Trois échantillons caractéristiques ont été retenus. Leur localisation est indiquée sur la Fig. 1b et leur composition chimique (éléments majeurs et traces) est présentée dans le Tableau 1.

Le granitoïde syncinématique TG24 (N5°00.00′, E10°40.12′) est composé de plagioclase, feldspath potassique, quartz, biotite et amphibole. Apatite, zircon, allanite, sphène et opaques constituent les phases accessoires. C'est une monzonite quartzique métalumineuse, ferrifère et potassique (K₂O/Na₂O = 1,26). Les roches plutoniques associées, toutes mises en place lors de la phase D2, sont généralement plus riches en silice. Elles sont souvent porphyroïdes. Elles présentent un caractère métalumineux à faiblement peralumineux ; la valeur du rapport molaire A/CNK comprise entre 0,83 et 1,03 en fait des granitoïdes de type I. Dans le diagramme K₂O en fonction de SiO₂ ([31] ; Fig. 2), ces roches correspondent à une série shoshonitique. Les teneurs en Ba (1504–6810 ppm), en Sr (281–2646 ppm) et en Zr (307–502 ppm) sont très élevées. Les spectres de terres rares sont très fractionnés et présentent un enrichissement marqué en terres rares légères (Fig. 3).

L'amphibolite à grenat TG5 (N5°02.38′, E10°42.09′) comprend : amphibole (50 %), grenat almandin (20 %), plagioclase (15 %), salite (10 %), quartz (5 %), sphène, apatite, zircon et opaques. C'est une roche riche en titane et en fer. La richesse en Nb contraste avec l'appauvrissement habituel en cet élément dans les basaltes d'arcs. Le spectre des terres rares (Fig. 3) est caractérisé par un enrichissement en terres rares légères, similaire à celui des basaltes alcalins ou des tholéiites continentales [41]. Il s'agit donc d'un ancien magma basaltique intraplaque ayant (re)cristallisé dans les conditions de la croûte inférieure.

Le gneiss plagioclasique à biotite et amphibole TG10 (N4°97.28′, 510°43.59′) comprend : plagioclase (45 %), quartz (20 %), amphibole (15 %), biotite (12 %), feldspath potassique (⩽5 %), apatite et magnétite (<1 %), ce qui correspond à un protolithe tonalitique. Les teneurs en éléments majeurs et traces indiquent un magmatisme calco-alcalin pauvre en K (Fig. 2). Le spectre de terres rares (Fig. 3) est très fractionné avec un appauvrissement relatif en HREE et une anomalie positive en Eu.

Une étude UPb sur zircons du granitoïde TG24, de l'amphibolite à grenat TG5 et du gneiss à biotite et amphibole TG10 a été faite par la méthode classique. Les zircons ont été séparés et abrasés [11], leur analyse étant réalisée sur un spectromètre Finnigan Mat 262 RPQ II à l'Institut de géochimie de l'université de Göttingen (Tableau 2). Dans le cas des roches encaissantes métamorphiques, les différentes fractions s'alignent sur une discordia, dont les intercepts supérieurs (Fig. 3) à 2080±15Ma (amphibolite à grenat) et 2110±13Ma (gneiss à biotite et amphibole TG10) et les intercepts inférieurs à 618±50Ma et 650±26Ma montrent que ces roches ont été d'abord impliquées dans une histoire paléoprotérozoïque vers 2,1 Ga, avant de subir l'empreinte panafricaine. Les zircons de l'échantillon TG24 sont des cristaux typiquement magmatiques ; les cinq fractions analysées s'alignent sur une discordia (Fig. 3), mais sont proches d'un intercept supérieur à 618±20Ma (MSWD = 1,1), qui est interprété comme l'âge de cristallisation du magma.

La région de Tonga est caractérisée par des granitoïdes panafricains plus ou moins orthogneissifiés, de nature calco-alcaline potassique à shoshonitique, mis en place lors d'une phase D2 correspondant à de grands cisaillements transcurrents de direction est–ouest à NE–SW. Ces granitoïdes syncinématiques fournissent un âge de 618 Ma, qui est l'âge de leur cristallisation, et donc aussi l'âge de la phase D2 correspondant au fonctionnement des grands cisaillements. Sur le plan géochimique, ils sont très semblables à d'autres granitoïdes panafricains syn-cisaillement, déjà caractérisés au nord-ouest comme au sud de la zone d'étude [23,34], ainsi qu'au Brésil [30]. Il est à noter que cet événement D2 est contemporain du métamorphisme de haute pression et haute température, reconnu dans le domaine sud-camerounais [28].

