An Iapetus origin for a layered eclogite complex in the northern Western Gneiss Region, Scandinavian Caledonides

The Western Gneiss Region (WGR) is a Precambrian basement domain in the Scandinavian Caledonides and one of the world's largest high‐ and ultrahigh‐pressure terranes. The south–central WGR underwent regional eclogite facies metamorphism 415–400 Ma ago when Baltica subducted beneath Laurentia, during the Scandian orogeny. Eclogites in the WGR group into two traditional types: (1) Precambrian mafic intrusions metamorphosed in situ during Scandian continental subduction and (2) eclogites, garnet peridotites and garnet pyroxenites within ultramafic complexes derived from the subcontinental mantle beneath Laurentia. We document, using field relations, petrography, whole‐rock geochemistry and secondary ion mass spectrometry (SIMS) zircon geochronology, a hitherto unrecognized third type of eclogite in the WGR that places new constraints on its tectonic architecture: an eclogitized fragment of oceanic crust from the Iapetus Ocean. The Kråkfjord eclogite complex is a km2‐sized body with an interior consisting of kyanite eclogite (meta‐troctolite) and subordinate layers and lenses of garnet peridotite, garnet websterite and kyanite–garnet leucotonalite. This interior is capped by Fe–Ti‐rich eclogite, which locally contains subordinate pockets of migmatitic aluminous gneiss. The elemental abundances and isotopic compositions of the Fe–Ti‐rich eclogites resemble those of mid‐ocean ridge basalt (MORB). In contrast, the interior kyanite eclogites, peridotites and pyroxenites have compositions similar to the gabbroic cumulates in the lower oceanic crust of slow‐spreading ridges. U–Pb SIMS dating of igneous zircon cores from a leucotonalite pod in the interior of the Kråkfjord complex yields Cambro‐Ordovician igneous ages of 500–440 Ma, with the ~500 Ma age interpreted as the isotopically undisturbed age. This age matches those of Iapetan oceanic rocks exposed elsewhere in the mountain belt. Metamorphic zircon from an Fe–Ti‐rich eclogite in the carapace of the Kråkfjord complex dates the eclogite facies metamorphism at 421.9 ± 2.2 Ma, synchronous with the continental collision. Zircon from a leucosome in Fe–Ti‐rich retro‐eclogite indicates an age of 408.5 ± 2 Ma for the crystallization of partial melt following the decompression. Detrital zircon core ages from a pocket of aluminous migmatitic gneiss in the carapace indicate derivation of sediment from the Baltic crust. Collectively, the data show that the eclogite complex (1) originated at an Iapetus spreading centre near the continent Baltica, (2) subducted to eclogite conditions during Scandian continental collision and (3) was tectonically intercalated with the Precambrian Baltica basement of the WGR.

continental subduction and (2) eclogites, garnet peridotites and garnet pyroxenites within ultramafic complexes derived from the subcontinental mantle beneath Laurentia.We document, using field relations, petrography, whole-rock geochemistry and secondary ion mass spectrometry (SIMS) zircon geochronology, a hitherto unrecognized third type of eclogite in the WGR that places new constraints on its tectonic architecture: an eclogitized fragment of oceanic crust from the Iapetus Ocean.The Kråkfjord eclogite complex is a km 2 -sized body with an interior consisting of kyanite eclogite (meta-troctolite) and subordinate layers and lenses of garnet peridotite, garnet websterite and kyanite-garnet leucotonalite.This interior is capped by Fe-Ti-rich eclogite, which locally contains subordinate pockets of migmatitic aluminous gneiss.The elemental abundances and isotopic compositions of the Fe-Ti-rich eclogites resemble those of midocean ridge basalt (MORB).In contrast, the interior kyanite eclogites, peridotites and pyroxenites have compositions similar to the gabbroic cumulates in the lower oceanic crust of slow-spreading ridges.U-Pb SIMS dating of igneous zircon cores from a leucotonalite pod in the interior of the Kråkfjord complex yields Cambro-Ordovician igneous ages of 500-440 Ma, with the $500 Ma age interpreted as the isotopically undisturbed age.This age matches those of Iapetan oceanic rocks exposed elsewhere in the mountain belt.Metamorphic zircon from an Fe-Ti-rich eclogite in the carapace of the Kråkfjord complex dates the eclogite facies metamorphism at 421.9 ± 2.2 Ma, synchronous with the continental collision.Zircon from a leucosome in Fe-Ti-rich retro-eclogite indicates an age of 408.5 ± 2 Ma for the crystallization of partial melt following the decompression.Detrital zircon core ages from a pocket of aluminous migmatitic gneiss in the carapace indicate derivation of sediment from the Baltic crust.Collectively, the data show that the eclogite complex (1) originated at an Iapetus spreading centre near the continent Baltica, (2) subducted to eclogite conditions during Scandian continental collision and (3) was tectonically intercalated with the Precambrian Baltica basement of the WGR.
K E Y W O R D S continental collision, eclogite metamorphism, oceanic extension, Sr-Nd isotopes, whole-rock geochemistry, zircon geochronology

| INTRODUCTION
In modern continental collision zones such as the Himalayas, most of the interface between the underthrusting lower continental plate and the overlying allochthonous sequences is deeply buried beneath the mountain belt and unavailable for direct observation.The tectonometamorphic processes that take place at this tectonic level can however be assessed in ancient collisional orogens that expose the formerly deep-seated basement in tectonic windows.The Western Gneiss Region (WGR) in Norway (Figure 1) is exposed in such a deep window within the Scandinavian Caledonides.The WGR covers an area of >50,000 km 2 from Bergen in the south to Namsos in the north and consists mainly of Precambrian granitic basement of the lower continental plate (Gee et al., 1985;Ramberg et al., 2008).The WGR was subducted to highpressure (HP) and ultrahigh-pressure (UHP) conditions during the Scandian continental collision 430-400 Ma ago (Cuthbert et al., 2000;Hacker et al., 2010;Krogh, 1982).With the exception of eclogites within garnet peridotites (Brueckner et al., 2010;Medaris et al., 2018), most eclogites in the WGR have been interpreted as meta-gabbroic components of the Precambrian basement that were metamorphosed in situ (e.g., Griffin, 1987).
The aim of this contribution is to document a previously unrecognized type of eclogite in the WGR-a piece of eclogitized Iapetan oceanic crust-and to discuss its tectonometamorphic implications.We provide data on the setting, metamorphic geology, geochemistry and U-Pb zircon geochronology of a km 2 -sized eclogite complex in Roan in the northern WGR, named the Kråkfjord complex.Collectively, the data constitute evidence of a Cambro-Ordovician Iapetus oceanic origin for the eclogite complex and relate its eclogite facies metamorphism to the Siluro-Devonian continental collision.The structural insertion of Iapetan ocean floor rocks within Precambrian Baltica basement rocks suggests tectonic imbrication of the ocean-continent transition (OCT) zone during Scandian collision.

| The Scandinavian Caledonides
The Scandinavian Caledonides formed as the Iapetus Ocean closed between 500 and 400 Ma ago (Corfu et al., 2014;Gee et al., 2008;Ramberg et al., 2008), culminating in the final ocean closure and collision between Laurentia and Baltica during the Scandian phase of the Caledonian orogeny.Between 430 and 400 Ma, Baltica was subducted westward (present coordinates) 'India style' beneath the Laurentian margin, while the overlying continental shelf deposits and oceanic sequences were thrusted eastwards (present coordinates) across the Baltica margin, forming a thick sequence of allochthons.The Caledonides in Norway and Sweden expose a unique tectonostratigraphic sequence through an ancient collisional orogen, from the overriding exotic continental rocks at the top (part of Laurentia), down through oceanic units (Iapetus), then through metamorphosed shelf rocks of Baltica, and finally into the deeply buried continental basement of the underthrust continental plate itself (Baltica).
The allochthonous thrust sheets-termed 'nappes'are traditionally grouped into four tectonostratigraphic levels (Gee et al., 1985; Figure 1).Tectonically higher allochthons were transported from successively more westerly locations, such that the nappe pile forms an overall in-sequence thrust wedge, later modified by vertical thinning and extension.Lowest in the tectonostratigraphy are metamorphosed Precambrian igneous rocks exposed in windows and generally regarded as part of the Parautochthon or Lower Allochthon.The WGR represents the largest of these basement windows.The Lower and Middle Allochthons are composed of Baltica-derived units: Neoproterozoic to Cambro-Silurian sedimentary rocks originally deposited on the Baltica margin, slices of Precambrian basement and elements of the rifted OCT zone (Gee et al., 2008).Although these nappes were transported over considerable distances, they are part of the lower tectonic plate and occupy a position below the ophiolites of the Upper Allochthon.The Upper and Uppermost Allochthons were derived outboard of Baltica.The Upper Allochthon consists of 500-430 Ma island arcderived sequences, ophiolites and related sedimentary rocks from the Iapetus Ocean (Bingen & Solli, 2009;Dunning & Pedersen, 1988;Furnes et al., 2012Furnes et al., , 2014;;Slagstad & Kirkland, 2018).The Uppermost Allochthon is composed primarily of metasedimentary rocks from the Laurentian margin, that is, from the upper tectonic plate of the collisional orogen (Augland et al., 2014;Barnes et al., 2007Barnes et al., , 2011;;Roberts et al., 2007).
Following the Scandian continental collision, the mountain chain underwent E-W extension during the Devonian.This extension resulted in westward backsliding of nappes towards the hinterland, upright folding, F I G U R E 1 Structural-tectonic setting.The Caledonides in south-central Scandinavia (adapted from Gee et al., 1985, andSolli &Nordgulen, 2008, with the 1 Ga Sveconorwegian Province from Möller & Andersson, 2018).The Caledonian thrust front is marked by black thrust marks.The Western Gneiss Region is the basement-dominated window between the Bergen and Namsos areas.MTFZ, Møre-Trøndelag Fault Zone.Red box marks the location of Figure 2a.

