Two Neoarchean tectonothermal events on the western edge of the North Atlantic Craton, as revealed by SIMS dating of the Saglek Block, Nain Province, Labrador Open Access

The dating of gneissic complexes is commonly complicated by the effects of multiple tectonothermal events, each of which can produce belts of highly strained, plastically deformed, partially melted and strongly metamorphosed rocks, with previous geological relationships commonly obscured or obliterated. As zones of crustal weakness and/or rheological contrast, such belts are also loci for tectonic reactivation, and it is important to decipher the sequence and extent of metamorphic and tectonic events that produced the gneisses if meaningful correlations are to be attempted for disparate terranes that underwent subsequently separate geological histories. Here we present new data on the timing of high-grade metamorphism and deformation in the Archean Nain Province of the North Atlantic Craton; namely, in the Torngat Mountains region of northern Labrador.

The Nain Province (Fig. 1a) consists of Archean gneisses that extend for 500 km along the Labrador coast, from Makkovik to Nachvak Fjord (Taylor 1971), with a likely extension to the Avayalik Islands near the tip of the eastern Labrador Peninsula (Scott 1995). It forms the western margin of the North Atlantic Craton, conjugate to southwestern Greenland prior to the opening of the Labrador Sea (Bridgwater et al. 1973). To the north and west, the Nain Province was reworked at c. 1.8 Ga by the north–south-trending Torngat Orogen, which juxtaposed the North Atlantic Craton with Archean continental terranes in the Churchill Province and Paleoproterozoic arc rocks in the Burwell Domain (Van Kranendonk 1996). The Nain Province has been subdivided into the Saglek and Hopedale blocks, north and south respectively of the town of Nain (Fig. 1a). In the vicinity of Saglek Bay, tonalite–trondhjemite–granodiorite (TTG) gneisses of the Saglek Block are the product of multiple episodes of high-T metamorphism and ductile deformation that produced a regional dome and basin pattern elongated in a north–south direction, and delineated by interspersed layers and tectonic enclaves composed of supracrustal (metavolcanic and metasedimentary) rocks (Bridgwater et al. 1975; Ryan & Martineau 2012). During the Torngat Orogeny, the Nain Province, along with an unconformably overlying sequence of Paleoproterozoic sediments that include the Ramah Group (Fig. 1), was overthrust by strongly reworked basement gneisses of unknown age (Van Kranendonk & Ermanovics 1990; Rivers et al. 1996).

A section of the Saglek Block investigated in recent studies by the same group (Kusiak et al. 2018; Sałacińska et al. 2018, 2019; Whitehouse et al. 2019) extends for 80 km along the Labrador coast, from Ramah Bay in the north to Hebron Fjord in the south (Fig. 1b). Gneisses in the southern part of the section formed in several stages throughout the Archean, with the oldest and most abundant TTG-type protoliths (the Uivak gneiss) formed between c. 3850 and 3600 Ma (Schiøtte et al. 1989; Komiya et al. 2017; Kusiak et al. 2018; Sałacińska et al. 2018, 2019), with lesser episodes of felsic plutonism at c. 3.3–3.0, 2.7 and 2.5 Ga (Schiøtte et al. 1990, 1992; Krogh & Kamo 2006; Komiya et al. 2015; Sałacińska et al. 2019). Krogh & Kamo (2006) suggested that TTG gneisses from outcrops around Saglek Bay differ in age across the Handy Fault (Fig. 1), with c. 3.6 Ga protoliths to the west and c. 3.3 Ga to the east. However, recent dating (Komiya et al. 2015; Kusiak et al. 2018; Sałacińska et al. 2018, 2019) revealed complicated and tectonized relationships between Eoarchean and younger TTG gneisses on both sides of the fault. Metamorphosed supracrustal assemblages of sedimentary and volcanic rocks, with associated ultramafic gneisses, occur as discontinuous layers within gneissosity typically less than 100 m thick, and have been divided into sparsely distributed pre- or syn-Uivak supracrustal rocks (the Nulliak assemblage) and post-Uivak Meso- to Neoarchean Upernavik supracrustal rocks (Bridgwater & Schiøtte 1991). Isotopic U–Pb and Hf data from detrital zircon from Upernavik metasediments indicate deposition after c. 3.0 Ga, and it has been suggested that this unit includes unrelated supracrustal packages with various ages of deposition (Schiøtte et al. 1992). Similarly, there is some uncertainty about the age of deposition of volcanic rocks and sediments that in part formed the Nulliak assemblage. Detrital zircons with an age of c. 3850 Ma (Nutman & Collerson 1991) supported deposition after that time, but Komiya et al. (2015) and Shimojo et al. (2016) favoured a greater antiquity, namely, >3.9 Ga, for these rocks. Whitehouse et al. (2019) have recently questioned this, suggesting that there is confusion over the assignment of metasedimentary and mafic gneisses to the Nulliak or Upernavik ‘assemblages’. Graphite-bearing metasediments, claimed by Komiya et al. (2015) to belong to the Nulliak assemblage, contain detrital zircon that demonstrate a much younger provenance (see Whitehouse et al. 2019, after Schiøtte et al. 1992). Also, mafic tectonic enclaves near Nulliak Is. that were assigned to the Nulliak assemblage on the map by Ryan & Martineau (2012) have Sm–Nd isotopic signatures that show that some of these are much younger than the enclosing Uivak gneiss (Morino et al. 2017). In the absence of more extensive direct dating of tectonic enclaves and belts, the true age of many of the supracrustal rocks in the Saglek Block remains in question.

