1. Introduction
Peralkaline, high-silica lavas and intrusive rocks are minor members of the alkaline rock spectrum of most continental rifts (e.g. Wilson, Reference Wilson1986), but they are prominent in some rift complexes, such as in parts of the East African Rift (Baker, Reference Baker, Fitton and Upton1987; Macdonald, Reference Macdonald, Fitton and Upton1987; Peccerillo et al. Reference Peccerillo, Gezaegn and Dereje1994) and in the late Palaeozoic Oslo Rift in Norway (Neumann et al. Reference Neumann, Wilson, Heeremans, Spence, Obst, Timmermann, Kirstein, Wilson, Neumann, Davies, Timmerman, Heeremans and Larsen2004; Larsen et al. Reference Larsen, Olaussen, Sundvoll and Heeremans2008). Although much of the compositional diversity of magmas in continental rift settings can be explained by fractional crystallization of mafic parental magmas derived from lithospheric or sublithospheric mantle (e.g. Peccerillo et al. Reference Peccerillo, Gezaegn and Dereje1994; Furman, Reference Furman2007; Upton, Reference Upton2013; Hutchison et al. Reference Hutchison, Mather, Pyle, Boyce, Gleeson, Yirgu, Blundy, Ferguson, Vye-Brown, Millar, Sims and Finch2018; Rooney, Reference Rooney2020), additional processes such as crustal anatexis, crustal contamination of mantle-derived magma and volatile-aided element transfer may be needed to generate peralkaline, silicic melt (Dietrich & Heier, Reference Dietrich and Heier1967; Neumann et al. Reference Neumann, Andersen and Hansteen1990a; Macdonald, Reference Macdonald, Fitton and Upton1987; Rasmussen et al. Reference Rasmussen, Neumann, Andersen, Sundvoll, Fjerdingstad and Stabel1988; Pedersen et al. Reference Pedersen, Heeremans and van der Beek1998). Furthermore, there is evidence from mantle xenoliths and melting experiments that liquids of phonolitic to trachytic composition may form directly by partial melting of metasomatized mantle peridotite (e.g. Ionov et al. Reference Ionov, Hofmann and Shimizu1994; Draper & Green, Reference Draper and Green1997; Laporte et al. Reference Laporte, Lambart, Schiano and Ottolini2014), evidence on which Ashwal et al. (Reference Ashwal, Torsvik, Horváth, Harris, Webb, Werner and Corfu2016) relied to suggest that phonolitic to trachytic lavas in an oceanic island environment at Mauritius were formed from primary, phonolitic mantle melts. This is an alternative petrological model which may possibly also apply to intermediate to felsic, peralkaline rocks in a continental rift setting.
The late Palaeozoic Oslo Rift in southeastern Norway (Larsen et al. Reference Larsen, Olaussen, Sundvoll and Heeremans2008) is host to rift-related extrusive and intrusive igneous rocks ranging in composition from nephelinite and various types of alkaline and subalkaline basalts, nepheline syenite and nepheline monzosyenites, through intrusive monzonite and latitic lavas to syenites and granites, including peralkaline types. The exposed igneous complex shows a dominance of intermediate to felsic rocks over mafic rocks (e.g. Barth, Reference Barth1945), which is compensated by the presence of high-density mafic–ultramafic rocks at depth, the presence of which is revealed by a positive gravity anomaly along the rift (Ramberg, Reference Ramberg1976) and by scarce enclaves of cumulate material in lavas and intrusive rocks (Neumann et al. Reference Neumann, Andersen and Mearns1988a; Andersen & Seiersten, Reference Andersen and Seiersten1994). Estimates based on combined geophysical and petrological evidence indicate that the volume of mafic-ultramafic intrusions and cumulate bodies at depth is similar to or greater than the exposed + eroded volume of igneous rocks in the shallow crust (Neumann et al. Reference Neumann, Wilson, Heeremans, Spence, Obst, Timmermann, Kirstein, Wilson, Neumann, Davies, Timmerman, Heeremans and Larsen2004). The intermediate part of the rock spectrum of the rift, comprising monzonitic and silica-undersaturated intrusive rocks, has been interpreted as products of differentiation of a common, mantle-derived, mildly alkaline basaltic parental magma, without significant contamination by crustal rocks (Neumann, Reference Neumann1980; Neumann et al. Reference Neumann, Tilton and Tuen1988b, Reference Neumann, Wilson, Heeremans, Spence, Obst, Timmermann, Kirstein, Wilson, Neumann, Davies, Timmerman, Heeremans and Larsen2004). In this scenario, silica-undersaturated derivative melts developed by fractional crystallization in staging chambers in the deep crust and silica-saturated-to-oversaturated monzonitic magmas at successively shallower depths, leaving behind mafic to ultramafic cumulates that account for the positive gravity anomaly. The petrological model was originally derived by comparison with experimental data on simplified systems (Neumann, Reference Neumann1980) and has more recently been supported by energy-constrained simulation of fractional crystallization processes by Rämö et al. (Reference Rämö, Andersen and Whitehouse2022).
There are large intrusive complexes in the rift consisting mainly of metaluminous to mildly peralkaline alkali feldspar syenite and alkali feldspar granite with late-crystallized sodic pyroxene and/or amphibole as characteristic mafic silicate minerals. Sr, Nd and Pb isotope data suggest an influence of a heterogeneous continental crust in at least some of these rocks (Neumann et al. Reference Neumann, Tilton and Tuen1988b; Sundvoll & Larsen, Reference Sundvoll and Larsen1990). Nevertheless, a mantle-derived component isotopically similar to that of basaltic and intermediate rocks remains prominent in most of the felsic intrusive complexes (Andersen & Knudsen, Reference Andersen and Knudsen2000).
Cumulates are igneous rocks formed by the accumulation of mineral grains precipitated from a silicate melt, characterized by frameworks of touching crystals (Wager et al. Reference Wager, Brown and Wadsworth1960; Irvine, Reference Irvine1982). Settling of crystals and formation of igneous layering are common, but not essential, processes in the formation of cumulate rocks (Irvine, Reference Irvine1982). Although cumulates are best known from mafic intrusions, anorthosites and agpaitic nepheline syenites (Wager & Brown, Reference Wager and Brown1968; Sørensen & Larsen, Reference Sørensen, Larsen and Parsons1986), cumulate networks have also been described from granitic and syenitic rocks (e.g. Vernon & Collins, Reference Vernon and Collins2011; Schaen et al. Reference Schaen, Singer, Cottle, Garibaldi, Schoene, Satkoski and Fournelle2018; Laurent et al. Reference Laurent, Björnsen, Wotzlaw, Bretscher, Pimenta Silva, Moyen, Ulmer and Bachmann2020; Shellnutt, Reference Shellnutt2021; Leandro et al. Reference Leandro, Stevens, Moyen, Kisters and Ferreira2025). In the Oslo Rift, trace element signatures indicating feldspar accumulation have been reported from some of the intermediate and silica-undersaturated intrusions and from porphyritic lavas (Neumann, Reference Neumann1980; Rasmussen et al. Reference Rasmussen, Neumann, Andersen, Sundvoll, Fjerdingstad and Stabel1988), but so far not from any of the syenitic or granitic intrusions. Igneous layering is locally prominent in intermediate intrusions (Oftedahl & Petersen, Reference Oftedahl and Petersen1978), but has also been reported from syenitic rocks (Andersen, Reference Andersen1984a). Mesocumulate textures (as defined by Wager et al. Reference Wager, Brown and Wadsworth1960) have been observed in several of the syenitic plutons (Andersen, Reference Andersen1984a; Rasmussen et al. Reference Rasmussen, Neumann, Andersen, Sundvoll, Fjerdingstad and Stabel1988). These observations suggest that crystal accumulation, mainly of feldspar, must be considered as a potential process in the evolution of syenitic and granitic intrusions in the Oslo Rift.
The focus of most published petrographic descriptions of plutonic rocks in the Oslo Rift has been the identification of magmatic mineral assemblages and characterization of crystallization sequences, fluid inclusion assemblages and alteration features, whereas less attention has been given to the overall, meso-scale textural features of the rocks. The purpose of this study is to re-examine the petrography and whole-rock compositions of quartz-normative monzonitic, peralkaline or near-peralkaline plutonic syenites and granites, and associated felsic peralkaline dyke rocks in the Oslo Rift. Questions to be addressed are whether whole-rock compositions of plutonic syenites and granites represent unmodified magmatic liquid compositions, to what extent they are affected by crystal accumulation, and if or how the magmas from which they formed can be related to the polybaric differentiation scenario for the intermediate rocks of Neumann (Reference Neumann1980) and Rämö et al. (Reference Rämö, Andersen and Whitehouse2022). For this purpose, the energy-constrained magmatic evolution model of Rämö et al. (Reference Rämö, Andersen and Whitehouse2022) is extended towards syenitic and granitic derivative melt compositions, and crustal contamination is incorporated into the model.
2. Geological setting
The Oslo Rift (Fig. 1) is the exposed, onshore part of a post-Variscan system of continental rift-related basins extending from northern Germany and Poland to the North Sea (Heeremans et al. Reference Heeremans, Faleide, Larsen, Wilson, Neumann, Davies, Timmerman, Heeremans and Larsen2004). It is built up by three half-grabens, namely, a from north to south, the non-magmatic Rendal Graben, the Akershus and Vestfold grabens with voluminous magmatism (Larsen et al. Reference Larsen, Olaussen, Sundvoll and Heeremans2008), and an offshore submerged magmatic graben segment (the Skagerrak Graben, Heeremans & Faleide, Reference Heeremans, Faleide, Wilson, Neumann, Davies, Timmerman, Heeremans and Larsen2004). The onshore rift cuts Palaeoproterozoic to Mesoproterozoic crustal rocks of southwestern Fennoscandia (Gaal & Gorbatschev, Reference Gaal and Gorbatschev1987; Bingen et al. Reference Bingen, Andersson, Söderlund and Möller2008; Andersen & Knudsen, Reference Andersen and Knudsen2000). Rift development in late Carboniferous to Permian time comprised six distinct tectonomagmatic stages, starting with (1) initial subsidence, sedimentation and sill emplacement, followed by (2) basaltic volcanism, (3) graben formation, intermediate volcanism and emplacement of the first monzonitic to nepheline syenitic plutonic rocks, (4) development of central volcanic complexes (preserved as cauldron structures), and terminating with two stages (5 and 6) of emplacement of syenitic and granitic intrusions (Ramberg & Larsen, Reference Ramberg and Larsen1978; Larsen et al. Reference Larsen, Olaussen, Sundvoll and Heeremans2008).
Simplified geological map showing the main intrusive complexes of the Oslo Rift. LPC: Larvik Plutonic Complex, SMC: Siljan–Mykle Complex, ESC: Eikeren–Skrim Complex, SCP: Sande Cauldron central pluton. EB: Eikeren–Bergsvann area with minor felsic intrusions related to the Sande Cauldron and younger SLG dykes (A-51, A-71 in Appendix 2), NHC: Nordmarka–Hurdal Complex, G: Gjerdingselva elpidite-bearing granite, GS: Grefsen syenite intrusion.