| The northern WGR and the Roan Window
The northern WGR, north of the MTFZ (Figures 1 and  2a), is also made up of Precambrian orthogneisses interfolded with supracrustal sequences of largely unknown origins (Birkeland, 1958;Ramberg, 1943;Solli et al., 1997).Another name for this area is Vestranden, originally coined for the coastal areas between Kristiansund and Namsos (Kjerulf, 1871;Ramberg, 1943).
The Precambrian orthogneisses in the northern WGR are in most places strongly deformed and migmatized and therefore challenging to correlate with Precambrian granitoids farther east.However, relict igneous textures and structures in granitic to quartzmonzodioritic rocks and younger dolerites are better preserved in the Roan peninsula on the coast, within the Roan Window (Möller, 1988;Figure 2b,c), which exposes the lowest structural and tectonic level of the northern WGR.Most spectacular are two large eclogite complexes: the Kråkfjord and Brandsfjorden bodies (and small eclogite lenses around these larger complexes).

| Caledonian metamorphic P-T-t-d records from the Roan area and the northern WGR
The metamorphic rocks in the Roan area record a clockwise-type P-T-t evolution suggesting a collisional scenario (Dallmeyer et al., 1992;Johansson & Möller, 1986;Möller, 1988Möller, , 1990)).All rocks, except the late pegmatites, were metamorphosed under eclogite, HP granulite or upper amphibolite facies conditions.HP granulite facies mineral assemblages correspond to conditions of ≥800 C and >12 kbar, within the kyanite stability field.Eclogite assemblages are restricted to the two km 2 -sized layered mafic-ultramafic complexes at Kråkfjord and Brandsfjorden (Figure 2b), the only known occurrences of eclogite in the northern WGR.Growth zoning in some garnet porphyroblasts in both complexes indicates a prograde and clockwise P-T path.
Secondary high-temperature minerals such as sillimanite, sapphirine, cordierite and andalusite constrain a stage of high-temperature decompression after the HP event followed by cooling at pressures <4 kbar (Möller, 1988(Möller, , 1990)).Many rocks record retrograde amphibolitization and migmatization associated with hydrous fluid infiltration related to high-temperature decompression (cf.Section 3.3).
Previous studies dated the metamorphic evolution of the northern WGR, including the Kråkfjord complex, using several methods.A Sm-Nd mineral-whole-rock isochron of 432 ± 6 Ma (Dallmeyer et al., 1992) was obtained from a well-preserved garnet websterite in the Kråkfjord eclogite complex (sample KR-10, Table 1), indicating that eclogite facies metamorphism was related to an early stage of the Scandian continental collision.Post-peak metamorphic pegmatites throughout the northernmost WGR yielded TIMS U-Pb zircon ages of 404 ± 2, 401 ± 3 and 398 ± 3 Ma (Dallmeyer et al., 1992;Schouenborg, 1988;Schouenborg et al., 1991).These ages were corroborated by LA-ICP-MS U-Pb ages of zircon in partial melts and pegmatites in the Berfjorden area in Roan (Figure 2b) that range from 410 ± 8 to 404 ± 8 Ma (Gordon et al., 2016).Thus, partial melt crystallization and pegmatite intrusions are both related to late-orogenic exhumation during Devonian extension.The Devonian pegmatites constitute structural markers for the latest stages of Caledonian ductile deformation: lower amphibolite facies shear zones (in many places localized along the contacts between pegmatite and gneiss) and upright gentle to open folding that formed the regional-scale NE-SW antiformal and synformal structures (Figure 2a; Möller, 1988;Schouenborg, 1989), including the domal F I G U R E 2 Geological setting of the northern Western Gneiss Region and the Roan area.(a) Bedrock map of the area north of Trondheimsfjorden (based on Solli et al., 1997, andNorges geologiske undersøkelse, 2022).Major structures, indicated by symbols, are synforms and antiforms (black solid lines with filled triangles), fold axes (arrows) with plunge and a Devonian extensional fault basin in the SW (black solid line with filled squares, lower left).(b) Bedrock map of the Roan area with (c) schematic cross section (both from Möller, 1988).Arrow shows the location of the Kråkfjord eclogite complex.shape of the Roan Window (Figure 2b,c). 40Ar-39 Ar ages of hornblende range from 413 ± 13 to 386 ± 6 Ma across the northern WGR, consistent with Devonian late-orogenic cooling (Dallmeyer et al., 1992). 40Ar-39 Ar muscovite from the same area yielded cooling ages from 395 ± 2 to 390 ± 2 Ma and mark the final stage of the orogeny.

| THE KRÅKFJORD ECLOGITE COMPLEX
The Kråkfjord complex forms an $1200 Â 500 m large, layered mafic-ultramafic body surrounded by migmatitic granitic orthogneiss (Figures 2b and 3).The enclosing orthogneiss contains small pods of amphibolite, some of which are clearly amphibolitized eclogite.Near the Kråkfjord complex (Berfjorden area, Figure 2a), these orthogneisses and amphibolite pods are strongly deformed (Section 3.3).The Kråkfjord complex itself is made up of a distinct interior dominated by bright green kyanite eclogite, capped by a mineralogically and geochemically distinct carapace of dark garnet-rich mafic rocks (Figure 3 and Table 1).The interior and the carapace of the Kråkfjord complex are both cut by lateorogenic pegmatite dykes (white bars in the bedrock map [Figure 3a]), up to 0.5 m wide and consisting of white quartz and beige feldspar.Kyanite eclogite is also present as small domains in the layered mafic and genetically related Brandsfjorden complex (Figure 2b).
The origin and igneous protolith age of the Kråkfjord eclogite complex have hitherto remained unresolved nor has its Sm-Nd age of eclogite metamorphism (Section 2.4) been corroborated by other methods, such as U-Pb dates of zircon.Provenance ages of detrital zircon in metasedimentary pockets in the carapace (Section 3.2) could provide further information on the origin of the complex.Moreover, partial melt patches in amphibolitized portions of the eclogite could potentially also date its exhumation.These are the targets of the present study, with the purpose of shedding light on the occurrences of eclogite in the northern WGR and, by implication, on the tectonic architecture of the WGR.

| The interior
The interior of the Kråkfjord complex forms an oblong layered body $750 Â 400 m large, exposed on three topographic hills (Figure 3a,b).Layers of medium-to coarse-F I G U R E 3 Overview of the Kråkfjord eclogite complex.(a) Bedrock map with locations of whole-rock and zircon geochronology samples (Table 1; sample KR-19 is from outside of this map area, cf. Figure 2b).Topographic contours are at 25 m intervals.(b) Photographic view from the west, showing the Kråkfjord eclogite complex as rounded hills in the centre, surrounded by Precambrian orthogneiss complex (steep cliffs).
grained bright green kyanite eclogite form the main rock type within the interior (Figure 4a,b).The next most abundant rock is garnet peridotite, which forms layers and lenses 0.5-5 m thick and are in many places spatially associated with garnet websterite.Less abundant rocks include dark green garnet clinopyroxenite, whitish zoisitite layers, light pinkish pods and layers of kyanite-garnet-bearing leucotonalite and, locally, red garnetite and garnet-plagioclase layers.The garnet-and pyroxenedominated assemblages of these rocks are relatively well preserved, but amphibolitized varieties also occur.In detail, most rocks show evidence of partial hightemperature re-equilibration after the eclogite facies event, mainly in the form of symplectitic coronas.Local deformation structures are typically associated with amphibolitization and include gentle folding of layers, boudinage of competent layers such as garnet peridotite and discrete shear zones.In places, a preferred orientation can be noted for HP minerals (e.g., porphyroblastic kyanite), which generally is concordant with the compositional layering.

| Kyanite eclogite
Well-preserved kyanite eclogite is bright green with blue kyanite and pink garnet (Figure 4a).The eclogite facies assemblage consists of clinopyroxene, garnet, kyanite, locally magnesio-hornblende and accessory rutile (Figure 3b).Veins of coarse-grained kyanite, quartz and hornblende, a few centimetres wide, are present locally.Very fine-grained decompression-related symplectites of sapphirine + anorthitic plagioclase + corundum ± spinel partly to completely replace kyanite (Johansson & Möller, 1986).Exsolution-like blebs in clinopyroxene consist of sodic plagioclase and, locally, quartz.Clinopyroxene and magnesio-hornblende grains are rimmed by symplectite of orthopyroxene and sodic plagioclase.Locally, very fine-grained symplectites, consisting of orthopyroxene + plagioclase + spinel, are present along the rims of garnet.Due to the relatively low Na contents of the rock and the (partly re-equilibrated) clinopyroxene, the kyanite eclogite was originally termed 'HP granulite' (op.cit.).

| Garnet peridotite
A layer of garnet peridotite, up to c. 5 m thick, is continuous for $500 m in the southern part of the interior (Figure 3a).In the central interior, the same type of garnet peridotite forms up to metre-thick lumps.Garnet peridotite is typically brownish-weathered and, where well preserved, contains up to 1.5 cm large knots of dark red garnet surrounded by dark rims (Figure 4c).This rock is dominated by olivine, garnet, clinopyroxene, and accessory apatite and opaques (Figure 4d).Dark green symplectite, consisting of two pyroxenes + spinel, separates garnet and olivine.

| Garnet websterite
Half a metre-thick layers and lumps of garnet websterite occur next to garnet peridotite in the central part of the body.The garnet websterite is comparably light-coloured and consists of medium-to coarse-grained bright red garnet, light brown orthopyroxene and light green clinopyroxene (Figure 4e).Locally, this rock contains 2-5 cm thick bands composed almost solely of bright red garnet.The garnet websterite has an equilibrium texture (Figure 4f) and was therefore used by Johansson and Möller (1986) to estimate the P-T conditions for the HP assemblage.The application of the garnet-clinopyroxene Fe-Mg exchange thermometer calibration by Ellis and Green (1979) and the Alin-orthopyroxene calibration by Harley and Green (1982) yielded an estimate of 870 ± 50 C and 14.5 ± 2 kbar.A Sm-Nd mineral-whole-rock isochron age of 432 ± 6 Ma was also retrieved from the same sample (Dallmeyer et al., 1992).

