Geology, geochemistry, and apatite/titanite U–Pb geochronology of ca. 1.88 Ga alkaline ultrabasic dykes in the Southern Province near Sudbury, Ontario

The giant impact that hit the boundary between the Archaean Superior Province and the Palaeoproterozoic Southern Province at 1850 Ma left behind not only the Sudbury Impact Structure with its associated Sudbury Igneous Complex (SIC) but also a major geophysical anomaly. Immediately to the northeast of the SIC is a similar-sized anomaly, corresponding to a surface area of some 1200 km², the Temagami Anomaly (Fig. 1). The cause of this regional conductive, strongly magnetic (11 000 nT), and gravity (+20 mGal) feature remains elusive but has been modelled to be a serpentinized ultramafic body at relatively great depth (5–20 km; Adetunji et al. 2021, for a review) that may or may not be related to the Sudbury impact event. Repeated attempts to drill into the Temagami Anomaly failed to clarify the anomaly’s cause (e.g., Meecham and Truscott 1992) although they did reveal the presence of impact-generated igneous rocks much further away from the SIC than previously known (Kawohl et al. 2019). This prompted increased interest in the magmatic history of the area northeast of the SIC (e.g., Easton et al. 2020; Adetunji et al. 2021; Kawohl 2022) because of a perceived elevated potential to host Sudbury-type Ni–Cu–PGE (±Co ± Au) sulphide deposits (e.g., Kawohl et al. 2020).

While exploring the Temagami Anomaly further, we discovered an unusual group of alkaline ultrabasic rocks both in outcrop and in a deep drill core. These rocks were clearly affected by the Sudbury impact but appear different from igneous rocks so far documented within the Southern Province of Ontario and the wider Sudbury region. Here, we provide a first field, petrographic, lithogeochemical, and isotopic characterization of these rocks. This is supplemented by U–Pb age data on titanite and apatite to place constraints on the rocks’ age and correlation with other magmatic events across the Superior Craton.

The study area is located about 50 km northeast of Sudbury, Ontario, within the Palaeoproterozoic Southern Province of the Canadian Shield (e.g., Percival and Easton 2007). Much of the study area is underlain by rocks of the Huronian Supergroup, an intra-continental rift succession (Bleeker et al. 2015) with a maximum preserved thickness of 12 km and a surface area of some 20 000 km² (e.g., Bennett et al. 1991). The Huronian Supergroup unconformably overlies the Superior Craton (locally represented by granite, gneiss, and greenstone belts) to the north, is thrust-bound by the metamorphic Grenville Province to the southeast, and borders the Great Lakes to the south (Fig. 1A). The Huronian Supergroup is made up mainly of siliciclastic sedimentary rocks, subordinate carbonates, and a basal sequence of volcanic rocks. Deposited between ca. 2.51 and 2.31 Ga (Bleeker, pers. comm.; Hill et al. 2018), the Huronian Supergroup was affected by multiple episodes of predominantly mafic magmatism at 2.48, 2.22, 1.75, and 1.24 Ga and again at 0.59 Ga (e.g., Corfu and Andrews 1986; Krogh et al. 1987; Kamo et al. 1995; Prevec 1995; Corfu and Easton 2001; Fedorowich et al. 2006; Easton et al. 2010; Ernst and Buchan 2010; Ketchum et al. 2013; Bleeker et al. 2015; Davey et al. 2019), most of which have been studied in great detail due to their inherent economic potential (magmatic Ni–Cu–PGE, hydrothermal Co–Ag), their potential as a metal source for the Sudbury impact melts (Lightfoot 2016; Keays and Lightfoot 2020, and references therein), and their overall geodynamic significance as it pertains to supercontinent cycles and palaeoenvironmental and atmospheric evolution (e.g., Kamo et al. 1995;Bleeker and Ernst 2006; Shellnutt and MacRae 2012; Ciborowski et al. 2015; Halls et al. 2015).

