Deep Crustal Flow Within Postorogenic Metamorphic Core Complexes: Insights From the Southern Western Gneiss Region of Norway

Viscous crustal flow can exhume once deeply buried rocks in postorogenic metamorphic core complexes (MCCs). While migmatite domes record the flow dynamics of anatectic crust, the mechanics and kinematics of solid‐state flow in the deep crust are poorly constrained. To address this issue, we studied a deeply eroded and particularly well‐exposed MCC in the southern Western Gneiss Region of Norway. The Gulen MCC formed during Devonian transtensional collapse of the Caledonian orogeny in the footwall of the Nordfjord‐Sogn detachment zone. We developed a semiquantitative mapping scheme for ductile strain to constrain micro‐ to megascale processes, which brought eclogite‐bearing crust from the orogenic root into direct contact with Devonian supradetachment basins. The Gulen MCC comprises different structural levels with distinct metamorphic evolutions. In the high‐grade core, amphibolite‐facies structures record fluid‐controlled eclogite retrogression and coaxial flow involving vast extension‐perpendicular shortening. Detachment mylonites formed during ductile‐to‐brittle noncoaxial deformation and wrap around the core. We present a sequential 3‐D reconstruction of MCC formation. In the detachment zone, the combined effects of simple shearing, incision/excision, and erosion thinned the upper crust. Internal necking of the ductile crust was compensated by extension‐perpendicular shortening within the deep crust and resulted in differential folding of distinct crustal levels. We identify this differential folding as the main mechanism that can redistribute material within solid‐state MCCs. Our interpretation suggests a continuum of processes from migmatite‐cored to solid‐state MCCs and has implications for postorogenic exhumation of (ultra‐)high‐pressure rocks.


Introduction
Postorogenic metamorphic core complexes (MCCs) show that the crust reequilibrates after orogeny and an entirely new crustal template may be created (e.g., Brun et al., 2017 ;Buck, 1991 ;Coney, 1980 ;Osmundsen et al., 2005 ;Platt et al., 2015 ;Vanderhaeghe & Teyssier, 2001 ;Whitney et al., 2013). Since the classic paper by Block and Royden (1990), it seems widely accepted that MCCs form by viscous crustal flow as a mode of isostatic compensation in response to upper crustal thinning by extensional faulting. Block and Royden [1990] elegantly explain geophysical observations made around MCCs, suggesting that the flow of lower crustal material forms a dome, while the Moho remains flat. Yet, what is the actual mechanism of this flow that supposedly redistributes material within the deep crust? For hot lithospheres, it has been suggested that mid-to lower crustal channels of partially molten low-viscosity crust can form (e.g., Rey et al., 2009 ;Rey et al., 2011 ;Whitney et al., 2013). However, there are many examples of MCCs, including classical Cordilleran MCCs, where high-strain rocks are in solid state and no such channels are observed (e.g., Cooper et al., 2017;Platt et al., 2015). Further complexity arises if we consider MCCs as three-dimensional structures. In settings that involve a 3-D component of strain, like transtension, 3-D numerical experiments show extension-perpendicular flow of material (Le Pourhiet et al., 2012;Rey et al., 2017) instead of extension-parallel flow as it is commonly shown in schematic cross sections of MCCs. Therefore, we need detailed constraints on the internal architecture of deeply eroded MCCs that can help us to unravel the 3-D kinematics and mechanisms of postorogenic crustal flow.
The Western Gneiss Region (WGR) of the SW Scandinavian Caledonides (Figure 1) is an excellent place to do this. It represents the deeply eroded core of a Silurian continent-continent collision orogen where large parts of the crust were turned around during Devonian postorogenic transtension (e.g., Krabbendam & Dewey, 1998). Material metamorphosed in the orogenic root at high-or even ultrahigh-pressure  Hacker et al. (2010). Eclogite localities from Andersen et al. (1994), Krabbendam and Dewey (1998), Hacker et al. (2003), Hacker et al. (2010),  and weakly deformed Baltic Shield basement (BSB) domain after Milnes et al. (1997). Structures of the extensional detachment system are shown in green. The location of cross sections in B is indicated, and the study area is marked by a rectangle. Abbreviations: BASZ = Bergen Arcs shear zone; MTFC = Møre-Trøndelag fault complex; NSDZ = Nordfjord-Sogn detachment zone. (b) Schematic cross sections of the WGR. A-A' redrawn from Milnes et al. (1997). B-B' based on Johnston, Hacker, and Ducea (2007b) and modified after Krabbendam and Dewey (1998), Braathen and Erambert (2014), this study and interpretation of regional structural data. (c) Satellite image of the Gulen MCC in between the Solund and Fensfjorden Devonian basins. Numerous islands in the western part of the area provide continuous and excellently exposed natural cross sections in E-W and N-S direction. Image from Google Earth.
conditions was brought back to the surface in the time it took to deposit Devonian supradetachment basins (e.g., Eide et al., 2005;Templeton, 2015). The processes responsible for this crustal revolution are recorded in MCCs in the footwall of large-magnitude extensional detachments along the western coast of southern Norway (Figure 1b). The Gulen MCC, which is the southernmost of these structures, offers particular exposure conditions. Numerous islands with an orthogonal set of ideally oriented coastlines provide almost continuous high-quality exposures, which can be easily accessed by boat (Figure 1c). In this study, we provide a detailed look inside the anatomy of a particularly well-exposed postorogenic MCC in order to understand the viscous flow of the crust during transtension. We approach this issue from the micro-to the megascale to constrain the timing, quantity, and kinematics of ductile strain as well as the metamorphic conditions and deformation processes.
The most important Devonian structure in SW Norway, the Nordfjord-Sogn detachment zone (NSDZ), is strongly undulating around E-W axes. Scoop-shaped Devonian basins occupy synformal positions, while 10.1029/2019TC005708 Tectonics the WGR forms several E-W elongated culminations (antiforms) in the footwall of the undulating detachment. The portion of the WGR situated between the Møre-Trøndelag fault complex and the Hornelen basin (ca. 62°N, Figure 1a) can be seen as the largest culmination of the WGR. A magnitude smaller are the culminations in between the Devonian basins ( Figure 1b). While the entire WGR can be seen as one giant MCC (Andersen & Jamtveit, 1990;Krabbendam & Dewey, 1998;McClay et al., 1986), we find it useful to apply the MCC concept (Brun et al., 2017;Platt et al., 2015;Whitney et al., 2013) also to the second-order antiformal culminations in the footwall of the NSDZ.
The Gulen MCC represents the southernmost culmination of the WGR and consists of strongly reworked Baltic Shield basement separated from overlying Devonian basins by the NSDZ and its southern continuation, the Bergen Arcs shear zone (BASZ; Wennberg et al., 1998). Precambrian protoliths were mainly formed during the 1.0 Ga Sveconorwegian orogeny as dated by U-Pb zircon ages of granitic leucosomes from migmatites (Wiest et al., 2019) and U-Pb zircon and monazite ages of local occurrences of granulite-facies rocks (Røhr et al., 2004). The area hosts the southernmost exposures of Caledonian eclogite in the Baltic Shield basement (Winsvold, 1996). Peak metamorphic conditions of the nearby Lavik eclogites have been constrained to 700°C, 2.3 GPa (Hacker et al., 2003). Previously published Ar-Ar ages group narrowly around 394 Ma (Boundy et al., 1996;Chauvet & Dallmeyer, 1992;Walsh et al., 2013), andHacker et al. (2003) suggested rapid exhumation of the eclogitebearing WGR along the southern segment of the NSDZ.

