Comptes Rendus Géoscience Sciences de la Planète

. The Mauritanian Tasiast unit in the Southwestern Reguibat Archean Shield (North of the West African craton = WAC) consists mainly of gneiss dated between 3.07 and 2.91 Ga. We present new ﬁeld and petrographic observations combined with whole-rock geochemical data on gneisses of Tasiast to understand their petrogenesis, tectonic setting, and the evolution of the continental crust of WAC. These data provide ﬁrm evidence of two distinct orthogneisses: trondhjemite and high-K granite a ﬃ nities. Geochemical characters suggest that (1) trondhjemites magma source were originally derived from polybaric partial melting at the thickened crust occurred over a range of P – T conditions, covering the stability ﬁelds of garnet amphibolite and rutile eclogite. The trondhjemite composition is attributed to mixing of two major melts: one originating from eclogitic facies source region (high Nb/Ta) and the other from a garnet-amphibolite facies (low Nb/Ta) leaving garnet, amphibole,andrutileintheresidue.(2)Comparedtotrondhjemite,thegraniteswithhighK 2 OandRb contents,andlowNa 2 O,Al 2 O 3 andSrcontentssuggestthatbasalticoceaniccrustwasnottheirsource material. Moreover, ﬁeld relationships suggest that the granites were derived from partial melting of trondhjemites. Hence, the trondhjemite and high-K granites marked together two distinct stages during the growth of the continental crust in the Tasiast area.


Introduction
Tonalite-trondhjemite-granodiorite (TTG) suites constitute an important part of outcropped Archean rocks [e.g. Martin et al., 2005], and the studies of these silicic magmatism can provide significant information about the origin and chemical evolution of the primitive continental crust [e.g., Arndt, 2013, Barker and Arth, 1976, Condie, 2005, Dey et al., 2017, Drummond and Defant, 1990, Jahn et al., 1981, Martin et al., 2005, O'Neil and Carlson, 2017, Rudnick, 1995. Mineralogically, TTGs are composed mainly of quartz + plagioclase and in lesser degrees, biotite and hornblende reflecting an evolved affinity with high content of SiO 2 (>68 wt%), Al 2 O 3 (>15 wt%) and Na 2 O (>3 wt%) and low K 2 O/Na 2 O (<0.5) ratios [e.g., Barker and Arth, 1976, Martin, 1994, Martin et al., 2005, Zhai and Santosh, 2011. In addition, Archean TTGs are also characterized by steep negative rareearth element (REE) patterns and low heavy REE abundances [Jahn et al., 1981, Martin, 1994, Martin et al., 2005. Also, the trace element concentrations of typical TTGs exhibit enrichments in incompatible elements and negative Nb-Ta anomalies [e.g., Huang et al., 2013, Martin, 1994, 1999, Martin et al., 2005. These Archean TTGs are generally interpreted to be the product of hydrous metabasalts partial melting [De Almeida et al., 2017, Foley, 2008, Foley et al., 2002, Martin, 1994, 1999, Martin et al., 2005, Moyen, 2009, Rapp and Watson, 1995, Rapp et al., 1999, 2003, Sen and Dunn, 1994, Winther and Newton, 1991. However, it is difficult to distinguish between melting of garnet-bearing amphibolite and eclogite in terms of major and trace element chemistry [Foley, 2008, Foley et al., 2002, Hoffmann et al., 2014, Klemme et al., 2002, Martin, 1999, Rapp and Watson, 1995, Rapp et al., 1999, 2003, Xiong, 2006. Furthermore, the geodynamic setting of TTG petrogenesis still remains controversial [e.g., Moyen et al., 2017 and references therein]. Consequently, three main models are proposed for the origin of TTGs and Archean crustal growth: TTGs can be formed by (1) partial melting of subducting oceanic crust. In this model, the hotter Archean mantle is responsible for melting (rather than dehydration) of metabasalts in subducted slabs; the Archean arc magmatism is therefore TTG in nature; (2) Partial melting of amphibolite/eclogite at the base of thickened arc crust. This model is probably the dominant petrogenetic model for the formation of the TTG suite [e.g., Moyen and Martin, 2012]; (3) Partial melting of wet mafic roots of an oceanic plateau [e.g. Van Kranendonk et al., 2007]. These last two models are primarily supported by the lack of clear arc-related features in the Archean geology [e.g., Hamilton, 1998Hamilton, , 2003, but it is difficult to admit a water supply at the base of the crust without evoking a subduction. In this paper, we present new field and petrographical observations, and whole-rock major and trace element data for the orthogneisses from the Tasiast Archean TTGs. The main purposes are (1) to provide new insights into the petrogenesis implications, (2) to assess the possible petrogenetic links between different orthogneisses, and (3) to understand the tectonic setting for the widespread Tasiast silicic magmatic units.

