Seve terranes of the Kebnekaise Mts., Swedish Caledonides, and their amalgamation, accretion and affinity

Abstract

A major allochthon of the Scandinavian Caledonides, the Seve belt has traditionally been considered to be derived from the rifted margin and continent-ocean transition (COT) of Baltica. However, geochronological results obtained from its inferred northern equivalent, the Kalak Nappe Complex (KNC), have been taken to indicate an exotic affinity of this complex and of also Seve terranes, an interpretation adopted in recent palaeogeographic models. In the Kebnekaise Mts., the COT is represented by the Kebnekaise terrane composed of amphibolitized dykes of gabbro and dolerite of depleted magma source and rare felsic and ultramafic rocks. Coronitic dolerite and gabbro with abundant rutile suggest high pressures before or during amalgamation with the underlying Mårma terrane, composed of quartzofeldspathic gneisses intruded by mafic and granitic rocks, the latter including a previously dated c. 845 Ma-old granite. The granite mingled with mildly alkaline dolerites chemically similar to transitional basalts of continental rifts. Following emplacement at shallow (andalusite stability) crustal levels, the igneous complex and host rocks underwent extensive deformation, metamorphism within the sillimanite-kyanite stability field and local migmatization. The Kebnekaise and Mårma terranes amalgamated in early Ordovician as indicated by the U-Pb age of 487±7 Ma obtained from titanite fabrics of deformed granite in thrust vicinity, and by ⁴⁰Ar-³⁹Ar results. The pressure increase in both terranes suggests that amalgamation occurred during initial subduction and imbrication. Correlation of the Mårma terrane with the KNC is discussed. Results obtained in this study give no reason to ascribe an exotic affinity to the Seve terranes of the Kebnekaise Mts.

Introduction

The Scandinavian Caledonides are built up by terranes derived from the pre-Caledonian margin of Baltica, its early Ediacaran rift basins and continent–ocean transition, and from marginal basins and various exotic systems of the Iapetus Ocean and its opposite Laurentian margin. Identification of indigenous terranes is fundamental to understanding the pre-collisional palaeogeography and the tectonic evolution of a mountain belt. A major allochthon of the Scandinavian Caledonides, the 1000-km-long Seve belt has ever since the reinterpretation of the Scandinavian Caledonides in terms of plate tectonics in the 1970s, as well as in recent works (Gee et al. 2014, 2015, 2017; Be`eri-Shlevin et al. 2011), been considered derived from the rifted margin of Baltica. However, geochronological results obtained from the belt′s equivalent continuation into Finnmark, the Kalak Nappe Complex (Fig. 1A), have been taken to indicate an exotic ancestry of this complex and, as a consequence of previous correlations, of also the Seve terranes (Kirkland et al. 2006, 2007; Corfu et al. 2007; Roberts et al. 2006). The disagreement can partly be ascribed to insufficient knowledge about lithologies and tectonostratigraphy within critical segments of the Seve belt. One such segment is the Kebnekaise area, the eastern part of which remained “poorly investigated bedrock” (Kulling 1964) until visited by M Sc students of Lund University during 1997–2001. Whereas previous work has been restricted to small areas around Tarfala and the Singis window, their mapping included also areas within central and eastern parts of the Kebnekaise Mts. A preliminary map compiled by the senior author and photos of rocks were put at the disposal of the Geological Survey of Sweden in connection with compilation of the map The Caledonides in northern Sweden (Thelander 2009a, 2009b). The unpublished results presented here have bearing upon the derivation of Seve terranes and Seve–Kalak relations, dealing with the tectonic setting of magmatism and the timing of amalgamation and accretion of the terranes. The compilation is also intended to serve as a basis for future studies, in particular concerning results where supporting evidence by quantitative studies is needed. The senior author compiled the map, petrological, chemical and structural data and wrote the text. He is solely responsible for inherent reinterpretations, and misinterpretations, if any, of the original M Sc texts.

Pending an adequate definition of the name ‘Seve nappe complex’, allowing its application on also northern segments of the Seve belt, we apply terrane nomenclature, which also addresses the juxtaposition of nappes (amalgamation) before their final emplacement (accretion). The Seve allochthon was treated as a suspect terrane by Stephens & Gee (1985, 1989) and Roberts (1988). It is suspect until proven (‘guilty of being’) either exotic, as proposed recently by Kirkland et al. (2006, 2007), Corfu et al. (2007) and Roberts et al. (2006), or indigenous, i.e., derived from Baltica. Andréasson (1996) grouped all terranes characterized by inferred Baltoscandian rift basin fill and magmatism (but scattered by numerous local names and by semantics) into the ‘Seve Superterrane’. This concept was later proposed to include also the Kalak Nappe Complex (KNC) of north Norway: ‘Seve-Kalak Superterrane’ (Andréasson et al. 1998). Subsequent work soon disclosed such wholesale correlation of Seve and Kalak units as premature. In the present study, Seve terranes are treated as units of the Middle Allochthon (Fig. 1A; Gee et al. 2008; Andréasson & Gee 2008). Focussing on Seve terranes, the paper treats very briefly underlying tectonostratigraphic units and the overlying Virisen terrane. The map (Fig. 1B) is based on mapping of selected areas by the students, and reconnaissance traverses made by the senior author connecting these areas to the gross tectonic stratigraphy of the Kebnekaise Mts. and the adjacent southern Abisko Mts. However, many valleys and peaks remain to be visited, in particular in northern parts of the mountains. Geographic coordinates used in the text refer to the WGS84 (decimals) system. Pending approval of new stratigraphic names by the Swedish Committee for Geological Nomenclature, lower case letters are used except in acronyms.

Autochthon, Parautochthon and Lower Allochthon

Precambrian autochthonous rocks, including a granodiorite dated at 2418 + 25/–15 Ma (Witschard et al. 2004), are exposed in the bottom of the Vistasvággi valley. Along the lower slopes of the valley, a marked, steep scarp of mylonitic and ultramylonitic rocks of the overridden autochthonous–parautochthonous Ediacaran–Cambrian sedimentary platform cover (Dividal Group; Føyn 1985) and the Parautochthon separate the autochthonous basement from the nappe pile. North-east of the Tjäktja mountain lodge, thin beds of conglomerate and quartzite cover Precambrian rocks of the Rombak–Sjangeli window, which is imbricate according to Bax (1989). Nappes derived from the inner telescoped margin of Baltica (Lower Allochthon) include mylonitic syenitic, granitic and mafic rocks (the latter as greenschists) intercalated by quartzite and grey or graphitic phyllite, reflecting the lithology of the underlying Autochthon–Parautotchthon. Locally, as in the Tarfalavagge gorge and in the Singis window, the dark blue quartzite, alum shale and grey phyllite typical of the Dividal Group are readily identified (Andréasson & Gee 1989).

Middle Allochthon below the Seve terranes

The Laddjuvagge nappe is composed of a lower sheet of mylonitic syenite and meta-arkose (Lower psammitic gneiss of Nilsson 1992), the latter often mylonitic and in places hosting dykes or boudins of metadolerite and an upper sheet of amphibolites (Lower Tarfala amphibolite of Nilsson 1992). In the upper part of the Tarfala valley, slices of a deformed rock with an assemblage typical of shear zones within granitoids of the Middle Allochthon (i.e., clasts of perthitic K-feldspar, green biotite, epidote and titanite) and with syenitic chemistry occurs intercalated with psammitic gneiss hosting metadolerites. In the Alesätno valley, the mylonitic and intensely folded bedrock of the Abisko Nappe (Kulling 1964) includes felsic and mafic crystalline rocks, quartzites, grey or graphitic phyllite and white-weathering, grey dolomitic marble, the latter as c. 10-m-thick sheets on two tectonic levels; elsewhere, also as numerous thin, intensely folded bands.

Seve terranes

Seve terranes include the Mårma terrane dominating the eastern part of the Kebnekaise Mts., and the overlying Kebnekaise terrane of the central massif. The former terrane is exposed also along floors and slopes of valleys dissecting the central massif. The terrane boundary is mostly located at c. 1200–1400 m.a.s.l. Both terranes thin rapidly westwards, the upper one from c. 1000 m at Tarfala to nil along the Tjäktjavaggi valley.

Mårma terrane

Units in the Mårma terrane include mafic and felsic intrusive rocks (Vassacorru igneous complex, VIC) and amphibolites (Vierručohkka amphibolite, VA) hosted by often mylonitic quartzofeldspathic and semipelitic gneisses (Leavasvággi gneiss). The gneisses correspond to the ‘gneissic rocks and porphyroblastic schists’ on Kulling’s (1964) map. Rare areas of low-strain preserve arkosic and subordinate calcareous sedimentary protoliths of the gneisses in the wall-rock of VIC intrusions. Porphyroblasts of andalusite were completely replaced by rozettes of sillimanite, and rimmed by idioblastic garnets (Fig. 2A, B), indicating a pressure increase following upon high-temperature metamorphism. Xenoliths of calcareous composition contain skarn, composed of wollastonite in addition to garnet and diopside, providing further evidence of initial high-temperature metamorphism of the Leavasvággi gneiss.

Figure 1. A . Regional tectonostratigraphic setting of terranes of the Kebnekaise Mts. Numbers are places referred to in the text. From Gee et al. (1985), Solli & Nordgulen (2008), Gee et al. (2014). B. Tectonostratigraphy of the Kebnekaise Mts. based on mapping of selected areas by the authors and reconnaissance traverses by the first author; previous mapping (Andréasson & Gee 1989a, 1989b; Tomas 1991; Page 1993) and regional maps (Kulling 1964; Witschard et al. 2004; Thelander 2009a).

Figure 2. A. Plane polarized light view of andalusite (chiastolite) porphyroblast which is replaced by radial bundles of prismatic sillimanite, and rimmed by idioblastic garnets (G). Leavasvággi gneiss, (N68.061303°/E18.943689°). B. Close-up of Figure A (frame) under crossed nicols. C. Leavasvággi gneiss with typical red-brown colour, and an extremely thin band of metabasite. East of Moarhmma glacier (N68.081593°/E18.734316°). D. Leucosomes wrapping disrupted metadolerite boudin. Eastern slope of Godučohkka (N68.148153°/E18.666002°).

Leavasvággi gneiss

The predominating gneiss is mylonitic, red-brown, rich in biotite and garnets, and often hosting bands or boudins of metadolerite (Fig. 2C, D) and lenses of calc-silicate gneiss (cf. Fig. 13 in Thelander 2009b). The less deformed psammitic gneiss preserves quartz-feldspathic and pelitic (red biotite) bands, and garnets with inclusions of quartz and biotite which are much coarser as compared to the matrix. In more pelitic varieties, kyanite and sillimanite occur in the fine-grained matrix deflected around garnets, but also as inclusions within the same garnets. With increasing strain, a matrix foliation developed, defined by quartz ribbons, fine-grained biotite and some muscovite wrapping around abraded grains of feldspar and garnet. Muscovite fish overgrown by nests of randomly orientated kyanite prisms suggest that the deformational event resulted in a pressure increase (cf. below). Undulatory extinction of quartz in ribbons and pressure shadows indicate that deformation continued also at lower metamorphic grades.

The metabasite boudins hosted by the gneiss vary strongly with regard to recrystallization (Supplementary material, Table 8), from those preserving subophitic texture to those coloured red by garnet (Fig. 2D) and lacking plagioclase. A metadolerite boudin (c. N67.921358°/E18.623814°) shows evidence of a pressure increase in terms of coronitic growth of garnet and 1.5 wt.% Na₂O of the texturally igneous, pale-green clinopyroxene.

Narrow zones of stromatic migmatite (Fig. 3A, B) are found at different tectonic levels within the Mårma terrane. The stromatic structure is parallel to the internal pervasive foliation of the gneiss indicating strain-enhanced segregation and migration of melt (cf. Brown, 1994 for review). Fig. 3A (right half) shows the S > L tectonite fabric of the migmatite. In addition to neosomes, well-defined bands of equigranular granite formed (Fig. 3B). Melt migrated also into low-pressure sites, such as necks between metabasite boudins (Fig. 2D).

