Neoproterozoic magmatic evolution of the southern Ouaddaï Massif (Chad)

– This paper presents new petrological, geochemical, isotopic (Nd) and geochronological data on magmatic rocks from the poorly known southern Ouaddaï massif, located at the southern edge of the so-called Saharan metacraton. This area is made of greenschist to amphibolite facies metasediments intruded by large pre- to syn-tectonic batholiths of leucogranites and an association of monzonite, granodiorite and biotite granite forming a late tectonic high-K calc-alkaline suite. U-Pb zircon dating yields ages of 635 ± 3 Ma and 613 ± 8 Ma on a peraluminous biotite-leucogranite (containing numerous inherited Archean and Paleoproterozoic zircon cores) and a muscovite-leucogranite, respectively. Geochemical ﬁ ngerprints are very similar to some evolved Himalayan leucogranites suggesting their parental magmas were formed after muscovite and biotite dehydration melting of metasedimentary rocks. A biotite-granite sample belonging to the late tectonic high-K to shoshonitic suite contains zircon rims that yield an age of 540 ± 5 Ma with concordant inherited cores crystallized around 1050 Ma. Given the high-Mg# (59) andesitic composition of the intermediate pyroxene-monzonite, the very similar trace-element signature between the different rock types and the unradiogenic isotopic signature for Nd, the late-kinematic high-K to shoshonitic rocks formed after melting of the enriched mantle and further differentiation in the crust. These data indicate that the southern Ouaddaï was part of the Pan-African belt. It is proposed that it represents a continental back-arc basin characterized by a high-geothermal gradient during Early Ediacaran leading to anatexis of middle to lower crustal levels. After tectonic inversion during the main Pan-African phase, late kinematic high-K to shoshonitic plutons emplaced during the ﬁ nal post-collisional stage.


Introduction
The Precambrian geology of central Africa can be described as Archean/Paleoproterozoic cratons assembled during the Pan-African orogeny. In particular, the Central Africa Orogenic Belt (CAfOB, Fig. 1) is delimited to the west and to the south by the West Africa and Congo cratons, respectively, which comprise an Archean cratonic nucleus and remnants of Paleoproterozoic Eburnean orogenic belts that have not been reworked during the Pan-African orogeny. In contrast, the northeastern boundary of the CAfOB is more enigmatic. In this region, the basement pointing from beneath the Phanerozoic sedimentary covers in Sudan, Chad and Lybia has been described successively as the Nile craton (Rocci, 1965), the East Sahara craton (Bertrand and Caby, 1978), the Ghost Saharan craton (Black and Liégeois, 1993), the Saharan metacraton (Abdelsalam et al., 2002), or as a large Neoproterozoic belt involving no craton but characterized by the presence of 1.0 Ga basement rocks in Sudan (de Wit and Linol, 2015).
The Ouaddaï massif in eastern Chad is one of the least known areas of Africa regarding its geological evolution. Penaye et al. (2006) consider it consists of Paleoproterozoic basement rocks while Abdelsalam et al. (2002), Begg et al. (2009) and Liégeois et al. (2013) consider that it corresponds to the reworked southeastern margin of the enigmatic Sahara metacraton made of Archean and Paleoproterozoic preexisting crust. Toteu et al. (2004) propose it is a Neoproterozoic juvenile crustal segment locally contaminated by Paleoproterozoic crust. All these assumptions are made without any geological or geochronological data, except for Nd model ages on crustal xenoliths from Cenozoic basaltic deposits in the neighboring Darfur massif in Sudan yielding 790 to 2800 Ma (Abdelsalam et al., 2002).
In this study, we present new field, petrological and geochronological data on magmatic rocks (Tab. 1) from the southern Ouaddaï massif (extending to the Darfur massif in Sudan), where Precambrian rocks are unconformably covered by Phanerozoïc sediments. These data are used to constrain the ages, compositions and sources of the plutonic rocks, and to discuss the geotectonic significance of the Ouaddaï massif within the puzzle of cratonic margins and Pan-African orogenic belts characterizing the geology of central Africa.