Geological setting
Reguibat Shield is the northernmost outcrop of the West African craton (Figure 1) [e.g., Ennih andLiégeois, 2008, Menchikoff, 1949]. It is surrounded on almost sides by Pan-African orogenic belts and covered in the south by the extensive intracratonic sediments of the Taoudeni Basin [e.g., Cahen et al., 1984, Ennih and Liégeois, 2008, Ouabid et al., 2017. The major part of this craton is buried beneath various metamorphosed sediments, and unconsolidated superficial deposits of Sahara Atlantic coastal. Dillon and Sougy [1974] and Bessoles [1977] distinguished two provinces within the Reguibat Shield: (1) a Southwestern province composed of rocks with an age older than ca. 2 Ga (i.e., pre-Eburnean cycle); and (2) Central and North Eastern province consists of Eburnean rocks (Figure 1). The Reguibat Shield was cratonized at the end stage of the Eburnean orogeny (cs. 2.1-2 Ga) and has been stable since about 1.7 Ga [e.g., Schofield et al., 2006, Montero et al., 2014. The Archean Shield contains grey gneiss, various acidic rocks, migmatites as well as mafic and ultramafic rocks (greenstone belt), quartzites and banded iron formations (BIF), and is characterized by regional foliations trajectories N-S to NE-SW in the southern part and NNW-SSE in its northern part [e.g., Chardon, 1997]. The southern part of the Shield is subdivided into three units which are from West to  [Chorlton, 2007].
The Tasiast-Tijirit terrane contains rocks with Nd model ages similar to Amsaga, between 3.6 Ga and 3.1 Ga, but zircon ages that only recorded magmatic events between 2.97 Ga and 2.89 Ga and a possible thermal imprint at ca. 2.7 Ga [Chardon, 1997, Key et al., 2008. The Tiris terrane has the youngest Archean crust with Nd model ages between 3.2 Ga and 3.0 Ga, and zircon ages that reflect several magmatic events at 2.95 Ga, 2.65 Ga, and 2.48 Ga [Schofield et al., 2012]. An age U/Pb, between ca. 3.06 and 2.95 Ga obtained from inherited zircons xenocryst in three quartz veins along Tasiast thrust [Heron et al., 2016]; this age corresponds to previous ages obtained from the Tasiast-Tijirit terrane gneiss and TTG rocks. According to our field observations (Figure 1b-d), the Tasiast granitoid gneisses can be easily subdivided into grey and leucocratic suites (Figure 2b-d). The leucocratic suite appears clearly younger than grey gneiss. The field relationships between both units are clear; the leucocratic gneisses are either wrapping the grey gneisses unit or intruding them with a sharp contact. New petrographic observations of more than 30 samples from these gneiss units permit to distinguish between two types of granitoid facies: TTG gneiss and granitic gneiss.

Petrography
Grey gneisses occur as grey-dark rocks with coarse-to medium-grained texture (Figure 3a-e). They have a homogeneous minerals assemblage made of plagioclase (dominant: 45-60% vol. of the rock), quartz (25-35% vol.), biotite (5-15% vol.), and minor interstitial potassium feldspar (K-feldspar), and accessory minerals including magnetite, ilmenite, titanite, apatite, and zircon. Generally, the grey gneiss occurs with two alternating layers: quartz and plagioclase-rich layer (clear) and biotite-rich layer (dark), which give the rock its gneissic foliation. Plagioclase crystals are euhedral to subhedral (1-4 mm in size) mostly altered to albite and sericite and enclose frequently biotite and apatite. Quartz ranges from interstitial to fine aggregates (0.1-0.5 mm in size), surrounding the large plagioclase crystals. Biotite occurs strongly as pleochroic flakes which are partially altered to chlorite, titanite, and opaque minerals, or exhibits elongated crystal around plagioclase or enclosed in this mineral. K-feldspar (mainly microcline) is interstitial between quartz and plagioclase grains.
