| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Journal of Petrology Volume 42 Number 12 Pages 2259-2277 2001
© Oxford University Press 2001
Gabbro-derived Granulites from the Northern Apennines (Italy): Evidence for Lower-crustal Emplacement of Tholeiitic Liquids in Post-Variscan Times
1DIPARTIMENTO DI SCIENZE DELLA TERRA, UNIVERSITÀ DI PARMA, PARCO AREA DELLE SCIENZE 157A, I-43100 PARMA, ITALY
2DIPARTIMENTO DI SCIENZE DELLA TERRA, UNIVERSITÀ DI PAVIA, VIA FERRATA 1, 27100 PAVIA, ITALY
3CNR–CENTRO DI STUDIO PER LA CRISTALLOCHIMICA E LA CRISTALLOGRAFIA, VIA FERRATA 1, I-27100 PAVIA, ITALY
Received July 30, 1999; Revised typescript accepted June 1, 2001
 |
ABSTRACT |
|---|
TOP
ABSTRACT
INTRODUCTIONThe mafic granulites from the Northern Apennines commonly contain significant amounts of either olivine or Fe–Ti-oxide phases. On the basis of mineralogy and whole-rock major and trace element compositions, their protoliths are recognized as cumulus rocks derived from variably evolved tholeiitic liquids. Clinopyroxenes from the olivine-bearing rocks show peculiar trace element compositions (e.g. low amounts of Cr, Zr and lanthanides, coupled with relatively high Sr concentrations) that record a process of granulite-facies trace element redistribution controlled by a reaction between olivine and plagioclase. The trace element compositions of clinopyroxenes from the Fe–Ti-oxide-bearing rocks point to igneous geochemical trends that argue for a common igneous parentage. The
Nd of the mafic granulites calculated at the age of emplacement of the gabbroic protoliths (290 Ma) ranges between +6·8 and -4·5. 87Sr/86Sr(290 Ma) varies between 0·7031 and 0·7077. The least evolved rocks show initial Nd–Sr isotopic compositions close to typical depleted mantle values. Two samples of felsic granulites associated with the mafic rocks yield age-corrected Nd of -8·0 and -5·7, and Sr isotope ratios of 0·7107 and 0·7109. As a whole, the Nd and Sr isotope data at the time of emplacement form a hyperbolic array. Whole-rock rare earth element and Nd–Sr isotopic compositions indicate that the parental liquids of the mafic granulite protoliths were tholeiites derived from a depleted mantle source. Isotopic compositions of the least evolved rocks are consistent with those of residual mantle lherzolites from the ophiolites of the Northern Apennines and the Western Alps. The protoliths of the Fe–Ti-oxide-bearing mafic granulites can be related to the least evolved rocks through fractional crystallization and assimilation of crustal material. The felsic granulites probably represent anatectic and migmatitic rocks resulting from multi-stage melting of lower-crustal basement rocks. KEY WORDS: granulite; gabbro; Northern Apennines; lower crust; crustal contamination; post-Variscan magmatism
|
INTRODUCTION |
|---|
The main insights into the magmatic processes occurring in the lower crust are provided by the study of (1) xenoliths brought to the surface by alkaline magmas, and (2) exposed granulite-facies terrains. These exposures are not necessarily representative of lower-crustal levels, because many of them are thought to derive from crustal thickening resulting from continent–continent collision (Mezger, 1992
). The gabbro-derived granulites from the Northern Apennines originated and re-equilibrated at deep crustal levels at 
290 Ma and were probably exhumed during the rifting event that led to the opening of the Ligurian Tethys (Marroni et al., 1998). These rocks locally preserve gabbroic textures (Marroni & Tribuzio, 1996) and are in places associated with felsic granulites. Both mafic and felsic granulites occur in Late Cretaceous sedimentary melanges from the External Liguride Units (Fig. 1), together with ophiolites and upper continental crustal rocks (Pagani et al., 1972; Braga et al., 1975). This rock association was interpreted to represent a continent–ocean transition between the Internal Liguride oceanic domain (Late Jurassic Ligurian Tethys) and the thinned continental margin of the Adria plate (Marroni et al., 1998).
|
Gabbroic rocks of Late Carboniferous–Early Permian age with granulite-facies metamorphic overprint are widespread in the remnants of the Adria plate margin at present exposed in the Western–Central Alps. Well-known examples are provided in the Ivrea Zone of the South Alpine domain (Voshage et al., 1990; Sinigoi et al., 1994, 1996; Quick et al., 1995) and in the Austroalpine nappes of Val Malenco (Hermann et al., 1997, 2001) and Sesia (Lardeaux & Spalla, 1991). The origin of these gabbroic rocks is commonly ascribed to magmatic activity at deep crustal levels, which developed in conjunction with post-Variscan lithospheric thinning (Costa & Rey, 1995; Schott & Schmeling, 1998). In this work, we have carried out trace element and Nd–Sr isotope analyses of selected gabbro-derived and felsic granulites from the Northern Apennines. In addition, clinopyroxenes from the gabbro-derived granulites have been analysed for trace elements by ion microprobe. The aim is to unravel (1) the affinity of the parental liquids of the gabbroic protoliths, (2) the role of fractional crystallization and crustal assimilation in the igneous differentiation, and (3) the possible genetic relations with the mantle rocks from the Northern Apennines and Western–Central Alps.
|
GEOLOGICAL AND PETROLOGICAL FRAMEWORK |
|---|
The Northern Apennines are a thrust and fold belt belonging to the Alpine orogenic system (Fig. 1). In the northernmost sector of the chain, two ophiolite-bearing units (Internal and External Liguride Units) are distinguished on the basis of lithostratigraphic and structural features. They are considered to represent intra-oceanic and peri-continental domains of the Ligurian Tethys, respectively (Abbate et al., 1980
; Piccardo et al., 1990; Molli, 1996; Marroni et al., 1998).
