Pre-Nagssugtoqidian crustal evolution in West Greenland: geology, geochemistry and deformation of supracrustal and granitic rocks north-east of Kangaatsiaq

The area north-east of Kangaatsiaq features polyphase grey orthogneisses, supracrustal rocks and Kangaatsiaq granite exposed within a WSW–ENE-trending synform. The supracrustal rocks are comprised of garnet-bearing metapelites, layered amphibolites and layered, likewise grey biotite paragneisses. Their association and geochemical compositions are consistent with a metamorphosed volcano-sedimentary basin (containing both tholeiitic and calc-alkali lavas) and is similar to other Archaean greenstone belts. The Kangaatsiaq granite forms a 15 × 3 km flat, subconcordant body of deformed, pink, porphyritic granite occupying the core of the supracrustal synform, and is demonstrably intrusive into the amphibolites. The granite displays a pronounced linear fabric (L or L > S). The post-granite deformation developed under lower amphibolite facies conditions (400 ± 50°C), and is characterised by a regular, NE–SW-trending subhorizontal lineation and an associated irregular foliation, whose poles define a great circle; together they are indicative of highly constrictional strain. The existence of a pre-granite event is attested by early isoclinal folds and a foliation within the amphibolites that is not present in the granite, and by the fact that the granite cuts earlier structures in the supracrustal rocks. This early event, preserved only in quartz-free lithologies, resulted in high-temperature fabrics being developed under upper amphibolite to granulite facies conditions.


Introduction and regional geology
The northern part of the Nagssugtoqidian orogen (NNO) is a domain of predominantly Archaean rocks that have been deformed and metamorphosed during Nagssugtoqidian orogenic activity at c. 1.8 Ga (Hollis et al. 2006, this volume;Mazur et al. 2006, this volume;Thrane & Connelly 2006, this volume;van Gool & Piazolo 2006, this volume).Palaeoproterozoic rocks are sparse, and apparently confined to some supracrustal belts, the most prominent one being the Naternaq supracrustal belt (Østergaard et al. 2002).A few small granitic-pegmatitic plugs and dykes are also related to the Palaeoproterozoic evolu-tion.Therefore, while the structures probably reflect Nagssugtoqidian deformation, the rocks themselves (and their protoliths) largely reflect Archaean formations and evolution.
Among the Archaean units, the ubiquitous orthogneissic basement has previously been studied (Moyen et al. 2003a;Steenfelt et al. 2005); it is mostly made up of classical tonalite-trondhjemite-granodiorite (TTG) gneisses, with minor components either related to TTG partial melting, or to the participation of peridotitic mantle in their petrogenesis.All these components are well known in the Archaean, and are probably related to the subduc-tion of hot oceanic lithosphere in an arc setting (Martin 1986(Martin , 1994;;Moyen et al. 2003b;Steenfelt et al. 2005).Several components of the gneissic basement have been dated (Kalsbeek & Nutman 1996;Connelly & Mengel 2000;Thrane & Connelly 2002, 2006, this volume), yielding ages in the range 2.9-2.6 Ga.
Supracrustal assemblages are common, and have been mapped in many places in the Kangaatsiaq, Aasiaat and Kangersuneq map sheet areas (Marker et al. 1995;Mengel et al. 1998;Østergaard et al. 2002;van Gool et al. 2002a;Garde 2004;Hollis et al. 2006, this volume).They are of two main types, either amphibolites or metasedimentary rocks, that may be either aluminous, biotite ± muscovite ± sillimanite ± garnet-bearing metapelites, or quartz-rich, psammitic rocks.The age of the supracrustal rocks is, however, poorly constrained.Whilst some of them are of demonstrated Palaeoproterozoic age (c.1.95 Ga, the Naternaq supracrustal belt, Østergaard et al. 2002;Thrane & Connelly 2002, 2006, this volume), others are likely to be of Archaean age, for instance anatectic metapelites in Saqqarput fjord in the southern part of the Kangaatsiaq map sheet area (Thrane & Connelly 2006, this volume).
Lastly, small intrusions, plugs and sheets of granite and pegmatite cut across the lithologies described above.Some of them have been dated (Kalsbeek & Nutman 1996;Thrane & Connelly 2002, 2006, this volume) and yielded late Archaean ages (2.7-2.6 Ga); it is commonly agreed that most magmatic activity in this region was related to late Archaean events, Palaeoproterozoic P-T conditions being such that anatexis was hardly achieved in the NNO (Mazur 2002;Piazolo 2002).The very homogeneous and porphyritic Kangaatsiaq granite north-east of Kangaatsiaq, 15 by 3 km in outcrop size, is among the largest granitic bodies of presumed late Archaean age in the southern Disko Bugt region.Altogether, the three components outlined above are representative of the usual trilogy of Archaean terranes (Windley 1995): grey TTG gneisses; volcanic and volcano-sedimentary deposits (greenstones); and late, K-rich granites.The area east and north-east of the town of Kangaatsiaq (Fig. 1) is dominated by a synform of supracrustal rocks (mafic and felsic volcanic rocks associated with sediments), into which the Kangaatsiaq granite was emplaced.It is, therefore, a good place to study the Archaean components and local history in the NNO.Fig. 1.Geological map of the Kangaatsiaq granite and surrounding synform, with sample localities from Table 1.Geology mostly from 2002 field work; some parts are drawn from 2001 data (J.A.M. van Gool, G.I. Alsop, S. Piazolo and S. Mazur).A-A, approximate position of section on Fig. 2; B, loc.89, see Fig. 3; C, locs 81-82, see Fig. 4; D, loc.80, see Fig. 5.