| Garnet clinopyroxenite
An outcrop of dark green garnet clinopyroxenite was found in a blasted building site in the western marginal part of the interior domain.It is coarse-grained and dominated by red garnet and dark green clinopyroxene with minor quartz, accessory rutile and opaque minerals, and locally olive-green hornblende at the rim of clinopyroxene.

| Kyanite-garnet leucotonalite
Fine-grained light pinkish leucocratic layers and pods occur scattered throughout the interior domain (Figure 4g).There is no obvious discordance to the layering of the mafic-ultramafic rocks and the rock contacts appear sharp.The leucotonalite is made up of plagioclase, quartz, garnet, dispersed grains of kyanite (Figure 4h), and accessory rutile and zircon.The kyanite grains are irregular in shape and mantled by plagioclase.

| The carapace
The carapace of the Kråkfjord eclogite complex is best exposed along the rocky parts of the coastline including along Kråkfjorden (Figure 3).It is dominated by dark green Fe-Ti-rich eclogite, rich in up to 10 mm garnet porphyroblasts (Figure 5a).The rock is layered on the centimetre to metre scale, shown mainly by variations in the amount and size of garnet crystals and the amount of quartz.Locally, the mafic rocks are complexly intermingled with and deformed together with layers and pockets of aluminous quartzofeldspathic gneiss.In some outcrops, the mafic rocks form partly amphibolitized pods within the felsic gneiss (Figure 5b).Both rock types have undergone the same style of small-scale tight folding with leucosome veinlets located along axial surfaces.The high-temperature, post-eclogite, amphibolite facies overprint makes it difficult to evaluate whether the original contacts between the mafic rock and gneiss are depositional, igneous or tectonic.However, some lobate contacts resemble pillows, which would suggest primary depositional contacts (Figure 5c).

| Fe-Ti-rich eclogite
Fe-Ti-rich eclogites are made up of dark red garnet, varying in size between 2 and 10 mm, dark green clinopyroxene, quartz, and accessory rutile, apatite and zircon (Figure 5d).Omphacite is broken down to form coarsely symplectitic intergrowths of clinopyroxene, secondary plagioclase and lesser amounts of orthopyroxene (Figure 5e).Retrograde replacement of clinopyroxene by hornblende is variable on the metre scale and typically more developed where the rock is deformed.We use the term retro-eclogite for a rock in which hornblende has largely replaced clinopyroxene.In some places, the Fe-Ti-rich retro-eclogites also show variable degrees of partial melting.At the shore of Kråkfjorden, small pockets of leucosome contain skeletal clinopyroxene megablasts (Figure 5f), interpreted as a peritectic mineral.

| Aluminous migmatitic gneiss
The quartzofeldspathic gneiss, occurring locally in the carapace, is migmatitic and typically small-scale folded with leucosomes oriented along axial plane-parallel surfaces (Figure 5g).The rock contains quartz, plagioclase, K-felspar, biotite, garnet and kyanite (Figure 5h) along with accessory rutile, zircon and apatite.As discussed further in Section 5.4, the rock contains detrital zircon cores, demonstrating a sedimentary origin.The higher abundance and larger size of garnet, the presence of aluminium silicate and the absence of hornblende make this rock distinctly different from the host Precambrian orthogneisses outside of the Kråkfjord complex (Section 3.3).

| WHOLE-ROCK GEOCHEMISTRY AND Sr-Nd ISOTOPE COMPOSITIONS
Analytical methods are reported in Appendix S1.Table 1 and Figure 3a provide an overview of the analysed samples and their locations.Chemical abundances of major and trace elements are listed in Table S1 and Nd-Sr isotope data in Table S2.

| Major elements
Major element abundances for the Kråkfjord body are plotted in diagrams in Figure 7a-h.All diagrams show distinct geochemical differences between the Fe-Ti-rich eclogites in the carapace and the layered mafic and ultramafic rocks in the interior part of the complex.The Fe-Ti-rich eclogites fall near average normal mid-ocean ridge basalt (N-MORB; Gale et al., 2013) but have slightly lower Si contents and higher Fe contents (consistent with seawater alteration, Staudigel & Hart, 1983; cf.Section 6.1).
The interior mafic and ultramafic rocks have significantly higher contents of Mg and Ni and lower incompatible element contents than the Fe-Ti-rich eclogites.None of the studied sample plots within the depleted abyssal peridotite fields; instead, they fall on fractionation trends from the Fe-Ti-rich eclogites (Figure 7a-f).In particular, the garnet peridotites have elevated Al and low Mg and Ni contents relative to abyssal peridotites, which suggests that their protoliths are troctolitic cumulates rather than depleted upper mantle residues (Paulick et al., 2006).
Their major element trends indicate that they are related to each other by fractionation of olivine and plagioclase (Figure 7a-f).The differences are particularly noticeable for the garnet peridotites, as would be expected given their high olivine concentrations.These characteristics suggest that the interior rocks are cumulates and that the kyanite eclogites originated as troctolitic cumulates.The kyanite-garnet leucotonalite in the interior has considerably higher contents of Si, along with lower contents of Ca, Mg and Fe, than the mafic and ultramafic rocks.

| Trace elements
The Fe-Ti-rich eclogites in the carapace are also clearly distinct from the interior rocks in their trace element geochemistry.All Fe-Ti-rich eclogites have trace element concentrations (Figure 7i) very similar to those of average global N-MORB of Gale et al. (2013).They show a small range of La/Sm values (0.98-2.3) that are equal to or only slightly higher than N-MORB (0.8-1.5; op.cit.).They generally lack the typical geochemical signatures of subduction zone magmas (Brenan et al., 1995;Miller et al., 1994) such as high Pb/Ce ratios, enrichment of the large ion lithophile elements (including Rb and Ba) and depletion in the high field strength elements (HFSEs, including Nb, Ta, Hf and Ti).Among the Fe-Ti-rich eclogites, two samples have slightly lower Nb than the others (Nb < 2 ppm), which are plotted separately as 'Fe-Ti Low Nb' in Figure 7j.Despite its evolved major element composition, the trace element concentrations of the kyanite-garnet leucotonalite from the interior part of the complex match closely with those of the Fe-Ti-rich eclogites.This similarity suggests a related origin between the leucotonalite and the Fe-Ti eclogites.
In contrast to both the leucotonalite and the Fe-Ti eclogites, the interior eclogite layers have very low trace element concentrations that plot well below the N-MORB levels, with Th and U below detection limits.For example, their Yb contents vary from 0.27 to 0.89 ppm.They do plot, albeit with considerable scatter, near the composition of a strongly depleted mantle of Salters and Stracke (2004;Figure 7i).This similarity is also shown on a depleted mantle normalized trace element diagram (Figure 7k), where most interior mafic and ultramafic rocks plot also close to olivine gabbro and olivine-rich troctolite, both of which occur as intrusive cumulate rocks below N-MORB-type tholeiitic basalts in a segment of the slow-spreading Mid-Atlantic Ridge (the Atlantis Massif, 30 N Mid-Atlantic-Ridge; Godard et al., 2009).The olivine gabbro and olivine-rich troctolite broadly encompass and mimic the shape of the trace element patterns of the interior eclogites, including the peaks at Pb and Sr, supporting our conclusion that the rocks of the Kråkfjord eclogite complex could have been generated in a slow-spreading mid-ocean ridge (MOR) environment, with the Pb and Sr anomalies resulting from seawater alteration.

| Sr-Nd isotope geochemistry
Figure 7l plots the present-day Sr and Nd isotope ratios of clinopyroxenes from the Kråkfjord eclogites and garnet peridotites that were also measured for major and trace elements (Tables 1 and S2).Again, there is a distinct difference in these ratios between the Fe-Ti-rich eclogites and the interior rocks.The Fe-Ti-rich eclogites have generally lower 143 Nd/ 144 Nd ratios (<0.51313) and strikingly higher 87 Sr/ 86 Sr (>0.7064) than all of the interior rocks (>0.51314 and <0.7046).There is some scatter to the data but, overall, the εNd values of the Fe-Ti eclogites resemble those of global N-MORB while the interior eclogites have slightly higher and much more varied εNd values.The hornblende separate, run in addition to clinopyroxene from Fe-Ti-rich eclogite sample KR-21, has a significantly higher 87 Sr/ 86 Sr (0.70723) than all other samples, but its εNd is within the range defined by the other samples.

