Introduction

Eclogites and amphibolites are widespread in sections of the Variscan basement exposed in the Alps and adjoining areas (Provence, Corsica and Sardinia). These basement slices are characterised by similar lithological units recording analogous metamorphic events. Eclogitic rocks usually occur in the cores of mafic boudins associated with amphibolite facies paragneisses (often migmatitic) and orthogneisses. There is a general consensus that most of these mafic rocks derive from Cambro-Ordovician igneous protoliths (Paquette et al. 1989; Rubatto et al. 2001; Schaltegger et al. 2003; Cortesogno et al. 2004; Palmeri et al. 2004; Giacomini et al. 2005), while the age of the high-pressure metamorphic overprint is debated.

Eclogite-facies rocks cropping out in the Gotthard massif were dated to 460–470 Ma by combining SHRIMP I data on zircons and the Sm-Nd isochron on whole rocks and garnets (Gebauer et al. 1988, Gebauer 1993). In the external Alpine Massifs of the western Alps, the eclogite-facies equilibration was instead ascribed to the Silurian (425–395 Ma) on the basis of intercept ages from conventional U/Pb analyses on zircons (Paquette et al. 1989). A not better defined “high-grade event” in the eclogites from northern Sardinia was dated to 403 ± 4 Ma on the basis of conventional U-Pb analyses on a metamorphic-looking zircon population (Cortesogno et al. 2004). Despite several other recent attempts (Rubatto et al. 2001; Palmeri et al. 2004; Giacomini et al. 2005), no other data are currently available for the timing of the eclogite facies overprint in the western Alps and Sardinia.

In this work, metagabbros and eclogites from the Savona Crystalline Massif in the Ligurian Alps, Italy (Fig. 1a, b) were analysed for major, trace element and Nd isotope compositions. The in situ LA-ICPMS trace element composition of relict igneous clinopyroxene from metagabbros was also determined. The U-Pb isotope and trace element composition of zircons from one gabbro sample and one eclogite were analysed in situ (LA-ICPMS) for geochronological purposes. The timing of pre-Alpine metamorphism in the Ligurian Briançonnais basement is poorly constrained. Published geochronological data only constrain the late, widespread amphibolite facies re-equilibration, which is thought to be coeval with the development of the main regional foliation. In particular, Rb/Sr and Ar-Ar data indicate that the amphibolite-facies metamorphism of the paragneisses and orthogneisses in the Ligurian basement occurred between 327 and 297 Ma (Del Moro et al. 1981; Barbieri et al. 2003).

Fig. 1
figure 1

a Tectonic map of the Western and Central Alps. The Penninic Zone comprises the Briançonnais Zone (or Grand Saint Bernard nappe system), which overthrust the Helvetic Domain. b Geological map of the Savona Crystalline Massif. c Geological sketch map of the Val Quazzola area. E, eclogite-facies relics; G, gabbro relics

Full size image

The principal goal of this study was to define the age of the still undated igneous protoliths and the high-pressure eclogitic overprint in the mafic rocks of this sector of the Variscan belt and to propose strategies for the interpretation of U/Pb zircon data in complex metamorphic systems. The study of mafic rocks cropping out in the Savona Crystalline Massif through a combination of in situ or bulk isotope data and geochemical analysis has provided new data for reconstructing the magmatic and tectono-metamorphic history of the basement of the Briançonnais zone. The comparison with published literature data on neighbouring southwest European Variscan units permitted a better understanding of the geodynamic evolution of the collisional belt.

Geological setting and field relations

The Savona Crystalline Massif (Fig. 1b) belongs to the Ligurian Briançonnais domain, a stack of pre-Alpine sedimentary sequences and basement slices thought to represent the Mesozoic passive margin of the European continent (Vanossi et al. 1984). The Savona Crystalline Massif comprises a strongly deformed mafic amphibolite-paragneiss complex in tectonic contact with orthogneisses. Locally, the amphibolites retain relics of gabbro and of eclogite facies assemblages of unknown age (Messiga 1987; Messiga et al. 1992; Cortesogno et al. 1993, Gaggero et al. 2004). Orthogneisses and paragneisses record white mica whole-rock and biotite whole-rock Rb-Sr ages of 327–297 Ma, which were attributed to cooling under amphibolite facies conditions (Del Moro et al. 1981). Ar/Ar ages of 311–302 Ma were obtained for white micas separated from paragneiss pebbles of the Tertiary molasse overlying the basement (Barbieri et al. 2003). A white mica separate from a paragneiss associated with the amphibolitised gabbros yielded a Rb/Sr cooling age of 308 ± 11 Ma (Del Moro et al. 1982). The Alpine metamorphic overprint does not exceed greenschist facies conditions (Messiga 1987; Messiga et al. 1992; Cortesogno et al. 1993, Cortesogno et al. 1997).

The study area comprises the valley of the Quazzola stream between Bersaggi and Richini villages (Fig. 1c). Undeformed bodies of either gabbro or partially retrogressed eclogite occur within the mafic amphibolites along the bed of the torrent. Gabbro- and eclogite-relics are not associated in the same outcrop. Near Richini village the mafic amphibolites contain decametre-scale metagabbro boudins. The gabbroic rocks preserve an igneous layering defined by modal and/or grain-size variations in plagioclase and clinopyroxene, the latter showing coronas of hornblende. Decimetre-scale layers of melagabbro are sometimes present. A coarse-grained fels rich in staurolite, garnet and magnetite occurs as a meter-scale lens within a deformed and amphibolitised gabbro. This lens is SiO2-poor and Al2O3-rich (both ∼30 wt%) and most likely related to digestion of crustal rocks by the basic magma (Braga and Tribuzio 1999). The contact between this enclave and the host amphibolitised gabbro is marked by the appearance of magnetite + quartz in the amphibolite paragenesis. Near Bersaggi village, about two kilometres northwest of the previous location, retrogressed eclogites form metre-scale, non-foliated domains within banded amphibolites. In the outcrop, the retrogressed eclogites show subhedral to anhedral garnet porphyroblasts (up to 1 cm in diameter) set in a light grey, fine-grained foliated matrix. Garnet porphyroblasts commonly bear coronas of dark-green amphibole aggregates. The transition from retrogressed eclogites to banded amphibolites is marked by a gradual decrease in the garnet-amphibole modal ratio and a progressive disappearing of the light matrix.

