ALC intrusion and peak metamorphic conditions

The temperatures recorded in ALC noritic plagioclase–amphibole and orthopyroxene–biotite pairs spanning ~800–1070°C probably reflect igneous crystallisation temperatures. Crystallisation of igneous cumulate/intercumulus amphibole in the 850–1030°C range has been inferred for modern arc cumulates (Arculus and Wills, Reference Arculus and Wills1980; Tollan et al., Reference Tollan, Bindeman and Blundy2012) and Caledonian intrusive rocks (Otten, Reference Otten1984). Had these amphiboles formed during a retrogression event they would have been expected to show lower temperatures similar to those recorded in host mafic granulite (~740–790°C) where green Ca-amphibole texturally overgrows pyroxenes. A prograde origin for the ALC noritic temperatures can be excluded, because clinopyroxene is expected to coexist with amphibole at such high temperatures (See ‘Age considerations and anatexis’, below; Green et al., Reference Green, White, Diener, Powell, Holland and Palin2016; Palin et al., Reference Palin, White, Green, Diener, Powell and Holland2016), which is not observed. Low Ti content in biotite is usually ascribed to re-equilibration of primary magmatic biotite (Bonin and Tatu, Reference Bonin and Tatu2016) and the variable TiO₂ in Amikoq noritic biotite also suggests re-equilibration. The lowest TiO₂ is observed in biotite from an intensely retrogressed norite (1.0–1.4 wt.%) with near-complete alteration of orthopyroxene to anthophyllite (Fig. 4h); whereas, two biotite grains from an unaltered norite (Figs 4g, S11) have high TiO₂ (~3.3 wt.%), even relative to biotite from the ilmenite-bearing sil–grt–crd gneiss (0.1–2.4 wt.%) suggesting that high-Ti norite biotites have retained at least some of their primary characteristics. The melanorite biotites have moderate TiO₂ (2.0–2.4 wt.%) suggesting re-equilibration; it is plausible that lower TiO₂ is a sign of lower magmatic TiO₂ content. Thus, it appears overall that the biotite has variably re-equilibrated, although this seems to have little effect on the calculated temperatures, in common with observations for noritic green Ca-amphibole (see section on Amphiboles, below).

Both the roof (mafic granulite, 193500) and floor (K-hastingsite gneiss, 191143) host rocks contain relict high-temperature assemblages, with a minimum peak assemblage of orthopyroxene + clinopyroxene + plagioclase + oxides (+ quartz for sample 191143?). Ca-amphibole is a later phase. Furthermore, clinopyroxene commonly contain exsolution lamellae of pigeonite with an estimated temperature of formation ≈ 975–1010°C, indicating that the host rocks experienced a high-temperature event, which probably resulted in their complete dehydration. Lower temperature recorded by orthopyroxene + clinopyroxene of 600–850°C (Lindsley, Reference Lindsley1983), and indeed all the ALC orthopyroxene, indicate that host rocks and the ALC shared a common re-equilibration at temperatures similar to the host mafic granulite green Ca-amphibole that overgrow pyroxenes (Fig. 6; see discussion of granulite- to amphibolite-facies conditions, below). However, it is difficult to pinpoint confidently the cause of the high-temperature event of the host rocks, as it could be associated with the intrusion of the ALC (intrusion contact aureole?), reflect the conditions at which the ALC intruded, or reflect peak temperature associated with partial melting of ALC (See ‘Age considerations and anatexis’, below). Garde (Reference Garde1997) suggested that some homogenous amphibolites had a subvolcanic intrusive origin. Such an origin is perhaps reflected by pigeonite exsolution in the host clinopyroxene.

Temperatures of >950°C should cause widespread and considerable partial melting throughout the MORB-like ‘amphibolitic’ host rocks during prograde metamorphism (~25–35% melt predicted for typical MORB-compositions at 950°C in the 5–10 kbar range; Palin et al., Reference Palin, White, Green, Diener, Powell and Holland2016). For a Lewisian amphibolite bulk composition, 12% melt extraction is expected to decrease in SiO₂ from ~50 to ~45 mol.%, K₂O from ~0.6 to ~0.01 mol.% and increase in TiO₂ from ~1.1 to 2.2 mol.% (Palin et al., Reference Palin, White, Green, Diener, Powell and Holland2016). Whereas bulk-rock K₂O shows no relation to SiO₂ in Amikoq host mafic supracrustals, SiO₂ (~55 to 47 wt.%) and TiO₂ (~0.4 to 1.2 wt.%) do show a crude correlation (unpublished data). K₂O could have been re-introduced by fluids during retrogression; whereas, the SiO₂–TiO₂ relation suggests that some rocks might have experienced at least 10% melt extraction (ignoring any pre-metamorphism geochemical heterogeneity).

Do ALC lithologies contain relict igneous minerals? Data presented in previous sections has important implications for identifying any putative primary igneous clinopyroxene in the ALC. Clinopyroxene is generally absent in ALC lithologies though Kristensen (Reference Kristensen2006) report a single analysis in an ultramafic sample (Fig. S15). This is unlikely to be due to a retrograde hydration replacement by amphibole, because: (1) the amphibole crystallisation temperatures are too high as discussed above; (2) remnants of clinopyroxene (e.g. relict cores) should have survived a metamorphic overprint similar to what is observed in the host rocks. These observations lead to the inference that clinopyroxene was an insignificant crystallising phase, if present at all.

The cumulate norites of the West Margin have major element compositions that are similar to the amphibolites except for higher Al₂O₃ and lower TiO₂ and P₂O₅ (Table 1). The norite differs from the supracrustal rocks in the absence of clinopyroxene and greater modal proportion of orthopyroxene (Figs 4b, 4g, S6, S11). Although garnet-bearing norites occur locally in high-strain zones (Fig. S2), garnet is not a significant constituent (Fig. 4g–h). Thus, it would appear that the norites have only recrystallised, rather than had their entire primary mineralogy overprinted by a secondary metamorphic mineralogy. As a comparison, Subramaniam (Reference Subramaniam1956) investigated the anorthosite–eclogite–gabbro complex of Sittampundi and observed significant metamorphic overprint in the form of Ca-amphibole, corundum, clinozoisite, garnet and anthophyllite in anorthositic rock;s and garnet and hornblende for gabbroic rocks, with the presence of common reaction microtextures deemed indicative of amphibolite- and eclogite-facies with some retrogression. Importantly, some of the metamorphosed Sittampundi rocks have major-element compositions comparable to Amikoq norites (Table 1), suggesting that the Amikoq norites have not been significantly overprinted by a secondary metamorphic mineralogy (barring retrogressed samples, see ‘Evidence for granulite-facies conditions and beginning retrogression’ below). An essentially primary mineralogy is supported by preservation of euhedral amphibole inclusions both in plagioclase and orthopyroxene.

