Introduction

Alkaline lamprophyres are considered to be the equivalent of basaltic rocks with high volatile contents, being related to partial melting of a hydrated mantle enriched in incompatible elements (Rock 1991). The incorporation of volatiles is thought to be due to metasomatism by infiltrating fluids or silicate melts ascending from lithospheric or sub-lithospheric sources (e.g. Hawkesworth et al. 1990).

Mafic alkaline lamprophyres were intruded during the Upper Permian in the Spanish Central System (SCS). Other moderately alkaline rocks from the Iberian Peninsula are broadly coeval with these lamprophyres (mafic dykes from Pyrenees; Debon and Zimmermann 1993), or are very similar in age (North Portugal; Portugal-Ferreira and Macedo 1977), though their exact geochronology has yet to be constrained. In any case, this intraplate magmatism represents a small volume of intruded magma. Previous studies, which focused on the Permo-Carboniferous magmatism from central Spain (Perini et al. 2004) and on the Hercynian and post-Hercynian basic rocks of the SCS (Bea et al. 1999), have advocated a lithospheric origin for the mantle sources of the SCS alkaline lamprophyres. Other studies have revealed that part of the SCS Permian alkaline basic intrusions show an isotopically depleted component, indicative of greater heterogeneity than previously suggested and the involvement of sub-lithospheric sources (Villaseca et al. 2004; Orejana et al. 2005).

The geodynamic setting proposed for these dyke swarms is also the subject of debate. Some authors link their generation to manifestations of Permo-Carboniferous magmatism related to the impingement of a mantle plume prior to the opening of the Atlantic Ocean (Doblas et al. 1998; Perini et al. 2004), whereas others consider that rifting in this region could result from lithosphere thinning and adiabatic upwelling of the hot asphenosphere (Orejana et al. 2006).

In this study we present new data, including major and trace element mineral analyses, bulk-rock composition and isotope ratios (Sr–Nd–Pb), which allow us to elucidate the petrogenesis of these alkaline dykes. The new sample set covers the whole outcropping region of every dyke swarm, and considerably enlarging the previously available analytical dataset. Particular emphasis is placed on the heterogeneity of the mantle sources, the possible role of crustal components and on the nature of any metasomatising agents. Furthermore, we also discuss the significance of the geochemical differences that exist in the widespread Permian basic magmatism developed at the end of the Hercynian orogeny in north-western and south-western Europe.

Geological setting

The SCS is located within the Central-Iberian Zone of the Iberian Massif (Fig. 1). This is a large batholith that consists of several granitic bodies emplaced into Palaeozoic to Neoproterozoic orthogneisses and metasediments. The first manifestation of this granite magmatism has been dated at 346 ± 63 Ma (Villaseca et al. 1998a), approximately 50 Ma after the beginning of the Hercynian collision (408–387 Ma; Gutiérrez Marco et al. 1990). In volume terms the most important plutons were emplaced in the range 323–284 Ma (Villaseca et al. 1998b; Bea et al. 1999; Zeck et al. 2007). The SCS Hercynian granitoids are mainly peraluminous felsic varieties displaying a narrow compositional range, and which do not show any marked compositional trend with time (Villaseca and Herreros 2000).

Fig. 1
figure 1

Schematic map showing the Permian alkaline lamprophyres and diabases from the SCS, together with other post-collisional magmatic intrusions of calc-alkaline and tholeiitic affinity. The names of the nine dyke swarms sampled in this study are indicated in italics

Full size image

Hercynian basic intrusions in the SCS are volumetrically small and occur as small gabbroic to quartz-dioritic masses. Ages range from 322 ± 5 Ma (Rb–Sr whole-rock isochron; Casillas et al. 1991) to 306 ± 2 Ma (SIMS U–Pb dating of zircon; Zeck et al. 2007). Based on their geochemistry these rocks have been linked to crustal recycling in the mantle or melting of a subduction-modified mantle source (Villaseca et al. 2004). In addition to these minor basic intrusives, the SCS is cross-cut by a number of different post-collisional dyke swarms (Fig. 1): (1) E-W oriented calc-alkaline microdiorites (290 ± 10 Ma, Rb–Sr isochron; Galindo et al. 1994), (2) shoshonitic microgabbros, and (3) N-S oriented alkaline dykes. The first two sets are associated with coeval granite porphyries. These three dyke swarm sets have been classified by Villaseca et al. (2004) into the following groupings, Gb2, Gb3 and Gb4, respectively. The Gb1 group was assigned to the previously mentioned Hercynian gabbroic to quartz-dioritic masses. The last magmatic event recorded in the SCS is represented by the intrusion of the large gabbroic Messejana–Plasencia tholeiitic dyke (named Gb5 by Villaseca et al. 2004), dated at 203 ± 2 Ma (Ar–Ar in biotite; Dunn et al. 1998), which is linked to the opening of the Atlantic Ocean.

The alkaline suite (Gb4 group) may be further subdivided into: (1) basic to ultrabasic lamprophyres and diabases, and (2) monzogabbroic to syenitic porphyries. The most recent geochronological data obtained for these rocks give Upper Permian intrusive ages between 264 ± 1 Ma (Ar–Ar on amphibole from lamprophyre dykes; Perini et al. 2004; Scarrow et al. 2006) and 252 ± 3 Ma (U–Pb on zircon from a syenitic porphyry; Fernández Suárez et al. 2006). The lamprophyres carry xenocrysts and megacrysts (mainly clinopyroxene, amphibole and plagioclase), and also xenoliths that are relatively abundant in some outcrops. The latter can be broadly subdivided into granulite and ultramafic xenoliths. Most of the SCS granulites have been interpreted as lower crustal restites formed as the result of extraction of the granite melts which led to the formation of the SCS batholith (Villaseca et al. 1999). Mafic to ultramafic xenoliths, on the contrary, are deep pyroxenitic cumulates that crystallised directly from Permian alkaline basic magmas or melts related to Hercynian calc-alkaline mafic underplating events (Orejana et al. 2006).

Analytical methods

The major element composition of minerals from SCS alkaline lamprophyres and diabases were determined at the Centro de Microscopía Electrónica “Luis Bru” (Complutentse University of Madrid) using a Jeol JZA-8900 M electron microprobe with four wavelength dispersive spectrometers. Accelerating voltage was 15 kV and the electron beam current 20 nA, with a beam diameter of 5 μm. Elements were counted for 10 s on the peak and 5 s on each of two background positions. Sillimanite, albite, almandine, kaersutite, microcline, ilmenite, fluorapatite, scapolite, Ni alloy, cromite, gahnite, bentonite and strontianite mineral standards were employed. Corrections were made using the ZAF method.

Trace element compositions (REE, Ba, Rb, Pb, Th, U, Nb, Ta, Sr, Zr, Hf, Y, Cr, Ni, V and Sc) were determined in situ on >130 μm thick polished sections by laser ablation (LA-ICP-MS) at the University of Bristol using a VG Elemental PlasmaQuad 3 ICP-MS coupled to a VG LaserProbe II (266 nm frequency-quadrupled Nd-YAG laser). Each analysis consisted of 100 s of counting time (including 40 s of background measurement), using a laser beam with a diameter of around 20 μm. NIST 610 glass was used for instrument calibration, and NIST 612 was used as a secondary standard (results are shown in eTable 1). Each analysis was normalised to Ca (clinopyroxene, amphibole, apatite) or Si (feldspars, phlogopite), using concentrations determined by electron microprobe.