Leucocratic gneisses are a medium-grained and yellow pale leucocratic rock with gneissic texture as well as the previous rock (Figure 3g-i). Major mineral phases are quartz (40-45% vol.), K-feldspar (35-40% vol.), plagioclase (10-15% vol.) and biotite (<4% vol.). Like the previous facies, titanite, Fe-Ti oxides (magnetite and ilmenite), apatite, and zircon are common accessory minerals in these granitic rocks. Quartz occurs as polycrystalline aggregates (0.2-2 mm in size) developed around primary crystals of feldspars or as fabric of alternated quartz and feldspar-rich layers. K-feldspar (principally orthoclase) and plagioclase occurs mainly as subhedral phenocrysts (1-3 mm in size) wrapped by the fine-to medium-grained matrix composed of K-feldspar and quartz aggregates.
Biotite is scarce and occurs as elongated crystals with preferential orientation parallel to the feldspar and quartz layers.

Analytical methods
Sixteen representative samples of the two distinct granitoid gneisses of the Tasiast unit were selected for whole-rock geochemical analyses, including eleven Grey gneiss and five leucocratic gneisses (compositions are given in Table 1). Rocks were analysed at the Instituto Andaluz de Ciencias de la Tierra (IACT, Granada, Spain). Rock powders were prepared using a jaw crusher and an agate ring mill. Wholerock analyses of major elements and V, Cr, Co, Zr, and Ni were determined by X-ray fluorescence (XRF). Loss on ignition (LOI) was determined by drying the samples at 900°C and is low (0.12-0.46 wt%). Trace elements (except V, Cr, Co, Zr, and Ni) were analysed using a Triple Quadrupole Agilent 8800 inductively coupled plasma mass spectrometer (ICP-MS) at the IACT laboratory. Sample digestion was performed following the HF-HClO 4 digestion process described by Garrido et al. [2000]. Element concentrations were determined by external calibration, except for Hf that was calculated using Zr measured by XRF and the chondritic Zr/Hf ratio [McDonough and Sun, 1995].
On the other hand, according to the petrographic observations and in contrast to previous gneiss, the granitic gneisses are silica rich with SiO 2 contents ranging between 71.99 wt% and 74.54 wt% and they are characterized by moderate Al 2 O 3 concentrations (13.39-15.08 wt%). Moreover, they are classified as high-K calc-alkaline to shoshonitic series with high K 2 O (3.25-6.28 wt%), moderate Na 2 O (2.71-4.69 wt%), and low CaO (0.64-1.43 wt%) concentrations ( Figure 5; Table 2). Furthermore, all the investigated samples of trondhjemite and granite gneisses are peraluminous [mol. Al 2 O 3 /CaO + Na 2 O + K 2 O = A/CNK > 1 and mol. Al 2 O 3 /Na 2 O + K 2 O = A/NK > 1; Maniar and Piccoli, 1989; Figure 6; Tables 1 and 2]. In contrast to the trondhjemite samples, the compositions of the studied leucocratic gneisses contrast starkly with Archean TTGs, and show a common trend for calc-alkaline granites (Figure 4a, b).
