Journal of Petrology Volume 41 Number 1 Pages 87-109 2000
© Oxford University Press 2000
The Cameroon Volcanic Line Revisited: Petrogenesis of Continental Basaltic Magmas from Lithospheric and Asthenospheric Mantle Sources
1DIPARTIMENTO DI SCIENZE DELLA TERRA, UNIVERSITY OF TRIESTE, VIA E. WEISS 8, 34127 TRIESTE, ITALY
2BERKELEY GEOCHRONOLOGY CENTER, 2455 RIDGE ROAD, BERKELEY, CA 94709, USA
3DEPARTMENT OF GEOLOGY AND GEOPHYSICS, UNIVERSITY OF CALIFORNIA, BERKELEY, CA 94720, USA
4DIPARTIMENTO DI MINERALOGIA E PETROLOGIA, UNIVERSITY OF PADOVA, CORSO GARIBALDI 37, 35137 PADOVA, ITALY
5MINISTÈRE DE RECHERCHE SCIENTIFIQUE TECHNOLOGIQUE, B.P. 4110, YAOUNDÉ, CAMEROON
6LABORATORY FOR EARTHQUAKE CHEMISTRY, UNIVERSITY OF TOKYO, HONGO BUNKYO-KU, TOKYO 113, JAPAN
Received October 5, 1998; Revised typescript accepted June 24, 1999
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONGEOLOGICAL SETTING40Ar/39Ar AGESANALYTICAL METHODSCLASSIFICATIONPETROGRAPHY AND MINERAL...GEOCHEMISTRYPETROGENESISSW MIGRATION OF THE...REFERENCES |
|---|
The volcanic activity of Mts Bambouto and Oku (Western Highlands) and of the Ngaoundere Plateau, in the continental sector of the Cameroon Volcanic Line, Equatorial West Africa, ranges in age from Oligocene to Recent. It is characterized by basanitic, alkali basaltic and transitional basaltic series. Mineral chemistry, major and trace element bulk-rock compositions, and geochemical modelling suggest that the magmatic series evolved mainly at low pressure (2–4 kbar) through fractional crystallization of clinopyroxene and olivine ± magnetite, at moderately hydrated (H2O = 0·5–1 wt %) and QFM (quartz–fayalite–magnetite) to QFM + 1 fO2 conditions. Basalts from Ngaoundere (Miocene to Quaternary) and from the early activity (31–14 Ma) of the Western Highlands have incompatible trace element and Sr–Nd isotopic compositions similar to those of oceanic Cameroon Line basalts, pointing to a similar asthenospheric mantle source. By contrast, the late (15–4 Ma) Western Highlands basanites and alkali basalts have anomalously high concentrations of Sr, Ba and P, and low concentrations of Zr, which are exclusive features of continental Cameroon basalts. The genesis of these latter magmas is consistent with derivation from an incompatible element enriched, amphibole-bearing lithospheric mantle source. Western Highlands basalts show a continuous spectrum from high to low Sr–Ba–P compositions, and may result from variable amounts of mixing between melts derived from an anhydrous lherzolite source (asthenospheric component) and melts from an amphibole-bearing peridotite source (lithospheric HSr component). New 40Ar/39Ar ages for Mts Oku and Bambouto basalts, combined with previous 40Ar/39Ar and K/Ar ages of basaltic and silicic volcanics, and with volcanic stratigraphy, suggest a NE–SW younging of the peak magmatic activity in the Western Highlands. This SW younging trend, extending from the Oligocene volcanism in northern Cameroon (e.g. Mt Oku) to the still active Mt Cameroon, suggests that the African plate is moving above a deep-seated mantle thermal anomaly. However, the age and location of the Ngaoundere volcanism does not conform to the NE–SW younging trend, implying that the continental sector of the Cameroon Volcanic Line cannot be easily interpreted as the surface expression of a single hotspot system.
KEY WORDS: Cameroon Line basalts;40Ar/39Ar geochronology; lithospheric and asthenospheric mantle source; hotspot
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGICAL SETTING40Ar/39Ar AGESANALYTICAL METHODSCLASSIFICATIONPETROGRAPHY AND MINERAL...GEOCHEMISTRYPETROGENESISSW MIGRATION OF THE...REFERENCES |
|---|
The role of the lithospheric mantle in the petrogenesis of continental basalts is difficult to evaluate, as its nature and composition are known to be highly variable, and not easily distinguishable from the asthenospheric mantle. Continental basalts have distinct geochemical signatures compared with mid-ocean ridge basalts (MORB), but exhibit many similarities to oceanic island basalts (OIB). Lithospheric contamination of the asthenospheric mantle may occur through lithospheric delamination (e.g. McKenzie & O’Nions, 1995
) or through ancient subduction processes (e.g. Hofmann, 1988). On the other hand, there is increasing evidence that melts originating from the asthenosphere interact with the lithospheric mantle en route to the surface (Chazot et al., 1996; Wulff-Pedersen et al., 1996; Class & Goldstein, 1997). Current debate concerns whether continental basalts are primarily derived from the asthenospheric mantle, and contaminated by lithospheric mantle, or if they originate from variably ‘metasomatized’ lithospheric mantle (e.g. Menzies & Hawkesworth, 1987; Francis & Ludden, 1995; Wulff-Pedersen et al., 1996; Comin-Chiaramonti et al., 1997).
