The 10 × 25 km Isua supracrustal belt is the oldest (3.8–3.7 Ga) supracrustal belt in the world (the older Acasta belt in Canada is gneissic and not strictly supracrustal), which mainly consists lithologically of (meta-) basaltic volcanic rocks including massive flows, pillow lavas, pillow breccias, volcaniclastic sediments, basaltic dykes and sills, gabbros and ultramafic rocks, minor felsic volcanic rocks, together with bedded cherts, magnetite-facies BIF, monomict and polymict conglomerates, and turbidites (Maruyama et al., 1991; Nutman et al., 1997a; Nutman et al., 1997b; Appel et al., 1998; Komiya et al., 1999; Nutman and Friend, 2009; Arai et al., 2015). These supracrustal rocks were intruded by a variety of tonalites and diorites, which are now mostly orthogneisses (Nutman et al., 2009).

The existence of surface water at 3.8–3.7 Ga is indicated in Isua by the presence of pillow lavas (Maruyama et al., 1991; Appel et al., 1998; Maruyama and Komiya, 2011) and of bedded sediments such as pelagic cherts and banded iron formation (BIF) (Moorbath et al., 1973; Nutman et al., 1997b). According to Maruyama and Komiya (2011) the total thickness of the largest unit of pillow basalts in Isua is up to 1.0 km, suggesting that the ambient ocean depth was at least 1.0 km; from the presence of vesicular basalts de Wit and Furnes (2013) suggested a depth of up to 2–4 km. By using the source mantle potential temperature of 1,450°C and the ocean surface temperature of 100°C in the Early Archaean by Komiya (2004), Komiya (2007) and Komiya et al. (1999), which compare with the corresponding temperatures of the modern Earth of 1,330 and 0°C, respectively, Maruyama and Komiya (2011) calculated the oceanic geotherm of the Early Archaean lithosphere and concluded that the lithosphere was sufficiently rigid to operate modern-style plate tectonics in Isua times. Frei and Polat (2007) used isotopic relations in the Isua BIF to demonstrate that two rare-earth element (REE) sources were derived from sea-floor vented hydrothermal fluids and from ambient surface seawater that were responsible for the iron- and silica-rich layers, implying that the silica was derived from a now-unexposed mafic landmass. Friend et al. (2008) demonstrated that the Isua BIF has modern seawater-like trace element signatures consistent with chemical sedimentation from seawater with little clastic input at c. 3.7 Ga. In a more detailed geochemical study of the Isua BIF Aoki et al. (2018) showed that with stratigraphic height the ratio of hydrothermal fluid to seawater component decreased throughout the period of deposition with increasing distance from a ridge to a trench in common with the changes in modern sediments on a spreading oceanic plate, confirming the role of ocean plate stratigraphy (OPS, see below) in the plate-tectonic history of Isua.

Akilia island at the southern end of the Itsaq Gneiss Complex consists of amphibolites, ultramafic rocks, a controversial Fe-Mg amphibole-quartz-pyroxene rock, a minor anthophyllite-garnet rock of possible sedimentary parentage, and TTG orthogneiss (McGregor and Mason, 1977; Nutman et al., 1996; Myers and Crowley, 2000; Nutman et al., 2002; Manning et al., 2006). The quartz-pyroxene rock, which looks like BIF in the field (Friend et al., 2008) consists of alternating layers of quartz vs. hedenbergitic clinopyroxene, Fe-Mg amphibole ± garnet, apatite and magnetite. The “Akilia controversy” (Whitehouse et al., 2009) arose when Mojzsis et al. (1996) and McKeegan et al. (2007) reported that the quartz-pyroxene rock contained biogenic graphite, apatite, and graphite inclusions in apatite, which has major implications for possible life on Earth at >3,800 Ma. Dauphas et al. (2004) supported the biogenic idea, but Fedo and Whitehouse (2002), Lepland et al. (2005), Nutman and Friend (2006) and Whitehouse et al. (2009) found no evidence for such an origin.

