Journal of Petrology Advance Access originally published online on February 25, 2005
Journal of Petrology 2005 46(6):1155-1201; doi:10.1093/petrology/egi013
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Post-Collisional Transition from Subduction- to Intraplate-type Magmatism in the Westernmost Mediterranean: Evidence for Continental-Edge Delamination of Subcontinental Lithosphere
1 GEOMAR RESEARCH CENTRE FOR MARINE GEOSCIENCES, DEPARTMENT OF VOLCANOLOGY AND PETROLOGY, WISCHHOFSTR. 1–3, 24148 KIEL, GERMANY
2 IFM–GEOMAR LEIBNIZ INSTITUTE FOR MARINE SCIENCES, DYNAMICS OF THE OCEAN FLOOR, WISCHHOFSTR. 1–3, 24148 KIEL, GERMANY
3 INSTITUTE FOR GEOSCIENCES, UNIVERSITY OF KIEL, OLSHAUSENSTR. 40, 24118 KIEL, GERMANY
RECEIVED NOVEMBER 12, 2003; ACCEPTED JANUARY 14, 2005
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONREGIONAL GEOLOGYANALYTICAL METHODSRESULTSDISCUSSIONGEODYNAMIC EVOLUTIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
Post-collisional magmatism in the southern Iberian and northwestern African continental margins contains important clues for the understanding of a possible causal connection between movements in the Earth's upper mantle, the uplift of continental lithosphere and the origin of circum-Mediterranean igneous activity. Systematic geochemical and geochronological studies (major and trace element, Sr–Nd–Pb-isotope analysis and laser 40Ar/39Ar-age dating) on igneous rocks provide constraints for understanding the post-collisional history of the southern Iberian and northwestern African continental margins. Two groups of magmatic rocks can be distinguished: (1) an Upper Miocene to Lower Pliocene (8·2–4·8 Ma), Si–K-rich group including high-K (calc-alkaline) and shoshonitic series rocks; (2) an Upper Miocene to Pleistocene (6·3–0·65 Ma), Si-poor, Na-rich group including basanites and alkali basalts to hawaiites and tephrites. Mafic samples from the Si–K-rich group generally show geochemical affinities with volcanic rocks from active subduction zones (e.g. Izu–Bonin and Aeolian island arcs), whereas mafic samples from the Si-poor, Na-rich group are geochemically similar to lavas found in intraplate volcanic settings derived from sub-lithospheric mantle sources (e.g. Canary Islands). The transition from Si-rich (subduction-related) to Si-poor (intraplate-type) magmatism between 6·3 Ma (first alkali basalt) and 4·8 Ma (latest shoshonite) can be observed both on a regional scale and in individual volcanic systems. Si–K-rich and Si-poor igneous rocks from the continental margins of southern Iberia and northwestern Africa are, respectively, proposed to have been derived from metasomatized subcontinental lithosphere and sub-lithospheric mantle that was contaminated with plume material. A three-dimensional geodynamic model for the westernmost Mediterranean is presented in which subduction of oceanic lithosphere is inferred to have caused continental-edge delamination of subcontinental lithosphere associated with upwelling of plume-contaminated sub-lithospheric mantle and lithospheric uplift. This process may operate worldwide in areas where subduction-related and intraplate-type magmatism are spatially and temporally associated.
KEY WORDS: post-collisional magmatism; Mediterranean-style back-arc basins; subduction; delamination; uplift of marine gateways
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONREGIONAL GEOLOGYANALYTICAL METHODSRESULTSDISCUSSIONGEODYNAMIC EVOLUTIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
Subduction-related and intraplate-type igneous rocks occur together worldwide (e.g. Mediterranean area, Colorado Plateau and Andes) (Kay & Mahlburg Kay, 1993
; Arculus & Gust, 1995; Wilson & Bianchini, 1999). Kilometre-scale continental uplift may occur contemporaneously with the magmatic activity and is thought to reflect lithospheric response to rapid changes in upper mantle geometry (Bird, 1979; Turner et al., 1993; Duggen et al., 2003). Large-scale reorganization of the upper mantle may be related to: (1) roll-back and detachment of subducted oceanic lithosphere (Innocenti et al., 1982; Keller, 1982; Davies & Blanckenburg, 1995; Wilson & Bianchini, 1999; Wortel & Spakman, 2000); (2) detachment or convective thinning of subcontinental lithosphere (Pearce et al., 1990; Platt & England, 1993; Turner et al., 1999; López-Ruiz et al., 2002); or (3) delamination (peeling-off) of subcontinental lithosphere (Bird, 1979; Serri et al., 1993; Docherty & Banda, 1995; Gîrbacea & Frisch, 1998; Duggen et al., 2003). Formation of magmas having a subduction-related geochemical signature can result from: (1) subduction of oceanic lithosphere (Hawkesworth et al., 1993; Pearce & Peate, 1995); (2) melting of metasomatically enriched subcontinental lithosphere with an inherited subduction signature (Turner et al., 1996; Benito et al., 1999); or (3) extensive crustal contamination of MORB-like (mid-ocean-ridge basalt) magmas (Turner et al., 1999). Generation of intraplate-type lavas in continental settings is generally related to upwelling sub-lithospheric mantle material (e.g. Western and Central Europe) (Wilson & Downes, 1991; Granet et al., 1995; Hoernle et al., 1995). However, the processes responsible for uplift and the generation of volcanic areas with geochemically diverse rock types such as subduction-related and intraplate-type igneous rocks are still poorly understood.