Previous work
Previous studies in the Kangaatsiaq area included reconnaissance mapping by Noe-Nygaard & Ramberg (1961), 1:250 000 scale mapping by Henderson (1969), and visits to key localities during the Danish Lithosphere Centre project (Marker et al. 1995;Mengel et al. 1998), as a result of which most published ages were obtained (Kalsbeek & Nutman 1996;Connelly & Mengel 2000).Mapping of the area was predominantly based on coastal exposures, while map information for large parts of the inland areas was based only on photogeological interpretation.Therefore, the Kangaatsiaq granite, which happens to crop out mostly inland and occupies the high grounds at the core of a synform, was at that time simply considered to be part of the polyphase gneissic basement.The Geological Survey of Denmark and Greenland (GEUS) and its partners undertook more detailed mapping of the Kangaatsiaq map sheet in the summer of 2001.This included limited inland work, and the Kangaatsiaq granite was recognised for the first time as belonging to the group of the late Archaean intrusives.Its overall shape was mapped, as well as the supracrustal rocks into which it intrudes.Metasedimentary rocks in the area were also sampled, allowing for metamorphic studies (Mazur 2002;Piazolo 2002).
Finally, re-evaluation of the area in the summer of 2002 by the present authors led to the refinement of geological boundaries and the production of the map of Fig. 1.Sampling of the whole supracrustal series was also undertaken.Thin sections were cut at Université Claude-Bernard (Lyon, France), and samples were analysed at GEUS using XRF as well as ICP-MS (Table 1).In addition, other supracrustal rocks from the same area (obtained from A.A. Garde, personal communication 2003) have been used for the interpretation presented here, as they show similar geochemical features.

Map pattern
As mentioned in the introduction, the studied area (Fig. 1) is mainly made up of three main components: basement orthogneisses discussed by Moyen et al. (2003a) and Steenfelt et al. (2005), a succession of supracrustal rocks which comprise a sequence of amphibolite and metasedimentary rocks described below (Figs 1, 2), and the Kangaatsiaq granite, an intrusion of pink, coarse-grained, strongly lineated (L or L > S fabric) granite with K-feldspar phenocrysts.The foliated basement gneisses and the supracrustal rocks, together with early folds and structures, are refolded into a complex synform which is locally overturned, in particular on its north-western rim (see below).The granite occupies the core of the synform; it is intrusive within the top amphibolitic layer of the supracrustal sequence (Fig. 3) and is also folded together with the supracrustal rocks.The geometry of the granite suggests that it constitutes a single sheet within the supracrustal unit, and that the original intrusion had an overall flat, laccolithlike shape (Fig. 2).We consider that the mapped contact always corresponds to the bottom of the laccolith, and that the top surface has been removed by erosion (Fig. 2).The early structures are associated with syntectonic aplites and pegmatites that cut across the amphibolite but occasionally occupy shear zones or fold hinges.