Fourteen new whole rock samples were analysed at the CNRS-CRPG Nancy for whole rock major and trace elements, adding to the existing seven dyke analyses (Villaseca et al. 2004; Orejana et al. 2006). The samples were fused using LiBO2 and dissolved with HNO3. Solutions were analysed by inductively coupled plasma atomic emission spectrometry (ICP-AES) for major elements, whilst trace elements have been determined by ICP mass spectrometry (ICP-MS). Uncertainties in major elements range from 1 to 3%, excepting MnO (5–10%) and P2O5 (>10%). Carignan et al. (2001) have evaluated the precision of Nancy ICP-MS analyses at low concentration levels from repeated analysis of the international standards BR, DR-N, UB-N, AN-G and GH. The precision for Rb, Sr, Zr, Y, V, Ga, Hf and most of the REE is in the range 1–5%, whereas they range from 5 to 10% for the rest of trace elements, including Tm. Analyses of BR (an independent standard) are shown in eTable 1.

Pb isotope ratios have been determined on seven lamprophyres and diabases and two hornblenditic xenoliths (previously analysed for Sr–Nd isotope ratios; Villaseca et al. 2004; Orejana et al. 2006) at the Laboratory of Isotope Geology of the Natural History Museum of Stockholm. These samples were dissolved using a 10:1 mixture of HF and HNO3 according to routine procedures of the laboratory. Pb was separated from the samples using element specific ion-exchange columns. The samples were spiked with a 205Pb tracer solution of known concentration to determine the lead concentrations. Isotope ratios were measured using a Finnigan MAT 261 multicollector TIMS in static mode. Empirical measurements of mass fractionation as a function of temperature were established using the international Pb isotope standards NBS 981 and NBS 982. Data from unknown samples, when corrected in an analogous way, are reproducible and accurate within 0.1% (Table 4) and analyses of BCR-1 gave results (206Pb/204Pb = 18.79, 207Pb/204Pb = 15.63, 208Pb/204Pb = 38.74) in agreement with literature values (206Pb/204Pb = 18.81, 207Pb/204Pb = 15.62, 208Pb/204Pb = 38.70).

We are also report Sr–Nd isotopic data for two new lamprophyre samples. These were determined at the Southampton Oceanography Centre, using a VG Sector 54 multi-collector TIMS with data acquired in multidynamic mode. Isotopic ratios of Sr and Nd were measured on a sub-sample of whole rock powder. Repeated analysis of NBS 987 gave 87Sr/86Sr = 0.710245 ± 06 (2σ) and the 143Nd/144Nd ratio of the JM Nd standard was 0.511856 ± 05 (2σ).

Petrography

The SCS alkaline lamprophyres and diabases display a heterogeneous porphyritic texture with abundant mafic phenocrysts (clinopyroxene, kaersutite, olivine pseudomorphs, phlogopite, ulvospinel). Nevertheless, the relative abundance and distribution of these minerals varies significantly from one dyke to another. Following the criteria of Le Maitre et al. (2002) we can classify the SCS lamprophyric dykes (leaving aside diabases) as camptonites, due to the abundance of clinopyroxene, amphibole and phlogopite phenocrysts, absence of felsic phenocrysts and predominance of plagioclase over alkali feldspar within the groundmass. No glass has been found.

Mafic dykes also carry deep-crystallised megacrysts (clinopyroxene, amphibole and plagioclase), which are more abundant in the diabase dykes, and xenoliths of distinctive provenance (wall-rock, lower crustal granulites and pyroxenite xenoliths from the upper mantle-lower crust boundary) (Villaseca et al. 1999; Orejana et al. 2006).

Phenocrysts do not usually exceed 3 mm and vary in their total mode up to 40% in sample 103811. Clinopyroxene and kaersutite are normally zoned, the former sometimes showing multiple rims and oscillatory zoning (Orejana et al. 2007). They are euhedral to subhedral, but may exhibit spongy inner zones infiltrated by groundmass. Fresh olivine has only been observed in one diabase dyke, whereas olivine pseudomorphs, which are mainly transformed to talc group minerals, are common in most lamprophyres and diabases. Plagioclase phenocrysts are only present in diabase dykes, which also occasionally contain mafic phenocrysts. In a analogous fashion to plagioclase, phlogopite phenocrysts are not present in all dykes, but are restricted to some K-rich lamprophyres. Two distinctive spinel phenocrysts are found: black ulvospinel, which transits to Ti-magnetite, and Cr-rich brown spinel microphenocrysts. Other phenocrysts found in variable quantities are ilmenite, apatite and Fe-sulphides.

The groundmass is typically holocrystalline, with a fine-grained inequigranular texture (grain sizes ranging from 10 to 800 μm), although trachytic texture is found in some samples. The main minerals found in the groundmass are clinopyroxene, kaersutite, biotite–phlogopite, Ti–magnetite–ulvospinel, olivine (normally pseudomorphosed), plagioclase and alkali feldspar; with lesser amounts of apatite, analcite, calcite, Fe–sulphides, barite, ilmenite and monazite. These crystals are euhedral to subhedral, with the exception of alkali feldspar, calcite and analcite, which are usually interstitial.

These alkaline rocks show different ocelli and vesicles types: (1) feldspatic, (2) carbonatic and (3) chlorite-rich (with acicular clinopyroxene). Feldspatic ocelli (syenitic) are irregular or semirounded, consisting of kaersutite, biotite and alkali feldspar laths, occasionally showing an inner calcite/analcite globule. Carbonatic ocelli are similar to these calcite globules, with minor amounts of feldspar and chlorite. Chlorite-rich vesicles are outlined by small clinopyroxene crystals that occur as prismatic laths. The inner zone of these ocelli shows chlorite, calcite, albite and minor barite. This kind of ocelli are typical of alkaline lamprophyres and are interpreted as volatile exsolution at shallow emplacement levels (Rock 1991).

Most of the SCS alkaline dykes are remarkably fresh. Alteration of the samples is mainly restricted to pseudomorphed olivine (by talc), although this secondary process is likely to be due to autometasomatism during volatile exsolution. Thus, it is likely that a selective low-P late-magmatic alteration of suspended solids (xenoliths and phenocrysts) occurred during lamprophyric magma devolatilization (Orejana and Villaseca 2008).