In the Internal Liguride Units, an ophiolite sequence is the base of a sedimentary cover ranging in age from Late Jurassic to Early Paleocene. The ophiolites consist of mantle ultramafic rocks intruded by a gabbroic complex and covered by N-MORB (mid-ocean ridge basalts) and sedimentary ophiolite breccias (e.g. Barrett & Spooner, 1977; Abbate et al., 1980; Cortesogno et al., 1987). The mantle rocks are mainly undeformed, clinopyroxene-poor (5–10 vol. %) spinel lherzolites showing incipient re-equilibration under plagioclase-facies conditions (Beccaluva et al., 1984; Rampone et al., 1996). The associated gabbroic complex formed by low-pressure fractional crystallization of N-MORB in Mid- to Late Jurassic time (Serri, 1980; Hebert et al., 1989; Tiepolo et al., 1997; Rampone et al., 1998; Tribuzio et al., 1999b, 2000). The lherzolites bear a residual geochemical fingerprint resulting from the extraction of N-MORB-type liquids (Rampone et al., 1996, 1998), although they are not the source of the associated gabbroic rocks and basalts. Rampone et al. (1998) showed that the extremely depleted Nd isotope signature of the Internal Liguride peridotites indicates a time of depletion earlier than Mid-Jurassic, probably Lower Permian.
The ophiolites from the External Liguride Units are mostly composed of mantle ultramafic rocks and normal to transitional MORB (Pagani et al., 1972; Marroni et al., 1998). These lithotypes crop out as slide blocks in Late Cretaceous sedimentary melanges, together with rarer slide blocks of continental origin (Braga et al., 1975; Marroni & Tribuzio, 1996). The latter mostly consist of either granulite-facies rocks or peraluminous granitoids, which also occur as clasts in polygenic breccias.
The mantle ultramafic rocks from the External Liguride Units mostly consist of fertile spinel lherzolites containing relatively high amounts of clinopyroxene (10–15 vol. %) and disseminated titanian pargasite (Beccaluva et al., 1984; Ottonello et al., 1984; Rampone et al., 1995). Temperature estimates for the spinel-facies assemblage point to relatively low values (mainly in the range 1000–1050°C), compatible with continental geothermal gradients (Beccaluva et al., 1984). The subcontinental origin of the lherzolites was confirmed by Sm–Nd and Rb–Sr isotope investigations (Rampone et al., 1995) that indicate a Proterozoic age as the time of accretion of the External Liguride mantle to the subcontinental lithosphere. According to Rampone et al. (1995), the External Liguride lherzolites underwent partial subsolidus recrystallization to plagioclase-bearing assemblages in Mid- to Late Jurassic time.
The granulite-facies rocks from the External Liguride Units were interpreted as slices of lower continental crust tectonically associated with the mantle ultramafic rocks (Marroni et al., 1998). Mafic granulites commonly show a metamorphic foliation that developed in granulite-facies shear zones. Rocks preserving gabbroic textural features at the hand-sample scale, such as igneous layering and coarse pyroxene porphyroclasts, are locally present (Marroni & Tribuzio, 1996). The mafic granulites have variable compositions and commonly contain significant amounts of either olivine or Fe–Ti-oxide phases (see also Montanini, 1997). Their gabbroic protoliths were emplaced at deep crustal levels, where they underwent slow cooling and recrystallization under granulite-facies conditions. Pressure and temperature values of 0·7–0·8 GPa and 800–900°C were estimated for the subsolidus re-equilibration of the gabbroic protoliths (Marroni & Tribuzio, 1996; Montanini, 1997). A gabbro-derived granulite gave a Sm–Nd clinopyroxene–plagioclase–whole-rock isochron of 291 ± 9 Ma, which was interpreted to date the granulite-facies recrystallization during post-magmatic cooling, probably close to the time of intrusion in the lower crust (Meli et al., 1996).
Mafic granulites are in places associated with felsic granulites (Balestrieri et al., 1997), but the original field relations between the two rock types are unclear. The felsic granulites commonly have quartz–feldspathic composition; the available petrological and geochronological data suggest a common P–T–t metamorphic evolution for mafic and felsic granulites (Marroni et al., 1998). Their exhumation to upper-crustal levels in Late Triassic–Mid-Jurassic time is testified by a pre-Alpine retrograde evolution to subgreenschist-facies conditions, which is accompanied by development of deformations progressively changing from plastic to brittle. The Late Cretaceous orogenic tectonics resulted in the widespread development of subgreenschist-facies assemblages in both mafic and felsic granulites (Marroni et al., 1998).
|
THE SELECTED SAMPLES |
|---|
Nine samples of mafic granulite and two samples of felsic granulite were analysed (Table 1). These rocks were selected on the basis of previous petrological investigations, by taking into account the variations in petrographic features and major element whole-rock and mineral compositions. The reader is referred to Marroni & Tribuzio (1996)
, Balestrieri et al. (1997), Montanini (1997) and Marroni et al. (1998) for an exhaustive description of mineral compositions.
|
The selected gabbro-derived granulites can be subdivided into three main types:
- olivine-bearing granulites are characterized by symplectites of orthopyroxene, clinopyroxene and Al-spinel (Fig. 2a) at the contact between olivine and plagioclase. Relatively coarse grains of orthopyroxene and clinopyroxene are present and commonly associated with anhedral Al-spinel (Fig. 2b), thus suggesting that these mineral aggregates might represent the annealed products of original symplectites. Plagioclase commonly forms coarse-grained polygonal aggregates. Sample A9 shows clinopyroxene-rich aggregates with triple junctions at 120°.