The first major model to explain the Isua geology came from the Maruyama-Komiya team, who with their experience of the Mesozoic OPS in Japan discovered essentially identical OPS in NE Isua (Maruyama et al., 1991; Komiya et al., 1999) followed by Maruyama and Komiya (2011) and Arai et al. (2015). Ocean plate stratigraphy (OPS) is a bulk term that describes the travel history of an opening-closing oceanic plate from basalts, gabbros and ultramafic rocks that form at a mid-oceanic ridge to pelagic cherts and/or carbonates that accumulate during the movement of the ocean toward its trench. When the outer trench slope is reached hemipelagic mudstones are deposited on the cherts, and succeeded by voluminous clastic turbidites, greywackes and conglomerates that may be derived by erosion of the accretionary wedge, the adjacent volcanic arc, or an eroded old craton behind the arc (Kusky et al., 2013). The key point relevant for this discussion is that the sequentially developed package of basalt-chert-turbidite can be recognised in many accretionary orogens from the Eoarchaean to the Cenozoic worldwide. As all the maps of any origin show, this tripartite package is clearly present in NE Isua (but see Geochemically-Constrained Models regarding basalt compositions). It represents the core of the argument that Isua developed as an accretionary orogen little different from those of the Phanerozoic.

The second major model to explain Isua geology comes from the Nutman-Friend team (Nutman et al., 1996; Nutman et al., 1997a; Nutman et al., 1997b; Nutman et al., 2004; Nutman and Friend, 2009; Nutman et al., 2010; Nutman et al., 2013; Nutman 2015a; Nutman et al., 2015b; Friend and Nutman, 2019), who used abundant zircon ages mostly of (meta-) tonalites and diorites to work out the sequential development of two main terranes in NE Isua; a southern c. 3.8 Ga and a northern c. 3.7 Ga terrane. The central boundary between the terranes is a mylonitised belt of chert, BIF and carbonate rocks that contain 3,940–3,750 Ma detrital zircons (Nutman and Friend, 2009). The northern, younger terrane is an imbricate package with no preserved stratigraphy, composed of arc components that young from c. 3,720–3,715 Ma boninites (Polat et al., 2002), tholeiites and picrites (Polat et al., 2002), to c. 3,710–3,700 Ma andesites, lesser dacites and sediments derived from them, to c. 3,700–3,695 Ma BIF and minor zircon-bearing siliceous detritus from a distal source (Nutman and Friend, 2009). The older southern terrane contains a similar arc assemblage and is similar in origin to the northern terrane, but it contains minor >3,850 Ma crust. This (now-thin) central boundary belt probably acted as a décollement when the northern younger, c. 3.7 Ga, terrane was thrust southwards over the southern older, c. 3.8 Ga, terrane.

Although Nutman and Friend (2009) and Nutman et al. (2009) considered that Isua developed by accretionary processes at a convergent plate margin similar to those operating in the modern Earth, their tectonic development by upward younging of the accretionary belt is the opposite of the younging downward polarity recognised in the accretionary wedges of modern subduction zones, and in accretionary orogens of any age (e.g., Leggett et al., 1979; Platt 1986; Kawai et al., 2007; Isozaki et al., 2010; Kusky et al., 2013).

Komiya et al. (1999) considered that the basalts in Isua were generated at a mid-oceanic ridge, and Komiya et al. (2004) reported that some basalts have MORB and others have OIB chemistry. However, detailed trace element geochemistry demonstrated that all the basalts in Isua have island arc chemistry (Polat et al., 2002; Nutman and Friend, 2009). This problem is related to the tectonic divisions in Isua as explained in The Lithology/Structure-Based Model and The Chronology/Lithology-Based Model above. Komiya et al. (1999) defined eight units (mostly across the northern 3.7 Ga zone of Nutman and Freind, 2009) of OPS, which are all separated by thrusts and underlain by massive and pillow basalts, representing the whole OPS history of Isua. Because the early traverses of Komiya et al. (1999) crossed some of the later isotopically-defined units and boundaries of Nutman and Friend (2009), and because Polat et al. (2002) and Nutman and Friend (2009) found that all the basalts have island arc chemistry, inevitably Nutman and Friend (2009) do not believe the OPS units and structural set-up of Komiya et al. (1999), but consider that their three zones developed separately and at different times and were juxtaposed by accretionary collisions. Kamber et al. (2001) made a critical assessment of the isotopic constraints on the evolution of Isua rocks. However, we will not comment further, as it is beyond the scope of this paper to present a comparative detailed analysis of these opposing geochemical arguments.

The Ivisaartoq metavolcanic belt at 64°45′N (Figure 1) contains deformed, metamorphosed pillow basalts with ocelli, cooling cracks and drainage cavities, volcanic breccias, picritic ultramafic flows with clinopyroxene cumulates, serpentinised ultramafic rocks, layered gabbros, and siliceous cherts (Chadwick 1990; Polat et al., 2007). The belt underwent two stages of calc-silicate metasomatic alteration between 3,075 and 2,963 Ma, namely epidotisation during seafloor hydrothermal alteration, followed by calc-silicate veins with diopside, garnet, amphibole, ± vesuvianite associated with late shear zones. Because of the remarkable similarity in all respects Polat et al. (2007) and Ordóñez-Calderón et al. (2009) interpreted the geology and isotope-trace element chemistry of the Ivisaartoq belt and its distinctive mineralisation as a Mesoarchaean supra-subduction zone forearc, similar to that in Tethyan ophiolites in the eastern Mediterranean (Dilek and Flower, 2003).