The Sudbury impact event, dated at 1849.5 ± 0.5 Ma (Davis 2008; Bleeker et al. 2015), resulted in the formation of one of the largest and oldest peak-ring (Grieve and Osinski 2020) or multi-ring (Spray et al. 2004) impact basins known on Earth, and simultaneously in the formation of some of the world’s largest economic concentrations of nickel (e.g., Lightfoot 2016). The apparent diameter of this now deeply eroded and tectonically deformed impact structure has been estimated at 180–260 km based on lineament analyses and the maximum outcrop extent of pseudotachylitic breccia, locally referred to as Sudbury Breccia (e.g., Spray et al. 2004). Allochthonous impact melt rocks, formed by instantaneous bulk melting and homogenization of the local continental crust (e.g., Kenny et al. 2017), are now preserved as an assemblage of breccias (the Onaping Formation; Muir and Peredery 1984), dykes (Grant and Bite 1984), and a voluminous (∼8000 km³) stratified igneous complex (Naldrett and Hewins 1984) but only within the former centre of the impact basin, collectively referred to as the SIC (Giblin 1984). The SIC and to a lesser extent its brecciated footwall are the principal host of magmatic Ni–Cu–PGE (±Au ±Co) sulphide mineralization. Consensus exists that the metals that are currently being extracted from the SIC were not derived from the impactor itself, either a comet or chondritic asteroid (e.g., Petrus et al. 2015; Mougel et al. 2017), but are solely due to reworking of an unusual mafic, pre-enriched crust in the target area of the impact (Keays and Lightfoot 2020). Consequently, there is an incentive to gain a better understanding of the pre-impact crustal architecture and target rock stratigraphy of the uniquely endowed Sudbury Impact Structure.

The Sudbury Impact Structure, the Huronian Supergroup, and most of the affiliated intrusive rocks in the Southern Province of Ontario were repeatedly subjected to regional metamorphism and affected by Proterozoic orogenic events (e.g., Easton 2000; Corfu and Easton 2001; Raharimahefa et al. 2014; Papapavlou et al. 2017). Metamorphism in the study area, the so-called Cobalt Embayment (Fig. 1A), did not exceed the lower greenschist facies (equivalent to the chlorite zone), and deformation is restricted to gentle folding and faulting (Meyn 1977; Dressler 1982, 1986), including sinistral strike-slip faulting related to the regional ca. 1.9 Ga NNW-striking Onaping Fault System (Buchan and Ernst 1994). Metamorphic grade and the intensity of Palaeoproterozoic compressional deformation increase to the southwest and reach a maximum (staurolite zone) just southwest of Sudbury (e.g., Easton 2000). In addition, there is evidence of basin-wide but poorly understood alkali metasomatism that affected the Huronian Supergroup sedimentary rocks at some time in the Palaeoproterozoic (e.g., Schandl et al. 1994; McLennan et al. 2000).

The first of two alkaline mafic dykes was discovered in 2018 between Laundry Lake and Bonesteel Lake (Mackelcan Township), 45 km northeast of the city of Sudbury (Fig. 1A). The dyke was emplaced into quartzite and arkose of the Lorrain Formation (Cobalt Group, upper Huronian Supergroup). The dyke is weakly to moderately magnetic, up to 7 m in width, strikes between N40°W and N65°W, and can be traced for 2 km in (sub-)outcrop. Contacts with the country rock are sharp, chilled, and vertical. On the eastern shore of Laundry Lake, the dyke is disrupted by a 100–200 m wide zone of megaclastic Sudbury Breccia (Fig. 2), the latter bearing all hallmarks of the typical pseudotachylitic breccia as described elsewhere within and around the Sudbury Impact Structure (e.g., Speers 1957; Rousell et al. 2003). Outcrop trenching near Laundry Lake revealed the existence of multiple mafic clasts in Sudbury Breccia, together with fragments of the local sedimentary rocks of the Cobalt Group. The mafic clasts range in shape from spherical to wispy and highly contorted, and in diameter from a few centimetres to as much as 12 m. In one place (Fig. 2D), the intrusive contact between the mafic lithology and the Lorrain Formation arkose appears to be preserved as a composite clast. All mafic clasts (e.g., Fig. 2A) are internally brecciated and pervaded by anastomosing veins and stockworks of altered pseudotachylite and (or) ultracataclasite. The extent of this brecciation decreases towards the east (with increasing distance from the SIC), where essentially undisturbed outcrops of the same mafic dyke can be found (e.g., at 46°52′17″N, 80°36′35″W).

