We report the discovery of one of the largest ancient (>3.0 Ga) crustal terranes on Earth. Granitoids with crystallization ages >3.0 Ga and/or Sm-Nd depleted mantle model ages ≥3.2 Ga define a ~1000 × 100 km belt on the western margin of the Rae craton, Canada, referred to herein as the Perry River terrane (PRT). Zircon U-Pb-Hf-O isotope and whole-rock geochemical data from granitoids show that the PRT is a predominantly juvenile 3.3–3.2 Ga terrane that was partially reworked by more evolved ca. 3.1 Ga magmatism. These findings call for a reassessment of the timing and extent of ancient continental growth on Earth. A global compilation of zircon Hf isotope data from 3.6 to 3.0 Ga igneous rocks reveals clusters of relatively juvenile (initial εHf −2 to +3) rocks at ca. 3.31 and ca. 3.23 Ga, which include samples from the PRT and 13 other terranes worldwide. Other global zircon data sets also document age peaks between 3.3 and 3.2 Ga, and a cluster of broadly chondritic initial εHf values around 3.2 Ga. The 3.3–3.2 Ga period may therefore have been a time of enhanced net continental growth on Earth, and the PRT is one of the largest terranes preserved from that time. Furthermore, zircon Hf isotope data from 3.3–3.1 Ga PRT granitoids and 3.5–3.0 Ga igneous rocks worldwide yield little evidence for parent magmas that interacted with or derived from pre–3.6 Ga continental material. Contrary to some continental-growth models, this latter observation suggests that the volume of continental crust established by 3.6 Ga was relatively small.

The timing, extent, and nature of continental growth prior to 3 Ga is poorly understood (Rollinson, 2017; Condie et al., 2018; Hawkesworth et al., 2019). The greatest obstacle in addressing these topics is the paucity of preserved ancient (>3 Ga) continental crust. To mitigate this issue, researchers have exploited proxy records, such as the Sm-Nd isotope composition of terrigenous sedimentary rocks (e.g., Garçon, 2021) and the U-Pb-Hf isotope composition of detrital zircon (e.g., Dhuime et al., 2012). These proxies arguably provide a more complete view of crustal evolution than the extant igneous rock record. They are limited, however, by a lack of information on the composition and petrogenesis of the source rocks and by an indirect and in some cases questionable link to continental volume (Korenaga, 2018). The reliability of the ancient detrital zircon record, in particular, is plagued by the effects of non-zero-aged Pb loss (Guitreau et al., 2012; Fisher and Vervoort, 2018). These limitations place a premium on the discovery and direct characterization of new ancient crustal terranes. Doing so would lead to an increase in the minimum amount of continental crust that was present before 3 Ga and to more robust constraints on pre–3 Ga continental growth.

Despite the considerable size of the Rae craton (Fig. 1A), the record of its ancient crust is virtually unscrutinized. Previous work has hinted at the presence of pre–3 Ga crust at the western (Henderson and Thériault, 1994; McNicoll et al., 2000; Davis et al., 2014) and southern (Hartlaub et al., 2005; Ashton et al., 2016) margins of the craton, and it has been speculated that this crust may constitute a single large terrane (Henderson and Thériault, 1994; Hartlaub et al., 2005). However, the true areal extent, age, and character of pre–3 Ga crust in the Rae craton remains unconstrained. We present new data for granitoid and orthogneiss rocks (“granitoids” hereafter) from the western Rae craton, including whole-rock Sm-Nd isotope data for 82 samples and whole-rock elemental plus zircon U-Pb-Hf-O isotope data for a subset of samples. These data and compilations of existing data are used to: (1) document one of the largest ancient crustal terranes on Earth; (2) characterize that terrane; and (3) discuss implications for global pre–3 Ga continental growth. All new and compiled data can be found in the Supplemental Material1, along with detailed descriptions of analytical methods, samples, and results.