| ZIRCON U-Th-Pb GEOCHRONOLOGY AND RARE EARTH ELEMENT (REE) CHEMISTRY
The goals for U-Pb zircon dating of samples from the Kråkfjord eclogite complex were to obtain ages of (1) protolith formation, (2) eclogite facies metamorphism, (3) posteclogite migmatization and (4) the source of detrital grains in the metasedimentary migmatitic aluminous gneiss.A total of $980 zircon crystals from six samples were documented in secondary electron (SE), cathodoluminescence (CL) and backscattered electron (BSE) images.One leucotonalite sample from the interior of the Kråkfjord complex and three samples from the carapace were chosen for secondary ion mass spectrometry (SIMS) analysis (Table 1).The kyanite-garnet leucotonalite sample KR-8 (Section 3.1; Figure 4g,h), which has a trace element geochemistry suggesting that it originated as a late acid differentiate within the mafic-ultramafic complex (Figure 7i), was selected to determine the igneous crystallization age of the Kråkfjord protolith.From the carapace, an Fe-Ti-rich eclogite with few retrogression features, CM16R-04A (Section 3.2; Figure 5d,e), was selected to date the age of eclogitization.A small sample containing leucosome with skeletal clinopyroxene megablasts, CM16R-11B (Section 3.2; Figure 5f), was analysed to date the time of leucosome crystallization, which occurred during the decompression and cooling of the Kråkfjord complex.A sample of migmatitic aluminous gneiss, CM16R-04D (Section 3.2; Figure 5g,h), was chosen primarily for dating the source terrane(s) of its detrital zircon grains and also to date its time of metamorphism.
Analytical methods are reported in Appendix S1.All SIMS data, summing to 267 U-Th-Pb isotope analyses and 91 REE analyses, are presented in Tables S3 and S4, respectively.Individual ages referred to in the text and figures are 207 Pb-corrected (Table S3); concordia and weighted average ages are reported at 2σ uncertainty (unless stated otherwise), the mean square of weighted deviates (MSWD) of the former being that of combined concordance and equivalence.Analyses of very low Pb content (≤0.6 ppm), high amounts of common Pb and/or >10% discordance (2σ) were omitted from age calculations and are listed under separate headings in Table S3.
F I G U R E 7 Geochemistry of rocks in the Kråkfjord eclogite complex (Tables 1 and S1).(a-f) Major and trace element covariance plots, suggesting that the interior rocks are related to the basaltic rocks in the carapace by fractional crystallization.Arrows mark accumulation vectors.(g) AFM diagram (Irvine & Baragar, 1971).(h) TAS alkali-silica diagram (Le Maitre et al., 2002).(i) Extended trace element patterns of Fe-Ti-rich eclogites and interior rocks, normalized to N-MORB (Gale et al., 2013).(j) Trace element patterns of the Fe-Ti-rich eclogites from the Kråkfjord carapace, normalized to N-MORB.Upper panel: two samples with Nb < 2 ppm (in red).For comparison (in grey) are Blåhø arc-type mafic rocks from the central-southern WGR (Hollocher et al., 2022;cf. Discussion, Section 6.3.2).Lower panel: two samples with Nb < 2 ppm (in red) and three with Nb > 2 ppm (in blue).For comparison (in grey) are MORB-type mafic rocks from the Blåhø Nappe (Hollocher et al., 2022).All these rocks have similar trace element patterns that differ from the arc-type rocks from Blåhø.(k) Extended trace element patterns of the interior rocks normalized to the depleted mantle of Salters and Stracke (2004).Note parallelism of the interior eclogite facies rocks patterns to olivine-rich gabbro and olivine troctolite from the slow-spreading Atlantis Massif of the Mid-Atlantic Ridge at 30 N (Godard et al., 2009).(l) Sr-Nd isotope relationships of clinopyroxenes from selected samples of Fe-Ti-rich eclogite and interior rocks (Tables 1 and S2).See main text for interpretations.Most CL-dark core domains are oscillatory-zoned or homogeneous, but some have blurred dark zoning patterns with minute light spots, typically in the innermost part of the crystals.The inner rim is distinctly CL-bright, and the outer rim is CL-dark.Commonly, the boundary between the core and the inner CL-bright rim is marked by a thin zone that is very bright in BSE images (dark and therefore not obvious in the shown CL images).Inclusions in zircon cores are apatite, quartz and sodic plagioclase, whereas inclusions in the rims are almandine-and pyrope-rich garnet that formed during high-grade metamorphism.

| U-Pb dates and U-Th contents
Spot analyses representing CL-dark core domains fall along the concordia between 498 ± 6 and 437 ± 6 Ma (Figure 8b).Four of these ages are from blurry CL-dark domains (orange ellipses in Figure 8b).The spread of the analyses along the concordia does not allow a significant concordia age calculation.Excluding the four spots of blurry dark domains, the weighted mean of 207 Pbcorrected data is 466 ± 6 Ma with MSWD = 7.4.
The 206 Pb/ 238 U ages of CL-bright inner rims range between 461 ± 13 and 416 ± 8 Ma (dashed lilac ellipses in Figure 8b).The spread of ages in the CL-bright rims precludes the calculation of a meaningful weighted average 206 Pb/ 238 U age or a concordia age.Analyses of CL-dark rims and single grains (including two domains that have sector zoning) range between 414 ± 5 and 393 ± 5 Ma.The limited spread in ages allows the calculation of a concordia age of 407 ± 3 Ma (Figure 8c; MSWD = 1.5, 95% confidence level); the weighted average age is also 407 ± 3 Ma (MSWD = 1.9, 95% confidence level).
The values of Th/U in oscillatory-zoned and CL-dark cores are 0.15-0.38,which are relatively high and indicate an igneous origin for these zircon domains (Figure 8d).By contrast, both Th contents and Th/U are low in the CL-dark rims (Figure 8a; Th/U ≤ 0.06), supporting their inferred metamorphic origin.Some of the analyses of CL-bright inner rims have very low Th concentrations (≤3 ppm; Th/U ≤ 0.11), while four are richer in Th, with intermediate values between igneous cores and metamorphic rims (Th/U = 0.06-0.18).

| REE contents
Chondrite-normalized REEs of core domains in KR-8 zircon form a narrow band on a REE diagram (Figure 8e), with a distinct positive Ce anomaly, a distinct negative Eu anomaly and a relative enrichment of heavy REEs (HREEs) ranging between 600 and 6300 times the chondrite values (Yb N /Gd N = 22-38).The one analysis of a blurry dark core domain has a similar pattern, however with higher values of light REEs (LREEs) (La, Pr and Nd; Yb N /Gd N = 97).
The CL-dark outer rim domains and single grains have flat HREE patterns (Yb N /Gd N = 3-5).Compared to the core domains, these CL-dark zircon domains also have lower values of both LREEs and HREEs and no or negligible Eu anomalies.
Analyses of CL-bright inner rim domains have Yb values in the same range as the CL-dark metamorphic rims, while LREE and middle REE (MREE) values scatter, mainly at low values.Yb N /Gd N of these ranges from 5 to 379.All spots except 81_01 show a distinct negative Eu anomaly.

| Fe-Ti-rich eclogite CM16R-04A1 (n5979)
Zircon grains from the Fe-Ti-rich eclogite in the carapace of the Kråkfjord complex (Section 3.2; Figure 5d,e) are up to 150 μm large, round to oval with smooth rounded grain boundaries; a few are irregular-shaped (210 mounted grains).The zircon crystals occur inside both garnet porphyroblasts and the matrix, the latter consisting mainly of clinopyroxene-plagioclase intergrowths (after omphacite) and some quartz.The zircon grains are dark in CL, many having a somewhat lighter core (CLgrey to CL-light; Figure 9a).Three grains have sector zoning and two show faint irregular zoning.Inclusions of silicates and oxides in the zircon grains are few: garnet (n = 3), rutile (n = 2), and in one zircon crystal (#40) sodic plagioclase (Ab 66 ) and a void.

| REE contents
REE analyses targeted both CL-dark (n = 12) and CLlight (n = 10) domains.Chondrite-normalized patterns show that the REE ratios of the different domains overlap (Figure 9d).All zircon domains are low in LREE, La-Nd having normalized values 0.04-3.3,with a distinct positive Ce anomaly but no Eu anomaly.The HREEs (Dy-Yb) show flat patterns (Yb N /Gd N = 1-3) with normalized values between 19 and 76.

| Tonalitic leucosome in retroeclogite CM16R-11B (n5976)
Fe-Ti-rich eclogite affected by amphibolitization and migmatization is well exposed along the shores of Kråkfjorden (Section 3.2; Figure 3a).Zircon grains (n = 58) were retrieved from a small hand sample of Fe-Ti-rich retro-eclogite containing tonalitic leucosome with skeletal clinopyroxene megablasts (Figure 5f).The zircon grains are rounded to irregularly oval, reaching a maximum length of 160 μm.The grains are dominantly dark or intermediate grey in CL, with some having a CLlighter core (Figure 10a).Among the imaged grains, two have CL-dark cores rimmed by lighter domains, and two have sector zoning.Only one inclusion, rutile, was found in one of the zircon crystals.

| U-Pb dates and U-Th contents
Analyses of CL-dark domains resulted in a concordia age of 408.5 ± 2 Ma (Figure 10b; MSWD = 1.4) and a weighted average age of 408 ± 2 Ma (MSWD = 1.05).The CL-grey to light domains gave an older and less precise concordia age of 417.5 ± 4 Ma (MSWD = 0.68) and a weighted average of 418 ± 4 Ma (MSWD = 0.74).Th/U values are low, ≤0.05, except for one higher value of 0.10 for a CL-light domain that has low contents of both U and Th (n5976_04_01; Figure 10c).CL-light and CL-dark domains have 11-57 and 93-1900 ppm U, respectively (Figure 10c).

| REE contents
The chondrite-normalized REE patterns of dated CL-dark and CL-light domains overlap (n = 10 and 2, respectively; Figure 10d), with La-Nd values between <0.005 and 4 and a distinct positive Ce anomaly.HREEs form relatively flat patterns with normalized values between 12 and 350 (normalized Yb N /Gd N = 3-10).The analyses show no or weakly negative Eu anomalies.

| Aluminous migmatitic gneiss CM16R-04D (n5978)
The garnetiferous felsic migmatitic gneiss that occurs locally within the carapace of the Kråkfjord complex (Section 3.2; Figures 3a and 5g,h) is interpreted as a metasedimentary rock based on its contents of garnet and kyanite and absence of hornblende.The presence of detrital zircon grains confirms its sedimentary origin and provides some information on the source region for the sediment.Separated and mounted zircon crystals (n = 186) from such a sample, CM16R-04D, are up to 180 μm long, rounded or oval, with length-to-width ratios up to 2. Core and rim domains can be distinguished in most crystals.Rims are commonly CL-dark or CL-grey, while core domains are CL-light, CL-grey and CL-dark or have oscillatory-type zoning (Figure 11a).U-Pb analysis targeted cores, rims and homogeneous crystals with the aim of dating both the initial crystallization of the detrital zircon grains and the metamorphic overprint.
Many zircon cores (n = 29) are distinctly older than $435 Ma, and most of these differ chemically from the Scandian zircon by having distinctly higher Th contents and Th/U values (0.19-1.9).Their ages range from 1991 ± 26 (one single analysis) to 437 ± 6 Ma (Figure 11d).Three pre-Scandian age groups can be discerned in Figure 11d: one Precambrian around 1.5 Ga (n = 4), another between 1.2 and 0.9 Ga (n = 15) and a third Ediacaran-Ordovician group (n = 8) between 600 and 440 Ma.Th/U values of the Precambrian cores that are 0.9 Ga or older range from 0.2 to 1.9 (Figure 11c), indicating igneous origins.The Th-U values of the Ediacaran-Ordovician aged core analyses (n = 8) scatter widely in the same diagram (Figure 11c) with Th/U = 0.03-0.94(most of them >0.19).
The REE patterns of the Ediacaran-Ordovician cores (n = 4) are subtly different from the older (>0.9 Ga) zircons.Their slightly positive Ce and negative Eu anomalies are less pronounced, and their Yb N / Gd N values are higher, 16-40.They also differ from the Scandian zircon domains, which have lower REE contents, particularly for the HREEs, where 13 analyses show flat patterns with normalized values ≤100.Yb N / Gd N values are 2-3, except two higher values of 12 and 14.They also lack or have only weak negative Eu anomalies.