- Olivine-free, coarse-grained granulites display high modal abundances of clinopyroxene (clinopyroxene-rich granulites). Pyroxene and plagioclase show 120° triple junctions that are in places associated with idioblastic Al-spinel (Fig. 2c). Clinopyroxenes locally exhibit aggregates that mimic a flaser texture.
- Fe–Ti-oxide granulites show relatively high modal amounts of ilmenite and magnetite. The texture is overall granoblastic, but the Fe–Ti-oxide phases locally display an anhedral shape (Fig. 2d). These rocks may include Al-spinel, biotite and apatite as accessory phases. Sample IC4/9 contains minor amounts of euhedral Fe-rich olivine (Fo50). Sample 229E is the rock dated at 291 ± 9 Ma by Meli et al. (1996).
|
The gabbro-derived granulite types commonly contain accessory to minor amounts of titanian pargasite (TiO2 3·3–4·5 wt %), as coronas around orthopyroxene, clinopyroxene and/or Fe–Ti-oxide phases. Coronitic titanian pargasite may be associated with plagioclase and, in Fe–Ti-oxide-bearing granulites, with garnet (sample IC4/8). The pyroxenes from the olivine-bearing and the clinopyroxene-rich gabbronorites have higher Al2O3 (up to 8 wt %) and mg-number than those from the Fe–Ti-oxide-bearing gabbronorites. The anorthite content of plagioclase varies between 62 and 34 mol %.
The selected felsic granulites have inequigranular granoblastic texture and consist of mesoperthitic to perthitic feldspar, quartz and garnet. Garnet is unzoned and has high contents of almandine and pyrope (56–60 and 34–42 mol %, respectively). Fe–Ti-oxide phases, rutile, apatite, zircon and monazite may occur as accessory phases. The felsic granulites also contain accessory amounts of chlorite pseudomorphs, interpreted to have replaced original grains of orthopyroxene and/or biotite (Balestrieri et al., 1997).
Some of the selected samples show significant extents of fine-grained recrystallization related to granulite-facies shear zones and low-temperature mineral transformations. The latter are mostly represented by sericitization of feldspars and subordinate development of chlorite and actinolite at the expense of mafic minerals.
|
ANALYTICAL TECHNIQUES |
|---|
Whole-rock compositions are reported in Table 2. Major and trace (Ni, Co, Cr, V, Cu, Zn, Ga, Rb, Sr, Ba, Zr, Y) elements were determined by conventional X-ray fluorescence (XRF) techniques. Rare earth elements (REE), U, Th, Pb, Nb, Ta and Hf abundances, and duplicate analyses of Rb, Sr, Zr and Y, were determined by inductively coupled plasma mass spectrometry (ICP-MS) at the Centre de Recherches Petrographiques et Géochimiques (CRPG, Vandoeuvre-les-Nancy). Because of the low concentrations, incompatible element analyses of samples 209B and A13 were duplicated by ICP-MS at Activation Laboratories (Ancaster, Ontario). Precision and accuracy of XRF and ICP trace element analyses are commonly within 10%.
|
Isotopic compositions (Table 3) were determined at the Centro di Studio per il Quaternario e l’Evoluzione Ambientale (CNR, Rome). Rocks were decomposed with a mixture of ultrapure HF and HNO3 in closed Savillex beakers at 70°C for 48 h. The resulting solution was evaporated and the residue redissolved in 6·2N ultrapure HCl for 12 h. Sr was separated on a 3 ml AG 50W-X8 resin column. Nd was extracted by ion-exchange chromatography, using hydrogen-diethyl-hexyl-phosphate (HDEHP) as the stationary phase. Isotope analyses were carried out with a Finnigan MAT262 RPQ multicollector mass spectrometer. Internal precision (within-run precision) of a single analytical result is given as 2 SE and is obtained as a mean of 350–400 ratio measurements. Repeated analyses of standards during the period of analysis gave averages as follows: NBS987, 87Sr/86Sr = 0·710237 ± 10 (n = 7); La Jolla, 143Nd/144Nd = 0·512875 ± 10 (n = 8). 86Sr/88Sr and 146Nd/144Nd were normalized to 0·1194 and 0·7219, respectively.
|
Clinopyroxenes from gabbro-derived granulites were analysed for REE and selected trace elements (Table 4) by secondary ion mass spectrometry at Centro di Studio per la Cristallochimica e la Cristallografia (CNR, Pavia), according to the method described in Bottazzi et al. (1994). Analyses were carried out on the cores of large clinopyroxene grains. Precision and accuracy are estimated to be better than ±10% above 1 ppm concentration. Below 1 ppm concentration, precision is constrained by counting statistics to be in the range 10–30%.
|
|
WHOLE-ROCK MAJOR AND TRACE ELEMENT COMPOSITIONS |
|---|
The gabbro-derived granulites
The gabbro-derived granulite types overall have low SiO2/Al2O3 values (Fig. 3), thus suggesting a cumulus origin for their protoliths. Olivine-bearing and clinopyroxene-rich granulites are characterized by higher mg-number, Ni and Cr contents than the Fe–Ti-oxide-bearing granulites. In general, mg-number values and TiO2 are negatively correlated, suggesting a tholeiitic differentiation trend (see also Marroni & Tribuzio, 1996
; Montanini, 1997).
|
The concentrations of high field strength elements (HFSE) and REE are low in olivine-bearing rocks and progressively increase in clinopyroxene-rich and Fe–Ti-oxide-bearing granulites. The Sr contents are relatively constant and do not show significant variations among the various rock types. Samples A13, A9 and 235G are characterized by high values of LOI and K2O, despite these rocks having low modal amounts of hydrous and K-bearing minerals in their granulite-facies assemblages. The high LOI and K2O values are probably due to the extensive sericitization of plagioclase, as already illustrated by Marroni & Tribuzio (1996) and Montanini (1997). Those studies also showed that the plagioclase alteration was accompanied by an increase in original whole-rock contents of Rb and Ba.