Stratigraphy
The supracrustal rocks that define the synform occur as largely discontinuous layers (Figs 2-4), that could either correspond to an original, discontinuous geometry (therefore suggesting lava flows), or simply be a result of tectonic stretching during the multiphase deformation witnessed by the area.Indeed, some of the contacts between the lithological units appear to be tectonic (Figs 4,5), suggesting that the present-day 'stratigraphy' might not be original.Nevertheless, our mapping suggests that three main units can be recognised, allowing the following tentative stratigraphic sequence (Figs 2, 4).
1.The lowermost, c. 100 m thick part consists of an association of amphibolite interlayered with garnet-sillimanite metapelites, sometimes with augen textures.Some of the amphibolites are garnet-bearing, while others contain centimetre-sized lenses of diopside-bearing gab-bro and small ultrabasic layers (pyroxenite or serpentinite, observed in the south-western part of the synform).
The pelitic rocks seem to be more abundant in the northern limb and north-eastern extremity of the synform, while the ultramafic rocks and gabbros were found only in its south-western part.
2. The middle part is a sequence about 100 m thick of layered biotite gneiss, i.e. quartzo-feldspathic gneiss with no discriminant minerals and a compositional layering at a scale of c. 10 cm.The layered biotite gneiss is commonly interstratified with layers and lenses of amphibolite 10-100 cm thick.The contact with the lower amphibolite is gradational.As will be discussed below, the layered biotite gneiss likely represents meta-rhyolite.The middle unit of layered biotite gneiss probably does not have a constant thickness; furthermore, in poor, inland outcrops, it is readily confused with basement orthogneisses.A detailed log of the lower and middle parts of the sequence as described in the foregoing was made in the overturned, north-eastern part of the synform, displaying its complex and composite nature (Fig. 3A, locs 81-82).upper boundary of this unit is not observed, since it is everywhere intruded by the granite.This 'top amphibolite' is continuous and can be traced all around the exposed granite contact; it is also rather homogeneous, much more so than any of the other components of the supracrustal sequence.In loc.80 (Fig. 5), it appears to be in tectonic contact with the underlying layered biotite gneiss.

Field description and petrology
As mentioned above, three main components are observed in the supracrustal succession: aluminous metapelite, layered biotite gneiss and amphibolite.Field aspects together with photographs of thin sections are presented in Fig. 6.The aluminous metapelites occur as slaty, fine-grained (0.5-1 mm), grey to yellowish paragneisses (Fig. 6A1).Garnet or sillimanite is commonly seen in outcrop.In thin section, they display biotite, plagioclase and quartz with either sillimanite or poikiloblastic garnet (Fig. 6A2) cutting across an earlier weak foliation marked by preferred orientation of biotite flakes and elongation of plagioclase crystals.
The layered biotite gneisses appear as grey, relatively massive, fine grained (0.5-1 mm), finely layered rocks.They are interstratified at all scales with amphibolite (Figs 3C, 6B1) and generally form discontinuous bodies on a 100 m scale.They consist of quartz, plagioclase, K-feldspar and biotite; the foliation is defined by the preferred orientation of biotite and elongation of quartz grains (Fig. 6B2).
The amphibolites are dark, massive rocks that also show a strong compositional banding (Fig. 6C1-C2).Regardless of their mode of outcrop either as a thick continuous layer, as in the 'top amphibolite', or as discontinuous layers interstratified with other lithologies, they are very similar in visual aspect and mineralogy.They mostly consist of a fine-grained (0.5-1 mm) hornblende-plagioclase assemblage, with preferred orientation of minerals defining the foliation.Commonly, small clusters of clinopyroxene surrounded by felsic (mostly plagioclase) rims are observed (Fig. 6C4).
At one locality, gabbroic lenses on a scale of 5-10 cm have been observed within the amphibolite (loc.92, Fig. 6C3).They are medium grained (2-5 mm) and greenish in aspect, and composed of a clinopyroxene-plagioclase association with diffuse contacts with the neighbouring amphibolite (Fig. 6C4).At the same locality, an ultrama- fic layer c. 0.5 m thick has been observed.It is slightly coarser grained (2-5 mm) than the amphibolite, and solely consists of amphibole grains (Fig. 6C5), which are optically similar to the hornblende in the surrounding amphibolite.