2.a. Intermediate and felsic intrusive rocks in the Oslo Rift
2.a.1. A note on petrographic nomenclature
An elaborate petrographic nomenclature system has been developed for the intrusive rocks of the Oslo Rift (Brøgger, Reference Brøgger1890, Reference Brøgger1894, Reference Brøgger1898, Reference Brøgger1906, Reference Brøgger1932, Reference Brøgger1933; Barth, Reference Barth1945), comprising many non-standard terms. A systematic use of IUGS-approved petrographic nomenclature (e.g. Le Maitre, Reference Le Maitre2002) for these rocks has proved unsatisfactory (e.g. Neumann, Reference Neumann1976) because of the presence of ternary feldspar rather than plagioclase + alkali feldspar in important rock types (Neumann, Reference Neumann1980). Furthermore, rock types that form mappable units, if classified using the traditional petrographic nomenclature (e.g. Pedersen et al. Reference Pedersen, Heaman and Holm1995; Bonin & Sørensen, Reference Bonin and Sørensen2003), tend to straddle field boundaries in the QAFP and TAS diagrams. To avoid such problems and to preserve compatibility with previous publications, we follow Rasmussen et al. (Reference Rasmussen, Neumann, Andersen, Sundvoll, Fjerdingstad and Stabel1988) in using a simplified version of the nomenclature system of Barth (Reference Barth1945). Non-standard terms are given in italics when first presented. Brief explanations of the most important non-standard terms are given in Table 1, and further details in Appendix 1 (in online supplementary data). More elaborate general petrographic descriptions of the major rock types can be found in the appendix of Rasmussen et al. (Reference Rasmussen, Neumann, Andersen, Sundvoll, Fjerdingstad and Stabel1988).
2.a.2. The major plutonic complexes
Intermediate to felsic plutonic rocks are concentrated in several multigenetic batholiths (Fig. 1). The southernmost of these is the Larvik Plutonic Complex (LPC), consisting of at least ten distinct intrusions of silica under- to oversaturated monzonite (larvikite), nepheline-bearing monzonite/nepheline monzosyenite (lardalite) and minor nepheline syenite, including agpaitic varieties and pegmatites (Neumann, Reference Neumann1980; Groome, Reference Groome2017; Sunde et al. Reference Sunde, Friis and Andersen2019; Buelens et al. Reference Buelens, Debaille, Decrée, Coint and Mansu2024). The LPC was emplaced at 302–288 Ma (Rämö et al. Reference Rämö, Andersen and Whitehouse2022), contemporaneously with fissure eruptions forming voluminous latitic rhomb porphyry lavas during stage 3 in the chronology of Larsen et al. (Reference Larsen, Olaussen, Sundvoll and Heeremans2008).
The northern part of the LPC is cut by intrusive rocks of the Siljan–Mykle Complex (SMC), comprising gabbro, larvikite, nepheline syenite to alkali feldspar syenite (nordmarkite) and peralkaline alkali feldspar granite (ekerite) (Pedersen et al. Reference Pedersen, Heaman and Holm1995; Holm & Larsen, Reference Holm and Larsen1997). Minor rock types of potential petrogenetic significance in the SMC are intrusions of microsyenite and porphyritic syenite that are petrographical intermediates between larvikite and nordmarkite (Petersen & Sørensen, Reference Petersen and Sørensen1997; Andersen & Sørensen, Reference Andersen and Sørensen2003; Andersen et al. Reference Andersen, Frei, Sørensen and Westphal2004). The SMC also provides a rare example in the Oslo Rift of an exposed, near-horizontal roof contact of an ekerite intrusion emplaced into larvikite, with evidence of fluid and melt infiltration into the overlying larvikite (Bonin & Sørensen, Reference Bonin and Sørensen2003).
The SMC is in turn cut by the Eikeren–Skrim Complex (ESC), consisting exclusively of ekerite. The complex was probably constructed by multiple intrusions of rather uniform composition (Neumann et al. Reference Neumann, Andersen and Hansteen1990a), which have not been separated on the published geological map of the area (Nilsen & Siedlecka, Reference Nilsen and Sieldecka2003). The central pluton of the Sande Cauldron (SCP) east of the ESC is an isolated, smaller, composite pluton consisting of a core of larvikite embedded in an outer zone of younger nordmarkite, which locally grades into quartz-rich nordmarkite (Andersen, Reference Andersen1984a; Rasmussen et al. Reference Rasmussen, Neumann, Andersen, Sundvoll, Fjerdingstad and Stabel1988). The contact between larvikite and nordmarkite in the SCP is marked by a 10 to 800 m wide (sub)vertical transition zone in which infiltration of syenitic melt into larvikite and local hybridization has taken place (Andersen, Reference Andersen1984b). On the published 1:50 000 geological map of the area, Gunby et al. (Reference Gunby, Heyer, Ihlen and Siedlecka2010) drew a profile through the complex showing a horizontal boundary between larvikite and underlying nordmarkite, as well as further, unexposed, hypothetic larvikite bodies. The bottom of the central larvikite body (Fig. 1) is nowhere exposed, and its geometry remains speculative. Rb-Sr isotope data on nordmarkite in the SCP suggest an influence by local crustal contamination (Rasmussen et al. Reference Rasmussen, Neumann, Andersen, Sundvoll, Fjerdingstad and Stabel1988; Sundvoll & Larsen, Reference Sundvoll and Larsen1990).
Two composite intrusions of metaluminous to mildly peraluminous biotite granite make up the central part of the rift. These represent a separate petrological lineage, which will not be considered here (Trønnes & Brandon, Reference Trønnes and Brandon1992; Haug, Reference Haug2007).
The intrusive rocks of the Akershus Graben form one large, composite batholith, the Nordmarka–Hurdal Complex (NHC), with larvikite and related monzonitic rocks and different varieties of syenite and granite (Sæther, Reference Sæther1962; Rasmussen et al. Reference Rasmussen, Neumann, Andersen, Sundvoll, Fjerdingstad and Stabel1988, Larsen et al. Reference Larsen, Lutro, Sæther and Nilsen2001; Lutro, Reference Lutro2001; Olerud, Reference Olerud2002). Intrusions of commonly fine-grained and porphyritic, sub-solvus monzonite to quartz monzonite (akerite) and plagioclase-bearing, and in part porphyritic syenite (the Grefsen syenite intrusion, indicated by GS in Fig. 1) make up the southern margin of the NHC (Oftedahl, Reference Oftedal1946; Sæther, Reference Sæther1962; Graversen et al. Reference Graversen, Nordgulen and Lutro2017). In the central part of the NHC, nordmarkite is cut by a ca. 2 by 4 km body of ekerite (the Gjerdingselva ekerite intrusion, G in Fig. 1) carrying elpidite (Na2ZrSi6O15·3H2O), janhaugite ((Na,Ca)3(Mn2+,Fe2+)3(Ti,Zr,Nb)2(Si2O7)2O2(OH,F)2), kupletskite ((K,Na)3(Mn,Fe)7Ti2Si8O26(OH)4F) and other rare Zr-Ti-Nb silicate minerals and fluorides (Sæther, Reference Sæther1962; Raade & Mladeck, Reference Raade and Mladeck1983; Larsen et al. Reference Larsen, Lutro, Sæther and Nilsen2001; Raade, Reference Raade2005; online database entry at https://www.mindat.org/loc-2502.html). Intrusive rocks in the northernmost part of the NHC (biotite-bearing syenites and granites) are influenced by local crustal contamination (Rasmussen et al. Reference Rasmussen, Neumann, Andersen, Sundvoll, Fjerdingstad and Stabel1988; Neumann et al. Reference Neumann, Tilton and Tuen1988b; Andersen & Knudsen, Reference Andersen and Knudsen2000) and are not considered here.
The syenitic to granitic plutons of the Oslo Rift were emplaced later than the LPC, during the final two stages of rift evolution (Larsen et al. Reference Larsen, Olaussen, Sundvoll and Heeremans2008). Rb-Sr geochronology (whole-rock and mineral isochrons) indicates a time span of emplacement from 265 Ma to 255 Ma for the main felsic rocks, but with the final intrusive activity as late as 245 Ma (Larsen et al. Reference Larsen, Olaussen, Sundvoll and Heeremans2008). This chronology has been modified by U-Pb dating of zircon, which suggests emplacement ages of 286-260 Ma for the main felsic plutons and termination of intrusive activity by 250 Ma (Pedersen et al. Reference Pedersen, Heaman and Holm1995; Haug, Reference Haug2007; Borg, Reference Borg2011; Olsen, Reference Olsen2018; T. Andersen unpublished data). Basalts, rhomb porphyries and LPC larvikites show initial 87Sr/86Sr ≤ 0.7045 and epsilon-Nd of + 0.5 to + 4, syenitic and granitic rocks tend to have epsilon-Nd values down to zero and initial 87Sr/86Sr of 0.7040–0.7067, with values up to 0.720 in the northernmost part of the NHC (Neumann et al. Reference Neumann, Tilton and Tuen1988b). Most of the plutonic complexes define separate Rb-Sr isochrons with well-constrained initial ratios that show minor differences between the complexes. This has been interpreted as a result of crustal contamination at depth at an early stage of magmatic evolution, before magmatic differentiation within the individual complexes (Rasmussen et al. Reference Rasmussen, Neumann, Andersen, Sundvoll, Fjerdingstad and Stabel1988; Neumann et al. Reference Neumann, Tilton and Tuen1988b; Neumann et al. Reference Neumann, Sundvoll and Øverli1990b; Sundvoll & Larsen, Reference Sundvoll and Larsen1990).
2.a.2. Dyke rocks
Brøgger (Reference Brøgger1894, Reference Brøgger1898, Reference Brøgger1932) described a suite of aegirine and/or arfvedsonite-bearing dykes assumed to be genetically related to the felsic plutons. The dykes range from nepheline- to quartz-bearing and typically show a flow-laminated microstructure, defined by platy alkali feldspar phenocrysts and needle-shaped aegirine. Brøgger classified these dyke rocks in an elaborate (and not always consistent, c.f. Sæther, Reference Sæther1947) nomenclature system, which included several newly defined rock names, none of which forms part of currently accepted petrographic nomenclature (Le Maitre, Reference Le Maitre2002). Among the locally defined names, only grorudite has found some use outside of the Oslo Rift (e.g. Noble, Reference Noble1948; Plechov et al. Reference Plechov, Ushakova and Scherbakov2023). Here, these dykes are informally grouped as SLG dykes (after the most prominent types in Brøgger’s nomenclature: sølvsbergite, lindøite, lestiwarite and grorudite; Appendix 1). SLG dykes are in general young features in the magmatic history of the Oslo Rift, intruding stage 4 lavas and intrusions, as well as stage 5 plutons (Brøgger, Reference Brøgger1894; Sæther, Reference Sæther1946; Naterstad, Reference Naterstad1978). When recalculated with the decay constant of 87Rb of 1.3972·10−11 a−1 (Villa et al. Reference Villa, De Biéver, Holdne and Renne2015), whole-rock and mineral Rb-Sr isotope data on a single SLG (grorudite) dyke from the southern margin of the NHC (Sundvoll and Larsen, Reference Sundvoll and Larsen1993) give a regression line with an age of 254 ± 7 Ma (MSWD = 6.3) and (87Sr/86Sr)i= 0.706 ± 0.010. This suggests that SLG dykes were emplaced towards the end of the magmatic history of the Oslo Rift, but the initial 87Sr/86Sr ratio of the regression line is too imprecise to be useful as a source indicator.