The arc around 64°30′N in Godthåbsfjord (Figure 1) hosts a large epithermal gold occurrence (Garde et al., 2012a). The Qussuk arc contains a folded and metamorphosed, shallow-crustal argillic to advanced argillic alteration zone with a minimum strike length of 20 km, embedded within the core of a less altered, bimodal andesitic-mafic metavolcanic belt. The leached and metamorphosed varieties of the andesite precursors contain abundant biotite, garnet and sillimanite, but very little or no feldspar. These rocks were originally mapped as pelitic metasediment (see also Schumacher et al., 2011). However, anomalously high contents of immobile trace elements including REEs, negative Eu anomalies (related to leaching of plagioclase), and anomalous Ga/Al ratios show that up to 80% of the precursor rock was lost by leaching. The geochemical signature is best explained by near-surface boiling and dissociation of hydrothermal fluids into a barren, highly acidic phase and an enriched brine, as found in many modern circum-Pacific arcs (Simmons et al., 2005). Garde et al. (2012a) recorded in these altered rocks, zoned zircon grains, which contain igneous cores with high Th/U ratios typical of primary volcanic zircons, yielding ages up to c. 3,070 Ma (Figure 4); the oldest nearby TTG rocks are c. 3,060 Ma in age (Garde et al., 2000). The surrounding zircon shells are strongly enriched in U and Th, with REE patterns typical of hydrothermal zircons. Thin metamorphic rims yield ages of c. 3,000 Ma, contemporaneous with a regional thermal event recorded in most of the Akia terrane (see below). Schlatter and Christensen (2010) had previously shown that the intensity of alteration in the gold-mineralised rocks is highly variable, as is commonly the case in Phanerozoic epithermal gold deposits. The latter authors preferred an interpretation as “orogenic gold” related to the Palaeoproterozoic Ataneq fault 10 km away (Thøgersen et al., 2006; Garde et al., 2012a), although this is barren of gold.

The Storø supracrustal belt is part of a complex zone of thrust-imbricated and folded slices of supra- and infracrustal rocks of Eo-to Neoarchaean age (Friend and Nutman 2019 and older references therein) at the northern margin of the Godthåbsfjord-Ameralik belt at around 64°30′N (Figure 1). It includes TTG orthogneiss, gabbro and anorthosite, mafic amphibolite, metasedimentary rocks, and altered garnet-rich lithologies. Dispersed gold mineralisation associated with quartz veins is hosted by supracrustal amphibolite in central Storø, in the vicinity of chemically altered rocks of volcanic origin (Scherstén et al., 2012; Szilas and Garde, 2013; Szilas et al., 2014, Szilas, 2016b). In spite of the geological complexity, it is widely agreed that the alteration of the aluminous and garnet-rich rocks was pre-metamorphic and that the gold was mobilised into its present location in the core of a late antiform (Szilas et al., 2016b). Arsenopyrite Re-Os and previously published zircon U-Pb age data led Scherstén et al. (2012) to propose that the gold mineralisation formed at 2.71–2.64 Ga, i.e. somewhat younger than its host rocks. Szilas et al. (2016b) argued on geochemical grounds that the altered and subsequently metamorphosed rocks represent transported and weathered sediment derived from mafic volcanic rocks and that these rocks were unrelated to the gold mineralisation, although field observations suggest that the alteration took place in situ in volcanic rocks (Szilas and Garde, 2013). Irrespective of the source, mode and age of the gold mineralisation, there is consensus that the Storø supracrustal belt formed in a convergent plate tectonic setting, where continentally derived sediments were deposited along with arc-related mafic and intermediate volcanic rocks.