Table 1.

Major-element compositions of Amikoq supracrustals compared to various lithologies from the literature.

* When no Fe₂O₃ or FeO is provided, FeO is given as FeO^tot or Fe₂O₃^tot. [1] Amikoq data: Aarestrup (unpublished data). [2] Storø amphibolite, SW Greenland: Yakymchuk and Szilas (Reference Yakymchuk and Szilas2018). [3] 1921 Kilauea olivine tholeiite: Helz (Reference Helz1973). [4] Plagioclase-eclogite (garnet–hornblende–plagioclase rock), India: Subramaniam (Reference Subramaniam1956).

Many metamorphosed ultramafic bodies have been inferred to have a de-serpentinized origin, i.e. primary minerals were completely altered at sub-greenschist-facies conditions and subsequently ‘de-serpentinized’ at amphibolite- to granulite-facies conditions (e.g. Evans, Reference Evans1977; Khedr and Arai, Reference Khedr and Arai2012; Nishio et al., Reference Nishio, Morishita, Szilas, Pearson, Tani, Tamura, Harigane and Guotana2019). Dehydration of serpentine in ultramafic rocks may generate high-Fo olivine with inclusions of magnetite (e.g. Nishio et al., Reference Nishio, Morishita, Szilas, Pearson, Tani, Tamura, Harigane and Guotana2019). Magnetite has been observed as inclusions in some ALC West Margin olivine, but where found, it is usually associated with chlorite, implying that magnetite is associated with chlorite formation and not breakdown of serpentine. Regardless, some magnetite is probably related to partial serpentinization. The Amikoq ultramafic rocks are unlikely to have been previously serpentinized, given their moderate Fo content (Fo_82–85 in West Margin ultramafic rocks), and the olivine is probably of cumulate origin, with a metamorphic overprint. This is similar to deductions for lower forsteritic olivine (Fo_79–84) in volumetrically minor ultramafic rocks hosted in a well-constrained lower crustal section in Italy (Schenk, Reference Schenk1984). This is consistent with petrographic observations with large relict crystals hosting abundant Al-spinel inclusions relative to the largely inclusion-free neoblasts (Fig. S7i–k).

Evidence for granulite-facies conditions and beginning of retrogression

The aluminous garnet–cordierite–sillimanite–biotite–ilmenite–plagioclase–quartz–spinel mineral assemblage observed in the marginal sil–grt–crd gneiss, either represents an altered metavolcanoclastic protolith, or a metasomatic product. Highly aluminous gneisses have been reported from Kangerdluarssuk in the Maniitsoq region (Yakymchuk and Szilas, Reference Yakymchuk and Szilas2018), as marginal rocks of the nearby Ulamertoq ultramafic complex (Guotana et al., Reference Guotana, Morishita, Yamaguchi, Nishio, Tamura, Tani, Harigane, Szilas and Pearson2018), and from the Neoarchean Storø Supracrustal Belt (Szilas et al., Reference Szilas, Maher and Bird2016b). At Amikoq, the sil–grt–crd gneiss at the West Margin is not associated with ultramafic lithologies and it is separated from the bottom noritic sequence of the ALC by a thin quartz-rich metasomatised ‘amphibolite’ layer (Fig. 2b; See Petrography section ‘Amphiboles’). The absence of alkali feldspar argues against a pelitic protolith and instead indicates a metavolcanoclastic protolith, which was metasomatised together with the adjacent mafic amphibolite, potentially along a shear zone.

Inclusions in garnet and cordierite, orientated perpendicularly to matrix minerals, could record a relict fabric and (minimum) mineral assemblage of sillimanite + biotite + quartz + Fe-sulfides + rutile (+ garnet?). However, the similarity between matrix- and inclusion mineralogy makes it dubious to relate the inclusions to a separate prograde assemblage as they could reflect a single protracted high-grade event (Vernon et al., Reference Vernon, White and Clarke2008). The absence of kyanite or kyanite pseudomorphs indicate a moderate pressure (‘non-Barrovian style’) metamorphic regime as suggested by Yakymchuk et al. (Reference Yakymchuk, Kirkland, Hollis, Kendrick, Gardiner and Szilas2020). The occurrence of cordierite in a vein found in an orthopyroxenite demonstrate that the ALC also experienced the peak metamorphic conditions recorded in the sil–grt–crd gneiss. The recorded peak metamorphic assemblage for the sil–grt–crd gneiss is surmised as coarse-grained cordierite–garnet_core–sillimanite–spinel–ilmenite_{euhedralInSeam} (+biotite?).

Experimental studies and observations of natural assemblages have shown that greater Mg contents in cordierite are equivalent to higher pressures, while cordierite is expected to increase in abundance in metapelitic rocks as temperature increases under isobaric conditions (Holdaway and Lee, Reference Holdaway and Lee1977). Cordierite is abundant (~40%; Figs 4a, S4) and Mg rich (Mg#: 0.73-0.76; Table S10), suggestive of high metamorphic grade. Conditions of ~7.5 kbar and ~820°C calculated from cordierite–garnet correspond to the calculated Mg#_Crd and garnet_Pyr of Aranovich and Podlesskii (Reference Aranovich and Podlesskii1983) (Fig. 10). They are also in agreement with other P–T estimates from Akia terrane metapelites, which range from ~8–10 kbar and ~800–850°C (Dyck et al., Reference Dyck, Reno and Kokfelt2015; Kirkland et al., Reference Kirkland, Yakymchuk, Hollis, Heide-Jørgensen and Danišík2018; Riciputi et al., Reference Riciputi, Valley and McGregor1990). Note that the barometer used here (Perchuk et al., Reference Perchuk, Aranovich, Podlesskii, LAvRANT'EvA, Gerasimov, Fed'Kin, Kitsul, Karsakov and Berdnikov1985) assumes a positive slope in P–T space, while using a calibration with a negative slope (e.g. Holdaway and Lee, Reference Holdaway and Lee1977) yields ~6.5 kbar and ~810°C (Fig. 10). However, the calculated temperature underestimates the actual peak temperature given orthopyroxene–clinopyroxene assemblages and pigeonite lamellae in host rocks as well as local anatectic conditions.