Petrographically, the clasts and the dyke are indistinguishable, although the degree of pseudotachylitic veining is higher in the clasts. The rock can be described as homogeneous, medium-grained (∼2 mm), equigranular, and melanocratic (see electronic supplementary material S1). Relatively fresh samples consist of ∼70 vol% mafic minerals (actinolite, pargasite) and ∼30 vol% anhedral interstitial plagioclase. Whereas the fibrous actinolite generation in this rock is clearly secondary after clinopyroxene, the short-prismatic Ti-Al-rich alkali amphibole (pargasite) likely represents a primary magmatic mineral, possibly a phenocryst (supplementary material S2). Accessory minerals, found in fresh and altered samples alike, are skeletal millimetre-sized semi-opaque Fe-Ti-oxides (leucoxene) and euhedral and transparent titanite. A characteristic feature of the rock that helps to distinguish it from other mafic rocks in the wider area is the abundance (1–2 vol%) of apatite. Apatite occurs as euhedral prismatic to acicular crystals, 0.4–1.5 mm in length and 40–70 µm in diameter. It is typically found in clusters of up to 30 grains poikilitically enclosed in plagioclase, amphibole, titanite, and locally overgrown by porphyroblastic epidote.

Retrograde metamorphism of the dyke is evident by the texturally destructive replacement of primary magmatic minerals by chlorite, actinolite, and especially epidote, the latter replacing plagioclase (as saussurite) or occurring as millimetre-sized porphyroblasts. No penetrative tectonic fabric is evident, which indicates that the dyke had largely escaped Proterozoic deformation. The observed secondary mineral assemblage of the dyke is diagnostic of the lower greenschist facies (≤350 °C), in agreement with the peak metamorphic conditions previously established for this region (e.g., Dressler 1982; Easton 2000). In addition, quartz-carbonate-chlorite-haematite veining can be intense, especially in the vicinity to Sudbury Breccia and along the intrusive contact between the dyke and its host rock. Some of this alteration seems to predate the brecciation.

A second mafic dyke was discovered during a re-examination of Falconbridge’s historic 1991 drill core M-SH-2 into the Temagami Anomaly (46°56′29″N, 80°30′01″W, Sheppard Township, 55 km northeast of Sudbury). The dyke occurs in the upper part of intersected Huronian Supergroup sedimentary rocks from 1369 to 1377 m depth. It was originally misidentified as Gowganda Formation wacke when the core was first logged (Meecham and Truscott 1992). The dyke has an apparent thickness of 8 m, and its orientation is not known. The dyke in drill core M-SH-2 is non-magnetic, green, mostly aphanitic with abundant margin-parallel veins and, therefore, difficult to distinguish from the crudely bedded and pervasively chloritized sedimentary rocks in its footwall and hanging wall; intrusive contacts are indistinct.

Metamorphosed under greenschist-facies conditions, the original magmatic protolith of the dyke is largely obscured. Where sufficiently coarse-grained (up to 2 mm), the rock exhibits a relict ophitic texture of saussuritized plagioclase (30–40 vol%) and fibrous actinolite (60–70 vol%), the latter arguably pseudomorphic after pyroxene. Leucoxene and biotite (∼2 vol%) occur disseminated throughout; apatite is an accessory mineral. No titanite, zircon, or monazite was observed in thin section. Veins of clinozoisite, chlorite, and actinolite are common, in places crosscutting each other, and thus representing multiple generations. In addition to this regional greenschist-facies metamorphism, which is evident throughout the entire core, the dyke and its host rocks have locally been affected by silicification, carbonatization, and chloritization.

Representative hand specimens from the two dykes were collected along strike and across the drill core, respectively. In addition, mafic clasts from several trenches (Fig. 2) were channel-sampled and compared to the undisturbed dyke. Their chemical composition was determined through conventional lithium borate fusion and energy-dispersive X-ray fluorescence spectrometry (XRF) at the Department of Geodynamics and Geomaterials Research (University of Würzburg), and through inductively coupled plasma – mass spectrometry (ICP–MS) at the Department of Geological Sciences (University of Cape Town, UCT) by following an analytical protocol as described by Kawohl et al. (2020). After conventional column separation, the Nd and Sr isotope ratios of selected whole-rock samples were measured on an MC–ICP–MS at UCT (see Kawohl et al. 2019, 2020 for details).