In addition to the SLG dykes, quartz-feldspar porphyry dykes and minor intrusions with sodic pyroxene and amphibole occur in different parts of the Oslo Rift, some of which are associated with cauldron structures, whereas others have no close spatial association to exposed, felsic plutonic rocks (e.g. Brøgger, Reference Brøgger1932; Sæther, Reference Sæther1947).
3. Modal mineralogy and microstructure
3.a. Plutonic rocks
Textural evidence of the presence or absence of accumulated minerals in medium- to coarse-grained intrusive rocks may be more easily observed in low-magnification whole-thin-section scans than at the higher magnification of a standard polarizing microscope (e.g. Schaen et al. Reference Schaen, Singer, Cottle, Garibaldi, Schoene, Satkoski and Fournelle2018). Thin sections from previous studies of Oslo Rift plutons have therefore been re-examined, using a modified optical slide scanner, examples of which are shown in Fig. 2. The material studied comprises thin sections from the W.C. Brøgger and G. Raade collections and published studies and academic theses kept at the Natural History Museum, University of Oslo (see Appendix 2), supplemented by sections from the first author’s collection.
Low-magnification photomicrographs of plutonic rocks. (a) Porphyritic syenite from the Siljan–Mykle complex (Andersen et al. Reference Andersen, Frei, Sørensen and Westphal2004). Phenocrysts of zoned anorthoclase (grey in hand specimen) embedded in a matrix consisting of a network of subhedral alkali feldspar crystals with interstitial quartz. (b) Porphyritic, quartz-bearing nordmarkite, outer contact of the Sande Central Pluton (SCP) against Silurian sandstone. Subhedral phenocrysts of alkali feldspar (total ca. 38 %) embedded in a fine-grained, granophyric quartz-alkali feldspar groundmass (ca. 62 %, excluding quartz xenocryst). Larger, rounded quartz grains are clastic grains picked up from the wall rock. Sample A6 of Andersen (Reference Andersen1984a). (c) Nordmarkite from the SCP showing a network of early, zoned alkali feldspar crystals with adcumulus overgrowths embedded in a matrix of smaller, subhedral alkali feldspar grains and interstitial quartz. Sample A89 of Andersen (Reference Andersen1984a). (d) Quartz-free nordmarkite from the SCP showing a well-developed mesocumulate microstructure, in which subhedral alkali feldspar crystals form a network of grains in contact with each other, with interstitial sodic-calcic amphibole (brown) and aegirine-augite (green). Sample R286 (Andersen (Reference Andersen1984a). (e) Quartz-bearing syenite from the Grefsen syenite intrusion at the southern margin of the NHC (sample 933 in the G. Raade collection in the Natural History Museum, University of Oslo), showing feldspar crystals in grain contact with only minor adcumulus overgrowth and interstitial aggregates of quartz, alkali feldspar, biotite and magnetite. (f) Quartz-bearing nordmarkite from the Nordmarka–Hurdal Complex (sample R796, Neumann, Reference Neumann1976), touching alkali feldspar crystals, make up a continuous network structure with interstitial albite, quartz and sodic pyroxene. (g) A cluster of subhedral alkali feldspar laths with interstitial aegirine embedded in a matrix consisting of smaller, subhedral feldspar grains and quartz, and granophyric intergrowths of alkali feldspar and quartz (lower right). Ekerite sample TH53 (Neumann et al. Reference Neumann, Andersen and Hansteen1990a) from the ESC. (h) Ekerite sample TH77 from the ESC (Neumann et al. Reference Neumann, Andersen and Hansteen1990a) consists of a network of touching subhedral alkali feldspar laths with interstitial quartz and sodic amphibole and pyroxene. Note that some of the quartz show crystal face terminations and are partly embedded in alkali feldspars. (i) Fine-grained ekerite from the ESC, sample R265 (Neumann et al. Reference Neumann, Andersen and Hansteen1990a). Subhedral grains of alkali feldspar and quartz are embedded in a fine-grained granophyric quartz-alkali feldspar groundmass.
Plutonic, porphyritic syenites from the SMC (Petersen & Sørensen, Reference Petersen and Sørensen1997; Andersen et al. Reference Andersen, Frei, Sørensen and Westphal2004) and the Grefsen syenite intrusion in the NHC (Sæther, Reference Sæther1962) contain prominent, grey plagioclase or anorthoclase phenocrysts that are commonly zoned and set in a finer-grained matrix built up by a network of subhedral, microperthitic alkali feldspar grains with interstitial quartz (Fig. 2a). The feldspar phenocrysts in the SMC porphyritic syenite are similar to the grey feldspar of larvikite, and the rock type has been considered a derivative of larvikite (Petersen & Sørensen, Reference Petersen and Sørensen1997; Andersen et al. Reference Andersen, Frei, Sørensen and Westphal2004). There are, however, no gradual transitions between such syenite and larvikite – where the rock types are in contact, the porphyritic syenite is sharply cross-cutting.
Fig. 2b shows an example of a porphyritic nordmarkite from the SCP, close to the contact to Silurian sandstone. The rock has 30–40 volume % of subhedral alkali feldspar phenocrysts, contained in a fine-grained groundmass of alkali feldspar and quartz, and contains rounded, clastic quartz grains picked up from the wall rock. This indicates some mechanical contamination during emplacement, but at 64 % SiO2, the net effect of contamination with sandstone must have had only a moderate effect on the whole-rock major element composition. This sample may be interpreted as a chilled margin facies of the pluton, in which the feldspar phenocrysts must have crystallized before emplacement to the present level. The nordmarkite samples shown in Fig. 2c and d, also from the SCP, consist of a network of zoned alkali feldspar crystals in point contact, in c with distinct overgrowths, approaching an adcumulate texture as defined by Wager et al. (Reference Wager, Brown and Wadsworth1960). The samples contain interstitial patches of quartz, magnetite, and sodic pyroxene and amphibole (c), or albite and sodic pyroxene/amphibole (d), and show alkali feldspar mesocumulate textures. Fig. 2e shows an example of a syenite from the southern margin of the NHC (Grefsen syenite intrusion), showing a network of touching feldspar crystals with minimal adcumulus overgrowth and minor interstitial quartz, biotite and magnetite. The example in Fig. 2f (from nordmarkite in the NHC) shows a grid of euhedral to subhedral alkali feldspar crystals in grain contact and with a tendency towards parallel orientation of the larger crystals, but without significant feldspar overgrowths. Interstitial minerals are quartz and sodic pyroxene/amphibole. Quartz is optically continuous over areas larger than the individual interstice. Fig. 2g, h and i show section scans of ekerite from the ESC (Neumann et al. Reference Neumann, Andersen and Hansteen1990a). Fig. 2g and h show rather closely packed aggregates of alkali feldspar intermixed with quartz crystals, some of which have euhedral terminations against feldspar grains. Mafic silicates are aegirine (dark green) and sodic amphibole (paler green, zoned). In g, there is also a patch of interstitial, granophyric quartz-alkali feldspar intergrowth. Interstitial granophyric textures are even better developed in i, which shows a relatively fine-grained sample of ekerite with subhedral grains of quartz and alkali feldspar embedded in a granophyric groundmass.
Some detailed microstructures in ekerite samples by conventional petrographic microscopy are shown in Fig. 3. Fig. 3a shows (at least) two generations of quartz crystallization: Subhedral quartz grains completely enclosed by laths of alkali feldspar (Q1) represent relatively early crystallization of quartz, postdated by anhedral quartz grains forming interstitial grain mosaics (Q2). In Fig. 3b, quartz grains embedded in alkali feldspar (Q1) are truncated by interstitial quartz (Q2). Interstitial quartz grains are in some examples subhedral and occur both embedded in rims of alkali feldspar grains, and surrounding grain edges of alkali feldspar, suggesting simultaneous growth (Fig. 3c). Aegirine and arfvedsonite are everywhere interstitial to both. Although sodic pyroxene and amphibole are by far the most abundant mafic silicate minerals in ekerite, a range of other minerals also occur (Dietrich et al. Reference Dietrich, Heier and Taylor1965; Neumann et al. Reference Neumann, Andersen and Hansteen1990a), with astrophyllite as the most widespread (Fig. 3d). Elpidite and other rare Zr-Ti-Nb silicate minerals have so far only been reported from the Gjerdingselva ekerite intrusion (e.g. Raade & Mladeck, Reference Raade and Mladeck1983).
Quartz-feldspar relationships and interstitial mineralogy in ekerite from the ESC. (a) Two generations of quartz in sample R265 (Neumann et al., Reference Neumann, Andersen and Hansteen1990a, cf. Figure 2i). Subhedral quartz grains are enclosed by alkali feldspar (Q1), which is in turn surrounded by an interstitial quartz-feldspar granophyric matrix (Q2). (b) A perthitic alkali feldspar crystal enclosing quartz grains (Q1), which are in turn truncated by quartz forming an interstitial mosaic of anhedral grains (Q2). Sample TH53 (Neumann et al. Reference Neumann, Andersen and Hansteen1990a). Crossed polarized micrograph enhanced by 1 λ accessory plate. (c) Interstitial aggregate in ekerite sample R225 from the ESC (Neumann et al. Reference Neumann, Andersen and Hansteen1990a). The quartz crystals are rounded-subhedral and partly embedded in alkali feldspar (Q2), suggesting simultaneous growth of quartz and late alkali feldspar, whereas sodic amphibole, pyroxene and mosaics of anhedral quartz grains (Q2) are interstitial to both. (d) Cluster of interstitial minerals in sample TH77 (Neumann et al. Reference Neumann, Andersen and Hansteen1990a). The interstitial mineral assemblage consists of quartz (Qz), aegirine (Aeg), arfvedsonite (Arf), astrophyllite (Ast), zircon (Zrn) and fluorite (Fl). Holes created during sectioning are marked with h.
Textural evidence as shown in the examples in Figs. 2 and 3 suggests that the presence of accumulated feldspar in nordmarkite and ekerite, and also quartz in ekerite, is the rule rather than an exception in the Oslo Rift felsic plutons. The rocks, therefore, should be regarded as melt-cumulus mixtures emplaced as crystal-laden slurries rather than as liquids, which has consequences for the interpretation of whole-rock geochemical data and for the modelling of liquid lines of descent.