Most samples display chondrite-normalized REE patterns with a slight steady decrease from light REE (LREE) to heavy REE (HREE) (Fig. 4). The olivine- and clinopyroxene-bearing granulites are characterized by positive Eu anomaly, which decreases in size with increasing total REE abundances. The positive Eu anomaly shows the role of plagioclase accumulation in these rocks, in agreement with their high Sr/LREE values (e.g. SrN/NdN = 6–31). Sample A9 differs from the other olivine-bearing granulites in the nearly flat LREE and in a less marked positive Eu anomaly. These variations are consistent with the fact that sample A9 has the lowest modal abundance of plagioclase and relatively high percentages of clinopyroxene + titanian pargasite (Table 1). The Fe–Ti-oxide-bearing granulites containing high amounts of apatite (samples IC4/8 and IC4/9) have the highest REE contents and show a barely appreciable Eu depletion. Sample 229E differs from the other Fe–Ti-oxide-bearing granulites in the slight LREE depletion that probably reflects the low plagioclase/clinopyroxene modal ratio and the absence of apatite.
|
The felsic granulites
These rocks are characterized by high SiO2 and K2O, and low CaO and MgO contents. Sample AM27E has higher MgO and FeOtot, and lower SiO2 than GA22, in agreement with its high modal abundance of garnet (Table 1). Both samples show high Ba contents, which are consistent with the local occurrence of Ba-rich alkali-feldspar (BaO up to 7·2 wt %) as exsolution product in mesoperthites. The REE patterns are LREE enriched, with no or slightly positive Eu anomalies and weak enrichments of HREE over the middle REE (MREE) (Fig. 4).
|
WHOLE-ROCK ISOTOPIC COMPOSITIONS |
|---|
Nd–Sr isotopic compositions were calculated at 290 Ma (Table 3), according to the age obtained by Meli et al. (1996)
for sample 229E. The olivine-bearing granulites A13 and A9, and the clinopyroxene-rich granulite (235G) have similar Nd(290 Ma), close to typical depleted mantle values (Fig. 5). The olivine-bearing granulite 209B differs in having a slightly lower Nd(290 Ma). These samples point to wide variations in the 87Sr/86Sr(290 Ma). The lowest value is close to typical depleted mantle compositions and pertains to the sample 209B, which is characterized by negligible plagioclase sericitization. The sample with the largest extent of plagioclase alteration (A13, Table 1) shows the highest 87Sr/86Sr(290 Ma). The Sr isotopic compositions of the selected olivine-bearing and clinopyroxene-rich granulites are therefore probably controlled by interaction with low-temperature fluids enriched in 87Sr/86Sr.
|
Fe–Ti-oxide-bearing granulites show wide Nd–Sr isotope variations. The sample containing olivine (IC4/9) displays the highest Nd(290 Ma) and isotopically approaches the olivine-bearing granulite 209B. The Fe–Ti-oxide-bearing granulite 229E shows the lowest Nd(290 Ma) and the highest 87Sr/86Sr(290 Ma) (-4·5 and 0·7077, respectively).
Nd and Sr isotopic compositions of the selected felsic granulites are respectively less radiogenic and more radiogenic than those of the gabbro-derived granulites. As a whole, if the Sr isotope data of markedly altered olivine-bearing rocks are not taken into account, the isotopic compositions of granulites at 290 Ma form a hyperbolic array. The isotopic compositions of felsic granulites approach those of the granulite-facies basement metasediments from the Ivrea Zone (Voshage et al., 1990).
|
TRACE ELEMENT COMPOSITIONS OF CLINOPYROXENES |
|---|
The clinopyroxenes from olivine-bearing granulites have low REE abundances. The pattern is characterized by flat LREE, a positive Eu anomaly and a weak depletion of HREE relative to MREE (Fig. 6). These clinopyroxenes display low Zr, Y, V, Sc and Cr, relatively high Sr and highly variable Ti (Fig. 7; Table 4). The clinopyroxenes from clinopyroxene-rich granulites are slightly richer in REE, Y, Zr, V, Sc and Cr. The REE patterns are similar, but the clinopyroxenes from the clinopyroxene-rich granulites do not show positive Eu anomalies.
|
|
The clinopyroxenes from Fe–Ti-oxide-bearing granulites are compositionally distinct relative to the clinopyroxenes from olivine-bearing and clinopyroxene-rich granulites. They have relatively high contents of REE, Y and Zr, which point to positive linear correlations (Fig. 7). In addition, they have relatively high V and Sc, and low Sr and Cr. The chondrite-normalized REE pattern is characterized by LREE depletion, a negative Eu anomaly and almost flat HREE. Although the size of the negative Eu anomaly and total REE contents vary significantly, the extent of the LREE depletion is nearly constant.
