| Journal of Petrology | Pages |
© 1999 Oxford University Press |
|
|
Tonalite-trondhjemite-dacite suites are generally formed at destructive plate boundaries (Ringwood, 1974; Smith et al., 1997). This tectonic setting gives rise to a geochemically complex spectrum of magmas as a result of a variety of potential source components (mantle wedge, subducted sediments, subducted oceanic crust) and processes (i.e. degree of melting, slab dehydration) in the source region. Ascending, high-Mg magmas are trapped beneath the relatively low-density island arc or continental margin crust, where they may fractionate and evolve until they become sufficiently buoyant to rise through the crust (Smith et al., 1997). Assimilation of continental crust in a continental margin setting or remobilization of pelagic sediments brought to depth by the subducted slab generally gives calc-alkaline magmas with elevated LILE/HFSE (large ion lithophile elements/high field strength elements) ratios (Perfit et al., 1980; Whalen, 1985; Hess, 1989; Janser, 1994; Smith et al., 1997). In an island arc setting, silicic magmas with low K2O may be generated in the mantle wedge by fractionation of high-Mg, low-K basaltic parent magmas with elevated Cr and Ni concentrations. Juvenile crust generated at a destructive plate margin is modified into a compositionally differentiated continental crust by processes including high-grade metamorphism and anatexis (Johannes & Holtz, 1996). Experiments show that high-grade partial melting of both mafic rocks and tonalites can give tonalite-trondhjemite-granodiorite (TTG) melt compositions (i.e. Johnston & Wyllie, 1988; Beard & Lofgren, 1991; Winther & Newton, 1991; Wolf & Wyllie, 1994; Singh & Johannes, 1996), leaving a pyroxene- or amphibole-rich granulitic restite behind (Beard & Lofgren, 1991). The present work is a restudy of the Precambrian Tromøy mafic-tonalitic-trondhjemitic gneiss complex, southern Norway, which gives a rare opportunity to study both of these processes within a single rock complex. The Tromøy complex is a classical example of low-K rocks metamorphosed at granulite facies (Field et al., 1980). The hitherto accepted petrogenetic model for the precursor of the gneisses is that they represent a suite of plagioclase-quartz dominated cumulates and trapped melts that were formed by fractional crystallization of a dacitic magma within the deep continental crust at 1·54 Ga (Rb-Sr whole-rock age; Field & Råheim, 1979; Field et al., 1980, , 1985). This has been taken as a classical example of processes generating an LILE-depleted lower continental crust (Touret, 1996). More recent Sm-Nd mineral ages and U-Pb zircon ages (Kullerud & Machado, 1991; Kullerud & Dahlgren, 1993) have, however, demonstrated that the high-grade metamorphism in the area occurred at 1100 Ma. These findings question a correlation between the emplacement of the Tromøy rocks and the high-grade metamorphism, and leave the genesis and metamorphic evolution of the Tromøy gneiss complex as open questions. The present study integrates field observations, geochemical data, isotopic data (Sr, Nd, Pb) and secondary ion mass spectrometry (SIMS) U-Pb zircon geochronology, and the results demonstrate that a more complex sequence of events is needed to account for the geochemical features of the Tromøy gneiss complex, where both crust-forming and crust-modifying processes play central parts. The Proterozoic part of the Baltic Shield (Fig. 1a, b) is built up by several crustal segments (traditionally named `Sectors' or `Belts'), separated by major shear zones that were active in Sveconorwegian times and possibly earlier (Hageskov, 1980; Park et al., 1991). Little is, however, known about the pre-Sveconorwegian evolution of the Kongsberg and Bamble Sectors: Rb-Sr whole-rock ages of 1536 Ma (Field & Råheim, 1979) for the tonalitic gneiss at Tromøy (Fig. 1c) and 1520 ± 50 to 1580 ± 50 Ma for dioritic gneiss and `enderbitic granulite' of the Kongsberg Sector (Jacobsen & Heier, 1978) indicate the possible influence of a Gothian (1750-1500 Ma) event that pre-dates the deposition of the sediments in the Bamble Sector and that occurred at 1370-1500 Ma (U-Pb detrital zircon ages; Knudsen et al., 1997a; Åhäll et al., 1998; de Haas et al., in review). Numerous gabbros intruded the southwestern part of the Baltic Shield at 1230-1110 Ma (Rb-Sr whole-rock ages, U-Pb and Sm-Nd mineral ages, Jacobsen & Heier, 1978; Munz & Morvik, 1991; Dahlgren et al., 1990;
de Haas et al., 1992, , 1993) during an extension-related magmatic event that is generally regarded as the first stage of the Sveconorwegian (1230-900 Ma) orogenic period (Starmer, 1990). A compressive, early Sveconorwegian tectonometamorphic event at 1100 Ma apparently affected the Bamble Sector only, whereas the main phase of the Sveconorwegian orogeny at 1000-900 Ma made no recognizable metamorphic imprint on the rocks of the Kongsberg-Bamble Sectors, but caused greenschist- to granulite-facies metamorphism in adjacent parts of the southern Baltic Shield (Fig. 1b; Johansson et al., 1991; Dahlgren, 1996; Bingen & Van Breemen, 1998). Post-orogenic magmatism in South Norway at 930 Ma (Rogaland) to 925 Ma (Bamble, Kongsberg and Østfold Sectors) (K-Ar, Rb-Sr, Pb-Pb and U-Pb datings; i.e. Pedersen & Falkum, 1975; Pedersen & Måløe, 1990; Schärer et al., 1996; Andersen, 1997) define a minimum age limit for Sveconorwegian orogenic processes in this area. Figure
The Tromøy gneiss complex crops out within the area of most intense Sveconorwegian metamorphism and deformation in the Baltic Shield (e.g. Field & Clough, 1976; Knudsen, 1996; Starmer, 1996). It is characterized by a granulite-facies mineralogy and low LILE and REE (rare earth elements) concentration levels (Moine et al., 1972; Field et al., 1980), and carries abundant evidence of the presence of syn-metamorphic carbonic fluids of possible mantle origin (Touret, 1971; Hoefs & Touret, 1975; Van den Kerkhof et al., 1994; Knudsen & Lidwin, 1996). The petrogenesis of these gneisses has been controversial ever since their first description by
Bugge, (1940). Early petrogenetic interpretations ascribe the geochemical characteristics of the Tromøy complex to metasomatism (Moine et al., 1972; Cooper & Field, 1977) or to loss of mobile elements during high-grade metamorphism (Field & Clough, 1976; Cooper & Field, 1977; Cameron, 1989; Touret, 1996). An alternative model suggesting that the gneisses originated as cumulates with varying fractions of trapped andesitic-dacitic melt, emplaced at high-grade P-T conditions in the deep crust at ~1540 Ma (Field & Råheim, 1979; Field et al., 1980), has been widely accepted during the last couple of decades. According to this hypothesis, the low LILE and REE concentrations are due to primary magmatic processes acting in the deep crust at ambient granulite-facies P-T conditions. The rocks along the present sampling traverses (A to C, Fig. 2), which reproduce those of
