The Farsund Shear Zone: geochemical evidence for lithological diversity in the wall rock of the Rogaland Anorthosite Province, South Norway

The Rogaland Anorthosite Province (RAP) – a typical Proterozoic Anorthosite–Mangerite– Charnockite (AMC) plutonic complex exposed in the Sveconorwegian orogen in South Norway – was emplaced diapirically through the crust, along a shear zone, at c. 933–916 Ma. The shear zone, recently defined as the Farsund Shear Zone (FSZ), crops out along the eastern flank of the anorthosite province, and it is made up of strongly foliated, steeply dipping, banded gneisses. The banded gneisses comprise a diversity of lithologies: metabasites, granitoid gneisses, augen gneisses and kinzigitic gneisses. Major and trace element compositions of samples mostly from the banded gneisses


Introduction
Diapiric rise of anorthosite through ductile migmatites is the dominant mechanism of emplacement of the large anorthosite massifs in the Rogaland anorthosite province (RAP) (Duchesne, 1984;Barnichon et al., 1999;Charlier et al., 2010).This diapirism may have been facilitated by a shear zone as suggested by the occurrence of a crustal size geophysical discontinuity revealed by deep seismic data (Andersson et al., 1996), and synthesised in the crustal tongue model of Duchesne et al. (1999).A component of this shear zone has been recently defined on the eastern flank of the magmatic province and called the Farsund Shear Zone (FSZ) (Bolle et al., 2010).This shear zone coincides with a unit of banded gneiss, interlayered with thin units of granitic gneiss and augen gneiss (Falkum 1982(Falkum , 1985;;Marker et al., 2003).

N o t e x p o s e d N o t e x p o s e d F a r s u n d s h e a r z o n e F a r s u n d s h e a r z o n e
various depths from tapping different sources (the crustal zone concept of Annen et al., 2006) have intruded this shear zone.Most intrusions are coeval with the RAP but we provide geochemical evidence that some could be related to older phases in the evolution of the gneiss complex, possibly up to 1500 Ma.The FSZ might have been active for a long interval of time.
The studied region, located between the SMB and the RAP, is part of the pre-Sveconorwegian basement (the Hardangervidda Rogaland sector) that continues to the north and merges in the Mesoproterozoic Suldal Arc (Roberts et al., 2013) that comprises c. 1500 Ma amphibolitic to rhyodacitic gneisses, gabbroic to granitic gneisses, as well as granitoids.
In Rogaland, the high-grade gneisses of the basement have suffered two metamorphic cycles (Kars et al., 1980;Tobi et al., 1985).The first one (M1) of regional extension has culminated in UHT conditions in the upper amphibolite to granulite facies (Drüppel et al., 2013) in the time window between c. 1045 Ma and c. 992 Ma (Laurent et al., 2018a), and the second one (M2) has developed in an aureole around the AMC complex at c. 930 Ma and has also reached UHT conditions (Laurent et al., 2018b).
An orthopyroxene isograd in quartzo-feldspathic rocks (Figs. 1 & 3) has been defined (Kars et al., 1980;Falkum, 1982) that shows a progressive E to W increase in metamorphic grade from amphibolite facies to granulite facies at the contact with the RAP.The two isograds that characterise the gneisses north of the Egersund-Ogna anorthosite and the Bjerkreim-Sokndal layered intrusion -an osumilite isograd in metasedimentary rocks and a pigeonite isograd in leucogranitic gneisses (Tobi et al., 1985) -have not been recognised in the studied area.However, Wilmart & Duchesne (1987) provided evidence ¬from a hercynite + quartz paragenesis armoured in porphyroblastic garnet in a kinzigitic gneiss (sample 81.12.1) ¬ that UHT conditions could have been reached in the banded gneisses close to the AMC complex (Fig. 2).
We mainly focused on rocks from the banded gneisses (Figs. 2 & 3) that comprise metabasites, augen gneisses, granitoids rocks and kinzigitic gneisses.Most samples were collected in the banded gneisses that crop out along the FSZ (Bolle et al., 2010) (Fig. 2).These samples were thus affected by the two cycles of metamorphism.Some samples (Fig. 3) are located near Lundevatn, more to the east, farther from the RAP, in another banded gneiss unit cropping out close to the SMB.These samples were clearly affected by the M2 cycle as shown by reaction rims between amphibolite and felsic lithologies (Tobi et al., 1985;Vander Auwera, 1993).We have also investigated samples from the granite gneiss unit that is exposed east of the FSZ (Fig. 2).All samples were strongly affected by several phases of deformation, particularly those located in the FSZ (Bolle et al., 2010).
The FSZ is described by Bolle et al. (2010).It is a NW-SE-trending, c. 70 km-long up to 3 km-thick, high-strain zone exposed along the northeastern contact of the RAP.It consists of granulite facies gneisses characterised by a steeply dipping tectonic lithological layering (or banding) (Fig. 1), recording extensional or transtensional deformation.The diversity of gneisses hosted in the shear zone are analysed hereafter.The timing of deformation clearly overlaps with the timing of emplacement of the Rogaland Anorthosite Province, i.e., 933-916 Ma (Schärer et al., 1996;Vander Auwera et al., 2011).