The Ameralik supracrustal belt at 64°05′N in the Sermilik block (Figure 1) was mapped at a scale 1:100,000 (Rehnström, 2011), but it has not been described in detail. Szilas et al. (2015) made a geochronological and geochemical investigation of the belt. It is not a coherent unit but consists of numerous inclusions of amphibolite, ultramafic rocks, garnet-mica schists, sillimanite- and cordierite-bearing schists, quartzites and remnants of layered anorthosite complexes that are sheared and boudinaged with the adjacent TTG gneisses. The sillimanite-cordierite-bearing rocks and the quartzites were interpreted as metasedimentary rocks, but might also represent hydrothermally altered (leached) rocks of volcanic origin as documented at Qussuk (see Qussuk-Bjørneøen Metavolcanic Belt); notably Szilas et al. (2015) showed that several of the rocks recognised as of volcanic origin have been hydrothermally altered. The least-altered metavolcanic rocks have a narrow major-element compositional range with SiO₂ contents between 48.0–51.2 wt%, but Szilas et al. (2015) identified two different geochemical groups based on trace elements. Like many other belts described here, as pointed out in the geochemical review by Szilas et al. (2018), the two groups do not represent a single magmatic suite, because they contain rocks with different crustal and mantle source components (Figure 2), as also indicated by the large range of initial εNd_2970Ma values from 0.0 to +5.7 and the initial εHf_{(2,970 Ma)} values from +0.7 to +10.4 in the least-altered amphibolites (Szilas et al., 2015).

As its name implies, Nunatak 1390 is a nunatak at 63°40′N in the Greenland ice cap in the Sermilik block, east-north-east of Sermilik–Figure 1 (Stendal and Scherstén, 2007; Scherstén et al., 2008; Szilas et al., 2012b). The nunatak contains a well-preserved bimodal volcanic succession of mafic to ultramafic rocks, as well as associated acid volcanic rocks and granite intrusions. The nunatak was first described by Escher and Pidgeon (1976) and later mapped in detail by Stendal and Scherstén (2007), who established a relative stratigraphy; the thickness of the entire volcanic package is about 2 km. Being only slightly to moderately deformed, the rocks show well-preserved primary pillow lavas, finely laminated tuffites, ignimbrites (Figure 5), and relict vesicles in pillows and flows.

The lower half of the supracrustal package is dominated by deformed mafic pillow lavas and pillow breccias that are extensively altered and contain calc-silicate minerals, and cut by an E–W-trending swarm of 1–5 m-thick, gabbroic to noritic dykes. In the upper pillow lavas ultramafic greenstones and soapstones have been interpreted as intrusive sills. Above the upper mafic pillow lavas there is a ca 80 m-thick unit of felsic volcanic and pyroclastic rocks that include ignimbrites with recognizable fiammé textures; these rocks are cut by felsic dykes (<1 m) interpreted as feeders to higher felsic lavas in the pile. The felsic rocks are overlain by mafic flows and ash beds that are intercalated with ultramafic sills and mafic rusty layers containing sulfides and tourmalinites. The top 700–800 m of the sequence consist of finely bedded tuffs, which pass upwards increasingly to subvolcanic granite intrusions.

Geochemically the lower mafic-ultramafic rocks have whole rock compositions that plot along a well-defined tholeiitic trend (high Fe/Mg series, Figure 2), whereas the felsic rocks, which are intercalated with tholeiites, form a loosely defined calc-alkaline group (low Fe/Mg) with an andesitic to dacitic composition. The tholeiites have Nb/La ratios of 0.6 ± 0.1 (1σ; n = 9), whereas the felsic volcanics have a substantially lower ratio of 0.14 ± 0.03 (1σ; n = 5); the tholeiitic ratios are thus reminiscent of those of lower crust, whereas the low ratios of the felsic volcanics are typical of arc-related volcanic rocks. Based on the close spatial relationship between the felsic and tholeiitic rocks, Stendal and Scherstén (2007) considered that an arc setting was most plausible. They contended that the intercalation of the mafic and felsic rocks could by analogous to that of a modern arc system, reflecting input from volcanoes in the trench side (tholeiites) and calc-alkaline rocks in an arc to back-arc.

The Grædefjord supracrustal belt, situated at 63°20′N on the southern side of Grædefjord in the Sermilik block (Figure 1), is bimodal comprising about 50% amphibolite (metabasalts) and 40–50% leuco-amphibolite (mostly meta-andesite) together with minor ultramafic rocks and late granitoids (see later). The Grædefjord area is characterised by intense cataclastic and brittle–ductile mylonites (Dziggel et al., 2017). The degree of deformation varies throughout the supracrustal belt, being strongest toward the sheared contacts with the surrounding TTG gneisses. Toward the interior of the belt deformation is less pronounced and primary pyroclastic lithologies such as agglomerates, tuffs and volcanic breccias are preserved.

