Thirty-two representative samples of plutonics (including mafic intrusions) from Mbengwi have been analyzed by ICP–OES at the “École des mines de Saint-Étienne” (France) for major elements. International geostandards have been used and details of the methods are available in Benoit et al. (1996) and Aries et al. (2000). Isotopic measurements of Sr and Nd on 13 representative samples have been done at Géosciences Environment de Toulouse (GET–OMP–University of Toulouse-3, France). The measurements were eluted after HF/HNO₃ digestion using Eichrom Sr-Spec, Thru–Spec and Ln–Spec resins, and performed on a Finnigan 261 multicollector thermal–ionization mass spectrometer. Sr isotopic ratios are corrected for mass fractionation with normalization to ⁸⁸Sr = 0.1194. Replicate analyses of the NBS 987 standard yielded an average ⁸⁷Sr/⁸⁶Sr value of 0.710255 ± 0.00002. Nd isotopic data are corrected for mass fractionation by normalization to ratio ¹⁴⁶Nd/¹⁴⁴Nd = 0.7219. Replicate analyses of the La Jolla Nd standard yielded an average ¹⁴³Nd/¹⁴⁴Nd value of 0.511850 ± 0.00001.

The Mbengwi plutonics can be subdivided into two not comagmatic and/or not cogenetic major series:

• • mafic plutonics displaying SiO₂ contents of 48–52 wt. %;
• • intermediate to felsic granitoids (SiO₂: 60.8–77.9 wt. %).

Mafic plutonics plot in the monzodiorite and monzogabbros field (although due to the absence of clinopyroxene, these rocks cannot be considered as monzogabbros but as monzodiorites), while granitoids form a continuous magmatic suite from monzonite to granite in the (Na₂O + K₂O) vs. SiO₂ diagram (Middlemost, 1994; Fig. 3A).

The normative composition (Table 1) of granitoids is characterized by (i) the occurrence of acmite in a Bt-granite sample (E₁₂₅); (ii) and the occurrence of corundum (0.08–2.37 wt. %) in almost all the samples except those containing normative diopside. Mafic plutonics are marked both by the lack of normative Qtz and the occurrence of olivine (2.02–18.9 wt. %) and nepheline (3.48–5.96 wt. %) in almost all the samples. Major elements Harker diagrams display a negative correlation of SiO₂ with Al₂O₃, TiO₂, Fe₂O₃^t, MgO, MnO, and CaO (Fig. 3B), and a positive correlation of SiO₂ with K₂O both for granitoids and associated mafic intrusions. Decreasing MgO (0.06–8.4 wt.%) and Fe₂O₃^t (0.2–12 wt. %) with increasing SiO₂, Na₂O and CaO reflect the inverse correlation of mafic minerals and feldspar modal contents, even exaggerated by the common replacement of plagioclase with Chl.

All granitoids apart from granodiorites have Na₂O–K₂O < 2 and therefore can be qualified as potassic according to Le Bas et al. (1986). In the K₂O vs. SiO₂ variation diagram (Fig. 3C) of Le Maître et al. (1989), they plot within the high-K to shoshonitic fields. Only granodiorites and a sample of Bt-granite (E₂₀) occupy the field of the fairly potassic rocks. Their CaO/Al₂O₃ molar ratios range from 0.06 to 0.68 and [(Na₂O + K₂O)/Al₂O₃] vary from 0.59 to 1.02. According to the diagram of Maniar and Piccoli (1989) and Chappell and White (1992) (Fig. 3D), granitoids are mainly metaluminous to weakly peraluminous in composition. Only a Bt-granite sample containing normative acmite (E₁₂₅) presents a molar sum (Na + K) > Al and can be considered as peralkaline according to Frost et al. (2001). The Mbengwi granitoids are mainly of type I, except a two-mica granite (E₆₁₀) that plots in type-S (A/CNK > 1.1) field.

Mafic intrusions are potassic (Na₂O–K₂O < 2) according to Le Bas et al. (1986) except monzodiorite (E₃₃). They all plot within the high-K to shoshonitic fields (Fig. 3C), and their CaO/Al₂O₃ molar ratios vary from 0.66–1.06, while [(Na₂O + K₂O)/Al₂O₃] vary from 0.47 to 0.62. They are metaluminous and of type I (A/CNK: 0.63-0.84) (Fig. 3D).

The I-type geochemical signature of the studied plutonics is supported by the mineralogy (predominance of Hbl, Bt as mafic silicate minerals, and the abundance of Spn and Mag as accessory phases). The FeO^t/(MgO + FeO^t) ratios of mafic plutonics (0.56–0.71) are lower than those of granitoids ranging between 0.70–0.94, but both series predominantly plot in the magnesian field (Fig. 3E).