For age determination, a 30 kg sample of the least altered, thickest, and non-brecciated part of Dyke #1 (450 m east of Laundry Lake, 46°52′17″N, 80°36′35″W) was subjected to a standard procedure of heavy mineral separation involving a commercial Wilfley table and a Frantz isodynamic separator, following initial grinding (jaw crusher, disc mill) and sieving to <250 µm grain size. Due to the limited amount of core material available, no attempt was made to extract accessory minerals from the fine-grained mafic dyke in drill core M-SH-2. A selection of suitable and representative mineral grains was handpicked from the least magnetic fraction, mounted in epoxy resin, polished, and studied by electron microprobe for internal structures. The instrument used for this was a JEOL JXA-8800L Superprobe at the Department of Geodynamics and Geomaterials Research (University of Würzburg), equipped with four wavelength-dispersive spectrometers and detectors for secondary electrons and backscattered electrons (BSEs).

The U–Pb dating was carried out at the Frankfurt Isotope and Element Research Centre (FIERCE) at the Goethe University Frankfurt, using a Thermo Scientific Element XR sector field ICP-MS. The instrument is coupled to a RESOlution 193 nm ArF excimer laser (ComPex Pro 102, Coherent), equipped with a two-volume ablation cell (Laurin Technic S155). Laser ablation-ICP-MS dating of apatite and titanite followed a modified version of the analytical protocols of Millonig et al. (2013), Müller et al. (2018), and Zulauf et al. (2021), using a 50 µm spot size, zircon GJ-1 as a primary reference material, and Durango Apatite and Namaqualand Titanite as secondary reference materials. Analytical details as well as analytical results for samples and reference materials are provided in the online supplementary material (S3). Uranium-Pb dates were calculated and graphs generated using IsoplotR (Vermeesch 2018). All uncertainties are reported at the 2σ level.

The least magnetic heavy mineral fraction of the mafic dyke consists of pyrite, apatite, and titanite; zircon and baddeleyite are conspicuously absent although the whole-rock analyses yielded relatively high concentration of >200 ppm Zr (Table 1). Absence of zircon or baddeleyite in spite of high Zr concentrations is, however, not uncommon in alkaline rocks (e.g., lamprophyres; Seifert and Kramer 2003; Craddock et al. 2007; Higgins et al. 2018) and necessitates the use of alternative geochronometers such as titanite or apatite.

The recovered titanite grains are 50–250 µm in size, euhedral, and display the typical sphenoid (wedge-shaped) habit dominated by {111}. Under the microscope, all grains are transparent, optically homogeneous, bright orange to amber in colour, and show smooth and lustrous crystal faces. Many titanite grains contain inclusions of apatite (Fig. 3) consistent with both minerals being part of the same paragenesis. Titanite shows a combination of faint, oscillatory, and sector zoning in BSE images. Resorption zones or even sieve-textured grains were not observed, and no inclusions of silicates or Fe–Ti oxides were noted. The apatite grains are very uniform with respect to length (90–120 µm), diameter (35–50 µm), habit (acicular simple prisms dominated by $(101¯0)$⁠), and optical appearance (colourless, transparent, homogeneous, vitreous lustre), suggesting that they all are part of the same generation (Fig. 3). Neither mineral inclusions nor compositional zoning were observed in the apatite.

A total of 50 titanite grains were selected for U–Pb geochronology, with one analytical spot per grain. All grains have relatively low U and high Th concentrations (2.9–8.76 ppm U; Th/U = 1.58–15.59; S3). In ²³⁸U/²⁰⁶ Pb vs. ²⁰⁷Pb/²⁰⁶Pb space (Tera and Wasserburg 1972), the data define a linear array that reflects binary mixing between common Pb and radiogenic Pb (Fig. 4A). An unanchored regression line defined by the 50 data points yielded a lower intercept at 1876.0 ± 8.7 Ma (2σ; MSWD = 1.20) and an initial ²⁰⁷Pb/²⁰⁶Pb ratio of 0.9176 ± 0.0127.

Analyses of 69 apatite grains (69 analytical spots) yielded consistently low U but higher Th concentrations (0.1–3.1 ppm U; Th/U 5.50–26.2; S3). When plotted on a Tera–Wasserburg diagram, the apatite U–Pb data define a regression line with a lower intercept at 1880.9 ± 8.3 Ma (2σ; MSWD = 1.12) and an initial ²⁰⁷ Pb/²⁰⁶ Pb ratio of 0.9024 ± 0.0069 (Fig. 4B). The upper and lower intercepts are within the uncertainty of the values obtained for the titanite fraction.