Geochemistry and origin
Figures 7-8 and Table 1 summarise the major and trace element (especially REE) characteristics and relationships of the three main supracrustal components: amphibolites, metapelites and layered biotite gneisses.There is little, if any doubt of the fact that the amphibolites correspond to metamorphosed and deformed mafic igneous rocks.Elsewhere, similar field characteristics in amphibolites as those observed here have been interpreted as corresponding to transposition of former pillow lavas in high strain domains (e.g.Myers 2001).The metapelites obviously have a sedimentary origin and probably represent terrigeneous sediments.The origin of the layered biotite gneisses, however, is less obvious.They could represent either sedimentary or felsic volcanic rocks.Therefore, they are plotted on geochemical diagrams for both magmatic and sedimentary rocks (see below), allowing comparisons.

Origin of the amphibolites
The supracrustal amphibolites and their counterparts, enclaves in the basement orthogneisses, appear to be very similar in composition.They plot mostly as basalts in a TAS diagram (Fig. 7; Le Maître et al. 1989), and an AFM diagram (Fig. 9; Irvine & Baragar 1971) reveals that they belong to a tholeiitic series.This, together with their spectacularly flat REE pattern at about 10 times chondritic values (Fig. 8), is consistent with the amphibolites corresponding to former MORB basalts, possibly formed as part of an oceanic crust.Many discriminant diagrams for basaltic rocks have been proposed on geochemical grounds (e.g.Pearce 1982;Shervais 1982;Mullen 1983).However, some caution should be exercised when using such diagrams for the Archaean, since the existence of modern-style tectonic settings in the Archaean is not certain, and the palaeogeodynamical contexts might not be similar to those of modern settings (Hamilton 1998;McCall 2003;van Kranendonk 2003).Nevertheless, in  such diagrams, the amphibolites plot either as MORB or as rocks originated in oceanic arcs (arc tholeiites), leaving some ambiguity about their original setting.

Origin of the aluminous metapelites
The geochemistry of metasedimentary rocks is commonly used to discuss their source, in terms of (1) the nature of the original sediments, (2) the nature of the weathered/ eroded source material, and (3) the degree of weathering of the source (see e.g.Taylor & McLennan 1985;Herron 1988;Roser & Korsch 1988;Nesbitt & Young 1989;Bohlar et al. 2005).While several authors also use the geochemistry of sediments to discuss their geodynamical setting (Bhatia 1983;Bhatia & Crook 1986;Roser & Korsch 1988), some caution should be exercised when dealing with Archaean environments, as mentioned above.
In terms of classification, the metasedimentary rocks from the Kangaatsiaq area plot mostly as shales or greywackes, using either of the two schemes proposed by Herron (1988).One of these is shown on Fig. 10A; the ambiguity and possible (chemical) confusion between the two groups, shales and greywackes, which are poorly separated by this diagram, has been outlined by these authors.Nevertheless, the conclusion points to relatively immature sediments which have undergone limited transport from their source.
The nature of the source itself can be discussed using major or trace elements.Roser & Korsch (1988) proposed a scheme for source determination of clastic sediments on the basis of major elements.In this instance, the studied samples straddle the P2-P3 boundary (Fig. 10B), suggesting a felsic to intermediate source.Also trace elements can be used to refine this conclusion.As pointed out by Taylor & McLennan (1985), some elements (high field strength elements, rare earth elements (REE), Y, Sc, Th) only undergo limited fractionation during sedimentary processes; thus, their ratios reflect the signature of their source.Plotting these elements against each other shows that the Kangaatsiaq metasedimentary rocks (Fig. 10E-H) have element ratios that are generally consistent with derivation from an orthogneissic source (amphibolites generally have too low trace element contents and incorrect ratios to be a possible source).The only exception is for heavy REE (Figs 8, 10G).Indeed, the relatively high Yb contents of the metasedimentary rocks precludes their derivation solely from a low-Yb gneissic basement, and implies that they must, at least in part, have been derived from higher-Yb rocks such as the amphibolites; this is hardly a surprise, since amphibolite occurs as enclaves intercalated within the orthogneisses.Modelling the REE contents of such a mixture shows that mixing of ortho-   gneisses with amphibolite (Fig. 10G, H) can explain the Yb contents of the sediments; since the amphibolites are, collectively, less enriched in trace elements than the gneisses, their involvement would only have little effect on the other incompatible elements in the metasedimentary rocks.
In contrast, the latter display higher Ni and Cr contents than the orthogneisses, also consistent with a contribution from amphibolite or its precursor rocks in their formation (Fig. 10H).Finally, the degree of alteration of the source can be discussed.The metasedimentary rocks display C.I.A. values (Chemical Index of Alteration, Nesbitt & Young 1989) of c. 60-70 (Table 1), slightly lower than for shales or similar rocks (70-75, Taylor & McLennan 1985).In the triangular diagrams proposed by Nesbitt & Young (1989;Fig. 10C, D), they also depart only moderately from their protoliths, suggesting a relatively unweathered source.Very little or no evidence for secondary Kenrichment is observed.
Taking the above-mentioned limitations into account, the geoynamic setting inferred from the geochemistry gives consistent results regardless of the classification scheme used.Both the major elements classifications of Bhatia (1983) and Roser & Korsch (1988) and the trace element systems of Bhatia & Crook (1986) suggest an oceanic or continental island-arc setting.However, this only reflects the characteristics outlined above: relatively immature sediment derived from poorly weathered felsic to intermediate magmatic rocks, with a possible mafic component.