In order to characterize the sources of the Mbengwi plutonics, we carried out Rb/Sr and Sm/Nd isotopic analyses on three monzodiorites and ten granitoids (two granodiorites, two amphibole–Bt- granites, three Bt-granites, two two-mica granites, and one leucocratic granite). The Nd, Sm, Sr, Rb contents and various isotopic ratios of the analyzed samples are presented in Table 1. Isotopic data have been corrected from an age of 620 My, according to the fact that no precise geochronological ages exist for our samples. All the analyzed samples have measured ⁸⁷Sr/⁸⁶Sr ratios higher than 0.7050 (average isotopic composition of the continental crust). However, one of the two-mica granite (E₆₁₀) has a very high ⁸⁷Sr/⁸⁶Sr value (0.78796). The initial ⁸⁷Sr/⁸⁶Sr isotopic ratios of mafic intrusions vary from 0.70371 to 0.70458, whereas the initial ɛ_Nd values range between -1.8 (E₁₄₁) and +0.6 (E₃₃). These same parameters vary respectively from 0.70436 to 0.70948 and from -15.9 to +0.7 in granitoids. The most enriched granitoids, in terms of ɛ_Nd, is a Bt-granite (E₁₂) and the most depleted sample is a two-mica granite (E₆₁₀). ⁸⁷Rb/⁸⁶Sr ratios vary from 0.31 to 0.88 in mafic intrusions, and from 0.42 to 8.85 in granitoids. The Sm/Nd ratios are rather constant, either in the analyzed mafic intrusions (0.11–0.14) or in the analyzed granitoids (0.07–0.16). Therefore, age corrections are critical for the Rb/Sr system and have a lesser impact on the Sm/Nd radioactive couple.

Using both the TAS diagram of Middlemost (1994; Fig. 3A) and AFM diagram (Fig. 4), it is noticeable that the Mbengwi plutonic suite occupies the fields of calc-alkaline series. In the AFM diagram, the representative points all integrate area I, forming a crescent shape characteristic of calc-alkaline series. This calc–alkaline affinity is also evidenced by a positive correlation of their Fe* (FeO^t/[FeO^t + MgO]) (Frost et al., 2001) with SiO₂ contents (Fig. 3E). All the mafic intrusions have magnesian affinity. Despite the presence of some Qtz–monzonite and granites having ferric affinity (0.85 ≤ Fe* < 0.95), almost all the granitoids samples are also magnesian. These same characteristics have been identified for Ngondo (Tagne-Kamga, 2003), Nkambe (Tetsopgang et al., 2008) and the western Cameroon granitoids in general (Nzolang, 2005).

Trace element and Sr isotope data can be used to more robustly decipher the nature of a magma source. Chemically mafic intrusions are metaluminous, have a low total alkali content (5.2–7.6) low FeO^T/MgO (1.2–2.5) and low Mg# values (29–44.3). Granitoids are weakly peraluminous, have a high total alkali content (mostly > 8.1 wt. %), low FeO^T/MgO (with very few > 5) and lower Mg# values (< 30). The overall low Mg# of the studied rocks rule out:

• • the possible direct melting of a peridotitic mantle source that was metasomatized via the addition of slab melt;
• • or the interaction of slab partial melts with peridotite during ascent through the mantle wedge (Martin et al., 2005);
• • or else magma mixing between crust- and mantle-derived melts (Jiang et al., 2009).

In addition, their high-K calcalkaline geochemical characteristics are not consistent with an origin via an interaction between slab partial melts and peridotite during ascent through the mantle wedge. Experiments of partial melting in conditions of temperatures and pressures close to reality (Patino Douce and Beard, 1996; Singh and Johanneses, 1996) proved that granitoids magmas can be produced from a wide range of parental crustal rocks. The REE patterns of Mbengwi granitoids and associated mafic intrusions exhibit LREE enrichment ([La/Yb]_N > 10), are mainly high in total REE (> 200 ppm), and have negative Eu anomalies. These REE characteristics are inconsistent with those of mantle-derived rocks, suggesting that these magmas originated in the crust. All the analyzed samples have ɛ_Sr > 0, which is symptomatic of a crustal source. Therefore, the investigated rocks may result from partial melting of the continental crust and the linearity of binary plots for the Mbengwi suite either for granitoids or for mafic intrusions can be explained mostly as a result of fractional crystallization. The monzodiorite sample (E₃₃) seems to be the less differentiated rock of the mafic series considering its lowest ⁸⁷Sr/⁸⁶Sr ratio, although it does not have a radiogenic ɛ_Nd. The type-S two-mica granite (E₆₁₀) with its high SiO₂ content (74.04 wt. %) is among the most evolved rock of the analyzed intermediate and felsic series; its lowest ɛ_Nd value (-15.86) is evocative of its long and complex crustal history.