Mineral chemistry

Clinopyroxene

The major element composition of representative clinopyroxenes from SCS alkaline basic dykes is summarized in Table 1 (see the whole data on eTable 2). They are Ti-diopsides or Ti-augites, following the criteria of Morimoto et al. (1988). Matrix clinopyroxene is similar in major element composition to phenocrysts, but display lower Mg# values (0.60–0.81) (Fig. 2). Clinopyroxene yields a heterogeneous composition characterized by high TiO2 and Al2O3 concentrations (up to 5.7 and 11.4 wt%, respectively). Na2O content ranges from 0.33 to 1.10 wt% and Cr2O3 may reach up to 0.9 wt%. Normal zoning is relatively common in the phenocrysts, and it gives trends of increasing Fe, Al, Ti and Ca concentrations and decreasing Mg#, Si and Cr from core to rim (Orejana et al. 2007). Clinopyroxene phenocrysts exhibit high concentrations of most trace elements (eTable 3): REE (48–120 ppm), LILE (Rb = 0.6–5.9 ppm; Ba = 1.5–3.6 ppm) and HFSE (Th = 0.1–0.4 ppm; Ta = 0.1–0.7 ppm; Nb = 0.5–1.6 ppm; Zr = 43–153 ppm). They display convex upward chondrite-normalised REE patterns (Fig. 2), similarly to clinopyroxenes crystallized at depth from basaltic melts (Irving and Frey 1984). Negative Ba, K, Nb, Pb and Sr anomalies may be found in their primitive mantle-normalised trace element patterns.

Table 1 Representative major element composition of clinopyroxene and amphibole from SCS alkaline lamprophyres and diabases
Full size table
Fig. 2
figure 2

Major and trace element composition of clinopyroxene from the SCS alkaline basic dykes. The normalising values of chondrite and primitive mantle are after Sun and McDonough (1989) and McDonough and Sun (1995), respectively

Full size image

Amphibole

The major element composition of representative amphiboles from SCS alkaline basic dykes is summarized in Table 1 (see the whole data on eTable 4). They are classified as kaersutites according to Leake et al. (1997). These amphiboles are Al–Ti-rich, with Al2O3 up to 14.8 wt% and TiO2 up to 7.7 wt% (Fig. 3). Phenocrysts show a homogeneous composition, with Mg# in the range 0.61–0.72. Matrix amphiboles overlap phenocrysts composition, but generally have more evolved compositions resembling the major element composition of amphibole from feldspatic ocelli (Fig. 3). Amphibole trace element compositions (eTable 3) mimic those of coexisting clinopyroxene phenocrysts, with characteristic convex upward chondrite-normalised REE patterns (Fig. 3). Nevertheless they show a slight enrichment in most trace elements with respect to clinopyroxene. They have high Ba (537–826 ppm), Nb (40–126 ppm), Ta (2.3–5.6 ppm) and Sr (>1,150 ppm) concentrations, this is reflected in positive anomalies for these elements in primitive mantle-normalised patterns.

Fig. 3
figure 3

Major and trace element composition of amphibole from the SCS alkaline basic dykes. The normalising values of chondrite and primitive mantle are after Sun and McDonough (1989) and McDonough and Sun (1995), respectively

Full size image

Olivine

Fresh olivine is only present in a scarce group of diabases (both as phenocrysts and within the groundmass). Nevertheless, the abundance of pseudomorphs in most SCS alkaline dykes highlights its potential importance during crystal fractionation. Forsterite content varies from Fo77 to Fo90 in phenocrysts (including slightly Fe-rich rims) (eTable 5). These values match the Mg/Fe ratios of clinopyroxene phenocrysts (0.67–0.88), indicating equilibrium crystallization. Olivine Mg# correlates positively with NiO content (0.05–0.28 wt%) (eFig. 1). Matrix olivine compositions overlaps those of phenocrysts, but they extend towards lower Mg (Fo75–Fo83) and NiO (0.05–0.13 wt%) concentrations.

Micas

They constitute a heterogeneous group ranging from Ti-biotite to Ti-phlogopite, whose main characteristic is high TiO2 (up to 10.5 wt%) and Al2O3 (up to 17.8 wt%) (eFig. 1; Table 2 and eTable 6). Similar to amphibole, mica phenocrysts (which are mainly phlogopites) display high Mg# values (0.65–0.78) and major element concentrations similar to those of groundmass micas, with the exception of Mg#, which may be as low as 0.28 in the matrix, resembling composition of biotites from feldspatic ocelli (eFig. 1). They may also have high F concentrations (up to 3.4%). Phlogopite phenocrysts exhibit high trace element concentrations (eTable 3), notably Ba (3,000–6,000 ppm), Rb (280–360 ppm), Nb (15–50 ppm) and Ta (0.9–2.2 ppm). These compositions are very similar to those of phlogopite phenocrysts from other alkaline lamprophyres (Foley et al. 1996) (eFig. 1).

Table 2 Representative major element composition of micas and olivine from SCS alkaline lamprophyres and diabases
Full size table

Feldspars

The major element composition of representative feldspars (plagioclases and alkali feldspar) in SCS alkaline dykes is summarized in eTable 7. Phenocrysts are exclusively plagioclases (from andesine to labradorite, An32–An70 (eFig. 2)) and are restricted to diabase dykes. The compositional field of groundmass feldspars is wider, with plagioclases overlapping the phenocrysts range, but plotting towards more albite and orthoclase-rich compositions, and alkali feldspar falling within the sanidine compositional field (Or54–Or100). Plagioclase phenocrysts are usually zoned with core compositions of An61–70 and rims of An49–62 (eFig. 2).

Other minerals

Spinel compositions are shown in eTable 5. Phenocrysts are Ti–magnetite–ulvospinel and Cr-spinel. The first group is more abundant and displays a relatively heterogeneous composition with FeO + Fe2O3 = 52–76 wt% and TiO2 in the range 12–36 wt%, whilst Al2O3 and MgO do not exceed 9 and 3 wt%, respectively (eFig. 3). Cr-spinels are euhedral and always overgrown by ulvospinel rims. Their Cr2O3 and Al2O3 concentrations range from 10 to 38 wt% and from 28 to 52 wt%, respectively. A negative correlation exists between Al and Cr (Mg) contents. Groundmass spinels are restricted to Ti–magnetite–ulvospinel series and overlap the compositional field of ulvospinel phenocrysts, although they tend towards slightly Fe-enriched compositions (eFig. 3).

Apatite may appear as phenocrysts or within the matrix. This phase is always F-rich (1.4–2.7 wt%) (eTable 5) and is classified as fluorapatite. Cl does not exceed 0.5 wt% and SrO ranges from 0.01 to 0.35 wt%. Ilmenite is restricted to the San Bartolomé de Pinares dyke and it is characterized by high MgO (0.02–10 wt%) and MnO (0.5–10 wt%) (eTable 5). Mg–Mn-rich ilmenites have been proposed by Rock (1991) as a diagnostic feature of alkaline lamprophyres.

Whole-rock geochemistry

Major and trace elements

The major and trace element composition of the SCS alkaline dykes is shown in Table 3. Although all samples have high LOI numbers (mainly 2.9–5.7 wt%), due to the abundance of hydrous minerals (amphibole and phlogopite), those displaying the highest values correspond with ocelli-rich varieties.