| Inferences from the whole-rock and Sr-Nd isotope geochemistry
One of the more striking petrological and geochemical features of the Kråkfjord complex is the spectacularly different major, trace element and isotopic compositions displayed by the Fe-Ti-rich eclogites that form its carapace and the mafic and ultramafic rocks that make up its layered interior.The geochemistry of the Fe-Ti-rich eclogites resembles that of N-MORBs, whereas the geochemistry of the interior eclogites and garnet peridotites matches cumulate gabbroic complexes of the lower oceanic crust from the slow-spreading Mid-Atlantic Ridge (Godard et al., 2009).The interior eclogites have higher Mg and Ni and mostly lower Al and incompatible element contents than the Fe-Ti eclogites, but these concentrations are lower and higher, respectively, than they are within abyssal peridotites, which suggest that their protoliths were troctolitic cumulates rather than depleted upper mantle residues (Paulick et al., 2006).Based on the elemental geochemistry and Nd isotopes, we propose that the protoliths of the Kråkfjord eclogite complex most likely came from a slow-spreading MOR environment, with an upper layer of N-MORB-type basalt overlying intrusives and/or cumulates of gabbros and troctolites.The trace element pattern of the leucotonalite closely matches those of the Fe-Ti-rich eclogites, which suggests that this rock originated as a late acid differentiate within the layered complex, comparable to, for example, plagiogranite in the Troodos Massif (Marien et al., 2019).This match suggests that dating the leucotonalite will also date the extrusion of the Fe-Ti eclogite protolith basalts.
The Fe-Ti eclogites show signatures of seawater alteration, given their enrichment in U (Figure 7i) and elevated 87 Sr/ 86 Sr ratios (Figure 7l).The absence of K, Sr and Rb enrichment in the Fe-Ti eclogites (Figure 7i) does not rule out seawater alteration.While seafloor alteration can substantially influence the Sr isotopic composition of oceanic crust, it may not significantly change its Sr content (Elderfield et al., 1999).K and Rb are generally added to the altered oceanic crust at low temperatures, but are lost at higher temperatures (Staudigel, 2014).Metamorphic processes deep in the crust can also lead to elevated 87 Sr/ 86 Sr ratios (in some cases >0.72 and up to 0.74 in eclogites from the WGR, Brueckner, 1977;Vrijmoed et al., 2006).However, the Sr isotope ratios of all the Kråkfjord samples fall within the range of seawater throughout the Palaeozoic and late Proterozoic (<0.7093, McArthur et al., 2020).Cambrian-Ordovician seawater had an estimated 87 Sr/ 86 Sr isotopic ratio of $0.709 (Denison et al., 1998;Ebneth et al., 2001).Therefore, it is also reasonable to find similar gradients in Sr isotopes between Kråkfjord (from $0.706 in the exterior to $0.704 in the interior) and present-day altered oceanic crust (Staudigel et al., 1995).The Nd isotopes in the Kråkfjord clinopyroxene samples also show considerable variation in their 143 Nd/ 144 Nd ratios (Figure 7l), and these differences are not readily explained by seawater alteration because Nd is relatively insoluble in seawater.
A recent isotopic study of clinopyroxenes of gabbros from a single core of the lower crust of the Atlantis Massif at ≈30 N (Lambart et al., 2019) showed variations in Sr and Nd ratios equal to or greater than those plotted in Figure 7l. 87Sr/ 86 Sr ratios in the core varied between 0.70217 and 0.70791, and 143 Nd/ 144 Nd ratios ranged between 0.512800 and 0.513235 ($8.5 ε units).These variations encompass most of the range of Mid-Atlantic Ridge basalts and are attributed to heterogeneities in the depleted source mantle (Lambart et al., 2019), and the same explanation could apply to the Kråkfjord complex.If so, the Sr and Nd isotope variations of Kråkfjord clinopyroxenes might be better explained as a combination of heterogeneities in the source mantle and seawater alteration.

| Kyanite-garnet leucotonalite
The textural and chemical characteristics of the CL-dark and oscillatory-zoned core domains in the kyanite-garnet leucotonalite (Figures 4g,h and 8) suggest that they are of igneous origin, formed during the primary crystallization of the leucotonalitic melt (Figures 4g,h and 8).The zircon dates range from 500 to 440 Ma (Figure 8b), that is, are Cambro-Ordovician in age, but the spread along the concordia unfortunately precludes calculation of a precise age.Some of the youngest core domains have blurred zoning, possibly indicating isotopic disturbance.Consequently, the data suggest that the primary age of the leucotonalite is older than 470 Ma and nearer to 500 Ma.
The origin of inner CL-bright rim domains (Figure 8a) is enigmatic.These domains proved difficult to analyse, partly because of their thinness but also because some have exceptionally low contents of U, Th, Pb and REEs.The U-Pb spot dates range between those of the igneous and metamorphic domains (Figure 8b).Immediately inside of some of these CL-bright inner rims is a micron-wide BSE-bright (CL-dark) seam, indicating enrichment of heavy elements.Whatever the process for formation of the CL-bright inner rims, it must have taken place after the igneous zircon crystallization and prior to the crystallization of the Scandian CL-dark outer rims.
The CL-dark outer rims are dated at 407 ± 3 Ma.These domains have textural and chemical characteristics, including low Th/U, typical of subsolidus metamorphic zircon (e.g., Rubatto, 2017) as well as of zircon formed during crystallization of partial melt under metamorphic conditions (e.g., Andersson et al., 2002).The low contents of HREE reflect the coexistence of garnet in the rock.The Eu anomalies are much weaker than those of the igneous cores, indicating a low amount or absence of coexisting plagioclase.The loss of plagioclase was the result of either eclogite facies conditions or partial melting.The mantles of plagioclase around kyanite grains also indicate that plagioclase was absent in the rock at some stage and thereafter formed by a kyaniteconsuming reaction, likely during decompression.The rock contains no distinct leucosomes (nor a mesosome), but considering the high-temperature metamorphic conditions and the leucocratic composition of the rock (Table 1), it is likely that it underwent partial melting, perhaps to a high degree.We suggest that the CL-dark outer zircon rims in the leucotonalite crystallized during an early stage of high-temperature exhumation, either from partial melt (which we consider likely) or by a subsolidus reaction.

| Fe-Ti-rich eclogite
Taken together, 37 analyses of zircon crystals in the Fe-Ti-rich eclogite yield a statistically valid concordia age of 420 ± 2 Ma (Figures 5d,e and 9).The textural and chemical characteristics of these crystals are typical of metamorphic zircon, including the low Th/U values.Low HREE contents with flat patterns and the absence of a Eu anomaly in zircon suggest the coexistence of garnet and the absence of plagioclase, respectively, that is, typical of eclogite facies zircon.Taken separately, the concordia ages of CL-light and CL-dark domains nevertheless indicate a small age difference, at the limit of the analytical resolution: The CL-light domains yield 421.9 ± 2.2 Ma, slightly older than the 416.7 ± 2.4 Ma age of the CL-dark (Figure 9b).The CL-light domains consistently occupy the cores of the zircon crystals (Figure 9a), which supports an actual age difference.An age difference is also in line with the two metamorphic stages recorded by the rock, eclogite facies metamorphism followed by high-temperature decompression (Section 3.2).The U-Pb SIMS data were collected during one uninterrupted session, eliminating possible session-to-session bias and further supporting a real age difference; there are also slight variations in the REE contents with the CL-light zircon domains having on average slightly lower Ce and higher Gd and Dy (Figure 9d).Therefore, we cautiously suggest the 421.9 ± 2.2 Ma age of the CL-greylight cores as the age of eclogitization, while the 416.7 ± 2.4 Ma age of CL-dark domains is interpreted to record somewhat later zircon growth, during either eclogite facies conditions or the early decompression, the latter involving partial consumption of rutile and omphacite.

| Tonalitic leucosome in retro-eclogite
The CL-dark zircon domains in the leucosome-rich sample CM16R-11B (Figures 5f and 10) are interpreted to date zircon crystallization from partial melt at 408.5 ± 2 Ma, during the high-temperature metamorphism that followed the eclogite facies stage (Sections 2.4 and 3.3; Figures 5f and 10).The CL-lighter domains, many of which occupy the zircon cores, gave a slightly older (and less precise) date, 417.5 ± 4 Ma.We find it difficult to assess the significance of this latter date.Most likely, the rock hosted eclogite facies metamorphic zircon crystals (just as the sample of Fe-Ti-rich eclogite, Figure 9) before undergoing decompression-related partial melting.For this reason, there is a possibility that the older zircon group includes inherited eclogite facies zircon, although this is difficult to prove.In both types of domains, the chondrite-normalized values of HREE are slightly higher than those of the eclogite sample CM16R-04A1, likely reflecting a lower modal content of garnet.The lack of or weakly developed negative Eu anomaly may be attributed to either eclogite facies conditions or that most plagioclase was still dissolved in the partial melt as the zircon crystallized.If the latter interpretation is correct, the zircon crystallized out of the melt shortly before the final melt completely crystallized.