Methods

Whole rock major, trace element and Nd isotope compositions were determined at Activation Laboratories (Ancaster, ON, Canada). Major and trace element analyses (Table 1) were carried out by ICP-MS; precision and accuracy are generally estimated to be better than 10%. Nd isotope analyses (Table 2) were performed using a Finnigan MAT 261 8-collector mass-spectrometer in static mode. Powdered samples were dissolved in a Hf-HNO3-HCl mixture. 143Nd/144Nd ratios are relative to the value of 0.511860 for the La Jolla standard.

Table 1 Bulk major and trace element compositions of selected samples
Full size table
Table 2 Isotope Nd composition and 147Sm/144Nd ratios of the selected samples
Full size table

The trace element composition of igneous clinopyroxene was determined by laser ablation (LA)-ICP-MS at the CNR-Istituto di Geoscienze e Georisorse– Unità di Pavia. The reader can refer to Tiepolo et al. (2003) for analytical details The LA-ICP-MS instrument couples a Nd:YAG laser operating at 213 nm with a double focusing sector field ICP mass spectrometer type Element I from Thermo. The laser was operated at a repetition rate of 10 Hz, with a spot diameter of 20–40 μm and a pulse energy of about 0.01–0.03 mJ, respectively. Data reduction was performed using the software package "Glitter" (van Achterberg et al. 2001). NIST SRM 612 was used as the external standard. 44Ca or 29Si were adopted as the internal standards, depending on the mineral composition. Precision and accuracy, assessed against the BCR-2 USGS reference glass, are better than 6% relative.

Zircon grains for the geochronological characterisation were separated using standard techniques starting from a sieved rock fraction of 60–250 microns; grains were mounted in epoxy resin and then polished down using 0.25 micron diamond paste. Pb geochronology of zircons was carried out at the CNR-IGG-Unità di Pavia using an ArF excimer laser ablation microprobe operating at 193 nm (Geolas200Q-Microlas) coupled with the previously described HR-ICPMS (Element-ThermoFinnigan). Laser was operated at 5 Hz, fluency was set at 12 J/cm2and the spot size to 25 or 10 μm according to the internal complexity of zircon. The method is basically that described in Tiepolo (2003) and in Miller et al. (2006). The signals of masses 202Hg, 204Pb, 206Pb, 207Pb, 232Th and 238U were acquired in magnetic mode. The 235U signal is calculated from 238U on the basis of the ratio 238U/235U = 137.88. Approximately 60 s of background and at least 30 s of ablation signal from zircon were acquired. The time resolved signal and U/Pb ratios were carefully inspected in order to detect perturbations related to inclusions, cracks or mixing between different age domains. Zircon showing detectable common Pb (from the 204Pb count rate) were neglected. The laser induced U-Pb fractionation was corrected adopting a matrix matched external standard (zircon 91500) and considering the same spot size and integration interval on the unknown and the standard zircon. Data reduction, isotope ratio and apparent age calculation was carried out with the GLITTER software (Macquarie Research Ltd, 2001) developed by Van Achterbergh et al. (2001). Concordia plot and concordant age values were calculated using the ISOPLOT/EX software by Ludwig (1999). In each analytical run zircon 02123 (Ketchum et al. 2001) was analysed as unknown as quality control and average accuracy is estimated close to 1%.

LAM-ICP MS trace element analyses were carried out on selected age-concordant zircons according to the method for trace element determination described above but using the 193 nm laser source. Ablation spots about 25 μm in diameter were located near the spots of U-Pb analyses.

Petrology of selected samples

This study determined the whole rock major and trace element chemistry and Sm-Nd composition of three metagabbro samples (GBSV2, GBSV3 and GBSV4) and four retrogressed eclogites (SV1, SV2, SV3 and SV4). The selected metagabbros preserve igneous textures and fresh clinopyroxene of igneous origin and record the lowest degree of amphibolite-facies recrystallisation. An additional metagabbro sample (GBSV1) was chosen for zircon separation. The retrogressed eclogites were selected in order to avoid samples with a greater degree of retrograde amphibolite-facies recrystallization (e.g., amphibole aggregates forming pseudomorphs after garnet).

Gabbros

The gabbroic rocks (Fig. 2) consist of euhedral plagioclase (about 55 vol%), subhedral to poikilitic clinopyroxene and a completely altered mafic mineral. The latter is replaced by rounded aggregates of colourless amphibole, which make up about 10 vol % of the rocks and have been interpreted as pseudomorphs after olivine. In all samples, clinopyroxene is rimmed by fine-grained aggregates of green amphibole and minor plagioclase. Accessory Fe-sulphides occur as inclusions within the major minerals. Scarce pseudomorphs of rutile and titanite interstitial to the major minerals suggest the presence of accessory Fe-Ti-oxide phases in the original igneous assemblage. Rare plagioclase grains with igneous texture are preserved and have an anorthite content of 60–63 mol %. Plagioclase is commonly altered to fine-grained aggregates of albite and epidote. The mg# value [Mg/(Mg+Fe*)] of clinopyroxene ranges from 0.76 to 0.89. The Al and Cr contents in clinopyroxenes generally decrease with decreasing mg#, whereas Mn increases.