In terms of acidic magma origin, many experimental and geochemical studies concluded that most TTGs are derived from a hydrated crust, although physical conditions are still debated [e.g., Foley et al., 2002, Hoffmann et al., 2011, Rapp et al., 2003. The TTG melting occurred over a range of P -T conditions, covering the stability fields of garnet amphibolite and rutile eclogite. There are two common tectonic scenarios to produce TTG magmas: (1) melting of subducted oceanic crust; and (2) melting at the base of a mafic crust [e.g., Condie, 2005, Drummond and Defant, 1990, Hoffmann et al., 2011, Martin and Moyen, 2002, Martin et al., 2005, Moyen, 2011, Moyen and Martin, 2012, Smithies, 2000. Subduction-related TTG magmas tend to possess high Mg#, Cr, and Ni contents due to interactions between the TTG melts and the mantle wedge [e.g., Martin and Moyen, 2002, Martin et al., 2005, Moyen and Martin, 2012, Smithies, 2000. Anyway, the studied Tasiast TTG units have low MgO concentration (0.39-0.83 wt%, Mg# = 37-47) indicating an absence of interaction of studied TTG magmas with peridotite in the mantle wedge [e.g., Condie, 2005, Martin et al., 2005, Rapp et al., 1999, Smithies, 2000. Therefore, it is more likely that the Tasiast trondhjemites were generated at the base of a thick mafic crust. The resulted melt involving basalts or amphibolites under various P H 2 O conditions below 10 Kbar were tonalitic in composition, without any trondhjemitic affinity; the garnet is not stable in these conditions [e.g., Beard and Lofgren, 1989, 1991, Helz, 1976, Holloway and Burnham, 1972, Rushmer, 1991. According to field and petrographic observations, and geochemistry data, there are no tonalitic rocks among Tasiast gneiss units [Figures 2-4;e.g, Key et al., 2008]. Also, we notice that the majority of studied Tasiast TTG samples have a positive or no Eu anomalies accompanied by pronounced Sr anomalies (Figure 7a, b) that could reflect melting at pressures above the plagioclase stability field. Furthermore, the high Sr abundances (304-516 ppm) and low Y (<3.18 ppm) generating the high Sr/Y ratios in the trondhjemite rocks of Tasiast region (Figure 8a), indicating that the plagioclase was not in the residue and the depth of melting is above plagioclase stability field [e.g., Martin et al., 2005]. In addition, the Tasiast trondhjemitic gneisses display a pronounced enriched LREE relative to HREE patterns with high (La/Yb) cn and low Yb cn (Figures 7b, 8b, Table 1), which are consistent with garnet in the residue [e.g., Martin et al., 2005].
Moreover, melts with pure amphibole in the residue would have low Nb/Ta, Gd/Yb, and Dy/Yb ratios, and high Zr/Sm ratios [e.g., Foley et al., 2002, Rapp et al., 2003, Zhou et al., 2014; while the amphibole and garnet as a residual phases will not only effectively raise the Sr/Y, La/Yb, and Zr/Sm ratios with variable Nb/Ta ratio, but also increase the Gd/Yb and Dy/Yb ratios in the produced melts [e.g., Davidson et al., 2007, Macpherson et al., 2006. The Tasiast trondhjemites have variable Nb/Ta (11.26-30.23), high Gd/Yb (4.5-1.4), Dy/Yb (2.3-4.7), and Zr/Sm (>75) indicating that the TTG gneisses were derived by partial melting of a source with garnet and amphibole in the residue. These characteristics make the trondhjemite rocks of Tasiast unit different from the  [Maniar and Piccoli, 1989] (symbols are as in Figure 4). low-pressure TTGs described by Moyen and Martin [2012].
In the primitive mantle-normalized trace element diagram (Figure 7a), the trondhjemites are strongly depleted in Nb and Ta, which are consistent with rutile or/and amphibole in the residue [e.g., Martin et al., 2005, Zhou et al., 2014. Consequently, these mineral phases involve in controlling the HFSE ratios such as Nb/Ta, Zr/Hf, and Zr/Sm because these elements pairs have different partition coefficients in different residual mineral phases [e.g., Foley et al., 2002, Jenner et al., 1994, Klemme et al., 2002, Rubatto and Hermann, 2003, Xiong et al., 2007. So these ratios can be used as powerful evidence to discriminate between different source regions [e.g., Hoffmann et al., 2011]. Rutile and amphibole are two diagnostic minerals in eclogite facies and amphibolite facies conditions, respectively [Foley et al., 2002, Rapp et al., 2003. Therefore, the Nb/Ta and Zr/Sm ratios can be used to recognize the conditions of melting: eclogite facies or amphibolite facies [Hoffmann et al., 2011]. Foley et al. [2002] investigated the distribution of HFSE in TTG liquids and concluded that the low Nb/Ta and high Zr/Sm ratios in Archean TTG preclude an eclogitic residue but are consistent with a garnet-bearing amphibolite. Both residues cannot explain the compositional trends found for trondhjemites of Tasiast. The variable Nb/Ta (11.26-30.23) and high Zr/Sm (75.08-120.17) ratios clearly in favour of polybaric melting occurred over a range of P -T conditions, covering the stability fields of garnet amphibolite and rutile eclogite as illustrated in Figure 9. Therefore, we attribute the TTG compositions to mixing of two major melts: the first is originating from eclogitic facies source regions (high Nb/Ta) and the second is formed from a garnet-amphibolite source (low Nb/Ta).