Semiquantitative Mapping of Ductile Strain
In order to reveal strain variations across a large area, we developed a scheme for the systematic description of ductile strain in outcrops, considering four main parameters: distribution, intensity, shape of the strain ellipsoid, and kinematics ( Figure 2). The acronym DISK makes it easy to remember the parameters, which can be easily recorded in a numeric scheme for each visited locality. With the help of mobile mapping devices (e.g., Midland Valley Field Move Clino™), a large number of data points can be efficiently collected and exported to GIS software for data analysis and visualization.
The first parameter in the DISK scheme describes whether the deformation is localized, intermediate, or pervasive on the scale of 1-10 m. The intensity of the strain is recorded in five steps (very weak -very strong) plus an additional value for undeformed rocks. It should be noted that the intensity of the strain refers to the deformed portion of the outcrop, and hence, this parameter needs to be read together with the distribution parameter. For example, an outcrop with pervasive strong deformation records more bulk strain than an outcrop with localized strong deformation. The shape of the strain ellipsoid follows the common classification from L-tectonites to S-tectonites. Our classification includes coaxial deformation and noncoaxial deformation. Because shear sense indicators are sometimes ambiguous, this parameter can be furthermore improved by additionally recording the certainty of the observation with a second parameter (e.g., uncertain; certain; very certain).

Dataset Presentation
The collection of data presented in this study is prepared for use in GIS software and can be accessed online (https://doi.org/10.6084/m9.figshare.c.4697006.v1). Our field data comprise 36 eclogite localities, 1264 structural measurements, and 216 structural data that have been digitized from Winsvold (1996). Around 190 localities have been classified with the DISK scheme, and in addition we present more than 400 georeferenced and commented field photos. The trace of the foliation has been interpreted from structural measurements and high-resolution orthophotos. The same accounts for the boundaries of mapped units. Our field results are presented in a newly compiled geological map and accompanying cross sections ( Figure 3). In addition, we present the results from DISK mapping in Figure 4 with symbols for each parameter overlaying the map.   Kildal (1970); Winsvold (1996); Ragnhildstveit and Helliksen (1997); Wennberg et al. (1998); Røhr et al. (2004). Abbreviations: BASZ = Bergen Arcs shear zone; NSDZ = Nordfjord-Sogn detachment zone.

10.1029/2019TC005708
Tectonics section A) that plunges shallowly toward the west. The MCC is bound to the NW by the shallowly dipping NSDZ, which is overlain by the extensive Solund basin in the hanging wall. To the south, the MCC is bound by the steep northern BASZ. In contrast to the NSDZ, the hanging wall of the BASZ is occupied by large remnants of Caledonian allochthons of the Bergen Arcs and Devonian sediments are preserved only in an isolated fault-bounded block of the Fensfjorden basin. Bathymetric lineaments seem to indicate that the NSDZ and the BASZ merge offshore, just west of Figure 3, and form a well-defined western boundary of . Distinguishing Precambrian protoliths from Caledonian shear zone rocks (a-c) and field occurrence of Caledonian eclogites (d and e). (a) Precambrian granodiorite intruded by granitic dykes. Weak Caledonian shear strain is localized at the dyke margins where they are preferentially oriented (red arrow). In case of less-suited orientations, the foliation transects the dyke (blue arrow). Ruler for scale. (b) Coarse-grained Precambrian migmatite (left side) is transformed into Caledonian banded gneiss in a meter-scale shear zone. The thin amphibolite layers represented originally mafic boudins in migmatite. Same locality as (a). Ruler is 40 cm. (c) Localized coaxial shear zone in Precambrian migmatite exposed in three dimensions. The straight trend and the sharp northern boundary may indicate that this shear zone has localized along a preexisting brittle feature. (d) The map of eclogite localities shows the highest concentration of eclogite bodies in the western core of the MCC. (e) Eclogitized mafic dyke in Precambrian migmatite. The dyke was boudinaged during migmatization, as boudin necks are filled with leucosomes. The protolith relationship was preserved because eclogitization was static and left the migmatite unchanged. However, eclogitization liberated fluids that were consumed by the host rock and led to garnet and white-mica growth in a narrow zone around the mafic body (green arrow). Note that minor shear deformation localized later at the hydrated contact zone.