Table 3 Major and trace element composition of SCS alkaline lamprophyres and diabases
Full size table

Due to their silica and alkalis contents (ranging from 41.2 to 47.4 wt% and from 3.0 to 7.4 wt%, respectively; Table 3) the SCS lamprophyres and diabases plot within the alkaline area in the TAS diagram (taken from Le Bas et al. 1986), mainly within the basanite compositional field. The K2O/Na2O ratio of the lamprophyres (a parameter used to differentiate potassic from sodic rocks) is higher than unity (Fig. 4) (Le Bas et al. 1986; Le Maitre et al. 2002), whereas the diabases are sodic. Nevertheless, two lamprophyre samples which yield K2O/Na2O > 2 could be included within an ultrapotassic group, following the criteria of Foley et al. (1987) and Le Maitre et al. (2002) (Fig. 4). Apart from being high in K, they are characterized by high TiO2 (1.4–4.1 wt%) and P2O5 (0.28–1.0 wt%) concentrations and moderate to low Mg# (0.43–0.7) (Table 3).

Fig. 4
figure 4

K2O versus Na2O (wt%) contents of the SCS alkaline basic dykes. The line for K2O/Na2O > 2, which separates potassic from ultrapotassic rocks has been constructed with the criteria of Foley et al. (1987)

Full size image

These alkaline dykes are markedly heterogeneous in their major and trace element composition (e.g. MgO = 4.5–12.2 wt%; Al2O3 = 10.3–16.8 wt%; CaO = 5–13 wt%; Rb = 46–145 ppm; Zr = 112–367 ppm; Y = 22–66 ppm) (Table 3). The diabases can be differentiated from lamprophyres due to their lower K–Ti–Ta–Zr–Hf–V and higher Mn–Cs contents (Fig. 5). When considered separately, samples from a single dyke usually show compositional variations characterized by an increase in Si, Al, Na, K and most incompatible trace elements, and a decrease of Ti, Ca, Cr, Ni and V towards lower Mg# (Fig. 5).

Fig. 5
figure 5

Major and trace element composition of the SCS alkaline basic dykes. Concentrations expressed as weight % (SiO2, Al2O3, CaO, TiO2 and K2O) and ppm (V, Cr, Rb, Sr, Nb, Zr and Ce). The black arrows indicate the compositional variation trend observed within samples from a single dyke. The solid grey line displays the output data of a fractional crystallization model applied on major and trace elements and made on the basis of the general Rayleight equation. The model input data consider fractionation of 20% cpx, 7% ol and 13% amph, obtained by averaging phenocrysts modal proportions. Clinopyroxene 103811-62, amphibole 80318-9 and olivine 104867-57 from Tables 1 and 2 have been used in major element modeling. As initial magma composition we have considered the most Mg-rich non cumulate samples from La Paramera dyke (76547 and 77753). Mineral/melt partition coefficients for clinopyroxene, olivine and amphibole used for modeling trace elements (except Cr, V) taken, respectively, from Foley et al. (1996), Zanetti et al. (2004) (except Nb after Taura et al. 1998) and LaTourrette et al. (1995). D Cpx/meltCr , D Cpx/meltV and D Ol/meltCr after Ringwood (1970); D Amph/meltV and D Ol/meltV after Dostal et al. (1983) and D Amph/meltCr after Matsui et al. (1977)

Full size image

The dykes are also characteristically enriched in incompatible trace elements. The REE are fractionated showing steep chondrite-normalised patterns with highly enriched LREE and depleted HREE (Fig. 6). The chondrite-normalised values are between 100 and 400 with respect to the La content, whereas Yb and Lu values are predominantly <20. This is reflected in their high (La/Yb)N values from 9.6 to 23.3. LILE and HFSE also display high concentrations, giving rise to positive Rb–Ba and Nb–Ta anomalies in primitive mantle-normalised patterns (Fig. 6). It is also worth noting the K peak and the Th–U and Pb troughs. These compositions resemble those of OIB, but show higher Rb, Ba, Th, U, K, Nb and Ta contents.

Fig. 6
figure 6

a Chondrite-normalised and b, c and d primitive mantle-normalised trace element composition of the SCS alkaline basic dykes. Lamprophyres containing phlogopite phenocrysts and diabasic samples have been plotted separately for trace element spidergrams, whereas they all have been plotted together for REE. The normalising values of chondrite and primitive mantle are after Sun and McDonough (1989) and McDonough and Sun (1995), respectively

Full size image

Sr–Nd–Pb isotope ratios

All isotopic data for SCS alkaline lamprophyres and diabases are corrected to 265 Ma (Table 4), taking into consideration the most recent geochronological results. Figure 7 shows both the new Sr–Nd analyses presented in this work and those recently obtained by our research group (Villaseca et al. 2004; Orejana et al. 2006). The data overlap the wide compositional field of OIB, falling within the mantle array. The most radiogenic compositions partially coincide with previous data from Bea et al. (1999) and Perini et al. (2004), confirming the marked heterogeneous composition of these Permian intrusive rocks, as previously suggested by Villaseca et al. (2004) and Orejana et al. (2006). This heterogeneity allows us to differentiate two groups: (1) one isotopically depleted, plotting close to MORB compositions and with (87Sr/86Sr)0 in the range 0.7023–0.7038 and εNd0 in the range +4.0 to +7.1, and (2) an isotopically enriched group, resembling bulk silicate earth (BSE) isotopic values, with (87Sr/86Sr)0 ranging from 0.7044 to 0.7052 and εNd0 from −0.9 to +1.4 (Fig. 7). All diabases and an ultrapotassic lamprophyre from the Sierra de la Paramera dyke plot within the depleted group. The isotopically enriched group corresponds to camptonitic dykes. There is a close resemblance between the Sr–Nd isotopic composition of the SCS alkaline basic dykes and that of the most depleted alkaline rocks from western Europe Permo-Carboniferous magmatism (Fig. 7), this is discussed below.

Table 4 Sr–Nd–Pb Isotopic composition of SCS alkaline lamprophyres and diabases
Full size table
Fig. 7
figure 7

Sr–Nd isotopic composition of the SCS alkaline basic dykes compared with that of other alkaline or moderately alkaline Permian basic rocks from western Europe, including previous analyses of Bea et al. (1999) and Perini et al. (2004) for the SCS alkaline lamprophyres. Maragato and Puerto del Pico samples represent new data, whilst the rest of plotted samples are taken from Villaseca et al. (2004) and Orejana et al. (2006). Compositional field of Pyrenees, Oslo Graben, Scotland, Corsica and Western Alps are taken from Lago et al. (2004), Neumann et al. (2004), Upton et al. (2004), Bonin (2004) and Monjoie (2004), respectively. It has been also plotted for comparison the composition of the Gb1 and Gb2–Gb3 calc-alkaline post-collisional dykes from the SCS, after Villaseca et al. (2004). The isotopic signatures of the SCS lower crustal felsic granulite xenoliths and metapelitic xenoliths are taken from Villaseca et al. (1999). SCS upper crustal materials (grey vertical bar) plot mostly out of the diagram, towards much higher 87Sr/86Sr values (Villaseca et al. 1998b). OIB and MORB fields after Wilson (1989)