An exceptional opportunity to determine the contribution of lithospheric mantle to the petrogenesis of alkali basalts is presented by the Tertiary–Quaternary Cameroon Volcanic Line (CVL), in Equatorial West Africa, along which both oceanic and continental alkali basalts were erupted. The CVL is a 1600 km long chain of Cenozoic volcanic and sub-volcanic complexes that straddles the continent–ocean boundary and extends from the Gulf of Guinea to the interior of the African continent (Fig. 1). The oldest magmatism is represented by poorly studied plutonic ring complexes of 60–30 Ma (Cantagrel et al., 1978). Volcanic activity (42 Ma to the present) ranges from basaltic to more evolved phonolitic or trachytic compositions.
|
Differences between the mean chemical and Sr–Nd isotopic compositions of the mainly alkali basalts of the two sectors are, at a first glance, negligible (Fitton & Dunlop, 1985; Halliday et al., 1988). This similarity has been used to infer an asthenospheric mantle source for the CVL alkaline basalts (Fitton & Dunlop, 1985). Small Pb isotopic differences between basalts from the continental and oceanic sectors were interpreted in terms of different contributions from HIMU, DMM and EM mantle source components (Halliday et al., 1990; Lee et al., 1994a).
The initiation of the volcanism in the oceanic sector of the CVL displays a time-related NE–SW shift, and magmatism on a single volcanic island may persist for as much as 30 My (Lee et al., 1994a). 40Ar/39Ar ages of silicic volcanic rocks from the continental sector (Marzoli et al., 1999) suggest NE–SW migration of the volcanism from Mt Oku (25 Ma) to Mt Bambouto (18–16 Ma). However, this younging trend does not apply to the entire silicic volcanism of the continental CVL, nor to published ages of the plutonic ring complexes (K/Ar ages: Cantagrel et al., 1978). In general, the volcanism along the CVL cannot be interpreted as a single hotspot track, but may be the result of a linear mantle upwelling zone, or hotline, extending from NE to SW (Lee et al., 1994a; Meyers et al., 1998).
Whatever the origin of the CVL basaltic volcanism, the common signature of continental and oceanic basalts points to a similar sublithospheric mantle source. Here we present new 40Ar/39Ar, major and trace element, and Sr–Nd isotopic data for three CVL continental volcanic complexes that show marked intra- and inter-volcanic differences in trace element concentrations and moderate differences in Sr isotopic composition. These differences suggest that a continental lithospheric signature overprinted the common asthenospheric signature of the CVL.
|
GEOLOGICAL SETTING |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGICAL SETTING40Ar/39Ar AGESANALYTICAL METHODSCLASSIFICATIONPETROGRAPHY AND MINERAL...GEOCHEMISTRYPETROGENESISSW MIGRATION OF THE...REFERENCES |
|---|
The crystalline basement of the CVL forms part of a mobile belt between the West Africa and Congo cratons (Fig. 1), and comprises Pan-African granitic rocks (Lassere, 1978
) that yield Middle Proterozoic Nd model ages (Toteau et al., 1994; W. R. Van Schmus, personal communication, 1994). These Pan-African granitoids are crossed by a major, still active fault belt, the Central African Shear Zone, that follows old Pan-African lineaments (Browne & Fairhead, 1983; Moreau et al., 1987; Wilson & Guiraud, 1992) and has its counterpart in the NE Brazil Pernambuco lineament.
The Y-shape of the CVL (Fitton, 1987) mirrors the shape of the adjacent Mesozoic sedimentary basin of the Benue Trough, in which volumetrically scarce alkaline and tholeiitic magmatism, younging from north to south, occurred between 147 and 49 Ma (Maluski et al., 1995). The alkali basalts of the Benue Trough are geochemically and isotopically similar to those of the CVL (Coulon et al., 1996).
This study focuses on the magmatic activity of Mts Bambouto and Oku (Western Cameroon Highlands) and of the Ngaoundere Plateau (Fig. 1). These regions were uplifted, probably in the early Cenozoic shortly before the onset of the CVL magmatism (Browne & Fairhead, 1983). A regional negative Bouguer anomaly characterizes the volcanic highlands, particularly the Ngaoundere Plateau, where it reaches -120 mGal (Fairhead & Okereke, 1987). Seismic and gravity data suggest a general crustal thickness of 
30–34 km along the CVL continental sector with a lithospheric thickness of 120–150 km, except in the Ngaoundere Plateau, where the crust and lithosphere are thinned to about 20–23 km and 80–100 km, respectively (Fairhead & Okereke, 1987; Plomerova et al., 1993; Poudjom Djomani et al., 1995, 1997).
Although paralleled in the north by the extensional structure of the Benue Trough, the relationship between regional tectonics and the CVL magmatism is still debated (Moreau et al., 1987; Halliday et al., 1988; Déruelle et al., 1991). Only the volcanism of the Ngaoundere Plateau is clearly associated with the extensional tectonics responsible for the Adamawa horst and the Djerem and Mbere grabens. The whole CVL continental sector instead seems to follow an old suture zone between two seismically and palaeomagnetically distinct lithospheric blocks (Fairhead & Binks, 1991; Smith & Livermore, 1991; Plomerova et al., 1993; Poudjom Djomani et al., 1995).
Sampling and stratigraphy
A total of 130 basic rocks were sampled from the Mt Bambouto (55 samples), Mt Oku (53 samples) and Ngaoundere Plateau (22 samples) volcanic complexes (Fig. 1). Mt Bambouto [2700 m a.s.l. (above sea level)] and Mt Oku (3000 m a.s.l.) are Oligocene to Quaternary stratovolcanoes characterized by large collapse calderas, particularly well preserved at Mt Bambouto. The Mt Bambouto complex is characterized by five magmatic events: (1) extrusion of voluminous early stratoid basalt lava flow sequences (up to 600 m thick), which represent the peak of basaltic magmatism; (2) eruption, shortly before caldera formation, of trachytic lava flows and minor rhyolitic ignimbrites (Marzoli et al., 1999); (3) intrusion of postcaldera basalt dykes into the trachytic flows, and volumetrically scarce basalt flows; and, finally, (5) a few pyroclastic and phreatomagmatic events representing the most recent volcanic activity.