This study focuses on the Rae craton in mainland Canada, the western and eastern boundaries of which are defined by the 2.0–1.9 Ga Taltson-Thelon orogen and the Snowbird tectonic zone, respectively (Fig. 1). The interior of the Rae craton is characterized by abundant 2.7–2.6 Ga granitoids and subordinate ca. 2.7 Ga greenstone belts (Sanborn-Barrie et al., 2014; Regis et al., 2017). Exposures of 3.4–3.0 Ga crust occur at the margins of the craton in the Queen Maud block, Taltson basement complex, Snowbird tectonic zone, and Southampton Island (Fig. 1). Archean crust in the Rae craton has been overprinted by widespread deformation, metamorphism, and magmatism related to both the 2.5–2.3 Ga Arrowsmith (Berman et al., 2013) and 1.9–1.8 Ga Trans-Hudson (Regis et al., 2021) orogenic events.

Zircon U-Pb crystallization ages and wholerock Sm-Nd depleted mantle model ages (TDM) from (meta)igneous rocks are used to characterize the extent of ancient crust in the Rae craton (Fig. 1). All samples with crystallization ages >3.0 Ga and for which Sm-Nd data are available yield TDM ≥3.2 Ga. Therefore, TDM ≥3.2 Ga are used to identify rocks that are or may have derived in part from crust that is >3.0 Ga in age. Sixteen samples with crystallization ages >3.0 Ga and 46 with TDM ≥3.2 Ga are distributed along the western and southwestern margins of the craton (Fig. 1). These samples appear to define a belt of ancient crust that stretches for >1000 km from central Canada to the Arctic coast. The existence of this ancient belt is strongly supported by along-strike similarities in the U-Pb age data. Granitoids from both the northern and southern portions of the belt yield crystallization ages of ca. 3.3, 3.2, and 3.1 Ga, and rare ca. 3.3 Ga inherited zircon (Fig. 1; Table 1). We refer to this belt as the Perry River terrane (PRT), after a major north-flowing river that empties into the Queen Maud Gulf (Fig. 1).

The full extent of the PRT is a challenge to evaluate. The western boundary of the terrane appears to lie within the Taltson-Thelon orogen (Fig. 1). In the Queen Maud block, the eastern boundary is clearly defined by a sharp transition from a mottled aeromagnetic anomaly pattern and abundant TDM ≥3.2 Ga in the west to a linear aeromagnetic anomaly pattern and TDM exclusively <3.2 Ga in the east (Fig. 1). South of the Queen Maud block, the eastern-northeastern boundary of the terrane may coincide with the Howard Lake shear zone and the northeastern extent of the Taltson basement complex (Ashton et al., 2016) (Fig. 1). The Howard Lake shear zone lies along strike of the well-defined eastern boundary in the Queen Maud block, and there is no evidence for ancient crust immediately east of it or northeast of the Taltson basement complex (Fig. 1). Furthermore, although deformation in the Howard Lake shear zone is Paleoproterozoic in age, it has been interpreted as a reactivated Meso-Neoarchean lithospheric boundary (Thiessen et al., 2022). In the Queen Maud block, the ancient crust appears to dominate an area of ~50,000 km2, whereas south of the block, numerous TDM <3.2 Ga indicate that it has been added to by a substantial amount of younger juvenile magmatism. Nevertheless, the spatial distribution of TDM ≥3.2 Ga and crystallization ages >3.0 Ga (Fig. 1) suggest that the PRT once occupied an area of as much as ~100,000 km2 or more. Based on this, its >1000 km length, and its minimum extent of ~50,000 km2, we conclude that the PRT is one of the largest remnants of ancient crust on Earth. For comparison, the areal extent of pre–3.0 Ga crust in the each of the well-studied Kaapvaal (South Africa) and Pilbara (Western Australia) cratons is ~60,000 km2 (Eglington and Armstrong, 2004; Kemp et al., 2015).

These observations pose a fundamental question: What is the relationship between the PRT and the dominantly Neoarchean crust in the interior of the Rae craton? Given that there is evidence of ancient crust on the eastern margin of the craton and that some ca. 2.7–2.6 Ga granitoids within the craton contain >3.0 Ga inherited zircon (Fig. 1), the terrane may have been part of an even larger ancient crustal nucleus upon which Neoarchean Rae craton crust formed. Such a scenario is consistent with the inference that Neoarchean greenstone belts in the interior of the craton formed during the extension of a cryptic, older crustal substrate (Sanborn-Barrie et al., 2014).