Fig. 2
figure 2

a Boudin of undeformed gabbro in a mylonitic metagabbro. b Thin section of an undeformed gabbro preserving relics of igneous plagioclase and pyroxene. Clinopyroxene is invariably rimmed by green amphibole, and the round-shaped amphibole aggregates are interpreted as pseudomorphs after olivine

Full size image

Deformed metagabbros (amphibolites to hornblendites) are characterised by the presence of both Mg-hornblende and plagioclase and by the development of porphyroblastic garnet in Fe-rich protoliths; pressure and temperature values of 0.4–0.7 GPa and 500–600°C have been proposed for the amphibolite-facies metamorphic re-equilibration (Braga and Tribuzio 1999).

Eclogites

High-pressure paragenetic relics (Fig. 3) consist of euhedral garnet porphyroblasts, omphacite, zoisite, rutile, kyanite, phengite and quartz. Omphacite (Jd32–41) is partially replaced by diopside-andesine symplectites. Garnet porphyroblasts are almandine-rich (49–62%) with lower amounts of grossular and pyrope (19–29% and 13–27%, respectively) and less than 3% spessartine; they are commonly zoned, with a decrease in Mn and Ca and a slight increase in Mg/(Mg+Fe) from core to rim (see also Messiga et al. 1992). Garnets are rimmed by green amphibole and plagioclase kelyphites. Kyanite is rimmed by fine-grained muscovite (± margarite) and quartz aggregates.

Fig. 3
figure 3

a Photograph of a polished eclogite sample highlighting the zoned syn- to post-kinematic garnet porphyroblasts rimmed by amphibole kelyphites; the deformed matrix mainly consists of omphacite, rutile and zoisite. b Back-scattered electron image of an eclogite sample: the relict eclogite-facies minerals are variably overprinted by lower pressure mineral parageneses

Full size image

Minimum pressure conditions of 1.7 GPa were proposed for the eclogite facies paragenesis (Cortesogno et al. 1997). Eclogite facies temperatures based on clinopyroxene-garnet Mg-Fe2+ exchange range from 650° to 750°C. The pressure-temperature conditions of the major late amphibolite-facies overprint are constrained to about 0.4–0.7 GPa and ∼600 °C (Braga and Tribuzio 1999; Cortesogno et al. 2004).

Chemistry

Whole rock geochemistry

Selected gabbros and eclogites have significantly different major and trace element chemical compositions. At nearly constant SiO2 contents, the gabbros have higher Al2O3, MgO and CaO contents and lower TiO2, Na2O and MnO contents than the eclogites. K2O contents are generally low (0.1–0.9 wt%), particularly in the gabbros where they never exceed 0.2 wt%. The mg# value of the gabbros (0.77–0.80; Fig. 4a) is significantly higher than that of the eclogites (0.51–0.67). The CIPW norm indicates that gabbros are olivine normative, supporting the hypothesis that the subrounded amphibole aggregates in the gabbros are pseudomorphs after olivine. Eclogites are hyperstene-normative.

Fig. 4
figure 4

a TiO2 vs. mg# plot evidencing the cumulus character of the gabbroic rocks. b Chondrite-normalised REE diagram and c trace element multivariation diagram (normalised to Primitive-Mantle values) of the investigated gabbros and eclogites. Ti* is recalculated from the TiO2 wt% in the major element analyses

Full size image

The REE pattern of the gabbros (Fig. 4b) displays depletion in LREE relative to MREE and HREE (LaN/SmN  =  0.6 − 0.7, YbN contents 3–5 times those in chondrites) and a clear positive Eu anomaly (Eu/Eu* = 1.6–1.8). The gabbros have low Nb, Ta, Zr, Y, Hf and Th contents.

The eclogites are slightly depleted in LREE with respect to MREE and HREE (LaN/SmN = 0.64–0.82), with a small to absent negative Eu anomaly (Eu/Eu*=0.89–1.06). Total REE contents are about 15–35 times those of a chondrite. The incompatible element pattern of the eclogites is smooth, with concentrations that are typically higher than in the gabbros. No significant Sr anomaly is observed. The LREE are lightly enriched relative to Th, Nb and Ta (Fig. 4b, c). U, Ba and Rb concentrations are highly variable and were not plotted in the multivariation diagrams: these elements are strongly mobile and their concentrations may have varied during the metamorphic events subsequent to the crystallisation of the igneous protolith. Present-day 143Nd/144Nd vary from 0.512965 to 0.513160 in the gabbroic rocks and from 0.513004 to 0.513129 in the eclogites; 147Sm/144Nd span from 0.193 to 0.241, thus yielding εNd(0) in the range +6.4 to +10.2.

Trace element composition of igneous clinopyroxene from the gabbros

The REE patterns of clinopyroxenes are characterised by LREE-depletion with respect to MREE and HREE (LaN/SmN = 0.07–0.17, for HREE contents 5–11 times those in chondrites) and by a weak negative Eu anomaly (Eu/Eu* = 0.7–0.9). The LaN/SmN ratio and the negative Eu anomaly slightly increase with total REE contents (Fig. 5; Table 3). Cr contents are relatively high (540–2600 ppm). Cr shows a rough positive correlation with mg# values. V and Sc range from 400 to 460 and from 130 to 160 ppm, respectively. The incompatible trace element patterns of the clinopyroxene (normalised to chondrite values, not reported) are characterised by low Nb, Ta, Sr, Zr and Hf contents with respect to the neighbouring REE. A remarkably similar incompatible element fingerprint is observed for clinopyroxenes from MOR-type cumulates (e.g. Tribuzio et al. 1999, 2004).

