The basement of Mesoarchean to Neoarchean greenstone basins in the Yilgarn craton is composed of fragments of evolved crust up to 3.7 Ga old. New cratonwide geochemical and isotopic data with unparalleled spatial resolution image a NE- to ENE-trending architecture in pre–2.73 Ga crust. These trends cannot be reconciled with plate-tectonic models, as they persist across younger NNW-striking structural fabrics, including a proposed suture previously interpreted to result from exotic terrane accretion. Our results suggest that, in spite of their substantial strike length, the NNW-trending structures have limited horizontal displacement and, although important for understanding regional geology, may be a geodynamically insignificant overprint of the primary ENE-trending architecture. We propose that these greenstone provinces or belts include individual basins formed in rifts with location, size, and orientation influenced by the interaction between basement fragments and regional crustal extension.

The geological history of the Yilgarn craton in Western Australia dates back to at least 4.37 Ga (Wilde et al., 2001), with preserved crustal fragments as old as 3.7 Ga (e.g. Ivanic et al., 2013), and preserves structural and geochemical evidence interpreted as being consistent with subduction processes at several stages. Old (>3.0 Ga) crust of the Narryer terrane is in contact with 3.0–2.6 Ga crust of the Youanmi terrane (YT; Fig. 1) along an interpreted NE- to ENE-striking suture marking the ca. 2.74 Ga amalgamation of the two terranes (Myers, 1995). Between 2.95 and 2.73 Ga, there is also geochemical evidence for melting of metasomatically enriched mantle sources similar to those found at modern subduction zones (Wyman and Kerrich, 2012; Smithies et al., 2018; Lowrey et al., 2019; Mole et al., 2019). NNW-trending structures such as the Ida fault, which separates the Eastern Goldfields superterrane (EGST) from the YT, are also presented as an exemplar that modern-style plate-tectonic processes were active by ca. 2.7 Ga (Krapěz and Barley, 2008; Czarnota et al., 2010). The history of the craton at this stage is widely considered to reflect E-W convergence and accretion of exotic “terranes” separated by translithospheric structures, typical of modern convergent plate margins, with a prominent regional (NNW) lithotectonic trend (e.g., Krapěz and Barley, 2008; Czarnota et al., 2010). In contrast to this view, other studies have suggested that the EGST represents an infilled rift that developed over continuous YT basement (Pawley et al., 2012; Mole et al., 2019; Masurel et al., 2022).

Much of the Yilgarn craton lies under cover, so distinctions between crustal fragments have largely relied on interpretations of geophysical data, mainly gravity and aeromagnetics, which are biased toward imaging upper-crustal NNW trends. However, using a large geochemical and isotope (Nd and Hf) data set, we identified regional variations in the composition of granite source regions that map an earlier cratonwide NE- to ENE-trending lithospheric architecture that remains recognizable despite the overprinting NNW-striking structures.

Our cratonwide granite data set included 5157 whole-rock geochemical samples (Fig. 1B), 485 of which were dated mainly by secondary ion mass spectrometry (SIMS) zircon U-Pb geochronology, 830 of which have accompanying whole-rock Sm-Nd isotope data, and 302 of which have zircon Lu-Hf isotope data (see Tables S1–S3 and Figure S1 in the Supplementary Material1). All samples were classified according to the geochemical groups recognized by Champion and Sheraton (1997) and Champion and Cassidy (2007) (see Supplementary Material), as outlined below.

High-Ca granites are broadly tonalite-trondhjemite-granodiorite (TTG) series and include high-Sr/Y (>40) and low-Sr/Y (<40) types, each separated into strongly sodic (K2O/Na2O <0.6), sodic (K2O/Na2O = 0.6–1), and rare potassic (K2O/Na2O >1) types. These granites mainly reflect melting of hydrated mafic crust (amphibolite), although some probably also had a direct large ion lithophile element (LILE)–enriched mantle source component (Johnson et al., 2017; Smithies et al., 2019).