A summary of representative whole-rock geochemical data is presented in Table 1, and the complete set of analyses, including geographical coordinates, is available as online supplementary material (S4). The data highlight two important aspects: First, there are striking similarities between the two different dykes. Second, the composition of Dyke #1 is identical to the mafic clasts in the Sudbury Breccia. Dyke #1 and its correlated clasts in Sudbury Breccia are ultrabasic (41.8–44.7 wt% SiO₂), olivine- and nepheline-normative, and have a Mg# between 44 and 50. A significant variability is observed in the concentrations of Rb, K, Ba, Sr, and Pb. This is likely a result of secondary element mobility associated with the retrograde metamorphic breakdown of feldspar and (or) metasomatic gain during carbonate alteration. Alkali and alkaline earth metals are thus presumably not representative of the protolith’s composition. In contrast, the high field strength elements (HFSE) Nb, Ta, Ti, P, Zr, Hf, and Th as well as the rare earth elements (REEs) are traditionally regarded as being relatively immobile during weathering, hydrous alteration, and low-grade metamorphism (e.g., Pearce 1996; Polat and Hofmann 2003; Ague 2017), and their concentrations are therefore considered more representative of the magmatic protolith. The mafic dyke and associated clasts are strongly enriched in these HFSEs and REEs, which conforms to an alkaline rock composition (Figs. 5A–5D). Of note, the dyke has particularly high concentrations of TiO₂ (≤3.6 wt%), Nb (≤50 ppm), Zr (≤240 ppm), and P₂O₅ (≤0.7 wt%). Dyke #2 is also ultrabasic (40.3–45.9 wt% SiO₂), alkaline, and olivine-normative. It is slightly more primitive with respect to Mg# (53–57), richer in Ni and Cr, and it has lower concentrations of incompatible lithophile elements. In primitive mantle-normalized plots (Fig. 5E), Dyke #2 shows a hump-shaped trace element pattern typical of modern ocean island basalts (OIBs) although with overall lower absolute concentrations than in Dyke #1. The trace element patterns for Dyke #1 and its supposed clasts in Sudbury Breccia are indistinguishable, supporting the hypothesis that the latter are derived from the former. Furthermore, there are no negative Nb-Ta-Ti anomalies in either rock (Nb/Th > 10), pointing to minimal crustal contamination. It is also evident from Fig. 5E that both dykes are characterized by strongly fractionated REE profiles and a depletion in HREE (La/Yb_N 7.6–15.5; Dy/Yb_N 1.5–2.1; TiO₂/Yb 1–2) likely due to low degrees of melting and the presence of residual garnet in their deep mantle source region (e.g., Pearce 2008).

Whole-rock Sm–Nd isotope data for selected samples are presented in Table 2, and, as Sm and Nd are both relatively insoluble in most fluids, their isotope ratios should be representative of the magmatic protolith as well. Dyke #1 has ¹⁴³Nd/¹⁴⁴Nd ratios corresponding to a present-day ε_Nd between –16 and –17. One-stage mantle extraction ages (for mafic dykes with ¹⁴⁷Sm/¹⁴⁴Nd < 0.13 that lack crustal contamination, these should approximate their emplacement age) are between 1911 and 1990 Ma. These values are, within the analytical error (±50 Ma), indistinguishable from the Nd model ages obtained from the mafic clasts in Sudbury Breccia (T_DM 1891–1974 Ma), thereby providing supporting evidence that the mafic clasts and the dyke are related and have a similar, if not the same, origin. Dyke #2 has similar ¹⁴³Nd/¹⁴⁴Nd ratios corresponding to a present-day ε_Nd of –14. Apparent mantle extraction ages for Dyke #2 are slightly older (T_DM 1970–2073 Ma) than those for Dyke #1. Initial ε_Nd(i) values, calculated for the inferred emplacement age of 1880 Ma, range from +1.72 to +3.60, indicating deviation from a slightly depleted mantle source. The uniform and exclusively positive ε_Nd(i) values are additional evidence that the parental magma cannot have been affected by extensive crustal contamination, which is further corroborated by a lack of negative Nb–Ta–Ti anomalies and overall OIB-like trace element patterns (Fig. 5), as well as low whole-rock SiO₂ concentrations (≤45 wt%), a lack of country rock xenoliths, and a lack of quartz and zircon xenocrysts.