Yakymchuck and Szilas (Reference Yakymchuk and Szilas2018) investigated the phase relations of amphibolites from Storø and found that at 7 kbar and temperatures ranging 800–850°C the stable assemblage was liquid + hornblende + augite + orthopyroxene + plagioclase + ilmenite ± quartz. This is in good agreement with observations in the Amikoq amphibolites (Figs 4b, S6), although the Amikoq amphibolites deviate slightly in geochemistry, e.g. higher MgO (Table 1). At lower temperatures (<800°C), orthopyroxene exits in the stable assemblage (Yakymkchuk and Szilas, Reference Yakymchuk and Szilas2018), similar to the phase assemblage modelling results of a typical MORB-composition (Palin et al., Reference Palin, White, Green, Diener, Powell and Holland2016) and metabasite (Beard and Lofgren, Reference Beard and Lofgren1991; Green et al., Reference Green, White, Diener, Powell, Holland and Palin2016).

Garnet-in estimates of ~8.8 kbar (Yakymchuk and Szilas, Reference Yakymchuk and Szilas2018), places an upper limit on the pressure conditions of Amikoq. Some garnet-bearing amphibolites have been observed at Amikoq, but they are probably a result of inhomogeneous strain and varying bulk rock composition throughout the supracrustals where, e.g. higher Al₂O₃, MnO and/or FeO^tot content are expected to stabilise garnet at lower pressures (O'Neill, Reference O'Neill1981; Palin et al., Reference Palin, White, Green, Diener, Powell and Holland2016). At temperatures above ~820°C, quartz leaves the stable assemblage of the Storø amphibolites to produce orthopyroxene (Yakymchuk and Szilas, Reference Yakymchuk and Szilas2018). This transition has been captured in an Amikoq mafic granulite, where rounded ilmenite grains bounded by clinopyroxene and orthopyroxene contain thin films consisting of quartz + amphibole + clinopyroxene (Fig. S6a-i); whereas, quartz tends to be rare elsewhere in the thin section. This suggests a retrograde reaction where quartz re-joins the assemblage by consumption of orthopyroxene. In typical MORB-compositions quartz re-joins the assemblage at 850–890°C in the 7.0 to 8.9 kbar range; above 8.9 kbar, rutile joins the assemblage (Palin et al., Reference Palin, White, Green, Diener, Powell and Holland2016) a phase that is not observed. The clinopyroxene-hosted pigeonite exsolution lamellae (~975–1010°C) were probably part of the peak temperature assemblage (Figs 6, S6).

Beginning retrogression at granulite-facies conditions is also indicated by the Al^IV/Al^VI in chlorite of Amikoq olivine rocks (0.99–1.05). Chlorite Al^IV/Al^VI is expected to be close to unity in ultramafic rocks close to the breakdown temperatures, while the ratio increases at lower grades (Frost, Reference Frost1976). The chlorite = olivine + orthopyroxene + spinel + H₂O reaction has been predicted to occur in the 770–837°C range between 5–10 kbar (Evans, Reference Evans1977; Zen, Reference Zen1972), which is in good agreement with the P–T conditions inferred from the sil–grt–crd gneiss (Fig. 10). Cr-magnetite enveloped in chlorite have been observed to contain relict spinel cores demonstrating that Cr-magnetite is coeval with chlorite formation and resulted from Al-leaching of spinel into newly formed chlorite. This is consistent with expectations of spinel alteration in ultramafic rocks during chlorite formation (Barnes and Roeder, Reference Barnes and Roeder2001; Evans, Reference Evans1977; Evans and Frost, Reference Evans and Frost1976). Importantly, it illustrates that chlorite + Cr-magnetite is not prograde in origin, and formed through the introduction of fluids.

Granulite- to amphibolite-facies conditions: Retrograde mineral assemblages

In the host rocks green Ca-amphibole (secondary hornblende) overgrowing pyroxenes in the roof mafic granulite (Fig. S6) provide textural evidence that amphibole post-dates the peak P–T conditions. This is verified by plagioclase–hornblende temperatures of 737–790°C and the amphibole chemistry.

Retrograde conditions of the sil–grt–crd gneiss could involve several steps that initially might comprise garnet_rim–sillimanite–biotite–plagioclase–quartz–ilmenite followed by or coeval with the beginning of garnet chloritization. Some biotite appears to postdate the cordierite while straight biotite–garnet_rim boundaries occur, suggesting that biotite and garnet rims attained equilibrium. The biotite–garnet and GASP results of ~500–560°C and 1.3–3.0 kbar are at best debateable, considering the abundant and stable sillimanite; whereas, in Al₂SiO₅ projections 500–560°C, 1.3–3.0 kbar is in the andalusite stability field (Fig. 10; e.g. Essene, Reference Essene1989; Thompson, Reference Thompson1976). This probably indicates that pressure is underestimated, and possibly also temperature, with the best possible estimate being ~560°C and ~3 kbar. The apparently erroneous pressure estimate based on garnet–plagioclase could be due to plagioclase being formed in some melt reaction at peak metamorphic conditions or during retrogression, while the Fe–Mg exchange between biotite–garnet might also have equilibrated at a lower temperature than at the time of garnet and biotite formation. Retrograde re-equilibration is especially well illustrated with calculated ilmenite–magnetite temperatures of 380–540°C (Lepage, Reference Lepage2003; Spencer and Lindsley, Reference Spencer and Lindsley1981; Stormer, Reference Stormer1983) from marginal rocks and amphibolites. Whereas the temperatures are unlikely to be reliable, they nevertheless support retrogression of the ALC.

Throughout the ALC lithologies, retrograde metamorphism is recognised by: (1) chlorite formation at the expense of olivine, orthopyroxene, spinel (s.s. and hercynite) and rarely amphibole; (2) formation of (Fe,Mg,Mn)-amphibole + magnetite at the expense of orthopyroxene; (3) pale-green high-Al Ca-amphibole + chlorite ± quartz formation at the expense of plagioclase and (4) green Ca-amphibole (principally hornblende melagabbro), rarely plagioclase and orthopyroxene (principally norites). The colourless (Fe,Mg,Mn)-amphibole aggregates and pale-green high-Al Ca-amphibole (hornblende) define retrograde hydration metamorphism, or a secondary event relatable to the high-strain zones of the West Margin, which has caused extensive deformation of the noritic bodies. The high-strain zones locally affecting some norites could have acted as fluid pathways explaining the heterogeneous distribution of orthopyroxene alteration to (Fe,Mg,Mn)-amphibole in the inspected samples, i.e. some are largely unaffected (compare Figs S11, S12). The retrograde hydration event locally resulted in a stable mineral assemblage in retrogressed norites consisting of amphibole + plagioclase + magnetite + chlorite + quartz, indicative of amphibolite-facies conditions (Fettes and Desmons, Reference Fettes and Desmons2011). Epidotization or sericitization (alteration to muscovite) of plagioclase (Owens and Dymek, Reference Owens and Dymek1997) is not observed indicating that retrograde facies exceeded greenschist.