Low-Ca granites are typically potassic (K2O/Na2O mainly >1) and enriched in incompatible trace elements, reflecting a highly inhomogeneous crustal source. One subtype (low-Ca [high-Ti] granites) has elevated TiO2 and P2O5 concentrations and zircon saturation temperatures up to 960 °C (Watson and Harrison, 1983).

High field strength element (HFSE)–rich granites are commonly hornblende-bearing, Fe-rich granites lying on a trend consistent with a strongly fractionated tholeiitic (dominantly mantle-derived) magma.

Mafic granites are hornblende-rich monzodiorite, diorite, and granodiorite and are dominantly sanukitoids derived from a metasomatized peridotitic source component (Smithies et al., 2019).

Syenites are spatially and temporally closely associated with sanukitoids and have similar Sm-Nd isotope (ɛNd mostly 0 to +3) and trace-element compositions (Fig. S2), suggesting a genetic relationship (Smithies et al., 2023).

The Sm-Nd isotopic composition of most sanukitoids, syenites, HFSE-rich granites, and many strongly sodic high-Ca (high-Sr/Y) granites is more radiogenic (i.e., juvenile) than bulk earth, with initial ɛNd values up to +4, reflecting a direct mantle source component. Other granite groups are less radiogenic (mainly initial ɛNd <0; Fig. S2), reflecting melting of predominantly crustal sources.

Except for areas in the Narryer terrane and a few other localities in the northern YT, the crystallization ages for ~70% of samples lie between 2.75 Ga and 2.65 Ga (Fig. 1C), irrespective of granite classification. Thus, our granite data set is a reasonable cratonwide reflection of a ca. 2.7 Ga time slice. Spatial patterns in compositional variables show a range of apparent trends, some persisting over multiple geochemical and isotopic data sets. Trends that formed prior to ca. 2.7 Ga are understandably erratic and commonly masked by the dominant ca. 2.7 Ga NNW structural trends. Nevertheless, a series of NE-ENE trends is preserved in the spatial distribution of compositional proxies for the evolution of Archean felsic crust, including Nd- and Hf-isotopic ratios (e.g., Champion and Cassidy, 2007; Mole et al., 2019) and Sr/Y and Gd/Yb ratios (Fig. 2). These trends are independent of contouring or interpolation methods (Fig. S3). Differences in Sr/Y and Gd/Yb can be proxies for regional variations in composition and residual mineralogy (e.g., Moyen, 2009) of granite source regions during crustal evolution. In our data set, NE-ENE and strong NNW trends were both observed, but consideration solely of rocks older than ca. 2.7 Ga (Fig. 3) eliminated any suggestion of a NNW structural trend, while NE-ENE trends remained (Fig. 3C).

High-Sr/Y granites, and, in particular, strongly sodic high-Ca (high-Sr/Y) granites, are concentrated in the northern part of the craton and most notably in the YT, into an ENE-trending “southern high-Sr/Y zone” (SHZ; Fig. 2C). A similar pattern is seen on a map contoured for Gd/Yb (Fig. 2D). Statistical modeling of point alignments (Hammer, 2009) suggests that trend orientations for all samples with Sr/Y >40 or Gd/Yb >4.5 are not normally distributed but instead show a dominant NNW orientation, with a significant NE-ENE fabric associated with a shorter, discontinuous set of lineaments (Figs. 2C2D).

Regions of mainly low-Sr/Y and low-Gd/Yb granites tend to correspond with felsic crust with nonradiogenic (evolved) Sm-Nd and Lu-Hf isotope compositions, as shown by old two-stage depleted mantle model ages (TDM2; Fig. 2). The northwestern part of the YT is a discrete NE-ENE accumulation of strongly sodic high-Ca (high Sr/Y) granites (Fig. 2C), corresponding to the juvenile “Cue isotopic zone” imaged by isotope maps (Champion and Cassidy, 2007; Ivanic et al., 2013).