Origin of the layered biotite gneiss (felsic volcanic rocks?)
The two samples analysed of the layered biotite gneisses give ambiguous geochemical signatures and can be interpreted either as sedimentary or igneous (Figs 7-10).In general, they seem to share more similarities with the granite or the orthogneisses than with any other member of the supracrustal group.In particular, Fig. 10 (C, D) shows that if these rocks are of sedimentary origin, they are indeed very similar to their source and were derived from a largely unweathered protolith.This implies that the layered biotite gneiss can be interpreted in two ways.It may represent very immature sediment derived from a mostly unweathered protolith with a very similar bulk composition, such as a conglomerate made of pebbles of unweathered orthogneiss, in which case the banding could be a trace of the transposed pebbles.Alternatively the layered biotite gneiss represents calc-alkali or TTG-type felsic lavas, whose composition would of course be very similar to that of their plutonic counterparts.

Origin of the supracrustal sequence as a whole
Based on the foregoing discussion two interpretations can be proposed for the supracrustal sequence.
1.The succession could represent a dismembered ophiolite sequence intermingled with clastic sediments eroded from a nearby continent.The combined sequence could then be interpreted as an accretionary prism.The likely tectonic nature of the contact between members of the sequence (see above and Figs 4, 5) supports this hypothesis.
2. The whole supracrustal pile consists of a bimodal, calcalkaline, probably subduction-related volcanic suite associated with immature terrigeneous sediments directly derived from their weathering.This is consistent with an arc situation, in which a back-or fore-arc basin is being filled with both volcanic products and detrital sediments largely derived from the weathering of these lavas.At the same time, tonalitic plutons are emplaced at depth from the same magmas.The plutonic rocks are quickly uplifted and eroded, and, besides intruding into the supracrustal pile, may in some cases also represent the basement for subsequent volcano-detritic basin fill.
In both cases, the rocks were formed in a convergent setting, probably above or close to an active subductions margin.In general, arc-or subduction-related origins for Archaean volcanic suites are preferred by most workers (e.g.Card 1990;Lowe 1994;Windley 1995;Chadwick et al. 1996), although the issue remains controversial (Hamilton 1998;McCall 2003;van Kranendonk 2003).