|
DISCUSSION |
|---|
The protoliths of olivine-bearing and clinopyroxene-rich granulites
Olivine-bearing and clinopyroxene-rich granulites display symplectitic intergrowths of clinopyroxene, orthopyroxene and spinel at the plagioclase–olivine interface, and/or triple junctions among clinopyroxene, orthopyroxene and plagioclase. Such features reflect subsolidus re-equilibration under granulite-facies metamorphic conditions. In particular, the symplectitic intergrowths show the occurrence of a metamorphic reaction between igneous plagioclase and olivine, which gave rise to clinopyroxene, orthopyroxene and spinel. The metamorphic re-equilibration under granulite-facies conditions is also indicated by the fact that plagioclase is relatively poor in anorthite component (40–50 mol %, Table 1). Presumably, the development of metamorphic clinopyroxene resulted in the partial consumption of the anorthite component of igneous plagioclase, thus leaving a residue enriched in albite component.
Whole-rock compositions (e.g. high mg-number and low SiO2/Al2O3 values, positive Eu anomaly in the REE patterns) indicate that the protoliths of olivine-bearing and clinopyroxene-rich granulites have a cumulus origin. In particular, sample A13 has high modal abundances of olivine + plagioclase and low whole-rock CaO contents, thus indicating that its protolith was a troctolite cumulate. A similar origin can be inferred for the olivine-bearing granulite 209B, as this sample is characterized by high amounts of normative olivine and lacks normative diopside and hypersthene (Table 5). Sample 209B has lower values of mg-number, Ni and Cr, and higher HFSE and REE concentrations than A13. These variations may be ascribed to crystallization from variably evolved liquids and/or to different proportions of liquid trapped between cumulus minerals. The trapped liquid could be related to the late growth of titanian pargasite (Tribuzio et al., 1999b). The latter is modally negligible in sample A13, thus suggesting that this sample originally was a troctolite cumulate with a low amount of post-cumulus material.
|
Olivine-bearing granulite A9 and clinopyroxene-rich granulite 235G contain clinopyroxene-rich aggregates which probably represent pseudomorphic replacements of original igneous clinopyroxenes. This inference agrees with their whole-rock compositions, which yield a significant amount of normative diopside. The occurrence of clinopyroxene in the original igneous assemblage is also consistent with the relatively high whole-rock concentrations of HFSE and REE. It is noteworthy that although the clinopyroxene-rich granulite 235G lacks olivine, its whole-rock compositions yield a high amount of normative olivine. The triple junctions between pyroxene, plagioclase and spinel (Fig. 2c) could thus reflect an annealing process starting from the products of the olivine–plagioclase reaction.
The clinopyroxenes from olivine-bearing and clinopyroxene-rich granulites: evidence for trace element granulite-facies redistribution
The trace element compositions of clinopyroxenes from olivine-bearing and clinopyroxene-rich granulites are difficult to reconcile with crystallization from silicate melts. The low Cr contents imply formation from extremely evolved liquids, which contrasts with the low contents of REE, Y and Zr, as well as with the high whole-rock mg-number values. In addition, most clinopyroxenes show positive Eu anomalies suggesting crystallization from basaltic liquids with an unusual positive Eu anomaly. Finally, hypothetical liquids in equilibrium with clinopyroxenes would be characterized by a marked enrichment of LREE over HREE (up to two orders of magnitude), which seems inconsistent with the slight LREE enrichment observed for the whole rocks.
Clinopyroxenes with similar REE patterns were reported by Mazzucchelli et al. (1992b) for coeval gabbro-derived granulites from the Upper Zone of the Ivrea mafic complex. The positive Eu anomaly in the clinopyroxenes from the Ivrea mafic complex was interpreted to reflect a geochemical feature of the igneous liquids that gave rise to the gabbroic rocks. These liquids would be derived from a hybridization process between crustal melts characterized by a positive Eu anomaly and mantle-derived melts (see also Mazzucchelli et al., 1992a). The clinopyroxenes reported by Mazzucchelli et al. (1992b), however, differ from those of the present study in their higher values of REE, Eu/Eu*, Y and Zr. In addition, the olivine-bearing and the clinopyroxene-rich granulites of the External Liguride Units have initial high Nd values, close to the depleted mantle, thus suggesting that the crustal contribution to the origin of these rocks was small or negligible (see also following section). On the other hand, the gabbro-derived granulites studied by Mazzucchelli et al. (1992a, 1992b) are characterized by low initial Nd (Voshage et al., 1990).
Clinopyroxenes with nearly flat REE and positive Eu anomaly were also reported by Gregoire et al. (1994) for spinel-bearing mafic granulites of the Kerguelen archipelago. The Kerguelen clinopyroxenes differ from the clinopyroxenes of the present study in the lower total REE contents, which approach chondritic values. The peculiar REE compositions of the Kerguelen clinopyroxenes were ascribed by Gregoire et al. (1994) to a metamorphic origin, through reaction between original igneous olivine and plagioclase.
A similar origin may also apply to the clinopyroxenes from the olivine-bearing and the clinopyroxene-rich granulites of the External Liguride Units. On the basis of mineral–liquid and mineral–mineral partition coefficients (e.g. Green, 1994), it is reasonable to assume that the concentrations of REE, Y, Zr, Cr, V and Sc in original igneous olivine and plagioclase were low or negligible. The low abundances of these elements in the clinopyroxenes, as well as their relatively high Sr contents and the positive Eu anomaly in their REE patterns, are thus consistent with the metamorphic hypothesis. However, the relatively high Ti concentrations retained by the clinopyroxenes cannot be reconciled with the olivine–plagioclase reaction and require chemical exchange with a Ti-rich phase. The Ti contents may therefore be related to a late re-equilibration of metamorphic clinopyroxene with igneous titanian pargasite and, possibly, with minor amounts of clinopyroxene originally present in the igneous assemblage of these rocks (see preceding section).