Geological setting
The Ouaddaï massif, cropping out in southeastern Chad, is located between the Saharan metacraton to the north and the Congo craton to the south (Figs. 1 and 2a). It represents the easternmost exposure of the Pan-African Central Africa Orogenic Belt (CAfOB, Fig. 1; Bessoles and Trompette, 1980;Toteu et al., 2004). Geological mapping carried out by Gsell and Sonnet (1960); Wolff (1964) and Van Osta (1991) led to the distinction of a northern Ouaddaï massif mainly made of granitoïds and migmatites, and a southern Ouaddaï massif dominated by metasedimentary rocks. Our field investigations in southern Ouaddaï document intensely deformed series of muscovite-bearing quartzite, muscovitebiotite metapelites intercalated by rare marbles and calcsilicate gneisses (Fig. 2b). These rocks recorded greenschist facies conditions grading toward amphibolite facies as shown by the uncommon occurrence of garnet and sillimanite defining the main regional foliation. Rare coarse grained amphibolite alternate with the metasediments and correspond to the oldest exposed magmatic rocks of the area (Fig. 2b). Preliminary U-Pb dating of detrital zircons in quartzites yielded maximum depositional age of 1 Ga (Djerossem, 2018) proving that sedimentation occurred during the Neoproterozoic. The main structural foliation in the metasediments dominantly strikes N30-40 and is associated with superimposed isoclinal folds refolded at the regional scale by open vertical folds with axial planes subparallel to the main foliation. The metasedimentary series is intruded by a variety of plutons comprising leucogranite batholiths affected by, and wrapped into the main foliation of the metasediments as well as smaller plutons of monzonite, granodiorite and granite that cut across the main foliation and display a magmatic fabric. The latter occasionally form very small apophysis intruding the large peraluminous batholiths. Accordingly, plutons of intermediate to felsic composition of the southern Ouaddaï are subdivided in two groups: (1) pre-to syn-tectonic large batholiths of muscovite and biotite leucogranites; (2) late to post-tectonic small plutons of metaluminous pyroxene monzonite, biotite-and hornblende-bearing granodiorite and biotite granite (Fig. 1b).
U-Pb dating of zircons in the neighboring areas in Cameroon and Chad; Fig. 1) indicate a protracted magmatic activity characterized by emplacement of plutons with dominant calc-alkaline signature associated with minor tholeiitic and peraluminous magmas from 750 Ma to 600 Ma ending with the emplacement of post-collisional high-K calc-alkaline to shoshonitic igneous rocks with minor anatectic and A-type granites at about 590-545 Ma (Toteu et al., 2004;Shellnutt et al., 2019).

Analytical methods
Rocks samples were sawed into chips for thin section preparation and trimmed to small blocks for geochemical investigations. About 200 to 500 g of each sample was crushed into a steel jaw crusher and then pulverized with an agate ball mill. Powders were digested using an alkali fusion procedure where the powder was mixed to lithium metaborate and melted to produce a glass pellet. The pellet was digested into diluted nitric acid before analyses. Analyses and digestions were made at the "Service d'Analyse des Roches et Minéraux" (SARM, CRPG, France); major elements were dosed by ICP-OES while trace-elements were determined by ICP-MS following the procedure detailed in Carignan et al. (2001). The isotopic composition of Nd was also measured at the SARM by MC-ICP-MS following the protocol exposed by Luais et al. (1997).
For U-Pb zircon dating, about 1 to 2 kg of sample was crushed and then sieved to keep the 100-500 mm fraction. The pulp was rinsed with water to wash out the fine particles and dried overnight. Quartz and feldspar were first separated using bromoform, and ferromagnetic and paramagnetic minerals were removed using a Frantz isodynamic separator. Apatite was sorted from zircons and other dense minerals in di-iodomethane heavy liquid. Final selection of zircon grains was made by hand-picking before mounting in epoxy resin. Prior to U-Pb dating the internal structure of zircon grains was investigated by cathodoluminescence (CL) and back scattered electron (BSE) imaging using a JEOL JSM-6490 scanning electron microscope (SEM) coupled with a Gatan Mini CL at the Goethe University (Frankfurt, Germany). U-Pb dating was carried out by laser ablationsector fieldinductively coupled mass spectrometry (LA-ICP-MS) also at the Goethe University (Frankfurt, Germany) following the method described by Zeh and Gerdes (2012). Laser spots were 30 mm in diameter and approximately 20 mm in depth. U and Pb content and Th/U ratio were calculated relative to GJ-1 reference zircon. Isotopic data were corrected for background, intra-run elemental fractionation, and common Pb using an inhouse EXCEL ® spreadsheet (for details see Gerdes and Zeh, 2009) that incorporates the Pb evolutionary model of Stacey and Kramers (1975). All data U-Pb age data were corrected by standard bracketing relative to the zircon standard GJ-1 (primary standard; Jackson et al., 2004), which yielded after background, daily drift, and common Pb correction a Concordia age of 603.6 ± 1.6 Ma (2 SD; n = 28; MSWD CþE = 0.46, Prob CþE = 0.99; SDstandard deviation; CþEconcordance and equivalence). The accuracy of the zircon U-Pb isotope method was verified by the analyses of reference zircons Ple sovice and OG-1 (secondary standards), which yielded Concordant ages of 339.0 ± 1.3 Ma (MSWD CþE = 0.83, Prob CþE = 0.71, n = 14), and 3466.8 ± 6.4 Ma (MSWD CþE = 1.1, Prob CþE = 0.31, n = 11), in agreement with published data (Sláma et al., 2008;Stern et al., 2009). The results of U-Pb dating and standard measurements are shown in Tables 2-4 and in Figures 5a-5f. 4 Petrography of (meta-)magmatic rocks from southern Ouaddaï massif

Amphibolites
Amphibolites occur as small isolated bodies (Fig. 2b). Contacts with surrounding metasedimentary rocks are nowhere exposed but the shape of these bodies underlines the regional foliation. The amphibolites consist of amphibole, plagioclase, titanite, epidote and minor quartz. They commonly show a grano-poikiloblastic texture and the preferred orientation of amphibole and plagioclase defines a pervasive schistosity (Figs. 3a and 3b). In the investigated samples, two generations of amphibole are identified. The first generation comprises coarse grained hornblende crystals, and the second generation is made of fibers of actinolite associated to plagioclase. The latter occurs either as very fine xenomorphic crystals in the matrix or forms inclusions in large hornblende crystals. Titanite, epidote and quartz are present as small crystals within the matrix.