Full size image

The Pb isotope analyses display more restricted compositional variation with (206Pb/204Pb)0 = 18.16–18.51, (207Pb/204Pb)0 = 15.54–15.64 and (208Pb/204Pb)0 = 37.93–38.44 (Table 4). The only previous Pb isotope data available on the SCS alkaline dykes is that of Perini et al. (2004), these rocks belong to the Sr–Nd isotopically enriched group of lamprophyres (Fig. 8). Overall, the Pb isotope compositions are close to PREMA (Prevalent Mantle) and BSE (mainly when considering 206Pb/204Pb), although they display some tendency towards being slightly 207Pb and 208Pb-enriched. Nevertheless, the Pb radiogenic ratios indicate derivation from a relatively depleted source. When considered in detail, the same division made on the basis of Sr–Nd isotope ratios can also be made on the basis of Pb isotope ratios; the Sr–Nd isotopically enriched lamprophyres clearly display the most radiogenic Pb values within the narrow compositional range of SCS alkaline dykes (Fig. 8). Absence of highly radiogenic Pb data implies that the PREMA component (and not HIMU) is dominant in the isotopically depleted SCS alkaline group (Orejana et al. 2005). A similar composition is observed in the Permian alkaline basic dykes from the western Alps (Fig. 8).

Fig. 8
figure 8

Pb isotope ratios of the SCS alkaline basic dykes. The single data of Perini et al. (2004) corresponds with a single SCS alkaline lamprophyre. It has also been plotted for comparison the compositional field of the Permian moderately alkaline dykes from western Alps (Monjoie 2004). Compositional fields of OIB, DMM, BSE, EMI, EMII, PREMA and HIMU are taken from Zindler and Hart (1986). NHRL: Northern Hemisphere Reference Line

Full size image

The Sr–Nd–Pb isotopic composition of the SCS alkaline magmatism represents the irruption of a new mantle-derived component in the Central-Iberian Zone. The PREMA-like alkaline dykes contrast with the previous calc-alkaline Hercynian basic magmatic suites (Gb1, Gb2 and Gb3; Villaseca et al. 2004) that are clearly lithosphere-derived (87Sr/86Sr0 ranging from 0.7045 to 0.7087 and εNd0 from −3.9 to +0.6; Fig. 7). This new component is not recorded previously in the whole Hercynian Iberian Massif, suggesting a significant geodynamic change during post-Hercynian times, this is discussed in more detail below.

Petrogenesis

Magma differentiation

One of the main features of the SCS Permian alkaline magmatism is its chemical and petrological heterogeneity, giving rise to the lamprophyre-diabase division and variable phenocryst modal composition. These characteristics are also manifested in the absence of a unique variation trend for the group of dykes as a whole for many major and trace elements (e.g. Ti, K, V, Rb, Zr), although there are both negative and positive correlations between these elements and Mg# when considering the samples of each dyke separately (Fig. 5). This heterogeneity could be explained by the intrusion of several magma batches derived from slightly different mantle sources, or it might be caused by variable amounts of melting. We regard the Sr–Nd isotopic differences to be due to a significant mantle compositional effect. This is discussed below.

Spanish Central System alkaline rocks do not show primary magma compositions according to the criteria of Frey et al. (1978): Mg# > 0.6–0.7; Cr > 500–1,000 ppm and Ni > 200–500 ppm (Table 3). Moreover, the forsterite content of olivines from primary magmas are normally in the range 88–94, whilst olivines from SCS lamprophyres and diabases do not exceed Fo86, suggesting the involvement of one or more differentiation processes. Of particular interest is the alkaline dyke from Maragato, which shows the richest composition in Mg, Cr and Ni (Fig. 5). These contents are probably influenced by the cumulative nature of this dyke, as deduced from its high proportion (>40%) of mafic phenocrysts (mainly clinopyroxene and pseudomorphosed olivine).

The positive correlation of Ca, Ni, Cr and V versus Mg# value points to ol + Cr-sp + cpx crystal fractionation. Additionally, the trends of Ti depletion observed in Fig. 5 are likely to be due to kaersutite fractionation. These premises are consistent with the petrographic observations. We have applied a crystal fractionation model based on major and trace elements for the lamprophyric melts; the results are plotted in Fig. 5. A 40% fractionation of cpx + amph + ol is enough to account for the compositional trends observed in the SCS lamprophyric magmas (more details in caption to Fig. 5). This model should be considered as a general indication of the fractionation process followed by the different alkaline dyke swarms (e.g. for diabasic magmas a different crystal fractionation process involving plagioclase should be considered).

Role of crustal assimilation

The entrapment of granulite and cumulate ultramafic xenoliths whose estimated pressure of equilibration ranges from 7 to 12 kbar (Villaseca et al. 1999; Orejana et al. 2006) seems to indicate that the SCS lamprophyric and diabasic melts might have stagnated at lower crustal levels. This is reinforced by the fact that some of those ultramafic cumulate xenoliths are genetically related with the Permian alkaline magmatism (Orejana et al. 2006), and also because some lamprophyres carry a complex high-pressure phenocryst population (Orejana et al. 2007). Thus, the question of a possible lower crustal contamination process arises. Ratios of trace elements with contrasting concentrations in mantle and crustal rocks may help to evaluate this possible influence. Ce/Pb and Ba/Nb ratios in the continental crust are around 3.9 and 57, respectively (Rudnick and Gao 2003) due to the high Pb and Ba concentrations in the crust. These values are close to those of the SCS granulite xenoliths from the lower crust (Ce/Pb ∼5 and Ba/Nb ∼66; Villaseca et al. 1999). The same ratios in the SCS alkaline lamprophyres and diabases are significantly different (Ce/Pb = 14–28 and Ba/Nb = 9–15), approaching the trace element composition of typical mantle-derived rocks (OIB: Ce/Pb ∼25 and Ba/Nb ∼7.3; Sun and McDonough 1989) (Fig. 9a). Three SCS lamprophyres show Ba/Nb from 15 to 29, although these high values are probably derived from the high proportions of phlogopite (Ba-rich) and clinopyroxene (Nb-poor) phenocrysts.

Fig. 9
figure 9

Whole-rock chemical composition of SCS alkaline basic dykes for a Ce/Pb versus Ba/Nb, b SiO2 versus Rb/Sr and c SiO2 versus (87Sr/86Sr.)0. Plot (a) also shows an AFC model made for three contrasting assimilation/fractionation ratios (0.2, 0.5 and 0.8). Initial model melt composition: sample 77753 (La Paramera dyke). Contaminant composition: averaged values of SCS lower crustal granulite xenoliths (Villaseca et al. 1999). The bulk DNb (0.06), DBa (0.05), DCe (0.07) and DPb (0.03) have been calculated using the mineral/melt partition coefficients and the proportions of fractionating phases described in Fig. 5 caption, with the exception of DPb, for which ol/melt and amph/melt partition coefficients have been taken from McKenzie and O’Nions (1991). MORB and OIB average composition in diagram (a) are taken from Sun and McDonough (1989)

Full size image

We have tested an AFC model using Ce/Pb and Ba/Nb ratios to quantify the degree of assimilation of lower crustal materials on the composition of the SCS alkaline dykes (see input data in Fig. 9 caption). The results shown in Fig. 9a are based on three different assimilation/fractionation (r) values. Leaving aside the three samples with high Ba/Nb discussed above, the small compositional differences displayed by the SCS alkaline dykes would require no more than 7% assimilation of lower crustal rocks. Moreover, there is not a positive correlation of silica content with Sr isotope radiogenic ratios or Rb/Sr values in these alkaline rocks (Fig. 9b,c), as would be expected if a silica-rich crustal component were involved in their genesis. Additionally, the negative Pb anomalies and Nb–Ta peaks shown by the SCS alkaline dykes when normalized to primitive mantle contents (Fig. 6), support the contention that contamination by lower crustal rocks did not exert a significant influence on their composition.