The retrograde event is also recorded in the green Ca-amphibole compositions of the retrogressed norites. Texturally, all the West Margin norites resemble metamorphic rocks with noteworthy similarities between the amphibolitic and norite stringer textures (compare Figs S6, S11), probably indicating that these were formed coevally. However, there is a marked difference between the edenite component (^ANa+^AK; Schumacher, Reference Schumacher2007), Ti and to a lesser extent, Al^IV, in green amphibole from the un-retrogressed norite and the retrogressed amphibole norites, where the former have higher content of all parameters (0.44–0.52, 0.16–0.20 and 1.75–1.87 apfu, respectively – excepting one amphibole with 0.39, 0.10 and 1.62 apfu) relative to the latter (0.18–0.41, 0.08–0.12 and 1.34–1.77 apfu, respectively). This suggests that the green amphiboles in the retrogressed norites have re-equilibrated at a lower temperature, judging from lower Al^IV and Ti, and possibly lost alkalis (see Discussion on Amphiboles; Fig. 12). Re-equilibration is corroborated by the ^B(Na+Ca) and ^TAl–^ANa–^AK in co-existing anthophyllite–Ca-amphibole pairs which, where extrapolated to the limbs of the two-amphibole miscibility gap (Schumacher, Reference Schumacher2007), co-exist at temperatures of approximately 500–650°C or somewhat higher and lower (Figs 11, S19). However, the (Fe,Mg,Mn)-amphibole might have formed at higher temperatures. The exsolved Ca-poor, Fe-rich amphibole lamellae (not analysed) observed in the South Margin hornblende melagabbro and in some norite amphibole was probably formed in the same temperature range. Plagioclase–hornblende thermometry results also yielded a slightly lower temperature for the retrogressed norites than the un-retrogressed norite. The temperatures are nevertheless high (890–958°C) and suggest that the thermometer is quite robust.

Fig. 11.

Temperature vs. ^B(Na+Ca) in amphibole with (Fe,Mg,Mn)-amphibole – Ca-amphibole solvus limbs, after Schumacher (Reference Schumacher2007). Bars labelled ‘Ath’ and ‘Ca-amp’ denote the anthophyllite and Ca-amphibole range, respectively. The compositional range is considerably smaller for ^B(Na+Ca) than ^TAl–^ANa–^AK (Fig. S18), which could indicate that the former is the more reliable parameter; especially for the norites, which form tight clusters indicating a temperature range of ~500–650°C. Symbols with black rims are cummingtonites, otherwise as in Fig. 11 – note that the plotted norites are retrogressed varieties.

Fig. 12.

Edenite component, Mg#, Al^IV and Ti in amphiboles. (a) Edenite component vs. Mg#. (b) Ti (apfu) vs. Al^IV (apfu). (c) Edenite-component vs. Al^IV (apfu). See text for details. ‘Alaskan Complexes’ amphiboles are from Himmelberg and Loney (Reference Himmelberg and Loney1995), ‘Arc xenolith’ amphiboles are from Arculus and Wills (Reference Arculus and Wills1980), DeBari and Coleman (Reference DeBari and Coleman1989) and Tollan et al. (Reference Tollan, Bindeman and Blundy2012), ‘Stillwater ultramafic’ amphiboles are from Page and Zientek (Reference Page and Zientek1987); average metamorphic amphibole composition from Schumacher (Reference Schumacher2007); experimental amphiboles are from Helz (Reference Helz1973, Reference Helz1979) – indicated 925°C Amp is likely to be an experimental outlier; small symbols are Amikoq amphiboles from Kristensen (Reference Kristensen2006) and Szilas et al. (Reference Szilas, Kelemen and Bernstein2015). All amphiboles have been recalculated using the spreadsheet of Locock (Reference Locock2014). Mg# = Mg / (Mg+Fe²⁺). Edenite component is the A-site occupancy (Schumacher, Reference Schumacher2007), i.e. ^AK + ^ANa + ^ACa; Ca is usually not found in the A site in Amikoq amphiboles.

Green Ca-amphibole: Insight from the Si, Al^IV, Ti and edenite component

With respect to ‘amphibolitic’ host rocks an important aspect of amphibole compositions is their Al^IV content (and implicitly, Si content), which is associated with the coupled Al-tschermakite substitution (Schumacher, Reference Schumacher2007). This substitution is highly temperature dependent (Beard and Lofgren, Reference Beard and Lofgren1991; Holland and Blundy, Reference Holland and Blundy1994) but independent of bulk composition in the mafic range (Helz, Reference Helz1973) and coexistence of minerals such as plagioclase (Gilbert et al., Reference Gilbert, Helz, Popp, Spear, Veblen and Ribbe1982). Experimental studies have demonstrated the edenite component to be dependent on bulk composition (Helz, Reference Helz1973). However, Gilbert et al. (Reference Gilbert, Helz, Popp, Spear, Veblen and Ribbe1982) and Beard and Lofgren (Reference Beard and Lofgren1991) considered the edenite component also to be temperature dependent. Helz's (Reference Helz1973) experimental data is plotted in Fig. 12, illustrating the temperature dependency, but also the significant impact of bulk composition on the edenite component. The Ti-tschermakite substitution is largely independent of bulk rock composition but dependent on temperature and oxygen fugacity ($f_{{\rm O}_ 2}$), which control Fe–Ti oxide stability (Helz, Reference Helz1973). Given the absence of hematite and presence of magnetite at Amikoq, $f_{{\rm O}_ 2}$ conditions probably approached the QFM buffer, supported by pleochroic amphibole as amphiboles formed under high $f_{{\rm O}_ 2}$ (HM-buffer) tend to be non-pleochroic (Helz, Reference Helz1973). The edenite component (and Si) plotted against Mg# in Fig. 12 (and S18a) reveal key differences between amphibole of the supracrustal host rocks and the ALC lithologies, where the former is considerably more Fe rich than the latter and generally have a greater edenite component although overlap occurs.