Muscovite leucogranite (sample GB16-3B, GB15-12) forms elongated bodies subconcordant to the regional foliation ( Fig. 3e). They have a medium grained texture and are composed of K-feldspar, plagioclase, quartz, and muscovite (Fig. 3f). Potassic feldspar (∼ 40 vol.%) is subhedral to anhedral and contains muscovite, plagioclase and quartz inclusions. Quartz (30 vol.%) is anhedral and shows microstructures indicative of dynamic recrystallization such as a rolling extinction and small polygonal subgrains along grain  Gsell and Sonnet, 1960): the study area is located in the southern part of Ouaddaï.

Granodiorite and biotite-granite suite
Plutons belonging to the granodiorite and biotite-granite suite cut across the structure of the metasediment-amphibolite intercalations but the main foliation is also locally deflected around these plutons (Fig. 2b). These plutonic rocks are dominated by K-feldspar and plagioclase, associated with quartz and biotite, and, in some samples, with clinopyroxene and green amphibole. Accessory minerals include titanite, epidote, zircon, apatite and oxides. Amphibole is restricted to biotite-granite (samples GB16-19A, GB16-24, GB15-20A), whereas granodiorite (sample GB15-13) contains higher amounts of apatite and zircon. Rocks of this suite are characterized mainly by a heterogranular to porphyric texture (Figs. 4a and 4b), locally overprinted by solid-state deformation (Fig. 4c). Localized shear bands are marked by elongated biotite and quartz with bended euhedral plagioclase crystals). Quartz porphyroclasts display kink bands and are surrounded by small polygonal neoblasts.

Pyroxene-monzonite
The pyroxene-monzonite (sample GB16-35) forms a small isolated plutonic body but the contact with host metasedimentary rock is not exposed. It consists of large euhedral crystals of plagioclase and K-feldspar with interstitial domains occupied by clinopyroxene and orthopyroxene (Fig. 4d). Brown amphibole is present along the rim of clinopyroxene and probably crystallized owing to late magmatic reactions between clinopyroxene and melt. Titanite is a very common accessory phase and is accompanied by apatite and zircon. The texture is purely magmatic and there is no evidence for solidstate deformation neither on outcrop scale nor under the microscope.

Geochronology
U-Pb dating was carried out on zircon grains from biotiteleucogranite sample GB16-5, muscovite-leucogranite sample GB15-12 and biotite-granite sample GB15-20A. Images of analyzed zircon grains with the Concordia diagrams are displayed in Figure 5, and the results of U-Pb dating presented in Tables 2-4. Zircon grains in the biotite-leucogranite GB16-5 are euhedral, elongated and prismatic and show a strongly resorbed CL-bright core with an oscillatory zoning, surrounded by an unzoned, CL-dark rim (Fig. 5a). High Th/U ratios (0.42-1.04, with two spots at 3.0) of cores and rims   Pl P P P P P Pl Pl l Pl P P P Pl Pl P P Pl Pl Pl P P P P P P Pl Pl P Pl P P P P P Pl Pl P  suggest a magmatic origin. The age spectrum recovered from these zircon grains reveals a complex history. Zircon cores yield mostly Paleoproterozoic and minor Archean 207 Pb/ 206 Pb ages between 3000 and 2048 Ma, and twelve rim analyses provide a rather well constrained concordia age of 635 ± 3 Ma with a MSWD of 0.3 (Figs. 5a and 5b). Some core analyses (n = 8) plot on or near to a discordia chord with an upper intercept at 2150 Ma, and one core gives a near concordant age at ca. 2200 Ma (analysis a576). The Neoproterozoic age (close to the Cryogenian/Ediacaran boundary) obtained from the rims is interpreted to date the time of granite emplacement.