Figure 3. A. Stromatic migmatite at Mårma glacier (68.078303°/18.729586°). Right half of Figure A shows the L>S tectonite character of the migmatite: upper left specimen with fresh parting along the pervasive foliation. Upper right: section cut normal to foliation. Bottom: section cut normal to lineation. B. Beneath knife: distinct, fine-grained granitic dyke, rich in small garnets. C. The fine-grained domain is composed of quartz, feldspar and biotite and overgrown by abundant small garnets. Rims of leucosomes contain peritectic garnets (at arrows). Palaeosomes include garnet and sillimanite in addition to biotite. Leavasvággi gneiss, northern slope of Alesätno valley (N68.199906°/E18.790317°). Spec. A90–20. D. Plane polarized light view of garnet porphyroclasts (within white lines) which preserves prism (2-mm long) and raft of sillimanite (S; within black lines), and dark brown biotite flakes with preferred orientation (arrows). Muscovite (Ms) occurs as large and numerous small fish. 1: Feldspar porphyroclasts. 2: Bands of mosaic quartz. Dark bands consist of very fine-grained biotite. SG at the type locality (N67.907414°/E18.591703°). Long side of photo is 3.5 mm. Spec. A87–256.

Figure 4. Vássačorru igneous complex. A. Typical net-veined outcrop, cut by late dyke. At Mårma glacier (c. N68.076431°/E18.711872°). B. Mingled mafic rocks and Vistas Granite (N68.078309°/ E18.541993°). The two boulders with mafic enclaves shown in Figure C occur downstream this outcrop. C. The inset figure in the upper right corner is a close-up of folded pillows in the boulder in the background. D. Mafic magmatic fragments in an inhomogeneous hybrid rock formed by mixing of dolerite and granite (N68.068772°/ E18.935717°; same locality for Figures E–F). Varying darkness of fragments reflects different stages of hybridization. The fragment below the hammer is partly angular suggesting some consolidation; note dark, rounded enclave inside the fragment. Above hammer, a xenolith of metasedimentary rock. Inset figures show fragments charged with (upper corner) or rimmed by (lower corner) porphyroblasts of garnet. E. Xenolith of psammitic rock in hybridic granitic rock. F. Xenolith of calc-cilicate rock cut by granitic veins with pegmatitic core. Mafic fragments are rimmed by, or invaded by (right), garnets. Sparse xenocrysts of feldspar.

Figure 5. A. Mafic rocks of the Mårma terrane according to the geochemical classification method of De la Roche et al. (1980). Crosses: dolerites of the VIC (n = 9). Circles: Vierručohkka amphibolite (n = 12). Legend: 1 = picritic basalt/ ultramafic rocks. 2 = alkali basalt/alkali gabbro. 3 = olivine basalt/olivine gabbro. 4 = tholeiitic basalt/noritic gabbro. 5 = trachybasalt/syenogabbro. 6 = latitic basalt/monzogabbro. 7 = andesitic basalt/ gabbrodiorite. B. Diagram designed by Pearce (1983) to estimate the extent of crustal contamination of a basaltic magma. Pure mantle enrichment concentrates Th and Ta with the same rate and their ideal variation therefore defines a band (grey field) with the slope of unity. Th is more sensitive than Ta to upper crustal contamination, Yb is not sensitive. Therefore, basalts contaminated by subduction zone fluids or continental crust tend to plot along trends indicated by arrows 1 and 2, respectively. Arrows 3 and 4 represent within-plate enrichment and fractional crystallization, respectively. Dots: Mafic rocks of the Kebne Dyke Complex, central massif. N, E and OIB are reference compositions for N-MORB, E-MORB and oceanic island basalts, respectively (Sun & McDonough 1989). CAB: calc-alkaline basalts; THO: tholeiitic basalts.

Figure 6. Mafic rocks of the Mårma and Kebnekaise terranes; symbols as in Figure 5. A-B: Discrimination diagrams according to Meschede (1986) and Wood (1980). 1: within-plate (WP) alkali basalts; 2: WP alkali and tholeiitic basalts; 3: E-MORB; 4: WP tholeiites and volcanic arc basalts; 5: N-MORB and volcanic arc basalts; 6: tholeiites of volcanic arcs; 7: calc-alkaline arc basalts; 8: tholeiitic WP basalts. Grey field in Fig. A: Lower Seve Nappe of the central Scandes (Z. Solyom, P.-G. Andréasson and I. Johansson, unpublished results). C. Primitive mantle-normalized multielement diagram for the VIC dolerites. D. REE contents of VIC dolerites. E–F. Ditto for the Vierručohkka amphibolite. Normalizing factors for primitive mantle and chondrite from Palme & O’Neill (2004); for E-MORB from Sun & McDonough (1989).

Figure 7. Vistas Granite. A. Typical coarse-grained, slightly deformed variety (N68.077075°/E18.542358°). B. A widespread protomylonitic variety of the Vistas Granite (N68.077592°/E18.561136°). Most of the porphyroclasts are Carlsbad twins, some mantled, with their twin planes subparallel to the mylonitic foliation.

Figure 8. A. Granitoid rocks of the Mårma and Kebnekaise terranes according to the geochemical classification method of Debon & Le Fort (1983). Diamonds: Vistas Granite. Filled diamonds: granitoid dykes and sheets of the KDC. G: Gaskkasjávri granite. B–C. Classification diagrams of Whalen et al. (1987). Boundary lines divide granites derived from recycled continental crust (A-type) from those derived from sedimentary (S-type), igneous (I-type) or subducted oceanic crust, or mantle (M-type); cf. Chappell & White (1992) and Collins et al. (1982). Small grey field (L) is the Litlefjord Granite (cf. Discussion). D. Primordial mantle-normalized incompatible element diagram for the Vistas Granite. Grey line in D and E: reference composition of the upper continental crust (Rudnick & Gao 2003). E. REE variation diagram for the Vistas Granite. F. Primordial mantle-normalized incompatible-element contents of granitoid dykes and sheets of the KDC. In diagrams F and G: Grey field: granodiorite and quartz-monzonite. Thin black line: granite. Fat black line: reference composition of upper continental crust (Rudnick & Gao 2003). G. REE variation diagram for granitoid dykes and sheets of the KDC. Normalizing factors for primary mantle and chondrite in all diagrams: Palme & O’Neill (2004).

Figure 9. Kebne Dyke Complex. A. Dolerite dyke chilled against a medium-grained, plagioclase-porphyritic dyke. Diameter of coin: 28 mm. (N67.926902°/E18.558531°). B-C. Normal appearance of the KDC. The deformed and amphibolitized complex preserves evidence of different dyke generations. Note isoclinally folded band of felsic rock along right margin of figure C. Vaktposten (B) and Knivkammen (1878), western wall (C). D. Parallel and crossed polarized light views of ‘necklace’ rims of idioblastic garnets growing along the interface of clinopyroxene (p) and plagioclase (white). r: rutile. a: amphibole. Quartz is seen (crossed nicols) as small, white spots. Height of figure is 1.5 mm. Spec. A87–258 g. (N67.909292°/E18.576192°). E. Foliated felsic dyke of the KDC. (N68.007581°/E18.315936°).

Figure 10. A. Mafic rocks of the KDC according to the geochemical classification method of De la Roche et al. (1980). Black dots: samples from the central massif (n = 7). Grey dots: samples from the Tarfala area (n = 17). Legend: 1 = picritic basalt/ ultramafic rocks. 2 = alkali basalt/alkali gabbro. 3 = olivine basalt/olivine gabbro. 4 = tholeiitic basalt/noritic gabbro. 5–7: cf. Fig. 5A. B. Incompatible element contents of mafic rocks from the KDC (central massif) normalized to primordial mantle. Grey line: E-MORB. C. Chondrite-normalized REE contents of mafic rocks from the KDC (central massif). Grey line: E-MORB. D. Chondrite-normalized REE contents of mafic rocks from the KDC in the Tarfala area; (n = 10). Heavy black line: E-MORB. Normalizing factors for primordial mantle and chondrite from Palme & O’Neill (2004); for E-MORB reference lines from Sun & McDonough (1989).

Figure 11. Stereographic plots of foliations and trends of fold axes and lineations of Seve and sub-Seve terranes in the Kebnekaise Mts. For linear structures, planes dipping > 5° were unfolded. White dots in left bottom diagram refer to the base of the entire nappe pile. Equal-area projections: “cont.” = contour percentage intervals. A denotes structures formed mainly during amalgamation of the Mårma and Kebnekaise terranes; S mainly during Scandian accretion of the composite terrane and P mainly during postaccretion warping and faulting. 1: calculated from limbs of large-scale steep folds in central part of the KDC (cf. Fig. 12B). 2: Transport-parallel fold at the base of the KDC (cf. Fig. 12C). 3: Refolded transport-parallel folds in shear zone within the Mårma roof zone. 4: Transport-parallel fold shown in Fig. 12D. 5: Open fold in the Mårma terrane, λ c. 50 m; axis orientation calculated from 25 planes (c. N68.014567°/E18.431133°). 6: Lineation of stromatic migmatite in Fig. 3A-B. 7: Stretching lineation and rodding of leucosome in neck of metabasite boudin (c. N68.148153/E18.666002). Stereonet: Allmendinger et al. (2013); Cardozo & Allmendinger (2013).

Figure 12. A. Typical fissile foliation of amphibolite of the KDC when outcropping close below the Köli/Seve thrust. Dip: 22° W. (N67.926105°/E18.324144°). B. Hinge zone with M-fold in a kilometre-scale system of steep, NNE-trending folds in KDC of the central massif (N67.935994°/E18.486218°). C. WNW-trending, recumbent transport-parallel fold at the base of the KDC (N67.956909°/E18.498331°). D. Psammitic gneiss of the Mårma terrane folded by steeply raised, isoclinal, transport-parallel (125°) folds (N67.90733°/ 18.80035°). E. 400 m to the east of the locality shown in D, the gneiss is overlain by the Vierrutjohkka amphibolite, with isoclinal, transport-parallel (100°) folds. F-G. Non-cylindrical fold above a stack of sheath folds. Leavasvággi gneiss (N68.093349°/ E18.985121°); view towards NNE. G. Close-up of lower part of Fig. F. (P is the same point as in Figure F). The white line traces the nose of a sheath fold. Note lineation (at hammer) on the vertical face of the non-cylindrical fold.

Figure 13. A. Left figure: spaced shear zones (black strokes at horizon) in mylonitic granite at the base of the Mårma terrane (N68.077610/E18.546437). View towards south, i.e., ramp towards west. Right figure (red frame in left figure seen from above) shows the typical transverse (compass shows 90°) lineation defined by stretched quartz and K-feldspar. B. Normal faults running N–S and displaying ductile shear. View towards south. Leavasvággi gneiss, Mårma terrane (N68.117884°/E18.806408°). C. Inferred imbrication of the basal part of the VIC. South-western slope of Vássačorru, view towards north-east. D. Close-up the c. 500-m-high wall within framed section of Fig. C.

In the thinly banded, fine-grained pelitic gneiss, sillimanite occurs aligned with a strong biotite fabric. Garnet grew as aggregates of small idioblastic crystals along the biotite fabric. Quartz and K-feldspar assembled around poikilitic garnet porphyroblasts and as narrow seams parallel to the biotite. These textures and the frequent evidence of peritectic garnet in leucosomes suggest biotite dehydration melting: biotite + sillimanite + plagioclase + quartz → garnet + K-feldspar + melt. Sillimanite is a reactant of this reaction at pressures <9 kbar (Spear 1993 and references therein). There is a difference in composition between garnets (Alm₇₅Pyr_17–19Grs_3–4Sps_3–4) with sigmoidal quartz inclusions and wrapped in the biotite–sillimanite foliation and the idioblastic garnets (Alm_73–82Pyr_7–13Grs₂Sps₈) overgrowing this foliation; a variation reported from migmatitic gneisses with incipient melting (e.g., Turkina & Sukhoukov 2017; Jung et al. 1999). As a characteristic component of the foliation of deformation zones in the gneiss, muscovite occurs within the S > L fabric of the migmatite. This muscovite was sampled for ⁴⁰Ar/³⁹Ar dating (cf. below). Samples of the Storglaciären gneiss (cf. below) show, in addition to the large fish of muscovite parallel the foliation, a fine-grained intergrowth of small muscovite flakes and quartz, which may correspond to ‘late’ muscovite formed by the reaction K-feldspar + sillimanite + H₂O → muscovite + quartz; where the water was released by crystallization of stagnant melt formed during peak of migmatization (Brown 2002). The stromatic structure of the gneiss is transposed into the subhorizontal tectonic foliation of protomylonite and eventually mylonite (Fig. 14B–D), as the basal thrust of the overriding Kebnekaise terrane is approached.