Combination of systematic geochemical and geochronological data allows reconstruction of the temporal and spatial transition from post-collisional subduction-related to intraplate-type magmatism in the western Mediterranean. The transition occurs regionally in the westernmost Mediterranean realm and locally in individual volcanic systems such as the Gourougou and Guilliz stratovolcanic complexes in northwestern Africa. The new geochemical and geochronological data for igneous rocks from the continental margins of southern Spain and northern Morocco are integrated with published data for rocks from the westernmost Mediterranean sea floor (Duggen et al., 2004) and northwestern Algeria (Louni-Hacini et al., 1995; Coulon et al., 2002) providing the basis for a case study of this transition.
|
REGIONAL GEOLOGY |
|---|
|
TOPABSTRACTINTRODUCTIONREGIONAL GEOLOGYANALYTICAL METHODSRESULTSDISCUSSIONGEODYNAMIC EVOLUTIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
Neogene post-collisional igneous rocks occur along the European and African margins of the western Mediterranean (Wilson & Bianchini, 1999
). In the Alboran region in the westernmost Mediterranean, Middle Miocene to Pleistocene volcanic rocks can be found in two lithospheric domains: (1) the Alboran Basin (hosting the Alboran Sea), which consists of strongly thinned continental crust progressing into Miocene oceanic crust to the east (Lonergan & White, 1997); (2) the southern Iberian and northwestern African continental margins (Fig. 1).
|
The Alboran Basin area includes Middle to Late Miocene igneous rocks from the Alboran Sea floor (ridges and seamounts), the Alboran Island and coastal stratovolcanic complexes bordered by major strike-slip faults in southern Spain (the Cabo de Gata and Aguilas block) and northern Morocco (Ras Tarf and Trois Furches) (Duggen et al., 2004
) (Fig. 1). The magmatic evolution of the Alboran Basin area is marked by the eruption of magmas of both the low-K (tholeiitic) series and the medium- to high-K (calc-alkaline) series in the Middle to Late Miocene (Figs 1 and 2). The mafic Alboran Basin lavas have major and trace element and O–Sr–Nd–Pb-isotope compositions characteristic of tholeiitic volcanic front and calc-alkaline rear-arc lavas in active subduction zones such as the Izu–Bonin island arc (Benito et al., 1999; Hoernle et al., 1999; Duggen et al., 2004).
|
The Iberian and African continental areas surrounding the Alboran Basin are marked by igneous activity with a different geochemical composition. Upper Miocene to Pleistocene post-collisional magmatism along the southern Iberian and northwestern African continental margins is characterized by Si-rich, K-rich igneous rocks and Si-poor, Na-rich lavas (Fig. 1). Upper Miocene to Lower Pliocene Si–K-rich igneous rocks include latites, shoshonites and ultrapotassic rocks such as lamproites (Fig. 2) (Bellon & Brousse, 1977
; López Ruiz & Rodríguez Badiola, 1980; Venturelli et al., 1984; Hernandez, 1986; Louni-Hacini et al., 1995; El Bakkali et al., 1998; Benito et al., 1999; Coulon et al., 2002). These Si–K-rich igneous rocks are often spatially associated with Si-poor, Na-rich lavas such as alkali basalts and basanites to hawaiites and tephrites. The first Si-poor, Na-rich alkali basalts appeared in the Upper Miocene and continued to erupt into the Quaternary (Figs 1 and 2) (Duggen et al., 2003). The post-collisional transition from Si–K-rich to Si-poor, Na-rich magmatism occurred close to the Miocene–Pliocene boundary that is defined by the Messinian Salinity Crisis and can clearly be seen in the field in the Guilliz volcanic area, where dark alkali basalt lava flows were erupted on top of light-coloured, Si–K-rich, pyroclastic deposits of an eroded stratovolcanic complex.
A large number of models have been proposed to explain the geodynamic evolution of the Alboran region in the westernmost Mediterranean since the Eocene and can be divided into three groups: (1) mantle diapirism (Weijermars, 1985); (2) subduction of oceanic lithosphere involving slab roll-back and slab detachment (Royden, 1993; Zeck, 1996; Lonergan & White, 1997; Hoernle et al., 1999; Coulon et al., 2002; Gutscher et al., 2002; Duggen et al., 2003, 2004); (3) convective removal (detachment) and peeling-off (delamination) of subcontinental lithosphere (Platt & Vissers, 1989; Docherty & Banda, 1995; Houseman, 1996; Seber et al., 1996a; Comas et al., 1999; Turner et al., 1999; López-Ruiz et al., 2002; Platt et al., 2003). In this study, the temporal and spatial geochemical evolution of the magmatism in the westernmost Mediterranean is integrated with geophysical and geological data from the literature to distinguish between these models. We developed a three-dimensional model that describes the mantle geodynamic setting in the westernmost Mediterranean area near the Miocene–Pliocene boundary.
|
ANALYTICAL METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONREGIONAL GEOLOGYANALYTICAL METHODSRESULTSDISCUSSIONGEODYNAMIC EVOLUTIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
40Ar/39Ar isotope ratios were measured on K-bearing phases such as amphibole, biotite and feldspar phenocrysts and glass and matrix chips, at the GEOMAR Tephrochronology Laboratory. The freshest possible glass and matrix chips and phenocrysts were hand-picked from crushed and sieved splits and cleaned using an ultrasonic disintegrator. Plagioclase phenocrysts were additionally etched in hydrofluoric acid (15%). For several samples, fresh matrix chips were separated and analysed together with phenocrysts to examine the reliability of age data measured for matrix chips from aphyric igneous rocks. The samples were neutron-irradiated at the 5 MW reactor of the GKSS Reactor Centre in Geesthacht (Federal Republic of Germany), with crystals and glass and matrix chips in aluminium trays and irradiation cans wrapped in 0·7 mm cadmium foil. Age determinations by laser 40Ar/39Ar analysis were performed by fusing single crystals and glass and matrix chips in a single step or by laser-step heating analysis when necessary. Purified gas samples were then analysed using a MAP 216 series noble gas mass spectrometer. Raw mass spectrometer peaks were corrected for mass discriminations background, and blank values determined every fifth analysis. Neutron flux during irradiation was monitored using TCR sanidine (Taylor Creek Rhyolite, 27·92 Ma) (Dalrymple & Duffield, 1988
; Duffield & Dalrymple, 1990) and an internal standard SAN6165 (0·470 Ma) (van den Bogaard, 1995). Vertical variations in J values were quantified by a cosine function fit. Lateral variations in J were not detected. Optical grade CaF2 and high-purity K2SO4 salt crystals that were irradiated and analysed together with the samples were used for corrections of interfering neutron reactions on Ca and K. For statistical reasons, 3–19 particles of each phase were separately dated to calculate mean apparent ages and isochron ages. For samples that did not produce statistically valid isochron or mean apparent ages, step heating ages on single matrix chips were additionally determined. The age with the highest confidence level was chosen for each sample (Table 1), taking into account mean square weighted deviates (MSWD), plateau portions for step heating analysis, deviation of the data in the isochron plots (see Electronic Appendix 1, which may be downloaded from http://www.petrology.oupjournals.org) and the initial 40Ar/39Ar ratio. Errors on the age data are reported at the 1
confidence level.