The muscovite-leucogranite sample GB15-12 contains only a few zircon grains, consistent with the low Zr content of this sample (38 ppm). The grains are subhedral and stubby except for a few euhedral elongated prismatic grains. The internal texture is patchy, which is typical for re-crystallized zircon grains (Fig. 5c). Th/U ratios between 0.19 and 1.71 suggest re-crystallization of primarily magmatic zircon grains affected by metamictisation due to extremely high U and Th contents (U = 1670-9460 ppm; Tab. 3). Generally high common Pb levels (up to 15.3%) suggest infiltration of fluids during re-crystallization (Geisler et al., 2007;Zeh and Gerdes, 2014). Most spot analyses yield discordant 206 Pb/ 238 U ages younger than 550 Ma. Only three analyses yield a concordia age of 613 ± 8 Ma with a rather low MSWD of 0.16 (Figs. 5c and 5d), which most likely represents a minimum age for magma emplacement. No inherited zircons were identified in this sample.
The biotite granite GB15-20A delivered numerous zircons. U-Pb dating yields two age clusters (Figs. 5e and 5f): five ellipses, obtained on zircon with a Th/U ranging from 0.03 to 1.37, are concordant between 1050 and 1100 Ma while 10 spots, with a Th/U ranging from 0.12 to 0.97, characterized by a concordance above 90%, yield an age of 540 ± 5 Ma with a  (Rickwood, 1989); (c) A/NK vs. A/CNK (Miniar and Piccoli, 1989); (d) MALI classification from Frost et al. (2001). See text for data sources on COAfB occurrences.

Geochemistry and Nd isotopic composition
The geochemical characteristics of the different metaplutonic and plutonic rocks of the southern Ouaddaï are illustrated first in binaryl major and trace element diagrams (Figs. 6 and 7) and then in multi-element and REE diagrams (Figs. 8 and 9). These diagrams allow identifying the main magmatic trends and are compared to the Nd isotopic signature of the different facies (Tabs. 5 and 6).
Muscovite-leucogranite samples (GB16-3B, GB15-12) are also silica-rich (74.5-75.7 wt% SiO 2 ) with low contents in Fe 2 O 3t (0.64-0.67) and very low concentrations in MgO, TiO 2 , P 2 O 5 (close or below detection limits). They are peraluminous (molar A/CNK: 1.05-1.15) and, according to their mineralogical characteristics, belong to evolved S-type granites (Fig. 6c) with NK/A ranging between 0.80 and 0.85. They have very low REE contents compared to other Ouaddaï samples with either a concave upward, HREE-rich pattern (garnet-bearing leucogranite GB16-3B) or a LREE-enriched pattern with a (La/ Yb) N ratio of 5.7. The multi-element plot is characterized by very low contents in Ba, LREE, Zr-Hf and Sr but similar trends for elements ranging from Rb to Ta compared to the monzonitegranodiorite-biotite granite suite. There is a negative anomaly for Nb and a positive one for Pb (Fig. 9d). Nd isotope composition measured on two samples yield eNd (620 Ma) of À10.2 and À2.2. The T DM model ages are 4195 and 1583 Ma, respectively (Tab. 6).
Granodiorite and biotite-granite samples (sample GB15-13, GB16-19A, GB15-20A, GB16-24) display variable high SiO 2 (68 to 72 wt%) negatively correlated to MgO content (0.95 to 0.77 wt%) with relatively high Mg# (39-44) for felsic rocks (Fig. 6a). This suite of samples also follows the trend of high-K calc-alkaline series (K 2 O: 3.63-4.79 wt%) and are metaluminous to slightly peraluminous (molar A/CNK ranging between 0.94 and 1.02, typical of I-type granitoïds; Figs. 6b and 6c). They show rather high maficity (molar FeO þ MgO: 0.04-0.06; Figs. 7a-7d) and moderate P 2 O 5 (0.12-0.21 wt%). The different samples display very similar REE patterns with strongly enriched LREE normalized abundances compared to HREE ((La/Yb) N : 29-44), a weak negative to no Eu anomaly (Eu/Eu*: 0.77-1.05) and a flat pattern for the heaviest REE from Er to Lu (Fig. 8b). The trace-elements plot (Fig. 9b) displays typical features of calc-alkaline subduction-related and of post-collisional igneous rocks, that are (i) higher contents in large-ion lithophile elements (LILE: Rb, Ba, Sr) compared to most HFSE, (ii) enrichment in very incompatible  (Kelemen, 1995). It displays a shoshonitic affinity (K 2 O: 5.57 wt%) and plots in the metaluminous field of the A/NK vs. A/CNK diagram (Fig. 6c). LREE are strongly enriched compared to HREE ((La/Yb) N : 23) with normalized abundances for La around 200 (Fig. 8b). The multi-element plot shows decreasing abundance from most incompatible to less incompatible elements with pronounced negative Nb-Ta anomalies and a positive spike for Pb (Fig. 9b). This pattern is strikingly similar to those of the high-K calc-alkaline granodiorite and biotite-granite described above, pointing to a single igneous suite. The Nd isotopic signature (eNd (540 Ma) : À11.8) is within the range of values for granodiorite and biotitegranite and the Nd T DM model age of 1.9 Ga.