Figure 14. The boundary between the Mårma and Kebnekaise terranes exposed in front of the Reida glacier at Vaktposten. Brown mylonitic Leavasvággi gneiss is overridden by amphibolites of the KDC. B–E. Progressive deformation of the gneiss towards the thrust shown in Fig. A.: stromatic migmatite of the Mårma terrane (B) passing into crenulated migmatite with subhorizontal axial planes (C) and Storglaciären gneiss (D) and, eventually, into mylonite (E). Knife in Figure C is 18-cm long.

The name Storglaciären gneiss (SG) was given to a conspiciuous mylonitic gneiss/schist characterized by strings of feldspar or quartz-feldspar clasts, coarse muscovite and large clasts of violet garnet (cf. Fig. 13 in Thelander 2009b), occurring in the Tarfala–Laddjuvagge area (Andréasson & Gee 1989a, 1989b; Tomas 1991; Nilsson 1992; Goerke 1993; ‘Mylonite Gneiss’ of Baird et al. 2014). Some tens to a hundred metres thick and hosting boudins or sheets of metabasite, the SG occurs within the Leavasvággi gneiss close below the floor thrust of the Kebnekaise terrane at 1200–1500 m. a. s. l.; in places at more than one level due to imbrication. The clasts of K-feldspar, plagioclase, quartz, garnet and kyanite are wrapped in a mylonitic foliation defined by blades of sutured or mosaic quartz, very fine-grained brown biotite, trails of kyanite prisms and sometimes bands of glassy material. Sillimanite occurs as inclusions (Fig. 3D) or growing as prisms together with kyanite in feldspar (Nilsson 1992, his Fig. 13). Muscovite occurs as fish overgrown by randomly oriented kyanite prisms. Evidence of incipient melting of a sedimentary rock is provided by small pods or streaks of arkose preserved within migmatitic gneiss (Fig. 3C). Available whole-rock chemical data of the SG (n = 3; Supplementary material, Table 4; Tomas 1991) are limited but uniform: 85.3–85.6 wt. % SiO₂; 6.4–8.3 wt. % Al₂O₃; 1.0–1.8 wt. % Fe₂O₃; 0.3–0.9 wt. % CaO; 0.5–1.3 wt. % Na₂O and 3.6–4.1 wt. % K₂O. Based on these data, the SG classifies as arkose (Herron 1988). The strings of coarse feldspar clasts and fist-size aggregates of quartz-feldspar found on some localities probably derive from coarse, stromatic migmatite formed from the gneiss (Fig. 14B), cf. further below.

Vássačorru igneous complex

The Vássačorru igneous complex (Mårma magmatic complex of Paulsson & Andréasson 2002; Mårma complex of Paulsson 1996 and Sandelin 1997) is a network of gabbroic and doleritic mafic rocks and granite (Fig. 4A, B). In addition to the main occurrence in the Kebnekaise massif, smaller bodies of the VIC occur scattered from Singis in the south (Manak gabbro; Page 1993, and Aurek gabbro; Tilke 1986) to Lake Torneträsk (Kålkuktjåkkå; N68.236058°/E19.231625°) in the north. Mafic rocks are cut by granitic dykes, some quartz-monzonitic; these are in turn cut by dolerite dykes. Narrow, very fine-grained mafic dykes with pilotaxitic plagioclase occur. Aphanitic dykes lacking chilling against country rock are interpreted as quenched liquids. Mingling and mixing between felsic and mafic magmas took place (Fig. 4). Fig. 4D shows a hybrid rock with xenoliths of metasedimentary rocks and with mafic fragments in different stages of hybridization. Fragments of all generations are rich in garnet and some are rimmed by garnet porphyroblasts and biotite. Scattered, sometimes mantled, xenocrysts of feldspar occur in both fragments and hybrid host rocks. Hybrid rocks occur also as narrow, fine-grained dykes of monzogabbroic composition. Rare evidence of igneous layering includes horizons of rusty-weathering olivine-rich gabbro, anorthositic gabbro and hornblendites. The gabbro is medium- to fine-grained, granular and composed of plagioclase, augite, hypersthene and olivine. Olivine crystals with narrow inner coronas of orthopyroxene and outer coronas of amphibole occur. Olivine-absent, orthopyroxene predominates among mafic minerals, often as hypersthene stained by iron-oxides. The dolerite is ophitic, often plagioclase porphyritic and dominated by plagioclase and augite with pigeonite exsolution lamellae. Hypersthene and inverted pigeonite are subordinate and mantled by clinopyroxene. Pyroxenes are always rimmed by hornblende. The plagioclase of some dykes is clouded by iron oxide. The clouding is confined to central parts of plagioclase laths, and the boundary to clear rims of laths is very sharp, suggesting magmatic origin (Johansson 1992). Towards the overriding Kebnekaise terrane, gabbros and dolerites of the VIC pass into amphibolites, often garnetiferous.

Chemical data of mafic rocks is restricted to nine samples of dolerites (Supplementary material, Table 1). Samples were taken from dykes with chilled margins and away from sites of mixing and mingling. Analysed samples contain 45.6–47.2 wt. % SiO₂ ( = 46.6); 2.4 – 3.4 wt. % TiO₂ ( = 2.7); 10.9–13.6 wt. % FeO* ( = 12.0) and 0.4–0.8 wt. % K₂O ( = 0.5). Mg/(Mg + Fe_tot) ranges between 0.46 and 55. The dolerite is mildly alkaline according to the norm (<5.3 wt. % nepheline) and mainly an alkali basalt according to the R1–R2 classification (Fig. 5A), but tholeiitic according to the Alkali Index (A.I.) and binary classification diagrams based on incompatible elements (not shown). Samples plot close below the boundary between ferroan (upper) and magnesian basalts (Frost & Frost 2008). The low SiO₂ and fairly high TiO₂ contents compare to those of transitional basalts of continental rifts as do the low Mg numbers and high Zr contents (average 270 ppm). According to methods used for discrimination between tectonic setting of the magma based on ternary relations of high field strength elements (Fig. 6A, B), the dolerite has within-plate tholeiitic affinity (Nb–Y–Zr-diagram of Meschede 1986) but corresponds to N-MORB according to the Hf/3-Ta–Th diagram (Wood 1980). However, the primary mantle normalized multielement diagram (Fig. 6C) shows enrichment compared to E-MORB, which is even more evident from the almost flat (La_ch/Yb_ch = 1.1–1.8) REE patterns (Fig. 6D). The conspicuously high values of HREE indicate derivation from levels above the garnet lherzolite source region of the mantle.

Table 1. U–Pb analytical data of titanite from the Gaskkasjávri granite. (Sample 94001, Laboratory for Isoptope Geology, Swedish Museum of Natural History, Stockholm. Analyst: Hans Schöberg).

Fraction			Concentrations			Radiogenic lead^c			Atomic ratios^a^,^b			Model ages^c			Error^c
No.	Size (μm)	Weight (mg)	U	Pbrad	Pbinit	^206Pb	(at.%) ^207Pb	^208Pb	^206Pb/^204U	^206Pb/^238U	^207Pb/^235U	^206Pb/^238U	(Ma) ^207Pb/^235U	^207Pb/^206U	Correlation
1	<74	0.335	299	30	0.51	85.45	5.57	8.98	2900	0.10044 ± 15	0.9032 ± 34	617	653	781 ± 07	0.48742
2	74–100	0.185	238	33	0.12	85.21	6.23	8.56	2900	0.13557 ± 23	1.3667 ± 49	820	875	1017 ± 06	0.55135
3	>100	0.042	123	21	0.03	83.03	6.42	10.55	1700	0.16774 ± 50	1.7874 ± 91	820	875	1129 ± 08	0.62802
4	<74^d	0.189	310	33	0.45	83.06	5.48	11.46	2900	0.10359 ± 17	0.9424 ± 70	635	674	806 ± 14	0.52518
5	74–100^d	0.059	261	34	0.56	81.69	5.81	12.50	1500	0.12345 ± 27	1.2114 ± 92	750	806	962 ± 14	0.46104
Corrections
Discrimination	U: 0.12%/AMU	Pb: 0.12%/AMU
Blank	U: 20 pg	Pb: 30 pg
Initial lead	^206Pb/^204Pb: 17.9	^207Pb/^204Pb: 15.6 ^208Pb/^204Pb: 37.7 (growth curve according to Stacey & Kramers (1975)). growth curve).

^{a 206}Pb/²⁰⁴Pb as measured, other atomic ratios corrected for mass discrimination, blank and initial lead

^b Errors are given as least significant digits at the 95% confidence level.

^c Data reduction and age calculation according to Ludwig (1991), using the decay constants recommended by the IUGS Subcommission on Geochronology (Steiger and Jäger (1977)) and correction terms given below.

^d Abraded

With regard to the intrusive setting of the mafic rocks of the Seve terranes, contamination with rift system wall-rock or basin fill may have modified the primary magma composition. In the following, the role of contamination and fractionation, if any, on samples used in this study is tested by an often used method shown in Fig. 6B (provided Th and Ta have been analysed) and also by some other criteria. With regard to the VIC dolerites, these extend between the N-and E-MORB reference compositions within the mantle enrichment array and accordingly indicate neither contamination nor fractionation.

Vierručohkka amphibolite

In the Mårma–Vierručohkka area, a pervasively foliated garnet amphibolite underlies the Leavasvággi gneiss, e.g., on the northern slope of peak 1667 (N68.103889°/E18.741111°), on the western slope of peak 1651 and on the south-western slope of the glacier cirques on Rassepautastjåkkå (c. N68.059722°/E18.866667°). It overlies the VIC at, e.g., Sarvesbakti (N68.053611°/ E19.028889°01). In the south, the 200–500-m-thick amphibolites on Savučhokkha (south of the map in Fig. 1B) and Skárttoaivi (east of Singis) probably belong to the same unit. On the latter locality, the amphibolite is separated from the overlying Kebnekaise terrane by the SG. In the following, we refer to this amphibolite as the Vierručohkka amphibolite. Its extension into the western Mårma region is not known. A previous interpretation of the VA as belonging to the mafic complex of the Kebnekaise terrane (Paulsson & Andréasson 2002) arrived in conflict with findings during subsequent field and laboratory studies. Interbanded grey marble occurs, and decimetre-thick bands of epidote are common in some areas. Petrography reflects thorough recrystallization: igneous minerals are completely replaced by blades of hornblende with interstitial mush of plagioclase and quartz, clinozoisite, titanite and poikiloblasts of garnet.

Analysed samples of the Vierručohkka amphibolite contain 46.4–49.5 wt. % SiO₂ ( = 47.8); 1.7 – 2.4 wt. % TiO₂ ( = 2.1); 11.5–14.2 wt. % FeO* ( = 12.8) and 0.09–0.25 wt. % K₂O ( = 0.21). Mg/(Mg + Fe_tot) ranges between 0.45 and 0.54 (Supplementary material, Table 2). Major element contents classify the protolith as olivine basalt (Fig. 5A). The analyses plot between N-MORB and E-MORB along the mantle enrichment array of Fig. 5B, i.e., showing no evidence of contamination, nor fractionation, and as a well-defined cluster within the N-MORB field of discrimination diagrams for tectonic settings (Fig. 6A, B). However, incompatible and RE element variation contents (Fig. 6E, F) demonstrate the same enrichment compared to E-MORB, and high HREE compositions as the VIC dolerites.