|
For bulk-rock geochemical analyses, whole-rock sample chips (0·1–1 cm in diameter) were hand-picked using a binocular to minimize the effects of alteration; these were then cleaned ultrasonically with de-ionized water to extract salts, and then dried and pulverized in an agate mill. Bulk-rock analyses of major and selected trace elements (Ba, Co, Cr, Cu, Ga, Rb, Sr, V and Zr) were performed on fused tablets with a Philips X'unique PW 1480 X-ray fluorescence spectrometer (XRF) at the GEOMAR Research Centre. International reference standards JB-2, JB-3 (basalts), JA-2, JA-3 (andesites), JR-2, JR-3 (rhyolites), JG-2, GM (granites) and JF-1 (feldspar) were used to evaluate the precision and accuracy of the measurements. Na contents were normalized to the standard values to account for X-ray tube drift for this element. H2O and CO2 were determined using a Rosemont Infrared Photometer CSA 5003. The new data are presented in Table 2 and Electronic Appendix 2. Values of major elements given in the text and used for plotting and rock classification were calculated on a volatile-free basis following a correction of Ca contents assuming that CO2 comes from CaCO3.
|
Additional trace elements (Li, Y, Nb, Mo, Sn, Sb, Cs, REE, Hf, Ta, W, Tl, Pb, Th and U) were determined from mixed acid (HF–aqua regia–HClO4) pressure digests by inductively coupled plasma-mass spectrometry (ICP-MS) at the Institute for Geoscience, University of Kiel, using an upgraded (High-Performance Interface, torch box) VG Plasmaquad PQ1. The precision of analytical results as estimated from duplicate measurements was better than 3% RSD. Details of the analytical procedure have been described in the literature (Garbe-Schönberg, 1993
). Blanks and international standards, BE-N and BHVO-1, were analysed with the samples, to evaluate the precision and accuracy of the measurements (Duggen, 2002). The ICP-MS trace element data are presented in Table 2 and Electronic Appendix 2.
Sr, Nd and Pb isotopic analyses were carried out on whole-rock powders. To minimize the effects of seawater alteration, the powders were leached with hot distilled 6 M HCl for 1 h (except the highly evolved samples). After dissolution with Merck Ultrapur HF, HNO3 and HCl, Sr, Nd and Pb were extracted by conventional ion exchange chromatographic techniques. Isotope ratios of all samples were measured by thermal ionization mass spectrometry (TIMS) in static (Sr, Nd, Pb) and dynamic (Sr, Nd) mode on a Finnigan MAT262 system at GEOMAR. Measured 87Sr/86Sr and 143Nd/144Nd ratios were normalized within-run to 86Sr/88Sr = 0·1194 and 146Nd/144Nd = 0·7219. The Sr and Nd isotope ratios were further normalized to 87Sr/86Sr = 0·71025 (NBS 987 standard) and 143Nd/144Nd = 0·51185 (La Jolla standard) or 143Nd/144Nd = 0·51171 (internal Nd-Spex standard calibrated to La Jolla) to correct for mass spectrometer drift. Replicate analyses of standards NBS 987 (Sr), La Jolla and Nd-Spex (Nd) used for the evaluation of the external precision gave mean values of 87Sr/86Sr = 0·710242 ± 0·000017 (2 SD) (n = 39), 143Nd/144Nd = 0·511841 ± 0·0000010 (2 SD) (n = 13) and 143Nd/144Nd = 0·511706 ± 0·000008 (2 SD) (n = 12). Replicate analyses of the NBS 981 (Pb) standard yielded a mean value of 206Pb/204Pb = 16·894 ± 0·004 (2 SD), 207Pb/204Pb = 15·433 ± 0·005 (2 SD) and 208Pb/204Pb = 36·515 ± 0·017 (2 SD.) (n = 23). Pb-isotope data were corrected for 0·1238
mass discrimination per atomic mass unit (a.m.u.). Total analytical blanks were on average <0·100 ng and are thus negligible given the generally high Pb content of the samples.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONREGIONAL GEOLOGYANALYTICAL METHODSRESULTSDISCUSSIONGEODYNAMIC EVOLUTIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
40Ar/39Ar geochronological data
Laser probe 40Ar/39Ar age data for 45 volcanic rocks from the Iberian and African continental margins in southeastern Spain and northern Morocco are reported in Table 1. Phenocryst ages are identical within error, or in very good agreement with each other, in all samples (e.g. feldspar and amphibole in sample CX110599 [GenBank] -3, biotite and amphibole in sample TA200400-1, biotite and feldspar in GZ280300-1). Matrix chip ages are also identical within error and in accordance with the phenocryst ages (e.g. matrix and feldspar in GG140699-8, matrix and feldspar and biotite in GZ280300-3b, matrix and feldspar in GZ180699-4, matrix and biotite in OD190699-1a), showing that analyses of carefully prepared matrix chips can provide reliable age data. Two Si–K-rich samples (lamproites) from southeastern Spain yielded phlogopite ages (FT120599-5, BQ170599 [GenBank] -9) with high MSWD values as a result of a large variation in single phenocryst ages. These phlogopite ages are significantly older than the matrix ages and are unlikely to reflect eruption ages. Three other Si–K-rich (lamproite) samples (CT270300 [GenBank] -2c, ZN130599-1, VE290699-10A) provided phlogopite ages having low MSWD and initial 40Ar/36Ar ratios close to the atmospheric value and are thus considered to represent eruption ages. Below we systematically discuss the laser probe 40Ar/39Ar age data for the Si–K-rich igneous rocks followed by the Si-poor lavas from four areas on the Iberian and African continental margins: southeastern Spain and the Gourougou, Guilliz and Oujda volcanic areas in northwestern Africa (Fig. 1).