L'encaissant métamorphique comprend des métabasites et des gneiss plagioclasiques à biotite et amphibole, donc des roches orthodérivées, qui ont préservé une structure D1 caractérisée par une foliation subhorizontale. Ces roches donnent des âges (intercepts supérieurs) paléoprotérozoïques (2,1 Ga), interprétés comme âges de cristallisation, et des âges (intercepts inférieurs) correspondant à un métamorphisme panafricain vers 650 et 618 Ma. Les paragenèses des métabasites permettent de préciser les conditions D1, qui sont de haute température (850 °C) et de haute pression (1 à 1,2 GPa), donc très différentes des conditions de pression attribuées à D2 (0,5 GPa). L'écart de pression (0,5 à 0,7 GPa) entre les métabasites et les granitoïdes syn-cisaillement implique l'érosion d'environ 20 km d'épaisseur de croûte continentale entre les phases D1 et D2, qui sont donc des phases distinctes et séparées dans le temps par au moins 20 Ma, sinon plus. D2 étant maintenant bien datée à 618 Ma, il est difficile de faire de D1 une phase de déformation panafricaine antérieure, car l'orogenèse panafricaine ne commence probablement pas beaucoup avant 620 Ma, en Afrique centrale comme en Afrique de l'Ouest [10]. Cela impliquerait alors que D1 corresponde à un épisode tectonique paléoprotérozoïque. L'âge de cet épisode (2,1 Ga) est proche des âges de 2,05 Ga obtenus au Sud-Ouest du Cameroun [29,39]. Ces âges coïncident avec le paroxysme de l'orogenèse éburnéenne–transamazonienne qui a soudé les parties brésilienne et africaine du craton du Congo–São Francisco et qui se manifeste aussi par une intense activité magmatique en de nombreux endroits du monde [13,14]. La région de Tonga et, probablement, tout le domaine centre-camerounais doivent être considérés comme un ensemble continental paléoprotérozoïque repris par l'orogenèse panafricaine. Par suite, l'orogenèse éburnéenne–transamazonienne, déjà reconnue dans le Sud du Cameroun [19,39], doit être maintenant mieux caractérisée sur l'ensemble du pays.

Syntectonic, locally orthogneissified, granitoids occur as syn-D2 intrusive bodies. Sample TG24 has been selected for geochronology; its location (N5°00.00′, E10°40.12′) is indicated in Fig. 1b. The rocks are medium-grained. Foliation is either magmatic or acquired during solid-state deformation. Foliation planes are underlined by the preferred orientations of feldspar, biotite and/or amphibole crystals and by the elongate shapes of quartz grains. Plagioclases have andesine to oligoclase compositions and are normally (sometimes reversely) zoned. Amphibole is mainly ferroedenite. Biotite crystals display Al, Ti and Fe-rich compositions. Accessory minerals are apatite, allanite, titanite and zircon. The rocks are silica-saturated, with SiO₂ ranging from to 57 to 70% (Table 1). They display high total alkali contents (7 to 9%). They are metaluminous to slightly peraluminous (molar A/CNK = 0.83–1.03), hence corresponding to I-type granitoids after [6]. In the K₂O vs SiO₂ diagram (Fig. 2), they plot in the shoshonitic domain of [31]. In addition, they are ferro-potassic rather than magnesio-potassic rocks. Large ion lithophile element (Rb and Ba) and high field strength elements (Zr and Th) display high to very high contents. REE patterns (Fig. 3) are very fractionated ((La/Yb)_N = 52–125) with a pronounced LREE enrichment and a small negative Eu anomaly (Eu/Eu∗=0.5–0.8).

The studied garnet amphibolite (TG5) from Maham III is a dark-green rock composed of hornblende (50%), pyroxene (10%), garnet (20%), plagioclase (15%) and quartz (5%). The rock shows two main deformation imprints associated with metamorphic recrystallisation. The first phase (D1) is represented by the S1 foliation, commonly flat-lying and striking N40°E, outlined by a compositional layering and by the preferred orientations of biotite lamellae, feldspar crystals and amphibole fishes. These foliation planes mould relictual (ante-D1?) clinopyroxene–garnet assemblages. The L1 mineral and stretching lineation trends N0–30°E with a nearly horizontal plunge towards the south-south-west. P1 folds occur as decimetre-sized isoclinal folds with S1 as the axial plane schistosity. Their axes are roughly parallel to L1. These D1 structures were not observed in the previously-described granitoids. The D2 deformation phase is responsible for a S2 schistosity (average strike N80°E) transposing S1, and for amphibolitic boudins with average lengths of 15 cm and pinch-and-swell structures. Quartzo-feldspathic leucosomes locally infill pressure shadows between boudins. Microprobe analyses reveal that hornblende is ferropargasite (XFe=0.59–0.62), and plagioclase is andesine (An33–38). Garnet is an almandine–pyrope–grossular solid solution, with high almandine content (up to 56%). The chemical composition of this garnet amphibolite is that of a common basic rock (Table 1). The rock is poor in Si and transition elements, but rich in Fe, K, Ti, Ba, Zr and Th, suggesting its derivation from continental tholeiitic magma [41]. REE patterns are moderately fractionated ((La/Yb)_N = 12), with LREE enrichment characteristic of intracontinental alkali or tholeiitic basalts (Fig. 3).