| Migmatitic aluminous gneiss
The existence of zircon cores with diverse ages verifies our initial interpretation that the aluminous migmatitic gneiss originated as sediment (Figures 5g,h and 11).Cores and single grains of detrital origin were the main objectives during analysis; still, the total number of obtained concordant dates of detrital zircon is limited (n = 29), and the majority of spots identified zircon domains with typical metamorphic chemistry (n = 70; Figure 11c).Nevertheless, a pattern can be discerned among the dated detrital grains.Precambrian zircon grains dominate with the oldest grain at c. 2 Ga, four around 1.5-1.4Ga and 15 between 1.2 and 0.9 Ga.These two latter age groups correlate well with Mesoproterozoic to Neoproterozoic granitic crust in the southwesternmost Baltic Shield (Sveconorwegia, Figure 1; age data in Bingen et al., 2021).Although the analyses are too few for statistical significance, it is likely that these zircons were derived from either SW Baltica or a micro-continent that had rifted off thereof.In addition, there is a group of eight Ediacaran-Ordovician aged zircon cores that differ chemically from the older ones (Figure 11c,e).These analyses are also too few; nevertheless, their differences in age, REE chemistry and scattered Th/U patterns indicate that they are Ediacaran-Ordovician detrital grains of igneous origin rather than Precambrian crystals that were partially reset by Scandian metamorphism.Such ages are diagnostic of rocks from the Iapetus Ocean, which formed and evolved from 600 to 440 Ma (Section 2.1) and make up most of the higher Caledonian allochthons (Figure 1a).We regard it as likely that the $500 Ma oceanic protolith of the Kråkfjord eclogite complex, as dated by zircon in a leucotonalite (Figure 8), shared environment with Ediacaran-Ordovician sediments.
Individual ages of the metamorphic domains of zircon crystals from the aluminous gneiss range between 442 ± 6 and 403 ± 5 Ma (n = 70; Figure 11b).A concordia age could not be calculated due to the age spread.Even though the weighted average age of 419 ± 2 Ma is close to the age of eclogite metamorphism, it is difficult to assign individual spot data to metamorphic recrystallization prior to or at eclogite facies conditions or to later re-equilibration in the presence of partial melt.HP metamorphism and partial melting at high and intermediate pressures both result in the stability of garnet and the consumption of plagioclase and therefore in low contents of HREE and the absence of a negative Eu anomaly in the coexisting zircon.

| Summary
The U-Pb ages of igneous zircon cores from kyanitegarnet leucotonalite suggest that its protolith is older than 470 Ma and probably formed near 500 Ma (Figure 8b).These dates provide an approximate protolith age of the Kråkfjord eclogite complex because the leucotonalite is chemically related to the mafic rocks (Section 4; Figure 7i).The aluminous migmatitic gneiss, which occurs in the carapace of the eclogite complex, contains detrital zircon cores ranging in age from ≥2000 to ≤500 Ma (Figure 11a,d), confirming that this rock is sedimentary in origin.There are too few dated detrital zircon domains for a statistically significant analysis; nevertheless, the limited data fit a source area of Precambrian igneous rocks typical of SW Norway and also cautiously indicate derivation from Ediacaran-Ordovician igneous rocks, which were typical of Iapetus and the Laurentian margin (Upper and Uppermost Allochthons in Figure 1; Section 2.1).
The eclogite facies metamorphism of the Kråkfjord complex is dated at 421.9 ± 2.4 Ma by the cores of metamorphic zircon crystals in an Fe-Ti-rich eclogite (Figure 9b).The interpretation is based on the textural context (zircon crystals occur included in garnet), the Th-U content and the REE chemistry (lack of Eu anomaly and flat HREE patterns) of this zircon.The growth of this metamorphic zircon was likely related to the metamorphic breakdown of primary Zr-bearing igneous minerals in the protolith.
Zircon in migmatites commonly allows U-Pb dating of the partial melting event (cf.Rubatto, 2017), either by in situ recrystallization/modification of zircon rims or by partial dissolution of the pre-existing zircon grains in the melt followed by precipitation of new zircon upon crystallization of the Zr-enriched melt (e.g., Andersson et al., 2002).Zircon crystals from a leucosome-rich sample within retro-eclogite gave an age of 408.5 ± 2 Ma (Figure 10b) that is significantly younger than that of the eclogite metamorphism.This age is interpreted to date zircon formation in a partial melt during the hightemperature decompression following the eclogite facies stage (garnet stable; plagioclase consumed by partial melt).The 407 ± 3 Ma age obtained from CL-dark zircon rims in a leucotonalite (Figure 8c) is also interpreted to reflect zircon crystallization during high-temperature decompression, either at subsolidus conditions or from a partial melt.We find the latter more likely because of the likelihood of partial melting of this leucocratic rock.Finally, the metamorphic zircon domains in the aluminous migmatitic gneiss range from $440 to 400 Ma (Figure 11b); these dates record Scandian metamorphism, but it has not been possible to determine during which metamorphic stage the individual domains grew.

| The origin of the Kråkfjord layered igneous complex-A fragment of Iapetus
The Kråkfjord complex was originally interpreted as an intrusion within the Precambrian orthogneiss complex in Roan (Dallmeyer et al., 1992;Johansson & Möller, 1986;Möller, 1988Möller, , 1990)), that is, a part of the ancient crystalline Baltica continental margin that was metamorphosed in situ during Scandian collision.However, the data presented herein contradict this interpretation and instead suggest that the Kråkfjord complex is a fragment of Iapetan oceanic crust that was tectonically inserted into the continental crust during the 430-400 Ma Scandian collision.The main arguments for the new interpretation are (1) the Cambro-Ordovician age for the igneous crystallization of the protolith, (2) the MOR basalt (MORB) character of the protoliths of the Fe-Ti-rich eclogites and (3) the depleted geochemical character of the oceanic gabbro protoliths of the interior eclogites, typical of a slow-spreading centre, and (4) the former presence of sediment (now aluminous migmatitic gneiss) within the basaltic carapace of the complex, with detrital zircons derived from Baltica and (suggested with caution) Iapetus.
A remaining question is in which part of Iapetus this igneous complex could have originated.Below, we discuss three possible scenarios: (1) a layered mafic intrusion into Baltic continental crust, (2) a back-arc basin on the upper oceanic plate and (3) an OCT zone along southern Baltica.

| Scenario 1: A layered mafic intrusion into Baltic continental crust
A possible hypothesis for the origin of the Kråkfjord complex is as a layered, differentiated igneous intrusion into Baltic crust (suggested by Dallmeyer et al., 1992;Johansson & Möller, 1986;Möller, 1988Möller, , 1990)).This scenario invites comparison with the Skaergaard complex of eastern Greenland, which shares chemical similarities with the Kråkfjord complex.Both show the familiar ironenrichment (i.e., 'Skaergaard' or tholeiite) trend, particularly when the Kråkfjord Fe-Ti-rich eclogite protoliths are considered along with the interior cumulates (Figure 7g).The Kråkfjord complex also invites comparison with the layered, eclogitized Eiksunddal complex in the central WGR, which intruded and differentiated as a layered complex and was subsequently eclogitized during the Scandian orogeny (Carswell et al., 1983;Jamtveit, 1987a;Jamtveit et al., 1991;Schmidt, 1963).There are mineralogical similarities between these two complexes, particularly with the cumulate interior layers of the Kråkfjord complex.Both display layers of bimineralic eclogite, orthopyroxene eclogite, garnet websterite and garnet peridotite.
However, the similarities between Kråkfjord and both Skaergaard and Eiksunddal should not be overemphasized.MOR tholeiitic basalts also follow an ironenrichment trend (le Roex et al., 1982), so a MOR hypothesis for the Kråkfjord complex is not geochemically excluded.Moreover, the layered intrusion model cannot explain the changes in Sr isotope ratios within the Kråkfjord complex or the observed large differences in Nd isotopes, as crystal fractionation alone cannot generate isotopic differences.For example, the entire Skaergaard complex has limited ranges of 87 Sr/ 86 Sr ratios (0.7041 to 0.7046) and εNd (+4.1 to +5.2 for the layered series and +1.7 to +5.2 for the upper border series; McBirney & Creaser, 2003).The overall chemistries of Kråkfjord and Eiksunddal differ as well.Cumulate peridotites in the Eiksunddal complex are more ferriferous and titaniferous and less chromiferous than the peridotites from the Kråkfjord complex (Carswell et al., 1983;Jamtveit, 1987b), andJamtveit (1987b) assigns a withinplate origin to Eiksunddal magmas, whereas this study for the Kråkfjord complex indicates a MOR origin.An intraplate setting should typically lead to preferential sampling of the more enriched components in the underlying mantle, particularly in the highly incompatible elements.This enrichment is not the case for the Kråkfjord complex.
Another significant feature is the association in the Kråkfjord complex with aluminous migmatitic metasediments in its basaltic carapace, containing detrital zircons derived from Baltica (and possibly also from Iapetus).The lithological contacts between the basaltic rocks and sediments are modified by high-grade metamorphism, but the field relations do indicate primary contacts between the two (e.g., the pillow-like structure in Figure 5c).This association implies that MORB of the Kråkfjord complex was exposed on the seafloor so that it could be covered with aluminous pelitic sediments or, alternatively, that the mafic rocks intruded into a sediment cover on oceanic crust.
Finally, the $500 Ma leucotonalites of the Kråkfjord complex do not occur within the enclosing orthogneisses, indicating that the mafic/ultramafic complex must have been tectonically juxtaposed with the Precambrian orthogneisses rather than intruded into them.