The southern Iberian continental margin
Laser probe 40Ar/39Ar age data for 13 Si–K-rich rocks (lamproites and their derivatives) from southeastern Spain range from 8·19 to 6·37 Ma (Table 1 and Figs 1 and 2) and are within the range of published K/Ar ages (8·6–6·2 Ma) (Nobel et al., 1981; Bellon et al., 1983). The 10·8 ± 1 Ma K/Ar age for a lamproite from the Vera basin (Bellon & Brousse, 1977) was not confirmed by laser 40Ar/39Ar age dating. The 6·76 ± 0·04 Ma matrix 40Ar/39Ar age from La Celia near Jumilla is identical within error to a published Ar/Ar whole-rock age of 6·8 ± 0·4 Ma (Turner et al., 1999). A systematic relationship between the new Ar/Ar ages and older K/Ar age data was not detected. As inferred from geological maps, the laser 40Ar/39Ar age data are consistent with stratigraphic evidence showing that lamproitic rocks are generally associated with Upper Miocene marine sediments (mostly marls and sandstones) in the Murcia area. Field observations in the Vera basin show that lamproitic lava flows, showing pillow structures, were emplaced and chilled in soft, wet Messinian marine muds (peperites). Two lamproitic peperites were dated at 6·44 Ma and 6·37 Ma and are consistent with the Messinian ages of their host marine sediments. A newly discovered lamproitic dyke (VE290699-10A) yielded a phlogopite age of 7·45 ± 0·08 Ma. It intrudes rhyodacitic breccias cut by the Palomares fault and the dyke itself has been tectonically deformed. Thus the new age data also provide constraints for the minimum age of the breccia and the tectonic history of the Palomares fault zone.
Si-poor, Na-rich volcanic rocks from the Tallante volcanic field (two hawaiites and one trachybasalt), which are spatially associated with the Si–K-rich (lamproitic) rocks, yielded amphibole + biotite, glass and matrix mean apparent ages and step heating ages ranging from 2·93 to 2·29 Ma (Table 1 and Figs 1 and 2). For Tallante, the ages are consistent with field evidence showing that the volcanic units overlie Tortonian, Messinian and Pliocene sediments and are in good agreement with published K/Ar ages ranging from 2·8 to 2·7 Ma (Bellon et al., 1983). The 10·5 Ma Ar/Ar phlogopite age for a Tallante lava reported by Turner et al. (1999) is not consistent with these Pliocene ages. Phlogopite is not a common phenocryst phase in basaltic magmas, as it is stable only at high pressures and becomes unstable upon extrusion (Deer et al., 1996). Ultramafic mantle xenoliths frequently associated with the alkali basalts from Tallante contain phlogopite (Dupuy et al., 1986; Capedri et al., 1989) and, therefore, the phlogopites dated by Turner et al. (1999) may be xenocrysts from disaggregated mantle xenoliths. In conclusion, the transition from Si–K-rich to Si-poor, Na-rich volcanic rocks occurred between 6·4 to 2·9 Ma at the southern Iberian continental margin.
The northwestern African margin
Si–K-rich igneous rocks from the Gourougou volcanic field (basaltic trachyandesites, trachyandesites, andesite and one diorite) yielded matrix mean apparent and isochron ages on biotite and feldspar ranging from 7·58 to 4·8 Ma (Table 1 and Figs 1 and 2). These ages are largely consistent with the range of whole-rock and biotite-separate K/Ar age data (9·0 to 4·6 Ma) (Bellon & Brousse, 1977; Hernandez & Bellon, 1985; El Bakkali et al., 1998) and 40Ar/39Ar age data (6·73 to 6·0 Ma) (Roger et al., 2000) for volcanic tuffs interbedded with Messinian marine sediments, which were deposited on the slope of the Gourougou stratovolcanic complex. Si-poor, Na-rich alkali basalts and trachybasalts were erupted on the flanks of the Gourougou stratovolcano between 6·3 and 3·73 Ma (Table 1). Lava flows directly overlie Messinian marine sediments (e.g. Haidoun flow, dated at 3·73 Ma). Previously determined K/Ar ages generally tend to be younger (e.g. 2·5 Ma for Haidoun) (Bellon & Brousse, 1977) than the laser 40Ar/39Ar age data. In the Gourougou volcanic system, the transition from Si–K-rich to Si-poor, Na-rich igneous activity occurred between 6·3 and 4·8 Ma.