The supracrustal belt has a poorly constrained Sm-Nd and Lu-Hf isochron formation age of ∼2,970 Ma, similar to that of the Fiskenæsset Complex (Szilas et al., 2013a). Whole rock geochemical data, which confirm the bimodal compositions inferred from field observations, define the two main groups that belong to different chemical differentiation series. The amphibolites have a tholeiitic (high Fe/Mg) affinity coupled with a moderately depleted incompatible trace element signature of (La/Sm) _N ∼1) and weak negative Nb-Ta anomalies of (Nb/Nb* = 0.6–1.1), which plot above the mantle array in a Th/Yb vs. Nb/Yb Pearce diagram (Figure 2). In contrast, the leuco-amphibolites are calc-alkaline with andesitic compositions that show moderately enriched incompatible trace element signatures with (La/Sm)_N = 2.7–5.7 and negative Nb-Ta anomalies of Nb/Nb* = 0.01–0.45) that plot well within the arc-field on a Th/Yb vs. Nb/Yb Pearce diagram. Geochemical modelling suggests that the trace element variation of the leuco-amphibolites is due to mixing between basaltic mantle-derived magma and crust-derived felsic magma with at least 50% contribution from the latter. Hf-isotope constraints point to contamination with pre-existing continental crust that has an age of c. 3,250 Ma, a model that is underpinned by the preservation of coeval inherited zircons in local TTG gneiss. Ultramafic rocks of the Grædefjord supracrustal belt have platinum-group element (PGE) patterns that are similar to those of magmas generated from a high-degree melting of mantle, but they have relatively enriched trace element patterns. The rocks are interpreted to represent arc-related picrites or alternatively high degree partial melts of metasomatised sub-continental lithospheric mantle. Szilas et al. (2013a) concluded that the proposed mixing model for the Grædefjord supracrustal belt is comparable to that in modern continental subduction zone environments, where mixing and crustal contamination processes are univocally important. A possible difference may be the extent of crustal melting, which was higher in the Archaean due to a hotter Archaean sub-continental mantle, compared to that of modern arcs where assimilation of pre-existing continental crust is relatively more important.

The Bjørnesund and Ravns Storø supracrustal belts that crop out at 63°N just north of Frederikshåb Isblink (Figure 1) consist of upper amphibolite facies, andesitic metavolcanic rocks with prominent pyroclastic lapilli tuffs, agglomerate bombs and pillows (Andersen and Friend, 1973; Keulen et al., 2014, including a summary of previous work). The two belts belong to the same mafic volcanic sheet that was subsequently folded. Leuco-amphibolites in the Ravns Storø belt have a whole-rock Sm-Nd isochron age of 3,020 ± 78 Ma and juvenile values of εNd_(2,970 Ma) of +2.1 to +3.5 and εHf_(2,970 Ma) of +3.5 to +4.3 (Szilas et al., 2012a), while Nutman et al. (2004) reported a SHRIMP zircon Pb-Pb age of 2,908 ± 13 Ma for a volcanic sediment in the same supracrustal belt. The supracrustal belts were intruded by the anorthositic-gabbroic-ultramafic rocks of the Fiskenæsset complex (see below), by subhorizontal sheets of diorite, and later protoliths of the granodiorite-trondhjemite orthogneisses (Keulen et al., 2014) that make up about 75–85% of the Archaean crust in West Greenland (Kalsbeek and Myers, 1973; Steenfelt et al., 2005).

The main regional deformation at Bjørnesund is associated with major folds (F2 in most literature), thrusts and shear zones. Large-scale regional F2 folds throughout the entire area between Fiskenæsset and Frederikshåb Isblink (e.g., Myers, 1985) belong to a deformation event that moved from south (Paamiut area), north-eastwards (Bjørnesund region), to north-west (north of Fiskenæsset) between 2.86 and 2.83 Ga (Keulen et al., 2011). The folding was associated with intrusion of granites in fold axial planes.

The Bjørnesund supracrustal belt is mineralised with gold, which represents an original arc-related mineralisation that was metamorphosed and variably remobilised. According to Kolb et al. (2013) thrusts and shear zones that were active during the main deformation event acted as pathways for fluids, which created interactions with the host amphibolites and transported the gold into the amphibolites. The gold accumulation is thought to have occurred during the intrusion of granites into the Bjørnesund supracrustal belt and during deformation associated with coeval thrusts and strike slip faults (Keulen et al., 2014). Metasomatic aluminous gneisses containing cordierite, garnet and sillimanite were generated during high-temperature hydrothermal alteration of the volcanic rocks (Schumacher et al., 2011; Schlatter and Stensgaard, 2014); these are difficult to distinguish from metamorphosed rocks that had already been altered in a surficial environment.