Cooper & Field, (1977), show gradational variations between the lithologies both on a 10 m and hand-specimen scale (Table 1) as a result of strong shear deformation and isoclinal folding post-dating the Sveconorwegain high-grade metamorphism (Knudsen, 1996; Knudsen & Lidwin, 1996). The green tonalitic gneiss consists of Pl + Qtz + Opx + Cpx + Hbl + Bt + Grt lithologies [mineral abbreviations from
Kretz, (1983)] with occasional anhydrous Pl + Qtz + Cpx + Opx domains, and grades into the associated mafic gneiss, consisting of Hbl + Pl + Qtz + Cpx + Opx + Bt + Grt. As the rocks are generally devoid of K-feldspar but are hornblende bearing, the terms charnockitic gneiss (Cooper & Field, 1977; Field et al., 1980, , 1985) and enderbitic gneiss (Van den Kerkhof et al., 1994; Knudsen & Lidwin, 1996) are formally incorrect. The gneisses are crosscut by decimetre- to metre-wide, relatively fine-grained mafic dykes (Fig. 3b), which have been metamorphosed with their country rocks and are partly tectonically broken up into separate lenses (Fig. 3d). Decimetre-wide, anhydrous veins and dykesof trondhjemite to leucotonalite (Fig. 3a, c, d; Pl + Qtz + Opx + Hbl + Grt; here referred to as `ordinary trondhjemite' for simplicity) are common within the tonalitic-mafic gneiss complex, but are also found in the metapelites of the islands in the nearby Tromøy-Hisøy-Torungen area, and as dykes up to 1 m wide crosscutting a Sveconorwegian coronitic gabbro in the Hisøy-Torungen area (Knudsen & Lidwin, 1996). Most trondhjemite intrusions are Opx bearing, with minor hornblende and minor to accessory garnet (i.e. enderbite sensu stricto). Decimetre-wide, coarse-grained dykes and veins with the primary assemblages of Pl + Qtz + Grt, Pl + Qtz + Hbl, or similar veins where hornblende is overgrown by orthopyroxene (Kullerud & Dahlgren, 1993) are found locally. At Tybakken (Fig. 2), pegmatitic hornblendite forms decimetre-wide veins and pods in the tonalite, spatially associated with veins of coarse-grained, garnet-rich trondhjemite. The intrusive orthopyroxene- and hornblende-bearing trondhjemitic veins have induced dehydration zones of 4-5 mm width in surrounding hornblende-bearing lithologies (arrow in Fig. 3c), and the trondhjemites carry abundant magmatic CO2 inclusions (Knudsen & Lidwin, 1996) with a carbon isotope mantle signature (Hoefs & Touret, 1975; Van den Kerkhof et al., 1994). Field observations demonstrate that the trondhjemite is intrusive into the other lithologies, and is thus unlikely to have formed as plagioclase-dominated cumulates, as was suggested by
Field et al., (1980). An Sm-Nd mineral age of 1073 ± 28 Ma (Kullerud & Dahlgren, 1993) on trondhjemitic veins from Hove (Fig. 2), indicates that the intrusion of the trondhjemites at Tromøy overlaps with the regional granulite-facies metamorphism (M2 at 7·5 kbar, 840°C; Knudsen, 1996) at 1100 Ma (U-Pb zircon age; Kullerud & Machado, 1991). Trondhjemite forming diffuse, millimetre-wide zones or pods in the tonalite (arrow in Fig. 3a), represents a similar in situ incipient melt unable to segregate. Table 1. The mail element compositions and normative calculations of tonalitic gneiss, mafic gneiss, mafic dykes and trondhjemitic gneisses from Tromøy
Figure
Figure
The inhomogeneous shear deformation of the high-grade Bamble rocks is particularly well expressed on certain wave-washed beach localities (i.e. at Hove and Sandum, Figs 2 and 3a, d). Here, trondhjemite veins have intruded the tonalitic gneiss in a NW-SE direction, and the white weathering colour of plagioclase makes the veins especially prominent. The trondhjemites form alternating less deformed, open folded bands (Fig. 3d) with NE-SW directed fold axes and strongly deformed, NE-SW directed remobilized bands parallel to the main foliation of the area. This pattern of low-strain zones alternating with strongly NE-SW deformed rocks is repeated at a larger scale. Several metre-wide, low-strain `pockets' (for example, in an old quarry at Tybakken, Fig. 2) reveal that the interbanded tonalitic to mafic gneisses are composed of a suite of medium-grained tonalitic gneiss and hornblende-bearing mafic granulite, crosscut by fine-grained mafic dykes and later by trondhjemitic veins. Major elements were analysed by X-ray fluorescence (XRF) on fused lithium borate glass pellets, and trace elements by XRF on pressed powder pellets. The analyses were performed on a Philip 2400 instrument with X47 software at the Department of Geology, University of Oslo. The XRF major element precision is within ±1%, whereas the trace elements have a detection limit below 1 ppm, except Rb (5 ppm), and U, Th and Nb (all 1 ppm). The REE were analysed by instrumental thermal neutron activation (Brunfelt & Steinnes, 1969) at the Institute of Energy Technology (IFE) at Kjeller, Norway. Accuracy and precision are better than 3% for La and Sm, 5% for Ce and Nd, 6% for Tb, 7% for Tb and Yb, and 13% for Lu. In the samples where a given trace element concentration is lower than its detection limit (Table 2), the detection limit is used in the normalized multi-element plots, and serves as an upper concentration limit for the element in the given sample. Table 2. The trace element compositions of tonalitic gneiss, mafic gneiss, mafic dykes and trondhjemitic dykes and veins from Tromøy
Zircons from the non-magnetic fraction of the samples were mounted in epoxy, and 29 of these were selected for SIMS analysis. The U-Pb zircon dating was performed in the NORDSIM laboratory located at the Swedish Museum of Natural History in Stockholm, using a CAMECA IMS1270 ion microprobe. Technical details regarding sample preparation have been given by Knudsen et al., (1997a) and the analytical conditions have been described by
Whitehouse et al., (1997). The procedures for the Rb-, Sr-, Sm-, Nd- and Pb-isotope analysis are identical to those of Knudsen et al., (1997b). Nd isotopic ratios are normalized to 146Nd/144Nd = 0·7219. During the period when the present analyses were made, the Johnson and Matthey Batch S819093A Nd2O3 gave 143Nd/144Nd = 0·511101 ± 0·000013 (2[sigma]), whereas the NBS 987 Sr standard yielded 87Sr/86Sr = 0·710190 ± 0·000050 (2[sigma]). The 2[sigma] error of the 147Sm/144Nd ratio was 0·025%. Lead isotope analyses were corrected for mass fractionation off-line, using correction factors derived from multiple runs of the NBS SRM 981 common lead standard and the standard composition determined by