Methods of investigation
Representative samples of the major lithologies have been collected.A short petrographic description of the samples is given in Electronic supplements 1 & 2. Detailed field relationships of most samples are described in guide books -locality 4.10 in Duchesne et al. (1987), and locality 4.3 in Duchesne & Korneliussen (2003) -and for the samples around the lake Lundevatn in Vander Auwera (1993).
The samples were analysed for major and trace elements.X-Ray fluorescence on a CGR Lambda 2020 spectrometer (University of Liège) was used to analyse Si, Ti, A1, total Fe, Mn, Mg, Ca, K and P on Li borate glass discs, as well as Na, Rb, Sr, Zr, Y, Ni, Co, Zn, V, Cr, Ba, Ce, La and Nd, on pressed powder pellets.FeO was measured by titration.
Instrumental neutron activation analyses for REE, U, Th, Ta, Hf, Sc, Rb and Cr were carried out at the Afdeling Fysico-Chemische Geologie, KUL, Leuven, under the supervision of Jan Hertogen.The agreement between the two methods is excellent.When several methods were used, the values from the NAA have been preferred.
A fourth type assembles samples from the other types that, whatever their mineralogy, have high P, Ti and Zr contents, characters that are typical of jotunites (Fe-Ti-P-rich hypersthene diorites or monzonorites).This group is called jotunitic metabasites.Jotunites are typical members of the AMC plutonism, particularly in the RAP where they occur as chilled margins of the Bjerkreim-Sokndal and Hidra intrusions as well as in a large dyke system (Duchesne et al., 1989;Duchesne, 1990).
Sample locations and major and trace elements compositions are given in Electronic Supplement 3 and short petrographic features are reported in Electronic Supplement 1.
The geochemical contrast between the various types is shown in Figs. 4 & 5.The group of jotunitc rocks, defined by P2O5, TiO2 and Zr (Fig. 4A, B, D), tends to have a lower Mg# and higher Ba and Ce compositions (Fig. 4E, F) than the other metabasites.For the other elements the jotunitic rocks are similar to the other types, thus of little help in the discrimination process.

Paragneisses
A paragneiss occurrence (sample 81.12.1;Fig. 2) in the banded gneiss unit close to the contact with the Apophysis (Trolldalen) has been documented.It is a kinzigitic gneiss that contains garnet, cordierite, sillimanite, K-feldspar, plagioclase, hercynite, Fe-Ti oxides and quartz.Armoured relics of spinel-quartz aggregates in garnet suggest that temperatures >900°C were attained at c. 6 kbar in the contact aureole of the RAP (Wilmart & Duchesne, 1987).This gneiss locally shows evidence of migmatisation processes: sample 81.12.1B has a leucosome (L) associated with a melanosome (M); sample 81.12.1A is taken from the homogeneous part of the gneiss.Their major and trace element compositions are reported in Electronic Supplement 3.