The Nigerlikasik metavolcanic belt at 62°05′N is the easternmost of three subparallel supracrustal belts in the Paamiut region of the Kvanefjord block (Figure 1). The entire sequence was metamorphosed at upper amphibolite facies and subjected to at least two stages of folding (Klausen et al., 2017). An E–W, c. 550 m-profile shown in Figure 6 demonstrates that the belt consists of a mafic to ultramafic lower part and an intermediate to felsic upper part. From base to top the belt comprises: 1. Ultramafic serpentinites (hornblende ± phlogopite ± diopside ± serpentinite/talc). 2. Mafic amphibolites, generally fine-grained, layered hornblende ± biotite ± plagioclase rocks. 3. Intermediate, andesitic–dacitic, fine-grained, layered biotite ± hornblende ± plagioclase ± quartz schists and gneisses. The surrounding felsic TTG gneisses are clearly intrusive into the supracrustal sequence. The Nigerlikasik belt has a minimum age of 2,930 Ma, constrained by zircon U-Pb ages of intrusive TTG-like sheets within the metavolcanic sequence (Klausen et al., 2017). Geochemically the Nigerlikasik belt records a distinct geochemical change in magmatism from tholeiitic (lower part) to calc-alkaline (upper part) (Figure 2). This indicates a fundamental change in the process of derivative magma generation, because it is not possible to link the two rock suites simply by magmatic differentiation. Klausen et al. (2017) interpreted the entire metavolcanic sequence in the framework of a single maturing island arc, with an initial tholeiitic stage characterised by more juvenile (mantle-sourced) magmatism, followed by a differentiated calc-alkaline stage when crustal sources were involved in the generation of the andesite-dacite suite. The final stage of TTG intrusion was interpreted as a separate stage (unrelated to the andesite-dacite suite) of melt formation in the lower crust, which by that time would have been substantially thickened by magmatic accretion and might constitute a proto-craton.

The Au-mineralised Tartoq Group at 61°45′N (Higgins and Bondesen, 1966; Higgins, 1968; Appel and Secher, 1984; Evans and King, 1993) comprises greenschist-to granulite-grade, mainly mafic metavolcanic rocks with well-preserved subaqueous lava flows, basaltic pillows, andesitic lapilli tuffs, shallow crustal mafic sills and dykes, gabbroic rocks, and serpentinites that may have been derived from harzburgitic cumulates or from the mantle, as well as sedimentary banded iron formation; this combined rock assemblage closely resembles the lithological units that typically occur in oceanic crust (Szilas et al., 2013b). The mafic lavas have chondrite-normalised REE patterns and Th/Y-Nb/Yb ratios indicative of an arc affinity when compared with modern island arcs (Figure 2). Nutman and Kalsbeek (1994) reported a SHRIMP zircon U-Pb age of 2,944 ± 7 Ma for an intrusive TTG, which provides a minimum crystallisation age, while Lu-Hf regression gives a model age of c. 3,189 ± 65 Ma for the Tartoq Group (Szilas et al., 2013b). Saintilan et al. (2020) obtained Re-Os model ages of c. 3.18–3.13 Ma from arsenopyrite in gold-sulphide-mineralised shear zones and interpreted the mineralisation as associated with arc volcanism (and thus likely originally epithermal), whereas Nutman et al. (2004) found detrital zircons as young as 2,842 ± 6 Ma. The various latter data may indicate that the Tartoq Group contains significantly different age components.

Kisters et al. (2012) and Polat et al. (2016) described the post-depositional geological history of the Tartoq Group. It is interpreted as beginning with burial and subduction of supracrustal rocks, followed by exhumation and expulsion, and finally overprinting by transcurrent shearing. Most of the observed features comply with those occurring during high-temperature subduction and retrogresssion during exhumation at a modern convergent margin (see also van Hinsberg et al., 2018). Different from current subduction settings are the absence of high-P metamorphic rocks and the close spatial and temporal association between granitoid plutonism and the structural burial and return flow of the Tartoq Group, both of which are explained by a higher geothermal gradient than today in this Archaean subduction zone. Slab melting and the formation of granitoids at a relatively shallow depth in the subduction channel move the magmatic front to the forearc region of the convergent margin. The high geothermal gradient would have caused a higher volume of felsic melts and a weaker crust, enhancing the disintegration of the downgoing slab and thus producing a short-lived subduction event alternating with slab break-off and backstepping of the subduction front in a wide accretionary complex. This scenario was replicated by Polat et al. (2016) who made a convincing comparison between the accretionary Mesoarchaean Tartoq Group and the Upper Cretaceous Chugach-Prince William accretionary complex in Alaska, having formed by subduction-accretion processes.