For this study, basic lava flows and dykes of the precaldera and postcaldera to Quaternary magmatic events were collected from the volcanic plains south, east and north of the main volcanic edifice of Mt Bambouto. Single basaltic flow units reach a thickness of >50 m, and the thickness of the sampled dykes varies from a few metres to 30 m. Quaternary basalts were collected close to the town of Foumbot, 50 km east of Mt Bambouto.
Field observations indicate for Mt Oku a succession of magmatic events similar to that of Mt Bambouto (see Njilah, 1991). Thick basalt lava sequences extend all around the main edifice and generally underlie silicic flows. Three 400–600 m thick volcanic sequences were sampled in detail at Kumbo, Jakiri and Foundong, i.e. east, south and north of the main edifice, respectively.
The general stratigraphy of the volcanic Ngaoundere Plateau is similar to that of the Mt Bambouto and Mt Oku regions, with stratoid basalt flows underlying the Miocene volcanics of Mt Tchabal Nghana stratovolcano (50 km east of Ngaoundere town), which is mainly composed of trachytic lava flows (Nono et al., 1994). Minor basaltic flows and pyroclastic deposits were erupted in Quaternary times. Miocene to Quaternary basaltic rocks were collected on the Ngaoundere Plateau, to the east and north of the Ngaoundere town, i.e. on the southern and western slopes of Mt Tchabal Nghana stratovolcano.
|
40Ar/39Ar AGES |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGICAL SETTING40Ar/39Ar AGESANALYTICAL METHODSCLASSIFICATIONPETROGRAPHY AND MINERAL...GEOCHEMISTRYPETROGENESISSW MIGRATION OF THE...REFERENCES |
|---|
Present (Fig. 2, Table 1) and previous 40Ar/39Ar (Lee et al., 1994b
; Marzoli et al., 1999) and K/Ar data (Fitton & Dunlop, 1985) indicate that the oldest magmatic activity in the Mt Bambouto region is represented by scarce quartz-normative mugearites of the Bangangte area (e.g. CA74, Table 2) dated at 42 Ma (Lee et al., 1994b; D. C. Lee, personal communication, 1995). However, the peak magmatic activity at Mt Bambouto is represented by slightly alkaline to subalkaline stratoid basaltic flows erupted between 21 Ma (sample CA51, 40Ar/39Ar plateau age of 20·8 ± 0·1 Ma) and 14 Ma (K/Ar; Fitton & Dunlop, 1985). The precaldera quartz trachyte lava flows and ignimbrites of the Bambouto main edifice range in age from 17·8 Ma to 15·5 Ma (40Ar/39Ar; Marzoli et al., 1999; A. Marzoli, unpublished data, 1999). After caldera formation, basalt dykes (sample CA27, 40Ar/39Ar plateau age of 15·1 ± 0·1 Ma) intruded the silicic flows of the main edifice, and basanitic flows erupted in the caldera. Younger volcanic activity is mainly represented by basanitic and interlayered alkali basalt flows occurring on the eastern slopes of Mt Bambouto (CA48, 40Ar/39Ar plateau age of 4·5 ± 0·3 Ma).
|
|
|
The stratoid basalt volcanism of Mt Oku commenced in the northeastern regions (31–23 Ma, Ndu area) and moved then to the SW (25–15 Ma, Bamenda area; Fitton & Dunlop, 1985; Njilah, 1991). A mugearite sample (CA162) belonging to the Jakiri basalt sequence, south of Mt Oku, yielded a 40Ar/39Ar plateau age of 23·4 ± 0·5 Ma. Finally, 40Ar/39Ar (Marzoli et al., 1999) and K/Ar ages (Gouhier et al., 1974) of Ngaoundere trachytes range between 11 and 7 Ma. Sporadic volcanism continued until the Quaternary in the Mt Bambouto, Mt Oku and Ngaoundere Plateau complexes.
|
ANALYTICAL METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGICAL SETTING40Ar/39Ar AGESANALYTICAL METHODSCLASSIFICATIONPETROGRAPHY AND MINERAL...GEOCHEMISTRYPETROGENESISSW MIGRATION OF THE...REFERENCES |
|---|
Major and trace elements (Table 2) of 130 samples were determined at the Dipartimento di Scienze della Terra, Trieste, Italy, using a PW 1404 XRF spectrometer and the procedures of Philips (1994)
for the correction of matrix effects. The analytical uncertainties are less than 5% and 10% for major and trace elements, respectively. FeO was measured by titration and loss on ignition (LOI), corrected for FeO oxidation, by gravimetry. Rare earth elements (REE, 35 samples) were determined by inductively coupled plasma-mass spectrometry (ICP-MS) at the Centre de Recherches Pétrographiques et Géochimiques, CNRS, Vandoeuvre, France (Govindaraju & Mevelle, 1987). Representative analyses are given in Table 2. The complete dataset may be downloaded from the Journal of Petrology website at http://www.petrology.oupjournals.org. Mineral compositions of 22 selected samples were obtained with a CAMECA-CAMEBAX electron-microprobe operating at 15 kV and 15 nA, at the Dipartimento di Mineralogia e Petrologia, Padua, Italy. A PAP CAMECA program was used to convert X-ray counts into weight percent of the corresponding oxides. Results are considered accurate to within 2–3% for major and 10% for minor elements. Representative compositions are reported in Table 3.
|
|
Samples for Sr (35 samples) and Nd (18 samples) isotopic analysis were first dissolved in HF, HNO3 and HCl in Teflon vials, followed by Sr and Nd collection by ion exchange and reversed-phase chromatography, respectively. The isotopic compositions were measured using a Finnigan MAT 262-RPQ mass spectrometer at the Centro di Studio per il Quaternario e l’Evoluzione Ambientale, CNR, Rome, Italy. Repeat analyses of NBS 987 and La Jolla standards gave average values of 0·71226(2) and 0·511857(8), respectively. No corrections were applied to the measured ratios for instrumental bias. The reported uncertainties on the Sr–Nd isotopic compositions are at the 2
confidence level. Some samples were reanalysed after HCl (6 N) leaching, to test the possible effects of alteration. The analyses yielded Sr and Nd isotopic compositions coincident with those of the unleached samples, within analytical errors [0·70365(2) vs 0·70366(1), and 0·512874 (2) vs 0·512874 (2), respectively].