Age of igneous events in southern Ouaddaï and the significance of inherited zircons
The Ouaddaï was considered by some authors as a piece of pre-Neoproterozoic crust reworked during the Neoproterozoic Pan-African event (Abdelsalam et al., 2002;Liégeois et al., 2013), as Paleoproterozoic basement  or as a segment of Neoproterozoic juvenile crust tectonically accreted during the Pan-African orogeny (see Toteu et al., 2004). These contrasting suggestions are here discussed in the light of the first geochronological data obtained on plutonic rocks of the southern Ouaddaï presented in this paper.  Djerossem et al.: BSGF 2020, 191, 34 Our results show that intrusive plutonic rocks of the Ouaddaï are Ediacaran to lowermost Cambrian, implying that the deposition age of host metasedimentary sequences is between 1.0 Ga (age of youngest detrital zircons; Djerossem, 2018) and 635 Ma.
Ages of inherited zircons range from 2.1 to 3.0 Ga in the Ediacaran biotite leucogranite and 1.05 Ga in the Early Cambrian biotite granite (Fig. 5). The high Th/U ratios of the analyzed zircon grains are consistent with magmatic crystallization. Old zircon cores found in the peraluminous biotite  Djerossem et al.: BSGF 2020, 191, 34 leucogranite (formed after melting of metasediments, see next section) can be derived from post 2.15 Ga basement metasediments forming the source of the leucogranitic magma (see below) in the middle or lower crust. The oldest Archean to Paleoproterozoic ages are common in the Saharan domain and have been interpreted to represent the basement of the Sahara metacraton and surrounding cratons (Abdelsalam et al., 2002). In contrast, Mesoproterozoic zircons found in the biotite granite are rare in Central Africa except as detrital grains or inherited zircons in post collisional magmatic rocks (Meinhold et al., 2011;Shellnutt et al., 2017Shellnutt et al., , 2019 found in the Saharan metacraton domain. Considering the proposed petrogenesis of the Early Cambrian high-K granite biotite (see next section), we suggest that they derive from assimilation of crustal material during differentiation. Occurrences of late Mesoproterozoic igneous rocks have not been discovered yet, but these detrital and inherited zircons in Chad and Lybia definitely indicate that the Saharan metacraton was affected by an igneous event around 1.0 Ga, although the nature and context of this magmatism remain elusive.

Petrogenesis and significance of the three (meta-) igneous suites
The geochemical characteristics of amphibolites (high Al and Mg contents, low REE with positive Eu anomalies; Figs. 6, 7, 8a) sampled in the southern Ouaddaï and described above indicate that they derive from cumulate gabbros. This implies that the geochemical fingerprint of the parental magma has been blurred by fractional crystallization, involving accumulation of plagioclase and/or ferromagnesian silicates. The isotopic signature indicates either a depleted mantle source if the igneous protolith is Neoproterozoic (eNd at 600 Ma: þ3 to þ4) and a slightly enriched source or a contaminated mantle derived magma if the igneous event is Paleoproterozoic (eNd at 2.0 Ga: À1 to À4). In the absence of a radiometric age for the mafic igneous rocks precursor to these amphibolites and considering that the geochemical signature does not represent the composition of a melt, no precise information can be extracted on the source and the tectonic context.
Leucogranites are the most abundant magmatic rocks in southern Ouaddaï forming large plutons (Fig. 2b). They are subdivided into biotite-leucogranite and muscovite-leucogranite. The latter have the characteristics of evolved peraluminous S-type granitoids (Chappell and White, 1974;Chappell, 1999) most often formed by partial melting of terrigenous meta-sedimentary sources (Patino Douce, 1999;Clemens and Stevens, 2012). They display low maficity (0.62-0.67) and P 2 O 5 (below detection limits), high Al (13.6-14.5 wt% Al 2 O 3 ) compared to biotite leucogranite (Figs. 6 and 7, Tab. 5). Interestingly, Ba, Sr, LREE and Zr contents are low (ex: 28-38 ppm Zr) while U, Th, Nb and Ta reach moderate to high concentrations (Figs. 8d and 9d, Tab. 5) defining a trace-element distribution similar to north Himalaya leucogranites formed by muscovite dehydration melting (Gao et al., 2017). Garnetbearing sample GB16-3B is enriched in HREE and Y (Fig. 8d) reflecting peritectic garnet entrainment in the anatectic melt, a common feature of S-type granites (Stevens et al., 2007). Despite an origin by partial melting of a pelitic metasediment, it contains no inherited zircon. The very low bulk rock Zr content (below 38 ppm, Tab. 5) suggests that the absence of inherited zircons is not due to zircon dissolution in the peraluminous melt because it would have led to high Zr content in the melt. The paucity of zircons, related to the low bulk Zr content, together with the low Rb/Sr ratio for these samples are compatible with a limited degree of muscovite dehydration-melting (Inger and Harris, 1993;Gardien et al., 1995;Harris et al., 1995;Gao et al., 2017). Harris et al. (1995) explain the low Zr content and of some other incompatible elements, including P, to the location of the melting site within the source rock (typically around muscovite