Todt et al., (1984). The instrumental fractionation amounted to 0·095%/a.m.u.; a 2[sigma] external precision of 0·2% (counting statistics + fractionation) is assumed in the lead isotope ratios. The major and trace element data reflect the difficulty of sampling pure end members. The terms `mafic gneiss', `tonalitic gneiss' and `trondhjemite' of Tables 1 and 2 indicate the dominant rock type present in the samples, which are often heterogeneous on a small scale. The classification of tonalite (SiO2 = 60-70 wt %) and trondhjemite (SiO2 > 70 wt %) is based on CIPW norms (Table 1). Figure 4a shows that the normative Ab/Or ratios are similar for the mafic gneisses, the tonalites and trondhjemites, and the rocks are generally low in Al (Al2O3 < 15 wt % for most samples, Table 1), metaluminous [Al/(Na + K + Ca) < 1] and show decreasing Al2O3 with increasing SiO2 content. They are characterized by normative quartz + orthoclase + albite + anorthite + diopside + hypersthene, except the four peraluminous samples, which are corundum normative. Additional normative minerals are ilmenite, magnetite and apatite. Figure
The mafic and tonalitic gneisses define two restricted fields within the Pl-Opx-Qtz triangle of the Pl-Ol-Qtz diagram (Fig. 4b). The linear trend defined by the trondhjemites is mainly due to the difficulty of separating trondhjemite mechanically from its host, and the best estimate of the trondhjemite end-member composition is given by the most silica-rich compositions. The trondhjemite end-member composition overlaps with minimum melts formed by partial melting of either a mafic or tonalitic source at moderate pressures (SJ1, SJ2, BL1 and BL2 in Fig. 4b). On the other hand, high-pressure melting within the stability field of garnet in tonalite compositions (of the order of 15 kbar), gives far more plagioclase-normative melts (CW in Fig. 4b). The Tromøy rocks define a moderate Fe-enrichment trend in the AFM diagram (Fig. 4c), straddling the tholeiitic to calc-alkaline division line. The MgO concentrations and mg-numbers are within the range commonly observed in evolved, low-magnesium lavas from modern oceanic arcs (Smith et al., 1997). Compared with recent analogues, most samples from Tromøy plot in the field of relatively low arc maturity (a to b in the figure; Janser, 1994), which is different from low-Fe trends formed in continental margin arc settings where significant assimilation of continental crust is involved (d in the figure; Grove & Kinzler, 1986). The mafic gneiss ranges from low-K basaltic to trachyandesitic compositions, and most samples of the mafic gneisses, tonalites and trondhjemites plot in the field of low-K magmatic rocks, typical of an immature oceanic island arc setting (Fig. 4d; Smith et al., 1997). The tonalitic gneiss and the trondhjemite show a clear `volcanic arc granite' signature in terms of Rb-Y-Nb. The mafic and tonalitic gneisses are enriched in LILE (Rb, Th, K, Pb) and depleted in Nb, Ti, Zr [i.e. high field strength elements (HFSE)], Y and heavy REE (HREE) relative to N-MORB (normal mid-ocean ridge basalt) [Fig. 5, normalized to N-MORB data of
Pearce & Parkinson, (1993)] and show the low HFSE/REE and low HREE/LILE values typical of subduction-related magmatism (McCulloch & Gamble, 1991; Hawkesworth et al., 1994). The range of Ni concentrations observed in the mafic gneiss (9-41 ppm) is similar to the level found in evolved magmas in modern oceanic arcs (e.g. Smith et al., 1997). Uranium concentrations below the 1 ppm detection limit are observed in a majority of samples, but it should be noted that this is not necessarily a magmatic feature, as the entire region has suffered uranium loss during high-grade metamorphism post-dating emplacement of the gneiss protolith (Knudsen et al., 1997b). The trondhjemite has an LILE and HFSE distribution similar to the tonalitic and mafic gneisses, but some practically pure trondhjemite samples have low potassium, and high to extreme K/Rb ratios of >13 000 [Table 2 and
Knudsen et al., (1997b)]. Figure
The mafic dykes plot solely in the tholeiitic field of the AFM diagram. Their lithophile element distribution patterns overlap with the tonalitic gneiss, except for slightly lower incompatible element enrichment, and a flat trend in the compatible end of the pattern. The Ni content of the mafic dykes (26-187 ppm) exceeds that of the mafic gneiss. The REE analyses of 13 carefully selected samples of practically pure end members of the different rock types of the Tromøy complex are given in Table 3. Chondrite-normalized REE distribution patterns of the tonalitic and mafic gneisses are overlapping, showing slight light REE (LREE) enrichment, no or a shallow negative Eu anomaly and flat HREE patterns (Nd/Sm ratios in the range 2·9-5·0 and La/Yb = 2·9-12·9), resembling the patterns of modern calc-alkaline magmatic complexes (Gill, 1981). The ordinary trondhjemite has lower REE contents, Nd/Sm = 4·9 and 6·3, its REE patterns are distinctly concave upwards in HREE, with La/Yb ranging from 3·6 to 6·7, and show positive Eu anomalies. The hornblendites and Grt-bearing trondhjemite have low REE levels compared with the tonalitic gneiss. REE data on the Tromøy rocks by
Field et al., (1980), which were interpreted to represent LILE-deficient, low total-REE cumulates from an andesitic-dacitic magma, overlap with the present REE data on the tonalites and trondhjemites. The two late mafic dykes analysed have nearly flat REE patterns at 30 to 50 times chondritic concentration level with La/Yb = 2·6 and 3·4 and weak negative Eu anomalies. Table 3. REE concentrations of the Tromøy complex
Zircon morphology and zoning Two samples of mafic gneiss mixed with tonalite and one tonalitic gneiss were selected for single zircon SIMS analyses. The intrusion of the trondhjemitic veins is well dated to 1073 ± 28 Ma from a previous Sm-Nd mineral study (Kullerud & Dahlgren, 1993), and therefore only one zircon from a trondhjemitic vein was included in the present SIMS analyses. All zircons were investigated by scanning electron microscopy (SEM) cathodoluminescence imaging before analysis, and by detailed backscatter electron (BSE) imaging after analysis. The images reveal zircons with complex internal structures that are unsuitable for conventional thermal IR multispectral scanning (TIMS) U-Pb dating. Three different episodes of zircon growth can be identified: (1) oscillatory zoned, magmatic cores (Fig. 6a-d); (2) overgrowths of zircon of up to 20 µm width, with homogeneous, moderately intense BSE brightness (Fig. 6c, d); (3) BSE bright zircon occurring as embayments or ~5 µm wide domains parallel to the oscillatory zoning (Fig. 6c), or also as micrometre-wide channels crosscutting the magmatic zoning (Fig. 6b). These textures suggest that the BSE bright areas are related to fluid-induced zones of zircon reworking. The following relative ages are indicated: (1) oscillatory zoned zircon (oldest), (2) metamorphic overgrowth and (3) reworked channels or domains (youngest). A contrasting type of apparently unzoned and homogeneous, metamorphic zircon has been identified in the mafic and tonalitic gneisses. Figure