The clinopyroxenes from the olivine-bearing granulite A13 show the lowest Ti concentrations, as well as the lowest abundances of REE, Y, Zr, V and Sc. This suggests that the contribution of igneous amphibole–clinopyroxene to the observed clinopyroxene compositions was negligible. This agrees with the inferred origin from a troctolite cumulate with a low amount of post-cumulus material. On the other hand, the clinopyroxenes from the clinopyroxene-rich gabbronorites (probably containing clinopyroxene in the original igneous assemblage) lack the positive Eu anomaly and are slightly richer in REE, Y, Zr, V, Sc and Cr than those from the olivine-bearing gabbronorites. We conclude that the trace element compositions of analysed clinopyroxenes were acquired as a result of a complex metamorphic crystallization, which involved all the major minerals of the original igneous assemblage, i.e. plagioclase, olivine, clinopyroxene and titanian pargasite.
Primary features of parental liquids
Olivine-bearing and clinopyroxene-rich granulites have initial Nd and 87Sr/86Sr close to typical depleted mantle values. In addition, samples A13, A9 and 235G define a Sm/Nd regression line age of 310 Ma, with an initial isotope ratio of 0·5126 (Nd = +6·4; Fig. 8). Relative to the regression line, the olivine-bearing granulite 209B is displaced towards slightly lower 143Nd/144Nd values. This sample could be related to a liquid that was derived from a different mantle source or, more plausibly, that underwent a small amount of crustal assimilation (see following sections). The obtained age has no meaningful geochronological implications, because the 147Sm/144Nd values determined by ICP-MS are affected by high uncertainty and the three rocks define a small range of 147Sm/144Nd ratios. Nevertheless, the regression line age is overall consistent with the mineral isochron Nd–Sm age obtained for the Fe–Ti-oxide-bearing granulite 229E (Meli et al., 1996). This implies that samples A13, A9 and 235G were formed by liquids with negligible or minor crustal contamination.
|
A quantitative determination of trace element compositions of parental liquids on the basis of whole-rock compositions [e.g. following the methods of Bédard (1994) or Cawthorn (1996)] is precluded by the uncertainties resulting from the subsolidus re-equilibration. In particular, the proportions of cumulus minerals and trapped liquid are poorly known. Furthermore, the composition of the igneous plagioclase is unknown, making it difficult to select an appropriate set of trace element plagioclase–basalt partition coefficients, as they vary considerably in relation to the amount of anorthite component (e.g. Bindeman et al., 1998). Olivine-bearing and clinopyroxene-rich granulites are overall slightly LREE enriched relative to troctolites and olivine-bearing gabbros with N-MORB affinity, from either ophiolites (e.g. Ottolini et al., 1992; Tiepolo et al., 1997; Rampone et al., 1998; Tribuzio et al., 1999b) or modern oceanic basins (e.g. Pedersen et al., 1996; Barling et al., 1997; Casey, 1997). The protoliths of olivine-bearing and clinopyroxene-rich granulites were thus probably derived from liquids that were weakly LREE enriched relative to typical N-MORB. The inferred weak LREE enrichment relative to typical N-MORB therefore either represents a fingerprint of the primary mantle-derived melts, or is related to the early igneous differentiation, as a result of minor crustal contamination or deeper fractional crystallization involving clinopyroxene as the major phase.
The involvement of tholeiitic liquids with Nd–Sr isotopic compositions close to or approaching depleted mantle was also reported for other post-Variscan gabbroic complexes of the Alpine belt (Voshage et al., 1990; Miller & Thöni, 1997; Tribuzio et al., 1999a) and of the Corsica batholith (Cocherie et al., 1994). In particular, primary tholeiitic melts with a slight LREE enrichment relative to N-MORB were possibly involved in the origin of the gabbroic complexes of Sondalo (Western Alps, Tribuzio et al., 1999a) and Pila Canale (Southern Corsica, Cocherie et al., 1994).
Implications for mantle sources
The initial Nd–Sr isotopic compositions of the least evolved gabbro-derived granulites are close to those of the clinopyroxenes from the mantle lherzolites of the Northern Apennines (Fig. 5). In particular, the mantle lherzolites from the Internal Liguride ophiolites (Rampone et al., 1996, 1998) display a residual geochemical signature and yield Nd model ages suggesting an Early Permian partial melting. Such a melting event was related by Rampone et al. (1998) to the Lower Permian gabbros with N-MORB affinity from Koralpe (Eastern Alps, Miller & Thöni, 1997). The compositions of the clinopyroxenes from the Internal Liguride lherzolites plot on the same 143Nd/144Nd vs 147Sm/144Nd linear array obtained for the olivine-bearing and the clinopyroxene-rich granulites from the External Liguride Units (Fig. 8). This suggests that the melting event recorded by the Internal Liguride lherzolites could have generated the primary liquids of the mafic granulite protoliths.