Approximately 5–10 mg of plagioclase mineral separates were used to determine 40Ar/39Ar crystallization ages at the Berkeley Geochronology Center, USA. Plagioclase crystals were carefully hand picked to avoid altered or inclusion-bearing crystals and were irradiated for 7 h in the Triga reactor at Oregon State University, along with Fish Canyon sanidine (FCs) neutron fluence monitors. Plagioclase separates were step heated using a defocused NdYAG or Ar-ion laser beam. Ar isotopic compositions were measured in static mode by a MAP215-50 spectrometer, adopting procedures described by Renne (1995). Apparent ages were calculated assuming an age of 28·02 Ma for the FCs neutron fluence monitor (Renne et al., 1998). Age uncertainties are given at the 2 level, and do not reflect systematic errors in the age of the standard or the 40K decay constants. All the analysed samples yielded plateaux (Fig. 2), defined by 6–9 contiguous steps and by 80–100% of cumulative 39Ar released. Calculated Ca/K ratios (12–23) are homogeneous for each sample and consistent with measured mineral compositions (Table 3c), suggesting that the analysed mineral separates were composed of fresh and compositionally homogeneous plagioclases. One sample, CA51, was analysed twice, and yielded two indistinguishable plateau ages (20·95 ± 0·18 Ma, and 20·75 ± 0·14 Ma; Fig. 2a and b) from which the weighted mean of 20·83 ± 0·11 is adopted herein. Ar/Ar analytical data are reported in Table 1.
|
|
CLASSIFICATION |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGICAL SETTING40Ar/39Ar AGESANALYTICAL METHODSCLASSIFICATIONPETROGRAPHY AND MINERAL...GEOCHEMISTRYPETROGENESISSW MIGRATION OF THE...REFERENCES |
|---|
The classification of the basaltic flows and dykes was based on the total alkali–silica diagram (TAS; Le Bas et al., 1986; Fig. 3; Table 2). Three main magmatic suites can be distinguished: (1) basanite–tephrite–phonotephrite (BSN series; CIPW nepheline, Ne = 9–24%); (2) alkali basalt–Ne hawaiite–Ne mugearite (AKB series, Ne = 1–11%); (3) CIPW hypersthene, Hy basalt (named hereafter transitional basalts)–Hy hawaiite–Hy or quartz mugearite (TRB series; Hy = 1–23% or Q = 2–3%).
|
), Ngaoundere (
) and Mt Oku samples (grey squares). Bsn, basanite; Teph, tephrite; PTeph, phono-tephrite; H, hawaiite; Mug, mugearite; Ben, benmoreite; B-And, basaltic andesite. |
PETROGRAPHY AND MINERAL COMPOSITIONS |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGICAL SETTING40Ar/39Ar AGESANALYTICAL METHODSCLASSIFICATIONPETROGRAPHY AND MINERAL...GEOCHEMISTRYPETROGENESISSW MIGRATION OF THE...REFERENCES |
|---|
The investigated basaltic samples are petrographically fresh, do not contain secondary minerals and are mostly aphyric or slightly porphyritic. Samples with more than 10 vol. % phenocrysts are rare and mainly belong to the alkaline volcanics of the Ngaoundere Plateau. The most common phenocrysts (Table 3) are olivine and clinopyroxene, with subordinate magnetite and plagioclase. Olivine predominates over clinopyroxene phenocrysts in strongly alkaline samples (BSN suite), whereas the opposite occurs in the moderately alkaline and subalkaline samples (AKB and TRB suites). Plagioclase is a common phenocryst in the TRB suite samples and, less frequently (
5%), in Ne hawaiites and Ne mugearites. Apatite is present as small microphenocrysts in all the suites. Olivine is commonly found as a phenocryst phase in most of the samples investigated, except Hy hawaiites and mugearites, where it represents a groundmass phase. The porphyritic samples may have large resorbed and moderately altered olivine phenocrysts, whereas the subaphyric ones sometimes contain skeletal crystals indicating high crystallization temperature and high cooling rates. In general, the phenocryst cores have forsterite (Fo) ranging from 75 to 89% (mean 81 ± 4%), whereas the Fo content of the groundmass varies between 29 and 85% (mean 67 ± 13%; Table 3a). A systematic Fo decrease from early to late-crystallized olivines was observed for each sample.
The relationship between the Fo content of the phenocryst cores and the mg-number [Mg/(Mg + Fe2+)] of the bulk rocks suggests that olivines generally crystallized under low pressure and equilibrium conditions with a Kd (Fe2+/Mg partition coefficient between olivine and melt) of 0·30–0·33 (Roeder & Emslie, 1970). However, several olivine compositions correspond to a Kd of 0·25, suggesting that this may reflect a compositional effect and/or a relatively high fO2 (e.g. Fe2O3/FeO 0·25), owing to the alkaline nature of the investigated rocks.