The coupling of Sr and Nd tracers in the diagram of Foucarde (1998; Fig. 5) enables to accurately locate the source of these Pan-African rocks. Almost all result from the melting of the lower continental crust, except the less differentiated mafic intrusion (E₃₃) and a sample of Bt-granite (E₁₂) having positive ɛ_Nd, symptomatic of the oceanic crust source. The ɛ_Nd vs. SiO₂ wt. % variation diagram (not shown) reveals a general lack of correlation between these two parameters for all the studied rocks. This suggests that the variability of isotopic contents is the consequence of the heterogeneity of sources. However, the positive and negative correlations observed respectively for the mafic intrusions and for intermediate to felsic granitoids (granodiorites, amphibole–Bt-granites and two-mica granites) may be furthermore an argument in favor of the involvement of AFC processes. In granites, the values of ɛ_Sr, ɛ_Nd, and of the ⁸⁷Sr/⁸⁶Sr and ⁸⁷Rb/⁸⁶Rb isotopic ratios vary significantly, contrary to their SiO₂ contents (71.5–75.4 wt. %). These characteristics also observed in the syn-orogenic granitoids from the Bandombaai complex in Namibia are likely related to the assimilation of intermediate rocks (probably granodiorite or tonalite) from the lower crust, which would have greatly modified the isotopic composition (Van de Flierdt et al., 2003).

The geodynamic context of the Mbengwi plutonics has been inferred using their amounts of major elements, Rb, Y and Nb. Major elements geochemistry of Mbengwi granitoids and associated mafic intrusions points to the high-K shoshonitic and calc-alkaline affinity of the investigated samples, implying a subduction- to collision-related geotectonic setting.

In the (Y + Nb) vs. Rb diagram of Pearce (1996), the studied rocks are mostly post-collisional, except two granites samples (E₆₆, E₁₂₉) that integrate the field of syn-collision. All the mafic intrusions plot in the field of volcanic arc granites (VAG), except E₁₄₂, which occupies the field of within plate granites (WPG). The granitoids clearly occupy as other western Cameroon Pan-African granitoids, such as those of Bantoum (Nzolang et al., 2003), Bafoussam (Djouka-Fonkwé et al., 2008), Nkambe (Tetsopgang et al., 2008) and even those of the NE–Brazil (Guimarães et al., 2004) both the areas of VAG and that of WPG (Fig. 6a). In the (Nb/Zr)_N vs. Zr variation diagram of Thiéblemont and Tegyey (1994; Fig. 6b), almost all mafic intrusions plot in the post–COLG domain and only two occupy the field of the subduction zone. Monzonites, Qtz-monzonites, granodiorites, amphibole–Bt-granites, and most of two-mica granites occupy the field of post-collisional rocks; while the majority of Bt-granites rather integrate the field of rocks associated with subduction zones. In general, we notice that samples are spread from the area of rocks associated with subduction zones (VAG) to that of post–collisional granitoids. These characteristics and mostly the high-K calc-alkaline to shoshonotic affinity of the Mbengwi granitoids and the associated mafic intrusions are commonly reported in granitoids from late-collision to post-collision continental settings (Black and Liégeois, 1993; Liégeois et al., 1994, 1998; Toteu et al., 2004). The evolution from high-K cal-alkaline to shoshonitic or alkaline-peralkaline compositions observed in the studied granitoids is typical of final stages of orogeny (post-collision) in a continental collisional setting where crustal delamination operated (Liégeois et al., 1994, 1998). The overall geochemical features are similar to those of western and central Cameroon Pan-African granitoids related to the Pan–African D₁ event in Cameroon (Kwékam et al., 2010; Nguiessi Tchankam et al., 1997; Nzolang et al., 2003) and to those of the Ngondo complex (Tagne-Kamga, 2003), but differ from common granitoids from the magmatic arc (e.g., Pearce et al., 1984; Pimentel et al., 1996) by their distinctive depletion in Sr- and Y-undepleted pattern (not shown). The Mbengwi plutonics are slightly different from the Ngaoundéré granitoids (Tchameni et al., 2006) by the absence of samples including the area of within plate alkaline rocks. Therefore, some within-plate features observed in some granitoids (Fig. 6a) may be interpreted as an evolved trend of calc-alkaline suites (Tchameni et al., 2006). The geochemical subduction-type signature shown by the studied rocks would probably be inherited from their protolithes, emplaced during the subduction phase preceding the collision.

The ongoing study, including mineralogy, trace element geochemistry, and further U/Pb dating on zircon, will enable us to characterize more tightly the petrogenesis and melting processes of the Mbengwi granitoids and the associated mafic rocks.