West Greenland contains the most extensive and best-known examples of these layered intrusions, of which the Fiskenæsset complex at 63°20′N (Figure 1) is the largest and most complete. Its original maximum stratigraphic thickness is close to 540 m. The current tectonic thickness reaches 2 km and some folded layers have an exposed length of at least 30 km (Myers 1985). Although isoclinally and tightly folded three times, metamorphosed up to granulite facies in some areas and commonly retrogressed to high amphibolite facies, the complex has a remarkably well-preserved magmatic stratigraphy, which from top to bottom is: upper gabbro, chromite-layered anorthosite with megacrystic plagioclase (An_75–95), upper leucogabbro, middle gabbro, lower leucogabbro, ultramafic unit (peridotite, dunite, hornblendite), lower gabbro; the weighted lithological average is leucogabbro (Myers, 1985). Excellently preserved primary structures include hundreds of mineral/size-graded layers, cumulate textures, scour channels and trough layering. The leucogabbros are characterised by a cumulate plagioclase-hornblende assemblage in which plagioclase megacrysts may reach 30 cm across, and undeformed chromitite layers in anorthosites are up to 20 m thick. The Fiskenæsset complex has a Sm-Nd age of 2,973 ± 28 Ma, and a Pb-Pb isotope regression age of 2,945 ± 36 Ma (Polat et al., 2010). Best-preserved anorthosites and leucogabbros contain zircons with a weighted mean U-Pb age of 2,936 ± 13 Ma, which provides a minimum crystallisation age of the complex (Souders et al., 2013), or 2,973 ± 28 Ma (Polat et al., 2010), or 2,985 Ma (Hoffmann et al., 2012). Huang et al. (2012) demonstrated that the complex is characterised by depletions of high field strength elements (HFSE; Nb, Ta, Zr, and Hf) relative to Th, U, LREE and LILE (Rb, Ba), as well as enrichments of Sr and Pb relative to MREE, which are consistent with generation in an oceanic island arc. They also showed that abundant magmatic hornblende, the high anorthite content of plagioclase, and negative anomalies of Nb in hornblende are collectively consistent with a sub-arc, hydrous, parental liquid of high-alumina basalt derived from a mantle source (see also, e.g., Hoffmann et al., 2012; Souders et al., 2013; Polat et al., 2015).

The Fiskenæsset complex was emplaced into island-type volcanic rocks (now amphibolites) at 2.97–2.95 Ga (Polat et al., 2010; Keulen et al., 2010), intruded by voluminous tonalites at 2,863 ± 10 and c. 2,802 Ma (Friend and Nutman, 2001; Steenfelt et al., 2005; Hoffmann et al., 2012), and likely underwent collision with other arcs and continental blocks in the late Neoarchaean. A characteristic and diagnostic feature of the Fiskenæsset complex is the presence over hundreds of kilometres of spectacular zones with a remarkable metasomatic assemblage with inter alia sapphirine, ruby, kornerupine, phlogopite, pargasite, enstatite, and red spinel–see Figure 7. The ruby and other aluminous phases were formed by metasomatic reactions at the contacts between the anorthosite and ultramafic rocks (Herd et al., 1969; Appel and Ghisler, 2014; Keulen et al., 2014; Smith et al., 2016; Keulen et al., 2020). The deformation and crustal thickening associated with early thrusting, subduction, and collision in the Fiskenæsset region gave rise to large-scale double and triple fold interference patterns and broadly synchronous amphibolite-granulite facies metamorphism (Windley and Garde, 2009).

Many smaller anorthosite-leucogabbro-gabbro-ultramafic complexes lack the diagnostic stratigraphy, chromitites and associated distinctive sapphirine-bearing rocks of the Fiskenæsset complex, and thus are considered to be derived from separate island arcs. Several were summarised by Windley and Garde (2009). These include.

The Naajat Kuuat complex in the eastern Tasiusarsuaq terrane (Figure 1), which has been subjected to the most detailed geochemical and isotopic study of all the anorthositic complexes in Greenland (Hoffmann et al., 2012).