Clinopyroxene (cpx) is generally represented by euhedral and weakly zoned pinkish phenocrysts in all the porphyritic samples, except the more evolved rocks (MgO < 3·0 wt %) where it is a groundmass phase. Some basanites may contain zoned cpx with a pink rim and a green core. Compositionally (Table 3b), cpx corresponds to salites and rarely to augites, which predominate in the more evolved samples. In terms of atomic Ca–Mg–Fe* (Fe* = Fe2+ + Mn + Fe3+) relationships all cpx, including those of the transitional basalt suite, define phenocryst core to groundmass trends characterized by increasing Fe* and by constant or slightly increasing wollastonite component, consistent with the alkaline nature or affinity of the parental magmas.
Using the geobarometer of Nimis (1995), based on crystal-structure modelling, the pressure of cpx crystallization was estimated. The structural parameters were calculated following the program of Ottonello et al. (1992), which allows simulation of the geometry of a C2/c pyroxene from its chemical composition. An intracrystalline temperature of 1100°C was assumed for the distribution of Fe2+ and Mg between the M1 and M2 sites. The pressure was calculated only for the megacryst and phenocryst cores, as they better approach a near-liquidus crystallization (see Nimis, 1995). The relationships between the cell and M1 site volumes indicate for the phenocryst cpx a low pressure of crystallization (0–4·3 kbar; Table 3b), except for the cpx of the hawaiite CA133, which yielded a pressure of 8·0 kbar. The pressure of the megacrysts (e.g. CA148) ranges from 3·9 to 4·6 kbar and is similar to the highest one obtained for the phenocryst cpx.
The olivine–clinopyroxene Fe2+–Mg exchange geothermometer of Loucks (1996) for early crystallized olivines and augites yielded temperatures ranging from 1250 ± 52 to 1106 ± 40°C (BSN suite), 1338 ± 54 to 1100 ± 43°C (AKB suite) and 1280 ± 35 to 1130 ± 63°C (TRB suite), respectively (Table 3b).
Plagioclase phenocrysts are commonly found in the rock types of the TRB suite and, sometimes, in those of the AKB suite. The anorthite (An) content (Table 3c) of the phenocryst core and groundmass plagioclases ranges from 82 to 42 wt %, but some transitional (e.g. CA50) and alkali (e.g. CA141) basalts may contain plagioclase megacrysts with relatively low An contents (e.g. 29–34 and 47–49 wt %, respectively), suggesting a xenocrystic origin. Finally, temperatures calculated according to Mathez (1973) for the early crystallized plagioclase phenocrysts range from 1289 to 1162°C for dry conditions and from 1180–1013°C for hydrous conditions (PH2O = 0·5 kbar). These temperatures, compared with those calculated for olivine–clinopyroxene pairs, suggest near-anhydrous conditions for plagioclase crystallization.
Opaque minerals (Table 3d) are mainly represented by magnetites with ulvöspinel content of 52–78%. The homogeneous groundmass magnetite–ilmenite pairs of the transitional basalt CA150 yielded -log fO2 of 13·25 and 15·75, and corresponding T values of 900 and 812°C [approximately the QFM (quartz–fayalite–magnetite) buffer].
|
In summary, from the mineral chemistry it appears that most early formed minerals, megacrysts included, crystallized within the crust, close to QFM buffer conditions and at low PH2O.
|
GEOCHEMISTRY |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGICAL SETTING40Ar/39Ar AGESANALYTICAL METHODSCLASSIFICATIONPETROGRAPHY AND MINERAL...GEOCHEMISTRYPETROGENESISSW MIGRATION OF THE...REFERENCES |
|---|
Major and trace elements
The variations of selected major and trace elements relative to MgO are shown in Fig. 4. In general, basanitic rock types predominate on the Ngaoundere Plateau, transitional basalts on Mt Oku, and alkali basalts on Mt Bambouto. Mt Bambouto volcanism tends to become more alkaline with time, similar to that of Mt Oku (Njilah, 1991
). The Ngaoundere Plateau samples are characterized, at similar MgO content, by higher amounts of Nb (up to 140 ppm; Table 2) relative to samples from Mts Oku and Bambouto, which, instead, have higher FeOt, Sr, Ba and P2O5 (Fig. 4).
|
Despite the general compositional similarity of the oceanic and continental CVL basalts (Fitton & Dunlop, 1985; Fig. 5), the minor and trace element concentrations in specimens from the continental sector exhibit a large range of variations (Table 2; Fig. 4). For example, primitive Mt Bambouto alkali basalts with almost identical major element compositions (compare CA40 and CA3, Table 2) have Sr concentrations that vary from 600 to 1300 ppm. Ba, P and Zr abundances display a similar variability. The ratios between Ba, Sr, P and Zr and other incompatible elements [e.g. light REE (LREE)] in the less evolved Mt Bambouto basalts (MgO > 6 wt %) reveal wide variations for similar (high) LREE concentrations (Fig. 6). These differences cannot be easily related to different degrees of fractional crystallization from a common parental magma (see Fig. 4), or to different degrees of partial melting of the same mantle source (see Fig. 6).
|
|
Samples with a mantle-normalized positive spike at Sr, relative to Ce and Nd (Fig. 5), are hereafter designated as high-Sr basalts (HSrB). Generally, HSrB are also characterized by strongly positive Ba and P spikes, and by negative Zr anomalies. In contrast, the basalts designated hereinafter as low in Sr (LSrB) are characterized by moderately negative or absent Sr and P anomalies, and by moderately positive Ba anomaly (Fig. 5).