Fig. 5
figure 5

Chondrite-normalised REE diagram of clinopyroxenes from the investigated gabbro samples

Full size image
Table 3 Average major and trace element compositions of relict magmatic clinopyroxene from the gabbrocic rocks
Full size table

Geochronology

Zircon features

U-Pb LA-ICPMS dating of zircons was carried out on one gabbro (sample GBSV1) and one eclogite (sample SV1) from the Savona Crystalline Massif. Back-scattered electron and cathodoluminescence image analysis was completed prior to laser ablation analysis to investigate zircon morphology and internal zoning. Selected cathodoluminescence images of zircons from the two dated samples are reported in Fig. 6.

Fig. 6
figure 6

Cathodoluminescence images of representative zircons from the gabbro (1–5) and eclogite (6–13) and location of geochronological and microchemical analyses (Tables 4, 5, 6). Images 13 and 14 are respectively CL and BSE images of the same grain: note the inclusion of clinopyroxene-plagioclase symplectite and the thin fractures in the host zircon

Full size image

Zircons from the gabbro are rare and small, not exceeding 100 microns in length. Back-scattered imaging revealed that almost all zircons are short prismatic, subhedral to euhedral. A few subrounded grains also occur. The cathodoluminescence is generally low: some zircons appear nearly structureless with faint convolute zoning, while others show well developed oscillatory zoning.

Zircons in eclogites are 50–150 microns in length and show subhedral, short prismatic to anhedral and subrounded morphologies. Quartz, apatite and subordinate zoisite inclusions occur in some grains. One fractured zircon rim hosts two small inclusions of clinopyroxene-plagioclase symplectites (Fig. 6). Cathodoluminescence analysis reveals a variety of internal structures. Zircons with internal patchy zoning and strong luminescence contrasts are frequent. Several grains have cores with low luminescence and bright rims. The dark cores may be zoned (patchy or oscillatory) or almost structureless. The bright rims are unzoned.

Geochronological results

The U/Pb isotope compositions of zircons are reported in Tables 4 and 5.

Table 4 LA-ICP-MS U-Th-Pb isotope analyses and calculated ages of zircons from gabbro GBSV1
Full size table
Table 5 LA-ICP-MS U-Th-Pb isotype analyses and calculated ages of zircons from eclogite SV1
Full size table
Table 6 Trace element compositions of selected zircons
Full size table

Twenty-two laser ablation spots were made on sixteen zircon grains from gabbro GBSV1. Eleven ages are concordant and span a time interval from 2,136 to 459 Ma (Fig. 7). Six concordant analyses, obtained from subhedral, short prismatic zircons with faint oscillatory zoning have ages of 481–459 Ma with a mean concordant age of 469 ± 6 Ma (MSWD =  0.6, probability = 0.5). Three grains have slightly older ages ranging from 494 to 515 Ma. They have variable internal structures, with complex convolute zoning or dark unzoned areas. One subhedral, rounded crystal has a concordant age of 2,128 Ma (average of 2 analytical spots). It shows sector zoning and alternating bright and dark areas. The other analyses yield discordant points and do not define reliable intercept ages.

Fig. 7
figure 7

U/Pb Concordia diagram (left) and mean Concordia age (right) for the analysed zircons from the gabbro

Full size image

Eighty-three laser ablation spots on fifty-nine grains were performed on zircons from the selected eclogite. Fifty-one analyses gave concordant ages with a wide scatter of 645–305 Ma (Fig. 8a). The generally complex relationships between measured ages and zircon textures are summarised below.

Fig. 8
figure 8

a Density plot of the U/Pb ages for the analysed zircons from the eclogite. b mean Concordia ages of the oldest (igneous) zircon populations. c mean Concordia age of the youngest (metamorphic) zircon population. See the Discussion for further details

Full size image

One subhedral zircon, three dark crystal cores and one unzoned domain in a zircon fragment yielded concordant ages of 645–520 Ma. Another twenty-five analyses of core and rim portions in euhedral-subhedral, short prismatic zircon crystals yielded concordant ages of 486–430 Ma. The internal textures vary from dominant faint oscillatory or patchy zoning, to rare fir-tree zoning. This large group of data presents an irregular bell-shaped distribution centred at about 460 Ma. Due to the large scatter, the data define two mean concordant ages (Fig. 8b): the seventeen oldest analyses cluster around a well-defined mean concordant age of 468 ± 4 Ma (MSWD = 1.2, probability = 0.3), whereas the other seven define a younger mean age of 446 ± 2 (MSWD = 1.8, probability = 0.2).

Nineteen analyses on both core and rim zircon domains with variable internal textures yielded ages in the 420–305 Ma range. These crystals are generally characterised by low luminescence and have variable internal zoning: the most frequent features are dark unzoned cores surrounded by brighter unzoned rims, or irregular crystal sectors characterised by faint luminescence contrasts. Four analyses performed on one zircon core and on three overgrowths resulted in a mean concordant age of 407 ± 9 Ma (MSWD = 0.11 probability = 0.7). Seven analytical points span from 392 to 374 Ma and form two small groups (comprising 3 and 4 analyses respectively): they define two mean concordant ages of 392 ± 4 Ma (high MSWD of 3.3) and 376 ± 2 Ma (MSWD = 1.8, probability = 0.2) respectively. In particular, the unzoned and fractured zircon rim containing two ovoid inclusions of clinopyroxene-plagioclase symplectites yielded a concordant age of 374 ± 18 Ma.

Lastly, eight analyses of irregularly zoned crystals and of bright unzoned rims on older cores yielded concordant ages ranging from about 346 to 305 Ma. Excluding the single younger analysis at 305 ± 6 Ma, the seven remaining points (Fig. 8c) define a mean Concordia age of 333 ± 7 Ma (MSWD = 0.104, probability 0.75).