The assembly of the Saglek Block, comprising the Eoarchean Uivak gneiss, Mesoarchean tonalitic to gabbroic gneisses, and supracrustal packages, may be attributed to c. 2.7 Ga tectonism during the widespread high-T metamorphism (Schiøtte et al. 1990; Krogh & Kamo 2006; Kusiak et al. 2018; Sałacińska et al. 2018). The above-cited studies were mostly focused around Saglek Bay, but c. 2.7 Ga metamorphic zircon was also identified as far south as Drachart Island (Schiøtte et al. 1990), and as far north as the Avayalik Islands near the tip of the eastern Labrador pensinsula, where Scott (1995) proposed an extension of the Nain Province as reworked crust within the Torngat Orogen. A corresponding c. 2.7 Ga craton-forming event is recognized in similar gneisses on the conjugate section of southwestern Greenland (e.g. Nutman et al. 2004, 2013; Kirkland et al. 2018).

Post-dating the formation of the gneisses in the Saglek Block, 2.5 Ga mineral ages (U–Pb zircon, monazite and titanite, and K–Ar hornblende; see Discussion for references) have been attributed to the thermal and hydrothermal effects of post-tectonic granitic magmatism (Baadsgaard et al. 1979; Schiøtte et al. 1992). Alternatively, c. 2.5 Ga monazite and titanite ages from offshore drilling samples collected by Wasteneys et al. (1996), along with c. 2.7 Ga ages of detrital zircon, led Connelly & Ryan (1996) to infer a north–south-trending tectonic boundary between the Saglek and Hopedale blocks. The Hopedale Block comprises late Paleoarchean to Neoarchean protoliths metamorphosed at c. 3.0–2.8 Ga. The inferred boundary with the Saglek Block extends offshore north of Nain; but has not been directly observed, being obscured by the extensive Proterozoic Nain Plutonic Suite. Connelly & Ryan (1996) suggested a link between the Saglek–Hopedale boundary and the Okak shear zone (van Kranendonk & Helmstaedt 1990), to the south. The Okak shear deforms a granitic pluton on Okak Island (Fig. 1), which was inferred by Schiøtte et al. (1992) to have an age of c. 2.5 Ga (based on unpublished data of Roddick and van Kranendonk and the age of metamorphic monazite in adjacent metasediments).

More recent monazite dating by Kusiak et al. (2018) has increased the known extent of high-temperature metamorphism at both c. 2.7 and c. 2.5 Ga in the Saglek Block between Ramah Bay and Hebron Fjord. However, it is not clear whether these ages represent a prolonged period of high-T metamorphism, a gneiss-forming event with subsequent passive thermal activity and granite emplacement, or two discrete tectonothermal events. The purpose of this paper is to re-evaluate the timing and significance of deformation and metamorphism in the Saglek Block, utilizing new U–Pb isotopic dating of zircon and monazite.

This study investigates the sequence of deformation events in the Saglek Block, based on coastal field work between Ramah Bay and Hebron Fjord (Fig. 1), conducted by our team in 2014 and 2017, and the scheme by van Kranendonk & Helmstaedt (1990) for the North River–Nutak area, 100 km south of Saglek Bay. In almost all localities, pre-deformation relationships between the diverse rock types have been transposed into a high-strain gneissosity and a multistage deformational history has long been recognized (Morgan 1975; Collerson & Bridgwater 1979; Schiøtte et al. 1990). Because gneissosity affects both Eoarchean and younger Archean TTG protoliths, as well as Mesoarchean Upernavik supracrustal rocks, and because published age data indicate widespread zircon and monazite growth during metamorphism at c. 2.7 Ga (Bridgwater & Schiøtte 1991), this may have been the tectonothermal event during which the Saglek Block was assembled. Commonly observed intrafolial folds of gneissic laminations are indications of high-strain ductile deformation (D₁) prior to that which produced the dominant gneissosity (D₂). The dominant gneissosity is in most places vertical to steeply dipping, with lesser domains of low-angle layering, such as that observed on the cliff face at Cape Uivak (Fig. 2a). The difference between D₁ and D₂ structures can be observed where S₂ flattens leucosome in gneisses with S₁ gneissosity and/or mineral foliation (Fig. 2b–d). Elsewhere, S₁ has been transposed into a high-strain, moderately dipping to subvertical gneissic S₂, which trends predominantly north–south. It is unknown whether D₂ significantly postdates D₁, but because it transposes D₁ fabrics in late Mesoarchean to early Neoarchean Upernavik supracrustal rocks, as well as older TTG gneisses, and no intrusive rocks separate D₁ and D₂, it is likely that they represent stages of a single tectonothermal event. Dating of gneisses with Eoarchean protoliths has established an additional, much older episode of high-grade metamorphism at c. 3.6 Ga (Sałacińska et al. 2018, 2019); however, no large-scale structures have been distinguished in the field that relate to this earlier event. In outcrops c. 100 km to the south of Saglek Bay, van Kranendonk & Helmstaedt (1990) described ductile thrusting (F₀) in the Upernavik supracrustal rocks and recumbent folding (F_n+1) in both supracrustal rocks and TTG gneisses during high-T, high-P metamorphism. These are low-angle features, in contrast to the predominantly steep north–south-trending nature of S₂ gneissosity in most of the Saglek area. Although it is possible that these relate to the low-angle macrofold at Cape Uivak, the latter, along with recumbent folds on nearby Big Is., have been attributed either to nappes produced during late Archean, regional, asymmetric folding that generated the present map pattern (Bridgwater et al. 1975) or else to a separate recumbent folding event superimposed on the gneiss pattern produced by the aforementioned asymmetric folding (B. Ryan, pers. comm. 2019).