The Al^IV in the supracrustal amphiboles range is 1.0–1.5 apfu, matching Al^IV in experiments conducted between 700–925°C (Helz, Reference Helz1973), which corresponds well with the expected metamorphic temperature conditions at Amikoq. This close resemblance between the green Ca-amphibole from Amikoq supracrustal rocks and experimental data is due to the common co-existence of ilmenite. Ilmenite buffers the Ti content in equilibrium amphibole, increasing as temperature increases until exhaustion of ilmenite (at ~1010°C in the 1921 Kilauea olivine tholeiite; Gilbert et al., Reference Gilbert, Helz, Popp, Spear, Veblen and Ribbe1982). An assemblage lacking a Ti-buffering phase will yield apparently low-Ti amphiboles, even at high temperature.

Amphiboles of ALC olivine-rich ultramafic parageneses change type as a function of metamorphic facies and also in lower grade facies in which tremolite is gradually changed to pargasite at granulite facies (Evans, Reference Evans1977; Robinson et al., Reference Robinson, Spear, Schumacher, Lairdeales, Klein’, Evans, Dollan, Veblen and Ribbe1982). The transition from a tremolite- to a pargasite-bearing assemblage occurs at upper amphibolite facies, coincident with sillimanite + garnet + biotite in metapelites (Robinson et al., Reference Robinson, Spear, Schumacher, Lairdeales, Klein’, Evans, Dollan, Veblen and Ribbe1982). Apart from the paragenesis, the metamorphic grade can be traced in terms of Si content of amphibole, as tremolites have 8 Si apfu and pargasite contains 6 Si apfu plus an edenite component which increases with decreasing Si (Robinson et al., Reference Robinson, Spear, Schumacher, Lairdeales, Klein’, Evans, Dollan, Veblen and Ribbe1982). The Amikoq ultramafic mineral association constitutes peak olivine + orthopyroxene + Al-spinel + magnesio-ferri-hornblende and retrograde chlorite + Cr-magnetite. The amphibole Si content range is 6.81–7.25, and as high as 6.7–7.5 apfu (Fig. S18a; Kristensen, Reference Kristensen2006). Such a Si range correlates well with expectations of upper amphibolite facies that should host olivine + orthopyroxene + spinel + magnesium-hornblende with amphibole Si content in the range of 6.5–7.6 apfu (Robinson et al., Reference Robinson, Spear, Schumacher, Lairdeales, Klein’, Evans, Dollan, Veblen and Ribbe1982). At higher grades, clinopyroxene joins the assemblage and Si in amphibole is <6.6 apfu (Robinson et al., Reference Robinson, Spear, Schumacher, Lairdeales, Klein’, Evans, Dollan, Veblen and Ribbe1982). The Si range of amphibole in Amikoq ultramafic rocks is substantial relative to amphibole from individual orthopyroxenite and norite samples which tends to be quite homogenous, possibly indicating a more heterogeneous re-equilibration in the ultramafic rocks related to the chlorite + Cr-magnetite retrograde event. This also suggests that amphibole pre-dates chlorite formation. Thus, even if the ultramafic amphibole were igneous in origin, they appear to have re-equilibrated at lower temperatures. However, the most probable origin for the amphibole in ultramafic rocks is deemed to be metamorphic.

Amphiboles of the ALC noritic and orthopyroxenite rocks, the ALC green amphiboles, generally have a lower edenite component at a given Al^IV content relative to the amphibolitic supracrustal rocks and also relative to: igneous amphibole from Alaskan layered complexes (Himmelberg and Loney, Reference Himmelberg and Loney1995); an olivine cumulate of the Stillwater Complex (Page and Zientek, Reference Page and Zientek1987); and arc xenoliths (e.g. Tollan et al., Reference Tollan, Bindeman and Blundy2012), which typically have large edenite components (Fig. 12a,c). In addition, most ALC amphiboles have edenite components (<0.6) significantly lower than generally observed in experimental amphiboles crystallised from basaltic to andesitic bulk compositions (Berndt, Reference Berndt2002; Helz, Reference Helz1973) all of which tend to have high edenite components (>0.5; Fig. 12). If the norites and supracrustal rocks equilibrated at similar P–T-$f_{{\rm O}_ 2}$ conditions, then the edenite component should be directly relatable to bulk-rock alkali content. The experiments of Helz (Reference Helz1973) demonstrated that at similar P–T conditions an increase in Na₂O+K₂O bulk rock composition of ~1.5 wt.% was required to increase the equilibrium amphibole edenite component by ~0.2 (Fig. 12c). As such, the bulk-rock median Na₂O+K₂O difference among the supracrustals (2.04 wt.%), norites (0.96 wt.%) and melanorites (0.59 wt.%; unpublished data) predict a difference in amphibole edenite component of 0.2 between the melanorites and supracrustals. However, the supracrustal amphibole have an edenite component (0.23–0.54; this study; Kristensen, Reference Kristensen2006) overlapping that of amphibole in the unretrogressed norite (0.39–0.52) and melanorites (0.32–0.55; Fig. 12) indicating that the bulk-rock alkali content is too low to support such a high edenite component. This suggests that some of the edenite component in norite and melanorite amphibole have been inherited from igneous precursors. Furthermore, the green amphibole from the retrogression-free norite generally overlap with igneous amphibole from the Alaskan complexes (Himmelberg and Loney, Reference Himmelberg and Loney1995) in terms of Si, Mg#, edenite component and Ti whereas Al^IV tends to be slightly higher at a given Ti and edenite component in the Amikoq norite (Figs 12, S18a). The elevated Al^IV is probably caused by the high Fe³⁺ content, which is involved in the supposedly temperature and $f_{{\rm O}_ 2}$ independent Fe³⁺ + Al^IV = Si + Mg coupled ferric-tschermakite substitution (Fig. S18b; Gilbert et al., Reference Gilbert, Helz, Popp, Spear, Veblen and Ribbe1982). Additionally, ilmenite exsolution in places is observed in the green Ca-amphibole (Fig. S11j); most commonly where proximal to orthopyroxene, reflecting subsolidus exsolution of Ti at lower temperatures. Such Ti exsolution probably reflects a previous high Ti content not stable under subsequent metamorphic conditions suggesting an igneous provenance. High Fe³⁺ content might also have been inherited.