Zircon trace element composition: magmatic and metamorphic zircons

Five zircon crystals from the gabbro were selected for trace element analysis: the specimens display oscillatory zoning in CL images and have ages of 459–477 Ma (Fig. 9). They have high Y (550–2,000 ppm), Hf (7,700–10,000 ppm), Th (80–470 ppm) and U (180–530 ppm) concentrations, with Th/U ratios of 0.35–1.07. The C1-chondrite normalised REE pattern is characterised by strong enrichment in HREE with respect to LREE, with CeN/YbN values ranging from 0.002 to 0.1. Ce is enriched with respect to Pr and La (La concentrations are often below the detection limit) and Eu is always depleted with respect to the neighbouring REE (Eu/Eu* = 0.02–0.55). This trace element fingerprint and the oscillatory zoning in CL images indicate zircon growth under igneous conditions (Bea and Montero 1999; Schaltegger et al. 1999; Rubatto and Gebauer 2000; Rubatto 2002).

Fig. 9
figure 9

REE compositions (normalised to chondrite values) of selected zircons from the gabbro

Full size image

Twenty-one dated zircon grains from eclogite SV1 were also selected for trace element analysis (Fig. 10). The zircons with concordant ages of 480 to 446 Ma are characterised by relatively high Hf (8,800–12,000 ppm), Y (250–500 ppm), Th (10–53 ppm) and U (68–900 ppm) concentrations. Th/U values are higher than 0.03, whereas the chondrite-normalised REE patterns display a steady HREE enrichment with respect to LREE (CeN/YbN  <  0.003) and a negative Eu anomaly (Eu/Eu* = 0.2–0.6). As in the case of the analysed zircons from the gabbro, this chemical fingerprint, together with their dominant subhedral shape and the internal texture of selected crystals (characterised by faint oscillatory or patchy zoning), is typical of zircons formed under igneous conditions.

Fig. 10
figure 10

a REE compositions (normalised to chondrite) of selected zircons from the eclogite: domains with igneous texture are on the left, those with metamorphic texture are on the right. b Y contents and Th/U ratio plotted against U/Pb ages of texturally different zircon domains

Full size image

The chondrite-normalised REE pattern of two zircon rims dated at 419 ± 10 Ma and 403 ± 8 Ma are similar to those of the magmatic zircons described above, with a negative Eu anomaly and HREE enrichments over MREE and LREE. However, the textures of these rims and of all analysed zircons with ages in the 420–400 Ma range are typical of metamorphic zircons: such ages were all measured in crystal overgrowths or in unzoned and badly-defined crystal sectors with generally low luminescence. The chemical features of these zircons thus partially overlap with those of the older igneous zircons, but their internal textures and the younger ages indicate solid state re-crystallisation under metamorphic conditions (Hoskin and Black 2000).

The trace element composition of seven zircon sectors spanning in age from 392 to 325 Ma are characterised by low Y, Th and U contents (35–132 ppm, 0.05–7.84 ppm and 76–197 ppm, respectively) and by Th/U contents usually below 0.02. The MREE and HREE patterns are nearly flat with low concentrations that are 70–100 times typical chondrite values; LREE contents are generally below the detection limit. These chemical features, together with their complex internal texture (presence of overgrowths, irregular sector zoning) are typical of zircons grown (or possibly recrystallised) under metamorphic conditions (Schaltegger et al. 1999; Hoskin and Black 2000; Rubatto 2002). In addition, the low abundance of HREE and Y suggests the presence of garnet as a chemical buffer during zircon growth/recrystallisation (Rubatto 2002).

Discussion

Protolith age and geochemical affinity of the mafic rocks

In the gabbro, zircons with oscillatory zoning and trace element compositions typical of crystallisation under igneous conditions (high Th/U ratios, high trace element contents, negative Eu anomaly) yielded a mean concordant age of 469 ± 6 Ma. This age is interpreted as the timing of zircon crystallisation in the basic magma. The presence of zircons with ages older than the inferred crystallisation age suggests the involvement of an old (crustal?) component in the petrogenesis of the gabbro.

Zircons from the eclogite yielded very scattered ages of 556–305 Ma. Zircon grains with typical igneous textures and trace element compositions concentrate in the 486–430 Ma range. These zircons define two mean concordant ages of 468 ± 4 Ma (17 analyses, MSWD = 1.2) and 446 ± 2 Ma (7 analyses, MSWD = 1.8), respectively. This large scatter does not allow a straightforward estimate of the emplacement age for the eclogite igneous protolith.

The spread towards younger ages could result from mixed ages in relation to the textural complexity of some zircon crystals. Nevertheless, we believe that the careful observation of CL images prior to geochronological analyses and the use of small laser spots (10 μm) for the most critical zircons substantially reduced the risk of mixed analyses. The fact that the age results of analyses performed at 25 μm are comparable (within error) to those of the spatially more precise analyses performed at 10 μm, suggests that the younger zircon ages are most likely “true” ages and not the result of mixed analyses. In addition, the dispersion of concordant ages with igneous-like characters is often observed in magmatic rocks that underwent slow cooling rates or metamorphic re-equilibration after crystallisation (Ashwal et al. 1999; Hoskin and Black 2000; Ghezzo, personal communication). This partial resetting of the zircon isotope system is variably explained with high-temperature solid-state recrystallisation (Ashwal et al. 1999; Hoskin and Black 2000) or with zircon/fluid interaction during younger metamorphic events (Hartmann 2000). This process may be facilitated by the occurrence of metamict areas in zircon with strongly enhanced isotope mobility. A thermal perturbation related to a younger metamorphic event can promote re-crystallisation of the metamict zones, thereby yielding apparent ages with no geological meaning (Pidgeon 1991; Geisler et al. 2001).