Post-D₂ structures tend to be localized. Upright minor folds are found with axial planes parallel to dominant gneissosity, and are interpreted as recording the waning stages of D₂ tectonism. There are abundant granitoid stocks, sills and dykes that cut the dominant gneissosity. Such granitoids have been classified by previous researchers (Bridgwater & Schiøtte 1991; Schiøtte et al. 1992) as ‘post-tectonic’, with metamorphism attributed to late syntectonic magmatism in the waning stages of Neoarchean tectonism. However, in many localities between Saglek Bay and Hebron Fjord, granitoid stocks and dykes are strongly deformed, especially on the islands and in eastern coastal regions. On Dog Is., coarse metagranite that intrudes Uivak gneiss has a steep S₃ foliation that is axial planar to open F₃ folds where granitic melt has pooled in fold noses (Fig. 2e). Intense L₃ defined by stretching of recrystallized fabrics in both pre-D₂ gneisses and post-D₂ granitoids indicates high rates of simple shear during the D₃ event. Dynamic recrystallization of granitoid produced augen gneiss 1 km to the east of St John's Harbour (Fig. 2f), and the augen show alignment with coarse-grained biotite in quartz and orthoclase. This alignment is parallel to that in the matrix, where quartz, feldspar and biotite have recrystallized into an anastomosing S₃ foliation (Fig. 2g). This is consistent with progressive crystallization of the granitoid during a high-strain ductile event. Mylonite micro-shear zones cut across S₃ foliation. There is an increase in D₃ strain eastwards and southeastwards from Saglek Bay to the coast, with increasing development of F₃ meso-to-macro folding with variably plunging fold axes and intense L₃ stretching and recrystallization parallel to fold axes. Such features are possibly related to D_n+3 structures described by van Kranendonk & Helmstaedt (1990) further south, which they attributed to a large amphibolite-facies shear zone that runs north–south along the coastal fringe of the Saglek Block south of Saglek Bay to Okak Island, where it deforms syntectonic granitic plutons assumed to have intruded at c. 2.5 Ga (Schiøtte et al. 1992). However, this shear zone involves retrogression of granulites to amphibolite-facies gneisses, whereas no such shear zone-related retrogression in association with D₃ structures is observed around Saglek Bay or in granulites around Hebron Fjord.

Several samples were collected for age determination between Ramah Bay and Hebron Fjord (Fig. 1b). Metamorphic grade at these localities varies from amphibolite to granulite facies (Ryan & Martineau 2012), albeit with varying degrees of later lower-grade overprinting, especially around Ramah Bay. The samples include felsic orthogneisses (L1414, L1488, L1489, L1491 and L1493), intermediate orthogneisses having the chemical characteristics of altered volcanic rocks (L1458 and L1487), mafic granulites (L1453 and L1490) and metapelitic gneiss (L1492). A sample of syn-D₃ granitoid (L1412) was also collected. Classification of orthogneisses is based on whole-rock geochemistry, using the total alkali v. silica diagram (Middlemost 1994) for igneous protoliths with <65 wt% SiO₂, and the ternary normative feldspar classification of Barker (1979; after O'Connor 1965). Plots are provided with geochemical data in the supplementary material. Here, samples are briefly described according to structural relationship and locality (Fig. 1b). Mineral modes and major element geochemistry are presented in Tables 1 and 2, respectively.

Fine- to medium-grained grey felsic orthogneisses matching the description of Uivak I gneiss were collected from Reichel Head (L1491, L1493), Little Ramah Bay (L1488, L1489) and Dog Island (L1414). All have granoblastic fabrics with S₂ defined by millimetre- to centimetre-scale quartzofeldspathic layers (leucosome) and aligned biotite with or without hornblende. Leucosome consisting of intergrown quartz and feldspar (Fig. 3a) or quartz and mesoperthite (Fig. 2d), which is coarser than the granoblastic fabric in the host gneiss, has resulted from crystallization of partial melt, with minor recrystallization on grain margins providing evidence of limited subsequent deformation. Samples L1488 and L1489 are representatives of trondhjemitic orthogneisses from Little Ramah Bay that have, respectively, an abundance and a scarcity of nebulitic leucosome. These two samples were combined for the purpose of dating. The leucosome of sample L1414 felsic orthogneiss from Dog Island (Fig. 2b) is stromatic and the host gneiss varies from a patchily heterogeneous texture, interpreted as the recrystallization of a coarse-grained granitoid (L1414A), to a finer-grained, homogeneous pale grey gneiss with few laminations of leucosome (L1414B). Parts A and B were therefore dated separately.

Mafic samples were collected from S₂ layers hosted by TTG orthogneisses at Little Ramah Bay (L1490) and the south shore of Hebron Fjord (L1453). The latter was tentatively assigned by Ryan & Martineau (2012) to the Nulliak supracrustal assemblage; however, the outcrop also contains aluminous metasediments more typical of the Upernavik supracrustal rocks. Both samples have granoblastic texture with two-pyroxene- and hornblende-bearing assemblages typical of mafic granulite generated from basaltic protoliths (Fig. 3b), but L1453 has a stronger foliation, with stromatic leucosome and associated garnet–biotite-rich selvages. A subvertical NW-trending S₂ gneissosity at the Hebron Fjord locality is crenulated by open to tight F₃ folds with SW-dipping axial planar S₃ defined by aligned biotite and NW-plunging axes. Some patches of garnet-leucosome truncate S₂ but are deformed by D₃ structures, indicating partial melting of mafic orthogneiss during both events.