Whole-rock Rb-Sr isotope data (Table 2) are more variable than those of the Sm-Nd isotopes, with measured ⁸⁷Sr/⁸⁶Sr ratios ranging considerably from 0.706 to 0.771 and 0.713 to 0.717 in Dyke #1 and Dyke #2, respectively; the calculated ⁸⁷Rb/⁸⁶Sr ratio ranges from 0.245 to 1.150 and from 0.254 to 0.406, respectively. These data do not permit to calculate any meaningful isochron age. We also noted that the large variation in Sr isotope ratios persists even when the initial ⁸⁷Sr/⁸⁶Sr(i) ratio is calculated for any arbitrary point in the Palaeoproterozoic. This apparent decoupling of the Sm-Nd and Rb-Sr isotope systems (Fig. 6) is a clear indication of sub-solidus element mobility of either Rb, or Sr, or both. In contrast to the Nd isotope ratios, the Sr isotope ratios therefore bear no petrogenetic information and will be excluded from further discussion.

The titanite dated in this study exhibits features typical of a magmatic origin, including a high Th/U ratio (Fedorowich et al. 2006; Gao et al. 2012; Rajesh et al. 2013; Zulauf et al. 2021), and complex compositional (oscillatory, faint, and sector) zoning (Paterson and Stephens 1992; McLeod et al. 2011). A high Th/U ratio of > 1 is neither expected nor particularly common in titanite grains that precipitated from relatively low-temperature hydrothermal or metamorphic fluids (e.g., Scibiorski and Cawood 2022) because of their low capacity to carry Th (e.g., Pearce 1996; Ague 2017). Besides, the titanite dated in this study differs from typical metamorphic titanite with respect to habit, optical appearance, textural association, type of mineral inclusions, or the lack thereof (cf. Corfu and Stone 1998; Corfu and Easton 2001; Gao et al. 2012; Papapavlou et al. 2017).

Apatite, although characterized by a lower closure temperature to Pb diffusion than titanite, and thus being more susceptible to isotopic disturbance and age resetting (e.g.,Corfu and Stone 1998; McGregor et al. 2019), yielded a U–Pb date of 1880.9 ± 8.3 Ma (Fig. 4B), which is within the analytical uncertainty identical to the age of the titanite fraction. Furthermore, the dated apatite grains exhibit features typical of a magmatic origin, including a high Th/U ratio and an acicular habit (e.g., Piccoli and Candela 2002). Consequently, we consider the U–Pb data of 1876.0 ± 8.7 Ma obtained on titanite and 1880.9 ± 8.3 Ma for the apatite as the best constraints on their crystallization age and thus on the age of their magmatic host rock. Such 1880 Ma intrusion age is fully consistent with the relative (crosscutting) age relations established in the field (Fig. 2).

Both the age and the alkaline affinity of the new dykes discovered by us are not in line with any of the Palaeoproterozoic dyke swarms and other occurrences of magmatic rocks in the Southern Province of Ontario. Also, the NW strike of the dykes is in disagreement with the orientation of other known Palaeoproterozoic dyke swarms in the area. We propose that the studied dykes are temporally and genetically related to the 1.88–1.87 Ga Circum-Superior Large Igneous Province (CSLIP). The CSLIP is a discontinuous magmatic belt (Baragar and Scoates 1981) that wraps for 3400 km around the Superior Craton and that has been interpreted as the erosional remnant of an originally more widespread LIP (e.g., Ernst and Bell 2010). The following—geochemically and lithologically quite diverse—magmatic units are currently considered part of the CSLIP (Figs. 7 and 8): the mafic-ultramafic (variably metamorphosed and mineralized) rocks of the Thompson Nickel, Fox River, and Winnipegosis belts (Manitoba); the Cape Smith Belt (including the economically important Raglan Formation) and the Labrador Through (northern Quebec); and the predominantly volcanic rocks of the Marquette Range Supergroup (Michigan, Wisconsin). In addition, there are a number of coeval but isolated carbonatite complexes and mafic dykes located in the interior of the craton. These include the Spanish River, Borden, Cargill, Goldray and Argor carbonatite complexes, and the Molson, Pickle Crow, and Fort Albany dykes (Fig. 7). For a detailed account on the different segments of the CSLIP, see Minifie (2010) and Ciborowski et al. (2017) and references therein.