Si–K-rich igneous rocks (trachyandesite to trachyte) from the Jebel Guilliz stratovolcano located in the centre of the Guilliz volcanic field yielded mean apparent ages for feldspar and biotite + feldspar ranging from 6·90 to 6·82 Ma (Table 1 and Figs 1 and 2). Mean apparent ages for matrix and feldspar + matrix from Si-poor, Na-rich alkali basalts in the Guilliz volcanic field range from 6·3 to 0·65 Ma (Table 1 and Figs 1 and 2). Most published K/Ar ages (8·0 to 4·9 Ma) (Hernandez & Bellon, 1985) fall within this range but some K/Ar ages tend to be older. Major discrepancies between the new laser 40Ar/39Ar age data (0·65 Ma and 0·88 Ma) and the K/Ar (1·74 Ma and 2·5 Ma) (Hernandez & Bellon, 1985) data from the literature exist for the Si-poor lavas from the Ain Zora and Jebel El Kehal lava flows in the westernmost Guilliz area. The new age data suggest that volcanism in the westernmost Guilliz volcanic field was active contemperanously with the Quaternary alkali basaltic volcanism of the nearby Middle Atlas mountains (1·8 to 0·5 Ma, K/Ar ages) (Bellon & Brousse, 1977; Harmand & Cantagrel, 1984). In the Guilliz area, the transition from Si–K-rich to Si-poor, Na-rich igneous rocks occurred between 6·8 and 6·3 Ma.
The Oujda–Algeria area includes igneous rocks from northwestern Algeria, the Oujda volcanic field and the Plateau du Rekkam area situated south of the Oujda area. Based on K/Ar age data, Si–K-rich magmatism (high-K calc-alkaline and shoshonitic series) in northwestern Algeria occurred at the Middle–Upper Miocene boundary and ceased in the Upper Miocene (11·7 to 7·5 Ma) (Louni-Hacini et al., 1995; Coulon et al., 2002). Si-poor, Na-rich rocks (basanites, alkali basalts and trachyandesites) from the Oujda volcanic field yielded matrix and biotite mean apparent ages ranging from 3·78 to 3·10 Ma (Table 1). These ages show that the eruption history of the Oujda volcanic field has a much narrower range than hitherto thought based on published whole-rock K/Ar age data (6·2 to 1·5 Ma) (Bellon & Letouzey, 1977; Tisserant et al., 1985; Andries & Bellon, 1989). Two altered lava flows outcropping in the vicinity of the Oued Isly southwest of Oujda were K/Ar dated at 6·2 and 5·6 Ma (Andries & Bellon, 1989). These flows overlie Tortonian blue marls (Marnes bleues) and so may correlate with Messinian marine sediments (Marnes jaunes) in the Oujda area (Jadid et al., 1999). However, redating of the flows in this area with the laser 40Ar/39Ar technique (3·78 to 3·10 Ma) was not able to confirm the Late Miocene K/Ar ages (Table 1). Based on K/Ar age data, Si-poor volcanic rocks from northwestern Algeria show a much wider age range (10·0–7·2 Ma and 4·3–0·8 Ma) (Fig. 2) (Louni-Hacini et al., 1995; Coulon et al., 2002) than Si-poor volcanic rocks from the Oujda volcanic field (3·78–3·10 Ma) (this study). However, additional laser 40Ar/39Ar age dating is necessary to confirm the older K/Ar ages from northwestern Algeria, in particular the onset and termination of Si–K-rich and Si-poor magmatism. Plateau du Rekkam volcanic rocks SE of the Oujda volcanic field probably erupted in the Plio-Quaternary (Saadi et al., 1985). In the Oujda–Algeria area, the transition from Si–K-rich to Si-poor, Na-rich magmatism apparently took place between 10·0 and 7·5 Ma.
In summary, the transition from Si–K-rich to Si-poor volcanism occurred between 6·4 and 2·9 Ma in southern Spain, between 6·3 and 4·8 Ma in the Gourougou area, between 6·8 and 6·3 Ma in the Guilliz area, and possibly between 10·0 and 7·5 Ma in the Oujda–Algeria area.
Major and trace element data
Major and trace element data for 145 igneous rocks from the southern Iberian and northwestern African continental margins are presented in Table 2 and Electronic Appendix 2. The Electronic Appendix includes the full dataset and is available from the Journal of Petrology website at http://www.petrology.oupjournals.org, whereas Table 2 is a reduced version that includes major and trace element data only for samples that were selected for Sr–Nd–Pb-isotope analysis.
The comprehensive dataset allows division of the post-collisional igneous rocks from the westernmost Mediterranean into a Si-rich and a Si-poor group (Fig. 3a). Based on the K2O content, the Si-rich samples can be further subdivided into a relatively K-poor group and a K-rich group (Fig. 3b). The Middle to Upper Miocene (12·1 to 6·1 Ma) K-poor group includes low-K (tholeiitic) and medium- to high-K (calc-alkaline) basalts through rhyolites outcropping in the Alboran Basin area (Hoernle et al., 1999; Duggen et al., 2004; Gill et al., 2004). The occurrence of Upper Miocene to Lower Pliocene (8·2 Ma and 4·8 Ma) K-rich magmatism, including high-K to shoshonitic series basaltic andesites and basaltic trachyandesites to dacites and trachytes and lamproitic rocks is restricted to the southern Iberian and northwestern African continental margins. Upper Miocene to Pliocene (6·3 and 0·65 Ma) Si-poor volcanic rocks are spatially associated with the Si–K-rich group rocks, and range from alkali basalts and basanites to hawaiites and tephrites and tend to have higher Na2O contents for a given K2O concentration than the K-rich group. Below we describe the major and trace element composition of the Si–K-rich and Si-poor group rocks from the continental margins of southern Iberia and northwestern Africa.