The S₂ gneissosity and stromatic leucosome found in the orthogneisses is also present in sample L1492 of metapelite from Reichel Head (Fig. 2c). The sample is rich in graphite, similar to metasedimentary rocks that have been claimed to be early Eoarchean in age by Tashiro et al. (2017). Garnet poikiloblasts enclose S₂-aligned flakes of biotite, graphite and sillimanite (Fig. 3c). Leucosome rich in K-feldspar and quartz is also flattened into S₂, and the mineral assemblage is characteristic of granulite-facies metamorphism.

Pyroxene–quartz-bearing samples were taken from Little Ramah Bay (L1487) and Upernavik Island (L1458). The former has been described by Kusiak et al. (2018), whereas the latter is a typical orthopyroxene–garnet gneiss (Fig. 3d) found interlayered with aluminous and mafic gneisses that form the Upernavik ‘assemblage’ (Ryan & Martineau 2012). It is more aluminous than typical andesite, but unlike metapelite from the same locality, it contains abundant orthopyroxene, and is chemically characteristic of altered metavolcanic or volcanoclastic rocks, similar to Mesoarchean deposits at Qussuk and Storø in southwestern Greenland (Szilas et al. 2016, 2017). Composition grades across S₂ gneissic layers from orthopyroxene–plagioclase (L1458A) to garnet–biotite (L1458B) gneiss; however, the difference is in modal proportion only, and all phases are present in both rock types. Abundant leucosome is present in L1458, as S₂-cutting layers with biotite-rich selvages and euhedral garnet poikiloblasts, as are commonly formed through incongruent melting of pelitic rocks (L1458C; Fig. 3d). Garnet is anhedral and slightly poikilitic in both types, with large inclusions of rutile and quartz (Fig. 3e). The leucosome is quartz-rich and moderately deformed, with warped and recrystallized quartz grains wrapping garnet poikiloblasts that contain S₃-aligned grains of biotite and monazite (Fig. 2h). The presence of coarse biotite flakes aligned with the foliation in biotite-rich selvages, and with the sub-mylonitic recrystallization of quartz in the leucosome, supports the interpretation that garnet formed through incongruent melting of the host gneiss, and that the melt crystallized under stress.

Detailed analytical protocols and data reduction procedures are provided in the Appendix. Determination of bulk-rock geochemistry for major elements was undertaken by Acme Labs in Vancouver, Canada, through Bureau Veritas, Poland. For Zr-in-rutile thermometry, electron microprobe (EMP) analysis was carried out on a Cameca SX-100 instrument at the Electron Microprobe Laboratory, State Geological Institute of Dionýza Štúra, Bratislava, Slovakia. For ion microprobe analysis, plugs drilled from polished thin sections, and monazite and zircon mineral grains separated from crushed samples, were mounted in epoxy, polished and imaged by scanning electron microscope (SEM) with backscattered electron (BSE) and cathodoluminescence (CL) detectors at the John de Laeter Centre, Curtin University, Western Australia. Isotopic U–Pb dating of zircon and monazite grains was by sensitive high-resolution ion microprobe (SHRIMP II) at the John de Laeter Centre, Curtin University in Perth, Western Australia, and by CAMECA IMS 1280 ion microprobe at the NordSIMS facility, Swedish Museum of Natural History, Stockholm. All ion microprobe data are quoted with 1σ analytical errors, whereas weighted mean and discordia intercept ages are quoted at 95% confidence levels, and include the decay-constant error of the concordia curve.

The abundance of rutile in sample L1458 allowed for in situ EMP analysis of zirconium contents to estimate temperatures of mineral growth. The formulation of Watson et al. (2006) was used; data are provided in Table 3. Three grains of rutile included in garnet from the garnet–biotite-rich part of the sample (L1458B, Fig. 3f) yielded ZrO₂ contents of 0.3394–0.3622 wt%, equivalent to 861–866°C, whereas grains enclosed in plagioclase or biotite yielded more variable, lower contents (0.1277–0.2450 wt%) equivalent to 747–817°C. The former temperatures are the best estimate for the temperature of garnet growth during metamorphism on Upernavik Is., and the estimate is consistent with granulite-facies metamorphism in the Saglek Bay area, as has been suggested by various researchers (e.g. Krogh & Kamo 2006; Ryan & Martineau 2012).

Zircon grains from all samples are subhedral to anhedral, with cathodoluminescence imaging (Fig. 4) revealing rims with anhedral, graduated or sector zoning typical of growth under high-grade metamorphic conditions. Cores having euhedral, graduated and/or oscillatory growth zoning, typical of crystallization from an evolving magma, were found in all samples except L1453 (Hebron Fjord) and L1490 (Little Ramah Bay), the latter two having rounded or irregular cores without distinct growth zoning. Sub-grain domains of zircon with features typical of growth during metamorphism were targeted for spot analysis in all samples, and zircon grains with magmatic growth features were targeted in samples L1412 (near St Johns Harbour), L1414A/B (Dog Island), L1487 (Little Ramah Bay) and L1491 (Reichel Head). Isotopic U–Pb data (Table 4) are presented in Tera–Wasserburg concordia plots (Fig. 5) along with ²⁰⁷Pb/²⁰⁶Pb mean ages for concordant populations and Model 1 discordia intercept ages for linear arrays. Older outliers from rounded or irregular cores, which are interpreted as xenocrystic or inherited zircon, were not included in the calculation of ages and statistics from the identified populations. For cores and grains with growth zoning characteristic of igneous zircon, Model 1 discordia chords were calculated with forced lower intercepts of 2720 ± 50 Ma, approximating the time period within which granulite-facies gneisses were estimated to have formed from older magmatic protoliths in the Saglek Block (Kusiak et al. 2018, and references therein). Statistical test values (mean square of weighted deviates; MSWD) and other details are provided with the concordia plots in Figure 5.