As can be seen from a compilation of published U–Pb ages (Fig. 8), most of the CSLIP-related magmatism took place, despite its wide spatial distribution, within a very short time span between 1885 and 1870 Ma. The U–Pb ages obtained in the present study overlap with previously published ages for the CSLIP and support a correlation of our alkaline ultrabasic dykes and CSLIP-related magmatism elsewhere on the Superior Craton. The relatively large uncertainty associated with the employed dating method does, however, not permit to distinguish between discrete magmatic pulses as identified by Bleeker and Kamo (2020). Theoretically, the dykes could have been an early manifestation of CSLIP-related magmatism as locally represented by the 1883 Ma ultramafic rocks in the Thompson Nickel and Cape Smith belts (Hulbert et al. 2005; Scoates et al. 2017; Bleeker and Kamo 2018, 2020), the Pickle Crow (Bleeker and Kamo 2020) and Molson dyke swarms (Heaman et al. 2009), and the various 1885–1880 Ma carbonatites (Rukhlov and Bell 2010; Bleeker and Kamo 2020). In this respect, it is worth noting that one of these carbonatites, the 1880.6 ± 2.4 Ma Spanish River Carbonatite (Rukhlov and Bell 2010), occurs just 100 km to the west of the studied dykes (Fig. 1A). Alternatively, the alkaline ultrabasic dykes could have been emplaced at a later stage of the CSLIP during a second magmatic pulse. For example, the eruption of komatiitic lavas in the Winnipegosis Belt has been dated at 1870.3 ± 7.1 Ma (Waterton et al. 2017). Furthermore, the age of the alkaline ultrabasic dykes is, within error, identical to 1877 ± 5 Ma lamprophyre dykes in northern Michigan (Craddock et al. 2007), which have also been assigned to the CSLIP (Ciborowski et al. 2017). Similar ages of 1874 ± 3 Ma and 1870 ± 4 Ma have been reported for magmatic rocks from the Hellancourt and Murdoch formations, Kokskoak Group, Labrador Through (Machado et al. 1997). A second magmatic pulse in the CSLIP is also evident by the mafic rocks that are distributed along the northern margin of the Superior Craton, specifically the 1870 ± 2 Ma Sutton Inlier Sills and the 1870.7 ± 1.1 Ma Fort Albany Dykes in the Hudson Bay Lowlands (Hamilton and Stott 2008) as well as the 1870 ± 4 Ma Haig Sills and the coeval Flaherty Formation basalts on the Belcher, Sleeper and Ottawa islands, Hudson Bay (Hamilton et al. 2009; Fig. 7). Interestingly, these ca. 1870 Ma magmatic rocks in the Hudson Bay region are (like the dykes discovered by us) of alkaline affinity—a feature that distinguishes them from all other known, either tholeiitic or calc-alkaline, Palaeoproterozoic dyke swarms in the Superior Craton (e.g., Ernst and Buchan 2010).

A mantle plume origin for the CSLIP, although called into question by some workers (e.g., Heaman et al. 2009; Herzberg 2022), seems now widely accepted. Arguments in favour of a plume origin include the short duration of magmatic activity (Bleeker and Kamo 2020; Fig. 8); the wide spatial distribution that coincides with the diameter of a typical plume head (∼2000 km); the high MgO content of some of the igneous rocks (e.g., komatiite), which requires a thermal anomaly in their mantle source (Ciborowski et al. 2017; Waterton et al. 2017); the ocean plateau-like geochemical signature of some of the rocks (Minifie et al. 2013; Ciborowski et al. 2017); the well-established link between plumes and carbonatites (Ernst and Bell 2010); the possible existence of a radiating dyke swarm comprising the Molson, Pickle Crow, and Fort Albany dykes (Minifie et al. 2013); and the linear arrangement of carbonatites (Fig. 7) possibly delineating a failed rift structure and (or) a hotspot track (Bleeker and Kamo 2020). Uncertainty exists, however, with respect to the exact location of the initial plume centre. Early research envisaged its location near Thompson, Manitoba, and suggested that all the other magmatic occurrences to the south and the east (Great Lakes, Quebec) were fed by lateral dykes extending over >1000 km (Ernst and Bell 2010; Minifie et al. 2013). More recently, Ciborowski et al. (2017), Waterton et al. (2017), and Bleeker and Kamo (2018, 2020) proposed a single plume beneath the central Superior Province. According to their hypothesis, the thick keel of the Superior Craton (∼300 km) would have caused lateral deflection of the plume head towards regions of thinner lithosphere, such as the plate margins. Only there would significant decompression melting have occurred, and the magmas been able to rise buoyantly through the lithosphere, hence the observed spatial distribution of large volumes of ultramafic, and in places Ni–Cu–PGE-mineralized, rocks surrounding the Superior Craton. Carbonatite melts, in contrast, would have been able to migrate even through the thicker inner part of the craton, because of their exceptional low viscosity and low density.