while zircons are often spatially associated to, or included into biotite in metasediments) and to the small volume of melt produced by fluid absent muscovite melting. Low volumes of melt formed around textural sites where zircons are scarce are unable to entrain zircons in the collected magma. The hypothesis considering these granites as near-solidus, low degree partial melts formed by muscovite dehydration melting is corroborated by their low zircon saturation temperatures ranging from 660 to 680°C (Watson and Harrison, 1983). These values, albeit low, are in the range of those determined for melts formed by  Djerossem et al.: BSGF 2020, 191, 34 muscovite dehydration-melting (see Harris et al., 1995;Gao et al., 2017) and consistent with the experimental data for muscovite breakdown conditions in the range 670-800°C between 5 and 10 kbar (see Vielzeuf and Schmidt, 2001). Biotite-leucogranite are also peraluminous, silica rich (∼ 76 wt% SiO 2 ) but show higher FeO contents (1.01-1.11 wt %) and lower alumina (Al 2 O 3 : 12.9-13.0 wt%) with very low P and Ti contents (Figs. 6 and 7, Tab. 5). The extremely low contents in plagioclase compatible elements (Sr, Ba, Eu) and the strong enrichment in very incompatible elements (Rb, Nb, Th, U) with the low Nb/Ta and K/Rb but high Rb/Sr ratios is similar to highly fractionated leucogranites from North Himalaya (Liu et al., 2016;Figs. 7, 8c and 9c) and Variscan French Massif central (Williamson et al., 1996). According to experimental data, the FeO and Rb contents together with lower Al 2 O 3 values (compared to muscovite-leucogranite) can be attributed to a partial melting reaction involving biotite in metasediments or, more generally, biotite-bearing metamorphic rocks in the lower crust source (Harris and Inger, 1992). This interpretation is corroborated by the presence of numerous inherited zircons. Both leucogranite types have low initial eNd, À2 and À10 for muscovite-leucogranite and À18 for the only biotite leucogranite analyzed (Fig. 10), compatible with old radiogenic crustal sources making the basement of the CAfOB (Toteu et al., 2001). Zircon saturation temperatures (Watson and Harris, 1983) are higher for biotite leucogranites (740-750°C) compared to muscovite leucogranites (660-680°C), supporting the hypothesis considering they derived from high temperature anatectic melt formed during biotite dehydration-melting reactions. Temperature required to melt biotite in fluid-absent conditions is in the range 750-850°C at a pressure equivalent to lower crustal depths (5-10 kbar; see Vielzeuf and Schmidt, 2001). Such temperatures in the crust during Early Ediacaran suggest a relatively high geothermal gradient (see Vanderhaeghe and Duchêne, 2010).
Granodiorite and biotite-granite form a high-K calcalkaline suite and belong to I-type metaluminous granitoids (Chappell and White, 1974;Chappell, 1999;Clemens and Stevens, 2012). Despite its peculiar major element composition (high Mg# and high-K), the shoshonitic pyroxenemonzonite, emplaced at 540 Ma, is interpreted to belong to the high-K calc-alkaline suite based on its trace-element signatures (Figs. 8b and 9b). The origin of I-type high-K intermediate to felsic magmas is attributed either to (i) partial melting of K-rich basic to intermediate rocks in the crust with entrainment of mafic peritectic minerals (Roberts and Clemens, 1993;Clemens et al., 2011) or (ii) partial melting of a fertile enriched mantle containing phlogopite and/or amphibole producing basic to intermediate high-K magma (Conceicao and Green, 2004;Condamine and Médard, 2014), which can further evolve to silicic compositions after fractional crystallization and assimilation (Castro, 2014). High maficity and high Mg# (59-39; Fig. 6a) found for this suite favors a mantle origin for the parental magma. Indeed, the most primitive pyroxene monzonite has a high-Mg# andesitic composition that is generally interpreted to derive from enriched mantle melts with or without crystal fractionation (Kelemen, 1995, Conceicao andGreen, 2004). This is consistent with experimental melting of phlogopite and amphibole lherzolite at uppermost mantle conditions (Foley et al., 1987;Condamines and Médard, 2014) that produces high-K to shoshonitic (3-5 wt% K 2 O) melts with basic to intermediate SiO 2 contents (52 to 63 wt% on anhydrous basis).  Shellnutt et al. (2018). The depleted mantle evolution is calculated as a one stage model with data from Salters and Stracke, 2004). COAB basement evolution is from Toteu et al. (2004). Subcontinental lithospheric mantle evolution has been recalculated with the data of Carlson and Irving (1994) acquired on enriched lherzolite xenoliths representative of the Wyoming craton. Fig. 11. Compilation of U-Pb (zircon and monazite) ages for igneous events in the Central Africa Orogenic belt in Cameroon and Chad. The data have been subdivided into: (i). early orogenic subalkaline and metaluminous meta-igneous rocks (mainly formed during the subduction-driven magmatic phases and the main orogenic event); (ii). anatectic (meta-) granites and leucosomes; (iii). late orogenic, post collisional rocks with high-K to shoshonitic signatures. Circles with red contour represent data acquired in this study. References are cited in the text.