For andesitic orthogneiss L1487 and trondhjemitic orthogneiss L1491, discordia chords yield upper intercept ages of 3664 ± 35 Ma and 3715 ± 26 Ma, respectively. The latter includes five concordant data with a mean ²⁰⁷Pb/²⁰⁶Pb age of 3714 ± 11 Ma. Data from zircon in trondhjemitic orthogneiss L1414B spread between c. 3650 and 3590 Ma. Mean ²⁰⁷Pb/²⁰⁶Pb ages were also derived from igneous zircon in meta-trondhjemite layer L1414A (2749 ± 3 Ma) and meta-monzonite L1412 (2547 ± 3 Ma). In all cases, the estimates are interpreted as the time of crystallization of igneous protoliths, with the exception of L1414B; in that sample, the cluster of analyses at c. 3590 Ma can be interpreted as a maximum age only for the protolith, assuming that they were not disturbed by later metamorphism. Igneous zircon from metapelite L1492 yielded scattered ages between c. 3280 and 2950 Ma, which are interpreted as dating detrital sources for the metasediment, although here again, the possibility of disturbance at c. 2.7 Ga cannot be discounted.

Zircon with metamorphic morphologies, either as rims with discordant boundaries to cores or as distinctly equant rounded and sector-zoned grains, yielded statistically valid (MSWD ≤ 1.3) mean ²⁰⁷Pb/²⁰⁶Pb ages for samples of andesitic orthogneiss L1487 (2742 ± 8 Ma), trondhjemitic orthogneiss L1488 (2750 ± 7 Ma) and mafic granulite L1490 (2739 ± 9 Ma). Slightly more scattered data were derived from metapelite L1492 (c. 2750–2720 Ma) and metagranite L1493 (c. 2740–2710 Ma). Two data from light-CL rims in zircon from sample L1414A yielded c. 2710 Ma ages. Together, these data from six samples are interpreted as dating zircon growth during high-T metamorphism between c. 2750 and 2710 Ma. The dataset from mafic granulite sample L1453 is more complex, with 51 analyses from unzoned, concentric and sector-zoned grains and cores ranging between c. 2740 and 2680 Ma (group 1 ages, Fig. 5), and 25 data from unzoned or gradationally zoned rims ranging between c. 2570 and 2510 Ma (group 2 ages). Analyses in group 1 record variable U contents (Fig. 5), whereas those from group 2 have uniformly low U contents. The groups represent periods of zircon growth during two separate metamorphic events. To better define the gap in time between the events, subsets of statistically equivalent data were extracted from the youngest ages in group 1 and the oldest ages in group 2. The 42 youngest out of 51 data in group 1 yield a mean ²⁰⁷Pb/²⁰⁶Pb age of 2702 ± 2 Ma, and the 20 oldest data out of 25 in group 2 yield a mean ²⁰⁷Pb/²⁰⁶Pb age of 2551 ± 6 Ma. These mean ages provide statistically robust estimates for the minimum age of zircon growth in the first metamorphic event, and the maximum age of growth in the second, respectively.

To constrain the timing of mineral growth in high-grade metamorphic assemblages, plugs containing monazite and surrounding minerals were drilled from polished thin sections, mounted and analysed by secondary ion mass spectrometry (SIMS; Table 5, Fig. 5). Monazite in andesitic orthogneiss L1487 occurs as xenoblastic grains in a granoblastic assemblage that is strongly retrogressed, as described by Kusiak et al. (2018). Owing to the marginal alteration of monazite grains in thin section (Fig. 6a), data were also taken from unaltered fragments of monazite separated from the orthogneiss and mounted in a polished epoxy plug. Excluding three slightly younger, discordant data points, 12 analyses from a combination of grains in drilled thin sections and separates yielded a mean ²⁰⁷Pb/²⁰⁶Pb age of 2709 ± 14 Ma. Monazite in metapelite sample L1492 is unaltered and has polygonal grain boundaries with other metamorphic phases (Fig. 6b), and yields a mean ²⁰⁷Pb/²⁰⁶Pb age of 2727 ± 6 Ma. Ages were also collected from each part of metavolcanic rock L1458 (A, B and C). Two 10 μm wide monazite inclusions in garnet porphyroblasts from the garnet–biotite-rich part (L1458B) yield spot ages of c. 2680 and c. 2670 Ma. Monazite occurs more abundantly in association with garnet-leucosome L1458C, in which millimetre-scale preferentially aligned inclusions in garnet poikiloblasts are parallel to S₃ defined by biotite inclusions (Fig. 6c). Four inclusions of monazite yield ages that range from 2550 to 2510 Ma. Excluding the two oldest analyses, nine data yield a mean ²⁰⁷Pb/²⁰⁶Pb age of 2522 ± 7 Ma, which can be considered as a robust statistical minimum age for the period of metamorphism. Age data were also obtained from monazite separated from the orthopyroxene-rich part (L1458A) and yielded a mean ²⁰⁷Pb/²⁰⁶Pb age of 2551 ± 5 Ma. Mean ages from samples L1487 and L1492 are attributed to monazite growth during the first period of high-T metamorphism. Monazite from sample L1458, dated as inclusions in garnet from parts B and C, indicates multiple stages of mineral growth. Those grains present in cross-cutting garnet-leucosome (L1458C) fall within the second period of mineral growth at 2.5 Ga identified in zircon from other samples in this study, as does monazite in the matrix of part A. The two monazite inclusions in the garnet–biotite-rich part (B) fall between the two stages of zircon growth in other samples, but agree with some monazite age estimates obtained by Kusiak et al. (2018). This may be an indication of monazite growth and/or disturbance continuing after 2700 Ma, but as a separate generation from the second stage of growth from 2550 to 2510 Ma.