SLG dykes and quartz-feldspar porphyries. (a) Microscope drawing by Brøgger (Reference Brøgger1894) of the type specimen of grorudite, showing long-prismatic grains of aegirine in a matrix of subhedral alkali feldspar (Afs, grey) and anhedral quartz (Qz, colourless) in which individual quartz and feldspar grains partly enclose aegirine needles (Aeg, green). This type of ‘tinguatitic’ texture was seen as a defining feature of grorudite by Brøgger (Reference Brøgger1894). (b) Photomicrograph of SLG dyke A51 (first author’s collection) from the Eikeren–Bergsvann area, with aegirine needles (Aeg) embedded in alkali feldspar (Afs) and quartz (Qz), differing from the type grorudite shown in a only by a moderately larger grain size. This particular dyke was identified as a ‘lindøite’ by Brøgger (Reference Brøgger1906), but its texture and an SiO2 content of 71.9 wt% are those of grorudite according to the criteria of Brøgger (Reference Brøgger1894). (c) Low-magnification photomicrograph of a trachytic SLG dyke (‘sølvsbergite’ in the terminology of Brøgger, Reference Brøgger1894), showing phenocrysts of biotite and magnetite with inclusions of apatite embedded in a fine-grained alkali feldspar - aegirine matrix. Sample HeII-6 of the W.C. Brøgger collection at the Natural History Museum, University of Oslo. (d) A closeup of the matrix of sample HeII-6, showing needles of aegirine (Aeg) and microphenocrysts of biotite (Bt) embedded in alkali feldspar. (e) Low-magnification photomicrograph of a quartz-alkali feldspar porphyry dyke from the SCP (sample A39, first author’s collection). Quartz phenocrysts are subhedral and somewhat rounded but show a pronounced tendency towards rhombic sections with extinction along the diagonals under crossed polarizers, which suggests that they originally crystallized as beta-quartz. The groundmass consists of dense quartz-alkali feldspar intergrowths with aegirine microphenocrysts (appearing black in the image). (f) Detail of e showing a quartz-aegirine segregation in the groundmass.
3.b. Dyke rocks
SLG dykes vary in colour from pale grey to green, depending on aegirine content. The microscope drawing of the type grorudite sample by W.C. Brøgger (Reference Brøgger1894, reproduced in Fig. 4a) shows a characteristic microstructure in which alkali feldspar and quartz grains partly enclose acicular crystals of aegirine. Such ‘tinguaitic’ texture (Le Maitre, Reference Le Maitre2002) is typical for the SLG dykes (Fig. 4b), also in quartz-free or quartz-poor varieties (Brøgger, Reference Brøgger1894). Fig. 4c and d show a trachytic SLG dyke with phenocrysts of biotite and magnetite (Fig. 4c) in a fine-grained matrix consisting of alkali feldspar enclosing needles of aegirine and laths of biotite (Fig. 4d).
Quartz-feldspar porphyries that do not have the foliated, trachytoid structure typical of SLG dykes have a groundmass consisting of fine-grained quartz-alkali feldspar intergrowths, with microphenocrysts of sodic pyroxene and/or amphibole, and larger, subhedral phenocrysts of alkali feldspar and quartz. The quartz phenocrysts commonly show somewhat rounded and corroded rhombic outlines in thin section (Fig. 4e). Extinction under crossed polarizers is typically parallel to the diagonals of the rhombic sections, suggesting that it crystallized as dipyramids of the beta polymorph. The groundmass of some quartz-feldspar porphyry intrusions contains scattered, polycrystalline segregations consisting of mosaics of quartz grains enclosing needles of aegirine (Fig. 4f).
4. Whole-rock major element data
The first major element analyses of felsic igneous rocks in the Oslo Rift were published by Kjerulf (Reference Kjerulf1855). The quality of the early analyses suffers from the technological limitations of the time and is now only of historical interest. Between ca. 1890 and the early 1930s, W.C. Brøgger, with the assistance of professional analytical chemists, accumulated a large set of whole-rock and mineral analyses using classical wet chemical methods. A compilation of more than 300 whole-rock analyses was published by Brøgger (Reference Brøgger1933). Brøgger’s data cover both the main volcanic and intrusive rock types and several minor varieties and dyke rocks that have received little attention since his time. Where Brøgger’s data can be directly compared to modern data, the accuracy of his analyses appears to be reasonably good. Like other analyses made by early twentieth century wet chemistry methods, the main potential inaccuracies are in minor elements (Mn, P) and in Al and Na, which were determined by difference (e.g. Potts, Reference Potts1987). Modern reinterpretation of his data may be complicated by issues with the size, quality and insufficient documentation of samples. Also, aspects of Brøgger’s very complex petrographical terminology and his emphasis on rock suites or ‘families’ (‘Reihen’ in German, e.g. Brøgger, Reference Brøgger1890, Reference Brøgger1933) are not always easy to relate to modern nomenclature and understanding of differentiation processes. The interest in the chemistry of the igneous rocks of the Oslo Rift was renewed from the 1960s. Modern, instrument-based analytical methods greatly simplified data accumulation, and the total number of individual analyses made since 1965 by far exceeds that of older data, but emphasis has been more on basaltic than felsic rocks (e.g. Neumann et al. Reference Neumann, Larsen and Sundvoll1985, Reference Neumann, Sundvoll and Øverli1990b, Reference Neumann, Dunworth, Sundvoll and Tollefsrud2002; Schou-Jensen & Neumann, Reference Schou-Jensen and Neumann1988; Dunworth et al. Reference Dunworth, Neumann and Rosenbaum2001).
4.a. The data compilation
Based on published data and data in academic theses available from the Library of the University of Oslo (references to unpublished academic theses are given in Appendix 2 in the online supplementary data), 235 whole-rock major element analyses of mainly quartz-normative monzonite (larvikite, akerite) from outside of the LPC, syenitic rocks (syenite/trachyte, nordmarkite), alkali feldspar granite (ekerite), SLG dykes and quartz-feldspar porphyries are considered in this review. The compilation of data is given as a downloadable spreadsheet file in the supplementary data (Appendix 2). Sixty-six of the analyses are wet chemical analyses from Brøgger (Reference Brøgger1933), and the remaining analyses were made between ca. 1965 and 2012 by different instrumental methods, mainly by X-ray fluorescence analysis (XRF). Most of the modern analyses have previously been plotted in TAS diagrams by Neumann et al. (Reference Neumann, Wilson, Heeremans, Spence, Obst, Timmermann, Kirstein, Wilson, Neumann, Davies, Timmerman, Heeremans and Larsen2004). In addition, ranges of 41 analyses of larvikite, lardalite and nepheline syenite from the LPC (Neumann, Reference Neumann1980) and 93 analyses of rhomb porphyry dykes and lavas (from sources given in Appendix 2) have been plotted for comparison.
4.a.1. Plutonic rocks
In Fig. 5, the data from Appendix 2 are plotted in conventional TAS (total alkali – silica) classification diagrams with fields according to Middlemost (Reference Middlemost1994) and R 1 R 2 plots after De la Roche et al. (Reference De la Roche, Leterrier, Grandclaude and Marchal1980). The latter diagram was used by Rämö et al. (Reference Rämö, Andersen and Whitehouse2022) to compare simulated liquid lines of descent with observed rock compositions from the LPC. It is a projection of the basalt tetrahedron, incorporating the constituents of a major element analysis except for manganese and phosphorus. The feldspar–olivine–diopside and the quartz–albite–nepheline joins are represented by straight lines, the former allowing easy discrimination between silica-saturated and -undersaturated rock compositions independently of the Fe2+/Fe3+ ratio. This makes it a useful representation of rock compositions in the Oslo Rift, where variations in silica saturation level are important. Neither the TAS nor the R 1 R 2 diagram allows an easy distinction between metaluminous and peralkaline rocks, which can be done in a plot of agpaitic index (A.I.) vs. SiO2 (Fig. 6). The agpaitic index is here defined by A.I.= (Na2O + K2O)/Al2O3 by mole, which is the original definition of the parameter from Ussing (Reference Ussing1912), differing from the one of Frost & Frost (Reference Frost and Frost2008).
Major element whole-rock data on intermediate to felsic rocks of the Oslo Rift (Appendix 2), shown in conventional total alkali–silica (TAS) plots with grid according to Middlemost (Reference Middlemost1994) and R 1 R 2 plots of De la Roche et al. (Reference De la Roche, Leterrier, Grandclaude and Marchal1980), the latter shows straight lines linking quartz, albite and nepheline, and the limit between silica-undersaturated and -saturated rocks. The data plotted in the diagrams are listed in Appendix 2, with references to the original sources. Contours represent probability density surfaces based on data from 1766 samples from 343 studies on monzonitic, quartz monzonitic, syenitic and granitic rocks worldwide, downloaded from the GeoRoc database (https://georoc.eu/georoc/new-start.asp, accessed February 2025, see Appendix 3). (a), (b) Monzonitic and latitic rock: quartz-normative larvikite from outside the LPC and akerite (quartz monzonite) from the NHC. Ranges of larvikite in the LPC and of rhomb porphyry lavas are shown by outlines. (c), (d) Syenites and quartz syenites, including trachyte/microsyenite and porphyritic syenite from the SMC, nordmarkite and other syenites. (e), (f) Ekerite and associated biotite granite (granite from the major biotite granite batholiths are not shown), quartz-feldspar porphyries and trachytic-rhyolitic SLG dykes.
Plot of agpaitic index vs. SiO2 content recalculated to 100% anhydrous composition. Sample symbols as in Figure 5. G denotes the Gjerdingselva elpidite-bearing granite (ekerite), analysis by Neumann (Reference Neumann1976). Contours as in Figure 5.
Larvikite from the LPC spreads from the monzonite to the syenite and foid monzonite fields in the Middlemost (Reference Middlemost1994) TAS classification, with lardalite and nepheline syenite plotting in the appropriate nepheline syenite field (Fig. 5a). Rhomb porphyries plot within the trachyandesite field of the volcanic TAS diagram (Neumann et al. Reference Neumann, Wilson, Heeremans, Spence, Obst, Timmermann, Kirstein, Wilson, Neumann, Davies, Timmerman, Heeremans and Larsen2004), but with wt % K2O generally larger than wt% Na2O–2.0, they are properly classified as latites (LeMaitre, Reference Le Maitre2002). Rhomb porphyries overlap with the LPC larvikite in terms of total alkalis and silica (Fig. 5a), but tend towards more potassic compositions than the plutonic rocks. Quartz-normative larvikite from outside of the LPC are monzodiorites to monzonites with SiO2 from 53 to 60 wt%. Akerite (SiO2: 57–62 wt%) from the southern part of the NHC (Oftedahl, Reference Oftedal1946; Borg, Reference Borg2011) classifies as monzonite to quartz monzonite (Fig. 5a). Differences in the silica saturation level among the monzonitic rocks are more clearly illustrated in the R 1 R 2 plot, with akerites tending to be more silica oversaturated than the other monzonitic rocks (Fig. 5b). All of the monzonitic and quartz monzonitic rocks are metaluminous, with A.I. from 0.56 to 0.8 and normative An, showing a positive correlation between SiO2 and A.I. (Fig. 6).