10.1029/2019TC005708
Tectonics the MCC. In contrast, there is no clear eastern boundary of the MCC, but rather a gradual transition into weaker deformed portions of the WGR (Figures 1a and 1b). This transition coincides with a rotation of structural trends from E-W in the MCC to NE-SW and the disappearance of eclogites toward the east. In E-W direction, the gneissic fabrics within the MCC appear shallowly undulating and rotate from generally easterly dip in the east to westerly dip in the west (Figure 3, cross section C). In three dimensions, thus, the shape of the Gulen MCC does not conform to a simple dome shape, but rather it must be described as a hyperbolic surface with marked N-S as well as E-W asymmetry.
Our mapping shows that the Gulen MCC comprises two fundamentally distinct structural units: first, mylonites that formed in relation to the extensional detachment system on the flanks of the MCC and, second, a high-grade metamorphic core. Low-grade retrogression overprinted higher-grade fabrics in the detachment mylonites. In contrast, the metamorphic core preserves amphibolite-facies fabrics, and low-grade retrogression is absent. The following sections will present detailed descriptions of how the crust was transforming inside this MCC.

How to Distinguish Precambrian Protoliths from Caledonian Shear Zones?
One of our main objectives was to quantify the amount of reworking of the Baltic Shield basement during the Caledonian orogeny. The distinctive feature of the Precambrian rocks is that they formed at high temperatures and involve large amounts of melt. Therefore, they are consistently coarse grained, which makes them easily discernible from rocks that experienced Caledonian solid-state shear deformation and associated grain size reduction ( Figure 4). The oldest rocks are a metasedimentary sequence that has been intruded by a large gabbro and countless mafic dykes. The metasediments have been almost entirely migmatized, and only quartzites are preserved as relicts within the migmatites. Large granite plutons intruded in the northern half of the area. A distinct belt of coarse-grained gneissic rocks with granulite facies assemblages (Røhr et al., 2004) runs along the boundary of the largest of these plutons.
We find features that indicate a considerable time gap of cooling between Sveconorwegian migmatization and the initiation of solid-state shear deformation. In low-strain areas, shear deformation is commonly localized along primary lithological heterogeneities like dyke margins ( Figure 4a) and probably preexisting brittle features ( Figure 4c). With increasing strain, these shear zones grow in width and transform the coarse-grained Precambrian protoliths into medium-grained banded gneisses, where primary lithological relationships are obscured ( Figure 4b).

Eclogites Record Processes in the Orogenic Root
Numerous eclogite localities ( Figure 4d) demonstrate that the crust within the MCC represents parts of the former orogenic root. They occur dominantly in the core of the culmination, and the concentration of eclogite bodies and pods increases from E to W. The eclogites occur mostly as isolated lenses within felsic gneisses, as it is typical all throughout the WGR. In contrast to other portions of the WGR, however, eclogitization in the Gulen MCC appears to have been a static process. There are no eclogite-facies fabrics, but preeclogite textural relationships are preserved. Figure 4e shows a mafic dyke that has been boudinaged during Mesoproterozoic migmatization. Later, the mafic dyke has been entirely eclogitized, while the migmatite remained unaltered. However, there are several centimeter-thick zones around the eclogitized dyke where white mica and garnet overgrow the migmatite fabric. These observations suggest that the eclogitization of hydrous mafic rocks liberated fluids that were consumed by the immediate host rock leading to overgrowth of eclogite-facies minerals. Phengite-quartz veins, found in many eclogites, appear to be related to the same process. Almost all of the eclogites in the Gulen MCC were retrogressed to some extent. The relation between eclogite retrogression and deformation of felsic gneisses is explained in section 6 and discussed in section 7.1.

Caledonian Shear Strain
The width of gneissic shear zones ranges from the centimeter scale to the kilometer scale. The distribution of deformation at the mesoscale classified with the DISK scheme shows no systematic variations within the MCC ( Figure 5). Localized mesoscale shear zones are restricted to low-strain domains, and most shear zones are tens of meters to kilometers wide. Therefore, variations in the distribution of shear zones become visible only at the map scale.
In the eastern part of the area, shear deformation localized in an anastomosing network of discrete shear zones cutting through little deformed protoliths. Along the N-S cross section B (Figure 3), 45% of surface exposures are shear zone rocks, but within the core (excluding the detachment mylonites), only 25% of the protoliths have been significantly deformed. Toward the west, these isolated shear zones merge and form a coherent mass of highly sheared gneisses containing only isolated blocks of undeformed protolith. Along cross section A (Figure 3), >80% of the protoliths have been transformed into shear zone rocks, and this value increases even further toward the west.
The mesoscale classification of strain intensity, shape of the strain ellipsoid, and kinematics shows systematic variations that distinguish the detachment mylonites from the core ( Figure 5). The detachment mylonites are dominated by noncoaxial top-W fabrics with a dominant S > L shape and strong deformation intensity. Strain intensity is generally increasing toward the top of the detachment. Large areas below the NSDZ mylonites are occupied by granite plutons with weak to moderate strain and a dominant L > S fabric.
Further away from the detachments, in the core of the culmination, the deformation intensity increases again and coaxial S > L fabrics dominate in strongly deformed domains. Strain intensity is mostly classified as strong, while very strong deformation is only found where rheological contrasts create particular conditions for strain localization (e.g., large eclogites contained in gneiss). Weakly deformed domains in the core show a dominant L-shape of the coaxial background strain. In general, stronger shearing appears to be Figure 5. Results from DISK mapping: Each parameter is presented by symbols overlaying the geologic map and a histogram. Note that undeformed localities have usually not been recorded with the DISK scheme and, hence, are underrepresented.

10.1029/2019TC005708
Tectonics associated with more planar fabrics, or in other words, the amount of E-W stretching correlates with N-S shortening. Noncoaxial fabrics occur to some extent also in the core of the MCC. They reflect overall topto-W shearing by sinistral and dextral fabrics at the northward-and southward-dipping flanks of the culmination, respectively.