Biotite–amphibole gneisses are the most abundant rocks in the Tonga area, where they constitute half of total outcrop areas. They are fine- to medium-grained, grey-coloured, with alternating millimetre thick ferromagnesian-rich and quartzofeldspathic layers. They are composed of plagioclase (45%), quartz (20%), hornblende (15%), biotite (12%), K-feldspar (⩽5%) and accessories, such as zircon, apatite and magnetite (⩽1%). In the studied sample (TG10), unzoned plagioclase is oligoclase (An25–30) and amphibole is ferropargasite or ferroedenite (XFe=0.5–0.6). The biotite–amphibole gneisses display the same two phases of deformation as those recognized in garnet amphibolites. They are intermediate to felsic rocks, with low Ti, Mg and Fe contents. The selected sample has a tonalitic composition, rich in alkalis (Na₂O + K₂O = 6.32%), with Na>K, hence its location at the boundary between the low-K and medium-K calk-alkaline fields (Fig. 2). REE patterns (Fig. 3) are well fractionated ((La/Yb)_N = 58) with LREE enrichment, HREE depletion (Yb_N = 1) and Eu positive anomaly (Eu/Eu∗=2).

In the amphibolites, pressures were calculated from the garnet–clinopyroxene–plagioclase barometer [20] and temperatures from the garnet–clinopyroxene thermometer [2], and the garnet–hornblende barometer [8]. Pressure and temperature estimates are 1–1.2 GPa and 850 (±50) °C, pointing to granulitic lower crustal conditions corresponding to ante- or syn-D1 conditions. Recognition of the appropriate assemblage in the synkinematic granitoids enables us to calculate P using the Al-in-hornblende barometer [9,32], yielding emplacement conditions around 5 kb, i.e. at mid-crustal level.

In Cameroon, only RbSr whole rock and mineral ages were available before 1987 [3]. Since 1987, zircon UPb dating has permitted to date the Pan-African orogeny in northern and southern Cameroon and to unravel an older inheritance [22,29,37–39]. To the south, in Yaoundé series, a high-pressure granulitic metamorphism occurred at about 616 Ma and the sedimentary protoliths may be not older than 1 Ga. By contrast, in northern Cameroon, the metasedimentary and meta-igneous rocks gave ages around 2.1 GPa, suggesting the presence of Palaeoproterozoic rocks in this area, but intense reworking in high-grade conditions also occurred during the Pan-African Orogeny. Ages around 2 Ga had already been recognized in the cratonic part of southern Cameroon [39]. Our new data enlighten the case of the central part of the fold belt.

Chemical preparation and analyses have been done at the ‘Geochemisches Institut’ of the University of Göttingen (Germany). Samples 15–20 kg in weight were collected from fresh outcrops, broken, pulverized, and sifted. Zircons were separated and zircons concentrates were purified using heavy liquids (bromoform), nitric acid washes and handpicking. Purified concentrates were split into various fractions (>160 μm; 160–150 μm; 150–120 μm; 100–90 μm; 80–60 μm; 60–40 μm; <40 μm). Zircon analyses (Table 2) were performed on fractions of 40 to 50 grains, previously submitted to air abrasion to eliminate their external altered zones [11]. After zircon dissolution, Pb and U were separated using procedures recommended by [15] and [17] and the solution was aliquoted, with one split spiked with a mixed ²³⁵U–²⁰⁸Pb tracer solution. In all cases, the samples are sufficiently radiogenic. Decay constants used are those of [33]. Data were assigned uncertainties of 0.06% for ²⁰⁷Pb/²³⁵U and 0.16% for ²⁰⁶Pb/²³⁸U. Mass-spectrometry was performed on a Finnigan MAT 262 RPQ II.