The new results from monazite and zircon associated with metamorphic assemblages and deformation fabrics, especially where supported by dating structurally constrained meta-granitoids, provide evidence of two distinct high-temperature tectonothermal events: at 2750–2700 Ma and 2550–2510 Ma (Fig. 7). The separation of the two episodes of high-T mineral growth is clearer than that observed in EMP monazite dating by Kusiak et al. (2018). There is evidence of partial melting and crystallization of anatectic melt in both the earlier and later stages of each of the events. Therefore, the growth of zircon and monazite after 2550 Ma is probably not due to the ‘thermal effects’ of granitic emplacement, as suggested by Schiøtte et al. (1992); rather, it is more likely that tectonothermal activity is the progenitor of granitic melts that were emplaced both during and after high-strain deformation. These include the c. 2530 Ma major granitic stockworks described by Baadsgaard et al. (1979) on the coast and islands outside Saglek Bay. A re-examination of localities in which c. 2.5 Ga magmatism and mineral growth occurs shows that such ages are scattered along the Saglek Block from Saglek Bay to Nain, and on both sides of the Handy Fault (Fig. 8). Further north there is a lack of data; however, zircon growth during metamorphic events at c. 2.7 and 2.5 Ga have been recognized by Scott (1995) in meta-tonalites at Home and Avayalik Islands, which may be part of the Nain Province. Nevertheless, it is likely that the effects of the 2.5 Ga event increase towards the south and east, as such ages were also obtained from zircon and monazite in drill-core samples taken c. 40 km outside Saglek Bay (Wasteneys et al. 1996; see Fig. 8). In that study, the authors hypothesized a north–south-trending tectonic boundary between the Saglek and Hopedale blocks that extends southward from offshore of Saglek Bay to the coast near Okak Island. The Saglek Block comprises c. 3.7 Ga protoliths metamorphosed at c. 2.7 Ga and the Hopedale Block contains c. 3.2 Ga protoliths metamorphosed at c. 2.5 Ga. Anomalies, such as the presence in the Saglek Block of the c. 3.2 Ga Lister gneiss, from which a migmatized sample yielded c. 2.5 Ga zircon and titanite, have been attributed to tectonic intercalation of fragments of the Hopedale block within the Saglek Block at c. 2.5 Ga (Schiøtte et al. 1992; Wasteneys et al. 1996). However, the presence of c. 2680 Ma granitoid sheets cutting folded and metamorphosed Lister gneiss constrains gneiss formation to c. 2.7 Ga (Schiøtte et al. 1989). This, along with the evidence for c. 2.7 and c. 2.5 Ga metamorphism in samples from the Saglek area in our study and in that of Kusiak et al. (2018), does not contradict the terrane boundary proposed by Wasteneys et al. (1996), but does suggest that the assembly of the Saglek and Hopedale blocks was earlier than c. 2.5 Ga.

The late Archean metamorphic events and the assembly of two different crustal blocks in northern Labrador may be analogous to the juxtaposition of terranes having differing structural and metamorphic histories in the Archean of southwestern Greenland (Nutman & Friend 2007; Friend & Nutman 2019). Major late Archean terrane boundaries along the coast of southwestern Greenland tend to run NE into the glacial cap, rather than following the general north–south trend of gneisses in the Saglek area. Circa 2.8–2.7 Ga high-grade metamorphism that strongly affects the Nain Province also does so in the vicinity of Nuuk, with grade decreasing towards the east (e.g. Nutman & Friend 2007; Dziggel et al. 2017). This part of the North Atlantic Craton contains a complex mixture of Eoarchean and Paleoarchean terranes, and Mesoarchean arc assemblages, and has many similarities in timing and composition to the gneisses of the Saglek Block. The extensive c. 2560 Ma Qôrqut Granite Complex that intrudes gneisses inland from Nuuk (Nutman et al. 2011; Næraa et al. 2014) is a potential correlative of syn- to late-D₃ magmatism in the Saglek block, as marginal tectonic reworking of the Qôrqut has been observed (Nutman et al. 2011). However, no significant granite metamorphic event at c. 2.5 Ga is recognized in this part of Greenland. Some 100 km to the north, dating of c. 2.5 Ga metamorphic monazite in gneisses near Maniitsoq and inland led Dyck et al. (2015) to define the Majorqaq Belt, a NE-trending mobile belt between the main 2.7 Ga assembled part of the North Atlantic Craton and the Mesoarchean Maniitsoq block further north. Dyck et al. (2015) suggested that the belt resulted from the collision of the Maniitsoq block subsequent to southward subduction of an ocean basin beneath the North Atlantic Craton, and that the Qôrqut Granite Complex is the product of slab dewatering. The Majorqaq Belt might well correlate with c. 2.5 Ga tectonothermal activity in the Saglek Block. In this case, the Qôrqut Granite Complex would correlate well with large plutons of the same age found to the south of Saglek in the Okak area (Schiøtte et al. 1992). Indeed, the Maniitsoq block itself was subjected to marginal reworking to the north by Paleoproterozoic tectonism, similar to the northern and western margins of the Nain Province (St-Onge et al. 2009). However, there is a lack of data from the Labrador coast north of Ramah Bay that would allow any clear correlation to be made with the hypothesized ‘Majorqaq’ Belt. In addition, this would dissociate the Eoarchean Uivak gneiss from terranes of similar age in the Itsaq Gneiss Complex (Fig. 8). The latter has complicated ductile structural relationships with Paleo- to Mesoarchean terranes (Hoffmann et al. 2014). If such relationships are similarly complicated in the Saglek Block, the intercalation of older and younger crust there may also be a product of amalgamation at c. 2.7 Ga. A simpler correlation would place c. 2.5 Ga tectonothermal activity as a reworking of c. 2.7 Ga gneisses in association with the Okak and Qôrqut granitic complexes, on the eastern and southern margins of a composite Eoarchean–Mesoarchean continent. To test this hypothesis, geochronological work needs to be extended further north along the Labrador coast, and to little-studied parts of southwestern Greenland.