Melanorite and orthopyroxenite amphiboles have elevated Cr content (Fig. S16b) with no apparent Cr phase present to buffer the amphibole, except for rare chromite/Cr-hercynite (five small grains observed) in the case of the orthopyroxenite. Similarly to the edenite component, this might indicate that the Cr₂O₃ is inherited from a melt, because amphiboles have ${\rm D}_{{\rm Cr}}^{{\rm Amp}/{\rm melt}} $ ≥3.5 (Adam and Green, Reference Adam and Green2006; Tiepolo et al., Reference Tiepolo, Oberti, Zanetti, Vannucci, Foley, Hawthorne, Oberti, Della Ventura and Mottana2007).

Petrographic observations also support an igneous provenance. In other metanorites or metagabbros, the reaction anorthite + enstatite + H₂O = hornblende + ‘Al phase’ occurred, where the Al phase is kyanite (Foslie, Reference Foslie1945) or corundum (Jasarová et al., Reference Jasarová, Racek, Jerabek and Holub2016; Owens and Dymek, Reference Owens and Dymek1997; Subramaniam, Reference Subramaniam1956). These Al phases are not observed at Amikoq, and chlorite, though occurring in the majority of investigated rocks, is not a ubiquitous phase associated with green Ca-amphibole. Thus, the green amphibole–orthopyroxene reaction textures (Fig. 4e–f) can be represented locally as:

(2)$${\rm Interstitial}\,{\rm melt} \lpar { \pm {\rm orthopyroxene}} \rpar \;\to {\rm green}\,{\rm Ca}-{\rm amphibole} \lpar { \pm {\rm plagioclase}} \rpar $$

The parental melt of Amikoq is as yet unidentified, but could be similar in character to the proposed high-degree mantle melts (>40%) of the nearby Seqi Complex (Fig. 2), possibly a type of komatiitic/picritic melt ± contamination and/or boninitic melt (Szilas et al., Reference Szilas, van Hinsberg, McDonald, Næraa, Rollinson, Adetunji and Bird2018). Such melts could easily be envisaged to crystallise minor late intercumulus amphibole in the noritic rocks with lower edenite component and Ti content than typical arc- and Alaskan-type amphiboles (Fig. 12). This hypothesis is also consistent with the increasing edenite component with decreasing Mg# in melanorites and unaltered norite (Fig. 11a). Nevertheless, the variable Si–Al^IV in the noritic lithologies (Figs 12, S18) probably records some metamorphic re-equilibration.

In summary, Al^IV–Ti characteristics indicate that the host supracrustal ‘amphibolitic’ green Ca-amphibole equilibrated in the temperature range ~700–825°C, corroborating ~740–790°C derived from the plagioclase–amphibole thermometer. Green Ca-amphibole of the olivine-bearing rocks equilibrated under similar conditions judging from the Si content, and these amphiboles are most probably of metamorphic origin. In contast, the noritic amphiboles probably have an igneous provenance, as suggested from plagioclase–amphibole thermometer results. The elevated edenite component and Cr₂O₃ content of noritic green Ca-amphibole, together with somewhat elevated Ti in the unretrogressed norite, indicate some inherited igneous compositions, implying that amphibole was a minor intercumulus phase.

Pale-green high-Al Ca-amphibole

This amphibole species has only been observed in retrogressed norites and as a retrograde phase in the hornblende melagabbro. In particular, the Al^IV/Al^VI ratio was found to decrease with increasing grade, with Al^IV/Al^VI of <2 corresponding to garnet-(staurolite) and kyanite zones. Similar inferences were made for Al^IV/Al^VI in high-P Ca-amphiboles from Sudbury where ratios of ~1.8 were deduced to reflect high pressure, middle amphibolite-facies conditions (Fleet and Barnett, Reference Fleet and Barnett1978). These inferences about Al in amphibole are consistent with experimentally-derived amphiboles where Al^IV, Al^VI and Al^tot tend to increase with increasing pressure (Gilbert et al., Reference Gilbert, Helz, Popp, Spear, Veblen and Ribbe1982). The 5 kbar experimental amphiboles of Helz (Reference Helz1973) all plot in the ‘Unaltered igneous’ or ‘Low-pressure’ domains, in agreement with the division proposed by Fleet and Barnett (Reference Fleet and Barnett1978) at pressures ≈ 5 kbar corresponding to Al^IV/Al^VI ≈ 2.0. That some experimental igneous amphiboles plot in the low-pressure metamorphic field (i.e. Al^IV/Al^VI <3.3) underlines the caution that should be exercised where interpreting amphibole Al^IV–Al^VI relationships (Fleet and Barnett, Reference Fleet and Barnett1978). A high proportion of Fe³⁺ could result in lower Al^VI than might be expected for a given pressure (Fleet and Barnett, Reference Fleet and Barnett1978), moving otherwise high-pressure amphiboles towards the low-pressure and igneous side of Fig. 13. In the case of the Amikoq pale-green high-Al Ca-amphibole, Fe³⁺ does not follow the general trend defined by the other amphiboles (Fig. S18b). The pale-green high-Al Ca-amphiboles are distinguished by their high Al^VI with Al^IV/Al^VI ≈ 1.5–2.0 and low Ti content (Figs 12, 13) and are similar in composition to sadanagaites and ferro-tschermakites from the Sudbury Igneous Complex and country rock metabasalts formed at high pressure, middle amphibolite-facies conditions related to a later orogeny (Fleet and Barnett, Reference Fleet and Barnett1978; Jørgensen et al., Reference Jørgensen, Tinkham and Lesher2019). This indicates that the Amikoq pale-green high-Al Ca-amphiboles formed at relatively high-pressure conditions. Comparing the Ti–Al^IV–Al^VI of Amikoq pale-green high-Al Ca-amphiboles to similar amphiboles from Sudbury and the Scottish Barrovian zones suggest a pressure of formation of >5 kbar, possibly ~7 kbar (Fleet and Barnett, Reference Fleet and Barnett1978; Jørgensen et al., Reference Jørgensen, Tinkham and Lesher2019). This probably coincided with the amphibolite-facies temperature decrease recorded in ‘amphibolitic’ host rocks (Fig. 12).

Fig. 13.

Al^IV vs. Al^VI (apfu) in amphibole. The plot has been subdivided into qualitative fields where ‘Unaltered igneous’ amphiboles have Al^IV / Al^VI >3.3, ‘Low-pressure’ metamorphic 2.0–3.3 and ‘High-pressure’ metamorphic have <2.0. The division of low and high pressure correspond roughly to 5 kbar (Fleet and Barnett, Reference Fleet and Barnett1978), consistent with the 5 kbar experimental data of Helz (Reference Helz1973). The roughly negative trend in Amikoq amphiboles, especially the host rocks, is probably related to a mixed igneous and high-pressure overprint. Note the distinct pale-green high-Al Ca-amphiboles. All other plotted data as in Fig. 12. After Fleet and Barnett (Reference Fleet and Barnett1978).