Nature and origin of the Kangaatsiaq granite
The Kangaatsiaq granite is pink, porphyritic, and displays a distinct rodding (Fig. 11A) showing that it has been intensely deformed (see below).While YZ sections (perpendicular to the main stretching direction) display a preserved magmatic texture, sections parallel to X clearly show the gneissic texture of the rock.The mineralogical paragenesis is K-feldspar + quartz + sodic plagioclase + biotite, with accessory zircon, titanite, apatite and oxides.The granite has moderate K/Na ratios (0.67-0.86), is slightly metaluminous with A/CNK ratios of 1.00-1.04,and has low Mg# of 32-41 (Table 1).Ni and Cr contents are also low, while Rb, Sr and Ba contents are moderate; this composition corresponds to the biotite-bearing granites of Moyen et al. (2003b), which are interpreted to have been derived from partial melting of TTG gneisses.This conclusion is consistent with the highly migmatitic nature of the surrounding gneissic basement (van Gool et al. 2002a).

Structure and deformation history
As mentioned above, the granite displays a strong rodding and L > S fabrics (Fig. 11A).The strain pattern in the granite (Fig. 11B) is consistent with highly constrictional deformation, with foliation poles plotting on a great circle, and lineations clustered near the pole of this great circle.This corresponds to subhorizontal, ENE-WSW stretching, consistent with the general orientation of the structures in Kangaatsiaq area (Fig. 1), and more general-ly with the structural grain of the region (van Gool et al. 2002a;Piazolo et al. 2004;Mazur et al. 2006, this volume).The surrounding gneissic basement and supracrustal rocks show the same strain pattern when plotted (Fig. 11C), although in the field, the rocks commonly have a LS or S > L fabric.This suggests that pre-existing foliations have been reoriented during the latest constrictional deformation event, leading to their present distribution.The fact that intense constriction (rather than shortening) can produce folded structures has previously been demonstrated by e.g.Leloup et al. (1995) in the Red River shear zone in Yunnan, China, where the ductile deformation in gneisses resulted in the development of elongate synclines and anticlines with axes parallel to the shear zone and the X-axis of deformation.
The study of deformation-related textures allows the conditions of deformation to be roughly constrained.In the granite and felsic components of the supracrustal se-  ries (Fig. 12A, B), the deformation led to the development of quartz subgrains and recrystallised quartz ribbons.This corresponds to deformation under lower amphibolite facies conditions (400 ± 50°C; Bouchez & Pécher 1976;Gapais & Barbarin 1986;Gapais 1989;Hirth & Tullis 1992;Vernon 2000).Under these conditions, only the quartz is ductile, such that all deformation is accommodated by quartz recrystallisation or deformation.In the Al-rich lithologies, deformation-related textures are mostly seen in the formation of poikiloblastic, syn-to post-tectonic garnets (Fig. 12C).Piazolo (2002) estimated that the chemistry of garnet in similar pelites nearby is compatible with a long duration of temperature conditions at around 500°C, which is in broad agreement with the above estimate.Willigers et al. (2002) described the cooling history of the NNO close to our study area from Ar-Ar dating of various minerals, and likewise concluded that the cooling history of the NNO was slow, from 400°C (mus-covite closure) at 1.7 Ga to 200°C (K-feldspar closure) at 1.5 Ga.Therefore, it can be considered that a constrictional deformation event post-dating the granite emplacement occurred during cooling to lower amphibolite facies conditions.Since this event is apparently responsible for the regional-scale structures (Mazur 2002;van Gool et al. 2002b;Piazolo et al. 2004), and is of lower Proterozoic age (Willigers et al. 2002), we propose that it corresponds essentially to the Nagssugtoqidian deformation proper.However, some textures are not compatible with the above conditions.In amphibolites, high-temperature fabrics with polygonal textures and 120° triple junctions are preserved (Fig. 12D).In some of the felsic supracrustal rocks or basement gneisses (but never in the granite), evidence is preserved for a similar high-temperature fabric, overprinted by later quartz recrystallisation (Fig. 12E).According to Kretz (1969), Gower & Simpson (1992), Kretz (1994) and Martelat et al. (1999), such fabrics are likely to develop under granulite facies conditions (600-800°C).This points to the existence of one or more older (D1?) deformation event(s).Since no evidence for this deformation is found in the granite, we suggest that it was pre-granite, and therefore likely corresponds to late Archaean deformation.P-T estimates for metapelites and metabasites in the Kangaatsiaq area by Piazolo (2002) also indicated the existence of an early metamorphic phase with P-T conditions between 650°C, 3-5 kbar and 780°C, P unknown.This estimate is in good agreement with the textural evidence for D1 deformation under lower granulite facies conditions.The pre-granite deformation is also evidenced by the early isoclinal folds, the existence of a foliation within the supracrustal rocks that does not exist in the granite, and the fact that the granite apparently cuts earlier structures (Fig. 5).At loc. 80, the granite is clearly observed cutting across the foliation and shear bands in the amphibolite; these shear bands are injected by pegmatites that might also be cut by the granite.This suggests that there were actually two pre-granite events, the first of which corresponds to the granulite facies formation of the foliation and isoclinal folds, and the second one to the pegmatite-injected shear bands.However, the floor of the granitic intrusion is also apparently offset by the shear bands (Fig. 5B).Furthermore, the geometry of the shear bands and the foliation suggests extensional deformation; since the cliff face studied here almost corresponds to a YZ section relative to the regional constrictional deformation, this geometry is likely to correspond to the original, preserved pre-constriction geometry.Finally, the fact that the granite both cuts across, and is offset by the shear bands, suggests that the granite emplacement may actually have been syn-extension as sketched in Fig. 5C.Altogether, the simplest possible deformation history (with the smallest number of episodes) can be summarised as follows (Fig. 13).
1.A first deformation event (D1) under lower granulite facies conditions (c. 5 kb, 600-800°C), resulted in the development of granulitic (polygonal) textures in all the existing lithologies, the formation of a main foliation, and isoclinal folding.It probably corresponds to compression of the original, likely accretionaly wedge or arc sequence.
2. This may have been followed by a second event (D1b) of probably extensional deformation, maybe associated with (or shortly followed by) the emplacement of the granite sheet.This event, only witnessed by the shear zones cutting the D1 foliation, e.g. at locality 80, is poorly recorded and probably just represents the final stage of D1 deformation.Assuming the granite has a late Archaean age, which is very likely in the regional context, this deformation could correspond to the later stages of the evolution of an arc or active continental margin, with strain relaxation and syn-extension granite emplacement.
3. A final event of constrictional deformation under lower amphibolite conditions (D2).Due to the relatively low-temperature conditions, only the quartz-bearing lithologies were affected.Therefore, the granite shows strong recrystallisation, the felsic supracrustal rocks display overprinting of the D1/D1b fabric by this event, and the quartz-free amphibolites were essentially unaffected by this event.The D2 event corresponds to the purely constrictional, regional structures which have been interpreted by Piazolo et al. (2004) and Mazur et al. (2006, this volume) as resulting from the indentation of the NNO by a solid, north-moving block immediately north of the Arfersiorfik shear zone (for the latter, see e.g.Sørensen et al. 2006, this volume).This Palaeoproterozoic deformation gave the studied area its present synformal structure.