In situ U–Pb ion microprobe dating of monazite and sub-grain dating of zircon and monazite provide clear evidence of high-T metamorphism at both c. 2.7 and c. 2.5 Ga in the Saglek Block. Both events involved ductile deformation, with the former producing the dominant gneissosity in the Uivak and other gneisses, including supracrustal metasedimentary and metavolcanic varieties. The effects of tectonothermal activity at c. 2.7 Ga are widespread in the Saglek Block, as they are in the parts of southwestern Greenland conjugate to the Nain Province before the opening of the Labrador Sea. The known extent of c. 2.5 Ga high-T metamorphism and deformation is more limited, and may be restricted to reworking of the margins of the North Atlantic Craton in both Greenland and Labrador. However, given the emplacement of large amounts of c. 2.5 Ga granitoid in both Labrador and southwestern Greenland, it is possible that parts of the North Atlantic Craton were assembled at this time, involving the juxtaposition of pre-existing continental crust in Labrador (the Saglek and Hopedale blocks of the Nain Province) and Greenland (between the Maniitsoq block and others to the south). The evidence for late Neoarchean final assembly is still very limited, and will require extensive new geochronological studies in both Labrador and Greenland.

This study also provides zircon ages for the formation of supracrustal precursors to gneisses in the Saglek Block 30 km to the north of Saglek Bay. A protolith age of c. 3.7 Ga for andesitic orthogneiss at Little Ramah Bay is comparable with Eoarchean ages for the Uivak gneiss that have been well established by many earlier studies. Circa 3 Ga ages from detrital zircon in graphite-bearing metapelitic gneiss from Reichel Head are comparable with those obtained from similar gneiss at St John's Harbour, and support the conclusion of Whitehouse et al. (2019) that such graphite-bearing gneisses were not deposited at the beginning of the Archean, as proposed by Tashiro et al. (2017). These ages provide new evidence for the northward continuation of the Saglek Block, and demonstrate the potential for new discoveries of Eoarchean crust along the Labrador coast.

Fieldwork was carried out with the permission and support of Parks Canada (special thanks go to D. Whitaker) and the Nunatsiavut Government. N. McNaughton, I. Fletcher and B. Rasmussen are thanked for assistance with monazite dating at the John de Laeter Centre, Curtin University, Australia. A. Gawęda of the Faculty of Earth Sciences, University of Silesia in Katowice, Poland, is thanked for assistance with whole-rock geochemical analysis. The staff of the Torngat Mountains Base Camp, and especially the boat pilots and bear guards who tirelessly supported and protected us in the field, are warmly thanked. Special thanks to B. Ryan for his thorough review that greatly improved the manuscript, and to Randall Parrish for his careful editorial handling.

The lead author (D.J.D.) was supported by a Polish National Science Centre (NSC) POLONEZ Fellowship grant 2016/23/P/ST10/01214, funded through the EU H2020 research and innovation programme under MSCA grant agreement 665778. This research was also supported by NSC OPUS grant 2014/15/B/ST10/04245 to M.A.K.; grants to M.J.W. from the Knut and Alice Wallenberg Foundation (2012.0097) and the Swedish Research Council (2012-4370); and funding to S.A.W. from the Australian Research Council Centre of Excellence for Core to Crust Fluid Systems (CCFS). The NordSIMS facility operates as a Swedish national infrastructure under Swedish Research Council grant 2014-06375; this is NordSIMS contribution 611. Work by A.S. was supported by an ING-PAN internal project for young scientists (2016–2017).

DJD: Conceptualization (Lead), Formal analysis (Equal), Writing – Original Draft (Lead); MAK: Conceptualization (Equal), Formal analysis (Equal), Writing – Original Draft (Supporting); SAW: Conceptualization (Equal), Data curation (Supporting), Writing – Review & Editing (Supporting); MJW: Conceptualization (Equal), Data curation (Supporting), Methodology (Supporting), Writing – Review & Editing (Supporting); AS: Formal analysis (Supporting); RK: Formal analysis (Supporting); PK: Formal analysis (Equal).

Scientific editing by Randall Parrish

Correction notice: The copyright has been updated to Open Access.

Polished thin sections from all samples were prepared and examined by optical microscope and SEM at the John de Laeter Centre, Curtin University, Western Australia. For in situ monazite analysis (samples L1458 and L1487), 3 and 5 mm discs were drilled out of thin sections and mounted in epoxy discs. For mineral grain dating, samples were crushed and sieved, and then processed by wet panning, magnetic separation and hand picking to isolate zircon and monazite grains. The grains were mounted, along with reference materials, in epoxy discs that were then polished to expose the mid-sections of grains. All grains were imaged for internal structure before and after analysis using an SEM fitted with BSE and CL detectors. The mounts were cleaned and gold coated prior to analysis by SIMS.