| Scenario 2: A back-arc basin on the upper oceanic plate
Another tectonic setting, within Iapetus, is suggested by Hollocher et al. (2022) for the origin of Blåhø amphibolites and subordinate eclogites in the central and southern WGR (Figure 1).Based on their whole-rock geochemistry, some of which show an arc signature for the protoliths (Figure 7j), the Blåhø rocks are interpreted to have originated within a back-arc basin on the upper oceanic plate, above a west-directed subduction zone west of the Baltican margin.
The elemental chemistry of the Kråkfjord rocks is, however, different from the Blåhø arc-type rocks, with little or no HFSE depletion and large ion lithophile element (LILE) enrichment (Figure 7j).Two of the Kråkfjord Fe-Ti-rich eclogite samples have slightly lower Nb than the others (plotted separately as 'Fe-Ti Low Nb' in Figure 7j) and could potentially represent arc values.However, they have much higher Ta and Ti contents than the arc-type rocks from Blåhø, suggesting a different origin.Metamorphism could change the budget of the LILE, but it would be highly coincidental for the Kråkfjord rocks to change their composition to N-MORB through metamorphism, which would require the removal of LILE and Pb from the system without removing U.

| Scenario 3: An OCT zone along southern Baltica
A third scenario for the origin of the Kråkfjord complex, which we favour, is an oceanic spreading ridge near to or within a Cambro-Ordovician hyper-extended OCT zone along southern Baltica (Figure 12).Andersen et al. (2012Andersen et al. ( , 2022) ) and Jakob et al. (2017Jakob et al. ( , 2019) ) propose that ribbons or microcontinents of Baltica crust were separated from the Baltica mainland by hundreds of kilometres of hyperextended crust during the Late Cambrian-Ordovician.The crystalline thrust sheets of the Middle Allochthon of southwestern Norway, for example, the Jotun and Lindås nappes, are thought to represent these ribbons or microcontinents.Beneath the Middle Allochthon is an 'extensional mélange' consisting of serpentinite, serpentinite clasts and both deep and shallow water clastic sediments, the latter typified by conglomerates (Andersen et al., 2012).Mafic intrusions and basalts occur only sparsely in these units, but pegmatitic metagabbro, metadiorite and augen gneiss record Early Ordovician, 480-470 Ma, igneous ages (Jakob et al., 2017).This mélange probably evolved in one or more of the ocean basins that comprised the OCT zone.
We suggest that the oceanic Kråkfjord complex might have originated within a spreading centre close to (or possibly within) this hyper-extended continental margin (Figure 12), as indicated by the bulk geochemical data and the age of leucotonalite, in an environment characterized by voluminous mafic magmatism.The Kråkfjord and Brandsfjorden complexes (Figure 2b) are km 2 -sized layered mafic rocks; this dominance of mafic rocks differs from the magma-poor extensional mélange in south-central Norway.Serpentinite and coarse clastic sedimentary rocks such as conglomerates are also absent in these Roan eclogite complexes.A further argument in favour of a spreading ridge related to an OCT zone rather than an upper plate back-arc basin setting (Scenario 2) is the intercalation of Precambrian continental crust with the Kråkfjord and Brandsfjorden oceanic complexes.This structural repetition is feasible for a shortened OCT zone in the lower tectonic plate but tectonically more complicated for sequences derived from the upper oceanic plate.
This scenario for the Cambro-Ordovician Baltica-Iapetus transition along southern Baltica is analogous to the Mesozoic Tethys Ocean along southern Europe, featuring a magma-poor hyper-extended continental margin with transitions outward into oceanic lithosphere.The Liguro-Piemont oceanic domains that occupy the Penninic Zone of the European Alps host km-scale slices of mafic and ultramafic rocks, including MOR-type basalt and gabbro-cumulate bodies typical of slow-spreading ridges (Agard & Handy, 2021).

| Caledonian metamorphism in the northern WGR
This study provides U-Pb zircon dates for the eclogite metamorphism as well as the following tectonothermal evolution.Below, we place these dates in their tectonometamorphic context based on the metamorphic and structural records.

| Eclogite metamorphism
Several of the mafic rocks in the Kråkfjord complex provide evidence of metamorphism in the eclogite facies: the garnet peridotite, garnet websterite, garnet clinopyroxenite, kyanite eclogite and Fe-Ti-rich eclogite (Section 4; Figures 4   and 5).Published P-T estimates made with conventional geothermobarometry in the late 1980s (Johansson & Möller, 1986;Möller, 1990); until new calculations are performed, we regard these estimates as indicative only.The garnet + orthopyroxene + clinopyroxene assemblage in a sample of garnet websterite gave 870 ± 50 C and 14.5 ± 2 kbar, which is at the boundary between the eclogite and HP granulite facies fields (Johansson & Möller, 1986).There is, however, a possibility that this estimate records re-equilibration between garnet and orthopyroxene during decompression (cf.Carswell et al., 2006).Importantly, the textures suggest that plagioclase in the mafic rocks of the Kråkfjord complex is secondary and a product of decompression and breakdown of omphacite, kyanite and, to a lesser extent, garnet.
The Kråkfjord and Brandsfjorden mafic complexes in Roan (Figure 2b) are to our knowledge the only known eclogite-bearing bodies in the northern WGR.This scarcity raises the question of whether the eclogite metamorphism of the Kråkfjord and Brandsfjorden complexes took place before their tectonic emplacement into the Precambrian Baltica gneisses.The metamorphic textures and peak-pressure assemblages in metadolerites within the Precambrian gneiss complex in Roan are different partly because metamorphic plagioclase was stable together with garnet and clinopyroxene (Möller, 1988).Stable plagioclase in mafic rocks suggests a peak pressure in the HP granulite facies, lower than eclogite facies F I G U R E 1 2 Sketch map (top) and vertical W-E cross section (bottom), illustrating the inferred origin of the Kråkfjord protolith at an Ordovician slow-spreading centre near continent Baltica (palaeogeography adapted from Andersen et al., 2022).The cross section outlines an extended transition zone between the continental shelf of Baltica in the east and the Iapetus Ocean in the west.In the east, the thinned Baltica crust consists of sediment-covered tilted blocks of Precambrian continental crust.Westwards, this association gradually gives way to Ordovician basaltic ocean floor with cumulate complexes.The distal part of the transition zone is capped by mafic tuffs and clay-rich sediments-a hypothetical origin for banded amphibolites, schists and paragneisses of the Vestranden Gneiss Complex.
conditions.Again, additional petrology and P-T modelling is required to test the indicated differences in peakpressure conditions.
The 421.9 ± 2.2 Ma U-Pb age of metamorphic zircon from Fe-Ti-rich eclogite (this study) clearly establishes eclogite facies metamorphism to be synchronous with the Scandian continental collision.This age is somewhat older than the 415 to 401 Ma range of metamorphic U-Pb zircon ages from eclogites in the central and southern WGR (Krogh et al., 2011, and references therein;Desormeau et al., 2015), nevertheless, within the more extended age range defined by Sm-Nd and Lu-Hf isochrons (reviews in Kylander-Clark et al., 2007, 2009).The difference between the U-Pb ages of zircon may indicate that the Baltica margin was subducted somewhat earlier in the north than farther south.Regardless, the 422 Ma age of eclogite metamorphism for the Kråkfjord complex is clearly within the 430-400 Ma period during which the protracted Scandian continental collision caused subduction of the Baltican margin beneath Laurentia.

| High-temperature decompression
Symplectitic reaction textures in Fe-Ti-rich eclogite, kyanite eclogite and garnet peridotite within the Kråkfjord complex record decompression at high temperature following the peak-pressure stage (Figures 4a-d and 5e; Johansson & Möller, 1986;Möller, 1988), similar to the decompression textures in eclogites within the central WGR (e.g., Engvik et al., 2018;Krogh, 1982).They are the result of partial exhumation, probably rapid, which transported the complex from eclogite to HP granulite/upper amphibolite facies conditions.The high-temperature overprint at intermediate to low pressures of the WGR, which also involved partial melting in the central and northern areas (e.g., Ganzhorn et al., 2014;Gordon et al., 2016;Möller, 1988), took place during Devonian regional extension and orogenic collapse, after the main collision (Section 2.4; cf.Wiest et al., 2021).Prior to this regional extension, either thrusting or tectonic extrusion must have intercalated the oceanic Kråkfjord and Brandsfjorden eclogite complexes with the Precambrian Baltica basement rocks.
Where hydrous fluid was available, the rocks in the northern WGR underwent partial melting, as evidenced by abundant partial melt structures (patch, dilation, net and stromatic migmatites; definitions by Sawyer, 2008).These migmatite features are present in most rocks: in the Precambrian orthogneisses and metabasites (Figure 6), in the carapace of the Kråkfjord complex (Figure 5) and in the amphibolites and paragneisses of the Vestranden Nappe Complex.The majority of partial melt structures are associated with amphibolitization, and the leucosomes are devoid of peritectic minerals (in contrast to the leucosome with peritectic clinopyroxene dated in this study, Figure 5d), suggesting that partial melting was triggered by the infiltration of hydrous fluid.Zircon in a leucosome (Figure 5f) in the Fe-Ti-rich retroeclogite in the carapace of the Kråkfjord complex yielded a concordia age of 408.5 ± 2 Ma (Section 5.3; Figure 10), distinctly younger than the 421.9 ± 2.2 Ma age of eclogite metamorphism.An identical age of 407 ± 3 Ma was obtained from outer zircon rims in garnet-kyanite leucotonalite (Section 5.1; Figure 8).Our preferred interpretation is that the $408 Ma ages date approximately partial melting and/or high-temperature recrystallization at the intermediate pressure conditions (garnet stable) that followed the eclogite facies metamorphism.These ages corroborate the oldest of the $410-400 Ma zircon ages obtained from leucosomes in the Precambrian granitic host gneisses by Gordon et al. (2016).