Petrogenesis of granitic gneisses
Compositional differences between granitic and trondhjemitic gneisses in the Tasiast area are remarkable. In contrast to the trondhjemite, the granites have high K 2 O and Rb, and low Na 2 O, Al 2 O 3 and Sr contents (see Figures 5-7 and Table 2). Unlike TTGs (K 2 O < 2.84), a high K 2 O (K 2 O ∼ 3.3-6.34) content in granites suggest that basaltic oceanic crust was not their source material [e.g., Smithies, 2000]. According to Patino Douce [2005], the partial melting of anhydrous TTG within the middle or lower crust could produce granitic melts. The depleted HREE, Y and high La cn /Yb cn (∼51-178) exhibited by studied Tasiast granite samples indicate the role of high fractions of residual garnet in their sources or an HREE-depleted source similar to that produced by the trondhjemites (Figures 7, 8, Table 2). In addition, the contribution of Watkins et al. [2007] highlighted that the melting of TTGs with low K 2 O/Na 2 O (<0.3) could not generate high-K granites. As the Tasiast trondhjemites have an average value of K 2 O/Na 2 O at about 0.36 (0.21-0.58), these trondhjemites seem to be the main suitable source of the granites studied here. Furthermore, the field relations between the trondhjemite and the granite clearly support this hypothesis (Figure 2).

Continental evolution in the Tasiast Archean unit
Archean TTGs show secular variation in composition indicating change in geodynamic setting of their formation or cooling of the Earth [e.g., Moyen, 2002, Smithies, 2000]. The geochemistry data of the Tasiast trondhjemite suggest that the most reasonable mechanism for the origin of these rocks is Figure 7. Primitive mantle-normalized trace elements patterns (a, c) and chondrite-normalized REE (b, d) of the trondhjemites, and granites in the Tasiast area [Sun and McDonough, 1989], (symbols are as in Figure 4).  Figure 4).
the partial melting at the base of a thickened mafic crust covering the stability fields of garnet amphibolite and rutile eclogite (see Section 6.1). On the other hand, the trace element variations of granitic rocks were probably controlled mainly by source compositions and melting conditions rather than tectonic settings [e.g., Wu et al., 2003]. Their geochemical characteristics are likely inherited from the trondhjemite and cannot be used for discriminating its tectonic settings.
In addition, the Tasiast unit occurs as Archean continental crust and is dominated by grey gneiss complexes (mainly trondhjemites, dated at about ∼3.07-2.91 Ga [Chardon, 1997, Heron et al., 2016, Key et al., 2008. Hence, the emplacement of these rocks is envisaged to represent the growth of new continental crust by partial melting of mantle that generate oceanic tholeiites (now amphibolites and eclogites), followed by the partial melting of these metabasalts to constitute the parental magma of studied Tasiast trondhjemitic gneisses. On the other hand, the Tasiast granitic gneisses are produced afterwards by partial melting of trondhjemitic gneisses (see Section 6.2). The age of granitization is unknown but it seems to be younger than 2.91 Ga and can be linked to an early stage of new crustal growth in the West African craton. These granites should be probably similar to the syn-tectonic high-K granites of the Tiris complex (Mauritania) that yielded an age of about 2.56-2.48 Ga [Schofield et al., 2012]. This comparison is strengthened by the recognition in Tasiast terrains of an important thermal imprint at that time [Chardon, 1997, Key et al., 2008.

Conclusion
Field and petrographic observations, and geochemistry data from the Tasiast Archean unit presented in this study allowed to formulating the following conclusions: (1) Tasiast area consists mainly of trondhjemitic gneisses which host the high-K granites.
(2) Geochemical composition of the Tasiast trondhjemites suggests that they are juvenile continental crust generated by polybaric melting of metabasalts occurred over a range of P -T conditions, covering the stability fields of garnet amphibolite and rutile eclogite, and the melt has occurred without any mantle wedge involvement.
(3) Field relationships and trace element composition provide a genetic link between the trondhjemite and the high-K granitic gneisses.
(4) Our new field and petrographic observations, and geochemical data highlight at least two stages of magmatic events in the Tasiast Archean unit as follows: (1) a trondhjemite gneiss which is the earliest known gneiss in the region; (2) high-K granite which was generated by a partial melting of trondhjemites.