Source enrichment and mantle heterogeneity

The primitive mantle-normalised Sm/Yb ratios in the SCS alkaline basic rocks range from 2.4 to 5.3, and these values are consistent with melting in the presence of residual garnet (e.g. McKenzie and O’Nions 1991). Furthermore, their high incompatible trace element contents probably indicate the involvement of an enrichment process in the mantle source.

Th/Yb and Ta/Yb ratios in basic rocks are a useful petrogenetic indicator. In Fig. 10a we have represented the compositional fields of mantle-derived rocks from depleted and enriched sources (following the mantle array) and those of island arc basalts (IAB) and active continental margins (taken after Wilson 1989). The involvement of crustal rocks or subduction components in IAB and continental margins is reflected in their Th enrichment and Ta depletion with respect to REE. The SCS alkaline basic dykes plot within the mantle array and completely overlap the enriched field of mantle rocks, with a composition similar to OIB (Fig. 10a), thus supporting a minor contribution of crustal assimilation and suggesting that primary magmas of these rocks were generated by partial melting of a metasomatised mantle.

Fig. 10
figure 10

Whole-rock chemical composition of SCS alkaline basic dykes for a Th/Yb versus Ta/Yb, b Rb/La versus K/La and c Ba/Nb versus Ba/Ce. Arrows represent the general trend described by the SCS alkaline dykes. MORB and OIB average composition in diagram (a) are taken from Sun and McDonough (1989), and Continental crust average composition taken from Rudnick and Gao (2003). The fields of Oceanic Island Arcs and Active Continental Margins after Wilson (1989)

Full size image

For SCS alkaline lamprophyres there are positive correlations between Rb/La and K/La ratios, and also between Ba/Nb and Ba/Ce ratios (Fig. 10b, c). These features might account for the presence of phlogopite in the source during partial melting as this mineral may preferentially incorporate LILE when compared to REE and HFSE. This is also supported by the potassic character of lamprophyres; potassic and ultrapotassic rocks are associated with the presence of phlogopite in the mantle (e.g. Foley 1992). The high Nb–Ta concentrations observed in the SCS alkaline dykes (Table 3) suggest that pargasitic amphibole has also participated in their genesis controlling the behavior of Nb–Ta during mantle melting (e.g. Ionov et al. 1997). Moreover, the SCS diabases have higher Nb/Ta ratios (13.6–16.1) when compared to the lamprophyres (11.2–13.1). Green (1995) found that pargasite is the only mineral involved in the genesis of basaltic magmas that shows DNb/Ta > 1, and thus the melt Nb/Ta ratio is controlled by amphibole during melting (Tiepolo et al. 2000). Therefore, although both potassic phases might have coexisted, amphibole was probably predominant in the mantle sources of SCS diabases, whilst the influence of phlogopite is evident mainly in the case of lamprophyres. The moderate P2O5 content of the SCS alkaline dykes (0.3–1 wt%; Table 3) indicate that a P-rich phase, such as apatite, might have been a stable metasomatic mineral in the mantle source. Slight negative P anomalies can be observed in these dykes (Fig. 6) indicating the presence of a residual P-rich mineral in the mantle. Samples showing positive P anomalies are those characterized by high proportions of accumulated apatite phenocrysts (e.g. sample 81839).

A diabase-lamprophyre association within the Permo-Carboniferous magmatism in western Europe has also been reported in southern Scandinavia by Obst et al (2004). Nevertheless, these diabases are tholeiitic in composition and they are derived from a depleted mantle source subjected to a metasomatic event that is genetically unrelated to the coeval alkaline lamprophyres. On the contrary, the SCS diabase-lamprophyre association consists of similar alkaline magma batches (e.g. Fig. 6), which also overlap in their Sr–Nd–Pb isotopic compositional fields, suggesting that they are derived from similar mantle sources with only slight differences in their modal accessory mineralogy.

The isotopic composition of the SCS alkaline magmas implies the involvement of at least two different mantle sources: one isotopically depleted (PREMA-like) and the other isotopically enriched. The depleted Sr–Nd isotope signatures of diabases and some lamprophyres indicate participation of a sublithospheric source (probably asthenosphere-related). However, the lamprophyres with more radiogenic Sr–Nd isotopic compositions clearly indicate the involvement of a lithospheric mantle. Metasomatism of both asthenospheric and lithospheric mantle sources is likely to have acted in a similar way, producing the same enrichment patterns; the highly incompatible trace element ratios in both lamprophyres and diabases are not significantly different (e.g. Fig. 10a).

Mantle metasomatism has been attributed to either (1) fluids or melts generated by subduction processes (e.g. Peacock 1990), or (2) volatile- and K-rich, low-viscosity melts that leak from the asthenosphere and accumulate in the overlying lithosphere (e.g. McKenzie 1989). The composition of these agents is believed to change continuously as they percolate thorough the mantle from their source regions (Navon and Stolper 1987). Normalised multi-element plots (Fig. 6) for the SCS alkaline basic dykes do not have subduction-related signatures, such as large negative Nb, Ta and Ti anomalies, and thus we attribute the source enrichment to metasomatising melts derived from the convecting mantle. McKenzie (1989) suggested that these melts would freeze in the lithosphere and might concentrate in thin zones over long periods of time, resulting in substantial volumes accumulated as veins, sills or dykes in a mechanical boundary layer. Melting of this metasomatised mantle might lead to the generation of potassic–ultrapotassic rocks (e.g. Chalapathi Rao et al. 2004). The potassic nature of the SCS alkaline lamprophyres is in accordance with this latter model.

The involvement of a continental component in the SCS lamprophyre mantle sources is difficult to assess. The Sr–Pb isotopic signatures of the enriched lamprophyres are indicative of the introduction of a 87Sr–207Pb–208Pb-rich component into the mantle. These characteristics cannot be explained by the generation of phlogopite in mantle sources during the metasomatic event. This mineral usually displays high Rb/Sr and low U/Pb ratios, thus an enrichment in radiogenic Pb would not be expected. Furthermore, the high Rb-REE concentrations in the isotopically depleted dykes argue against extended evolution of the mantle sources after being metasomatised. We consider that the 207Pb–208Pb-rich composition of these rocks could be derived from a component similar to the reservoir EMII, which could be related to the incorporation of continental crust into the mantle (Zindler and Hart 1986). The subcontinental lithospheric mantle source of these lamprophyres has been slightly modified by subducted crustal components, now only detectable because of their isotopic signatures. But the origin and age of this metasomatic event is difficult to establish with the current data.