Plutonic rocks classified by the original authors as nordmarkite range from mildly silica-undersaturated compositions (SiO2 = 55–64 wt%), but without modal nepheline, through the syenite and quartz monzonite fields in the TAS diagram, to the granite field, at SiO2 = 60 to 72 wt% (Fig. 6c). In some of the plutonic complexes, there are local gradual transitions between rocks that have been mapped as ‘quartz syenite’, ‘nordmarkite’ and ‘ekerite’ (Dietrich et al. Reference Dietrich, Heier and Taylor1965; Andersen, Reference Andersen1984a).
In the R 1 R 2 diagram, nordmarkite samples straddle the critical silica saturation line, clustering around the intersection with the quartz-albite-nepheline line (Fig. 5d). The A.I. ranges from 0.82 to 1.18 (Fig. 6). Porphyritic syenite from the SMC and plagioclase-bearing syenite from the southern margin of the NHC (Grefsen syenite intrusion) overlap with the low-silica-low-alkali part of the nordmarkite field in the TAS diagram at 59–64 wt% SiO2 (Fig. 5c), but are quite distinct in the R 1 R 2 plot (Fig. 5d) and are metaluminous with A.I. = 0.79 to 0.96 (Fig. 6). Syenites, including quartz-free nordmarkite with less than 65 wt% SiO2, form a continuation of the broad, positively correlated trend of A.I. and SiO2 seen in metaluminous monzonitic rocks (Fig. 6), whereas quartz-normative nordmarkites with A.I. between 0.95 and 1.18 show no systematic increase of A.I. with increasing SiO2.
With the exception of a few quartz-poor samples that classify as quartz monzonite, ekerite samples (SiO2 = 61-78 wt%) plot within the granite field in the TAS diagram (Fig. 5e), and along the quartz-alkali feldspar line in the R 1 R 2 plot, forming a continuation of the field of nordmarkite (Fig. 5f). As pointed out by Neumann et al. (Reference Neumann, Andersen and Hansteen1990a), there is no systematic increase of A.I. with increasing SiO2, but a scatter between 0.95 and 1.2, similar to the variation in quartz-normative nordmarkite. With A.I. of 1.23 at 73.4 wt% SiO2 (Neumann, Reference Neumann1976), the Gjerdingselva ekerite has the highest A.I. reported from any plutonic rock in the Oslo Rift (G in Fig. 6).
4.a.2. Dyke rocks
Brøgger (Reference Brøgger1933) listed 19 analyses of SLG dykes. The felsic dykes have attracted little research interest since Brøgger’s time, and only four SLG dykes (all of which have SiO2 > 66 wt% and would be grorudites in his terminology) have been analysed using more modern methods (Appendix 2). Based on the existing analytical data, the dyke rocks span a range of SiO2 contents from 56 to 76 wt% SiO2, plotting in the phonolite, trachyte and rhyolite fields in the TAS classification diagram (Fig. 5e); the higher-silica members (grorudite) classify as pantellerite (Macdonald, Reference Macdonald1974). The presence of aegirine and/or sodic amphibole indicates a peralkaline composition, which is confirmed by the presence of normative acmite in most of the samples (Appendix 2) and A.I. >1 (Fig. 6). Two analyses of trachytic SLG dykes from Brøgger (Reference Brøgger1933) have A.I. <1.0, which is in conflict with the modal mineralogy of the rocks and suggests some postmagmatic alkali loss, or analytical problems (one of the analyses was in fact marked as problematic by Brøgger, Reference Brøgger1933, p. 100). In the R 1 R 2 diagram (Fig. 5f), the SLG dykes plot at or close to the quartz-albite-nepheline line, with the least silicic samples straddling the silica saturation line. The silica-undersaturated SLC dykes overlap with quartz-free nordmarkite in this diagram, whereas the more siliceous dykes duplicate the total range of quartz-normative nordmarkite and ekerite. The SLG dykes (with the exception of altered samples) differ from the plutonic nordmarkite and ekerite by a distinctly higher agpaitic index at comparable SiO2 content (Fig. 6).
Of the five available analyses of fine-grained, non-SLG quartz-feldspar porphyry dykes and minor intrusions (Appendix 2), two are metaluminous at SiO2 > 75 wt% (analyses from Brøgger, Reference Brøgger1933) and possibly affected by fluid-induced alkali loss or low-temperature alteration. The remaining three are less siliceous (72.6–73.5 wt% SiO2) and cluster at A.I.=1.1, which is between the main concentration of ekerite samples and the SLG dykes in the A.I. vs. SiO2 diagram (Fig. 6).
For comparison, Figs. 5 and 6 also show contoured probability density surfaces calculated from a set of 1761 metaluminous and peralkaline monzonites, syenites, quartz monzonites and granites (according to the TAS classification of Middlemost, Reference Middlemost1994) downloaded from the GeoRoc database (https://georoc.eu/georoc/new-start.asp, accessed February 2025); the geographical distribution and GeoRoc source identification numbers are given in Appendix 3 in the online supplementary data. The monzonitic rocks from the Oslo Rift (larvikite outside of the LPC and akerite) show marginal overlap with the global pattern in the TAS, R 1 R 2 and A.I. vs. SiO2 diagrams, whereas nordmarkite and other syenitic rocks, including trachytic SLG dykes, fall near the maxima of the global probability density surfaces. Ekerites and rhyolitic SLG dykes trend towards more silicic compositions, overlapping with the high-R 1 part of the global field in the R 1 R 2 plot but ranging outside of the field in the other plots.
4.b. Trace element distributions
The data coverage for trace elements in the felsic intrusive rocks in the Oslo Rift is scarcer than for major elements. Outside of the LPC (Neumann, Reference Neumann1980), only two of the plutonic complexes considered here are covered by comprehensive and systematic trace element data: the ESC (Neumann et al. Reference Neumann, Andersen and Hansteen1990a) and the SCP (Appendix 2). Most of the data are 1970s–80s instrumental neutron activation analyses, with a restricted selection of elements and sensitivity and analytical precision inferior to modern ICPMS analyses.
Trace element distribution patterns of the Oslo Rift intrusive rocks (Fig. 7) normalized to average upper continental crust (Rudnick & Gao, Reference Rudnick and Gao2014) are generally enriched in Rare Earth Elements (REE), Zr and Hf relative to the average upper continental crust. The monzonitic and syenitic rocks fall at or above the 75 percentile limit for these elements in metaluminous and peralkaline monzonite to granite from ‘syenite complexes’ worldwide (Fig. 7a, b), whereas the ekerites and SLG dykes are distinctly higher (Fig. 7c). The larvikites show no depletion in Sr, whereas the syenitic and granitic rocks are increasingly depleted, and also in P and Ti. Ta is enriched, whereas U and Th are within the interquartile range of the global comparison data. Ba is moderately high in the larvikites and drops to lower values relative to the upper continental crust and the comparison data through nordmarkite and ekerite, but with concentrations spreading over an order of magnitude. Cs is anomalously low in the intrusive rocks of the Oslo Rift, with absolute concentrations in the range of 0.3 to 4 ppm, which is within the lower range of concentrations in the comparison data.
Distribution for lithophile trace elements in Oslo Rift felsic intrusive rocks, compared to lines representing 25 (lower quartile), 50 (median) and 75 (upper quartile) percentiles of data from monzonitic, quartz monzonitic, syenitic and granitic rocks worldwide (Appendix 3). All data have been normalized to the average upper continental crust values of Rudick and Gao (Reference Rudnick and Gao2014). Data for the Oslo Rift rocks are shown as summary boxplots showing median values as horizontal lines, interquartile distances as shaded boxes and the 5 to 95 percentile ranges as whiskers, with outliers beyond this shown as separate points. (a) Monzonitic rocks, comprising larvikite from outside of the LPC and akerite from the NHC. (b) Nordmarkite, with or without quartz. (c) Ekerite (mainly from the ESC, in blue) and SLG dykes (in red).
5. Simulation of fractional crystallization trends
In rocks where crystal accumulation has taken place, the purpose of modelling of fractional crystallization trends is restricted to testing whether or not a given parental magma can develop into residual liquid compositions that may form cumulate–liquid mixtures matching the observed rock compositions. The purpose of Rhyolite-Melts modelling carried out in this study is therefore to evaluate (1) whether or not fractional crystallization of a parental magma similar to that of the LPC can generate liquids relevant for the quartz-normative monzonites and peralkaline syenites reviewed here (Figs. 5 and 6); (2) the effects of crustal contamination at depth for the compositions of derived liquids; and (3) whether or not the internal compositional variation observed in peralkaline granite (ekerite) reflects liquid evolution, and how it can be related to the less silicic magmas in the rift.
5.a. The approach
Liquid lines of descent at constant pressures have been modelled using the Linux GUI version of the Rhyolite-MELTS 1.10 software (Gualda et al. Reference Gualda, Ghiroso, Lemons and Carley2012; https://melts.ofm-research.org/unix.html), running under the Windows Subsystem for Linux on an MS Windows 11 PC. Results of simulations are visualized and compared to observed variations in rock composition in the R 1 R 2 diagram as in Rämö et al. (Reference Rämö, Andersen and Whitehouse2022) and in plots of agpaitic index vs. SiO2 content, after recalculation to 100 % water-free compositions (Figs. 8 and 9).
(a) Model liquid lines of descent based on Rhyolite-MELTS (Gualda et al., Reference Gualda, Ghiroso, Lemons and Carley2012) simulations for starting melt compositions given in Table 2, at 5, 4.5, 3 and 2 kbar. Data from the individual modelling runs are given in Appendix 4. The 2 and 3 kbar lines terminate at 1100 oC, the 4.5 and 5 kbar curves at 1000 oC. Outlines of fields of variation of quartz-normative larvikite, akerite, syenite, nordmarkite with and without quartz and ekerite are from Figure 5, and compositions of SLG dyke rocks are shown as triangles. The effect of deep crustal contamination of the FCIM starting composition with 5 and 10 weight percent of primitive (T, representing TIFP in Table 2) and evolved (C, representing UC in Table 2) continental crust is highlighted in the inset. Curves marked LFCIM, 86062, TH16A and WCB91–6 are liquid lines of descent modelled for a second stage of fractionation at 3 kbar and 2 weight % water, starting from compositions given in Table 2. Black crosses in the main part of the figure are phonolitic to trachytic lavas from Mauritius (Ashwal et al. Reference Ashwal, Torsvik, Horváth, Harris, Webb, Werner and Corfu2016), shown for comparison (see section 6.c in the text). (b) Similar liquid lines of descent for the first fractionation stage calculated at different combinations of oxygen fugacity and initial water content. Curves at QFM are as in a. Variations in liquid lines of descent at f O2 between QFM-1 and QFM + 1 are shown by shading (at 5 kbar) and hachuring (at 4.5 kbar), respectively.