Vistas Granite (Paulsson & Andréasson 2002)

The granite occurs as a large body in the Vássačorru mountain and as slices along the upper Vistasvággi valley, on Rassepautastjåkka and Suorrivarri, and has been observed outside the map area as far north as Lake Torneträsk (Kålkuktjåkkå; N68.238906°/E19.238736°). Mixing with mafic rocks of the VIC resulted in variable composition and appearance. The predominating rock is grey, medium- to coarse-grained biotite granite (Fig. 7A) with phenocrysts of orthoclase, often mesoperthitic, and smaller discrete grains of plagioclase and microcline. A variety is medium-grained with pinkish K-feldspar. In addition to accessory minerals (ilmenite, magnetite, apatite, zircon), secondary titanite and garnet grew on the biotite fabrics of deformed granite. Aplitic and palingenetic dykes, often zoned, accompany the Vistas Granite. With increasing deformation, the granite passes into mylonitic varieties with K-feldspar crystals mantled by narrow rims composed of discrete grains of albite-oligoclase, or recrystallized grains. Plagioclase clasts often display bent and tapered twins. Eventually, a characteristic mylonitic gneiss (matrix > 0.05 mm) formed (‘Boginjira mylonite’ of Lundgren 2002), with dark bluish matrix (Fig. 7B) and spaced porphyroclasts of Carlsbad twins with their twin planes subparallel to the mylonite foliation. Rims of myrmekitic intergrowth of K-feldspar and plagioclase at crystal margins exposed to highest strain indicate temperatures of 500–600 °C (Harlov & Wirth 2000). This augen gneiss is widespread in the upper Vistasvággi valley and Mårma area. Along deformed contacts between granite and gabbro, a conspicuous dark gneiss with white stripes and abundant garnet formed. Along the western margin of the main body at Vássačorru, the Vistas Granite passes into a grey gneissic rock with occasional thin metabasite bands (Nallo gneiss of Boman 2001; Lundgren 2002). Biotite, muscovite, titanite, epidote and garnet define a spaced fabric.

Analysed samples of the Vistas Granite contain 68–74 wt. % SiO₂; 2.6–3.2 wt.% Na₂O and 3.3–4.8 wt.% K₂O (Supplementary material Table 3; n = 8). Standard methods, as those proposed by Debon & Le Fort (1983; Fig. 8A) and De LaRoche et al. (1980; not shown), classify the rock as quartz-monzonitic to granitic. The granite is ferroan according to the FeO_tot/(FeO_tot + MgO) versus SiO₂ ratio but calc-alkalic according to the modified alkali–lime index; the latter criterion being less diagnostic for SiO₂ contents above 70 wt. % (cf. Frost et al. 2001). With an alumina saturation index (molecular ratio Al/(Ca- 1.67P + Na + K)) below or equal to 1.0, the Vistas Granite is slightly metaluminous (Frost et al. 2001). The granite is of A-type according to both classification methods of Whalen et al. (1987) shown in Fig. 8B, C. Other A-type characteristics include distinctly positive Th and U and negative Sr anomalies in the chondrite-normalized incompatible element variation diagram (Whalen et al. 1996; not shown). It belongs to the ‘postcollisional, postorogenic or anorogenic’ (as different from rift, plume and hot spots) sub-type of A-type granites according to the classification (not shown) based on a. o. Rb/Nb versus Y/Nb ratios proposed by Eby (1992). Displaying the negative spikes of Nb, Ta, Sr, P and Ti typical of many granites, the multi-element diagram (Fig. 8D) compares closely that of the upper continental crust. REE patterns (Fig. 8E) display moderate LREE enrichment (La_n/Sm_n = 2.0–3.2; n = 8), small to moderate negative Eu anomalies (Eu/Eu* = 0.42–0.66; n = 7) and high HREE contents. With regard to tectonic setting of the magma, the Vistas Granite shows within-plate setting in the diagrams with Rb versus Nb + Ta (or Nb) and Ta versus Yb as discriminants (not shown) of Pearce et al. (1984), in agreement with an A-type affinity.

Kebnekaise terrane

Kebne Dyke Complex (KDC)

From a thickness of c. 1000 m in the east, the Kebnekaise terrane wedges out rapidly westwards to a few tens of metres along the Tjäktavagge valley; in the Rombak–Sjangeli Window, the overlying Virisen terrane rests directly on the window basement. The Kebne Dyke Complex (Andréasson & Gee 1989a, 1989b) is dominated by amphibolites and subordinate granitoid dykes or sheets. In a few places, the amphibolites preserve a network of several generations of massive mafic dykes. Repeated injection of magma within single dykes occurred, but a sheeted dyke complex sensu stricto has not been observed so far. The reader is referred to Kirsch & Svenningsen (2016) for a detailed description of intrusive relations within a single outcrop in the Tarfala valley. Towards the contact with the underlying Mårma terrane, the amphibolite carries garnet and large sparkling porphyroblasts of hornblende aligned with the transport lineation. In contrast, the amphibolite along the roof thrust is fine-grained and fissile (Fig. 12A). The thickness of pervasively foliated amphibolites normally reaches 300–400 metres below eroded peak plateaus. Internal large-scale folding locally raised the pervasive foliation to steep angles. Local preservation of less deformed dolerite and gabbro may be related to the hinge zones of such folds. Primary intrusive relationships between dykes can often be inferred also when the rocks are deformed and amphibolitized (Fig. 9A–C). Evidence of contact metamorphism of rare calcareous rocks is preserved as thin bands of skarn composed of plagioclase, diopside, scapolite, phlogopite and poikiloblasts of amphibole. A small (c. 30 × 50 m) body of ultramafic rocks and gabbro cut by narrow mafic dykes occurs at Tjäktjatjåkka (N68.009936°/E18.323375°). A single thin section obtained from this body displays enstatite replaced by tremolite, phlogopite and seams of serpentine.

Massive dykes are fine- to medium-grained and mostly subophitic or intergranular; ophitic and equigranular textures also occur. Medium-grained dykes are often plagioclase porphyritic. The variation of modal composition is considerable (Supplementary material, Table 9), but characteristic of tholeiitic basalts. Clinopyroxene and orthopyroxene (hypersthene and subordinate inverted pigeonite) may occur in about equal amounts, the former often with exsolution lamellae of orthopyroxene and stained by Fe–Ti oxides. While altered dykes may contain only clinopyroxene, hypersthene predominates in younger dykes. Both varieties are often rimmed by amphibole. Olivine is absent in samples studied by us, but was reported from the outcrop studied by Kirsch & Svenningsen (2016) where igneous layering was observed, suggesting a cumulate appearance of the mineral.

A conspicuous and important feature is the coronitic growth of garnet along the interface of texturally igneous plagioclase and pyroxene (Fig. 9D and Supplementary material, Table 9), indicating an increase in pressure after crystallization, suggested also by abundant rutile. A second generation of garnet grew as porphyroblasts, visible to the naked eye, along shear bands and fractures, suggesting growth facilitated by access of fluid phase.

The transition from pristine dolerite to completely recrystallized amphibolite may occur over a distance of a metre only. The proportions between subidioblastic hornblende (60–90%) and domains (streaks), composed of recrystallized plagioclase and subordinate quartz, vary with the intensity of foliation. Fairly abundant clinozoisite and zoisite and abundant idioblastic titanite (sometimes with cores of rutile) occur, all with preferred orientation. Along the contact with the Mårma terrane, garnet porphyroblasts abound.

Felsic rocks occur as rare sheets or disrupted dykes of granite, granodiorite and tonalite. Rare outcrops display felsic dykes cutting contacts between coarse- and fine-grained mafic dykes, and mingling or agmatite relation with mafic rocks. Plagiogranite occurs, as indicated by modal analysis of coloured thin sections (Boman 2001). However, the frequency of plagiogranite, as first inferred by the senior author and reported to the Geological Survey of Sweden (Thelander 2009a, 2009b), is overestimated. The granitoid rocks are light grey, fine-grained and mostly foliated (Fig. 9E). Granodiorite and tonalite contain quartz, plagioclase and orthoclase, sometimes with excellently preserved equigranular groundmass texture, a faint fabric defined by biotite and titanite, and spaced poikiloblasts of garnet, hornblende and rare scapolite. Some gneissic granites (Gaskkajávri granite) display a distinct fabric defined by brown biotite (Fig. 15) in a groundmass of quartz, microcline, orthoclase and albite; sparse garnet porphyroblasts may also occur.

Figure 15. A. Fabric of the Gaskkasjávri granite. Arrow: garnet porphyroblast. Long side of picture is 35 mm. B. Photomicrograph of the biotite fabric in A. Titanite (arrows) occurs as small idioblastic crystals often aligned with the biotite fabric, or as larger fragments. C. Concentrate of titanite with smaller lenticular grains and larger fragments, the latter often pitted by resorption and carrying opaque inclusions (sulphides). 255X. D. U–Pb concordia diagram for titanite analysis.

The intrusive age of the KDC is bracketed between 608 and 578 Ma (Baird et al 2014; Kirsch & Svenningsen 2016)

Analysed samples of mafic rocks from the central massif (Supplementary material, Table 5) contain 48.0–53.0 wt.% SiO₂ ( = 49.65); 1.25 – 1.95 wt.% TiO₂ ( = 1.66); 10.7–13.0 wt.% FeO* ( = 11.8) and 0.22–1.26 wt.% K₂O ( = 0.57). Mg/(Mg + Fe_tot) ranges between 0.43 and 0.55. Major element contents classify the protolith as olivine basalt and tholeiitic basalt according to the R1–R2 method (Fig. 10A); according to the norm calculation, most samples represent olivine tholeiite (Supplementary material, Table 5). Variation diagrams against Zr (not shown) suggest that mobilization of major elements due to postmagmatic alteration was, with the exception of K, negligible. Correlation between Zr and trace elements is strong to very strong, except for the LILE. Some analyses deviate from the mantle enrichment array of Fig. 5B indicating contamination, in this case probably with sedimentary wall-rock as reflected by higher SiO₂, K₂O, and Zr/Y than other samples of the population, and plausible with regard to field relations. This reduced the data available for interpretation of magma character to seven samples. Variation patterns of incompatible and RE elements (Fig. 10B, C) run close above the E-MORB reference line, but display weak or no enrichment of the LILE (except Rb) and the LREE (La_n/Sm_n = 0.6–1.1; average 0.9; n = 6). Discrimination diagrams based on ternary HFSE relations (Fig. 6A, B) are inconclusive, or possibly suggest a transitional affinity with regard to N- and E-MORB. Not shown Y–Nb–La relations (Cabanis & Lecolle 1989) display a scatter including affinities to N-MORB, back-arc basins and continental rift basalts.

Chemical data of mafic rocks from the Tarfala area (Supplementary material, Table 6) include, in addition to major elements, Zr and Y and RE elements. The samples plot as olivine basalt and subordinate tholeiitic basalt according to the R1–R2 classification shown in Fig. 10A. Picritic basalts include hornblenditic samples with low contents of SiO₂ (43.6–44.5 wt.%), mg# values between 59 and 64, but high contents of also CaO (13 wt.%) and fairly high Cr and Ni contents. Fractionation of olivine and plagioclase is suggested also by the variation of Al₂O₃ with MgO (not shown). Alkali-silica and TiO₂ versus Zr/P₂O₅ ratios (not shown) confirm tholeiitic affinity. REE variation lines (Fig. 10D) follow, as a broad band, the E-MORB reference composition (La_n/Sm_n = 0.8–1.6). With increasing enrichment of HREE, CaO and MgO contents decrease, suggesting that crystal fractionation may have influenced enrichment. In a diagram with Ti–Y–Zr as discriminants (not shown) all samples cluster within the MORB field.

Interpreting the metadolerites and amphibolites in the Tarfala valley as “enriched with respect to N-MORB, but not to the degree of an E-MORB”, Baird et al. (2014, p. 11) concluded that the KDC is the least enriched of the COT dyke swarms. From the central massif, they reported a metagabbro, enriched by fractionation. Kirsch & Svenningsen (2016) published chemical data from a single outcrop at the base of the complex in the Tarfala valley, amplified by analytical data reported by Pettersson (2003) which, however, includes also the VA samples. Screened for crustal contamination based on Th/Yb versus Ta/Yb relations, their own data give REE contents close below, MREE contents close above, E-MORB reference compositions. Primitive mantle-normalized incompatible multielement data plot along the E-MORB line except LILE, which drop below the line. Discrimination diagrams based on ternary HFSE relations show both N-MORB and E-MORB affinites.