We suggest that the time of crystallisation of the eclogite igneous protolith is reasonably recorded by the older mean concordant age of 468 ± 4 Ma defined by the largest “igneous” zircon population. Note that this radiometric date is within error of the inferred emplacement age of the gabbro. The younger ages clustering around the mean value of 446 ± 2 Ma are most likely due to partial isotope mobilisation during the subsequent metamorphic evolution.

The gabbros have high mg# (0.77–0.80) and positive Eu and Sr anomalies in the primordial mantle-normalised multivariation diagram, thus indicating a cumulus origin controlled by clinopyroxene ( ± olivine) and plagioclase segregation. We calculated the incompatible trace element composition of the melt in equilibrium with the clinopyroxene by applying the clinopyroxene/liquid partition coefficients for a basaltic system (Tiepolo et al. 2002). The MORB-normalised REE patterns of computed melts show no significant LREE enrichment (LaN/SmN = 0.8–1.8) and nearly flat HREE profiles.

The Nd isotope compositions at the time of igneous crystallisation were calculated on the basis of the U-Pb zircon data from the present study. The gabbros have initial ɛNd ranging from +5.4 to +8.8. The variability in the initial ɛNd values is most likely related to the interaction of a fractionating melt derived from a depleted mantle source with a low ɛNd crustal component. This is indicated by the fact that the initial ɛNd is roughly correlated with chemical features of clinopyroxene (e.g. it decreases with decreasing mg# and with increasing LaN/SmN). We thus propose that the mantle melt evolved through a process controlled by fractionation and concomitant assimilation of crustal material. The crustal contamination process is consistent with the occurrence of staurolite-, garnet- and magnetite-rich fels within the amphibolitised gabbros, which are considered evidence for country rock assimilation during magma emplacement (see also Braga and Tribuzio 1999). The presence of zircons with complex zoning patterns and ages (495–2129 Ma) older than the inferred magma crystallisation age provide further proof of inheritance from a crustal source.

The eclogites differ from the gabbros in their bulk chemical composition. Their chemical fingerprint may be representative of a melt composition and rules out derivation through a cumulus process. The variable mg# (0.51–0.67) and relatively high REE and incompatible element contents (15–35 times typical chondrite values) suggest an origin from evolved melts. The slight LREE depletion (LaN/SmN  =  0.6 − 0.8) and the Zr/Nb and Y/Nb ratios (33–43 and 10–17, respectively) are consistent with the chemical features of N- or T-MORB. The initial ɛNd of eclogites is similar to that of the analysed gabbros and ranges from +7.0 to +9.2, showing a rough negative correlation with the whole rock LaN/SmN ratio. There is good correspondence between the bulk trace element composition of the eclogites and the composition of the melts in equilibrium with the clinopyroxene from the gabbros of cumulus origin (Fig. 11). In addition, the U-Pb zircon ages indicate that the eclogite protoliths and gabbros are coeval (469 ± 6 and 468 ± 4 Ma, respectively). This sustains the hypothesis that the eclogite protoliths and gabbros were cogenetic, i.e. they formed from variably evolved and crustally contaminated melts that were derived from similar depleted mantle sources. The lack of high pressure parageneses in the metagabbros and associated paragneisses suggests that the crustal sections containing the relics of gabbros and eclogites were juxtaposed during the amphibolite facies Variscan evolution. An Alpine juxtaposition of the two sections cannot be demonstrated due to absence of know shear zones or faults in the region of interest.

Fig. 11
figure 11

Trace element (normalised to N-MORB values) of calculated liquids from clinopyroxene (metagabbro) compared to the whole rock compositions of the eclogites (average values)

Full size image

Cambro-Ordovician magmatism in the western European Variscides

Basic rocks of mid-Ordovician age are not a novelty in portions of the Variscan basement from the western Mediterranean area. The occurrence of mafic rocks with Cambro-Ordovician protoliths has already been reported in the Western-Central Alps, Sardinia and Provençe (Lancelot et al. 1998; Ménot et al. 1988; Paquette et al. 1989; Oberli et al. 1994; Abrecht et al. 1995; Poller 1997; Rubatto et al. 2001; Schaltegger et al. 2003; Cortesogno et al. 2004; Palmeri et al. 2004; Giacomini et al. 2005). Similarly to the Savona crystalline massif, these mafic rocks mostly occur as amphibolite or retrogressed eclogite bodies within paragneiss and orthogneiss sequences, and the protolith ages range from about 495 to 450 Ma. In particular, in-situ U/Pb zircon data (Rubatto et al. 2001; Palmeri et al. 2004; Giacomini et al. 2005) yield ages in the range of 460–450 Ma, which are usually younger than those obtained by conventional methods. These rocks commonly have E- to N-MORB chemical affinity, with initial ɛNd values ranging from +4 to +9. The low ɛNd values are alternatively attributed to processes of crustal contamination (e.g. Paquette et al. 1989) or to an origin in active margin/island-arc environments (Oberli et al. 1994; Schaltegger et al. 2003).