Fabric Orientations and Folding
Across the entire MCC, we find subhorizontal to moderately plunging lineations, lineation-parallel fold axes, and a corresponding girdle of foliations ( Figure 6). For structural analysis, we distinguished four domains with distinct fabric orientations. Mylonites belonging to the NDSZ show WNW-plunging lineations and subhorizontal foliations. Mylonitic foliations are folded around NE-SW as well as NW-SE trending tight recumbent folds, consistent with vertical shortening. The mylonites of the BASZ show consistently WNWplunging lineations and steeply SSW-dipping foliations. The mylonitic foliation is folded around isoclinal folds with SW-dipping axial planes that are parallel to the foliation. Lineations and fold axes in the core follow an E-W trend and plunge shallowly toward the east and the west. Foliations are mostly steep to subvertical and folded around E-W trending tight to isoclinal upright folds ( Figure 6). These folds are observed from the outcrop scale to the map scale (compare Figure 3) and involve passive folding as well as active multilayer buckling. Fold tightness appears to be correlated to strain intensity, and fold amplitude seems to relate to shear zone width. However, it should be noted that even in an almost continuously exposed area like the one studied, the exact amplitude of such steep folds is practically impossible to constrain. Our cross section reconstruction implies amplitudes up to several kilometers in the most strongly deformed domain in the west of the MCC. It is also impossible to quantify the amount of E-W extension and N-S shortening in these kinds of rocks, but conservative estimates imply at least 70% of N-S shortening by upright folding in individual outcrops of strongly deformed gneisses.Toward the east, fabrics rotate into an orogen-parallel trend: lineations plunge mostly toward the NE and foliations dip mostly steeply toward the SE. Besides this change in orientation, the NE-SW-oriented shear zones show the same characteristics as the shear zones in the core domain. Systematic variations in fabric orientation, ductile strain characteristics, as well as distinct metamorphic evolutions suggest that the detachment mylonites of the NSDZ and the shear zones in the core represent endmembers of shear zones that formed at different structural levels of the MCC (Figure 7). The BASZ seems to be an intermediate case.

Endmember 1: Detachment Mylonites of the NSDZ
In the detachment mylonites, which belong to the NSDZ, foliations are shallowly dipping to subhorizontal, and folds have subhorizontal axial planes (Figure 7a). Most fabrics are noncoaxial and consistently show topto-W transport (Figure 7b). Another striking characteristic of the detachment mylonites is their record of progressive retrograde deformation from high-grade to low-grade conditions. Amphibolite-facies fabrics are partly preserved as relicts and comprise coarse grain boundary migration fabrics in quartz-rich lithologies, which indicate high temperatures and low flow stresses (Regime 3 microstructures, Platt et al., 2015). Yet, most samples show complex microstructures with greenschist-facies fabrics overprinting higher-grade  Platt et al., 2015). The coarsegrained fabric is overprinted by fine-grained low-temperature, high-stress (Regimes 1-2) micro shear zones with bulging and subgrain rotation recrystallization. Both fabrics show asymmetry related to top-to-W shearing. (d) Schematic sketch of vertical and coaxial core shear zone, involving simultaneous N-S shortening and E-W extension. (e) Outcrop photo of a typical core shear zone. The mylonitic foliation is subvertical, fabrics are coaxial, and alternating feldspar-and quartz-rich layers show identical rheological behavior, indicative of high-temperature deformation. (f) Photomicrograph (cross-polarized light) of quartzite from a core shear zone with a simple microstructure of coarse-grained quartz. Grain boundary migration microstructures are observed, but most grains are strain-free and abundant 120°triple junctions indicate high-temperature recovery. The absence of a clear shape or crystallographic preferred orientation in quartz and parallel mica grains indicates coaxial deformation.

10.1029/2019TC005708
Tectonics fabrics (Figure 7c). Subgrain rotation and bulging recrystallization of quartz to very fine grain sizes indicate high flow stress and low-temperature deformation conditions (Regime 1-2 microstructures). Progressive retrograde deformation appears to lead to increasing strain localization that is strongly related to fluid-rock interaction. There are large tracts of phyllonitic shear zones, both in mafic and felsic lithologies, and chloritization, saussuritization, and epidotization are abundant. Embrittlement of feldspar-rich layers is recorded by greenschist-facies quartz veins. Layers of amorphous greenish cataclasites, with thicknesses up to several meters, are parallel to the shallow-dipping foliation and indicate a progressive fault rock evolution from ductile to brittle conditions. 5.2.2. Endmember 2: Core Shear Zones Shear zones in the core of the MCC have distinct structures and record a different metamorphic evolution. Foliations are steep to vertical and folded into tight to isoclinal upright folds. Subhorizontal lineations and mostly coaxial fabrics indicate overall pure shear deformation with simultaneous E-W stretching and N-S shortening (Figure 7d). Core shear zones preserve amphibolite-facies fabrics without further retrogression. In quartz-rich lithologies, these commonly involve grain boundary migration recrystallization of quartz at high temperatures and low flow stresses resulting in large grain sizes (Regime 3 microstructures, Figure 7 f). Feldspar layers in the gneisses are behaving entirely viscously and show identical rheological behavior as quartz-dominated layers (Figure 7e). Even though the core shear zones do not show evidence for lowgrade fluid-related retrogression, there is clear evidence that they represented important fluid conduits at amphibolite-facies conditions. Fractures occur at sites of rheological heterogeneity, commonly related to extension-parallel folding, and are mineralized as quartz-feldspar veins. The importance of fluids during amphibolite-facies deformation is furthermore recorded by the relationship between shearing and eclogite retrogression, which is presented in section 6.

The Intermediate Case: The BASZ
The detachment mylonites of the BASZ show noncoaxial deformation and a similar metamorphic fabric evolution to the ones from the NSDZ, with low-grade retrogression overprinting higher-grade fabrics. Yet, there are some important differences between these two segments of the detachment system.
The northern BASZ represents an oblique dextral strike-slip segment of the detachment system (Wennberg et al., 1998) and along the southern boundary of the Gulen MCC, the shear zone dips consistently with ca. 60°toward SSW. Large domains of weakly deformed rocks, as found below the NSDZ, are absent in the footwall of the BASZ. In contrast, the BASZ mylonites show a gradual transition into core shear zones, and their fabric orientation is more similar to those observed in the core than those in the NSDZ. Folds in the BASZ indicate N-S shortening and do not show vertical shortening like the NSDZ mylonites. No remnants of the Caledonian nappe pile are exposed in between the Fensfjorden basin and the BASZ mylonites, and the Devonian sediments appear directly juxtaposed to the eclogite bearing crust. These observations suggest that the amount of excision and incision was larger along the southern bounding fault of the Gulen MCC than along its northern boundary. Incision of the strike-slip segment of the detachment has allowed parts of the high-grade core to become overprinted by retrograde strike-slip shearing along the BASZ. Yet, the same may account for the westernmost part of the NSDZ mylonites that contain retrogressed eclogites (Figure 4d).