The PRT represents a rare, but critical, record of continental crust formation and evolution on the early Earth. Most 3.3–3.1 Ga granitoids from the terrane are similar in composition to typical Archean tonalite-trondhjemite-granodiorite, but some mafic (<53 wt% SiO2) and potassic granitoids are also present (Figs. 2A and 2B). On average, the ca. 3.1 Ga granitoids are more evolved (higher K2O/Na2O) than the 3.3–3.2 Ga granitoids (Figs. 2A and 2B). Granitoids spanning a range of crystallization ages yield zircon δ18O values that are almost exclusively within the mantle field (Fig. 2C), indicating that little or no supracrustal rocks altered by surface water were involved in their genesis. By contrast, weighted mean initial eHf values vary with granitoid age and are discussed in detail below.

Zircon Hf isotope data were acquired for five 3.3–3.2 Ga and four ca. 3.1 Ga granitoids (Table 1). The older group includes ca. 3.31 Ga and ca. 3.24–3.21 Ga sodic tonalites with initial eHf values of +0.5 to +2.7 and a ca. 3.22 Ga mafic rock with an initial eHf value of +0.6 (Fig. 3A). The former are interpreted to have derived from partial melting of mafic rocks with relatively short crustal residence times, whereas the latter may be a direct mantle melt. The variable initial eHf values from the ca. 3.24–3.21 Ga granitoids (Fig. 3A) probably reflects mixing between juvenile and older, more evolved, crustal components. This interpretation is consistent with the presence of rare ca. 3.3 Ga inherited zircon (Table 1) and the Hf isotope composition of the ca. 3.31 Ga tonalite, which would have been sub-chondritic by 3.24 Ga (Fig. 3A). The younger group includes three samples of ca. 3.13 Ga sodic tonalite-granodiorite with initial eHf values of ~0 and a ca. 3.07 Ga granitoid (no whole-rock elemental data) with an initial eHf value of -1.5 (Fig. 3A). These granitoids likely derived from partial melting of broadly similar mafic crustal rocks that gave rise to the ca. 3.3–3.2 Ga tonalites (Fig. 3A), although the presence of inherited zircon in two of them (Table 1) indicates that at least some felsic-intermediate crustal material was also involved as a magma source or contaminant.

The geochemical and isotopic data summarized above are altogether consistent with the PRT being a relatively juvenile 3.3–3.2 Ga terrane that was partially reworked to produce more evolved ca. 3.1 Ga granitoid melts. The extent to which the PRT granitoids are juvenile depends in part on the isotopic composition of the mantle reservoir from which they formed. There is significant evidence that, on a global scale, juvenile Eo- to Mesoarchean magmas were extracted primarily from mantle reservoirs with chondritic to only moderately radiogenic Hf isotope compositions (Guitreau et al., 2012; Fisher and Vervoort, 2018; Kemp et al., 2023), broadly similar to the 3.8 Ga depleted mantle model (Fig. 3A). Considering this, and that there is no direct evidence (igneous or inherited zircon) for crust older than ca. 3.3 Ga in the PRT (Table 1), we suggest that the mafic source rocks to the sodic 3.3–3.1 Ga PRT granitoids are unlikely to have been extracted from the mantle earlier than ca. 3.45–3.23 Ga (Fig. 3A).

The discovery of an ancient terrane, similar or greater in size to some of the best-studied pre–3.0 Ga crustal blocks, warrants a reassessment of the timing and extent of continental growth on the early Earth. Therefore, we integrate our new data from the PRT with a global compilation of initial zircon eHf values from 3.6 to 3.0 Ga igneous rocks (Fig. 3A). Several studies have highlighted the importance of using the comparably robust igneous zircon Hf isotope record, instead of or in addition to detrital zircon Hf isotope data, to constrain ancient continental growth (Guitreau et al., 2012; Vervoort and Kemp, 2016; Fisher and Vervoort, 2018). Two important observations are discussed.