SIMS data Six U-Th-Pb SIMS zircon analyses have been performed on the tonalitic gneiss, 22 on the mafic gneiss and one on the trondhjemitic vein, with altogether 20 and nine analyses of magmatic and metamorphic zircons, respectively. The maximum 204Pb/206Pb ratio observed is 0·00111 (Table 4), which gives a 206Pbnon-radiogenic/206Pbtotal ratio of 1·03%. Of the analysed spots 81% give 204Pb/206Pb ratios below 0·0002, which gives a maximum 206Pbnon-radiogenic/206Pbtotal ratio of 0·19%, and only minor corrections for common lead. The complex internal textures of most zircons suggest that special care must be taken in interpreting the results. Careful BSE investigations of all spots analysed have shown that the ~30 µm ion beam is commonly too large to resolve the complex internal zonation pattern of most zircons, and thereby gives mixed ages. Also, apparently simple, metamorphic zircons display a spread in ages. Many spots are inversely discordant, suggesting U loss (or radiogenic Pb gain), and 45% of all zircons analysed are concordant within the 1[sigma] error (Fig. 7a, b). The data set is not suitable for an exact age determination, but it demonstrates clearly that the intrusive tonalite-mafic complex is Sveconorwegian rather than Gothian as was assumed previously (e.g. Field & Råheim, 1979). It is suggested that the maximum 207Pb/206Pb ages obtained for the magmatic and metamorphic zircons are the best estimates of the ages of these events, giving the following ages for the three zircon-forming processes: (1) oscillatory zoned, magmatic zircons formed at 1198 ± 13 Ma or slightly earlier (analysis 23 B, Table 4)-this dating is referred to as 1200 Ma in the following text; (2) metamorphic zircon growth at 1125 ± 23 Ma (analysis 02A, Table 4), which overlaps with the metamorphic zircon ages of 1122 and 1133 Ma from metasediments in the Hisøy-Torungen area (U-Pb SIMS zircon ages; Knudsen et al., 1997a); (3) later zircon reworking and U loss. The intrusion age of the mafic and tonalitic gneisses cannot be separated by the present U-Pb data. Table 4. Single zircon ion-probe data
Figure
Rb-Sr Regression of the whole-rock Rb-Sr isotope data for the tonalites and mafic gneisses gives a very poorly defined correlation line with (87Sr/86Sr)i = 0·7031 and an apparent age of 1480 Ma. This is comparable with the Rb-Sr whole-rock isochron age of 1536 Ma obtained by
Field & Råheim, (1979). Both ages pre-date the present U-Pb SIMS zircon data by ~300 my, suggesting that any Rb-Sr whole-rock isochron age calculated for the Tromøy rocks is geologically meaningless. Statistically valid Rb-Sr correlation lines without age significance may form by two-component mixing (e.g. Faure, 1986) and in the present case, potential end members include a juvenile, mantle-derived and a crustal component. The calculated intercept and slope of such a mixing line is dependent on the isotopic compositions of the end members, and not on their proportions (Faure, 1986). The mantle-derived component (J) can be approximated by the least radiogenic mafic gneiss sample (7.95; Table 5), whereas a strongly LILE-enriched upper-crust component (U) can be represented by the average of nine metasediment samples from the region analysed by Knudsen et al., (1997b). The Rb-Sr systematics of the local deep continental crust (L) is constrained by data on granitoid intrusions (Andersen, 1997; Simonsen, 1997); it is distinctly different from the metasediments by a less extreme LILE-enriched composition, but does not represent a typical LILE-depleted rock from the lower continental crust. Mixing of the mantle-derived component with either of the two crustal end members results in present-day Rb-Sr correlation lines (Fig. 8) with slopes and intercepts indistinguishable from the 1480 Ma correlation line based on the present data and the 1·54 Ga isochron of
Field & Råheim, (1979). The present data suggest that the precursor of the Tromøy gneisses formed at ~1200 Ma by mixing of a juvenile, mantle-derived component with material having a prolonged prehistory in an LILE-enriched crustal reservoir, but the Rb-Sr data can neither characterize this crustal end member in more detail nor prove that it originated from the Baltic Shield. As the whole-rock system of the Tromøy complex was not initially homogeneous in strontium isotopic composition, any linear correlation reported is likely to be devoid of chronological significance, and ages around 1500 Ma reflect the composition and history of the end members rather than the emplacement age of the magmatic protolith. Figure
Table 5. Isotopic characteristics of calc-alkaline tonalitic gneiss, enderbitic intrusive veins and mafic granulites from Tromøy
Sm-Nd The Tromøy gneiss complex shows a wide range of present-day Nd isotopic compositions (143Nd/144Nd from 0·51207 to 0·51286 and 147Sm/144Nd from 0·0981 to 0·2789; Table 6). However, three samples of trondhjemite and mafic gneiss with 147Sm/144Nd > CHUR (Chondrite Uniform Reservoir) have experienced Sm-Nd differentiation because of garnet crystallization at 1100 Ma, and are not further considered. Furthermore, tonalite samples 19.95Tro and 10Ma show excessively high 143Nd/144Nd at 1·2 Ga at low but reasonable 147Sm/144Nd ratios. This is due to the presence of minor metamorphic (1100 Ma) garnet in these samples, which failed to dissolve completely during analysis. The high [epsilon]Nd at 1200 Ma of these samples is thus an analytical artefact, and Nd data for samples 10Ma and 19.95 are not considered further. Table 6. Isotopic characteristics of the endmembers of the two-component Rb-Sr isotope mixing calculations
Most mafic and tonalitic gneiss samples have [epsilon]Nd(1200) values in the range -2 to 6 and a negative evolution of [epsilon]Nd with time to the present (Fig. 9), reflecting their LREE-enriched REE patterns. [epsilon]Nd(1200) values close to the depleted mantle curve indicate the presence of a depleted mantle derived component in the Tromøy complex. The trondhjemites have [epsilon]Nd(1100) values that overlap with the values of the mafic and tonalitic gneisses, and this suggests that the rocks may be genetically related. Despite the low La/Yb ratio of the trondhjemite, the distinct concave-upwards curvature of the REE patterns causes low Sm/Nd ratios, and thus a steep trend of [epsilon]Nd with time (Fig. 9). Figure
