Middle Atlas (Morocco) peridotite xenoliths: evidence for coupled introduction of metasomatic sulphides and clinopyroxene. Lithos 115, 15–26. Wittig, N., Pearson, D.G., Duggen, S., Baker, J.A., Hoernle, K., 2010b. Tracing the metasomatic and magmatic evolution of continental mantle roots with Sr, Nd, Hf and Pb isotopes: a case study of Middle Atlas (Morocco) peridotite xenoliths. Geochim. Cosmochim. Acta 74, 1417–1435. Zeyen, H., Fernandez, M., 1994. Integrated lithospheric modeling combining thermal, gravity, and local isostasy analysis: application to the NE Spanish Geotransect. J. Geophys. Res. Solid Earth 99, 18089–18102.