Age and origin of the zircon-hosting quartzo–feldspathic pockets

The zircon U–Pb ages presented in this investigation reflect either: (1) igneous crystallisation associated with the emplacement of the ALC in evolved melt pockets; or (2) crystallisation in small anatectic melt pockets related to peak metamorphic conditions. These scenarios are discussed below with emphasis on Amikoq.

Zircon can grow from evolved, trapped late-stage melt pockets in mafic systems (Heaman and LeCheminant, Reference Heaman and LeCheminant1993; Wall et al., Reference Wall, Scoates, Weis, Friedman, Amini and Meurer2018) encouraging an examination of the Amikoq zircons hosted in quartzo–feldspathic pockets observed in orthopyroxene-dominated ALC lithologies (Fig. 3g). Corroded, partly recrystallised <250 μm zircons displaying well-preserved igneous-style oscillatory zoning from homogeneous norite samples FOS2 and QUA yield a pooled age of 3011 ± 4 Ma. It is contended that 3011 Ma is a best estimate of the age of intrusion of the Amikoq and similar northern bodies, assuming a coeval relation (Waterton et al., Reference Waterton, Hyde, Tusch, Hollis, Kirkland, Kinney, Yakymchuk, Gardiner, Zakharov, Olierook, Münker, Lightfoot and Szilas2020). In contrast, zircon domains in these samples that indicate recrystallisation of oscillatory-zoned zircon yield an age of 2997 ± 8 Ma, which is interpreted as reflecting recrystallisation during high temperature (amphibolite–granulite facies) metamorphism. The 3004 Ma age from the zircon megacryst with igneous zonation, but complex margins (Fig. S3), points to the onset of granulite-facies metamorphism at that time. The common occurrence of euhedral zircon megacrysts hosted in small quartzo–feldspathic pockets (Fig. 3g) is consistent with zircon growth in a silica-saturated melt. This melt is probably unrelated to trapped evolved melt pockets associated with crystallisation of the ALC, because K-feldspar usually crystallises together with quartz and hydrous phases in such evolved pockets (Heaman and LeCheminant, Reference Heaman and LeCheminant1993; Wall et al., Reference Wall, Scoates, Weis, Friedman, Amini and Meurer2018), whereas Ca-rich plagioclase crystallised with the Amikoq zircon (See Results section ‘Zircon petrography and geochronology’). Furthermore, zircon from such evolved pockets are usually ≤150 μm in grain size (Wall et al., Reference Wall, Scoates, Weis, Friedman, Amini and Meurer2018) and thus unlike the ≥5 mm Amikoq zircon. The patchy appearance of the pockets (Fig. 3f–g) makes distally-derived melts (e.g. TTG) unlikely, as these would be expected to be more akin to dykes.

Yakymchuk et al. (Reference Yakymchuk, Kirkland, Hollis, Kendrick, Gardiner and Szilas2020) presented evidence for partial melting of metabasites in western Akia, with development of orthopyroxene- + clinopyroxene-bearing leucosomes and amphibole-rich melanosomes. The partial melting was modelled to have occurred at <8 kbar with temperatures proposed locally to have been >900°C. These leucosomes are strikingly similar to field observations at Amikoq, as shown by the clinopyroxene–orthopyroxene–amphibole–plagioclase–quartz assemblage (compare Fig. 3f with figure 3 of Yakymchuk et al., Reference Yakymchuk, Kirkland, Hollis, Kendrick, Gardiner and Szilas2020). Thus, the assemblage observed in the field at Amikoq is interpreted to represent a leucosome. Uranium–Pb zircon geochronology of the west Akia pyroxene-bearing leucosomes yielded crystallisation ages of c. 3010–2990 Ma (Yakymchuk et al., Reference Yakymchuk, Kirkland, Hollis, Kendrick, Gardiner and Szilas2020) statistically indistinguishable to the 3004 ± 9 and 2990 ± 13 Ma zircon U–Pb ages derived from Amikoq zircon megacrysts (this study; Nilsson et al., Reference Nilsson, Söderlund, Ernst, Hamilton, Scherstén and Armitage2010). Consequently, the Amikoq quartzo–feldspathic pockets and associated zircon megacrysts are proposed here to represent partial melting. These zircons grew in an anatexis event shortly after the probable emplacement of the igneous bodies at 3011 ± 4 Ma, but before an event at 2997 ± 8 Ma leading to local zircon recrystallisation.

The obtained ages of 3004 ± 9 Ma and 2997 ± 8 Ma for thermal events matches the c. 3010–2970 Ma age inferred for the Mesoarchean granulite-facies event(s) affecting the Akia terrane (Friend and Nutman, Reference Friend and Nutman1994; Garde, Reference Garde1997; Garde et al., Reference Garde, Friend, Nutman and Marker2000, Reference Garde, Dyck, Esbensen, Johansson and Möller2014; Kirkland et al., Reference Kirkland, Yakymchuk, Hollis, Heide-Jørgensen and Danišík2018). However, several studies have highlighted that zircon crystallisation commonly post-dates peak pressure and slightly peak temperature in classical clockwise P–T–t paths (Jørgensen et al., Reference Jørgensen, Tinkham and Lesher2019), and the presented Amikoq age could date the waning stages of peak P–T. Nevertheless, the zircon investigated provides a robust minimum age of 3004 ± 9 Ma for the ALC and a time constraint on peak metamorphism, or the waning stages of peak P–T, at ~3000 Ma, predating local tonalite formation as evident from Seqi tonalites with U–Pb zircon ages of 2978 ± 8 and 2969 ± 5 Ma (Szilas et al., Reference Szilas, van Hinsberg, McDonald, Næraa, Rollinson, Adetunji and Bird2018; Whyatt et al., Reference Whyatt, Peters, Pack, Kirkland, Balic-Zunic and Szilas2020). We note that the zircon age determined by Nilsson et al. (Reference Nilsson, Söderlund, Ernst, Hamilton, Scherstén and Armitage2010) originated from a similar leucosome hosted in a North Margin orthopyroxenite, which we consequently also suggest to date peak-metamorphic conditions at Amikoq consistent with previous suggestions (Schersten and Garde, Reference Scherstén and Garde2013; Szilas et al., Reference Szilas, Kelemen and Bernstein2015).