In a plot of [epsilon]Nd vs initial 87Sr/86Sr at 1200 Ma (Fig. 10a), most samples of mafic and tonalitic gneiss plot at or close to a binary mixing curve between a mafic component derived from a depleted mantle reservoir (DePaolo, 1981) and a crustal component, at ~10-50% mantle contribution. Only one sample of tonalitic gneiss indicates a much higher crustal contribution. The crustal component is poorly defined from the Sr-Nd data. The component shown in the figure represents the deep crust in the SW Baltic Shield (Andersen, 1997; Simonsen, 1997), but serves as an example only, as other, more strongly Rb-enriched reservoirs would account equally well for the variation in the present data. At 1100 Ma, the majority of trondhjemite samples and the hornblendite plot well within the range of variation of mafic and tonalitic gneisses, strongly suggesting that the older lithologies of the Tromøy complex were involved in the genesis of the trondhjemitic melts. Figure
Lead Lead isotope data (Table 5) for 19 samples of tonalite, mafic gneiss, trondhjemite and late mafic dykes are plotted in Fig. 11, together with data for relevant global reservoirs at 1200 Ma. Both mafic and tonalitic gneisses span a considerable range of lead isotope compositions (206Pb/204Pb from 16·26 to 20·33), with two samples of mafic gneiss defining the unradiogenic end of the array. Six out of seven of the trondhjemite samples analysed show limited variation of Pb composition, with 206Pb/204Pb in the range 17·60-19·95. The late mafic dykes overlap with the tonalitic gneiss, indicating that their U-Th-Pb systematics are completely controlled by contamination with their wallrocks. 207Pb/204Pb is moderately well correlated with 206Pb/204Pb, giving rise to a positively inclined array in the 207Pb/204Pb vs 206Pb/204Pb diagram (Fig. 11). Regression of all lithologies together yields a poorly fitted regression line [mean square weighted deviation (MSWD) = 13] with a spurious age of 1703 ± 290 Ma. Such a distribution of present-day lead compositions suggests a special case of two-component mixing, in which the lead isotopic compositions and U/Pb ratios of rock volumes intermediate between the two end members are positively correlated at the time of mixing (Whitehouse, 1989; Romer & Bridgewater, 1997). Accumulation of radiogenic lead since closure of the system at ~1200 Ma has led to a clockwise rotation of the mixing line, but also to increased scatter around this line. Figure
The U-Th-Pb systematics of the mantle beneath the southwestern part of the Baltic Shield can be described by a single-stage 238U/204Pb ratio in the range 7·90-7·96 (Andersen et al., 1994; Andersen, 1997). The resulting mantle composition at 1200 Ma is slightly less radiogenic than the theoretical mantle composition of
Zartman & Doe, (1981), but as the two models are nearly collinear along a 1200 Ma isochron, the difference between them is insignificant for the present discussion; 1200 Ma mantle-derived rocks would today plot on the line marked `100% mantle' in Fig. 11. The lead isotope evolution of the upper continental crust in South Norway is comparatively well known from studies on metasediments and their protoliths, and common to most metasediments in the area are high present-day 238U/204Pb ratios and a pre-Sveconorwegian crustal history in a reservoir with elevated 238U/204Pb (Andersen & Munz, 1995; Andersen et al., 1995; Knudsen et al., 1997b). Neodymium isotope systematics on metasediments and SIMS U-Pb dating of clastic zircon grains indicate that the crustal protolith formed at 1750-1900 Ma (Andersen et al., 1995; Knudsen et al., 1997a, 1997b); its lead isotopic composition at 1200 Ma is similar to the 1200 Ma theoretical upper continental crust end member of
Zartman & Doe, (1981). The lead isotope characteristics of the deep continental crust in southern Norway are constrained by data on Sveconorwegian granites (Andersen et al., 1994; Andersen, 1997; Simonsen, 1997). Although the continental crust in South Norway is compositionally stratified, the deep crust is not depleted in LILE (Andersen, 1997), differing significantly from the much less radiogenic global `depleted lower crust' end member of
Zartman & Doe, (1981), indicated in Fig. 11. At present, rocks formed by remobilization of SW Baltic Shield deep crust would plot on the line marked `100% deep crust' in Fig. 11. All but two of the samples of mafic and tonalitic gneiss as well as all of the trondhjemite samples plot well above the `100% deep crust' line in Fig. 11, showing that binary mixing of a mantle-derived component and the deep crust of the SW Baltic Shield cannot account for the variation in 1200 Ma initial lead composition in the Tromøy complex. To account for the elevated 207Pb/204Pb of these rocks, significant amounts of a component similar to the global `upper continental crust' or to SW Baltic Shield sediments are required. The widely accepted petrogenetic model of
Field et al., (1980), which implies a single-stage evolution of the entire Tromøy gneiss complex during a Gothian orogeny (~1600 Ma), is contradicted by the field observations, U-Pb geochronology and geochemical data of the present study. Although the tonalitic and mafic gneisses are depleted in LILE relative to average values of the upper continental crust (Taylor & McLennan, 1985), the rocks are enriched relative to N-MORB, giving the LILE/HFSE and LILE/REE patterns typical of subduction-related magmatism (McCulloch & Gamble, 1991; Hawkesworth et al., 1994). Furthermore, the mafic and tonalitic members are metaluminous, low-K rocks which have characteristic negative spikes in niobium and positive spikes in lead, are moderately enriched in middle REE (MREE)-LREE and have relatively flat MREE-HREE patterns that resemble evolved magmas in modern oceanic island arcs. The data presented suggest that the protoliths of the mafic-tonalitic gneiss association formed by differentiation of a subduction-zone related magma at ~1200 Ma. Fractionation and emplacement of the parent magma took place at pressure conditions where garnet was unstable, and with a reduced water activity (restricted amounts of hornblende), in accordance with earlier interpretations of abundant mantle-derived magmatic CO2 inclusions in the complex (Hoefs & Touret, 1975; Van den Kerkhof et al., 1994). The rocks were affected by the regional metamorphism ~100 my later, at P-T conditions of the order of 7·5 kbar, 840°C [the M2 event of
Knudsen, (1996)], causing anatexis and emplacement of trondhjemite dykes and veins with accessory, or more rarely, major amounts of garnet. The major and trace element characteristics of different petrogenetic scenarios in the Tromøy complex have been quantified from data in Table 1, using built-in multivariate linear regression tools of Microsoft Excel. Trace element distributions have been estimated for different models using standard equations of fractional crystallization and non-modal batch melting, as given for example by
Rollinson, (1993). Partition coefficents for `tonalitic' systems, as compiled by
Martin, (1987), have been used, supplemented by data on rhyolitic systems from
Rollinson, (1993) (Table 7).