Analysed granitoid rocks of the KDC (n = 8; Supplementary material, Table 7) contain 58–78 wt.% SiO₂; 2.0–5.5 wt.% K₂O and 2.4–3.5 wt.% Na₂O and range from granodioritic to quartz-monzonitic in classification diagrams based on major element compositions (Fig. 8A). Strong negative correlations between SiO₂ and Al₂O₃ (r = –0.96), Fe₂O₃ (–0.93), MgO (–0.91), TiO₂ (–0.88) and P₂O₅ (–0. 88) would suggest magmatic evolution of a single suite, which is, however, modified by a strong negative correlation with also Nb (–0.93). Most samples classify as calc-alkalic and magnesian according to Frost et al. (2001). The rocks are borderline cases with regard to alumina saturation (ASI = 0.98–1.07; normative corundum = 0–0.45; n = 7) and also according to the I/S-type classification based on the Al₂O₃/(CaO + Na₂O + K₂O) ratio (0.97–1.05). In the Whalen et al. (1987) diagrams (Fig. 8B, C) the samples plot as both ISM-types and A-type, but mostly as borderline cases. This disparity is repeated by the Rb versus Nb + Ta diagram (not shown) for discrimination between tectonic settings. The incompatible and RE element contents (Fig. 8F, G) are higher than the reference composition for the upper continental crust (Rudnick & Gao 2003). The negative Eu anomaly (Eu/Eu* = 0.60–0.73) is significant for the granite only (0.30).

Virisen terrane (Stephens & Gee 1985; Lower Köli terrane of Roberts 1988)

The unit exposed along the western slope of the Tjäktjavagge valley (Fig. 1) resembles closely the Lower Köli Nappe (Stephens 1977; Stephens et al. 1985) with regard to both lithofacies and structures. Tightly foliated amphibolites of the KDC and, locally, gneisses of the Mårma terrane, underlie the Köli sequence. However, at the northern end of the valley, only a few tens of metres of Seve rocks separate Köli rocks from the Ediacaran cover of the Dividal Group resting on Precambrian crystalline rocks of the Rombak–Sjangeli window. Garnetiferous schistose phyllites and garben schist typical of the Lower Köli Nappe run along the bottom of the valley; graphitic or calcareous phyllites and schists, grey or brown calcareous or dolomitic marble and metabasalts alternate on the lower slopes. Micaceous quartzite (‘Patta quartzite’ of Kulling 1964) follows on the upper slope and crest of the ridge. The quartzite may preserve primary sedimentary structures, but the schists, carbonates and metavolcanic rocks are pervasively foliated and folded (cf. further below). Metamorphism is of greenschist to lower amphibolite facies.

Structures

Transported, the early structural features of the VIC and the KDC rift basins have little bearing upon the tectonostratigraphy of the Kebnekaise Mts.; this study is restricted to structures related to terrane amalgamation and accretion. Although the Seve terranes of the Kebnekaise Mts. escaped the exceedingly complex late Scandian structural evolution of the hinterland (c.f. Robinson et al. 2014), strain-partitioning implied that structures of one and the same event vary strongly with regard to scale, style, orientation and fabric development. Type of deformation (flattening, constriction) may shift rapidly and, locally, incipient melting released stresses. Adding to the complexity is the extensive Scandian transposition of earlier structures. For this reason and with regard to the premises of this study, we do not attempt a conventional type of event chronology (D1-n, S1-n, F1-n), but restrict the interpretation to structures related to pre-Scandian amalgamation of terranes (A), Scandian accretion of the composite terrane (S) and postaccretion warping and faulting affecting the outcrop pattern (P).

Figure 11 is a compilation of the pervasive foliation, fold axes and lineations at different levels within the stack of Seve and sub-Seve thrust sheets. The foliation diagrams (left column) are dominated by dips of 10–30° towards WNW, representing schistosity, gneissosity and imbrication related to Scandian thrusting, but include also axial plane foliations of recumbent, isoclinals folds in areas devoid of Scandian reworking and high-grade metamorphism. Steeper dips scatter as weak, NNE–SSW trending girdles; these dips disappear towards the floor of the Mårma terrane. Pervasive foliations of sub-Seve terranes dip < 10° towards west but also include a late warping towards NNE (the maxima possibly partly a result of skewed recording).

Trends of folds (Fig. 11, middle column). Structures of the Mårma terrane indicate a continuous evolution under large strains resulting in tight isoclinal folds, in turn folded by northerly trending folds passing into non-cylindrical folds, in places accompanied by sheath folds (Fig. 12F, G). Transport-parallel folds occur as steep folds (Fig. 12D; “4” in Fig. 11) or tight isoclinal folds with subhorizontal axial planes, the latter developed in terrane boundary vicinity or between bodies of different competence (Fig. 12E). NE-trending folds of the roof of the Mårma terrane include asymmetric ESE-vergent folding of transport-parallel folds and of the transverse stretching lineation, resulting in steep to vertical fold axes (“3” in Fig. 11) and lineations. A component of vertical shortening in addition to simple shear is indicated by isoclinally folded mafic dykes passing into segmented bands or boudins, the necks of the latter sometimes invaded by leucosomes (Fig. 2D), providing excellent targets for future dating of this event of deformation. Boudinage is a common structural feature of Seve terranes and usually interpreted as due to flattening related to collapse following upon (Scandian) stacking of thrust sheets. However, in eclogite-bearing terranes, extensional deformation resulting in boudins accompanied pre-Scandian exhumation of the eclogites (Andréasson et al. 1985; Albrecht 2000). Less complex in comparison with those of the Mårma terrane, folds structures of the Kebnekaise terrane include rare steep, fairly open to isoclinal NNE-trending folding of the pervasive foliation and felsic dykes and sheets (Fig. 12B). Over a distance of 500 m across strike, the limbs of the fold marked “1” in Fig. 11 dip between 60° and vertical. We tentatively interpret these folds as related to the amalgamation. Near the base of the terrane, the folds are transposed into tight-isoclinal, transport-parallel folds (Fig. 12C; “2” in Fig. 11). Reflecting the girdles of foliation diagrams, these trends scatter around a maximum at N55°W. Detailed studies of strain character and development at the base of the terrane at Tarfala have been made by Srivasta et al. (1995) and Bhattacharyya & Hudleston (2001).

Post-accretion structures include NNW-NE-trending gentle folding of the entire terrane stack displayed by the diagram of the floor of the Mårma terrane (Fig. 11, central column, 5a, “P”). These folds can be observed in the landscape with, for instance, amplitudes of c. 5 m and wavelengths of c. 50 m, but have also been reconstructed from recorded foliations (“5b” in Fig. 11). Some NE-trending folds at the floor of the Mårma terrane include asymmetric east-vergent folds which fold the WNW–ESE stretching lineation. The lack of N–NNE fold trends in the diagram of sub-Seve terranes in Fig. 11 (bottom of middle column) is probably due to restricted observations. NW–SE folds in sub-Seve terranes include mesoscale folds parallel to the transport lineation but, close to the south-eastern corner of the Rombak–Sjangeli Window, also large-scale, open folding.

Lineation diagrams (Fig. 11, right column) display stretching lineations defined by mineral elongation, quartz fibres and rods (Fig. 13A). We interpret WNW–ESE trends (“6–7” in Fig. 11) of quartz-feldspar stretching lineations and rodding observed in the stromatic migmatites and in leucosomes entering necks of metabasite boudins as related to amalgamation. Whereas, trends recorded from the Mårma (internal) and the Kebnekaise terranes have maxima towards N55–60°W, lineations of the Mårma floor and the sub-Seve terranes show a spread towards more westerly trends or two maxima (N70–85°W), typical of Scandian transverse lineations. Also lineations recorded from the Mårma roof show such a spread, suggesting Scandian reactivation of the amalgamation boundary.

Imbrication occurred at the base of the Mårma terrane as demonstrated by Fig. 13A and suggested by Fig. 13C, D, and in places within sub-Seve terranes. Imbrication also at the roof of the Mårma terrane is indicated by repetition of the SG and subjacent gneisses, e.g., on the western slope of the Tarfala valley (Fig. 1B). ⁴⁰Ar–³⁹Ar cooling ages of muscovite (431 Ma; Ekestubbe 2004) and low metamorphic grade of the fabrics of the fault zones indicate Scandian deformation. Some late normal faults displaying both ductile and brittle deformation show extensional faulting towards west (Fig. 13B); however, similar faults with dips towards east as well as due north and south have also been recorded.

In the Leavasvággi valley (Fig. 1B), the > 10-km-long, steep Leavasbahta shear zones, run N75–85°W, cutting the large-scale, transport-parallel folding of the Leaavasvággi gneiss and the Vierručohkka Amphibolite. The fissile, rusty (Fe-oxides) foliation indicates deformation at more shallow crustal levels; probably during Scandian accretion, but prior to final emplacement, since the zones do not cut rocks of sub-Seve terranes.

Sense of displacement

Eastward emplacement characterizes the entire allochthon. Thus, the basal thrust above conglomerates, quartzites and phyllites of the Dividal Group resting on the Precambrian granitoids of the Rombak–Sjangeli Window (68.061744°/18.253019°) displays east-vergent folds and winged clasts. An eastward direction and sense of displacement during amalgamation of the Kebnekaise and Mårma terranes can be inferred from the non-cylindrical folds passing into transport-parallell folds and from east-vergent sheath folds described above (Fig. 12D). Eastward displacement is also indicated by kinematic indicators often displayed by the SG. The ESE emplacement of the composite Mårma–Kebnekaise terrane during Scandian accretion is obvious from a rich variety of meso- and microscale kinematic indicators and from stretching lineations (Fig. 13A).

Along the Tjäktjavagge valley, the boundary between Seve and Köli terranes dips 10–40° towards west–south-west (cf. Fig. 12A). At the Tjäktja Pass (N68.025428°/E18.242685°), tightly foliated, often undulating amphibolite and schistose granite of the KDC are overlain by muscovite schists carrying large (<3 cm), rotated and helicitic garnets indicating eastward displacement of the Köli unit (Supplementary Fig. 1A), as do large, winged amphibole porphyroclasts of the garben schist. The transport-related stretching lineation was folded by eastward closing sheath folds (Supplementary Fig. 1C) and by kink folds with axes trending N20–40°E.Typical of the Lower Köli Nappe elsewhere, incompetent greenschists responded to gravity-induced vertical shortening by developing crenulation and kink-folds with subhorizontal axial planes (Supplementary Fig. 1F), conjugate extensional crenulation cleavage and east- or west-vergent mesoscale folds of marble bands. Boudins of quartz formed in arenitic layers (Supplementary Fig. 1C). Tilke (1986) interpreted west-vergent folds as related to backfolding of the Köli and presented textural evidence of westward normal slip along the terrane boundary, following upon eastward transport of the Köli.

Correlation with deformational events in adjacent areas

In the Singis area, Page (1992b, 1993; cf. also Tilke 1986) interpreted the ‘eclogite’ facies schistosity of the ‘Aurek Assemblage’ as evidence of early Caledonian deformation (‘D1’), but reported no macroscopic features related to the event. The ‘D2’ event of upper amphibolites facies conditions resulted in the ‘main isoclinal axial-planar foliation’ and mineral lineations trending N60°W. These structures were largely transposed during juxtaposition of Seve and Köli terranes (‘D4’) and emplacement (‘D5’) onto the platform margin. Tilke (1986) interpreted the gentle north-west trending antiform of the Singis window as related to late (‘D6’) thin-skinned ‘detachment on a lower thrust, and development of lateral anticlines’ (p. 128). He also proposed westwards backfolding and normal slip along the base of the Virisen terrane (‘D7’) commented upon above. In terms used in this study, ‘D2’ corresponds to the amalgamation of Mårma and Kebnekaise terranes and ‘D4–6’ to the Scandian accretion of the composite terrane. The ‘D6’ late warping is recorded also from the Alesätno and Leavasvággi valleys, reflected by a weak girdle in the diagram for sub-Seve foliations (Fig. 11, bottom left column).