True oceanic crust sequences are scarce in the Variscan chain. In the southern-western portion of the belt they could be represented by the Cambro-Ordovician Chamrousse ophiolite from the Belledonne Massif of the western Alps and possibly by some sequences in the French Maures Massif (Bellot 2005, and references therein). The Chamrousse ophiolite was interpreted as a crustal fragment of an “immature” oceanic basin (Guillot et al. 1992 and references therein), in agreement with the lack of pelagic sediments in the stratigraphic cover, which instead contains significant quantities of metatuff (Pin and Carme 1987). Based on conventional U-Pb geochronology applied to zircons from a plagiogranite, the Chamrousse ophiolite was dated at 496 ± 6 Ma (Ménot et al. 1988). The ophiolite has a complex geochemical signature, as the presence of non-cumulitic mafic rocks with variable LREE enrichment and initial ɛNd of +5 to +9 testify (Bodinier et al. 1981; Pin and Carme 1987). The geochemical signature of the Chamrousse sequence was attributed to mixing of N-MORB and E-MORB-like end members with a third component having subduction zone affinity (Pin and Carme 1987).

The protoliths of metagabbros and eclogites from the Savona Crystalline Massif originated from depleted mantle melts emplaced within a continental crust. They most likely record a process of contamination by crustal material. We have no evidence to sustain the hypothesis of their origin in a supra-subduction or a mid-ocean ridge setting. We propose that the Ligurian Briançonnais basement records a mid-Ordovician extensional phase characterised by intrusion of mantle-derived melts into the continental crust. The widespread occurrence of mantle melts intruding the pre-Variscan crust in the Mediterranean area has been taken as evidence for the development of an oceanic domain at the northern margin of Gondwana (Stampfli and Borel 2002; Stampfli et al. 2002; von Raumer et al. 2003), which marks the beginning of the new Wilson cycle leading to the Variscan orogeny. Considering the scattered occurrence of mafic rocks in the basement outcrops of the southern Variscides and their variable chemical and isotopic fingerprints, we suggest that the Cambro-Ordovician basin did not evolve into a true ocean, but was likely a basin on (thinned?) continental crust.

Timing of Variscan metamorphism in the Ligurian eclogites

Dating the eclogitic peak by means of zircon geochronology is not straightforward, mainly because it is often difficult to relate zircon U-Pb ages to the P-T conditions of metamorphic equilibration. Zircons grown or re-crystallised under eclogite and amphibolite facies conditions may have almost indistinguishable trace element patterns. Their composition (especially HREE and Y) is controlled by the presence of garnet (Rubatto 2002), which often remains stable (or metastable) during decompression to amphibolite-facies conditions. In addition, the common dispersion of “metamorphic” ages along the Concordia does not facilitate the interpretation of data and demonstrates that zircon may be an open system under particular metamorphic conditions (e.g. Timmermann et al. 2004; Gray et al. 2004; Giacomini et al. 2005).

In the studied eclogite, the age of zircons younger than the crystallisation age of the inferred igneous protolith and with typical metamorphic texture spans from 420 ± 17 to 305 ± 6 Ma. The zircons with ages of 420 ± 17–402 ± 8 Ma underwent a textural resetting that cancelled the igneous textures: their trace element composition is similar to that of the older igneous zircons (negative Eu anomaly, HREE enriched with respect to LREE and MREE). These grains probably represent a further step in the process of annealing during prograde metamorphism, as already observed in the rejuvenated igneous zircon (mean age of 446 ± 2 Ma) described in the previous paragraphs. The mean concordant age of 407 ± 9 Ma obtained from the four oldest zircons with metamorphic textures is therefore considered a prograde stage of subduction-related metamorphism (amphibolite facies conditions?) which promoted a process of solid-state recrystallisation in zircon. This is consistent with the evidence for prograde metamorphic crystallisation in the Savona eclogites (i.e. hornblende grains within the cores of garnet porphyroblasts) reported by Messiga et al. (1992).

All analysed zircons with ages younger than 392 Ma are texturally and compositionally indistinguishable. They are euhedral to subrounded or mostly anhedral and have flat HREE patterns (generally 10–100 times typical chondrite values) with no negative Eu anomaly, overall low trace element abundances and low Th/U ratios. These zircons were derived from equilibration of igneous zircons or new precipitation in a system containing stable or metastable garnet and in the absence of high modal percentages of plagioclase. In particular, a fractured zircon rim dated at 374 ± 18 Ma contains an inclusion of diopside-plagioclase symplectite, a typical product of omphacite breakdown during decompression. There is no clear evidence to establish whether the zircon enclosed an omphacite crystal during the eclogite peak or a symplectite portion during the early stages of decompression. However, the presence of small fractures around the inclusion could be attributed to an increase in volume related to the omphacite breakdown reaction; in this case, the development of the diopside-plagioclase symplectite post-dated the measured zircon age. Following this hypothesis, the seven ages ranging from 392 ± 7 to 374 ± 8 Ma are related to equilibration under eclogite-facies conditions.

The trace-element patterns of zircons with ages in the 346–320 range (giving a mean Concordia age at 333 ± 7 Ma) are similar to those of older metamorphic zircons and are typical of growth or re-crystallisation in the presence of garnet. The comparison with published data on the neighbouring Variscan outcrops of the western Alps and Sardinia supports the notion that the zircons younger than 350 Ma are related to post-eclogite exhumation. The exhumation of high-pressure rocks and the process of nappe stacking related to the main continental collision are constrained to about 350–310 Ma by different geochronological methods (Ferrara et al. 1978; Rubatto et al. 2001; Di Vincenzo et al. 2004; Palmeri et al. 2004; Giacomini et al. 2005; Giacomini et al. 2006). We therefore propose that the flat HREE patterns of these zircons are linked either to the presence of stable-metastable garnet in the post-eclogitic assemblages or to the inefficiency of the re-crystallisation process in completely erasing the trace element signature of older “eclogitic” zircons.

One thin rim (no trace element composition for this spot) yielded a young concordant age of 305 ± 6 Ma. This age is consistent with available geochronological data for amphibolite facies equilibration recorded by associated paragneisses and orthogneisses, which indicate 327–297 Ma (Del Moro et al. 1981; Barbieri et al. 2003). We conclude that such a young zircon rim from the selected eclogite is related to late-stage growth during the final stages of exhumation.