Finally, it should be emphasised that the intrusive ages we report here for samples QUA and FOS2 are not actually from the ALC, which is the focus of the present study, but they are rather for intrusive noritic rocks that do not have a verified co-magmatic relationship to the ALC. Ravenelle et al. (Reference Ravenelle, Weiershäuser and Cole2017) and Waterton et al. (Reference Waterton, Hyde, Tusch, Hollis, Kirkland, Kinney, Yakymchuk, Gardiner, Zakharov, Olierook, Münker, Lightfoot and Szilas2020) found similar ages of 3014 ± 2.8 Ma to 3002 ± 5.4 Ma for the Maniitsoq Norite Belt, but a robust magmatic age of the ALC remains to be established.

Anatectic textures versus relict igneous poikilitic textures

If suprasolidus conditions are correct, then an examination of poikilitic plagioclase in melanorites as potential anatectic leucosomes might be warranted (Fig. 4e–f). A pseudosection of a South Margin melanorite (191234) with typical poikilitic plagioclase textures was prepared (Table 1; Fig. 14). A thin section of the modelled composition show a mineral association consisting of orthopyroxene (~80%, En_82.4–83.8) + poikilitic and zoned plagioclase (~10%, An_69.3–84.9, excluding An_52.9–57.9, see Petrography section ‘Plagioclase’) + green Ca-amphibole (~10%, Mg#_0.90–0.96) with a few rare grains of biotite. Zoned plagioclase and its textural relation to orthopyroxene and green Ca-amphibole suggest that plagioclase ± Ca-amphibole essentially represent a former melt phase. Modelled orthopyroxene shows no appreciable variation (~En₈₅ throughout).

Fig. 14.

(a) Pseudosection results of melanorite 191234 (Fig. 4e–f) from the South Margin in P–T space. Small insets from top to bottom. (b) Plagioclase anorthite content contours. (c) Quartz content (in vol.%). (d) Degree of partial melt in vol.%. Inferred P–T conditions experienced by the rock correspond to the Opx + Amp + Qz field, below the solidus temperature. This phase assemblage is inconsistent with that observed in the thin section, suggesting that the observed texture is most likely to be relict igneous rather than anatectic. An alternative plausible assemblage is indicated in bold red in (a), (b) and (c) – see text for details. Transparent fields correspond to recorded peak metamorphic estimates of the sil–grt–crd gneiss (Fig. 12). Orange full lines are plagioclase–hornblende thermometry (Holland and Blundy, Reference Holland and Blundy1994) results of melanorite 191234 and 191146, thick dashed lines are the minimum and maximum temperature results of retrogressed norites and the thin dashed line is the minimum temperature result of norite (maximum temperature is at >1000°C).

The plagioclase–hornblende temperatures for sample 191234 yielded 795–850°C, similar to the temperature derived from the sil–grt–crd gneiss and an expected pressure of 7–8 kbar seems reasonable judging from the sil–grt–crd gneiss and pale-green high-Al Ca-amphiboles. At ~820°C and 7–8 kbar, considered the minimum P–T experienced by the melanorite, a stable mineral assemblage composed of orthopyroxene + amphibole + quartz is predicted. Plagioclase is not stable in the model at these conditions for this bulk rock composition. Furthermore, no liquid is predicted while detectable quartz should be present (~4–7 vol.%; Fig. 14). However, quartz was not observed in the thin section. Any anatectic melt produced from an essentially basaltic protolith is expected to be quartzo–feldspathic (Beard and Lofgren, Reference Beard and Lofgren1991; Helz, Reference Helz1973; Palin et al., Reference Palin, White, Green, Diener, Powell and Holland2016). This is also evident from the phase relations modelled here, where the first melt increments correlate well with a decrease in quartz content. Such a SiO₂-rich melt has not been recognised petrographically, though this could in principle merely be due to the putative produced liquid not being present in the thin section utilised in this study. Jørgensen et al. (Reference Jørgensen, Tinkham and Lesher2019) observed networks of quartz + low-Ca plagioclase rims on high-Ca plagioclase cores in a basaltic hornfels from Sudbury and interpreted these as former anatectic melts. The Sudbury plagioclases bear some resemblance to the ALC melanorite plagioclase (Fig. S10g) excepting for the presence and absence of quartz, respectively. It is plausible that temperatures exceeded 820°C in which case a maximum temperature is indicated by the appearance of clinopyroxene at ~900°C. In the scenario where quartz is not observed petrographically due to low modal proportion (<5%) and the small degrees of melt is either removed or represented by the lower An plagioclase rims, then the modelled melanorite might correspond to the Opx + Amp + Pl + Qz + liq field, corresponding to ~890–915°C at 7.5 kbar. Regardless, these small degrees of melting cannot account for the observed cumulate textures, which require the interstices between orthopyroxene crystals to have been occupied by liquid (~10–20% melt); had this liquid been anatectic in origin, quartz should be a significant interstitial phase. Thus, the observed textures in this specific thin section have a dominantly relict igneous origin and any putative partial melting of ALC rocks with these compositions must have been insignificant.

Relict igneous textures and mineralogy are also supported by plagioclase–hornblende and biotite–orthopyroxene thermometry (800 to 1070°C in norites and melanorites) and green Ca-amphibole geochemistry. These temperatures are considerably higher than the temperature conditions under which biotite is predicted stable in the pseudosection, suggesting these biotites are igneous in origin. This is supported texturally as biotite existing as discrete subhedral–euhedral grains and not as overgrowths or blebs. Late interstitial biotite is quite common in layered complexes, e.g. Stillwater (Page and Zientek, Reference Page and Zientek1987), Skaergaard (McBirney, Reference McBirney1996), Bushveld (Eales and Cawthorn, Reference Eales and Cawthorn1996) and Alaskan type complexes (Himmelberg and Loney, Reference Himmelberg and Loney1995).

Preservation of igneous textures probably reflects these remaining as structurally competent units as a consequence of an essentially anhydrous subsolidus history. After emplacement and crystallisation, the ALC was possibly dry relative to its host rocks, which will make any reaction at lower P–T sluggish and prevent formation of hydrous metamorphic minerals. In other cases, such as the retrogressed norites, fluids were introduced resulting in the partial breakdown of orthopyroxene to (Fe,Mg,Mn)-amphibole. This fluid introduction was probably related to local high-strain zones acting as fluid pathways.