Amphibolite-Facies Reworking of the Exhuming Crust
Pristine eclogites are preserved only in undeformed protolith domains (e.g., Figure 4e). Most eclogites were variably retrogressed at amphibolite-facies conditions. Larger eclogites within shear zones show progressive retrogression from the rim, with amphibolitized margins and eclogite in the core. The state of retrogression appears to be related to the position of the eclogites with respect to shear zones, the availability of fluids, and the size of the eclogite body. A particularly well-exposed locality on Kversøyna at the western apex of the MCC (see Figure 3 for location) shows that eclogite retrogression was fluid induced and simultaneous with shearing and folding of felsic gneisses.

Structure of the Kversøyna Locality
The 200-m-long and 100 m-wide coastal section (Figure 8) exposes strongly pervasively sheared gneisses and eclogite bodies of variable size, ranging from few meters to >100 m. The gneisses are folded around E-W trending fold axes into tight to isoclinal upright folds with undulating hinge lines. The eclogite bodies are boudinaged, indicating E-W stretching, and occupy synformal positions within the folded gneisses. A

10.1029/2019TC005708
Tectonics weakly pronounced structural asymmetry could indicate a minor component of dextral simple shear, but the gneissic fabrics are symmetric and imply dominant coaxial deformation with synchronous E-W stretching and N-S shortening, like in the rest of the core domain. High-strain zones formed at the borders of the largest eclogite body and show a marked increase of fold tightness compared to the surrounding gneisses. The folded gneisses wrap around the eclogite lenses (Figures 8b and 8c). The folds in the gneisses are generally harmonic; however, in proximity to the eclogites, they get increasingly disharmonic (Figure 8c). The synformal positions of the eclogites within the gneisses, their boudinage, and their influence on folding as well as shear zone localization indicate that the eclogites formed rigid and dense anomalies that were perturbing the else homogeneous flow of the felsic gneisses.

Fluid-Induced Eclogite Retrogression and Rheological Weakening
Even though the eclogite lenses appear as homogeneous rigid blocks at first sight, a closer look reveals that their rheological behavior has been highly variable depending on their state of retrogression. The typical pattern of eclogite retrogression is a reaction front that moved from the rim of the body toward the core (Figure 8c). Retrogressed amphibolite margins have been sheared progressively together with the surrounding gneisses and indicate that rheological weakening accompanied retrogression. The rheological contrast within the partially retrogressed mafic bodies leads to fracturing and boudinage of the eclogite domains (Figure 8c). Two conjugated sets of subvertical fractures strike roughly parallel and perpendicular to the maximum stretching direction, respectively. The extension-parallel set of fractures is mineralized with amphibole, plagioclase, and quartz (marked as veins in Figure 8c). In a large (> 20 m) eclogite, these veins are surrounded by decimeter-wide fluid alteration haloes. Within these fluid alteration zones, the eclogite has been retrogressed to undeformed amphibolite (Figure 9a). These mineralized fractures apparently represent corridors where fluids could enter the core of the eclogite and trigger retrogression, while dry domains of pristine eclogite, only centimeters apart, survived metastably until final exhumation.
On the microscale, retrogression is most advanced within fractures, where fluids have entered the rock (Figures 9b and 9c). From these fractures, the retrogression front spreads through the rock along microfractures and grain boundaries (Figure 9d). Microfractures and grain boundaries in omphacite show successive generations of symplectites with coarser symplectites in the core and finer symplectites at the rim. In some cases, symplectites are cut by fractures themselves and then replaced by amphibole (Figure 9c).
Like the retrogression at the rim of the eclogite bodies, retrogression in the core of the mafic lenses gives evidence for significant rheological weakening. The widest retrogression zone, measuring roughly 1 m in thickness, has been entirely transformed into an amphibolite-facies shear zone that cuts through the center of a large eclogite body (Figure 9a). Within this shear zone, not a single relict of the eclogite assemblage is preserved. The shear zone is cut itself by a mineralized fracture, giving evidence for brittle-ductile deformation cycles at amphibolite-facies conditions, probably caused by large rheological heterogeneity within the mafic body. This behavior is furthermore recorded by pseudotachylite formation in partially retrogressed and sheared domains of the eclogite.
Our observations show that the eclogite-bearing crust from the orogenic root was largely reworked at amphibolite-facies conditions. Shear zones appear to have served as conduits for amphibolite-facies fluids that entered undeformed domains through fractures. The availability of fluids clearly controlled the retrogression of eclogites and thereby rheological weakening that enabled viscous flow. These observations strongly resemble the processes observed in the famous Bergen Arcs eclogites (e.g., Austrheim et al., 1997), however, with the difference that they apply to retrogression and not during eclogitization.