The isotopic compositions of the least evolved gabbro-derived granulites are also consistent with those of the clinopyroxenes from the mantle lherzolites of the Southern Lanzo body (Western Alps, Bodinier et al., 1991). The Southern Lanzo peridotite is another mantle fragment retaining a residual geochemical fingerprint, which was uplifted in relation to the rifting event that led to the opening of the Ligurian Tethys. In particular, the clinopyroxenes from the Southern Lanzo lherzolites differ from those of the Internal Liguride ophiolites in the slightly lower 147Sm/144Nd values. The Southern Lanzo lherzolites could have thus generated liquids in Early Permian times with depleted isotopic compositions and a slight LREE enrichment relative to typical N-MORB. We thus conclude that also the Southern Lanzo lherzolites represent suitable refractory residua of the original mantle source that gave rise to the primary melts of the mafic granulite protoliths.
The protoliths of Fe–Ti-oxide-bearing granulites
The protoliths of the olivine-bearing and the clinopyroxene-rich granulites were gabbroic cumulates of plagioclase + olivine ± clinopyroxene, derived from relatively primitive melts. On the other hand, the Fe–Ti-oxide-bearing granulites show low mg-number values and high concentrations of incompatible elements, thus indicating that their protoliths formed from evolved liquids. Similar to the least evolved rocks, the protoliths of the Fe–Ti-oxide-bearing granulites have a cumulus origin. This is shown by samples IC4/8 and IC4/9, which have high TiO2 and P2O5, reflecting the precipitation of Fe–Ti-oxide phases and apatite, respectively. The cumulus origin of samples 225C and 229E is suggested by their low SiO2/Al2O3 values (Table 2). The Fe–Ti-oxide-bearing granulites do not show evidence for pyroxene-forming reactions, thus suggesting that their original igneous assemblage mainly consisted of plagioclase, clinopyroxene, orthopyroxene and Fe–Ti-oxide phases. In particular, sample IC4/9 differs from the other Fe–Ti-oxide-bearing granulites in the presence of Fe-rich olivine and in the negligible amounts of orthopyroxene, thus recalling the highly evolved olivine-bearing rocks of tholeiitic intrusive complexes (e.g. Eales & Cawthorn, 1996; Miller & Ripley, 1996).
The clinopyroxenes from the Fe–Ti-oxide-bearing granulites and those from the ilmenite-bearing gabbronorites of the coeval Sondalo gabbroic complex (Tribuzio et al., 1999a) have similar trace element compositions. Clinopyroxenes with similar trace element compositions were also reported for the Fe–Ti-oxide-bearing gabbro-derived granulites from the coeval Val Malenco complex (Hermann et al., 2001). These clinopyroxenes show wide variations in both total REE contents and size of the negative Eu anomaly, which were mainly ascribed to different percentages of melt trapped among the cumulus minerals (Hermann et al., 2001). The trace element compositions of clinopyroxenes from the Fe–Ti-oxide-bearing granulites of the External Liguride Units are consistent with an origin from evolved liquids, as they show high values of REE, Y and Zr. The low Sr contents and the negative Eu anomaly may be related to substantial plagioclase removal. Despite the recrystallization process under granulite-facies conditions, the clinopyroxenes from the External Liguride Fe–Ti-oxide-bearing granulites therefore seem to retain the geochemical signature of their igneous precursors and suggest crystallization from liquids with a similar geochemical fingerprint. This implies that the variability in the whole-rock REE patterns (Fig. 4) is mostly related to the cumulus nature of these rocks.
Origin of felsic granulites
Whole-rock compositions indicate that the felsic granulites were not derived from residual melts that originated through fractional crystallization of basaltic liquids, i.e. they are not genetically related to the liquids that formed the protoliths of the gabbro-derived granulites. For instance, the felsic granulites are characterized by the absence of Eu depletion, but this feature contrasts with a fractional crystallization process controlled by plagioclase removal. In addition, the felsic granulites show low Nd and high 87Sr/86Sr at the time of the gabbroic intrusion, thus suggesting a crustal origin. These rocks probably represent either metasedimentary country rocks, possibly occurring as xenolith bodies within the gabbroic complex, or crystallization products of anatectic melts.
The studied felsic granulites display close geochemical similarities to the orthopyroxene-bearing silica-rich granulites (‘charnockites’) cropping out as septa within the Ivrea mafic complex (Sinigoi et al., 1991, 1994, 1996). The felsic granulites from the External Liguride Units and the charnockites from the Ivrea mafic complex have subparallel incompatible element patterns (Fig. 9) and similar Sr isotopic compositions. On the other hand, the felsic granulites from the External Liguride Units differ from the granulite-facies basement metasediments of the Ivrea Zone (Schnetger, 1994) in the higher Ba, the lower U, Th, LREE, Ti and Nb, and in the lack of negative Eu anomaly.
|
Sinigoi et al. (1994, 1996) argued that the Ba enrichment in the charnockites from the Ivrea mafic complex was due to melting of K-feldspar-rich sources that originated through early dehydration melting (e.g. Beard et al., 1994; Carrington & Watt, 1995) of biotite-bearing gneisses. Melting of a K-feldspar-rich depleted source could result in the development of crustal melts with a positive Eu anomaly (Sinigoi et al., 1994, 1996), similar to that observed for sample AM27E. The marked U and Th depletions are also consistent with removal of these elements from the source rocks through previous extraction of anatectic melts. Sinigoi et al. (1994, 1996) concluded that the charnockites from the Ivrea mafic complex were derived from the crystallization of anhydrous anatectic melts extracted from restitic xenolith bodies of metasedimentary origin.