Table 7. Partition coefficients used in modelling of tonalitic-trondhjemitic magmas
Tonalite and mafic gneiss The present-day mineralogy of the mafic gneiss, characterized by orthopyroxene and hornblende, reflects high-grade metamorphism rather than igneous crystallization. The mafic gneiss and tonalite have overlapping REE concentration levels and near-parallel distribution patterns, which is inconsistent with fractionation or accumulation of a mineral assemblage dominated by hornblende, or with garnet as a major phase, as this would modify the HREE levels beyond what is observed. The tonalite and mafic gneiss undoubtedly represent igneous precursors that are closely genetically related to each other, but as any original structural relationship between the two has been obliterated by later deformation, field observations cannot help identify the actual process involved. The mafic gneiss may represent a magma that was parental to tonalitic magmas, it may be a mafic cumulate from a tonalitic liquid, or the two lithologies may both be cumulates from a common parent magma (dominated by pyroxene with minor Fe-Ti oxides and hornblende, and plagioclase, respectively). Although it is possible to generate a liquid similar to average tonalite from a magma with average mafic gneiss composition by 25-30% fractionation of plagioclase, mafic silicates, apatite and iron-titanium oxides, in proportions depending on mineral compositions and on constraints on the crystallizing mineral assemblage, this does not exclude the other possible mechanisms. To preserve the parallel REE distribution patterns, hornblende and garnet, respectively, cannot exceed 10-15% and 1% of the accumulated or fractionated solid assemblage, suggesting that the rocks crystallized at pressures lower than the limit of the stability field of garnet. Ordinary trondhjemite The trondhjemite is close to minimum melt compositions in mafic-tonalitic experimental systems at the pressure conditions of the M2 granulite-facies event at Tromøy (Fig. 4b), suggesting that the older, less silicic lithologies in the complex may have acted as the source rock for trondhjemitic partial melts. The distinct positive europium anomaly and the rising HREE distribution pattern of the ordinary trondhjemite (Fig. 12a) indicate that neither plagioclase nor garnet remained in the residue after extraction of trondhjemite melt. Otherwise, these minerals would have retained enough Eu, Yb and Lu to give a negative Eu anomaly and a declining HREE distribution pattern. Figure
Simple mass-balance estimates on average compositions (derived from Table 1), indicate that the maximum yield of trondhjemitic liquid is of the order of 45-50% from a mafic gneiss protolith, and as high as 80% from an average tonalite, assuming that all plagioclase is consumed. The restite consists of 95-100% hornblende, with minor clinopyroxene and iron-titanium oxides. However, even with plagioclase completely removed from the residue, none of the lithologies observed at the present section through the Tromøy complex would be able to produce partial melts with positive europium anomalies and the overall low REE level observed (Fig. 12b). Generation of a low-REE liquid with a positive Eu anomaly requires partial melting of a protolith that itself has a moderate positive Eu anomaly, such as a plagioclase-rich cumulate. A model cumulate consisting of 50% plagioclase, 30% clinopyroxene and 20% orthopyroxene would be able to produce 35-45% of trondhjemitic liquid, depending on its plagioclase composition, and a 100% hornblende restite by partial melting. A leucogabbroic to dioritic cumulate of this character would naturally form by low-pressure fractional crystallization of a tonalitic magma. It differs only slightly from the mafic gneiss, mainly in having higher CaO + Na2O. The REE pattern of a leucogabbroic-dioritic model cumulate would mimic the mafic gneiss, but for a slight positive Eu anomaly and marginally lower LREE (Fig. 12a). Ten to 40% partial melting of a leucogabbroic-dioritic cumulate would in turn produce a liquid with a positive Eu anomaly and an LREE distribution mimicking that of the trondhjemite dykes. However, its HREE distribution would be flat, significantly underestimating the observed Yb and Lu concentrations (Fig. 12a). Accumulation of 1-3% of garnet in the trondhjemite, in agreement with the minor modal abundance of garnet, adequately reproduces the observed increase in Yb and Lu concentrations of ordinary trondhjemite. Hornblendite and garnet-rich trondhjemite The field relations and the coarse-grained texture of the hornblendite and garnet-rich trondhjemite suggest that these rocks are genetically related to the intrusive trondhjemite dykes. In silica-rich liquids, hornblende has partition coefficients for REE well above 1·0 (Martin, 1987; Rollinson, 1993). Nevertheless, the pegmatitic hornblendite in the Tromøy complex has REE concentration levels comparable with the normal trondhjemite dykes (Table 3). However, the garnet-rich trondhjemite (7/96) spatially associated with hornblendite (8/96) has distinctly lower REE concentrations than any of the other samples analysed. The extreme REE distribution patterns of these rocks are thus most probably due to late-stage differentiation processes in anatectic melts that otherwise form the ordinary trondhjemite. The coarse-grained hornblendite does not represent the (unexposed) hornblendite residue after trondhjemite formation, as this is likely to show substantially higher REE concentrations in accordance with the relatively high REE partition coefficients for hornblende (Martin, 1987; Rollinson, 1993). The present data strongly suggest that most of the geochemical characteristics of the Tromøy complex can be explained by magmatic fractionation processes and subsequent anatexis. The tonalites and mafic gneisses at the present level of exposure have generally retained their primary whole-rock REE characteristics through the ~100 my high-grade metamorphism. REE fractionation took place at mineral scale only, during growth of metamorphic garnet. These findings disagree with earlier interpretations of the Tromøy rocks as severely metasomatized or as depleted in LILE and REE by syn-metamorphic fluids during the high-grade metamorphism (Moine et al., 1972; Touret, 1985, 1996; Cameron, 1989). The present findings are also in conflict with earlier ideas about severe element mobility, which would have caused resetting of the Rb-Sr system (Field & Råheim, 1981, 1983; Weis & Demaiffe, 1983; Field et al., 1985), which inevitably has been related to high-grade metamorphism. The high-grade event affecting the coastal part of the Bamble Sector at ~1100 Ma reached P-T conditions of 7·5 ± 0·5 kbar, 840 ± 40°C (Knudsen, 1996), which corresponds to a level well within the lower continental crust. At the thermal maximum, the Tromøy complex was in a state of partial melting, generating a trondhjemiticanatectic melt. This process did not, however, involve extraction and upwards migration of hydrous, potassic and LILE-enriched granitic magmas, as envisaged by