The south-western part of the Fiskefjord block contains several kilometre-sized lenses of ultrabasic rocks comprising peridotite, dunite and occasional layered chromitite. They are older than the regional TTG orthogneisses and mostly associated with layered metagabbroic and noritic rocks; a few are isolated enclaves in orthogneiss. Some contact zones are jumbled mixtures of ultrabasic rocks, amphibolite and orthogneiss. The dunitic component of the Seqi body (Figures 1 and 8) is highly refractory (Fo_91–93; Szilas et al., 2018). The dunitic bodies may represent residual upper mantle like the nearby kimberlite xenoliths (Bernstein et al., 1998; Bernstein et al., 2007), but their common association with metagabbros suggests they are more likely basal parts of layered complexes as indicated by their chromite compositions and platinum-group element patterns (Garde 1997; Guotana et al., 2018; Szilas et al., 2018). The dunitic bodies are clearly not metamorphosed komatiites and do not yield support for plume or sagduction models.

About 75 noritic intrusions ∼3.0 Ga old with hybrid margins to their hosts occur in the northern Fiskefjord block, and related post-kinematic diorites farther south (Secher 1983; Garde 1991; Garde et al., 2013b; Waterton et al., 2020). The very variable geochemistry of these highly unusual rocks with high Ni and Cr contents suggests derivation from ultramafic magmas contaminated with 30% TTG material. They are unrelated to the older mafic–ultramafic complexes and difficult to explain, since the crustal evolution of the Fiskefjord block up to 3.0 Ga was normal for the NAC. Origins by crustal thinning and extraterrestrial impacting have been suggested (see below).

New evidence of extensive high-pressure metamorphism (Dziggel et al., 2014, 2017; Friend and Nutman 2019; Nutman et al., 2020) has reinforced the importance of collisional tectonics in the boundary region between the Godthåbsfjord-Ameralik belt and the Sermilik block (McGregor et al., 1991), supplementing low-pressure metamorphism interpreted by Wells (1979) as a result of injection of voluminous tonalitic magma into the middle crust. Prior to the final assembly between ca. 2,720–2,700 Ma (Nutman and Friend 2007; Dziggel et al., 2014; Dziggel et al., 2017; Dziggel, 2019), the structurally upper Tasiusarsuaq terrane underwent a prolonged period of medium- and high-pressure granulite facies metamorphism from ca. 2,800 Ma, which terminated in near-isobaric cooling and amphibolite facies reworking (Crowley 2002; Kolb et al., 2012; Dziggel et al., 2012; Dziggel et al., 2014; Dziggel et al., 2017). An excellent explanation of the history of the terrane concept in the Godthåbsfjord region and how it solved controversies in the 1980s between field observations and geochronological data is by Friend and Nutman (2019).

The Tasiusarsuaq terrane (Sermilik block, Figure 1) extends from the southern Nuuk region at Ameralik fjord all the way southward to Frederikshåb Isblink (Figure 1) and comprises TTG gneisses with intrusion ages of 2,880–2,860 Ma (Friend and Nutman 2005a; Næraa and Scherstén 2008). Geochronological data suggest that a significant earlier crust-forming event took place at around 2,920 Ma (Schiøtte et al., 1989; Kokfelt et al., 2011). McGregor and Friend (1992) recognised crustal tilting in the present Bjørnesund block and reported a rare prograde amphibolite to granulite facies boundary in inner Bjørnesund.

Dziggel et al. (2017) described deep-crustal granulite-facies rocks in the central Tasiusarsuaq terrane as ductile fold nappes like those in the orogens of the Grenville and Himalayas. The interpretation by Dziggel et al. (2017) was based on the estimated thickness of crust in the Neoarchaean (≥60 km), an apparent extrusion of deep-crustal nappes into the middle crust, and the evidence for contemporaneous underplating of cool and older continental crust, as suggested by McGregor et al. (1991) for the Godthåbsfjord region. Dziggel et al. (2019) argued that deformation of the Tasiusarsuaq terrane evolved from initial accretion to terminal continent-continent collision.

The Kvanefjord block was studied by McGregor and Friend (1997), Friend and Nutman (2001) and Nutman et al. (2004) who proposed several new tectono-metamorphic terranes (the Sioraq, Paamiut, Neria and Sermiligaarsuk blocks in their terminology).

Using published maps and subsequent observations, Windley and Garde (2009) re-examined the regional distribution of amphibolite facies, granulite facies and amphibolite facies rocks retrogressed from granulite facies within the Kvanefjord block and showed that the region can be viewed as one contiguous crustal segment of prograde amphibolite to granulite facies rocks, overlain by a major fold nappe, the Neria nappe, which was thoroughly retrogressed from the granulite facies and refolded by younger, upright folds. In contrast to early, recumbent isoclinal folds in other areas, which document horizontal tectonics during the crustal accretion, the proposed nappe affected rocks that had already been metamorphosed at granulite-facies conditions. It may well constitute evidence for collisional tectonics, comparable to that described from the Sermilik block farther north.