The structure of the entire allochthon undergoes a dramatic change westwards. Based on inter alia observations in a hydropower tunnel only 25 km to the west of the Singis Window, Björklund (1989, p. 160; Plate 1 B) described folding of the Seve thrust and sub-Seve sheets by deep, inclined and transport-parallel folds (‘F3’), which folded the thrusting-related pervasive foliation (‘S2’) and mylonites of the Akkajaure Complex (Middle Allochthon) and the Lower Allochthon, including the sole thrust. F3 is folded by late N–NNE-trending synforms and antiforms (‘F4’). In Kebnekaise, post-accretion folds with these trends include also mesoscale structures.

Bax (1989) studied the influence exerted by the imbrication at depth in the Rombak–Sjangeli Window on the structures within the overriding thrust sheets during the structural evolution of the window. Following development of the Seve–Köli floor thrust during (his) D1, thrusting of the Abisko Nappe Complex (Middle Allochthon) took place during D2 with development of the characteristic transverse lineation, and continued during D3, when the window “started to play an active role” and “initiate and influence structures in the overriding nappe pile” (p. 96). He ascribed NE-trending folds, overturned to the south-east, in the Seve–Köli allochthon to D3 imbrication at depth. His D4 event included N–NNE-trending, high-angle faults dipping both east- and westwards and cutting all units. Similar faults cutting the Mårma terrane (Fig. 13B) may correspond to this event.

Timing of amalgamation and accretion

U–Pb titanite dating

Figure 15A, B shows the fabric of a sheared granite occurring at the boundary between the Mårma and Kebnekaise terranes. Outcrops appear as a lens in the wall above the Kebnepakte glacier and as disrupted sheets south of the small lake Gaskkasjávri (N67.926167°/E18.554653°). The granite was sampled from an outcrop adjacent to the lens in the wall and from boulders on the glacier beneath the lens. In terms of morphology, size and colour, the sample contained two distinct groups of titanite crystals: (a) large, stumpy and light brown fragments and (b) small, light yellow, lenticular and idioblastic crystals, the latter often grown in rows along the biotite fabric (Fig. 15B). Some fragments display overgrowth. The sample was separated into three fractions of which the two finest ones were abraded (Table 1). From each fraction, grains with best morphology and transparency were selected. Regression of all fractions resulted in a good discordia line with an upper intercept at 1217 ± 11 Ma, and a lower intercept at 487 ± 7 Ma (MSWD = 0.9); the finest fraction plotting close to the lower intercept. The age of 487 ± 7 Ma is interpreted as the growth of titanite during deformation-induced recrystallization of the Gaskkasjávri granite and cooling of the terrane boundary through the 650–700 °C U–Pb closure temperature of titanite. The upper intercept suggests an inherited Grenvillian component.

⁴⁰Ar–³⁹Ar dating

An attempt was made to date hornblende and muscovite fabrics at the base of the Vierručohkka amphibolite and the Mårma terrane, respectively. Samples were taken also from the muscovite-rich L > S-tectonite developed from the shear zone with stromatic migmatite between the VIC and the Leavasvággi gneiss describe above (Fig. 3A, B), and from a gneiss at the base of the Mårma terrane, rich in muscovite, epidote group minerals and titanite, but lacking garnet. From the amphibolite, eight samples with porphyroblasts of hornblende and garnet were selected. Only muscovite samples gave interpretable spectra. The spectrum of the stromatic migmatite shows initially slightly descending increments followed by a plateau of 488 ± 2 Ma defined by four increments, representing 80% of the total ³⁹Ar evolved. The inverse isochron intercept is 486 ± 4 for 8 out of 10 increments. The diagram of the schist at the base of the Mårma terrane shows a plateau of 435 ± 1 Ma defined by six increments, representing 93% of the gas. The inverse isochron intercept value is 435 ± 2 Ma (increments 4–10).

Discussion

Tectonostratigraphy

Witschard et al. (2004) mapped the psammitic and pelitic gneisses with metabasites and bodies of deformed granite occurring to the east of the lower Vistasvággi valley as units of the sub-Seve Middle Allochthon. We interpret these units as southern extensions of the Leavasvággi gneiss and the VIC.

In their reconnaissance maps, Andréasson & Gee (1989a, 1989b) interpreted the amphibolite at the bottom of the Tarfala valley (‘Tarfala amphibolite’) as equivalent to the amphibolite of the Lower Seve Nappe in the central Scandes, where the amphibolite occurs associated with quartz-feldspathic gneiss, garnet-biotite schist and marble (or calcareous meta-arkose/phyllite). Subsequent work has shown that, at Tarfala, the associated rocks are absent, the chemistry of the amphibolites is different from the basal Seve amphibolites (Fig. 6A), and that the metamorphic grade is higher as compared to the Lower Seve Nappe.

Tilke (1986) placed the Aurek gabbro and associated amphibolite and paragneiss (‘Aurek Assemblage’) in the Singis area above the ‘Kebnekaise amphibolite’. Page (1992a, 1992b, 1993) referred to the latter as the Savotjåkka Assemblage from the Savocohkka hill in the south-eastern corner of the Kebnekaise Mts. (outside the map of Fig. 1B). In addition to well foliated garnet amphibolites (‘Savotjåkka Amphibolite’), the assemblage includes intercalated quartzofeldspathic gneiss (‘Savotjåkka Quartzofeldspathic Gneiss’) with lenses of meta-gabbro and subordinate calcareous rocks. The intercalations vary in thickness from a few to several hundreds of metres (Page 1993). A third unit, overlying the Aurek Assemblage, named the ‘Viddja assemblage’, includes psammtic gneiss passing into a mylonitic gneiss, interpreted as the SG by Andréasson & Gee (1989b). Based essentially on the observation that “plagioclase has been completely consumed” in some parts of the (locally layered) Aurek gabbro body, Tilke (1986, p. 144) inferred eclogite facies metamorphism; however, he also obtained pressures of 10–15 kbar and temperatures of 700–730 °C for the assemblage. Page (1992a) reported data indicating upper amphibolite to granulite facies conditions for the Savotjåkka Assemblage. Results obtained from the Viddja Assemblage indicated pressures of c. 7–9 kbar and temperatures of 571–766 °C, conditions allowing for the incipient melting (<9 kbar; c. 765 °C) of the Leavasvággi gneiss described in the present study. In conclusion, with regard to overall similarities of rocks, structures and field relations, we include the Aurek, Viddja and Savotjåkka assemblages in the Mårma terrane; the garnet amphibolite of the Savotjåkka Assemblage corresponding to the VA and the quartzofeldspathic gneiss to the Leavasvággi gneiss of the present study.

Magmatism

Conclusions reached in this study concerning sources and tectonic settings of the magmas of the VIC, KDC and VA based on geochemical signatures need supporting evidence by isotope signatures. A second uncertainty is related to the palaeogeography of the North Atlantic prior to Rodinia breakup, which is currently subject to debate. The igneous rocks of the Kebnekaise and Mårma terranes have been transported at least 500 kilometres from their original igneous provinces and corresponding mantle sources. Below we favour the concept that the Sveconorwegian (Grenvillian) Orogen reached into the high Arctic (Lorenz et al. 2012; Gee et al. 2014, 2015, 2017), but take into account also older Fennoscandian belts.

Host rocks of the VIC demonstrate that the complex was emplaced into continental sedimentary rocks, and at shallow (andalusite stability) crustal levels. The weakly alkaline character of the dolerites, their higher K contents as compared to MORB and their REE enrichment suggest a continental rift setting. The high HREE contents of both dolerite and granite suggest generation above the garnet–spinel transition in a thinning continental lithosphere (e.g., Ellam 1992). In discrimination diagrams based on LREE versus Nb enrichment (Fig. 16B), their compositions approach the fields of flood basalt reference compositions. Heat from the intrusions melted the upper crust and the Vistas Granite and hybridic varieties were formed.

Figure 16. A. The VIC, VA and KDC plot along a line between the estimated composition of < 1.1-Ga subcontinental lithospheric mantle (grey fields) and a depleted magma source. Upper grey field represents the Blekinge-Dalarna Dolerites (n = 20; Solyom et al. 1992), lower grey field the Dal Group basalts (n = 10; Brewer et al. 2002). Symbols as in previous figures. E, N = E-MORB and N-MORB reference compositions (Palme & O′Neill 2004); OIB from Sun & McDonough (1989). Discrimination method from Hooper & Hawkesworth (1993). B. Variation of various groups of Baltoscandian rift and COT magmatism with regard to LREE enrichment and Nb depletion/enrichment, and comparison of these groups with reference compositions of N-MORB (N), E-MORB (E), ocean–island basalts (OIB and flood basalts (K = Karoo; C = Columbia River; D = Deccan). ICE = Iceland. Method and most reference compositions adopted from Hollocher et al. (2007). Upper grey field: Volyn flood basalt, n = 16 (Shumlyanskyy & Andréasson 2004). Lower grey field: Siberian traps, n = 20 (Hawkesworth et al. 1995). Samples of both fields were filtered to the composition: 46–49 wt.% SiO₂; 1.3–2.9 wt.% TiO₂; 0.3–1.0 wt.% K₂O. Field with thick line: KDC filtered for contamination, n = 16 (this study; Baird et al. 2014; Kirsch & Svenningsen 2016). Field 1 : Alkaline Särv dolerite, Leksdal Nappe, n = 5 (Solyom et al. 1985, Table 1; Nb = 0. 76*Nb¹⁹⁸⁵). 2 : Egersund dyke swarm; aphyric dykes, n = 6 (Bingen & Demaiffe (1999). 3 : Hedmark Group basalts adjacent beneath the Moelv tillite, corresponding to the c. <596 Ma Late Neoproterozoic glaciations; n = 10 (Furnes et al. 1983; Kumpulainen et al. 2016; Bingen et al. 2005). 4 : Amphibolites of the Sylarna Mts., n = 20 (Pettersson 2003); 5 : Särv and Saetra type dyke swarms cutting sediments and basement plinths of rift basins (Solyom et al. 1985; n = 5; Hollocher et al. 2007; n = 91). 6 : Sarek Dyke Swarm, n = 20. (Andréasson et al. 1992; Svenningsen 1994b). 7 : Troms county dyke swarm, n = 20 (Stølen 1997). 8 . Coronitic dolerites and eclogites of the Grapesvare Nappe, Seve terranes, n = 10 (Andréasson & Albrecht 1995). Crosses: dolerites of the VIC. C. The KDC (dots) magma was influenced neither by subduction processes, as indicated by low contents of elements with preferential transport in fluids (Sr), nor by elements inherited from melted sediments (Woodhead et al. 1998). The trend shown by the Vistas Granite (diamonds) could indicate melting of crustal wall-rock. Triangles: Litlefjord and Revsneshamn granites (Kirkland et al., 2006).

Ternary relations of incompatible elements classify the VA as N-MORB and the REE patterns are flat, but the contents of the REE are far too high, even for E-MORB. The MORB affinity may indicate a later stage of the rift magmatism suggested by the dolerites, but available chemical data are inconclusive.

Normally, incompatible elements do not fractionate from each other during partial melting and the following crystallization processes, and ratios of the elements are considered to represent their mantle source. Hooper & Hawkesworth (1993) used relations of HFSE (Nb, Zr) and LILE (Ba) in order to estimate the influence of asthenospheric and lithospheric source regions on the magma of the Columbia River Basalt. The composition of the corresponding subcontinental lithosperic mantle (SCLM) was estimated from intrusions of the preceeding (Mesozoic) magmatism. Accepting that the Sveconorwegian–Grenvillian belt extended at least along the length of the present-day mountain belt, the SCLM related to Seve magmatism would have been influenced by Sveconorwegian processes. The arcuate dyke swarm of the 0.95–0.98-Ga-old Blekinge-Dalarna Dolerites (BDD) ought to have extended into the Sveconorwegian belt and is selected here as reference composition of the post-Sveconorwegian SCLM in Fig. 16A (grey field). However, since the northern extension of the Sveconorwegian belt is debated, basalts of the c. 1.1-Ga Dal Group, considered to derive from the SCLM and represent the last magmatic event before Sveconorwegian accretion to Baltica (Brewer et al. 2002), are also used as reference.