The geodynamic context of the SCS alkaline magmatism within the Permian magmatic province of western Europe

The intrusion of the SCS alkaline magmas is considered to be part of the widespread magmatism developed in western Europe at the end of the Hercynian orogeny, with extensive alkaline magmatic manifestations from the northern foreland to the internal zones (Wilson et al. 2004 and references therein). In Fig. 11 we have highlighted the location of the most important Permian basic magmatic regions in western Europe, together with the main structural lineations that were active at that time (see references in Fig. 11 caption). According to different studies, this Permian magmatism coincided with a period of incipient regional rifting (Bonin 1988; Ziegler 1993; Heeremans et al. 2004). Nevertheless, a palaeomagnetic study on Permian volcanic rocks from the western Alps has shown that a single geodynamic setting might not be applicable to the whole area (Muttoni et al. 2003). These authors propose that part of the S-Europe region would have been assembled with Africa (Gondwana) during Early Permian times. This is in agreement with the model of Irving (1977), which suggests that a significant change might have occurred in the palaeogeographic configuration of the Pangea supercontinent from early to late Permian, placing Gondwana farther to the East by approximately 3,000 km with respect to Laurasia at the beginning of this period. This transition from Pangea ‘B’ to Pangea ‘A’ has been associated with an intraplate dextral megashear system and with the reactivation of Hercynian shear zones (Muttoni et al. 2003). The Hercynian front could be related to one of these megashear bands.

Fig. 11
figure 11

Sketchy map made after Franke (1989) showing the location of the main magmatic regions within western Europe during Permian. WIM West Iberian Margin, after Gardien and Paquette (2004); P Portugal, after Portugal-Ferreira and Macedo (1977); PY Pyrenees, after Debon and Zimmermann (1993) and Lago et al. (2004); IC Iberian Chain, after Lago et al. (2005); AM Armorican Massif, after Bellon et al. (1988); CS Corsica-Sardinia, after Bonin (1988); ALP western and southern Alps, after Cortesogno et al. (1998) and Rottura et al. (1998); SCT Scotland, after Upton et al. (2004); NS North Sea, after Heeremans et al. (2004) and Stemmerik et al. (2000); NEGB North East German Basin, after Neumann et al. (2004), OG Oslo Graben, after Neumann et al. (2004) and SC Scania, after Neumann et al. (2004). Structural lineations after Bonin (1988), Ziegler (1993), Heeremans et al. (2004) and Ziegler et al. (2004). Hercynian front after Kirstein et al. (2006)

Full size image

Figure 12 shows the time intervals of Permian basic magmatism as a function of magmatic affinity (calc-alkaline, alkaline and tholeiitic) within western Europe. Calc-alkaline rocks are clearly confined to lower Permian or older ages within SW Europe and do not coexist with alkaline intrusions in the internal zone of the orogen, with the exception of Corsica-Sardinia (Cocherie et al. 2005) (Fig. 12). Additionally, Carboniferous alkaline rocks are confined to NW Europe. In Scotland this magmatism starts in the Dinantian (from 342 ± 1 Ma; Monaghan and Pringle 2004) and persists during Permian times (Upton et al. 2004; Kirstein et al. 2006). Something similar is observed in the Oslo Graben, where alkaline basalts were intruded between 305 and 299 Ma and larvikitic and basanite lavas follow in several stages until 243 Ma (Neumann et al. 2004). Highly alkaline basalts and basanites are characteristic of the lower Permian in Scania (294–274 Ma) and the NE German basin (294–302 Ma) (Neumann et al. 2004). A similar age range has been obtained for basic volcanics (strongly alkaline basalts and basanites) from the North Sea basins (broadly 299–260 Ma; Stemmerik et al. 2000; Neumann et al. 2004). Minor tholeiitic volcanics may accompany the more abundant alkaline rocks in NW Europe Permo-Carboniferous regions. Moreover, a regional tholeiitic magmatic event recorded around 295 Ma characterizes these northern areas (Heeremans et al. 2004), but was not seen in SW Europe (Fig. 12). On the other hand, tholeiitic magmas occur in SW Europe in continental areas at the Triassic–Jurassic boundary, as exemplified by the SCS and Pyrenees intrusions.

Fig. 12
figure 12

Chronology of the main Permian-Triassic basic magmatism in western Europe. PY Pyrenees, after Alibert (1985), Debon and Zimmermann (1993) and Lago et al. (2004); WIM West Iberian Margin, after Gardien and Paquette (2004); IC Iberian Chain, after Lago et al. (2005); AM Armorican Massif, after Bernard-Griffiths et al. (1985) and Bellon et al. (1988); CS Corsica-Sardinia, after Traversa et al. (2003), Bonin (2004) and Cocherie et al. (2005); ALP western and southern Alps, after Rottura et al. (1998), Eichhorn et al. (2000), Carraro and Visonà (2003) and Monjoie (2004); SCT Scotland, after Upton et al. (2004); NS North Sea, after Stemmerik et al. (2000); NEGB North East German Basin, OG Oslo Graben and SC Scania, after Neumann et al. (2004). NW Europe tholeiitic magmatic event at 295 Ma are taken from Heeremans et al. (2004)

Full size image

Accordingly, a geochemical contrast exists when comparing SW and NW Europe basic magmatism during the Permian. In SW Europe tholeiitic and alkaline magmatism is mostly absent during the Lower Permian (excepting Corsica-Sardinia; Fig. 12), whereas calc-alkaline magmatism is dominant and it does not appear in NW Europe. This discrepancy supports the possibility outlined by Muttoni et al. (2003), that NW and SW Europe were geographically farther apart during the Early Permian than previously thought. The widespread alkaline magmatic event recorded in both SW and NW Europe (SCS, PY, CS, ALP, SCT, OG, SC, NS and NEGB) during the Mid Permian might represent the assembly of all western Europe, to configure Pangea ‘A’ within the framework of a common general geodynamic setting.

In the SCS, the intrusion of the alkaline suite during Upper Permian marks the change from the last Hercynian extensional stages to widespread post-Hercynian rifting. This latter global faulting might represent the break up of Pangea, but it has also been linked to the opening of the Atlantic Ocean and the development of the Central Atlantic Magmatic Province (Bonin 1988). This latter possibility could be supported by the alkaline-tholeiitic transition during the Upper Triassic. This mainly occurs near ocean margin regions (the ocean–continent transition zone in the NW of the Iberian Massif and Armorican Massif; Fig. 12).

The marked change in the geochemical composition of the magmatism from Lower to Upper Permian in SW Europe, means the introduction of deeper mantle-derived melts with isotopically depleted components, as evidenced by their positive ε(Nd) values (Corsica, Bonin 2004; western Alps, Monjoie 2004; Pyrenees, Lago et al. 2004, and the SCS) (Fig. 7). Although this new isotopic signature is also recorded in NW Europe (Oslo Graben, Neumann et al. 2004; Scotland, Upton et al. 2004), it appears as early as 342 ± 1 Ma in Scotland (Monaghan and Pringle 2004), indicating a different geological history for this northern sector when compared to southern Europe, reinforcing the geodynamic hypothesis proposed above.