Augen gneisses
Two augen gneiss occurrences have been sampled in the banded gneiss unit of the FSZ: 81.12.16 in Trolldalen and 81.16.3 in Eikeland (Fig. 2).They both contain K-feldspar megacrysts in a granoblastic matrix of equigranular grains of plagioclase, quartz, clinopyroxene, orthopyroxene, rare dactylitic biotite, apatite and zircon.The chemical compositions (Electronic Supplement 3) are compared to selected samples of the Feda and Liland augen gneisses (Bingen, 1989;Bingen et al., 1993) in Fig. 6.Both samples are calc-alkalic (Fig. 6A), magnesian (Fig. 6B), and belong to the high-K calcalkaline series (Fig. 6C).In Harker diagrams they are similar to the less evolved samples of the Feda suite (but have slightly lower Mg#) (Fig. 6D).The REE content (Fig. 6E) is typically higher than in the Feda suite, and the spidergram (Fig. 6F) is similar with a less pronounced Nb negative anomaly and higher Zr contents.Although the Liland augen gneiss intrusion is located close to the FSZ (Fig. 2), Fig. 6 suggests that it is compositionally more differentiated and distinct.
Figure 6.Augen gneiss compositions from the Farsund Shear Zone compared to the Liland intrusion and to selected samples of the Feda suite (data from Bingen, 1989;Bingen et al., 1993) (Frost et al., 2001); E to J: Harker diagrams of significant elements.

Granitoids
Several types of granitoids can be distinguished on a geochemical basis.Firstly, granite gneisses; secondly, granitoid layers in the banded gneisses which can be subdivided into granite (s.s.) and leucosome layers.All types belong to the high-K calc-alkalic series (except for sample 89-75 which is calc-alkaline) (Fig. 7A) but the granites are exclusively ferroan, the granite gneiss and the leucosome layers being magnesian to ferroan (Fig. 7D).Compared to the granite layers, most granite gneisses are richer in Th (Fig. 7G), Pb (Fig. 7I) and Rb (Fig. 7J) and lower in FeOt, TiO 2 (Fig. 7E, F) and Zr (Fig. 7H).The REE distributions show high contents, high [La/Yb] N ratios and high Eu negative anomalies in granite gneiss (except the intriguing 73-48) (Fig. 8B), slightly lower REE contents and La/Yb ratios and small Eu anomalies in granitic layers (Fig. 8C), and the lowermost REE contents with large positive Eu anomalies in leucosome layers (Fig. 8D).Sample 81.27.1D (Fig. 8B) is a granite gneiss with a high HREE content which resembles the profile of a granite band but its high Pb and Th concentrations preclude its belonging to this group.Two hololeucocratic granites (samples 81.27.2 and 83.22.2C) are very low in Fe, Ti and Zr (Fig. 7).

Metabasites
The jotunitic metabasites can be related to the RAP magmatism.They have indeed strong similarities to the jotunite compositions (Duchesne et al., 1989;Vander Auwera et al., 1998, 2011) that are intimately linked to the AMC plutonism, irrespective of whether they are interpreted as parental magmas (e.g., Duchesne et al., 1999;Liégeois et al., 2002) or residual liquids (Bybee et al., 2014) (sample 84-24 from Moi) typically shows the development of a hypersthene rim at the contact with the neighbouring felsic band.As shown by Vander Auwera (1993), the rim development took place after the folding of the banded gneisses which implies the early intrusion of the jotunitic magma in the envelope prior to the M2 contact metamorphism and its associated deformation in the shear zones.This jotunite could be contemporaneous with the early phase of the RAP formation and be older than 930 Ma.
The other types of metabasites are more difficult to interpret.The classical view that these amphibolites could have a metasedimentary origin can be readily rejected because there are no associated carbonate rocks in the banded gneisses.Thus, the amphibolite protoliths must have an igneous origin and could have been affected to various degrees by migmatisation processes, indicated in particular by a sharp decrease of F and an increase of Nb and Ta when passing into granulite-facies conditions (Vander Auwera,1993).In addition, this author concluded that the REE distributions are not significantly affected by metasomatic processes and must reflect geochemical properties of the protolith.The flat REE distributions of the amphibole metabasites and the relatively low [La/Yb] N ratios of the biotite-bearing amphibolites and two-pyroxene norites (Fig. 5) are consistent with an oceanic origin, a feature which is corroborated by the Ti-V diagram of Shervais (1982) and the Ti-Zr diagram of Pearce & Cann (1973) (Fig. 9).(1982).B: Ti vs. Zr diagram of Pearce & Cann (1973).