(a) Agpaitic index of modelled liquid lines of descent as a function of water-free SiO2. Fields of variation of plutonic rocks from Figure 6, SLG dykes, are shown separately as triangles. Stars: Composition of alkali feldspar and of alkali feldspar + quartz in cotectic proportions of a water-saturated haplogranitic system at 2 kbar (Johannes and Holtz, Reference Johannes and Holtz1996). Codes on model curves as in Figure 8. (b) Liquid lines of descent at variable f O2 relative to the QFM buffer, at initial water contents of 0.1 and 0.5 wt%.
Neumann et al. (Reference Neumann, Dunworth, Sundvoll and Tollefsrud2002) estimated the composition of a possible parental magma for subalkaline basalts in the Vestfold Graben by compensating the least fractionated of a series of aphyric basalts analysed by Neumann et al. (Reference Neumann, Sundvoll and Øverli1990b) for 30% of fractionation of olivine. The resulting composition (their FCI) is picritic by the definition of Le Maitre (Reference Le Maitre2002). Rämö et al. (Reference Rämö, Andersen and Whitehouse2022) modified this model parental magma composition by reducing TiO2 and K2O to better match the composition of ‘low-Ti’ basalts in the Oslo Rift and the monzonitic rocks of the LPC. This modified parental magma estimate (FCIM in Table 2) was also used by Buelens et al. (Reference Buelens, Debaille, Decrée, Coint and Mansu2024) and is the model parental magma composition for the simulation of first-stage fractional crystallization in the present study.
Starting compositions for Rhyolite-MELTS simulations
1LPC parent magma (Rämö et al., Reference Rämö, Andersen and Whitehouse2022).
2Average upper crust (Rudnick & Gao, Reference Rudnick and Gao2014).
3Orogenic crust, Talkeetna Arc, Alaska (Kelemen et al., Reference Kelemen, Hanghøj and Greene2003).
4Microsyenite, SMC (Andersen & Sørensen, Reference Andersen and Sørensen2003).
5Quartz syenitic aplite, ESC (Neumann et al., Reference Neumann, Andersen and Hansteen1990a).
6Trachytic SLG dyke («Sølvsbergite» from type locality, Brøgger, Reference Brøgger1933, p. 91).
Estimates of f O2 in intermediate to felsic intrusive rocks in the Oslo rift indicate values of QFM-1 to QFM for monzonitic rocks (larvikite and lardalite) and up to QFM + 1 for syenitic rocks (Neumann, Reference Neumann1976; Andersen, Reference Andersen1984a). Fayalite-bearing rocks are scarce in the Oslo Rift, having only been reported from a few localities the LPC (Kvamsdal, Reference Kvamsdal2023 and references therein), whereas quartz-magnetite assemblages are of widespread occurrence in the silica-oversaturated rocks of the rift. The QFM buffer thus gives a reasonable minimum limit for the evolution of syenitic to peralkaline granitic melts, and is the first choice used in the present simulations of liquid lines of descent. Supplementary simulations have been made at QFM-1 and QFM + 1, and for second-stage liquids also at higher f O2 levels (up to the HM buffer level).
Clinopyroxene is the most abundant ferromagnesian silicate mineral in the monzonitic rocks, with less abundant amphibole (magnesio-hastingsite to richterite) and biotite, indicating a relatively low water content in the initial magma (Rämö et al. Reference Rämö, Andersen and Whitehouse2022). In the peralkaline syenite (nordmarkite) and granite (ekerite), sodic-calcic amphibole occurs with Na-rich clinopyroxene. Quartz in nordmarkite and ekerite contains aqueous fluid inclusions; these rocks also have abundant miarolitic cavities suggesting increasing water content with fractionation and eventually separation of a free aqueous fluid phase (Neumann, Reference Neumann1976; Rasmussen et al., 1980; Andersen et al., Reference Andersen, Rankin and Hansteen1990; Hansteen & Burke, Reference Hansteen and Burke1990, Reference Hansteen and Burke1994). For consistency with the LPC model of Rämö et al. (Reference Rämö, Andersen and Whitehouse2022), the initial water content of the parental magma is set to 0.1 wt%. Supplementary simulations were made with initial water up to 0.5 wt%. The starting temperature of Rhyolite-Melts runs was well above the liquidus, and calculations terminated at 1100 °C (for quartz monzonite/akerite) and 1000 °C for nordmarkite.
Different starting compositions were used to simulate the internal evolution of nordmarkite and ekerite: (1) model residual melts derived from the FCIM composition by fractional crystallization at 4.5 to 5.0 kbar to 1000 °C, (2) analysed fine-grained syenitic and granitic rocks that are likely to approach melt compositions in the SMC and ESC complexes (Neumann et al. Reference Neumann, Andersen and Hansteen1990a; Andersen & Sørensen, Reference Andersen and Sørensen2003), and (3) a trachytic SLG dyke composition reported by Brøgger (Reference Brøgger1894, Reference Brøgger1933). These simulations were made at pressures of 2 and 3 kbar with water contents of 1 to 3 wt% and oxygen fugacity at QFM and higher values, the model runs terminating at 750–700 °C. Summaries of results from individual simulation runs are given in Appendix 4 in the online supplementary data.
5.b. Mid-crustal fractionation trends
Independently of pressure, the first mineral to crystallize from the FCIM parental magma is olivine with Fo91, at temperatures of 1490 °C at 4.5 to 5 kbar and 1470 °C at 2 kbar. At pressures higher than ca. 3.2 kbar, olivine is succeeded as a fractionating mineral by orthopyroxene at ca. 1340 °C at 5.0–4.5 kbar (Fig. 8b). Under fractional crystallization conditions, there will be no reaction between residual melt and early-crystallized olivine, and orthopyroxene crystallization forces melt compositions towards increasing silica undersaturation, that is, away from the critical saturation line in the R 1 R 2 plot (Fig. 8a). Clinopyroxene crystallization follows orthopyroxene at 1250–1260 °C and is accompanied by iron-titanium oxides (titanomagnetite and minor ilmenite) at 1170–1180 °C, and eventually by plagioclase at 1130 °C. Fractionation of Fe-Ti oxides deflects the liquid lines of descent towards the critical silica saturation line in the R 1 R 2 diagram, but peralkaline, syenitic residual liquids close to the line itself will only be formed by fractionation at pressures of 4.5 to 5.0 kbar (Figs. 8a, 9a). At pressures above 5 kbar, liquids at 1100–1000 °C will be more strongly silica-undersaturated, nepheline normative monzonite as shown by Rämö et al. (Reference Rämö, Andersen and Whitehouse2022). Residual melts at 4.5 to 5.0 kbar will reach peralkalinity at 1060–1070 °C, but at ca. 1000 °C, they will have A.I.≈ 1.3, which is higher than observed in nordmarkite (Fig. 9a). Residual liquids become quartz-normative at 1110 °C at 4.5 kbar and 1060 °C at 5 kbar, but none of the modelled liquid lines of descent starting from the FCIM composition will intersect the ekerite field in either the R 1 R 2 or the A.I. vs SiO2 diagram (Figs. 8a, 9a).
At pressures lower than ca. 3.2 kbar, orthopyroxene does not fractionate, and early development towards stronger silica undersaturation is suppressed (Fig. 8a). The critical SiO2 saturation plane is crossed at a temperature of ca. 1140 °C, with residual melts having lower A.I. than at higher pressure (Fig. 9a). The liquid lines of descent will intersect the field of metaluminous, quartz-normative larvikite at ca. 1100 °C, which may be a reasonable emplacement temperature for such monzonitic magmas (Rämö et al. Reference Rämö, Andersen and Whitehouse2022).
5.b.1. Effects of variations in oxygen fugacity and initial water content
Increases in f O2 and initial water content each cause residual liquids to become less silica-undersaturated. The effect of oxygen fugacity is relatively minor; at f O2 = QFM + 1, residual liquids at 5 kbar remain marginally silica-undersaturated, whereas at 4.5 kbar they become mildly silica-oversaturated, intersecting the field of quartz-normative larvikite (Fig. 8b). Increases in oxygen fugacity compatible with measurements from the syenitic plutons will, however, not cause liquid lines of descent to intersect the silica-oversaturated part of the nordmarkite field or the ekerite field. At f O2 = QFM-1, residual liquids will shift towards increasing silica undersaturation, but at 4.5 to 5 kbar, they will remain within the compositional range of nordmarkite (Fig. 8b).
The effect of increasing initial water content on silica saturation is stronger than that of oxygen fugacity: at 0.5 wt% initial water, liquid lines of descent cross the silica saturation boundary at 1120–1130 °C at 5–4.5 kbar and QFM oxygen fugacity (Fig. 8b). Residual liquids formed under these conditions will, however, be significantly less alkaline than at lower water content (Fig. 9b). They will remain metaluminous at low temperature and will intersect the fields of quartz-normative larvikite and akerite in the agpaitic index. vs. SiO2 diagram, but not the nordmarkite or ekerite fields (Fig. 9a, b). An increase of initial water content from 0.1 to 0.5 wt% will largely compensate for a reduction of f O2 to QFM-1.
5.b.2. Effects of crustal contamination at depth
Crustal contamination at depth, before fractional crystallization of the parental magma (Reference Neumann, Tilton and TuenNeumann et al. 1988b), will shift the starting point from FCIM towards less silica-undersaturated compositions. There is no direct evidence from crustal xenoliths indicating the major element composition of potential lower crustal contaminants in the Oslo Rift plutons. Data on the Precambrian rocks, including Mesoproterozoic anatectic granites on either side of the rift, suggest the absence of a ‘depleted lower crust’ in this region (Knudsen & Andersen, Reference Knudsen and Andersen1999; Andersen & Knudsen, Reference Andersen and Knudsen2000; Andersen et al. Reference Andersen, Andresen and Sylvester2001, Reference Andersen, Griffin and Sylvester2007). An average upper continental crust major element composition from Rudnick & Gao (Reference Rudnick and Gao2014) and a primitive orogenic crust composition based on an average of felsic to intermediate rocks in the Talkeetna Arc, Alaska (Reference Kelemen, Hanghøj and GreeneKelemen et al. 2003), are used as proxies of crustal contaminants, added in amounts up to 10 wt% to the FCIM mafic component before onset of fractional crystallization (Table 1). Contaminated parental magmas will remain olivine-hypersthene normative up to at least 10 weight percent contamination (Fig. 8a).
The main effects of contamination at depth are to lower the liquidus temperature and increase the temperature interval and amount of orthopyroxene crystallization (Fig. 8c). This will cause an early shift of the liquid lines of descent towards increasing silica undersaturation, which will be further enhanced once plagioclase starts fractionating. Residual liquids at 1000 °C are therefore shifted towards lower R 1 values compared to the uncontaminated systems (Fig. 8a) and show increasing silica undersaturation with increasing early contamination. The potential spread in melt composition at 4.5 to 5 kbar and 1000 °C, with zero to 10 wt% contamination, corresponds approximately to the total compositional range seen in quartz-free nordmarkite (Fig. 8a, inset). Residual melts in contaminated systems are less peralkaline than their uncontaminated counterparts but still have higher A.I. than nordmarkite and ekerite (Fig. 9a).