Page 17 of 22 F. Djerossem et al.: BSGF 2020, 191, 34 All the samples from this suite have unradiogenic Nd signatures (eNd: À5 to À14) with T DM model ages ranging from 1.3 to 2.0 Ga. Such values can be interpreted as tracing (Fig. 10): (i) a crustal source, (ii) an old enriched mantle source such as a subcontinental lithospheric mantle (SCLM) or (iii) a suprasubduction mantle wedge formed during Late Cryogenian and reactivated during the post-collisional period, or (iv) mixing between crustal and mantle-derived magmas. The first hypothesis can be discarded as the suite includes high-Mg# rocks that necessarily derive from mantle melts. For example, a cratonic SCLM formed in the Archean or in the Paleoproterozoic can evolve toward very negative eNd values at 540 Ma ( Fig. 10; Carlson and Irving, 1994). Mixing of crustal and mantle components in the mantle (for example during Pan-African subduction) or assimilation of crustal rocks by mantle-derived magmas are also possible but it is difficult to test this hypothesis in absence of data on presubduction mantle, Neoproterozoic pelagic sediments from the Central Africa Orogenic Belt and basement rocks in Ouaddaï. Nevertheless, the presence of inherited 1.05 Ga zircons in the biotite granite suggest that parental magmas have assimilated crustal material but there is not enough isotopic data on this area (and especially on 1.05 Ga rocks that have yet to be found) to quantify crustal assimilation with mixing equations. However, the sample with the lowest eNd (À14) is also the one with low Th (18 ppm) compared to the most radiogenic sample (eNd: À3, Th: 32 ppm) while assimilation of crustal material with low eNd and high incompatible element content should increase the traceelement content while decreasing eNd values in the mélange (DePaolo, 1981). Crustal assimilation can therefore not fully explain the trace-element variations in the high-K to shoshonitic igneous suite. Despite variable silica contents (58-72 wt%), the Early Cambrian suite display limited traceelement variations (Figs. 8b and 9b) except for Sr/Y (40-140, Tab. 5) and La/Yb ratios (23-44; Fig. 8b). Crystal fractionation at high pressure (> 1 GPa) stabilizes garnet over plagioclase (Sisson and Grove, 1993;Müntener et al., 2001;Alonso-Perez et al., 2009) and lead to magma with high Sr/Yand high La/Yb ratios (MacPherson et al., 2006;Profeta et al., 2015) with no or little Eu anomaly (Fig. 8b). Garnet fractionation, as deduced from the variations in Sr/Y ratios, suggest that high-K and shoshonitic magma fractionated in the deep levels of a rather thick crust (probably > 30 km; Chiarada, 2015). To sum up, we propose that Early Cambrian high-K to shoshonitic suite of Ouaddaï was formed after partial melting of an enriched and heterogeneous K-rich mantle followed by high pressure fractional crystallization in the deep crust with minor impact of crust assimilation before final emplacement in the supracrustal metasediments. Such a scheme is often proposed for I-type intermediate to felsic plutonic rocks (Castro, 2014) in post collisional context (Dilek and Altunkaynak, 2007).

The place of Ouaddaï in the Pan-African belts of Western Gondwana
The southern Ouaddaï is located in the prolongation of the boundary between the Western Cameroon and Central Cameroon domains marked by the Tcholliré-Banyo shear zone and by a gravity anomaly at depth ( Fig. 1; Poudjom Djomani et al., 1995;Toteu et al., 2004;Ngako et al., 2008). Petrological, geochemical and geochronological characteristics of magmatic rocks of the southern Ouaddaï presented in this paper provide new insights on the nature of the Pan-African CAfOB in this region. The thick metasedimentary series deposited between 1 Ga and 635 Ma. They were deformed and recrystallized in greenschist to amphibolite facies conditions suggesting inversion of the sedimentary basin during the Pan-African phase. Early Ediacaran leucogranites, formed by dehydration melting of micas, testify for a temperature broadly between 670 and 850°C in deep crustal levels and thus for a high geothermal gradient. The absence of HP metamorphic rocks and of an oceanic suture does not favor an active margin context in Ouddaï during Early Ediacaran. Alternatively, such large detrital basins associated with a high geothermal gradient characterize back-arc basins forming tens to hundreds kilometers from the trench (Collins, 2002;Hydmann et al., 2005). Ouaddaï leucogranites could have formed in the hot lower crust of such a continental back arc basin. 750-600 Ma arc magmatism in the CAOfB is evidenced by juvenile mafic and intermediate rocks forming the Mayo Kebbi massif (Fig. 1) interpreted as an island arc Isseini, 2011) and calc-alkaline batholiths associated with exhumed, often migmatitic gneisses and metasediments formed in an Andean-type active margin (Figs. 1, 10 and 11;Toteu et al., 2004;Bouyo et al., 2015Bouyo et al., , 2016Nomo et al., 2017;Tchakounté et al., 2017;Saha Fouotsa et al., 2019). Plutons fed by anatectic magmas emplaced during the same period between 640 and 600 Ma and are intrusive in these metamorphic host rocks (Figs. 1, 10 and 11) (Tchameni et al., 2006;Fosso Tchunte et al., 2018;Li et al., 2017;this study). Parallelism between the Cryogenian-Ediacaran evolution of the CAOfB in Cameroon and the magmatic evolution of southern Ouaddaï massif strongly support that it belongs to the eastern prolongation of the Pan-African belt, but that it was located in a back arc position far from the trench where the suture zone will be located. Moreover, the presence of old inherited zircons, the unradiogenic signatures of Ediacaran to Cambrian magmatic rocks, the old Nd model ages and the absence or paucity of juvenile magmas or accreted rocks in Ouaddaï suggest that the Ouaddaï back arc basin formed by rifting and reworking of an older lithosphere that could correspond to the margin of the Saharan metacraton .