Zircons from the garnet amphibolite TG5 are colourless, cloudy and elongate; they preserve a magmatic zonation, but no inherited core. In the biotite-amphibole gneiss TG10, zircons are colourless, light yellowish and cloudy; crystals are either prismatic elongate or small, rounded, multifacetted, without internal structure. Isotopic data give concordant ages to the upper intercept at 2080±15Ma and 2110±13Ma and lower intercepts at 618±50Ma and 650±26Ma, respectively for TG5 and TG10 (Fig. 4).

Zircons from the synkinematic granitoid TG24 are yellow in colour, euhedral, elongate, cloudy, devoid of any inclusions, internal structure or zonation. They are typical magmatic crystals. The discordia for five different fractions defines an upper intercept with an age of 618±20Ma (Fig. 4).

The syntectonic granitoids are shoshonitic intrusives emplaced along D2 shear zones, hence a more or less pronounced orthogneissification acquired at pressure conditions around 0.5 GPa. Therefore, the 618-Ma age of TG24 corresponds to this peculiar magmatic event, as well as to the age of this D2 transcurrent event, hence clearly related to the Pan-African orogeny. High-K calk-alkaline to shoshonitic magmatism closely associated with various major Pan-African shear zones has been described in many places of the Pan-African realm, either in Cameroon [23,34] or elsewhere, e.g., in Nigeria and Brazil [7,30]. There, this magmatism occurred in post-collisional conditions during transcurrent tectonics at 580–550 Ma. By comparison, the synkinematic potassic plutonism in Cameroon occurred earlier, between 620 and 580 Ma, and therefore it appears nearly coeval with the high P–high T metamorphism and nappe tectonics recognized along the southern edge of the Pan-African belt [29].

Country rocks are characterized by a flat-lying foliation, that is attributed to a D1 event and that has been locally transposed by the D2 deformation. In the studied amphibolite (TG5), the lower intercept at 618 Ma very likely corresponds to this D2 Pan-African transcurrent tectonics. This amphibolite shows a typical high-pressure granulite facies assemblage, pointing to much higher pressure conditions for D1 than for D2 at the outcrop level. The pressure gap (0.5–0.7 GPa) is such that it implies an uplift of about 20 km. Hence, D1 must have occurred at least 20 Myr (or more) before D2. However, it is difficult to admit that the Pan-African orogeny began much earlier than 620 Ma in central Africa as in western Africa [10]. Therefore, the D1 structural features may not have been acquired during a Pan-African event, but rather during an earlier history. Besides, the regular zonation, the absence of a core and the general subeuhedral form of the elongate zircons from TG5 suggest magmatic growth conditions. Thus, the upper intercept age of the amphibolite (2080±15Ma) is regarded as the crystallisation age of the mafic protolith. It is suggested that the zircons grew during a HP–HT event, synchronous with basaltic magma emplacement in the deep crust at about 2.1 Ga.

In TG10 gneiss, the euhedral forms of the zircons also call for a magmatic origin, despite the lack of zoning. Thus, the upper intercept dates the emplacement of the magmatic protolith at 2110±13Ma. This low-K calk-alkaline magmatism represents juvenile continental crust, possibly generated by partial melting of a mafic protolith, such as subducted young oceanic crust or the roots of an oceanic plateau. In both cases, restitic amphibole (± garnet) would explain the major and trace element contents of the melt [16]. The lower intercept at 650±26Ma is attributed to Pb loss and resetting during the Pan-African tectono-metamorphic imprint.

The Tonga area of central Cameroon is a Palaeoproterozoic domain, which was reworked during the Neoproterozoic orogenesis at around 0.6 Ga. The 2.1–2.0 Ga ages are similar to those obtained from the Nyong Series, which constitute the western part of the Congo craton (Fig. 1a). Grey gneisses from the Nyong Series are regarded as reworked Archaean TTG (tonalites–trondhjemites–granodiorites) [19]. Isotopic data point to the derivation of most of the Nyong Series from 2.9-Ga-old protoliths [39]. In Tonga area, no Archaean inheritance is evidenced, unlike in northern Cameroon [38]. Hence, we regard the central part of the Cameroon basement as a Palaeoproterozoic juvenile crust, which has been accreted to the Congo craton during the Eburnian–Transamazonian orogeny. This orogeny is responsible for the linkage of the Brazilian and African parts of the São Francisco–Congo craton and involved various continental fragments on both sides of the present Atlantic Ocean [13,14]. Our data emphasize the importance of the Eburnian–Transamazonian orogeny in Cameroon, which had been somewhat unappreciated so far. The existence of large Palaeoproterozoic crustal segments in central Cameroon raises the question of the northern limits of the Congo craton, which may extend up to north of the Pan-African fold belt [1]. Moreover, a detailed scheme of the Eburnian–Transamazonian orogeny in Cameroon remains to be drawn.