Strongly sodic high-Ca (high-Sr/Y) granites tend to be more closely associated spatially and temporally with greenstone belts than do other granites, except for sanukitoids and syenites (Figs. 1 and 2), most of which also have Sr/Y >40. None of these granite types form major components in granite bodies distal to greenstone belts, which are dominated by crustal melts of low-Ca, or of less sodic high-Ca, granite composition (Fig. 1B).

Low-Ca (high-Ti) granites are among the youngest granites (Fig. 1C). The main concentration forms a broad band within the western half of the YT that coincides with a region of felsic crust with TDM2 older than 3.3 Ga (Figs. 2A and 2B). This band of granite and old felsic basement is compartmentalized along NE-ENE trends. The granites themselves do not show NE-ENE fabrics, and so the spatial patterns in the granites probably mirror preexisting NE-ENE deformation trends and offsets in the distribution of their basement melt sources. Similar NE-ENE trends align with sharp gradients in Sm-Nd and Lu-Hf isotope maps and in regional granite geochemical data. In particular, they extend into the EGST as sharp NE-ENE–trending steps in the shape of the NNE-trending isotopic juvenile zone (Figs. 2A and 2B), orthogonal to the NNW-oriented structural grain. Mole et al. (2019) also noticed that the 2.7 Ga komatiite occurrences of the EGST appeared to be offset along similar NE-ENE–trending steps in the EGST isotope contours.

Most sanukitoids, syenites, and HFSE-rich granites intrude into or are peripheral to greenstone belts along fault systems, which, by implication, are mantle-tapping structures, and in many cases, these fault systems are the structural control on late sedimentary basins and potentially also on the early basalt magmas forming the greenstone basins themselves (Smithies et al., 2022). These mantle-derived granites form regionally extensive NNW-trending belts, several of which abruptly terminate, in the southern EGST, against the ENE continuation of the northern boundary of the SHZ, which in this region also corresponds with one of the main NE-ENE steps in isotope contour maps (Figs. 2A and 2B).

All isotope maps of the Yilgarn craton (Champion and Cassidy, 2007; Champion and Huston, 2016; Mole et al., 2019; Hartnady and Kirkland, 2022; Lu et al., 2022a, 2022b) identify the EGST as a region of anomalous juvenile felsic crust. Magmatism forming the greenstone belts and much of the granite of the EGST began after 2.73 Ga (e.g., Czarnota et al., 2010). Regional compositional data from older granites (Fig. 3) show no evidence for a distinct basement to the EGST, but they reveal broad ENE-trending compositional basement bands in Youanmi-aged crust across all but the westernmost part of the craton (Morris and Kirkland, 2014). Virtually the entire juvenile contribution to the felsic EGST crust was incorporated between 2.73 and 2.65 Ga into a basement isotopically similar to the YT. This contribution was largely via sanukitoid, syenite, and strongly sodic high-Ca (high-Sr/Y) magmatism in and around greenstone basins, and this alone accounts for the NNW-trending region of high-Sr/Y crust (Fig. 2). Low-Ca granites with εNd values >0 mainly lie close to EGST greenstones and probably reflect remelting of a “sanukitoid-infused” crustal source or of isotopically juvenile strongly sodic high-Ca (high-Sr/Y) granite. The strong spatial and temporal association of sanukitoid, syenite, and strongly sodic high-Ca (high-Sr/Y) magmatism with greenstone basins possibly reflects an overarching structural control on basin formation, and on intrusion pathways for both the mafic magmas forming the greenstone belts and for the later mantle lithosphere–derived sanukitoid and syenite. Associated high-Ca (high-Sr/Y) granites reflect either fractionated sanukitoid (Smithies et al., 2019) or deep crustal melts resulting from heat related to the ascent of greenstone-forming magmas, explaining why they are less common away from greenstone belts.