As elegantly as the above model might account for the observed distribution of CSLIP-related magmatic rocks (Fig. 7), a major drawback of the model has been seen in the absence of 1.88 Ga magmatism in the southeastern part of the craton, specifically within the Huronian and Mistassini–Otish basins. As further noted by Heaman et al. (2009), no rocks with an OIB-like trace element signature were, prior to this study, known to be associated with the CSLIP, leaving room for speculations about a plume-unrelated mantle source. In this context, the discovery of 1.88 Ga magmatic rocks having an OIB-like affinity and cutting across strata of the Huronian Supergroup adds critical detail to our current understanding of the CSLIP as it fits perfectly with the models of Waterton et al. (2017), Ciborowski et al. (2017), and Bleeker and Kamo (2018, 2020), reinforcing the idea of a single plume head beneath the central Superior Craton.

The petrological, geochemical, and geochronological study of recently discovered NW-trending alkaline ultrabasic dykes in outcrops above and a drill core into the Temagami Anomaly northeast of Sudbury, Ontario, makes it possible to draw a number of conclusions pertaining to the tectono-thermal evolution of the Superior Craton and the adjacent Palaeoproterozoic Southern Province:

• The dykes were emplaced at 1876.0 ± 8.7 Ma and 1880.9 ± 8.3 Ma based on U–Pb dating of magmatic titanite and apatite, respectively.

• The above ages obtained for the dykes are indistinguishable from the 1885–1870 Ma CSLIP.

• The alkaline chemistry with OIB-like composition of the dykes is best explained by a mantle-plume origin and their genetic relationship with the CSLIP, specifically the 1.88 Ga carbonatites in the interior of the Superior Craton.

• Our data provide evidence of a previously unrecognized Geon 18 magmatic event in the Cobalt Embayment and in the footwall of the 1850 Ma Sudbury Impact Structure.

Especially the latter point invites speculation on the reasons for the unusually high metal endowment of the Sudbury Complex. Maybe it is related to the fortuitous overlap in space of at least three distinct mafic/ultramafic, metal-rich magmatic units in the target area of the Sudbury impact (East Bull Lake Suite, Nipissing Suite, CSLIP). The geological cause of the Temagami Anomaly remains open to debate.

The authors wish to thank the following people for their collaboration on this project: Jacob VanderWal and Winston Whymark (Inventus Mining) for field assistance and logistic support, Wolfgang Dörr (Univ. Frankfurt) for help with the mineral separation, Stefan Höhn (Univ. Würzburg) for supervising the XRF and EMP analyses, Petrus Le Roux and Christel Tinguely (Univ. Cape Town) for performing the isotope and trace element ICP-MS analyses, respectively, and Gregg Snyder (Glencore) for providing access to the historic Falconbridge drill core. This study is part of a Ph.D. project by AK, funded by the German Research Foundation DFG (grant FR 2183/12-1; project number 418960271). The dating facility FIERCE is financially supported by the Wilhelm and Else Heraeus Foundation and by the DFG (INST 161/921-1 FUGG and INST 161/923-1 FUGG), which is gratefully acknowledged. This is FIERCE contribution No. 102. Much appreciated are the detailed comments by Fernando Corfu, Wouter Bleeker, and an anonymous reviewer, who helped to improve the quality of the original manuscript significantly.

Data generated or analyzed during this study are provided in full within the published article and its supplementary materials. Funding agencies are listed in the acknowledgements.

Alexander Kawohl: conceptualization, data curation, formal analysis, investigation, methodology, visualization, writing – original draft; Hartwig Frimmel: data curation, funding acquisition, supervision, writing – review & editing; Wesley Whymark: investigation, resources, writing – review & editing; Leo Millonig: formal analysis, investigation, methodology, writing – review & editing; Axel Gerdes: formal analysis, methodology, writing – review & editing.

Supplementary data are available with the article at https://doi.org/10.1139/cjes-2022-0030.