Conclusions
While the present-day synclinal structure of the Kangaatsiaq area essentially results from N60 constriction related to the Palaeoproterozoic Nagssuqtoqidian deformation, the lithologies together with early preserved structures give insight into the late Archaean crustal evolution.The basement gneisses genetically belong to the TTG suite (Moyen et al. 2003a;Steenfelt et al. 2005), which is generally interpreted as generated by partial melting of a subducting slab (e.g.Martin 1994).Some components of the basement display implications of mantle wedge involvement in their genesis (Steenfelt et al. 2005), which is unusual in the Archaean but nevertheless consistent with an active margin setting.The supracrustal succession is composed of discontinuous layers of mafic MORB-like or arc tholeiite lavas, and together with immature, terrigeneous shales or greywackes derived from erosion of the basement TTG gneisses or volcanic counterparts to them, with a likely small contribution from tholeiitic lavas.Part of the succession could also have been felsic rocks derived from erosion of the basement TTG gneisses or volcanic counterparts to the latter, with a likely small contribution from tholeiitic lavas.The whole series is capped by a layer c. 100 m thick of mafic volcanic rocks likewise of tholeiitic affinity.All these lithological components are in good agreement with either an arc-related setting, with a plutonic arc developing simultaneously with the filling of volcano-detritic basins with lavas of similar affinities and immature sediments; or with an accretionary wedge environment involving ocean floor juxtaposed together with similar sediments.In both cases, they correspond to an active subduction margin.Intense migmatisation of the basement is probably associated with the emplacement of the anatectic, likely synkinematic Kangaatsiaq granite.This was apparently synchronous with an early, lower granulite facies (D1/D1b) deformation event that may have ended with strain relaxation and exhumation of the rocks from the active margin at the end of the Archaean cycle.
The supracrustal association and the sequence of events in the Kangaatsiaq area are comparable to the evolution of many Archaean greenstone belts (e.g.Card 1990;Chadwick et al. 1996;Hunter et al. 1998).On the other hand, classical Archaean components such as orthochemical sediments and plume-related komatiites (Arndt 1994) or orthochemical components (Lowe 1994) are completely missing from the Kangaatsiaq area.However, this apparently rather uncommon absence is known from other midto late Archaean greenstones, also in West Greenland (e.g.Garde 1997).The setting is sometimes interpreted as being arc-related (Card 1990;Lowe 1994).In contrast, widespread melting and granite emplacement at the end of the Archaean is a very common situation, which has been described in many studies (among others, e.g.Gorman et al. 1978;Card 1990;Sylvester 1994;Windley 1995;Chadwick et al. 1996;Moyen et al. 2003b).