Frost et al., (1989), and did not generate a `classical' compositional stratification in the continental crust with a `depleted' and dehydrated lower crust, as suggested by
Field et al., (1980) and
Cameron, (1989). In fact, the opposite evolution took place, as trondhjemitic melts forming below the present erosional section through the tonalite complex had lower concentrations of LILE and REE than their source rocks (Figs 5a and 11). As hornblende was an important restite phase, water was kept back in the solid residue, whereas CO2 was dissolved in the anatectic melts, to be released during crystallization of the trondhjemite intrusions. The free fluid phase in the Tromøy gneisses during the high-grade event is well characterized, consisting of carbon dioxide with a mantle [delta]13C signature (Hoefs & Touret, 1975; Van den Kerhof et al., 1994). This fluid phase is abundant also in the 1100 Ma trondhjemite dykes and veins (Knudsen & Lidwin, 1996), which suggests that the carbonic fluid was contained within the rock complex since its primary crystallization at 1200 Ma, without exchange with the country rocks or dilution by externally derived fluids. The stabilization of hornblende in the restite after generation of trondhjemite melt indicates some introduction of water-bearing fluids to deeper levels of the complex after its primary crystallization, but the mafic and tonalitic gneisses at the present level of exposure have stayed remarkably closed to fluid-induced element exchange with their surroundings during Sveconorwegian metamorphism. The reason may be found in the rheologic properties of the rather massive mafic and tonalitic gneisses and that lithologies affected by late external fluids have been systematically avoided during sampling. The tonalites and metapelites exposed at the present section through the Tromøy-Hisøy-Torungen area experienced high-grade incipient melting only [Fig. 3a and
Knudsen, (1996)], producing millimetre-wide melt pods or veinlets unable to segregate (Knudsen, 1996). This process was locally controlled by mineral reactions giving a strongly reduced water activity. A fluid composition of XH2O = 0·3 can be estimated for nearby metapelites (Knudsen, 1996), again with CO2 as an important free fluid (Van den Kerkhof et al., 1994; Knudsen & Andersen, 1997). There is textural evidence of zircon reworking in the Tromøy complex (Fig. 6), and a high number of inversely discordant zircons giving 206Pb/207Pb ages younger than 1100 Ma suggest a separate episode of U loss after 1100 Ma. This relatively late process might also have affected whole-rock U concentrations, but as calc-alkaline rocks generally have low LILE concentrations of a few parts per million (i.e. McCulloch & Gamble, 1991; Hawkesworth et al., 1994), and a majority of the samples have concentrations at or below the XRF detection limit, this cannot be evaluated from the present data. A study of element mobility in the metasediments and mafic granulites of the Bamble Sector (Knudsen et al., 1997b) has demonstrated that the surrounding area experienced a large-scale U loss, which may be a parallel to the process observed in the Tromøy complex. Only geochemical parameters can give any indications of the 1200 Ma tectonic setting of the Tromøy gneisses, as the present-day mineralogy and field relations reflect the ~1100 Ma metamorphic overprint. The major and trace element compositions of the mafic and tonalitic gneisses suggest that the protoliths were similar in composition to present-day magmatic rocks formed at destructive plate margins. The observed low-K series is most commonly associated with immature island arcs and is much less frequent in mature arcs or continental margin settings (e.g. Hess, 1989; Smith et al., 1997). At modern destructive plate margins, magmas are generated by partial melting within a hydrated and metasomatized mantle wedge above a subducted slab (Barker, 1979) and primitive melts may undergo fractional crystallization in sub-arc magma chambers (e.g. Smith et al., 1997). A crustal signature in oceanic island arc magmas is commonly interpreted as the result of metasomatic enrichment of the mantle wedge in LILE-enriched components with a crustal prehistory, derived from subducted sediments (e.g. McCulloch & Gamble, 1991; Pearce & Parkinson, 1993; Hawkesworth et al., 1994). In a magmatic arc underlain by continental crust, mantle-derived magmas can easily be contaminated by assimilation during ascent (Hildreth & Moorbath, 1988; Barnes et al., 1995, , 1996; Galán et al., 1996; Mason et al., 1996). The magma(s) forming the Tromøy igneous protolith contained components derived from a depleted mantle and an evolved, LILE-enriched upper-crustal component, but the present lead isotope data indicate that the deep continental crust of the SW part of the Baltic Shield was not involved in the petrogenesis of these rocks. Metapelitic lithologies make up an important constituent of the crust surronding the Tromøy complex and represent a possible continental crust source. However, the effects of contaminating a mantle-derived magma with (1) high-K granitic melts derived from deep-seated metapelites, (2) aqueous fluids or (3) the metapelites are not observed. These processes generally produce negative [epsilon]Ndi values (Hildreth & Moorbath, 1988; Barnes et al., 1995, , 1996; Galán et al., 1996; Mason et al., 1996) and medium- to high-K rocks, which are not observed. The high to extreme K/Rb ratios observed in the trondhjemites do not favour a massive influx of aqueous fluids or brines derived by dehydration of metapelitic lithologies at this