There is a continuous compositional spectrum between HSrB and LSrB (Fig. 6). Approximately 35% of the Mt Bambouto basalt samples are of HSr type. These samples mainly belong to the BSN suite and represent the late magmatic events (15–4 Ma): dyke intrusions in the trachytic lava flows of the main edifice, intracaldera flows and postcaldera flows to the east of Mt Bamouto. The early (21–14 Ma) basaltic flows that cover the plains around Mt Bambouto are generally of LSr type and belong to the AKB and TRB suites. Only 10% of Mt Oku basaltic volcanics are of HSr type (BSN and AKB suites), whereas HSrB are absent among those from the Ngaoundere Plateau. Notably, nephelinitic rocks from Mt Etinde, at the continent–ocean boundary on the SW slopes of Mt Cameroon, are characterized by extreme enrichments of Sr and Ba (up to 10 000 and 3000 ppm, respectively; Nkoumbo et al., 1995), recalling HSr-type compositions. We emphasize that the compositional characteristics of HSrB have not been described among the oceanic CVL basalts, which, instead, are compositionally close to the LSrB of the CVL continental rocks (Fig. 5).
All the analysed samples have steep chondrite-normalized REE patterns [(La/Yb)cn = 7–30; Fig. 7; Table 2], suggesting that they may have been generated by low melting degrees of a garnet peridotite mantle source. Consistently, the highest LREE/HREE (heavy REE) ratios are generally shown by BSN samples. BSN and AKB samples for the Ngaoundere Plateau have (La/Yb)cn slightly higher (23–28 and 13–30, respectively) than the corresponding rocks from the Western Highlands (15–18 and 12–16, respectively). In general, there is a tendency for the REE concentrations to increase with differentiation, whereas LREE/HREE ratios remain virtually constant. An exception is represented by the evolved Bangangte mugearites (Bambouto area, e.g. CA74; Fig. 7c), which have distinct REE patterns with low (La/Yb)cn (6·82).
|
Sr and Nd isotopic data
In general, the investigated basalts yielded initial 87Sr/86Sr (0·70288–0·70382) and initial 143Nd/144Nd (0·51292–0·51284, 
Nd +3·7 to +5·6; Fig. 8a, Table 2) that reflect time-integrated depleted source mantle compositions, and are similar to those of other CVL continental and oceanic basalts (Fitton & Dunlop, 1985; Halliday et al., 1988, 1990). Only a few of the most evolved rocks, i.e. mugearites from Mt Bambouto (CA13, CA73 and CA74), yielded relatively high initial 87Sr/86Sr (0·70504–0·70775) and low initial 143Nd/144Nd (0·512614–0·512498, Nd -0·4 and -2·6; Fig. 8a). Finally, we note that initial 87Sr/86Sr compositions of Mt Bambouto samples are slightly variable, with basanites having higher Sr isotopic composition (0·70350–0·70364) than alkali basalts (0·70311–0·70342). In general, initial 87Sr/86Sr values of HSr basalts are higher than those of LSr basalts (Fig. 8b), whereas Nd isotopic compositions are indistinguishable.
|
|
PETROGENESIS |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGICAL SETTING40Ar/39Ar AGESANALYTICAL METHODSCLASSIFICATIONPETROGRAPHY AND MINERAL...GEOCHEMISTRYPETROGENESISSW MIGRATION OF THE...REFERENCES |
|---|
Fractional crystallization
The petrography, mineral chemistry, and major and trace element variations suggest that differentiation of the basanite (BSN), alkali basalt (AKB) and transitional basalt (TRB) suites is compatible with fractional crystallization at low pressure. The BSN suite is dominated by olivine, clinopyroxene and subordinate magnetite removal; the AKB suite by clinopyroxene, olivine and subordinate magnetite fractionation, and plagioclase during the late stages of evolution; whereas the TRB suite probably differentiated through olivine, clinopyroxene, plagioclase, and moderate magnetite fractionation.
The fractional crystallization model was further tested by using the MELTS code of Ghiorso & Sack (1995). MELTS allows the modelling of liquid lines of descent of silicate magmas as a function of P, T, fO2 and H2O, and provides the compositions of both the derivative magmas and liquidus minerals. As an example, the modelled dry and hydrous (H2O = 1·0 wt %) differentiation trends for Mt Bambouto AKB are shown in Fig. 9. In general, for all suites the best match between modelled and observed differentiation trends and mineralogical compositions was obtained for Pt = 2–4 kbar, H2O = 0·5–1·5 wt % and fO2 = QFM (TRB and AKB) and QFM + 1 log unit (BSN). These fO2 values are consistent with the analysed Fe–Ti oxide compositions (Table 3d) and with literature data for alkali basalts (e.g. Dixon et al., 1997). The crystallization temperatures of coexisting olivine–clinopyroxene phenocrysts calculated for a total pressure <4 kbar using MELTS modelling are virtually the same as those obtained (Table 3b) following Loucks (1996), except for few alkali basalts (Fig. 10).
|
|
The MELTS-modelled fractional crystallization process (Fig. 10) from the least evolved basanites to phonotephritic compositions (P = 2–4 kbar, fO2 = QFM + 1 log unit, H2O = 1 wt %) requires 50 wt % fractionation of an assemblage comprising olivine (27 wt %), clinopyroxene (64 wt %), magnetite (7 wt %) and apatite (2 wt %). The transition (2 kbar, QFM, H2O = 1·0 wt %) from primitive alkali basalts to mugearites may result from 40 wt % fractionation involving clinopyroxene (57 wt %), olivine (19 wt %), plagioclase (12 wt %), and magnetite (12 wt %). The Hy-normative mugearites may derive from transitional basalts at 2 kbar, fO2 = QFM and H2O = 0·5 wt %, through about 60 wt % fractionation involving olivine (16 wt %), clinopyroxene (27 wt %), plagioclase (36 wt %), magnetite (12 wt %) and apatite (1 wt %). Notably, early plagioclase fractionation is exclusive to the transitional basalt suites. These results are consistent with major element mass balance calculations (Stormer & Nicholls, 1978), and with trace element Rayleigh fractionation modelling.