Initial dating of mounted zircon grains was carried out by SHRIMP II at the John de Laeter Centre, Curtin University in Perth, Western Australia. A spot size of 20–25 μm was used with an O₂⁻ primary beam intensity of 3–4 nA. The secondary ion beam was focused through a 100 μm collector slit onto an electron multiplier to produce mass peaks with flat tops and a mass resolution (1% peak height) better than 5100 M/ΔM. Data were collected in sets of six scans, with reference standards analysed after every five sample analyses. Count times per scan for Pb isotopes 204, background position 204.1, 206, 207 and 208, were 10, 10, 10, 30 and 10 s, respectively. U–Th–Pb ratios and absolute abundances were determined relative to the zircon reference standard BR266 (559 Ma, 903 ppm U; Stern (2001). Instrumental mass fractionation (IMF) of ²⁰⁷Pb/²⁰⁶Pb was monitored during each session by repeated analysis of the zircon reference standard OGC (Stern et al. 2009). No IMF correction was required because the measured values of OGC were in agreement with the reference value within 2σ uncertainty. Raw data were processed using the SQUID 2 add-in (v. 2.50.12.03.08) for Excel 2003 (Ludwig 2009) and plotted using the ISOPLOT 3.70 add-in of Ludwig (2001). Measured compositions were corrected for common Pb using measured ²⁰⁴Pb and contemporaneous common Pb composition according to the terrestrial Pb evolution model of Stacey & Kramers (1975). Owing to the low proportion of common Pb detected in standards and samples (<1% of measured ²⁰⁶Pb, as estimated from ²⁰⁴Pb measurement), the choice of modelling age for common Pb composition did not have a statistically significant effect on age estimates. Mean ages are quoted with 95% confidence levels.

Further analysis of zircon in samples L1412, L1414, L1453, L1488-89, L1490, L1491, L1492 and L1493 was carried out by CAMECA IMS 1280 ion microprobe at the NordSIMS facility, Swedish Museum of Natural History, Stockholm. Protocols for U–Pb data closely follow published methods (Whitehouse & Kamber 2005). Zircon grains were analysed using a c. 15 µm, 6 nA O₂⁻ primary beam, and peak-hopping monocollection in an ion counting electron multiplier (EM) at a mass resolution of c. 5400 M/ΔM. Reference material 91500 (1065 Ma, 80 ppm of U; Wiedenbeck et al. 1995) was used for calibration of Pb/U ratios using the Pb/UO v. UO₂/UO calibration protocol of Jeon & Whitehouse (2015). Common Pb was corrected using the ²⁰⁴Pb counts assuming a present-day terrestrial Pb-isotope composition model (Stacey & Kramers 1975) following the rationale of Zeck & Whitehouse (1999) that this is largely surface contamination introduced during sample preparation and not common Pb residing in zircon and/or micro-inclusions. Very low amounts of common Pb were detected during the spot analyses with (<0.1% of total ²⁰⁶Pb), in many cases below detection limit for ²⁰⁴Pb based on the electron multiplier background. Where common Pb corrections were deemed necessary on the basis of measurable ²⁰⁴Pb (>3× standard deviation on the average background), these were small and therefore insensitive to the precise composition of common Pb. Data reduction was performed using the NordSIMS-developed suite of software of M. J. Whitehouse. All ion microprobe data are quoted with 1σ analytical errors, whereas weighted mean and discordia intercept ages are quoted at 95% confidence levels, and include the decay-constant error of the concordia curve.

For in situ (i.e. within polished thin section) and grain mount monazite analysis, the SHRIMP II was operated with a primary beam of O₂^– ions focused through a 50 μm Köhler aperture to produce an oval 10 μm wide spot with a surface current of 0.2–0.4 nA. Secondary ionization was measured without energy filtering on a single electron multiplier on 13 mass stations from 202 (LaPO₂) to 270 (UO₂), with a mass resolution of >5200 for the latter. Secondary ion retardation was used to eliminate ion scatter. Mass stations 202 (LaPO₂), 203 (CePO₂), 205.9 (NdPO₂), 232 (Th), 244.8 (YCeO) and 264 (ThO₂) were analysed for matrix corrections and interference on ²⁰⁴Pb, following the protocols outlined by Fletcher et al. (2010). Mass stations were measured through six cycles, with typical count times of 10 s per cycle for ²⁰⁴Pb, background (at 204.04 a.m.u.) and ²⁰⁶Pb, 30 s for ²⁰⁷Pb and 5 s for ²⁰⁸Pb. Reduction of raw data for standards and samples was performed using the SQUID 2.5 and Isoplot 3.70 add-ins for Microsoft Excel 2003 (Ludwig 2001, 2009). Age (²⁰⁶Pb/²³⁸U) and abundance of U were calibrated against reference monazite French (514 Ma; 1000 ppm U). High La and high Y–Nd–U standards Z2234 and Z2908, respectively, were used for matrix and interference corrections, following the method described by Fletcher et al. (2010). Corrections for common Pb on isotopic U/Pb values and ages were carried out with common Pb estimated from ²⁰⁴Pb counts and the composition of Broken Hill lead.

For Zr-in-rutile thermometry, electron microprobe analysis was undertaken at the Electron Microprobe Laboratory, State Geological Institute of Dionýza Štúra, Bratislava, Slovakia, utilizing a Cameca SX-100 electron microprobe equipped with four wavelength-dispersive spectrometers. Large high-sensitivity, LPET and LLIF crystals and a conventional TAP crystal were used for analysis. Analytical conditions were chosen to balance the best analytical conditions against reasonable acquisition times. An accelerating voltage of 15 kV was used, with a probe current of 200 nA. Zirconium contents were calibrated against an in-house standard.