Fig. 5 .Fig. 6 .
Fig.5.Photomosaic (A) and structural interpretation (B) of the cliff face at loc. 80 (photo facing east).Stippled: pegmatites; grey: high-strain zones.Evidence for pre-to syn-granite, apparently extensional deformation is preserved in the amphibolite bodies intruded by the granite.Details of the cliff face display the apparently extensive deformation in the amphibolite.Cross-cutting pegmatites (see photo 4-12) are occasionally affected by this deformation, suggesting that it is synchronous or nearly synchronous with granite emplacement.C: Schematic relationships between the granite, the early extensional deformation, and the supracrustal pile, inspired from loc. 80.

Fig. 11 .
Fig. 11.A: Macroscopic view of the Kangaatsiaq granite at loc. 75 (corresponding to sample 485529), showing strong rodding.Hammer shaft about 4 cm wide.B, C: Stereograms of poles to foliation (circles) and lineations (squares) within and outside the granitic intrusion.The strain patterns are similar in both units and define a highly constrictional, NE-SW-trending and subhorizontal deformation.

Fig. 12 .
Fig. 12. Deformation textures either related to the latest, constrictional deformation (A-C) or not compatible with low-T deformation (D, E).See comments in the main text.A: Quartz ribbons in the Kangaatsiaq granite (sample 485527).B: Quartz subgrains in felsic supracrustal gneiss (sample 485531).C: Poikiloblastic garnet in metapelite cutting across an earlier foliation (sample 485535).D, E: High-temperature recrystallisation with 120° triple junctions in amphibolite (sample 485540) and felsic rocks (sample 485530).In E, the quartz also shows low-temperature deformational features such as undulating extinction and quartz subgrains, indicating that this rock witnessed two successive deformation events.

Fig. 13 .
Fig. 13.Summary of the deformation history of the Kangaatsiaq synform and Kangaatsiaq granite.See comments in the main text.Photos from Fig. 12.

B. Generalised stratigraphic columns in the Kangaatsiaq syncline A. Coastal section, locs 81-82
Fig. 7. Total alkali vs. silica (TAS) diagram (Le Maître et al. 1989) for the magmatic components of the supracrustal rocks and the surrounding orthogneisses.

Table 1 .
Chemical analyses of rocks in and around the Kangaatsiaq syncline Sample numbers refer to GEUS databases; localities are shown on Figs 1 and 4, except gia 2001-217 outside the map area (UTM 396540, 7575984).Major elements in wt%; trace elements in ppm.Analytical data obtained at GEUS by XRF (major elements and approximate trace elements in 470529) and ICP-MS (all other trace elements).