Juvenile crust in the NE-trending Cue isotopic zone (Fig. 2) evolved between 2.82 Ga and 2.76 Ga (Ivanic et al., 2022) and has been attributed to back-arc development before the ca. 2.74 Ga inferred amalgamation of the Narryer and Youanmi terranes (Rowe et al., 2022). It includes 2.82–2.76 Ga mafic-intermediate stratigraphy with compositions reflecting a metasomatically enriched peridotite source and a compositional range typical of modern subduction initiation assemblages (Lowrey et al., 2019; Smithies et al., 2018). The juvenile isotopic signature is marked by 2.76–2.69 Ga sanukitoid and strongly sodic high-Ca (high-Sr/Y) granites.

Although less distinct, the southern high-Sr/Y band also represents a NE-ENE accumulation of strongly sodic high-Ca (high-Sr/Y) granites and sanukitoids within a zone of slightly more juvenile felsic crust.

The NE-ENE compositional architecture reflects variations in proxies for both composition and melting conditions in granite source regions, implicating deep NE-ENE–striking compositional domains. The local preservation of the NE-ENE trends in spite of transpressional deformation along NNW-oriented structures (Zibra et al., 2022) suggests that, although the strike length of these later fault zones is up to hundreds of kilometers in scale, displacements were relatively small. The preservation of the NE-ENE compositional domaining across the Yilgarn craton (Fig. 2) cannot be reconciled with the idea that the craton amalgamated via lateral E-W accretion of exotic terranes. The prominent NNW structural and isotopic trends marking the EGST are more consistent with development in a rift zone as a series of discrete greenstone basins forming an upper stratigraphic component of the YT, and surrounded and separated by granite that was emplaced dominantly coeval with or after greenstone formation.

Based on the patterns in regional Lu-Hf isotope data sets, Mole et al. (2019) suggested that the Yilgarn craton basement is segmented into compositionally discrete blocks, a feature also demonstrated in our data. Although there is no evidence that the NE-ENE trends reflect structures that deform greenstone belts of the EGST, they clearly controlled the distribution of juvenile crust that evolved synchronously with the greenstones (Fig. 2). We suggest that the NE-ENE–trending architecture influenced the location, orientation, and evolution of both the EGST greenstone basins and the spatially related sanukitoid, syenite, and very sodic high-Ca (high-Sr/Y) magmas. The same is probably true of the NE-ENE trends that characterize the Cue isotopic zone and SHZ regions. Future work can test this hypothesis by investigating kinematic scenarios of basin formation that take into account progressive deformation and reactivation of the NE-ENE–trending architecture.

Large data sets of geologically well-constrained geochemical, age, and isotopic data have been used to investigate the deep crustal geology of other Archean cratons (Mole et al., 2021; Harris et al., 2021; Vandenburg et al., 2023). The point of such exercises is not necessarily to challenge the role of modern-style arc/terrane accretion processes, although this appears to be the outcome so far. Such integrated data sets provide a powerful means of testing whether geodynamic models constructed largely from surface and near-surface data are consistent with deep crustal architecture, and for understanding the extent and geological reasons for any inconsistencies that might be detected.

We show that, although once the basis of an E-W terrane accretion paradigm for the Yilgarn craton, the prominent NNW structural trends that dominate the EGST are instead more likely a late overprint on a continuous basement already characterized by an ancient NE-ENE–trending architecture, which itself was possibly at least partly a result of some form of local, incipient, subduction-like process. Greenstone belts in the EGST most likely developed in continental rift zones, and the evolution and locations of the individual basins were probably fundamentally influenced by early NE-ENE YT-basement architecture. The dominant NNW trends were not a direct or immediate product of exotic terrane accretion within the region of the Yilgarn craton, although they potentially, but not necessarily, might relate to plate-tectonics occurring elsewhere.

1Supplemental Material. Tables S1–S3 and Figures S1–S3. Please visit to access the supplemental material; contact with any questions.

Comments from Peter Cawood, two anonymous reviewers, and editor Urs Schaltegger were very helpful and are greatly appreciated. R.H. Smithies, K. Gessner, Y. Lu, J. Lowrey, T. Ivanic, J. Sapkota, and R. Quentin de Gromard publish with the permission of the Executive Director of the Geological Survey of Western Australia. We acknowledge funding from the Government of Western Australia Exploration Incentive Scheme.

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