|
The southern Iberian continental margin
Si–K-rich volcanic rocks from southeastern Spain (lamproites and their derivatives) show a large range in most major and trace element contents. MgO (1·5–12·9 wt %) correlates inversely with SiO2 and Al2O3 and positively with TiO2, Nb, Ta, Nd, Sm, Zr, Eu, Hf, Gd, Sn, Cr and Ni (with r2 for the regression line >0·7, except for Ni, which shows a curved correlation with MgO) (Fig. 4). Most of these systematic geochemical variations can be explained by fractional crystallization of common phenocryst phases in the Spanish lamproites such as olivine (containing Cr-spinel), phlogopite, apatite and sanidine (Venturelli et al., 1988
; Toscani et al., 1995). There is a compositional continuum between high-MgO lamproites, their low-MgO derivatives (some of which contain crustal xenocrysts such as garnet and cordierite) and cordierite–garnet-bearing volcanic rocks from southeastern Spain (e.g. Cerro Hoyazo) (Fig. 4a–d). The Si–K-rich, lamproitic rocks from southern Spain have enriched incompatible element contents and exhibit spiked patterns on primitive mantle normalized multi-element diagrams (Fig. 5a), resulting from an enrichment of fluid-mobile, incompatible elements such as Rb, Ba, (± Th), U, K, Pb (and Cs, Sn, Sb and Li) relative to incompatible elements that are less fluid-mobile or fluid-immobile [e.g. rare earth elements (REE), Nb, Ta]. Strong enrichments of the light and middle REE relative to heavy REE result in high (LREE, MREE)/HREE ratios (e.g. La/Yb = 34–71). A strong LREE enrichment is associated with depletion of Eu relative to Sm and Gd as reflected by negative Eu anomalies (Eu/Eu* = 0·52–0·67). Enrichment of fluid-mobile relative to less fluid-mobile elements and pronounced troughs in Nb and Ta are typical features of subduction zone magmas (e.g. Aeolian Islands and Izu–Bonin arc lavas in Fig. 5c) (Ellam et al., 1988, 1989; Taylor & Nesbitt, 1998).
|
|
Si-poor volcanic rocks from Tallante in southeastern Spain have MgO contents ranging from 3·6 to 8·5 wt %. MgO correlates negatively with Al2O3, Nb, Ta, Zr and Hf and positively with TiO2, Cr and Ni (r2 > 0·7) (Fig. 4a–d), generally consistent with the fractionation of the observed phenocryst assemblages (e.g. olivine, clinopyroxene and magnetite). The Si-poor group samples from the Iberian continental margin generally exhibit contrasting multi-element patterns to the Si–K-rich group showing relative enrichments of fluid-immobile Nb and Ta and relative depletion of fluid-mobile elements K, ± Pb, and have lower Rb, Ba, Th, U, K and Pb, but higher Sr and Ti concentrations than the Si–K-rich volcanic rocks. The Si-poor group lavas have geochemical signatures similar to those of ocean-island basalts (OIB) and lavas from continental intraplate volcanic settings (e.g. Massif Central, St. Helena and Canary Islands in Fig. 5d). The Tallante lavas, however, show deviations from typical intraplate-type, multi-element patterns with positive Th–U anomalies, highly variable K anomalies and Pb anomalies ranging from negative to positive.
The northwestern African continental margin
The Si–K-rich igneous rocks from the northwestern African margin also show a large variation in major and trace element composition. MgO ranges from 0·1 to 6·6 wt % and correlates positively with CaO, TiO2, FeO, CaO/Al2O3, Co (Gourougou) and MnO, Co, V (Guilliz) and negatively with SiO2, K2O, total alkalis, Rb (Gourougou) and K2O and Li (Guilliz) (r2 > 0·7) (Fig. 4e–h). In multi-element diagrams, Si–K-rich group rocks from Gourougou at the northwestern African margin show spiked patterns as a result of enrichment of fluid-mobile elements Rb, U and K relative to less fluid-mobile or fluid-immobile elements (e.g. REE, Nb and Ta) (Fig. 6a). However, the behaviour of Pb is more complex, showing both positive and negative anomalies, and the troughs in Nb and Ta of some samples are only moderately pronounced. The more evolved Si–K-rich samples from Guilliz have smooth patterns and show neither positive nor negative anomalies for the fluid-mobile elements Rb, Th, U, K and Pb and fluid-immobile elements Nb and Ta. The only negative anomalies are for Ba, Sr and Ti, and there is a slight positive anomaly for Zr (Fig. 6c). The Si–K-rich sample from the Oujda–Algeria area (Ahfir sample; open circle, Fig. 6e) shows enrichment of fluid-mobile Rb, Ba, Th and U relative to fluid-immobile Nb and Ta but lacks peaks for fluid-mobile elements K and Pb. The multi-element pattern of the Ahfir sample is largely subparallel to that of both the mafic and evolved Si–K-rich samples from northwestern Algeria (Fig. 6e) (Louni-Hacini et al., 1995; Coulon et al., 2002). Volcanic rocks having distinct troughs in Nb and Ta but only slight positive peaks for fluid-mobile elements can be found in the active Aeolian subduction zone (Fig. 5c).
|
For the Si-poor, Na-rich lavas, MgO (1·8–12·6 wt %) correlates positively with Cr, Ni, Ga (Gourougou), Cr, Ni (Guilliz) and CaO, CaO/Al2O3, Cr, Ni (Oujda area) and inversely with Al2O3, K2O, Th (Gourougou), Al2O3 (Guilliz) and Al2O3, MnO, Cs, Zr, Hf, Eu, Yb, Lu and Y (Oujda) (r2
0·7, except for Cr and Ni, which show curved correlations with MgO) (Fig. 4e–h). The Oujda samples also show a strong decrease in TiO2, FeO, Co and V as MgO decreases below 3 wt %. These systematic correlations between major and trace elements point to the fractional crystallization of the observed phenocryst assemblages (olivine, clinopyroxene and magnetite). In multi-element diagrams (Fig. 6b, d, f), the Si-poor lavas from northwestern Africa generally show contrasting patterns to the Si–K-rich group as a result of a relative depletion in fluid-mobile elements Rb, Th, U, K and Pb, compared with less fluid-mobile elements such as the REE, Nb and Ta. These patterns are similar to those of lavas from oceanic intraplate-type volcanic settings (e.g. St. Helena and Canary Islands in Fig. 5d) (Chaffey et al., 1989; Thirlwall, 1997). Several Si-poor samples from Gourougou and Guilliz, however, show only weak Nb and Ta anomalies, and a distinct negative K anomaly is lacking. It should be emphasized that several samples from both the Si–K-rich group and the Si-poor group show deviations from typical subduction-related or intraplate-type geochemical fingerprints and have hybrid compositions between the two end-member geochemical signatures (Figs 5 and 6). Some of the Si–K-rich samples lack relative enrichment and even show slight relative depletion in some fluid-mobile elements (in particular Th, U, K and Pb for Guilliz and Pb for Gourougou lavas). These samples also display large ranges in fluid-mobile to fluid-immobile element ratios, which have values between those found in OIB and subduction zone lavas. The Si–K-rich volcanic rocks from Gourougou, for example, have U/Nb = 0·09–0·33 and K/La = 732–1143, whereas similar rocks from Guilliz have U/Nb = 0·03–0·07 and K/La = 461–582. Some mafic Si-poor samples exhibit less pronounced depletion or even slight enrichment in fluid-mobile elements (e.g. Th, U, K and Pb for Guilliz and Gourougou lavas and Th, U, K and Pb for Tallante lavas). Selective addition of strongly fluid-mobile elements relative to much less fluid-mobile elements, which have almost identical partition coefficients (such as Pb relative to Ce) cannot be significantly controlled by differences in the degree of partial melting or by fractional crystallization processes in mafic magmas but could reflect binary mixing of magmas from different sources, derivation from a subduction-modified mantle source or crustal assimilation.