Although it is possible to model the mineralogy and liquid lines of descent of the rocks from the BSN, AKB and TRB suites, the compositional variations within each suite require different parental magmas. For example, the calculated differentiation sequence of a typical LSr alkali basalt (e.g. CA40; Ne = 6, MgO = 8 wt %) shows that the differences between HSr- and LSr-type basalts cannot be explained through fractional crystallization, even assuming different P and fO2 conditions and H2O contents (Fig. 9). Similarly, the compositional differences between the rocks of Mt Bambouto and Ngaoundere Plateau demand different parental magmas.
Crustal contamination
Crustal contamination may be important in alkali basalt petrogenesis (e.g. Wilson et al., 1995), despite the high ascent velocity of the parental magmas. The broad similarity between CVL oceanic and continental basalts (Fitton & Dunlop, 1985) and their unradiogenic Sr and radiogenic Nd isotopic compositions (Halliday et al., 1988) suggest that crustal contamination was probably negligible.
However, a few evolved basaltic rocks, i.e. quartz-normative Bangangte mugearites (e.g. CA74, 42 Ma) and one Hy mugearite (CA13), from the Bambouto area, have isotopic compositions that fall far from the CVL basaltic field (Fig. 8a), i.e. they have high initial 87Sr/86Sr (0·70500–0·70775) and low initial 143Nd/144Nd (0·51257–0·512499). These isotopic compositions, coupled with relatively high SiO2 (53–55 wt %) and relatively low Nb (39–89 ppm) may be indicative of contamination with silicic crustal material, possibly the Pan-African granites of the basement. These granitoids are characterized by high SiO2 (64–77 wt %), low Nb (30 ppm), high 87Sr/86Sr (0·71043–0·72099) and low 143Nd/144Nd (0·51215–0·51165; Halliday et al., 1988; Marzoli et al., 1999; A. Marzoli, unpublished data, 1999). The relatively low LREE/HREE ratio of the mugearite CA74 [(La/Yb)CN = 6·82], compared with the high ratios of associated basalts [(La/Yb)CN = 12–30; Fig. 7], is compatible with the assimilation of basement Pan-African granites that have (La/Yb)CN = 4–7 (Toteau, 1990). An assimilation–fractional crystallization process (AFC; DePaolo, 1981) can explain the high Sr and low Nd isotopic values of the Bangangte mugearites, starting from the transitional basalt CA4, and assuming granite CA18 (Marzoli et al., 1999) as contaminant, with an r value (i.e. rate of assimilated mass/rate of fractionated mass) of 0·3.
In contrast, AFC modelling was not successful in explaining differences between HSrB and LSrB compositions. Common crustal materials are not suitable to change the concentrations of Ba, Sr and P significantly, without substantially affecting major and other trace elements. In summary, crustal contamination is not appreciable in most of the investigated basaltic rocks of the CVL continental sector.
Lithospheric and asthenospheric mantle sources
About 10% of the analysed CVL basaltic rocks have compositions that approach those of primary melts (mg-number values 0·60–0·68; Ni 86–321 ppm; Cr 200–493 ppm; see Table 2). In general, primitive (MgO > 7 wt %) basanites and alkali basalts of Ngaoundere are depleted in FeOt and slightly enriched in SiO2 (Fig. 4a), and have generally higher LREE/HREE ratios than the Western Highlands volcanics with similar MgO (Fig. 7). These differences cannot be due only to fractional crystallization of olivine and clinopyroxene (see Figs 9 and 10). High-pressure experiments on dry lherzolites (e.g. Adam, 1990; Hirose & Kushiro, 1993; Green & Falloon, 1998) suggest that the SiO2 concentration of mantle partial melts are pressure and temperature (i.e. melting degree) dependent, whereas FeOt concentrations are mainly pressure dependent (i.e. increasing with pressure). Therefore, we suggest that Ngaoundere basanites and alkali basalts may have been generated at lower pressure than Bambouto ones. This is consistent with geophysical evidence that the lithosphere is thinned beneath the Ngaoundere Plateau (Plomerova et al., 1993; Poudjom Djomani et al., 1995, 1997). Mt Bambouto basanites are relatively depleted in SiO2 and have high LREE/HREE, compared with Mt Bambouto alkali basalts, whereas FeOt concentrations are similar (Figs 4a and 7; Table 2). We interpret these compositional differences between basanites and alkali basalts to reflect essentially increasing melting degrees.
Incompatible trace element ratios in the least evolved basaltic samples were considered, to provide further constraints on their mantle source(s). REE modelling shows that the least evolved CVL volcanics (basanites, alkali basalts and transitional basalts) fall between the equilibrium batch melting curve for a depleted garnet lherzolite and that for an amphibole–garnet lherzolite with primitive mantle concentrations (Fig. 11a).
CVL basalts of the oceanic sector (Fitton & Dunlop, 1985), of the Ngaoundere Plateau, and LSrB from Mts Oku and Bambouto have similar incompatible element patterns, suggesting that they were generated in the asthenospheric mantle. Their Rb to Y concentrations are similar to those derived by 0·5–1% melting of a depleted garnet lherzolite (McKenzie & O’Nions, 1991; Fig. 11b).