5.b.3. Composition of mid-crustal cumulate
Fractionation of an FCIM-type liquid at 4.5 to 5 kbar leaves behind a bulk cumulate with 70–85 % of mafic minerals (olivine, orthopyroxene, clinopyroxene, spinel and ilmenite) and plagioclase (Fig. 10a). This is a gabbroic rather than ultramafic cumulate. The proportion of orthopyroxene will decrease with decreasing pressure of fractionation, to disappear at ca. 3.2 kbar. Contamination with crustal material at depth restricts early crystallization of olivine, leaving an ultramafic cumulate consisting of orthopyroxene, clinopyroxene, Fe-Ti oxides and ilmenite with only minor olivine and less than 10 wt % plagioclase (Fig. 10b).
(a) Remaining liquid fraction of the FCIM composition as a function of temperature at 5, 4.5, 3 and 2 kbar, with fractionating minerals. FTO: Fe-Ti oxides (titaniferous magnetite and ilmenite). (b) Remaining liquid fraction as a function of temperature with 10% of the crustal components added.
5.c. Shallow fractionation trends and peralkaline, granitic residual liquids
Batches of residual liquids after mid-crustal fractionation will undergo further fractionation during ascent, possible residence in shallow-crustal staging chambers, and after final emplacement. Here, such secondary liquid evolution is modelled in terms of fractional crystallization of melt compositions given in Table 1 at 2–3 kbar pressure.
Residual liquids from fractionation to 1000 °C of the FCIM parental at 4.5-5.0 kbar (LFCIM in Table 2) will remain within the range of nordmarkite in the R 1 R 2 plot during secondary fractionation at lower pressure (Fig. 8a), but the residual melts will have A.I. > 1.3, well outside of the range of peralkalinity observed in nordmarkite. However, accumulation of alkali feldspar will reduce the A.I. of the total system.
Liquids represented by whole-rock compositions of fine-grained quartz-syenitic facies of the SMC and ESC complexes (microsyenite 86062 and quartz syenite aplite TH16A in Table 2) with 2 wt% initial water will at 3 kbar develop towards residuals approximating part of the ekerite trend in the R 1 R 2 diagram (Fig. 8a), but the peralkalinity will still exceed that observed in ekerite, except for the Gjerdingselva elpidite-bearing ekerite (Fig. 9a). Increasing f O2 will to some extent compensate for the excess alkalis but still cause only marginal overlap (Fig. 9a). Melts like these would eventually develop into the quartz-feldspar cotectic.
A melt corresponding to the trachytic SLG dyke composition will, at 3 kbar, fractionate towards peralkaline rhyolitic (pantelleritic) compositions similar to the grorudite members of the SLG suite both in terms of R 1 R 2 , SiO2 and peralkalinity (Figs. 8a, 9a), but liquid compositions are significantly more peralkaline than whole-rock compositions of ekerite.
5.c.1. Composition of shallow cumulates
Cumulates formed from an FCIM-derived liquid during secondary, low-pressure fractionation consist of 80 to 90 weight percent alkali feldspar, with minor amounts of clinopyroxene and Fe-Ti oxides. The bulk cumulate deposited at 3 kbar by a derivative liquid formed from an FCIM parental liquid at 5 kbar will have a composition quite close to a quartz-free nordmarkite (Fig. 8a). For systems reaching the alkali feldspar-quartz cotectic, solid assemblages will develop along a line with A.I. = 1.0 in the SiO2-A.I. diagram from the composition of alkali feldspar to that of quartz and alkali feldspar in eutectic proportions (Fig. 9a).
6. Discussion
6.a. Melt and cumulate in Oslo Rift felsic plutons
As shown by Neumann (Reference Neumann1980) and Rämö et al. (Reference Rämö, Andersen and Whitehouse2022), the main petrological features of intermediate rocks of the LPC, including its range of silica-saturated to undersaturated compositions, can be accounted for by deep- to mid-crustal fractional crystallization of a common parental magma, on the assumption that the intrusive rocks approximate liquids in composition. Previous studies have shown that this type of process cannot be directly extrapolated to the transition between monzonitic rocks (larvikite and related rocks) on the one hand and peralkaline syenite and granite (nordmarkite and ekerite) on the other (Dietrich et al. Reference Dietrich, Heier and Taylor1965; Neumann, Reference Neumann1976; Andersen, Reference Andersen1984a; Rasmussen et al. Reference Rasmussen, Neumann, Andersen, Sundvoll, Fjerdingstad and Stabel1988; Neumann et al. Reference Neumann, Andersen and Hansteen1990a). The energy-constrained modelling carried out in this study indicates that liquid lines of descent directly from a monzonitic (larvikite) precursor to peralkaline syenite (nordmarkite) or from nordmarkite to peralkaline granite (ekerite) are unrealistic, but that the three main types of residual liquid composition may still originate from a common parental magma by sequences of fractional crystallization and accumulation processes acting in succession at different levels in the crust. Crustal contamination has a significant effect on radiogenic isotope systematics (Neumann et al. Reference Neumann, Tilton and Tuen1988b; Sundvoll & Larsen, Reference Sundvoll and Larsen1990), but its consequences for major element evolution trends are less important than suggested in some earlier studies (e.g. Rasmussen et al. Reference Rasmussen, Neumann, Andersen, Sundvoll, Fjerdingstad and Stabel1988).
One feature that has been problematic in terms of liquid evolution controlled by fractional crystallization is the restricted variation in the agpaitic index of syenitic and granitic rocks (quartz-normative nordmarkite and ekerite) and the lack of a positive correlation between A.I. and SiO2 in these rocks, which is difficult to reconcile with feldspar fractionation from magmas that were originally mildly peralkaline. This pattern has been explained by alkali loss to late- to post-magmatic fluids (Dietrich et al. Reference Dietrich, Heier and Taylor1965; Dietrich & Heier, Reference Dietrich and Heier1967; Neumann et al. Reference Neumann, Andersen and Hansteen1990a). Escape of alkali-chloride and sulphate brines has indeed taken place after solidification of the ESC (Hansteen & Burke, Reference Hansteen and Burke1990, Reference Hansteen and Burke1994), but it is unlikely that loss of relatively small volumes of such solutions can account for an alkali deficiency amounting to several weight percent of the total rock. The textural evidence reviewed in this study suggests that accumulation of alkali feldspar or ternary anorthoclase in the felsic plutons is more important than previously realized. Accumulation of feldspar and quartz after final emplacement, or entrainment of alkali feldspar ± quartz in an ascending batch of highly peralkaline liquid represented by one of the secondary fractionation trends in Figs. 8 and 9, is a more efficient mechanism preventing whole-rock compositions from evolving towards elevated peralkalinity than loss of aqueous fluid.
Early crystallized alkali feldspar, and in some examples also quartz, are present as phenocrysts in porphyritic rock types (Fig. 2a, b and Fig. 3a). The example of a porphyritic nordmarkite shown in Fig. 2b has a content of alkali feldspar phenocrysts of close to 40 volume percent. A melt carrying a crystal load up to 55 volume percent will be physically mobile as a slurry (Vigneresse et al. Reference Vigneresse, Barbey and Cuney1996) that can rise buoyantly as diapirs (Copley et al. Reference Copley, Weller and Bain2023) or migrate along fractures (Bons et al. Reference Bons, Dougherty-Pager and Eburg2001). In non-porphyritic varieties, alkali feldspar forms networks of touching crystals, that is, a cumulate texture in the sense of Irvine (Reference Irvine1982). Systems with more than 55 % solids form rigid mushes that can only move as solid, buoyancy-driven diapirs (Copley et al. Reference Copley, Weller and Bain2023), but the interstitial liquid would still be able to migrate through connecting spaces between crystals (Vigneresse et al. Reference Vigneresse, Barbey and Cuney1996), which would be the case for the examples shown in Fig. 2e, f and g. On further solidification (Fig. 2d), interstitial spaces in the cumulate structure become discontinuous at ca. 71 volume percent of crystals (Vigneresse et al. Reference Vigneresse, Barbey and Cuney1996), after which interstitial liquids can no longer move, except through fractures. Highly peralkaline interstitial mineral assemblages with sodic pyroxene, amphibole and astrophyllite (Fig. 3d) must have formed from a sample of highly evolved, peralkaline silicate liquid trapped in interstitial spaces, with A.I. far higher than that of the solid alkali feldspar + quartz matrix of the rock. Only the Gjerdingselva ekerite intrusion provides an example where such residual liquids have been retained on an intrusion-wide scale. The SLG dykes are aphyric or only sparsely porphyritic and therefore more likely to represent liquid compositions than any of the plutonic rocks and are thus better representatives of highly evolved magmatic liquids than any of the coarse-grained plutonic rocks.
6.a.1. Comparison to the haplogranitic, experimental system
Neumann et al. (Reference Neumann, Andersen and Hansteen1990a) pointed out that quartz syenite and alkali granite (ekerite) samples from the ESC plot along the thermal valley on the liquidus surface of the haplogranitic Ab-Qz-Or experimental system. This, in fact, applies to all of the quartz-normative nordmarkite and ekerite samples considered here, and to the SLG dykes, which are on average slightly more potassic than the plutonic rocks (Fig. 11). An array of this kind will represent a liquid line of descent of a felsic magma only if full equilibrium between residual liquid and precipitated solids is preserved throughout. In a fractional crystallization scenario, points on the line represent random mixtures between a residual liquid composition and that of the total entrained or accumulated crystals in the sample. The modelled liquid lines of descent for the second-stage fractionation are slightly displaced towards higher normative Or compared to the plutonic whole-rock trend, intersecting it only at the high normative Qz end, corresponding to crystallization of a cotectic alkali feldspar + quartz assemblage. On the other hand, the modelled trends fall closer to the trachytic to rhyolitic SLG dykes, the most evolved dyke compositions overlapping with the high normative Qz termination of the modelled lines.
Rock compositions plotted in a normative Ab-Or-Qz diagram (weight percent) with liquidus boundaries in haplogranitic systems at 1 and 10 kbar (Johannes and Holtz, Reference Johannes and Holtz1996). Dotted lines are liquid lines of descent modelled from secondary starting compositions in Table 1 at 3 kbar.