The difference between the two alternatives does not affect the main conclusion drawn from application of Hooper & Hawkesworth (1993) method (Fig. 16A), i.e., that the VIC, VA and KDC magmas did not derive from partial melting of an OIB source asthenospheric mantle, nor do samples trend towards the OIB in the La_cn/Sm_cn versus Nb/La diagram (Fig. 16B). Tegner et al. (2016) inferred an asthenospheric influence for the ‘Baltoscandian dyke swarms’ based on comparison with the contrasting ∆Nb values (∆Nb = 1.74 + log(Nb/Y) − 1.92 × log(Zr/Y); Fitton et al. 1997) of plume-enriched Icelandic basalts (∆Nb > 1) and N-MORB (∆Nb < 1). However, they noted a change to dominantly negative values towards the north along the mountain belt, i.e., in the Sarek (Fig. 1A) and Kebnekaise Mts., which is in agreement with the KDC samples analysed in this study, of which all except one show negative ∆Nb values (mean value: minus 0.1). Our mapping of central parts of the Kebnekaise Mts. has confirmed observations at Tarfala (Andréasson & Gee 1989a) that sedimentary rocks are very subordinate. This fact and the body of ultramafic rocks at Tjäktjatjåkka could indicate that the KDC derived from a more distal part of the passive margin as compared to the dyke swarms of the Sarektjåkkå Nappe in the south (cf. further below) and the Váivvančhkka and Rohkunborri nappes, adjacent in the north (localities 3–4 in Fig. 1A; Kathol 1987, 1989; Stølen 1994a, 1994b; 1997), in harmony with the strong Nb depletion. Baird et al. (2014, p. 11) described the KDC of the Tarfala area as “the least enriched member of the tholeiitic continent–ocean transition”.

The scatter displayed by the granitoid dykes and sheets of the KDC need not imply different orogenic settings of magmas, but merely reflect processes at different stages and levels in a rifted continental margin. The A-type affinity, if practicable, speaks against volcanic arc setting. We propose fractional crystallization of the already REE-enriched KDC magma in a chamber of the rift system. The tendency to S-type affinity could reflect formation by melting of rift wall-rock, as suggested by Th/Yb ratios (Fig. 16C), and proposed by Baird et al. (2014) for the dated granite (WP-65).

Amalgamation and accretion

The Leavasvággi gneiss underwent a pressure increase sometimes after emplacement of the VIC, as indicated by the replacement of andalusite in the contact aureole by sillimanite and kyanite, and by nests of kyanite growing on muscovite fish of protomylonites and stromatic migmatites of the Leavasvággi and SG gneisses. Coronitic metadolerite in boudins hosted by the gneisses indicate pressure increase. In the VIC gabbro, garnet coronas developed along the pyroxene-plagioclase interface and as the outermost rim of olivine-pyroxene coronas. Also the KDC underwent a pressure increase as suggested by garnet–corona dolerites. Srivastava et al. (1995) obtained c. 800 °C from garnet–clinopyroxene and garnet–hornblende assemblages and 9–14 kbar for a garnet–clinopyroxene–plagioclase–quartz assemblage of a sheared metagabbro at Tarfala. They wrote: “Textures in the rocks are consistent with an equilibrium metamorphic mineral assemblage … Similar textures inside and outside the shear zones indicate that deformation was synchronous with metamorphism or that peak metamorphism outlasted deformation” (p. 1218).

The high grade of metamorphism and varied structures of adjacent terranes require that amalgamation took place prior to the greenschist lower amphibolites facies grade and pervasive deformation characteristic of Scandian accretion. Juxtaposition of the two terranes occurred at deeper crustal levels, a conclusion confirmed and refined by geochronological results obtained in this and previous studies. Page (1992b) obtained ⁴⁰Ar–³⁹Ar ages of 450–458 Ma of muscovite from shear zones between the Aurek and Manak gabbro bodies and their host rock, i.e., locations corresponding to those of VIC/Leavasvággi gneiss shear zones. Baird et al. (2011) reported an age of c. 452 Ma (U–Pb TIMS) for titanite in an amphibolite from the Tarfala valley. Using CO₂-laser technique, Ekestubbe (2004) obtained ⁴⁰Ar–³⁹Ar ages of 443 Ma for muscovites from the SG, and 460 Ma from garnetiferous schistose gneiss of inferred Leavasvággi gneiss affinity within the telescoped nappe sequence at the corner of the Rombak–Sjangeli Window.

For the Scandian juxtaposition of Seve and Köli terranes and their transport above sub-Seve terranes, a number of ⁴⁰Ar–³⁹Ar muscovite cooling ages around 430 Ma have been reported by Page (1992b), Tilke (1986) and Ekestubbe (2004). The muscovite age of 435 ± 1 Ma obtained in this study from the base of the Mårma terrane is consistent with these results. The high ⁴⁰Ar–³⁹Ar age (488 ± 2 Ma) obtained for muscovite from the transport-parallel stromatic migmatite at the VIC/Leavasvággi gneiss contact must be repeated or confirmed by other method before any geological significance can be ascribed to it.

The evidence of increased pressure could indicate that amalgamation of the Mårma and Kebnekaise terranes took place during subdution and imbrication of the Baltoscandian margin. Localization of migmatites to the boundary between the terranes could indicate that melting enhanced emplacement.

Correlation and affinity

Sarek Mts

On maps (Kulling 1964; Thelander 2009a), the Mårma terrane extends southwards to the Akkajaure culmination (1 in Fig. 1A). South of the culmination, in the Sarek Mts., the Sarektjåkkå Nappe carries the 608-Ma-old dyke swarm (Svenningsen 2001). The nappe is underlain by gneisses similar to the Leavasvággi gneiss, and hosting lenses of metadolerite, including also retroclogites (Andréasson 1986; Rehnström 1998; Lund 1999), calc-silicate gneiss, amphibolite and slices of deformed granitoids. A unit composed of deformed gabbro and granite occurring close below the COT dyke swarm may correspond to the VIC (Rehnström 1998, and personal communication during her visit to Kebnekaise in 1999). The eclogite-bearing Grapesvare Nappe underlying the Sarektjåkkå Nappe carries slices of c. 945–Ma-old crystalline rocks (Albrecht 2000), interpreted as basement plinths of the rift basins, cut-off during nappe translation. In the COT dyke swarm of the Sarek Mts., metasedimentary host rocks are much more predominant as compared to the KDC, including also marbles and meta-evaporites (Svenningsen 1994a). Chemical analyses of samples, obtained from walls and ablation zones of glaciers, published up to now (Andréasson et al. 1992) and from a single locality (Svenningsen 1994b) indicate subduction influence on magma composition suggested by Th/Yb versus Sr/Nd, and Th/Yb versus Ta/Yb ratios (not shown); they do not follow the normal mantle enrichment trend, and plot as volcanic arc basalt in HFSE-based discrimination diagrams (not shown). High metamorphic grade during emplacement of the Sarektjåkkå Nappe is indicated by kyanite and sillimanite in foliated metasedimentary screens at the base of the dyke complex. ⁴⁰Ar/³⁹Ar ages of hornblende from samples taken at, or close to, the base of the Sarektjåkkå Nappe, yielded ages of 464–469 Ma (Dallmeyer et al. 1991). Narrow zones of SG-type gneiss occur along the base of the nappe, but migmatites have not been observed so far. Mica schists, graphitic phyllites, marbles, mafic and acid metavolcanic rocksand garben schists typical of the Lower Köli Nappe of the Virirsen terrane overlie the Sarektjåkkå Nappe in its north-western corner (Sierkavagge valley; Andréasson 1986).

Lake Torneträsk and Indre Troms

A klippe of lithologies comparing closely to those of the Mårma terrane occurs at Kålkuktjåkkå (7 km S of Lake Torneträsk). Kathol (1984) reported andalusite from isoclinally folded metaarenite at this locality. Gneisses of high metamorphic grade (kyanite + sillimanite) and large strains (sheath folds) hosting boudins of retro-eclogite have been mapped from Lake Torneträsk and 50 km northwards to Mt. Njunis (Fig. 1A: localities 3–4; Kathol 1989; Stølen 1994a, 1994b; 1997). Kathol (1989) reported retro-eclogites preserving clinopyroxene with < 10 formula % jadeite and garnet with < 10% pyrope. He considered if the lack of preserved Ordovician subduction-related sedimentary successions ‘reflected a lesser intensity of subduction processes and corresponding uplift within the Torneträsk section, compared with areas further to the south’ (p. 63–64). An associated amphibolite unit bears petrographical and chemical (major element) similarities to the VA. On the other hand, despite the short distance from Kebnekaise, the dyke swarms of the overlying Váivvančohkka and Rohkunborri nappes differ markedly from the KDC with regard to dyke density (65–70%) and the lithofacies of the screens (impure marble, dolomite and calcareous, graphitic and psammitic schists). However, the dolerite is almost as depleted as that of the KDC (Fig. 16B).

Finnmark county. Seve–Kalak relations

Main differences between the Seve terranes and the KNC include: (a) the absence, in the KNC, of sediments deposited after c. 840 Ma (Kirkland et al. 2007), of the COT and of the HP/UHP metamorphism; (b) the absence, in the Seve terranes, of late alkaline magmatism comparable to the Seiland Igneous Province (SIP; Fig. 1A). However, with regard to the Seve terranes of the Kebnekaise Mts., the protolith of the Leavasvággi gneiss was deposited before c. 840 Ma. Paulsson & Andréasson (2002) compared age data of the Vistas Granite with the c. 830-Ma old (Daly et al. 1991) Litlefjord Granite of the KNC (Fig. 1A: locality 5) and proposed an origin of the magmatism in a region of initial break-up of Rodinia, allowing for granites of similar age to occur today on the Laurentian side of the suture, e.g., in the Scottish Highlands. The Litlefjord Granite has since been redated and three additional granites have been included in the list of c. 840-Ma granites intruding the Sørøy Succession with a depositional age of 910–840 Ma (Kirkland et al. 2006; Corfu et al. 2007). Inherited components of the Vistas Granite have analogues among the KNC granites but also with intrusions of the Central Taimyr Belt (Paulsson & Andréasson 2002; their Table 4). One inherited component (1778 ± 11 Ma) compares to the protolith U–Pb age of 1776 ± 4 Ma (Rehnström et al. 2002) obtained for the granitic gneiss in the Sarek Mts. However, Kirkland et al. (2007) inferred a volcanic-arc setting for the 840-Ma granites of the KNC; a view later referred to as “evidence of proximity to a subduction zone” in a recent palaeogeographic model (Cawood et al. 2010, p. 101). Realizing that the monotonous psammitic–pelitic gneisses of the KNC successions hosting the granites bear little resemblance to volcanic arc successions, Kirkland et al. (2007) suggested successor basins “located on top of the developing Grenville Orogen after successive cycles of accretionary orogenesis and erosion. Later these basins were intruded by arc complexes”. Since such a tectonic setting would exclude correlation with the Vistas Granite as interpreted in this study, we examine here the geochemical evidence (Kirkland et al. 2006, Table 2) of the proposed volcanic arc. According to the classification scheme of Frost et al. (2001) the 840-Ma granites are ferroan (i.e., ‘tholeiitic’), and straddle the boundary between alkali-calcic (n = 3) and calc-alkalic granites (n = 2) according to the modified alkali–lime index. They plot on the boundary line between I-S-M and A-type granites (Fig. 8B, C). In the Rb versus trace element discrimination diagrams based on HFS and LIL elements of Pearce et al. (1984) the granites plot very close to the triple points for the within-plate, syncollisional and volcanic-arc fields. The very weak calc-alkaline affinity displayed by discrimination diagrams based on HFS elements and the transitional behaviour in I-S-M-A classification diagrams could reflect influence of calc-alkaline lithologies in the rift systems of the pre-existing crust (Fig. 16C). Such an origin is supported by the Rb(Sc)/Nb versus Y/Nb diagram (not shown) proposed by Eby (1992) where the granites plot among A₂-type granites, i.e., a sub-type of the A-group “generated from crust that had been through a cycle of subduction-zone or continent-continent collision magmatism” (p. 643), i.e., in this case, Sveconorwegian crust. Kirkland et al. (2006) also refer to a single sample of a small gabbro body cut by the granite. If discussion of a single analysis is at all meaningful, the very high content of MgO (15 wt.%), high contents of Cr and Ni and low contents of TiO₂, Al₂O₃, Na₂O and Sr indicate crystal fractionation; thus, the employed discrimination diagram of Pearce & Cann (1973) based on Ti–Zr–Y relations and the Nb–Y–Zr diagram of Meschede (1986) should be interpreted with great caution. The Th–Hf–Ta diagram referred to by the authors is more appropriate, unfortunately, Hf and Ta are lacking in the published data table. Moreover, evidence that the gabbro and granite are coeval is not presented. In conclusion, we question a volcanic arc setting and compressional orogenic context of the c. 840-Ma granites of the KNC.