The incompatible element patterns of the basaltic volcanics of HSr type, instead, are exclusive to the continental sector, and approach those of modelled low-degree (1%) melts of an amphibole-bearing garnet lherzolite (Fig. 11c). Compared with the depleted mantle source of the LSrB, the mantle source of HSrB must have been rich in Sr (40 ppm) and Ba (14 ppm), and slightly depleted in Zr (3.0 ppm), and trace amounts of apatite (<1 wt %) in the mantle source of HSrB could account for their high P2O5 concentrations (not shown in Fig. 11). The Sr–Nd isotopic compositions of HSr and LSr rocks suggest that the HSr mantle component was slightly enriched in 87Sr/86Sr, but similar in 143Nd/144Nd isotopic composition to the LSr component (Fig. 8).
|
It should be noted that amphibole-bearing peridotite xenoliths were found in Western Highlands (A. Marzoli, unpublished data, 1999) and in Ngaoundere Plateau volcanics (Lee et al., 1996). The amphiboles in these mantle xenoliths are characterized by positive Ba and Sr anomalies, and by relatively enriched LREE concentrations (Lee et al., 1996). Considering the low P–T stability of amphibole (<30 kbar and <1050°C; see Mengel & Green, 1989; Foley, 1991) and of apatite (Watson, 1980; Baker & Willey, 1992), and considering the 100 km lithospheric thickness (e.g. Plomerova et al., 1993), the amphibole-related chemical signature of the continental CVL magmas suggests a lithospheric mantle source component in the petrogenesis of these HSr basalts. As Western Highlands basalts represent a continuous spectrum from HSr to LSr compositions, we suggest that they may result from mixing of variable amounts of asthenospheric melts derived from an incompatible element depleted anhydrous lherzolite (i.e. LSr component) and of ‘small-volume’ melts from amphibole-bearing lithospheric mantle peridotite (i.e. HSr component).
Mantle enrichment
Incompatible element modelling suggests that the amphibole- and apatite-bearing mantle source of HSrB was enriched in Sr and Ba, and depleted in Zr. Typically, mantle lherzolites affected by carbonate (CO2-rich fluids or carbonatitic magmas) metasomatism may contain amphibole and sometimes apatite, and are characterized by amphibole (and clinopyroxene) enriched in Sr and Ba, and depleted in Zr (e.g. Ionov et al., 1996). It is possible that carbonate metasomatism may have affected the lithospheric mantle source of Mt Bambouto HSrB (and of Mt Etinde nephelinites; Nkoumbo et al., 1995), allowing for Ba, Sr and P enrichments and Zr depletion. As (carbonate) metasomatism produces generally enriched LREE/HREE ratios (e.g. Ionov et al., 1996), a relatively recent enrichment process would be consistent with the 143Nd/144Nd isotopic data, which show no variation between HSrB and LSrB and require a time-integrated LREE-depleted source.
Although the age, origin and nature of the metasomatic event that possibly affected the mantle source of the continental HSrB remains poorly constrained, a Mesozoic enrichment may be hypothesized according to studies on CVL mantle xenoliths and on basalts from the continent–ocean boundary region (Halliday et al., 1990; Lee et al., 1994a, 1996). The CVL mantle xenoliths show evidence of Mesozoic incompatible element enrichments (Lee et al., 1996), which may be related to the activity of the St Helena hotspot. During the Early Cretaceous, the St Helena hotspot was located underneath the Benue Trough, where it generated alkali basaltic magmatism 200 km to the north of the Western Cameroon Highlands (Coulon et al., 1996). Subsequently, the hotspot migrated towards the region at present located underneath southern Cameroon and Nigeria (O’Connor & Le Roex, 1992), where the anomalously high Pb isotopic compositions of the CVL basalts at the continent–ocean boundary, including the HSr Mt Etinde nephelinites (Halliday et al., 1990; Lee et al., 1994a), provide further record of a Mesozoic enrichment event. It may be suggested that magmas or fluids associated with the St Helena mantle plume and the Early Cretaceous alkaline magmatism in the Benue Trough enriched the continental mantle lithosphere, which became later the source of the HSr basalts at the continental–ocean boundary (Mt Etinde) and in the Western Highlands. This scenario is consistent with the absence of HSr basalts on the Ngaoundere Plateau, which was too far from the location of the hotspot, and in the CVL oceanic sector, as no oceanic crust was formed before the Late Cretaceous in the Equatorial Atlantic.
In summary, low degrees of partial melting of a dry, depleted, lherzolite asthenospheric mantle source can account for the common signature of the oceanic and continental CVL basalts, whereas the anomalous chemical characteristics of HSrB, exclusive to CVL continental basalts, suggest the involvement of a metasomatically enriched, amphibole- (and apatite-)bearing lherzolite from the continental lithosphere.
|
SW MIGRATION OF THE VOLCANISM AND PLATE MOTION |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGICAL SETTING40Ar/39Ar AGESANALYTICAL METHODSCLASSIFICATIONPETROGRAPHY AND MINERAL...GEOCHEMISTRYPETROGENESISSW MIGRATION OF THE...REFERENCES |
|---|
The 40Ar/39Ar ages of the silicic volcanism of the Western Highlands (Marzoli et al., 1999
) and the age of the peak basaltic activity at Mt Oku (Ndu, 31–23 Ma, to Bamenda, 25–15 Ma; Njilah, 1991) and Mt Bambouto (21–14 Ma) are consistent with a general SW younging of the volcanism in the Western Highlands. Previous K/Ar data (Gouhier et al., 1974; Fitton & Dunlop, 1985) suggested that this younging trend extends to all the NE–SW aligned continental Cameroon volcanoes, from the Oligocene Mandara volcanic mountains (northern Cameroon) to the mainly Pliocene–Quaternary volcanism of Mts Manengouba, Cameroon and Etinde, and of Bioko island (southwest Cameroon; Fig. 1). K/Ar ages of intrusive rocks of northern Cameroon (e.g. Golda Zuelva and Mboutou, 66–55 Ma; Jacquemin et al., 1982) are broadly consistent with the above NE–SW trend.
Notably, the African plate moved from SW to NE with a velocity of 1·5–2 cm/y in Oligocene–Pliocene times (Pollitz, 1991; Silver et al., 1998</