The metasedimentary character of the kinzigitic gneiss
The major element compositions of the kinzigitic gneiss and its migmatitic leucosome and melanosome, projected in the Eskola triangles (Eskola, 1920), show that the homogeneous sample 81.12.1A was initially a pelitic sediment essentially made up of quartz, chlorite and muscovite (Fig. 10).The pelitic character is corroborated by the REE distribution compared to the Post-Archaean Argillaceous Shales (PAAS) (Taylor & McLennan, 1985) (Fig. 11).It is also very close to the upper continental crust composition of Taylor & McLennan (1985) except for an intriguing low Sr content (Fig. 11C).It is worth noting that the leucosome composition is particularly depleted in HREE and shows a positive Eu anomaly (Fig. 11B), features characteristic of migmatitic leucosomes (see below).

Migmatitic leucosomes
Some of the granite layers in the banded gneiss are interpreted as migmatitic leucosome essentially on the basis of their REE distributions (Fig. 8D) that show low REE concentrations and a clear positive Eu anomaly.The same is also observed in sample 81.12.1BL (Fig. 11B), the leucosome associated with the kinzigitic gneiss 81.12.1A.Such REE profiles are commonly observed in migmatitic rocks (Barbey et al., 1989;Vander Auwera, 1993;Duchesne & Wilmart, 1997), and are usually interpreted as resulting from selective anatexis of feldspar and quartz, with heavy REE retained in the residue by an accessory mineral such as zircon and common mafic minerals.A detailed discussion is outside the framework of the present paper.

Charnockites
Two possible origins of charnockitic compositions have been proposed, and they both seem to be compatible with available major and trace element data.On the basis of experimental data in a pure CO 2 gas phase, Wendlandt (1981) has argued that charnockitic melts could be produced by anatexis.
Another process for the formation of charnockites has been documented in the neighbouring Farsund intrusion (Vander Auwera et al., 2014a).A differentiation series starting at around 65% SiO 2 (a quartz monzonitic composition AD011) and extending up to 72% SiO 2 has been defined.It can also be explained by classical concepts of plutonic crystallisation.The chemical variation trends are shown in Fig. 12 and compared to the compositions of our samples in the FSZ.These trends globally mimic our compositions if natural variations in compositions are considered, namely in the Breimyrknutan body and in two samples, 89-57g1 and 89-57g2, from the same layer.

Comparison with the Sirdal magmatic belt
In the high-grade gneiss domain of South Norway, the banded gneiss formations have been intruded by plutons of the Sirdal magmatic belt (SMB) and are thus older than 1070 -1020 Ma (Coint et al., 2015;Slagstad et al., 2018).Could some magmatic products from the SMB have intruded the FSZ?The occurrence of the Liland augen gneiss, dated at c. 1051 Ma (Bingen & van Breemen, 1998), that crops out close to and parallel to the FSZ (Fig. 1) is evidence that the SMB intrusive process can inject magma batches at some distance (10 -15 km) outside the main belt.It is therefore plausible (and the simplest interpretation) that augen gneisses with compositions analogous to the Feda suite could also represent SMB offshoots emplaced in the 1070-1020 Ma interval.
On a more general basis, it is worth comparing our granitoids to the SMB database of Slagstad et al.
(2013) (Fig. 13).It turns out that charnockites with their high contents in FeOt (Fig. 13A) and TiO 2 (Fig. 13B) together with very low Th (Fig. 13C) are not found in the SMB.Many SMB granitoids are rich in Th, Pb and Rb, a feature characteristic of our granite gneiss unit, although samples 72.158 and 81.15 have [La/Yb] N ratios much higher than any SMB rocks.More samples are needed to constrain the origin of our granite gneisses but, globally, as for the SMB, an arc origin cannot be precluded considering their high-K calc-alkaline character.Moreover, Figs.13A, B, D show that our granite layers with their high FeOt, TiO 2 and Zr contents have very few equivalents in the SMB.Their formation and intrusion during the SMB event is thus highly unlikely.We show below that they are possibly related to the older Suldal Arc.