Eclogites as Time Markers of Postorogenic Viscous Flow
Eclogites within the Gulen MCC are robust time markers for processes that took place in the orogenic root as well as those associated with exhumation. Static eclogitization and the absence of eclogite-facies deformation fabrics indicate that the crust in the southernmost part of the Caledonian eclogite province was still relatively cold and rigid and did not deform pervasively during peak metamorphism. This is quite different from other parts of the WGR where eclogite-facies fabrics are well developed, probably because the crust was buried deeper and higher metamorphic grades were reached (e.g., Andersen et al., 1994;Braathen & Erambert, 2014).
In the Gulen MCC, extensive ductile deformation localized into shear zones and in broad zones of pervasive viscous flow. We find a consistent relationship pattern of eclogites being retrogressed in shear zones, while eclogites in undeformed domains remain mostly preserved. This indicates that the shear zones formed during the transtensional collapse of the overthickened crust (Mode II extension by Fossen, 1992) and after exhumation from eclogite facies conditions, possibly through eduction (Mode I extension). Obviously, we cannot exclude the possibility that some of the shear deformation may have occurred at an earlier stage. However, fabrics show a consistent orientation from the top of the detachment to the lowermost exposed parts of the metamorphic core. The rotation of fabrics from E-W into a more orogen-parallel trend toward the East represents an exception from the rule. Yet, shear zones in this domain show a metamorphic evolution similar to that in the rest of the core, the same relation between linear and planar fabrics and, most importantly, they also contain retrogressed eclogites. Hence, the NE-SW trending shear zones are likely to be postorogenic as well, and their orientation might relate to local orogen-parallel flow during unloading of the orogen (Duclaux et al., 2007). Our interpretation of postorogenic viscous flow in the Gulen MCC is supported by available thermochronological data (Boundy et al., 1996;Chauvet & Dallmeyer, 1992 Figure 8a. Around mineralized veins, eclogite (red-green) was retrogressed statically to amphibolite (dark green to black). The largest retrogression zone in the center of the photo developed into an amphibolite-facies shear zone, which is cut again by a mineralized brittle fracture. Hammer for scale. (b) Thin section photomosaic (cross-polarized light) of partially retrogressed eclogite. Retrogression domains are easily recognized by the dark color of nontransparent symplectites and green amphiboles in contrast to the light-colored eclogite facies minerals. Retrogression is most advanced along a conjugate set of fractures (red arrows). Note how the retrogression front follows grain boundaries. (c) Close-up of the intersection of the mineralized fractures (red arrows). The larger fracture displays an amphibolite-facies mineralogy. Omphacite is surrounded by a rim of fine grained symplectites, which become coarser further away from the grain boundary and show progressive symplectite growth. Note how symplectites are transformed into amphibole, where they are cut by the small fracture. This relationship clearly demonstrates that even on the microscale, retrogression is a function of fluid availability. (d) Scanning electron microscope image of the least retrogressed eclogite domain. Retrogression advances along fractures and grain boundaries and leads to growth of submicrometer size symplectites. et al., 2013) and interpretations of previous works in the surrounding areas (Andersen et al., 1994;Hacker et al., 2003;Milnes et al., 1997).

Viscous Flow at Distinct Structural Levels: Kinematically Linked, Mechanically Decoupled
Our results from the Gulen MCC fit very well within the general framework of postorogenic exhumation of the WGR by shearing along the NSDZ during sinistral transtension (e.g., Krabbendam & Dewey, 1998). Yet, our results are not fully explained neither by simple shear detachment models (e.g., Hacker et al., 2003;Souche et al., 2013), nor by homogeneous constriction as predicted from transtension (e.g., Dewey, 2002;Fossen et al., 2013;Krabbendam & Dewey, 1998), nor by simple vertical thinning of the overthickened crust (e.g., Andersen et al., 1994). All of these elements need to be integrated in a dynamic model of detachment and footwall evolution, to explain the distinct deformation that we observe at various levels of the extending crust.
At the presently exposed level, crustal flow within the Gulen MCC is entirely solid state. The crust appears to represent a coherently stretched block with significant strain variations in three dimensions ( Figure 10). We find distinct structural levels that appear kinematically linked while mechanically decoupled. They formed within the same kinematic boundary conditions of sinistral transtension but represent different metamorphic conditions and hence different rheologies. On top, the detachment level shows simple shear with vertical thinning and the characteristic overprint of greenschist facies and semibrittle fabrics on highergrade fabrics. In contrast, the lowest level shows pervasive coaxial amphibolite-facies viscous flow of the crust reaching extreme amounts of simultaneous E-W stretching and N-S shortening. A key finding of this study is that the coaxial high-grade fabrics in the core experienced no further low-grade overprint. As such, it must represent a structural level that is distinct from the noncoaxial detachment zone with its consistent retrograde overprinting relationships. In between these two levels, we find minor remnants of an intermediate domain of localized amphibolite-facies deformation, which involves constriction and minor vertical thinning. These distinct amphibolite-facies domains resemble the localized-distributed transition within the viscous crust postulated by Cooper et al. (2017). Today, we find all these different structural levels along the same exposure level as observed along cross section A (Figure 3). Their systematic variations in metamorphic conditions and strain characteristics imply, however, that they formed at different vertical levels of the crust and were later juxtaposed during core-complex exhumation.

Inward Flow Through Differential Folding ("Stockwerk Folding")
The N-S cross sections through the WGR and the Gulen MCC show intensive extension-parallel folding (compare Figure 1B and Figure 3) in contrast to shallowly undulating, parallel fabrics seen in E-W sections. According to our mapping in the Gulen MCC and the interpretation of large-scale fold patterns along the west coast of the WGR ( Figure 1B), different crustal levels experienced apparently different amounts of N-S shortening. The detachment zone and the overlying basins show mostly open folds, although thrusts and reverse faults within the basins record locally larger amounts of syn-sedimentary N-S shortening . The WGR footwall, however, is pervasively folded into tight to isoclinal upright folds from the scale of meters to tens of kilometers. Thus, the amount of N-S shortening in the footwall of the NSDZ appears to be significantly larger than in the overlying nappes and the detachment zone itself (Braathen & Erambert, 2014). These extension-parallel folds have been related to a distinct tectonic phase of regional N-S compression (e.g., Chauvet & Seranne, 1994), but they are also adequately explained as a result of transtension (Dewey, 2002;Fossen et al., 2013;Krabbendam & Dewey, 1998). However, our observations are not in agreement with a simple model of progressive transtensional folding during exhumation. The tight upright folds formed due to strong extension-perpendicular shortening during internal necking within the amphibolite-facies crust, while the upper crustal levels experienced only weak extensionperpendicular shortening. This differential folding might just represent the actual mechanism of extension-perpendicular crustal flow that is predicted from 3-D numerical models of transtensional corecomplex formation (Le Pourhiet et al., 2012). We interpret the high-grade extension-parallel folds as the structural expression of isostasy-driven Poiseuille-type flow in the deep crust that moved low-viscosity solid-state material from the north and south into the dome and hence as a less drastic case of the mechanism postulated for migmatite-cored MCCs (e.g., Kruckenberg et al., 2011;Rey et al., 2017;Roger et al., 2015). As an interesting side note, this kind of differential folding was already described by Wegmann (1935) and Haller (1956) in the Greenland Caledonides. Distinct deformation regimes in relation to different metamorphic 10.1029/2019TC005708 Tectonics conditions in vertical levels of the crust were the essence of the "stockwerk folding hypothesis." However, as pointed out by Wernicke (2009), these authors worked within a "fixist" paradigm that did not permit major horizontal movements. Therefore, it took many years until researchers realized that this "stockwerk folding" occurred in the footwall of massive extensional detachment shear zones (Hodges, 2016;White et al., 2002).