Sr–Nd–Pb-isotope data
Sr–Nd–Pb-isotope data for 52 igneous rocks from the southern Iberian and northwestern African continental margins are reported in Table 3. Figures 7, 8, 10 and 11 display age-corrected Sr–Nd–Pb-isotope ratios.
|
18O = 5·6–6·5
|
|
The southern Iberian continental margin
Si–K-rich, lamproitic rocks have high initial 87Sr/86Sr ratios (0·7161–0·7230) and low initial 143Nd/144Nd ratios (0·51215–0·51223) (Fig. 7a, c, d). The Sr-isotope ratios reported here slightly extend the range of published data (0·7166–0·7207) (Nelson, 1992
; Turner et al., 1999) but the new Nd-isotope ratios are significantly higher and show much less scatter than published data (143Nd/144Nd = 0·51197–0·51214, normalized to 146Nd/144Nd = 0·7219 and age corrected to 7 Ma) (Nelson, 1992; Turner et al., 1999). The Si–K-rich, lamproitic rocks show a relatively narrow range in initial 206Pb/204Pb (18·71–18·85), 207Pb/204Pb (15·66–15·69) and 208Pb/204Pb ratios (38·90–39·13) and have high 
7/4 (14·3–16·5) and 8/4 (61·4–76·8) (Fig. 7b) [7/4 and 8/4 notations denote the vertical deviation from the Northern Hemisphere Reference Line (NHRL), in the 206Pb/204Pb vs 207Pb/204Pb and 208Pb/204Pb diagrams, respectively] (Hart, 1984). It should be noted that the Si–K-rich, lamproitic rocks, however, have Nd and Pb isotopic compositions very similar to those of cordierite-bearing rocks from southeastern Spain (Fig. 7a, b, d).
Si-poor, Na-rich lavas from southeastern Iberia have relatively low initial 87Sr/86Sr ratios (0·7037–0·7046) and relatively high initial 143Nd/144Nd ratios (0·51283–0·51298) (Fig. 7a, c, d). Despite higher initial 206Pb/204Pb (18·88–18·98) compared with the Si–K-rich, lamproitic rocks, the Si-poor lavas from Tallante have similar 207Pb/204Pb and 208Pb/204Pb ratios (15·65–15·67 and 38·9–39·0) to the Si–K-rich rocks from the same area and therefore plot well above the NHRL, also having relatively high 7/4 and 8/4 (10·2–13·3 and 38·6–54·7; Fig. 7b).
The northwestern African continental margin
Si–K-rich igneous rocks from the African continental margin show a wide range of initial 87Sr/86Sr ratios (0·7040–0·7100) and initial 143Nd/144Nd ratios (0·51236–0·51284), similar to the range reported for the Oujda–Algeria area (87Sr/86Srinitial = 0·7043–0·7096 and 143Nd/144Ndinitial = 0·51226–0·51254) (Coulon et al., 2002) (Fig. 7a, c, d). On Pb-isotope diagrams, they show a limited range for initial 206Pb/204Pb (18·7–19·0) and 207Pb/204Pb (15·63–15·70) ratios and plot above the NHRL (7/4 = 7·1–12·9, 8/4 = 30·7–56·9; Fig 7b).
Most Si-poor, Na-rich lavas from the northwestern African continental margin have lower initial 87Sr/86Sr (0·7029–0·7045) and higher 143Nd/144Nd (0·51274–0·51303) than the spatially associated Si–K-rich rocks and overlap with the range reported for Si-poor rocks from northern Algeria (87Sr/86Srinitial = 0·7032–0·7060 and 143Nd/144Ndinitial = 0·51255–0·51292) (Coulon et al., 2002) (Fig. 7a, c, d). The African Si-poor rocks exhibit a large range in initial Pb-isotope ratios (206Pb/204Pb = 18·85–20·75, 207Pb/204Pb = 15·65–15·67, 208Pb/204Pb = 38·9–39·0), although the range within the individual volcanic areas is much more limited, with the exception of lavas from the Plateau du Rekkam volcanic field in the Oujda–Algeria area (Fig. 7b). Si-poor samples from Gourougou and Guilliz plot well above the NHRL (except one sample) (7/4 = –0·1–9·2, 8/4 = –2·2–43·8) and the data fields overlap those of the spatially associated Si–K-rich rocks but extend to more radiogenic Pb isotopic composition. Most Si-poor volcanic rocks from the Oujda–Algeria area plot below the NHRL (7/4 = –7·7–5·8, 8/4 = –42·1–34·8) and fall within the field for Ahaggar volcanic rocks (southern Algeria) (Allègre et al., 1981) and two samples have 206Pb/204Pb ratios as high as St. Helena OIB (Chaffey et al., 1989). The variation in 206Pb/204Pb ratio for Si-poor lavas from the northwestern African margin almost covers the field for the modelled composition of young HIMU plotting below the NHRL, reflecting the composition of MORB with a 238U/204Pb of 10 and a 0·5–1·5 Gyr recycling time (Thirlwall, 1997) (Fig. 7b). In contrast to the young HIMU field, most Si-poor lavas plot closer to or lie above the NHRL; Si-poor Oujda lavas, however, overlap with the low-velocity component (LVC) (Hoernle et al., 1995).