Relationships with the Suldal arc lithologies
Could the banded gneisses in the study area have any relationships with the older c. 1520-1480 Ma Suldal arc lithologies that outcrop in the northwestern part of the Rogaland Hardangervidda sector (Roberts et al., 2013)?In the Suldal arc, the major and trace element compositions of grey gneisses ranging from 50 to 78% SiO 2 and granitoids from 65 to 76% SiO 2 have been reported by Roberts et al. (2013).These two series of rocks and our granite gneisses and granite layers are high-K calc-alkaline and may thus have been formed in the same arc context.There are, however, significant differences when trace elements are considered.We compare granite gneissses and the granite layers in the banded gneisses to the Suldal granitoids and grey gneisses >70% SiO 2 in Fig.   (data from Vander Auwera et al., 2014a).
We can speculate that our granite gneiss may have formed in an arc context but with a source or evolution different from that of the Suldal arc.Our metabasites with oceanic characteristics could belong to altered oceanic material (with overlying sediments, the kinzigitic gneiss) that, in Roberts' model, were accreted to a pre-existing continental margin and that were in part underthrust to the magma generation zone of the Suldal arc.Finally, the only vestige of the c. 1500 Ma Suldal arc in the south of the Hardangervidda-Rogaland sector would be the granite layers in the banded gneisses.Slagstad et al., 2013).
Timing of the magmatic events in the shear zone The FSZ was particularly active during the waning stages of the Sveconorwegian orogeny with the intrusion of anorthosite and related rocks.Anorthosite massifs are commonly related to major lithospheric structures (Emslie et al., 1994;Scoates & Chamberlain, 1997;Wiszniewska et al., 2002;Bogdanova et al., 2004;Shumlyanskyy et al., 2017).On a large scale it has been suggested that the RAP intruded along a lithospheric-scale weakness zone (Duchesne et al., 1999) detected through deep seismic data (Andersson et al., 1996).The eastern-most part of this structure is now represented by the FSZ (Bolle et al., 2010).
The Hidra anorthosite massif dated at 932 ± 9 Ma (Vander Auwera et al., 2011) intruded as a wedge in the FSZ.It is the youngest intrusion in the area as evidenced by its undeformed structure.The charnockitic part of the Farsund intrusion at 931 ± 2 Ma has been emplaced mainly through the same channel.This age is within errors similar to the Hidra intrusion but blastomylonitic structures in the Farsund intrusion close to the contact with the Hidra intrusion clearly point to a slightly older age than this massif.
The Breimyrknutan charnockite and possibly the charnockitic layers were formed and emplaced contemporaneously with the Hidra body and with the Apophysis.The heat flow from both intrusions provided the high temperatures required for anatexis.Roberts et al., 2013).
We suggest that the jotunite metabasite intrusions have taken place over a large time interval.They can be related to the emplacement of the RAP and more precisely either to the main magmatic stage at 933-929 Ma (Schärer et al., 1996;Vander Auwera et al., 2011) or to a later pulse of jotunitic magma in dykes at 920-916 Ma (Vander Auwera et al., 2011).Undeformed reaction rims between jotunitic and granitic layers at Moi (sample 84-24) (Vander Auwera, 1993) point to the early occurrence of jotunite in the envelope of the RAP prior to the deformation linked to the anorthosite emplacement.Such jotunite could thus be coeval with the very early stage of the HAOM (High Aluminum Orthopyroxene Megacryst) formation dated at 1041 ± 17 Ma (Bybee et al., 2014;Vander Auwera et al., 2014b).
Do we have evidence of magmatic events in the FSZ during the main Sveconorwegian orogeny (1070 -1020 Ma)?In the building up of the Sirdal magmatic belt, the banded gneiss units, coined by Falkum (1982), were described as "N-S-oriented zones rich in xenoliths" or as "screens of metamorphosed and deformed rocks" by Coint et al. (2015).Slagstad et al. (2018) show a geological map of the SMB on which kilometre-wide screens can be followed for several tens of kilometres.The xenoliths comprise a large variety of rocks from metabasites to granitoidic grey gneisses in various stages of migmatisation.
The banded gneiss units of Falkum were intruded by the SMB products and are thus at least partly older than the 1070-1010 Ma SMB.
The augen gneisses in the FSZ have calc-alkaline affinities comparable to the Feda trend and could have intruded the banded gneiss unit at the time of the SMB magmatism in the same way as the 1051 Ma Liland augen gneiss intrusion (Bingen, 1989;Bingen & van Breemen, 1998) at some distance from the main magmatic belt.However, they are geochemically distinct from the Liland intrusion but analogous to the Feda augen gneiss.Roberts et al., 2013).
Finally, we have some indications that the FSZ could have integrated lithologies as old as 1500 Ma.Our granite layers are compositionally similar to the Suldal arc grey gneisses and granitoids.In addition, our amphibolite-and biotite-bearing metabasites and our metapelitic rocks could represent fragments of an oceanic crust and its sedimentary cover cotemporaneous with or even older than the c. 1500 Ma Suldal arc.