The pre-Alpine metamorphic history: a regional comparison

The U/Pb age distribution in the selected eclogite of the Savona Crystalline Massif further constrains the Palaeozoic metamorphic evolution of the European Variscan belt in the Mediterranean area. For the first time in this sector of the chain, the high-pressure metamorphic overprint has been dated to 392–374 Ma. The post-eclogitic evolution is recorded by several zircon ages that yield a mean Concordia age of 333 ± 7 Ma. These new ages are complementary to previously published data (Rubatto et al. 2001; Cortesogno et al. 2004; Palmeri et al. 2004; Giacomini et al. 2005) on high-pressure rocks in the basements of Sardinia and the Western Alps. All these eclogite relics have similar geological settings and peak parageneses, and mainly occur as lenses and boudins within dominant amphibolite-facies paragneiss and orthogneiss sequences.

In northern Sardinia, SHRIMP and LA-ICP-MS U-Pb zircon data constrain the post-eclogite exhumation stage (granulite- to amphibolite-facies) from about 350 to 320 Ma (Palmeri et al. 2004; Giacomini et al. 2005). New data on two samples of kyanite-bearing eclogites from the Sardinian basement (Giacomini et al. unpublished) confirm the presence of a dominant metamorphic zircon population with ages spanning from 367 to 301 Ma. In the external massifs of the western Alps (Belledonne and Argentera), high-pressure metamorphism occurred from 425 to 395 Ma (conventional U/Pb zircon dilution, Paquette et al. 1989). In addition, muscovite Ar-Ar ages of 375–350 Ma for the Argentera Massif were attributed to the amphibolite-facies retrograde overprint (Monié and Maluski 1983). In the same area, however, SHRIMP and ID-TIMS U/Pb zircon ages from an eclogitised metagabbro (Rubatto et al. 2001) indicate that the post-eclogite amphibolitic overprint occurred at 323 ± 3 Ma. Older metamorphic ages (∼470 Ma) were proposed by Gebauer et al. (1988) and Oberli et al. (1994) for some eclogites cropping out in the Aar-Gotthard massif (central Alps). On the basis of an upper intercept discordia age at 870 Ma, Gebauer et al. (1988) proposed that these eclogites have igneous protoliths of Precambrian age and that the eclogite-facies overprint refers to Caledonian accretion.

Geochronological estimates from the western Alps and Sardinia constrain the beginning of the Variscan prograde metamorphic evolution to Siluro-Devonian times. The broad time span proposed for eclogite facies equilibration (425–375 Ma) can be attributed to either diachronous subduction in the different zones, or to uncertainties related to the application of different dating methods. We favour the latter hypothesis, in agreement with the complex internal features and the domains with different U/Pb ages observed in metamorphic zircons through recent micro-analytical studies (e.g. Rubatto et al. 1999; Song et al. 2005; Giacomini et al. 2006; this work). In addition, the different areas yield consistent post-eclogite U/Pb zircon datings at 340–320 Ma, which are commonly interpreted as the timing of amphibolite facies equilibration at middle crustal levels (Rubatto et al. 2001; Di Vincenzo et al. 2004; Palmeri et al. 2004; Giacomini et al. 2005; Giacomini et al. 2006). Thus, old “eclogitic” ages like proposed in the Belledonne and the Argentera massifs would imply very slow rates (0.6–0.4 mm/year) to exhume the high pressure rocks to the middle crustal levels (∼0.5 GPa) inferred for the widespread amphibolite facies overprint. Exhumation rates are almost 1.5- to 2-fold quicker (0.9–1 mm/a) if calculated taking into account the youngest eclogitic ages proposed in this contribution.

Conclusions

The Savona Crystalline Massif was intruded at ∼468 Ma by basaltic melts that originated from depleted mantle sources, probably in response to regional Cambro-Ordovician crustal extension at the northern margin of Gondwana. The mantle melts underwent fractional crystallisation and crustal contamination in relation to their emplacement in the continental crust.

In Devonian times at least part of the Savona Crystalline Massif was involved in a subduction zone that produced eclogite-facies assemblages pointing to equilibration at minimum pressures of 1.7 GPa and temperatures of 650–750 °C (Fig. 12). In particular, LA-ICP MS analyses of metamorphic zircons from one eclogite show that the high-pressure metamorphic overprint occurred between 392 and 374 Ma. Based on a comparison with other published data, the younger ages of metamorphic zircons that cluster around a mean Concordia age of 333 ± 6 Ma (i.e. in the Late Carboniferous) are ascribed to the amphibolite facies overprint related to the main Variscan collision and the chain exhumation.

Fig. 12
figure 12

Schematic P-T-t evolution of the eclogites from the Savona Crystalline Massif (pressure-temperature estimates from Messiga et al. 1992 and Cortesogno et al. 1997). *Ar-Ar data from Barbieri et al. (2003)

Full size image

The present study also shows that the U/Pb isotope system in zircons from eclogites is a powerful tool for reconstructing the chronology of entire orogenic cycles. Within a single eclogite sample from the Variscan Savona Massif, the evaluation of U-Pb zircon microanalyses in association with CL images and trace element signatures allowed us to identify five petrogenetic events: (1) old zircons inherited from the pre-Variscan basement, (2) igneous zircons dating the crystallisation of the eclogite protolith during an extensional tectonic event, (3) variably rejuvenated zircons, at least partly related to a prograde metamorphic stage which marked the onset of plate convergence, (4) peak eclogitic zircons dating the involvement of the crustal slice at deep subduction levels, and (5) retrograde zircons associated with eclogite exhumation under amphibolite facies conditions in response to continental collision.