stage, as such fluids would be expected to carry significant amounts of Rb into the Tromøy complex, lowering the K/Rb ratio of the anatectic melts. The conspicuous absence of sediment-derived, pre-1360 Ma inherited zircons from the samples of tonalitic and mafic gneiss analysed in the present study adds to the evidence against direct interaction of Tromøy magmas with their present country-rocks during emplacement at 1200 Ma. The absence of a discernible local crustal input to the magmas suggests that the protoliths of the mafic and tonalitic gneisses were emplaced as part of an island arc system, somewhere off the margin of the Baltic Shield. The present geochemical evidence indicates that the Tromøy gneisses represent a metamorphosed and deformed fragment of an ~1200 Ma low-K igneous complex, which probably formed in an island arc setting. If this is the case, there must be a significant tectonic break between the Tromøy area and adjacent parts of the Bamble Sector. Tonalitic rocks metamorphosed in the amphibolite facies form a discontinuous belt along most of the south coast of Norway southwest of Tromøy (Starmer, 1987), but unlike at Tromøy, trondhjemitic intrusions are not observed in these medium-grade gneisses (authors' unpublished field observations). In the granulite-facies area immediately southwest of Tromøy island, no tonalitic gneisses are exposed at the present surface, but the metasediments and an early Sveconorwegian gabbro are crosscut by trondhjemitic dyke intrusions indistinguishable from those at Tromøy (Knudsen & Lidwin, 1996), which indicates the presence of Tromøy-like gneisses at depth at ~1100 Ma in this area as well. These observations suggest that a substantial amount of Tromøy-type juvenile crust may have formed south of the present coast of Norway at ~1200 Ma. The 1100 Ma Sveconorwegian medium- to high-grade event recognized in the Bamble Sector including Tromøy (Kullerud & Machado, 1991; Kullerud & Dahlgren, 1993) appears to be specific for this area. It pre-dates the main Sveconorwegian orogeny recognized in the rest of South Norway and Southwest Sweden by ~100 my (Johansson et al., 1991; Dahlgren, 1996; Bingen & Van Breemen, 1998), but this has remained unexplained. We suggest that the 1100 Ma collisional event represents a likely time for the accretion of the Tromøy island arc system onto the southwestern margin of the Baltic Shield. The Tromøy calc-alkaline gneisses have retained many features of their magmatic precursors despite the fact that the rocks have suffered a granulite-facies metamorphism. A suite of tonalites, mafic rocks (of basaltic-trachyandesitic compositions) and unpreserved or unexposed plagioclase-rich, pyroxene-bearing cumulates (leucogabbro-diorite) represent remnants of an island arc formed south of the present coast of southernmost Norway at ~1200 Ma. The tonalites probably formed by fractional crystallization of a mafic parent magma similar in composition to the mafic gneiss, with plagioclase, pyroxene, magnetite, biotite and apatite as crystallizing phases; alternatively, both rock types are differentiates from a common mafic parent magma. Fractionation and emplacement of the parent magma took place at pressure conditions where garnet was unstable, and with a reduced water activity (restricted amounts of hornblende), in accordance with earlier observations of abundant mantle-derived magmatic CO2 inclusions in the complex. Accretion of the arc fragment onto the Baltic Shield continental margin took place at medium- to high-grade metamorphic conditions, during the early phase of Sveconorwegian metamorphism at 1100 Ma. In this process, trondhjemitic melts formed by anatexis of a plagioclase-rich facies of the complex, some of which aggregated and intruded as dykes in the older lithologies. The trondhjemites represent 10-45% melting of plagioclase-rich cumulates, leaving a hornblende residue behind. Late-stage differentiation processes of the trondhjemite melt caused formation of rare garnet-rich trondhjemite and hornblendite with extremely low REE concentrations. Rb-Sr and Pb-Pb correlation lines with apparent `ages' of 1703 ± 290 Ma and 1480 Ma pre-date SIMS U-Pb ages of magmatic zircons by several hundred million years. Sr, Nd and Pb isotope data indicate mixing between a mantle-derived component and an LILE-enriched, upper-crustal component. The Pb-isotopic signature of the crustal component is distinctly different from the lower continental crust present in the southwestern part of the Baltic Shield prior to 1100 Ma, and the data suggest that the LILE-enriched crustal component was introduced from subducted sediments into the source of magmas in a mantle wedge. The present LILE, HFSE and REE concentrations of the tonalites and mafic gneisses are connected to the primary crystallization conditions and not the results of element mobility during the high-grade metamorphism. However, an event of U loss occurred after the high-grade metamorphism at 1100 Ma and caused zircon reworking along fluid-induced channels and produced negatively discordant magmatic and metamorphic zircon grains. The thorough and constructive reviews of Ken Johnson and an anonymous reviewer are gratefully acknowledged. We are greatly indebted to several persons in the NORDSIM laboratory: Martin Whitehouse for assistance with the SIMS data reduction, Torbjörn Sunde for assistance during zircon analyses and Jessica Vestin for zircon mounting. Turid Winje kindly assisted with the BSE imaging of the zircons, and Gunnborg Bye-Fjeld and Toril Enger assisted during sample preparation. The advice and comments of Else-Ragnhild Neumann are appreciated. The partners of the NORDSIM consortium, in particular the Norwegian Reseach Council, are thanked for making this work possible. The analytical work was supported by grants from the Nansen Foundation to T.-L. Knudsen and from the Norwegian Reseach Council to T. Andersen.INTRODUCTION
Geological setting
Field relations and petrography
GEOCHEMISTRY AND GEOCHRONOLOGY
Analytical procedures
Major and trace elements
U-Pb geochronology
Whole-rock radiogenic isotope systematics
DISCUSSION
Magmatic differentiation and anatexis in the Tromøy complex
Element mobility during 1100 Ma high-grade metamorphism
Tectonic setting of the 1200 Ma Tromøy magmatism
Regional significance
CONCLUSIONS
ACKNOWLEDGEMENTS
REFERENCES