Conclusions
The FSZ has been intruded by various types of magmatic rocks, besides the anorthosite massifs and the BKSK layered intrusion and more specifically in the studied area by the Hidra anorthosite and the Fig.2

Figure 3 .
Figure3.Geological map afterFalkum (1982) around the Lundevatn with location of 3 samples in a banded gneiss unit.

Figure 4 .
Figure 4. Harker diagrams showing the discriminating characters of the jotunitic metabasites compared with the other metabasites.A typical jotunite composition (the Tjörn parental magma ofDuchesne & Hertogen, 1988) is also plotted to show its similarities with the jotunitic metabasites.
14.It appears that our granite gneisses (with typical high Th, Pb, Rb and [La/Yb] N ratios; Fig.14A, B, C, D) are not represented in the Suldal arc while our granitic layers cannot be distinguished from the Suldal lithologies (Fig.14A, B, C, D).Thus, we cannot preclude that our granite layers could have been formed by the same process and at the same time as the age of the Suldal arc.

Figure 12 .
Figure 12.Harker diagrams comparing the charnockitic rocks (Breimyrknutan intrusion and charnockitic layers in banded gneiss) with the charnockitic chemical variation trends (LLD) and assumed parental magma composition AD011 of the Farsund intrusion (data from VanderAuwera et al., 2014a).

Farsund
massif.Charnockitic layers or bodies were possibly derived by local anatexis, such as the Breimyrknutan granite, or derived from fractionation of a quartz monzonitic magma, contemporaneous with the RAP magmatism in the 933-916 Ma interval.Jotunites were also emplaced during this event or slightly before.Augen gneisses similar to the Feda gneiss are coeval with the 1070-1010 Ma SMB.It cannot be precluded that granitic banded gneisses, other than migmatitic leucosomes, could have been produced by the same arc-related magmatism as the 1500 Ma Suldal arc.Metabasites with oceanic signatures may represent fragments of the ocean crust involved in the Suldal arc subduction and the kinzigitic gneisses could be relics of sedimentary material deposited on that crust.The shear zone has thus been active during more than 500 Myr and has tapped a variety of sources.It is a major component of the Sveconorwegian orogen.