6.a.2. The Ba-SiO2 correlation
Neumann et al. (Reference Neumann, Andersen and Hansteen1990a) reported a negative correlation in ESC ekerite between SiO2 and a group of elements that are compatible in alkali feldspar, sodic amphibole, titanite, Fe-Ti oxides and apatite, including Ba, Ca, Na, Eu, Al, Ti and P (their ‘Ba-group’). The correlation between Ba and SiO2 was attributed to removal of alkali feldspar by late-stage fractional crystallization. However, the observed variation of Ba and other alkali feldspar-compatible elements could equally well be a result of variable degrees of alkali feldspar accumulation in already fractionated liquids. Ba in the nordmarkite shows a considerable range of variation, from 86 to 1467 ppm at SiO2 = 65 ± 3 wt% (Fig. 12; see also Fig. 7), without any significant negative correlation. Some of the ekerite samples that have been analysed for Ba have concentrations above 300 ppm, whereas another group of samples has low Ba (<150 ppm), as have the few peralkaline rhyolitic dykes that have been analysed. These low-Ba rocks are likely to reflect a liquid line of descent, whereas those with higher Ba contain accumulations of alkali feldspar. The monzonitic rocks form two distinct clusters at lower SiO2 at Ba > 1500 ppm (akerite) and 600–1200 ppm (larvikite). From the modelling of the major element evolution done in this study, neither of these can be a direct precursor of the more silicic rocks.
The relationship between Ba (as an example of the ‘Ba-group’ of feldspar-compatible elements of Neumann et al. 1990) and SiO2. The overall negative correlation of Ba and other alkali feldspar-compatible elements with silica observed in the ESC ekerite by Neumann et al. (Reference Neumann, Andersen and Hansteen1990a) is likely to be a result of a combination of a low-Ba trend of liquid compositions and variable amounts of accumulated alkali feldspar enriched in Ba and other elements with high partition coefficients for alkali feldspar.
6.b. A petrogenetic model for the felsic intrusive rocks
Combining petrographic observations and Rhyolite-MELTS modelling results allows a revised petrogenetic scenario for the peralkaline syenitic to granitic rocks in the Oslo Rift to be envisaged (Fig. 13). One observation is that the entire spectrum of intermediate and felsic intrusive rocks, from larvikite and nepheline syenite through nordmarkite and ekerite to highly peralkaline felsic dyke rocks, can be accounted for by successive polybaric fractionation and alkali feldspar accumulation processes, starting with a primary magma similar to that assumed for the LPC by Rämö et al. (Reference Rämö, Andersen and Whitehouse2022).
Cartoon illustrating the evolution of peralkaline syenite and granite magma in the Oslo Rift. See section 6.b in the text for an explanation.
Although crustal contamination has certainly taken place and is locally important (Rasmussen et al. Reference Rasmussen, Neumann, Andersen, Sundvoll, Fjerdingstad and Stabel1988), its effect on major element bulk composition is surprisingly small compared to that of fractional crystallization and feldspar accumulation. It should, however, be noted that biotite granite and biotite syenite intrusions known from radiogenic isotope data to be strongly influenced by local crustal contamination (Rasmussen et al. Reference Rasmussen, Neumann, Andersen, Sundvoll, Fjerdingstad and Stabel1988; Trønnes & Brandon, Reference Trønnes and Brandon1992; Haug, Reference Haug2007) are not considered in this model.
Quartz-bearing monzonitic and quartz-monzonitic rocks outside of the LPC (i.e. larvikite in the SCP and larvikite and akerite in the NHC) can be accounted for by fractional crystallization from an FCIM-type parental magma at 3 kbar or less, with evolution towards the more silicic quartz monzonite (akerite) compositions being facilitated by a higher water content in the initial melt. Local wall rock assimilation processes, as suggested by Sæther (Reference Sæther1962), are not necessary to account for the major element composition of akerite at the southern margin of the NHC.
Evolved liquids able to produce a quartz-free nordmarkite cumulate at low pressure can develop by fractionation of an FCIM precursor at ca. 5 kbar pressure. Quartz-free porphyritic syenite in the SMC and silica-undersaturated SLG dykes could represent samples of this type of liquid evolving from FCIM after ca. 80 % fractionation of the parental liquid.
FCIM-type magma evolving at slightly lower pressure (e.g. 4.5 kbar) would form more silicic, peralkaline residual liquids that would eventually precipitate quartz. Such liquids formed fine-grained, quartz-bearing microsyenitic intrusions and also trachytic SLG dykes. Slurries with this type of liquid component formed nordmarkite with interstitial quartz when emplaced into the shallow crust.
Final evolution at low pressure of melts derived from FCIM-type precursors would reach quartz saturation to give rise to ekerite, again emplaced as slurries with random mixtures of alkali feldspar and quartz in a melt saturated in both, and to pantelleritic SLG dykes (grorudite) formed from liquids without cumulate load. The ekerite, for example, in the ESC, essentially represents a cumulate-dominated rock system from which much of the final liquid was lost, except where it was retained as granophyric interstitial aggregates or formed highly alkaline interstitial mineral assemblages with astrophyllite and other high-alkali minerals. Aqueous fluid or residual melt lost in this process has caused alkali metasomatic alteration in the cap rocks of plutons (Bonin & Sørensen, Reference Bonin and Sørensen2003), which are now nearly everywhere removed by erosion. The Gjerdingselva ekerite intrusion is an anomalous case, where the late magmatic melt or fluid was retained on an intrusion-wide scale.
The SLG dykes have largely been neglected since the time of W.C. Brøgger. This is unfortunate, as these dykes appear to reflect the evolution of felsic, peralkaline magmatic liquids more closely than any of the cumulate-influenced plutonic rocks in the Oslo Rift. These dykes should be a target for renewed research.
6.c. Implications beyond the Oslo Rift
The felsic peralkaline rocks of the Oslo Rift may be fairly representative of a class of metaluminous to moderately peralkaline ‘syenitic’ rocks found in many other places on Earth, in intraplate, rift-related, oceanic and even subduction-related settings (Figs. 5, 6, and 7). Such rocks tend to be enriched in alkali feldspar but are commonly relatively depleted in trace elements that are normally regarded as incompatible in the feldspar lattice – Cs being a prime example. Although the role of accumulation processes in the formation of syenitic rocks has been recognized in principle (e.g. Wolff, Reference Wolff2017), it may generally have been overlooked or underestimated, with studies by Shellnutt (Reference Shellnutt2021) and Leandro et al. (Reference Leandro, Stevens, Moyen, Kisters and Ferreira2025) as important exceptions. Furthermore, the peralkaline, near-silica saturated syenites of the Oslo Rift do not require the formation of trachytic melts directly from the partial melting of metasomatized mantle peridotite, as was suggested for similar magma compositions in Mauritius by Ashwal et al. (Reference Ashwal, Torsvik, Horváth, Harris, Webb, Werner and Corfu2016), plotted for comparison in Fig. 8a. Whereas experimental studies and work on mantle xenoliths leave no doubt that melts of intermediate to silicic composition may form in the lithospheric mantle and that they are involved in mantle metasomatism, the present findings suggest that it may be unnecessary to invoke such primary melt compositions for trachytic and phonolitic magmas reaching the surface. The Mauritius trachytes and phonolites have molecular MgO/(MgO + Fe) ≤ 0.16 (Ashwal et al. Reference Ashwal, Torsvik, Horváth, Harris, Webb, Werner and Corfu2016), which is very low for both primary mantle melts and glass compositions reported from experimental studies (e.g. Laporte et al. Reference Laporte, Lambart, Schiano and Ottolini2014). The similarity in composition of Mauritius trachyte and Oslo Rift nordmarkite (Fig. 8a), and the association with olivine-hypersthene normative, mildly alkaline basalts in both provinces suggest that a similar, polybaric fractionation scenario should also be investigated for Mauritius. A Moho depth of 10–15 km at Mauritius (Singh et al. Reference Singh, Kaviani and Rümpker2016) suggests that fractionation processes could take place over a range from 3–4 kbar to near-surface pressures, which may be sufficient for basaltic parental magmas to evolve towards trachytic/syenitic compositions straddling silica saturation.
7. Conclusions
Fractional crystallization of a magma similar to the feasible parental magma of the LPC monzonitic rocks in the Oslo Rift (larvikite and lardalite) will yield metaluminous, silica-saturated residual melts corresponding to quartz-bearing larvikite outside of the LPC and to quartz monzonitic akerite in the NHC at ca. 1100 °C when fractionating at pressures of 2 to 3 kbar and possibly at higher initial water content than the LPC magmas.
The presence of accumulated alkali feldspar - and in the most siliceous compositions, also quartz - is the rule rather than an exception for the plutonic alkali feldspar syenites (nordmarkite) and granites (ekerite) in the Oslo Rift. Felsic, peralkaline residual liquids straddling the silica saturation plane can form from the same initial melt composition as larvikite by fractional crystallization to ca. 1000 °C, but only in a quite narrow pressure interval of 4.5 to 5 kbar. At 5 kbar, residual melts are slightly silica-undersaturated, at 4.5 kbar, marginally silica-oversaturated. These melt compositions are distinctly more peralkaline than any analysed plutonic syenitic rock in the Oslo Rift. Mid-crustal liquid lines of descent are sensitive to differences in oxygen fugacity and (in particular) initial water content, but changing f O2 within reasonable limits (QFM-1 to QFM + 1) does not change this pattern significantly. Increasing water content has a stronger influence on silica saturation but at the same time prevents residual liquids from approaching peralkalinity, contrary to what is observed in nordmarkite and ekerite.
Quartz-free and quartz-bearing, plutonic nordmarkite varieties do not represent samples of this mid-crustal residual melt but mixtures of residual liquid and solid cumulate formed during ascent or a second stage of fractionation at a shallower level in the crust (2–3 kbar). Emplacement to the final, shallow level was as slurry or mush consisting of alkali feldspar entrained in a residual melt. Adcumulus alkali feldspar growth continued after final emplacement.
Ekerite does not represent a simple derivative of plutonic nordmarkite, but liquids able to form alkali granitic alkali feldspar ± quartz cumulates may have developed from trachytic liquids represented by minor fine-grained quartz syenitic intrusions in the SMC and ESC, or from liquids akin to trachytic SLG dykes. Fractional crystallization of such liquids at P ≤ 3 kbar will yield residual liquids that are far more peralkaline than the ESG ekerite but which fall close to rhyolitic SLG dykes in composition. Similar to nordmarkite, plutonic ekerite itself consists of a mixture of entrained or accumulated alkali felspar ± quartz with minor amounts of trapped, highly peralkaline, silica-rich liquid. Post-magmatic loss of aqueous fluid may have caused alkali loss, but the main factor behind the relatively low A.I. of plutonic ekerite is buffering by alkali feldspar cumulate.
The SLG dykes are the most reliable representatives of progressively more silica-rich and peralkaline differentiated liquid compositions in the Oslo Rift and represent batches of relatively cumulus-free melt ejected from fractionating magma chambers below the present level of erosion, or possibly from the solidifying plutons themselves.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S001675682510040X.
Acknowledgements
The authors thank the Natural History Museum of the University of Oslo for unrestricted access to library, rock and thin section collection and the University of Johannesburg for support through a Distinguished Visiting Professorship to TA. Apart from this, the study has received no economic support. The authors also thank editor Sarah Sherlock and two anonymous reviewers for their helpful comments.
Open access