A similar origin applies to the External Liguride felsic granulite GA22, as it shows major element compositions resembling those of S-type granitoids. On the other hand, sample AM27E has relatively high MgO and FeO contents, thus arguing against a pure anatectic origin. This sample is characterized by low CaO and Nd(290 Ma), and high 87Sr/86Sr(290 Ma), and therefore is not related to hybridization of the anatectic melts with basalt-derived liquids. The preferred hypothesis for the protolith of sample AM27E is a migmatite-type rock, i.e. a rock of anatectic origin which enclosed a small amount of restitic material.
Origin of Fe–Ti-oxide-bearing granulites: isotope evidence for crustal contamination
The suite of gabbro-derived granulites can overall be related to a fractional crystallization process, with early separation of olivine, plagioclase and clinopyroxene, followed by the replacement of olivine by orthopyroxene at the liquidus. This is consistent with increasing REE and HFSE contents and decreasing mg-number values from the olivine-bearing to the Fe–Ti-oxide granulites (Figs
3 and 4). On the other hand, the initial isotopic compositions of the gabbro-derived granulites form a hyperbolic array suggesting mixing between a basaltic magma derived from a depleted mantle source and a crustal end-member characterized by low Nd and high 87Sr/86Sr (Fig. 5). In the post-Variscan gabbroic complexes of the Ivrea Zone (e.g. Sinigoi et al., 1994) and Sondalo (Tribuzio et al., 1999a), similar Nd–Sr isotope variations are accompanied by considerable 
18O increase, thus testifying to assimilation of crustal material. A process of crustal contamination is consistent with the occurrence of orthopyroxene in the original igneous assemblage of the Fe–Ti-oxide-bearing granulites, as orthopyroxene crystallization might be promoted by SiO2 enrichment through mixing with anatectic liquids. The felsic granulites have Nd–Sr isotopic compositions that are consistent with those of the crustal contaminant.
Isotopic compositions and trace element ratios that are sensitive to crustal contamination (e.g. LREE/HFSE) are not correlated in gabbro-derived granulites, but this is probably due to the fact that their protoliths were cumulates. It is, however, noteworthy that the Fe–Ti-oxide-bearing granulite IC4/9 has Nd–Sr isotopic compositions that are close to those of the least evolved rocks, despite its low mg-number and its high concentrations of incompatible trace elements. The original igneous assemblage of sample IC4/9 was probably characterized by the occurrence of olivine and by negligible modal amounts of orthopyroxene, thus suggesting an origin by a liquid that underwent minor crustal contamination. The lack of correlation between crustal isotope signature and degree of magma evolution indicates that the protoliths of the Fe–Ti-oxide-bearing granulites were not derived from a single-stage assimilation–fractional crystallization (AFC)-type process.
The lack of such correlation was also observed for the Ivrea Mafic Complex (Voshage et al., 1990) and for the coeval noritic–monzogranitic pluton of Cima d’Asta (Southern Alps; Barth et al., 1993). In particular, the lower portion of the Ivrea Mafic Complex (the so-called Layered Series) is mostly made of pyroxenitic to gabbroic rocks with abrupt Nd–Sr isotope heterogeneities and initial compositions that may be close to depleted mantle (Voshage et al., 1990). On the basis of Nd–Sr isotope variations as a function of igneous stratigraphy, the development of the Layered Series was ascribed to an AFC mechanism that operated erratically (Voshage et al., 1990). In the Ivrea Mafic Complex, moreover, a striking enrichment in radiogenic Sr approaching the enclosed paragneiss septa was shown by Sinigoi et al. (1996). In the External Liguride Units, the lack of information about country rock material and primary relations among the various gabbro-derived granulite types precludes a detailed evaluation of the igneous differentiation mechanism.
|
CONCLUDING REMARKS |
|---|
The protoliths of the mafic granulites from the Northern Apennines formed by fractional crystallization at deep crustal levels. The least evolved rocks show that the role of crustal assimilation in the igneous differentiation was minor. The trace element compositions of clinopyroxenes from these rocks do not provide information about the parental liquid compositions, as they record a process of trace element redistribution under granulite-facies conditions controlled by a reaction between olivine and plagioclase. Whole-rock REE and Nd–Sr isotopic compositions indicate that the parental liquids of the mafic granulite protoliths were tholeiites derived from a depleted mantle source.
The protoliths of the Fe–Ti-oxide-bearing mafic granulites can be related to the least evolved rocks through fractional crystallization and assimilation of crustal material. The associated felsic granulites probably represent anatectic and migmatitic rocks, resulting from multi-stage melting of basement crustal rocks, and show Nd–Sr compositions that are overall consistent with those of the crustal contaminant.
The post-Variscan gabbroic complexes from Western–Central Alps (Voshage et al., 1990; Miller & Thöni, 1997; Tribuzio et al., 1999a; Hermann et al., 2001) and Northern Apennines (this study) indicate that the lithospheric thinning related to the collapse of the Variscan orogen was accompanied by substantial production of tholeiitic liquids. The residual mantle lherzolites from the Northern Apennines–Western Alps ophiolites are suitable refractory rocks of such a post-Variscan magmatic event (see also Rampone et al., 1996, 1998).
|
ACKNOWLEDGEMENTS |
|---|
We are indebted to F. Castorina who carried out the isotope analyses. P. Bottazzi is thanked for assistance at the ion microprobe. We are grateful for reviews and comments by J. Beard, V. Chavagnac, J. Hermann, P. D. Kempton, M. Mazzucchelli, A. Peccerillo, E. Rampone, M. Scambelluri and C. Villaseca. This work was supported by MURST, FAR and CNR funds.
|
FOOTNOTES |
|---|
*Corresponding author. Telephone: +39 0521 905342. Fax: +39 0521 905305. E-mail: alessandra.montanini{at}unipr.it