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONREGIONAL GEOLOGYANALYTICAL METHODSRESULTSDISCUSSIONGEODYNAMIC EVOLUTIONCONCLUSIONSSUPPLEMENTARY DATAREFERENCES |
|---|
Temporal evolution of volcanism at the Iberian and African margins
Dating of the westernmost Mediterranean igneous rocks suggests that the transition from Si–K-rich to Si-poor, Na-rich compositions occurred between 6·4 and 2·9 Ma in Spain along the southern Iberian continental margin, between 6·8 and 4·8 Ma in Morocco and possibly between 10·0 and 7·5 Ma in Algeria within the northwestern continental African margin (Fig. 8a–h) (Louni-Hacini et al., 1995
; El Bakkali et al., 1998; Coulon et al., 2002).
The transition also occurred in individual volcanic systems such as the Gourougou and Guilliz stratovolcanic complexes over shorter timescales (Figs 1, 2 and 8a–h). At the Gourougou stratovolcanic complex, Si–K-rich igneous activity occurred from the Late Miocene to Early Pliocene (7·58 to 4·8 Ma) partially overlapping the Latest Miocene to Early Pliocene Si-poor volcanism (6·3 to 3·73 Ma), which erupted from localized vents on the flanks of the stratovolcano. In the Guilliz volcanic field, a short interval of c. 0·4 Myr separates Late Miocene Si-rich explosive and intrusive volcanism (6·90 to 6·69 Ma) from the eruption of lava flows from scoria cones with Si-poor compositions in the Latest Miocene to Pleistocene (6·3 to 0·65 Ma). The Si–K-rich Gourougou igneous rocks show a systematic change in their geochemical characteristics with decreasing age over a time span of about 3 Myr (7·58 to 4·8 Ma). This is observed in their major and trace element and Nd isotope ratios (e.g. K2O/Na2O, K2O/Al2O3, Ba/Sr, Th/Nb, K/Nb, K/La, Pb/Ce, 87Sr/86Sr, 7/4Pb decrease and Nb/La, Nb/Zr and 143Nd/144Nd increase) (Fig. 8a–c, f, h). A similar geochemical evolution is observed in the Si-poor Gourougou lavas, which show a systematic decrease of incompatible element ratios (e.g. Ba/Sr, K/La, K/Nb, Pb/Ce) with decreasing age (Fig. 8a, f). The systematic temporal variations of both isotope and incompatible element ratios with very similar partition coefficients cannot be explained by differences in the degree of partial melting or by fractional crystallization, but instead must reflect crustal assimilation or binary mixing between different components from a heterogeneous mantle source.
Mantle source identification
As illustrated in the Sr–Nd–Pb-isotope diagrams (Fig. 7a–d), a number of distinct mantle and/or source components appear to play a role in the petrogenesis of the Si–K-rich and Si-poor groups: (1) lithospheric components such as the continental crust and subduction-modified lithospheric mantle (e.g. Enriched Mantle 2, EM2) having high 87Sr/86Sr, 7/4 and 8/4, moderate 206Pb/204Pb ratios and generally low 143Nd/144Nd ratios, and (2) sub-lithospheric components such as young and old HIMU mantle with low 87Sr/86Sr, high 143Nd/144Nd and moderate 206Pb/204Pb ratios combined with negative 7/4 (e.g. Canary Islands and Ahaggar) or high 206Pb/204Pb ratios combined with positive 7/4 (e.g. St. Helena) (Thirlwall, 1997). Below we explore the role of the continental crust, the subcontinental lithospheric mantle, sub-lithospheric mantle reservoirs and the possible depths of melting in the origin of Si–K-rich and Si-poor group magmas from the southeastern Iberian and northwestern African continental margins.
The Si–K-rich group from the southern Iberian margin (8·2–4·8 Ma)
The Si–K-rich volcanic rocks (lamproites) from the southeastern Iberian continental margin form a compositional continuum with cordierite–garnet-bearing lavas from southern Spain, which are thought to result from crustal anatexis of a metapelitic protolith (Fig. 4a–d) (Zeck, 1970; Munksgaard, 1984; Cesare et al., 1997; Cesare & Maineri, 1999). The geochemical similarity of the Spanish low-MgO lamproites (some of which contain cordierite and garnet, e.g. the lamproitic rocks from Mazarron based on petrographic observations) and the cordierite–garnet-bearing lavas from Cerro Hoyazo suggests that the cordierite–garnet-bearing igneous rocks themselves may be highly crustally contaminated derivatives of mafic lamproitic melts rather than pure crustal anatexites. This is consistent with the results of geochemical modelling by Benito et al. (1999), who proposed a mantle-derived origin for the parental melts of the Cerro Hoyazo lavas. However, as inferred from the previously mentioned compositional continuum between high-MgO lamproites, low-MgO lamproites and cordierite–garnet-bearing lavas, the Si–K-rich, lamproitic lavas with MgO below 6–8 wt % appear to have apparently assimilated significant amounts of continental crust. Although the high-MgO lamproites (MgO >8 wt %) seem to be unaffected by crustal contamination, they have isotopic compositions very similar to crustal materials such as local continental crust and marine sediments (Figs 5a and 7a–d). High-MgO lamproites also show extreme enrichments