The host rocks of the VIC (Leavavággi and Storglaciären gneisses and migmatites) compare closely to the Klubben psammitic gneiss and underlying Eidvågeid Paragneiss of the KNC with regard to rock type, petrography and structures, and the overall metamorphic evolution. The KNC preserves evidence of extensive metamorphism at c. 850–820 Ma and c. 710–680 Ma (Kirkland et al. 2006, Roberts et al. 2006; Corfu et al. 2007; Gasser et al. 2015). While the high T/P-event of the Seve terranes manifested by the andalusite pseudomorphs is related to the 850-Ma emplacement of the VIC, the interpretation of the ensuing pressure increase must await quantitative thermobarometric and additional geochronologic work in Kebnekaise. In their detailed metamorphic and geochronological study of fabrics of pelitic gneiss of the KNC, Gasser et al. (2015) found no evidence of early metamorphism within the andalusite/sillimanite stability field. The authors drew the far-reaching conclusion that the apparent absence of pre-Iapetan high T/P metamorphism in the crust, from which the KNC derived, indicated a collisional instead of extensional regime from at least c. 800 Ma. However, their study was restricted to a narrow shear zone exposed by a road-cut. As demonstrated by the present study, evidence of high-T metamorphism can be found in areas which resisted subsequent penetrative deformation, such as the contact-metamorphic aureoles of the 840-Ma intrusions in Kebnekaise Mts.

KDC and Baltoscandian rift magmatism

Baltoscandian rift magmatism (Andréasson 1994 for review) is hosted by a tectonostratigraphic sequence displaying upwards increasing frequency of intrusions: (a) rare dyke swarms cutting the basement (Egersund dykes); (b) nappes derived from rift basin fill and basement plinths and cut by dolerite dyke swarms (the 596-Ma-old Ottfjället dolerite dykes swarm of Särv and equivalent nappes; some eclogite-bearing Seve nappes); (c) nappes with swarms of high dyke density (>c. 70%) cutting psammites, marbles, black schists and evaporites (Indre Troms Mts; Sarek Mts.); and (d) nappes almost (Kebnekaise Mts.) or entirely (Sylarna Mts.; Pettersson 2003) composed of mafic and subordinate plagiogranitic dykes, and ultramafites.

Based on discriminant analysis of comprehensive geochemical data from the dyke swarms of Särv and equivalent nappes within the central Scandinavian Caledonides, Hollocher et al. (2007) combined criteria of LREE enrichment (La_cn/Sm_cn) with arc or continental affinity (Nb/La) in order to identify different groups and their relations to various orogenic reference settings. Based on these parameters, Fig. 16B examines the KDC in the context of Baltoscandian rift and COT magmatism along the mountain belt. The Egersund dyke swarm (field 2), cutting autochthonous basement, represents the most proximal magmatism. The mildly alkaline Hedmark basalt (field 3) adjacent beneath the Moelv tillite is considered short-transported as compared to the rift basin dyke swarms, which vary from rare alkaline basalts (field 1) to the predominating, enriched MORB (field 5). Analyses of the coronitic dolerites and eclogites of the Seve terranes (field 8) plot at the extreme end of the Nb/La axis due to very low LREE contents (average La_n/Sm_n = 0.9) rather than Nb enrichment (average 10; n = 14). Subducted to depths approaching 100 km (Albrecht & Andréasson 2000), these E-MORB metadolerites may represent rift basins detached form a more distal part of the diving margin. Among COT basalts, the KDC samples (fat line field) plot between N-MORB and arc compositions, slightly more depleted than the Troms county dyke swarms (field 7) and markedly more depleted than the Sarek (field 6), the latter group requiring more study (cf. above) and therefore interpreted here with caution. Plotting close to the OIB reference composition, the Sylarna complex (field 4) deviates from all other COT swarms. If the Baltoscandian rift system was related to a plume, the more pronunced oceanic character of the Sylarna lithofacies and the ∆Nb value of 0.35 could indicate that this segment of the COT was located closer to the plume head as compared to the KDC. However, as long as geochronological data is lacking, the Sylarna complex must be treated with caution.

Kirsch & Svenningsen (2016) inferred a temporal trend towards depletion of Seve magmatism, but enrichment of KNC magmatism, leading the authors to the suggestion of an exotic origin of Seve terranes. Their interpretation is based on a plot of HFSE relations (100*(2Nb/(2Nb + Zr/4 + Y)); Hollocher et al. 2007) versus a few available more or less precise age determinations. The diagram is selective and the trend towards enrichment of the KNC is based solely on the c. 610-Ma Corrovarre dyke swarm and the alkaline rocks of the SIP as end members. Tholeiitic intrusions belong to the evolution and narrow time span of the SIP; moreover, crustal contamination and assimilation influenced magma evolution (Tegner et al. 1999; Roberts 2007). A temporal trend is not supported by the evidence that the enriched 596-Ma-old Ottfjäll dolerite of the rift-basins (Nb–Zr–Y-value = 26–30) is coeval with the c. 600-Ma-old depleted (Nb–Zr–Y-value = 17) dyke swarm intruding carbonates of the outermost margin (Rohkunborrhi Nappe). The Sylarna complex, with typical COT lithofacies, is strongly enriched (average Nb–Zr–Y-value = 35) as compared to the KDC (average Nb–Zr–Y-value = 15). Also within single rift-basin swarms, the variation of Nb depletion and LREE enrichment is considerable (Fig. 16B). Hollocher et al. (2007) demonstrated such a variation across 400 km of the Särv–Saetra nappes and compared it to the variation along the present-day Mid-Atlantic ridge axis due to hot spot volcanic islands.

Summary and conclusions

In the Kebnekaise Mts., the Seve allochthon is represented by the Mårma and Kebnekaise terranes derived from rifted continental crust and the continent–ocean transition, respectively. The terranes amalgamated in the early Ordovician and accreted, in late Silurian, as a composite terrane onto sub-Seve terranes of the Middle Allochthon. Seve terranes of the Kebnekaise Mts. are overlain by a sequence of quartzites, schists, phyllites, marbles and metavolcanics equivalent to the Lower Köli Nappe (Virisen terrane).

The Leavavággi gneiss of the Mårma terrane is composed of quartzofeldspathic, semipelitic and subordinate calc-silicate gneisses and intruded by the Vassačorru igneous complex, a network of mingled and mixed mafic and granitic rocks, the latter including the 845-Ma-old Vistas Granite. The mildly alkaline dolerite of the complex compares to transitional basalts of continental rifts. The Vierručohhka amphibolite shows N-MORB affinity in discrimination diagrams for tectonic settings, while incompatible and RE element contents demonstrate the same enrichment compared to E-MORB and high HREE reference compositions as the dolerites. The high HREE contents suggest generation above the garnet–spinel transition in a thinning continental lithosphere. Heat from the intrusions melted the upper crust and generated the quartz-monzonitic to granitic Vistas Granite and hybridic varieties. The granite classifies as ferroan and A-type. The proposed correlation of the Vistas Granite with the Litlefjord and other 840–850-Ma granites of the KNC is justified. The volcanic arc setting and compressional orogenic context of the Litlefjord Granite referred to in recent palaeogeographic modelling are questioned. Emplaced at shallow (andalusite stability field) crustal levels, the Vassačorru igneous complex and host rocks underwent deformation and metamorphism in the sillimanite–kyanite stability field, and local migmatization. The Mårma terrane displays a complexity of tight-isoclinal recumbent folding, non-cylindrical folds, sheath folds and transport-parallel folds, all inferred related to, or partly predating, Ordovician amalgamation. Mylonitic paragneisses and augen gneisses, formed from the Vistas Granite, abound. Imbrication of the base of the Mårma terrane occurred. The Leavasvággi gneiss may correspond to the basement preserved as plinths of the rift basins in the eclogite-bearing nappes.

The Kebnekaise terrane consists of amphibolite formed from dolerite and gabbro of the Kebne Dyke Complex, which includes also subordinate felsic dykes or sheets, and rare small bodies of ultramafic rocks. Coronitic dolerites and gabbros suggest high pressures before or during amalgamation with the Mårma terrane. Samples analysed in this study classify as olivine basalts. Incompatible and RE elements indicate enrichment as compared to N-MORB. However, combined with the variation of the LREE (La_cn/Sm_cn vs Nb/La), the magma source of the KDC of the central massif appears as the most Nb–La depleted among the COT complexes, supporting results obtained from the Tarfala valley by Baird et al. (2014). According to chemical compositions, felsic dykes are mainly granodioritic to quartz monzonitic and originated by melting of continental crust or sediments. However, modal compositions indicate also the presence of rare trondhjemitic dykes. Large-scale longitudinal folds in the internal parts of the complex are transposed by transport-parallel isoclinal folds at the base of the complex. Imbrication occurred at the base of the terrane.

Amalgamation of the Mårma and Kebnekaise terranes occurred in Early Orodovician and at higher temperatures and pressures as compared to those prevailing during Scandian accretion. The pristine character of rocks of some nappes (Särv Nappes) derived from the rift basins suggest that slices of the margin detached and slided on top of the diving, imbricated margin. We speculate that the COT of the Kebnekaise terrane detached in the same way, but was dragged to some depth on top of subducting slices (Mårma terrane) arriving from more proximal, basement-dominated, parts of the margin. Locally, Silurian accretion reactivated the terrane boundary. Whereas, transvere linear structures recorded from the internal Mårma and Kebnekaise terranes have maxima towards N55–60°W, those of the Mårma floor and the Sub-Seve terranes show a spread, or two maxima, towards more westerly trends (N70–85°W) reflecting Scandian accretion. Postaccretion deformation includes N–NNE-trending open folding and warping and late normal faults and flexures, verging north, south, east and west and cutting the entire nappe pile.

Results obtained in this study give no reason ascribe an exotic affinity to the Seve terranes of the Kebnekaise Mts.

Funding

This work was partly supported by the Swedish Natural Science Research Council [grant number G5103-672] to P. G. A. (The Baltoscandian volcanic rifted margin: fragment of a Neoproterozoic-Cambrian Large Igneous Province?).

Supplemental data

Supplemental data for this article can be accessed here: https://doi.org/10.1080/11035897.2018.1470200.

Acknowledgements

Dr. Josef Tomas, Geological Survey of former Czechoslovakia, Prague, kindly put whole-rock chemical data and thin sections from his study (Fleetwood & Tomas 1990) of weathering and ion flow in the Tarfala drainage basin at our disposal. We are indebted to Torgeir B. Andersen and Reinhard O. Greiling for constructive reviews, and to GFF editors Christian Skovsted and Magnus Ripa for helpful advice and editing and Taylor & Francis editor Sarah Minchin. The first author thanks David Gee, Benno Kathol, Laurence Page, Emma Rehnström, Lars- Kristian Stølen and Olaf Svenningsen for useful discussions on the outcrops. Figs. 2C, 7A, 9A, C, 14B are found (with somewhat larger size) in Thelander (2009b). The following authors have changed their names: Ann Allen corresponds to ‘Pettersson, A.’ in the Reference list; Oskar Aurell corresponds to ‘Paulsson, O.’ in the list.