Permian alkaline melts from the SW Europe (e.g. Corsica-Sardinia, SCS area) display magma types with lower ɛNd values than NW Europe alkaline regions, plotting close to BSE (Fig. 7). Thus, an enriched lithospheric mantle component has been involved in the genesis of alkaline magmas in those southern European areas (e.g. Perini et al. 2004). This is in agreement with Pb isotope data, as discussed in the previous section. Nevertheless, Upper Permian alkaline magmatism in western Europe is considered to be mainly derived from mantle sources without the involvement of significant subduction-contaminated components (Wilson et al. 2004). Intracontinental rifting at the end of the Hercynian orogeny within central–western Europe has been suggested, implying lithosphere thinning and ascent of hot deep mantle (Wilson et al. 2004 and references therein). Four possible mechanisms may account for this general process of rifting and associated mantle melting: (1) breakoff of a subducting slab; (2) delamination of the lower crust; (3) impingement of a mantle plume; and (4) passive upwelling of asthenosphere as continental fragments diverge and melting occurs by adiabatic decompression.

Models involving crustal sinking (slab breakoff or lower crustal delamination) are unlikely due to the absence of significant subduction-related components in the SCS lithospheric mantle sources.

The origin of the Permo-Carboniferous volcanic province in Europe and NW Africa has been interpreted in terms of a superplume impinging on the base of the lithosphere (Ernst and Buchan 1997). According to this model, the head of this superplume would have spread northward asymmetrically from the incipient Atlantic rift (Oyarzun et al. 1997; Wilson 1997), covering a roughly elliptical area of 4,000 km from Morocco to the Oslo Graben. Its influence in western Europe is speculated to continue during Mesozoic and Cenozoic times and is further related to the opening of the Atlantic Ocean (e.g. Burke et al. 1973; White and McKenzie 1989; Oyarzun et al. 1997). The small magma volumes represented by these alkaline dykes and the absence of magmatism between 252 ± 3 Ma (the youngest age determined for the SCS alkaline suite; Fernández Suárez et al. 2006) and 203 ± 2 Ma (intrusion of the tholeiitic Messejana-Plasencia dyke; Dunn et al. 1998) in central Spain, contradicts the classical theory of a deep mantle plume, continuously generating large magma volumes over a long period of time (Campbell 2005 and references therein). Moreover, no anomalously thick crust due to igneous accretion (underplating) of mafic rocks is known to have existed beneath the SCS. In fact, the composition of the lower crust seems to be felsic as indicated by the granulite xenoliths carried by the alkaline dykes (Villaseca et al. 1999). Recently, based on the absence of large magma volumes, thermal anomalies and low 3He/4He ratios, Kirstein et al. (2004) found no compelling evidence of mantle plume involvement in the genesis of Permo-Carboniferous magmatism in NW Europe. This criticism with the plume model has also been extended to the opening of the central Atlantic Ocean (McHone 2000).

We suggest that the most probable geodynamic setting for the generation of the SCS alkaline magmas is that of lithosphere thinning, passive upwelling of hot asphenospheric mantle and melting forced by adiabatic decompression. Asthenosphere-like melts might be identified with the isotopically depleted SCS dykes, whilst the lamprophyres with BSE-like isotopic compositions are recording melting at the base of the lithosphere. The present lithospheric thickness in central Spain is around 110 km (Fernández et al. 1998; Tejero and Ruiz 2002) and the region seems to have been tectonically stable since Permian times. Moreover, alkaline melt generation occurred at great depth (the stability field of garnet in peridotites ranges down to around 80 km; Nickel 1986) close to the proposed lithosphere–asthenosphere boundary. The occurrence of alkaline mafic dykes in the western Alps at 260 ± 1 Ma (Monjoie 2004) has also been ascribed to melting of an asthenosphere-like mantle as a consequence of lithosphere thinning. The close similarities, which exist in the geochronology and isotope geochemistry of the alkaline magmas from the SCS, Pyrenees and western Alps support the regional scale extension of this geodynamic context in southern Europe during the Permian.

Summary and conclusions

The alkaline lamprophyres and diabases from the SCS constitute a petrographically and geochemically heterogeneous suite of dykes. The abundance of mafic phenocrysts, the absence of primary magmas and presence of variation trends characterized by decreasing Ca, Ti, Ni, Cr and V towards lower Mg#, is in accordance with fractionation of olivine + clinopyroxene + kaersutite ± Cr-spinel ± ulvospinel ± plagioclase (this latter mineral only in the case of diabases).

Their bulk chemistry has not been influenced significantly by assimilation of crustal rocks, as their incompatible trace element ratios are similar to those of rocks typically derived from mantle sources (Ce/Pb and Ba/Nb in OIB and MORB). Moreover, they do not show positive correlation of silica content with Rb/Sr or 87Sr/86Sr ratios.

The clear positive Nb–Ta and negative Pb anomalies indicate that enrichment of their sources was not caused by any subduction-related component. The highly fractionated chondrite-normalised REE patterns and high Sm/YbN ratios suggest that they formed within the garnet stability field. The potassic nature of the SCS lamprophyres and their Rb/La and K/La ratios point to phlogopite dominating their generation, whereas amphibole prevailed in the case of the sodic diabases. It is likely that the enrichment event was caused by infiltration of K- and volatile-rich fluids or melts. The high REE contents shown by the isotopically depleted dykes imply that melting occurred shortly after metasomatism.

Two isotopic groups of SCS alkaline dykes are observed: (1) a PREMA-like (asthenosphere) component (εNd = +4 to +7.1; 87Sr/86Sr = 0.7029–0.7037; 206Pb/204Pb = 18.1–18.5); and (2) a BSE-like or slightly enriched lithospheric component (ɛNd = +1.4 to −0.9; 87Sr/86Sr = 0.7043–0.7051; 206Pb/204Pb = 18.3–18.5). The slight enrichment in 87Sr, 207Pb and 208Pb isotope ratios can be ascribed to the introduction into the mantle sources of continental or subduction-modified components.

A clear geochemical contrast can be observed when comparing lower Permian basic magmatism from SW and NW Europe. In areas north of the Hercynian front, no calc-alkaline or peraluminous magmatism is recorded at that time and alkaline manifestations are widespread. In areas affected by the Hercynian orogeny alkaline magmas are scarce before Upper Permian times. Moreover, northern areas show a regional tholeiitic event around 295 Ma, which does not exist in SW Europe. This difference agrees with the model outlined by Muttoni et al. (2003) in which NW and SW Europe were geographically assembled during mid Permian times.

Though a mantle plume has been proposed as the principal factor responsible for the rifting process during Permo-Carboniferous times there are features that are not in accordance with this model when applied to the SCS. This has also been argued for the magmatism in Northern Europe (e.g. Kirstein et al. 2004, 2006). Thus we favor a passive model in which rifting would follow from lithosphere thinning and upwelling of the hot asthenosphere. This tectonic regime might be transitional between the general Permo-Carboniferous extension in western Europe and the final opening of the Atlantic Ocean during the Mesozoic.