| The structural record
We propose that the contact between the Kråkfjord complex (including its pockets of aluminous migmatite in the carapace) and the host Precambrian orthogneiss complex is tectonic, not igneous.However, such an early-formed tectonic contact is very difficult or impossible to identify because of the regional-scale high-temperature metamorphic and deformational overprint that followed the insertion of the complex.Most of the strongly deformed, isoclinally folded and refolded orthogneisses near the Kråkfjord complex (Figure 6b-f) are amphibolitized and migmatized, suggesting that these structures developed during post-eclogite exhumation.They clearly overprinted and modified whatever pre-existing structures might have existed, including the postulated tectonic contact between the eclogite complex and host orthogneiss.It is possible that the earliest recorded structures (i.e., S1-S2 foliations and F2 isoclinal folds; Figure 6b,c) were formed during this proposed imbrication, but this hypothesis is difficult to prove.Similar relations apply on a regional scale to the structures in the northern WGR (Figure 2a).Allochthons (Figure 1).They originated as Iapetus oceanic arcs and ophiolites (the Upper Allochthon) and intrusions within the overriding Laurentian margin (the Uppermost Allochthon; cf.Section 2.1).However, sparse Ordovician intrusive rocks do occur at a lower tectonostratigraphic level (Andersen et al., 2012;Jakob et al., 2017;cf. Section 6.3.3),thought to have originated within the OCT zone that fringed southwest Baltica during the Ordovician (Figure 12).The data presented herein from the Kråkfjord complex also imply that Iapetan oceanic complexes of Cambro-Ordovician age occur among the lower plate allochthons, even within the lowermost tectonostratigraphic levels (e.g., the WGR).

| A connection between the eclogite complexes in Roan and the Vestranden Nappe Complex?
It is yet unknown whether or not the eclogitized mafic layered complexes in Roan and the voluminous layered amphibolites of the Vestranden Nappe Complex (structurally above the Roan Window, Figure 2a-c), originated within the same Iapetan environment and to what extent they shared metamorphic evolution.To our knowledge, eclogites have not been found within the Vestranden Nappe Complex.This is the main reason for, at present, keeping the Roan eclogite complexes distinct from the Vestranden Nappe Complex (legend in Figure 2a).
Layers of garnet-bearing leucocratic tonalite, reminiscent of those in the Kråkfjord complex (Figure 4g), do occur inside the layered amphibolite bodies of the Vestranden Nappe Complex (Figure 2a; but are absent in the host Precambrian orthogneisses).Geochronology and bulk geochemistry can be used to test whether these mafic complexes (and the associated metasedimentary rocks) shared their origin with the eclogite complexes in Roan and potentially give further information of the tectonic interface between Baltican basement and Iapetan allochthons.Previous attempts to date rocks within the Vestranden Nappe Complex provide some insight.A U-Pb zircon TIMS date of 430 ± 12 Ma of a rootless leucocratic granodiorite dyke in a schistose paragneiss sequence at an eastern location (Fosslia in Figure 2a; Johansson et al., 1987) was the first date confirming allochthoneity of a paragneiss sequence within the northern WGR. Gordon et al. (2016) reported 490-430 Ma ages of complex zircon grains from two samples of leucosome in paragneiss at a southern locality (Førsholman, Figure 2a), indicating an Ordovician source.Furthermore, C and Sr isotope compositions of marble occurrences indicate that they are Cambrian-Ordovician in origin (Trønnes & Sundvoll, 1995).Thus, limited data indicate the possibility that the rocks of the Vestranden Nappe Complex (Figure 2a) originated at the same time period as the Kråkfjord and Brandsfjorden complexes.The dominance of mafic rocks and the scarcity of serpentinite and coarse clastic rocks within the Vestranden Nappe Complex stand in contrast with the magma-poor extensional mélange in south-central Norway (Section 6.3).6.5.3 | Seve or not Seve?Solli et al. (1997) correlate the Vestranden Nappe Complex with the Seve Nappe Complex of the upper Middle Allochthon.The lithological association of large amphibolite bodies, high-grade schists and paragneisses, subordinate marble and rare peridotite matches that of the Seve Nappe Complex.The high-grade metamorphism and the occurrences of eclogite in the Kråkfjord and Brandsfjorden complexes also superficially resemble the eclogitebearing Seve units.However, the geochronological record of the Kråkfjord eclogite complex is quite different from that of the Seve eclogites (as also noted by Hollocher et al., 2022, for the Blåhø units in central and southern WGR, Figure 1).
In contrast to this record, the igneous protolith of the Kråkfjord complex crystallized at 500-470 Ma, $100 Ma later than the intrusion of the Seve mafic dykes.At this time, Iapetus was already fully developed and contracting, and the mafic and sedimentary Seve rocks were undergoing subduction and metamorphism under eclogite facies conditions.
In short, the collective data from the Kråkfjord eclogite complex indicate an independent evolution, different from that of the eclogite-bearing Seve nappes, encompassing (1) a Late Cambrian-Ordovician oceanic origin, (2) no Cambro-Ordovician HP/UHP metamorphism, (3) no other record of Cambro-Ordovician metamorphism, (4) eclogite facies metamorphism related to the Scandian continental collision and (5) tectonic intercalation with Precambrian Baltica basement.

| CONCLUSIONS
Geochemical and igneous U-Pb zircon data from the layered eclogite complex at Kråkfjord-and implicitly the nearby Brandsfjorden complex-in the Roan area of the northern WGR, provide strong evidence that these rocks originated within a slow-spreading ocean floor environment within Iapetus.This spreading centre was possibly situated either outside of or within the distal part of a Late Cambrian-Ordovician OCT zone along the outermost margin of Baltica (Figure 12).We hypothesize that the Cambrian-Ordovician OCT environment included rifted ribbon microcontinents of Baltica basement and that the layered amphibolites and metasedimentary rocks of the Vestranden Nappe Complex (Figure 2a) also originated within Iapetus.
The Kråkfjord and Brandsfjorden layered mafic complexes in the Roan area underwent eclogite facies metamorphism during protracted continental collision and subduction of the Baltican continental margin during the Scandian orogeny.U-Pb SIMS analysis of zircon from an eclogite dates this event at 421.9 ± 2.2 Ma, somewhat older than most eclogite ages from the central and southern WGR, indicating that the northern WGR may have subducted somewhat earlier than the south-central WGR.At some stage during collision and subduction, these oceanic complexes were tectonically transferred onto and intercalated with Precambrian Baltica gneisses.They subsequently shared a metamorphic evolution through HP granulite and upper amphibolite facies conditions, partial melting, polyphasal folding and ultimate exhumation.
The fact that eclogitized Iapetus oceanic crust exists within the northern WGR raises the possibility that similar Iapetan sequences are also present within the central and southern WGR.A few published Ordovician spot dates of zircon in eclogites and metapelites in the central WGR appear to support this hypothesis (DesOrmeau et al., 2015;Walsh et al., 2007), as does the proposed oceanic origin of Blåhø mafic rocks (Hollocher et al., 2022).These data encourage renewed investigations of eclogite associations within the WGR.

F
I G U R E 6 Field photographs of Precambrian orthogneiss inside of the Roan Window.(a) Quartzmonzonitic orthogneiss at comparably low degrees of strain and migmatization.Leucosome, some with hornblende megablasts, forms a patch to net migmatite structure (migmatite terminology of Sawyer, 2008).Photographs in (b)-(f) show deformed orthogneiss at Nes, Berfjorden (Figure 2b), next to the Kråkfjord eclogite complex and mapped as undifferentiated migmatitic gneiss.(b) Highly strained orthogneiss with fragment/enclave of darker grey fine-grained gneiss, isoclinally folded and refolded by small-scale tight folds.Isoclinal folds are indicated in red.(c) Close-up of the same outcrop as (b).Most leucosome is parallel with the gneissic foliation and refolded by small-scale tight folds (indicated by red line; F3 folds of Möller, 1988).Leucosome occurs in lesser amounts as short cross-cutting veinlets parallel with the F3 axial surfaces (op.cit.).(d) Disrupted and isoclinally folded mafic layer (former dyke?) in orthogneiss.(e) Mafic pod in migmatitic orthogneiss.The mafic rock is garnet-rich in the interior of the pod and amphibolitized and garnet-poor at the rim.(f) Late-orogenic granitic pegmatite cross-cutting the structural elements illustrated in (b)-(e).F I G U R E 7 Legend on next page.

F
I G U R E 8 Zircon geochronology of kyanite-garnet leucotonalite, sample KR-8, in kyanite eclogite, in the interior domain of the Kråkfjord complex.(a) CL images of representative zircon crystals with dark igneous cores and double metamorphic rims.Colour code for SIMS spots and ages: igneous cores (pink), CL-bright rims (lilac) and CL-dark metamorphic rims (blue).Orange dashed lines in two crystals outline domains with blurry pattern within the CL-dark igneous cores.(b) Inverse concordia diagram showing analyses of igneous domains.Data point error ellipses in this and following concordia diagrams are 2σ.(c) Inverse concordia diagram showing analyses and concordia age of metamorphic domains.(d) U versus Th in analysed zircon domains.(e) REEs in analysed zircon domains.

F
I G U R E 9 Zircon SIMS geochronology of Fe-Ti-rich eclogite in the carapace of the Kråkfjord complex, sample CM16R-04A1.(a) CL images of zircon crystals with CL-grey/light and CL-dark domains.SIMS spots and ages are marked in blue.(b) Inverse concordia diagram showing analyses and concordia ages of metamorphic CL-grey/light and CL-dark zircon domains.(c) U versus Th in analysed zircon domains; scale identical to that of Figure 8d.(d) REEs in analysed zircon domains.

F
I G U R E 1 0 Zircon SIMS geochronology of leucosome-rich sample of Fe-Ti-rich retro-eclogite, CM16R-11B.(a) CL images of zircon crystals.SIMS spots and ages are marked in blue.(b) Inverse concordia diagram showing analyses and concordia ages of metamorphic CLgrey/light and CL-dark zircon domains.(c) U versus Th in analysed zircon domains; scale identical to that of Figure 8d.(d) REEs in analysed zircon domains.F I G U R E 1 1 Zircon geochronology of aluminous migmatitic gneiss in the carapace of the Kråkfjord complex, sample CM16R-04D.(a) CL images of zircon crystals.Colour code of SIMS spots and ages: >0.9 Ga igneous cores (pink), Ediacaran-Ordovician cores (orange) and Scandian metamorphic rims (blue).(b) Inverse concordia diagram showing SIMS analyses of metamorphic domains.(c) U versus Th in analysed zircon domains.(d) Inverse concordia diagram showing SIMS analyses of detrital zircon cores.(e) REEs in analysed zircon domains.