Early Cambrian late kinematic high-K calc-alkaline to shoshonitic rocks form the second main magmatic pulse in Ouaddaï. Parental magmas of these plutonic rocks might have tapped either the old enriched subcontinental lithospheric mantle of the Saharan metacraton (Fig. 11) during the post collisional period of the Pan-African orogeny (Figs. 10 and 11) or the suprasubduction mantle previously enriched during the Page 18 of 22 F. Djerossem et al.: BSGF 2020, 191, 34 Cryogenian subduction(s). Late orogenic, post collisional magmatism is also well expressed in the CAfOB, from 590 to 545 Ma Isseini et al., 2012;Kwékam et al., 2013;Shellnutt et al., 2017Shellnutt et al., , 2018Shellnutt et al., , 2019Saha Fouotsa et al., 2019;Figs. 1, 10 and 11) and the southern Ouaddaï comprises the youngest traces of this magmatism, dated at 540 Ma. The igneous pulses in the southern Ouaddaï are coeval with the emplacement of Ediacaran to Early Cambrian magmas in the central part of the CAfOB in Cameroon (Figs. 1, 10 and 11).

Conclusion
The first geochronological data on rocks from southern Ouaddaï presented in this work, demonstrate that this massif is composed of Neoproterozoic metasedimentary and igneous rock affected by the Pan-African orogeny. It is located along the eastern termination of the Central Africa Orogenic Belt and on the southern margin of the Saharan metacraton. The clastic metasediments deformed and recrystallized in the greenschist to amphibolite facies during the Ediacaran, and were intruded by several intermediate to felsic plutonic bodies. Large batholiths made of peraluminous leucogranites emplaced between 635 and 612 Ma; they formed after melting of metasedimentary rocks in the deep crust during the paroxysmal Pan-African event and correlates with similar evolution in neighboring Precambrian massifs in Cameroon. We suggest that thermal conditions required to melt the deep crust at the Cryogenian/Ediacaran transition were reached in the thinned lithosphere of a continental back arc basin later inverted during the main Pan-African orogenic phase. The leucogranites preserve Archean to Paleoproterozoic zircons probably inherited from the metasedimentary sources in the middle to lower crust. High-K to shoshonitic intermediate and felsic magmatic rocks emplaced at the Ediacaran/Cambrian boundary after partial melting of an enriched mantle with further differentiation in a relatively thick (> 30 km) crust. This enriched mantle could either correspond to the old subcontinental lithospheric mantle belonging to the southern edge of the Saharan metacraton or to an orogenic mantle previously metasomatized by slab-derived fluids and melts. The late Pan-African magmatism of Ouaddaï is coeval with an orogenwide phase of late orogenic post collisional magmatism recorded in Cameroon and Chad, which tapped a heterogeneously enriched mantle underlying the whole Central Africa Orogenic Belt.
These data show that an Archean to Paleoproterozoic metaigneous or metamorphic basement is not cropping out in Ouaddaï, in contrast to what was proposed by some authors. This old basement is however potentially present below the thick supracrustal sequence made of clastic metasediments. Inherited 1.05 Ga zircons in a biotite-granite suggest the contribution from Mesoproterozoic igneous rocks that might represent parts of the unexposed basement under the Early Neoproterozoic metasedimentary sequence. Exploration of U-Pb age and eHf signatures in detrital zircons would give insight into phases of juvenile, mantle-derived magmatism and crustal reworking from the Archean to the Neoproterozoic in Ouaddaï. Further studies should also emphasize the relative timing between metamorphism, deformation and magmatism to better integrate the Ouaddaï massif within the tectono-metamorphic evolution of the Central Africa Orogenic belt.