When Distant Relatives Become Close Neighbors: A Dynamic Model for Transtensional MCC Formation
During MCC exhumation, large parts of the original crust can be removed through the combination of incision, excision, and erosion. In the final structure of the Gulen MCC, syntectonic sedimentary rocks are directly juxtaposed with deep crustal levels, as demonstrated by the juxtaposition of the Fensfjorden Devonian basin with the eclogite-bearing crust across the BASZ. During the progressive evolution, material within the MCC is moving through different metamorphic conditions and deformation regimes, for Figure 10. The internal structure of the Gulen MCC reveals a coherently stretched crustal section; however, distinct structural levels show very different deformation fabrics, kinematics, and fluids. Differences between these structural levels appear to relate to vertical metamorphic variations and horizontal strain gradients in N-S as well as E-W direction. Yet, they are all interpreted to have formed within the same kinematic boundary condition of sinistral transtension, which is represented by the divergence vector . Below the brittle crust, the detachment mylonites show retrograde deformation from amphibolite-facies to semibrittle conditions, involving noncoaxial shearing and vertical shortening. The amphibolite-facies core consists of a domain of localized deformation, which is necked by pervasively sheared gneisses. The localized domain shows weak noncoaxial shearing and minor vertical shortening. In contrast, the distributed domain is characterized by coaxial shearing and N-S shortening.

10.1029/2019TC005708
Tectonics example, ductile rocks become brittle. In addition, the thermal conditions of the system are transient. Heat advection will change the geothermal gradient during MCC evolution  and hence the spatial distribution of deformation regimes.
We incorporated these basic principles of MCC formation with existing models of postorogenic collapse of the Scandinavian Caledonides. Thereby we try to explain our field observations in accordance with the available information from other culminations of the WGR, the NSDZ, and recent work in the Devonian basins (Osmundsen & Péron-Pinvidic, 2018;Osmundsen et al., 2019). Our model is presented in Figure 11 as a

10.1029/2019TC005708
Tectonics series of N-S and E-W cross sections to illustrate the 3-D kinematics. The time evolution is portrayed in three stages, which do not represent distinct phases but are an inevitable simplification made to illustrate the incremental evolution. The model that we come up with, of course, is highly simplified. We do not consider any internal structural geometries, rheological heterogeneities, or topographic variations that were possibly there at the end of Mode I extension. Our cross sections show neither E-W extension of the entire crust nor the associated extension-orthogonal shortening of the crustal section. Furthermore, we do not show the interference of different MCCs and the rotation of the divergence vector during progressive transtension.
Our model starts with a strongly thickened, horizontally layered crust. We show three rheological layers ( Figure 11): brittle upper crust and the ductile crust separated between greenschist facies and amphibolite facies or higher grades. The crust consists of a thick pile of nappes overlaying the eclogite-bearing Baltic Shield basement (BSB). The unit boundaries serve only as passive markers for the structural evolution but do not represent rheological boundaries, except for the distinction between localized and pervasive deformation domains within the BSB. For reasons of simplicity, we overlap the greenschist-/amphibolite-facies boundary with the nappe-basement contact, but we note that extensional shearing in the lower nappes started at amphibolite-facies conditions (Johnston, Hacker, & Ducea, 2007b and references therein). 7.4.1. Stage 1: MCC Initiation Initial extension formed steep, laterally limited normal faults that grade into mid-crustal shear zones at depth. Initial necking occurred at rheological transitions and involved extension-perpendicular folding within the amphibolite layer. It should be noted that both the depocenter of the initial basin and the future MCC formed in the center of the fault, because this is the zone of maximum stretching (e.g., Kapp et al., 2008;Osmundsen & Péron-Pinvidic, 2018;Osmundsen et al., 2019).

Stage 2: MCC Formation
During the main phase of MCC formation, the MCC started to rise in the zone of maximum stretching, and the basins were displaced to the flanks of the MCC as argued by Osmundsen et al. (2019). The synchronous formation of several MCCs along-strike (not shown in Figure 11) made the basins occupy synforms in between the rising antiforms. Different processes occurred at the different levels, which are still far apart from each other. The brittle crust, comprising the upper part of the nappe pile, was exposed, eroded, and deposited in the basins. Steep faults incised at the flanks of the MCC and form steep topography, which promoted catastrophic mass movements through large landslides as observed in the Solund and Kvamshesten basins (Hartz et al., 2002;Osmundsen et al., 1998). This shows that erosion can contribute significantly to upper crustal thinning during MCC exhumation.Greenschist facies top-to-W shearing within the detachment zone itself occurred mostly in the nappes, but as the upper crust was thinning progressively, also the BSB got involved. At the same time, at a deeper level, the amphibolite-facies crust was necking internally. Lateral inward flow occurred through extension-orthogonal shortening of the pervasively deforming lowviscosity crust. This process formed the high-grade core with steep strike-slip shear zones at its flanks. Differential exhumation at amphibolite-facies conditions is a consequence of this process. Rocks originating from different depths may be juxtaposed in relation to their position along major fold structures (compare position of eclogite bodies between stages 1 and 2). The high-grade core was exhumed progressively to greenschist facies conditions, but greenschist facies deformation was restricted to the top and the flanks of the culmination. Apparently, retrograde deformation stalled suddenly within the core, high-grade fabrics became "frozen-in," and active flow moved to weaker layers.
A possible explanation for this may be that the internal necking of the ductile layer and the associated exhumation of the high-grade core represent a form of vertical heat advection. The rising core apparently contributed heat and fluids to the detachment system, which lead to weakening and increasing strain localization in the detachment layer. Furthermore, greenschist facies phyllonites, which formed previously through fluid-related retrogression within the detachment mylonites, are brought into the realm of brittle faulting. Such fault rocks with particularly low shear strength (Braathen et al., 2004), the abundance of fluids, and progressive heating within the detachment zone through head advection and possibly shear heating (Souche et al., 2013) imply that the brittle layer became weakened at this stage, promoting brittle faulting at progressively lower angles. At the same time, the rising core itself was cooling down, and therefore the active amphibolite-facies flow moved to a deeper and hotter level where the viscosity was lower. Furthermore, the possible occurrence of partial melting at a deeper, unexposed level of the MCC must be noted.