First, the compilation reveals clusters of relatively juvenile (initial εHf -2 to +3) igneous rocks at ca. 3.45, 3.31, and 3.23 Ga. The ca. 3.45 Ga cluster is dominated by samples from the Pilbara and Kaapvaal cratons. By contrast, when considered together, the ca. 3.31 and 3.23 Ga modes comprise samples from the PRT and 13 other cratons or crustal blocks (Fig. 3A; Fig. S4 in the Supplemental Material). Similar patterns emerge from other global zircon data sets. A detrital zircon U-Pb-Hf isotope compilation reveals a cluster of broadly chondritic initial eHf values at ca. 3.2 Ga (Fig. 3B). Furthermore, two igneous zircon U-Pb age compilations document density peaks between 3.3 and 3.2 Ga (Fig. 3C) that mirror the ca. 3.31 and 3.23 Ga modes in the igneous zircon U-Pb-Hf isotope data set (Fig. 3A). These findings suggest that ca. 3.3–3.2 Ga was a period of increased net continental growth, and the PRT may be one of the largest terranes created during that period. This conclusion is independent of whether that net continental growth reflects a period of enhanced continental crust production (Condie et al., 2018) or a period of enhanced continental crust preservation (Hawkesworth et al., 2019).

The second observation pertains to how much continental crust existed on the early Earth. Detrital zircon U-Pb-Hf isotope data have been used to argue that ~40% of presentday continental crust volume was established by 3.6 Ga (Dhuime et al., 2012; Hawkesworth et al., 2019), although the methods that were used to convert those data into a measure of crustal volume have been questioned (Korenaga, 2018). If such a large amount of continental crust was present by 3.6 Ga, then on a global scale, it is reasonable to infer that a significant fraction of igneous rocks emplaced from 3.5 to 3.0 Ga should bear the isotopic fingerprint of that voluminous pre–3.6 Ga crust. This is not the case. Rather, the overwhelming majority of 3.5–3.0 Ga igneous rocks worldwide yield initial eHf values well above the modeled evolution of 3.6 Ga felsic crust and above the modeled evolution of 3.6 Ga mafic crust (Fig. 3A). An important consideration is that ancient continental crust may have been more mafic, on average, than modern continental crust (Tang et al., 2016; Hawkesworth and Jaupart, 2021; cf. Garçon, 2021), which would make its interaction with younger magmas less perceptible in Hf isotope data (Fig. 3A). However, the Hf budget of continental crust is strongly weighted toward felsic rocks. For instance, mass balance calculations indicate that if ~70% of ancient continental crust were mafic (<57 wt% SiO2; Hawkesworth and Jaupart, 2021), ~49% of the Hf in the total continental crust would still reside in the felsic component (Table S6 in the Supplemental Material), and that felsic component would evolve to very negative eHf values with time (Fig. 3A). Yet there is a striking paucity of evidence for reworking of pre–3.6 Ga felsic crust in the 3.5–3.0 Ga igneous zircon Hf isotope record (Fig. 3A). Furthermore, the samples that do have initial εHf values consistent with reworking of pre–3.6 Ga crust are largely from the Acasta gneiss complex (Northwest Territories, Canada) (Fig. 3A), a tiny area that has been sampled extensively due to the known presence of Hadean to Eoarchean rocks. These points, along with other evaluations of the zircon Hf isotope (Kemp et al., 2015; Fisher and Vervoort, 2018), detrital zircon U-Pb age (Parman, 2015), and rock (Rollinson, 2017) records, suggest that the volume of continental crust established by the end of the Eoarchean was relatively small.

1Supplemental Material. New and compiled data, along with detailed descriptions of analytical methods, samples, and results. Please visit https://doi.org/10.1130/GEOL.S.22335541 to access the supplemental material, and contact editing@geosociety.org with any questions.

Funding and logistical support was provided by Natural Sciences and Engineering Research Council of Canada grants to Chacko and Heaman, the Northwest Territories Geological Survey, the Polar Continental Shelf Program, and the Geo-mapping for Energy and Minerals program. We are grateful to Kent Condie, David Mole, and Chris Hawkesworth for constructive reviews that greatly improved the manuscript; Urs Schaltegger for efficient editorial handling; Rebecca Canam for field assistance; Robert Dokken for preparing zircon grain mounts; and the Geological Survey of Canada for archival material of J.B. Henderson and V. McNicoll.

Gold Open Access: This paper is published under the terms of the CC-BY license.