10.1029/2019TC005708
Tectonics 7.4.3. Stage 3: MCC Exhumation The "frozen-in" high-grade core was exhumed progressively through localized deformation in the weakened detachment system. Brittle faulting occurred at progressively lower dips. At this stage, the deep crust had reached the surface and became eroded itself, as demonstrated by detrital Ar-Ar cooling ages from Hornelen basin (Templeton, 2015). The basins have reached a massive size at this stage, but only minor remnants are preserved at the present erosion level, which expose the oldest strata. Reconstructions imply a maximum basin depth between 9 and 13.5 km with a geothermal gradient of 38°C/km (Souche et al., 2012;Souche et al., 2013;Svensen et al., 2001). This implies that in parts of the system, the brittle crust had been entirely replaced by supradetachment basins. Devonian sediments were directly juxtaposed to detachment mylonites, and in the case of the Solund Basin, the lowest strata of the basin were even mylonitized themselves (Seranne & Seguret, 1987). At this stage, most of the nappe pile and even parts of the basement have been removed by ductile thinning, excision, and incision. The detachment system can be envisaged as a selective rock grinder: Weak rocks were smeared out ductilely in the detachment shear zone, while strong lithologies were removed through brittle faulting, footwall uplift, and erosion. In the synforms of the detachment zone, we find mostly weak rheologies like mica schists and phyllonites preserved. At the flank of the MCC, steep strike-slip faults incised the core and exploited previously formed ductile fabrics. In the case of the BASZ, the basin was directly juxtaposed to the deep crust. Further isostatic compensation may have occurred through folding at deeper amphibolite-facies levels.
At the end of this evolution, our reconstruction implies a crustal thickness of around 40 km (10 km thick basin fill + 30 km present crustal thickness). Final juxtaposition of the different units occurred through brittle fault reactivation during North Sea rifting (e.g., Eide et al., 1997;Torsvik et al., 1992).

Conclusions and Implications for Postorogenic Exhumation of (Ultra-)High-Pressure Rocks
Our study shows that vertical metamorphic variations and lateral strain gradients during postorogenic transtension may cause a highly variable behavior of the crust, leading to differential exhumation and the formation of MCCs. We think that our field-based model ( Figure 11) encapsulates important general aspects of MCC formation in 3-D strain settings and can contribute to understand MCCs where the deep levels of the crust are not exposed. Large magnitudes of extension can amplify differential extension-orthogonal shortening of distinct crustal levels. This process results in characteristic fold structures, which resemble the historic "stockwerk folding hypothesis" (Wernicke, 2009) and might easily be mistaken for contractional structures formed during convergence. In the case of the WGR, however, they clearly formed during postorogenic transtension. Thus, like in the case of migmatite-cored MCCs (Rey et al., 2017), extension-orthogonal contraction within the deep crust is the main mechanism of isostasy-driven crustal flow also in solidstate MCCs.
This crustal flow mechanism has the potential to redistribute material within the extending crust ( Figure 11). Hence, it might be highly important for the exhumation of (ultra-) high-pressure rocks in postorogenic settings. A comprehensive look at the WGR demonstrates that this might have been the case in the Caledonides. The Gulen MCC is the southernmost of a series of evenly spaced MCCs ( Figures 1A and 1B); however, also the ultrahigh-pressure domains north of the Hornelen Basin have similar orientations, spacing, and wavelengths. If the pressure estimates from these rocks reflect burial depth, the occurrence of such discrete domains (e.g., Root et al., 2005) implies that the rocks inside these domains have been exhumed from greater depth than the rocks in the surrounding areas. We suggest that our model of crustal flow could offer a plausible explanation for this differential exhumation pattern. This would imply that the processes across the WGR have been self-similar over large areas of the giant high-pressure terrane.
Even if we zoom out to the scale of the entire orogen, we can find notable self-similarities. Already Ramberg (1981) pointed out the existence of two distinct belts of basement domes that run parallel along the entire length of the Scandinavian Caledonides. Osmundsen et al. (2005) showed that these domes can be seen as different types of MCCs. The belt in the foreland consists of cylindrical domes with long axes perpendicular to the extension direction, which comprise low-grade metamorphic rocks and record low amounts of exhumation. In contrast, the domes in the hinterland of the orogen are noncylindrical with long axes parallel to the extension direction and expose high-grade metamorphic rocks, which were exhumed from great depth. The different shapes of these domes resemble exactly the a-type versus b-type classification that has been suggested for Aegean domes (Jolivet et al., 2004). Le Pourhiet et al. (2012) attributed the coeval formation of both dome types in the Cyclades to distinct kinematic boundary conditions related to a slab tear during rollback of the Aegean slab. In the case of the Scandinavian Caledonides, however, the different dome-types appear systematically with respect to their position toward the foreland or the hinterland of the orogen. Furthermore, as they occur along the entire length of the orogen, the domes in the Scandinavian Caledonides are likely to have formed within the same kinematic boundary conditions. Thus, the different dome types apparently relate to variations in metamorphic conditions at the onset of extension, as well as different amounts of crustal stretching and associated exhumation. At least in the case of the Scandinavian domes, noncylindricality appears to be a function of crustal stretching, but we speculate that this relationship might also apply universally. During the formation of large normal-fault systems, there will always be along-strike variations, which can trigger differential crustal flow in the deep crust. This creates a feedback loop that can be amplified during progressive crustal stretching. Hence, an alternative explanation for the coeval formation of a-and b-type domes might be that noncylindrical MCCs will be exhumed from the deep crust if a critical amount of extension is reached (Stage 2 in Figure 13). A way to test this hypothesis could be to analyze systematic variations in metamorphic grade and exhumation recorded in a-and b-type domes around the world, respectively.