The Proterozoic assembly of Australia, the understanding of which is critical for reconstructing Proterozoic supercontinents, involved amalgamation of the West Australian (WAC), North Australian (NAC), and South Australian cratons (SAC). However, the basement between these Archean to early Proterozoic lithospheric blocks is mostly buried beneath younger basins; hence, its composition and age and the timing of Proterozoic assembly remain uncertain. In situ zircon U-Pb-O-Hf analyses of igneous rocks from drillholes that intersected basement beneath the northwestern Canning Basin reveal the presence of a substantial domain of juvenile Proterozoic lithosphere, the Percival Lakes province, between the WAC and NAC. Although isotopically distinct from the neighboring WAC and NAC, the Percival Lakes province is strikingly similar to other juvenile Proterozoic tectonic elements between the WAC, NAC, and SAC. Combining isotope and seismic data, we interpret the Percival Lakes province as part of an ~1700 × 400 km Proterozoic lithospheric domain that lacks evidence of Archean provenance but consists mainly of reworked remnants of Mesoproterozoic oceanic crust that survived WAC-NAC-SAC convergence. The apparent absence of Archean lithosphere between the cratons implies they never directly collided or that complete collision was prevented by impingement of three-dimensional promontories in the converging lithospheric blocks. Instead, the Percival Lakes province and other Proterozoic elements between the WAC, NAC, and SAC consist of oceanic lithosphere extracted from Earth's mantle in the Proterozoic. Our results imply that WAC-NAC convergence was younger than Columbia amalgamation at ca. 1.8 Ga and that Proterozoic Australia formed during the earliest phases of Rodinia assembly at ca. 1.3 Ga.

The Proterozoic assembly of Australia involved three major cratons: the West Australian craton (WAC), North Australian craton (NAC), and South Australian craton (SAC), all of which are of key importance for Columbia, Rodinia, and Pangea supercontinent reconstructions (Cawood and Korsch, 2008; Johnson, 2021; Wang et al., 2021). However, the timing of amalgamation of Proterozoic Australia is uncertain, given that several models postulate events between ca. 1800 and 530 Ma that fail to provide robust constraints for supercontinent reconstructions (Cawood and Korsch, 2008; Martin et al., 2017; Gardiner et al., 2018; Johnson, 2021; Payne et al., 2021). This uncertainty is largely because Proterozoic orogenic belts that surround the Australian cratons are mostly hidden beneath basins (Fig. 1) that obscure the nature of basement rocks.

The intervening crust between the WAC and SAC is covered by Neoproterozoic to Cenozoic basins (Fig. 1). Analyses of rocks recovered from deep stratigraphic drilling characterized the underlying Madura and Coompana provinces as reworked Proterozoic oceanic and island-arc crust, revealing that continental collision did not occur between the WAC and SAC (Kirkland et al., 2015; Spaggiari et al., 2018). Isotopic evolution of the Madura and Coompana provinces is comparable to that of the Mesoproterozoic Musgrave province exposed at the junction of all three cratons. The basement of these three provinces was interpreted to be juvenile crust formed in the 1.95–1.9 Ga Mirning Ocean (Kirkland et al., 2015, 2017). Farther north, however, the age and composition of the intervening crust between the WAC and NAC are more enigmatic. There, a distinct tectonic domain imaged in seismic data and referred to as the Percival Lakes province (PLP) is entirely covered by the Phanerozoic Canning Basin (Zhao et al., 2022) and represents the most significant missing link in understanding the assembly of Proterozoic Australia.

Igneous rocks from the few drillcores that intersected the PLP provide a valuable opportunity for constraining its composition, age, and crustal architecture. Zircon U-Pb-O-Hf data from samples of these cores reveal, for the first time, that continuous Proterozoic lithosphere is present between the WAC, NAC, and SAC. Therefore, these cratons did not completely collide during Rodinia formation, and a significant part of the continental lithosphere in the PLP and other Proterozoic elements between the WAC, NAC, and SAC was likely generated by extraction from the mantle in the mid-Proterozoic (ca. 1.9–1.3 Ga) rather than by reworking of Archean lithosphere. Our results indicate that WAC-NAC convergence was younger than Columbia amalgamation and that Proterozoic Australia formed during the earliest phases of Rodinia assembly.

To understand the age, composition, and crustal architecture of the PLP, we analyzed the O-Hf isotope composition of U-Pb–dated zircons in six igneous samples (five felsic and one mafic) recovered from six drillholes as deep as ~2.8 km (drilled by various petroleum and mineral exploration companies between 1958 and 2012 CE; Fig. 1; Table 1). Two additional samples were collected from granitic rocks of the 654–603 Ma O'Callaghans Supersuite that crop out in the Neoproterozoic Yeneena Basin on the southwestern margin of the Canning Basin (Fig. 1). This new data set includes 145 oxygen isotope analyses and 160 Lu-Hf analyses (Tables S1 and S2 in the Supplemental Material1). We synthesized all available O and Hf data from ~730 samples in Western Australia, including ~3000 O and ~12,900 Lu-Hf analyses (Fig. 2). The new U-Pb-O-Hf results are summarized in Table 1; analytical methods and detailed descriptions of results are presented in the Supplemental Material.

The Lu-Hf isotopic compositions of most magmatic zircons in the PLP fall within the crustal evolution array of the Musgrave province, its closest neighboring Proterozoic terrane (Fig. 2A). Both regions have Hf isotope arrays distinct from those of the WAC and NAC (Fig. 2A). The Archean Pilbara and Yilgarn cratons of the WAC, although partially reworked in the marginal Proterozoic orogens (Kirkland et al., 2015; Johnson et al., 2017; Gardiner et al., 2018), are unlikely to have been major sources of Neoproterozoic magmatism in the PLP given that the evolution array of the Archean rocks indicates ɛHf(i) < −25 (i—initial) at <700 Ma, significantly less radiogenic than results for the PLP (−10 to +3; Fig. 2A). The NAC, which is currently separated from the PLP by a northwest-trending suture (Zhao et al., 2022), yields older two-stage depleted mantle model ages (Tdm2, median = 2.5−2.1 Ga) than in both the Musgrave province and the PLP (median Tdm2 = 1.9−1.3 Ga), indicating that the NAC is an unsuitable source for most PLP zircons (Fig. 2A). The distinction is best illustrated spatially by the zircon Lu-Hf isotope map, which highlights similar median Tdm2 (1.93–1.30 Ga) for the PLP and the Musgrave, Madura, and Coompana provinces, whereas the WAC and NAC are dominated by older median Tdm2 of 3.7–2.1 Ga (Fig. 3; Fig. S3).

Our new Lu-Hf data support the interpretations of previous seismic studies in the region (Fig. 3; Fig. S3) that the PLP is structurally and compositionally distinct from the adjacent cratons. For example, the 870-km-long 18GA-KB1 deep seismic reflection line deployed by Geoscience Australia, which extends from the eastern Pilbara craton to the Aileron province, imaged a province beneath the Canning Basin with seismic characteristics distinct from those of the WAC and NAC (Doublier et al., 2020). Modeling of passive seismic data acquired along the coastal margin of the Canning Basin (the 900-km-long 19CWAS line, surveyed by Chinese Academy of Sciences, northwest of active line 18GA-KB1) identified a compositionally distinct segment of lithosphere separated from the WAC and NAC by two southeast-trending sutures (Zhao et al., 2022). Combining information from new zircon Lu-Hf isotope data with existing seismic surveys, we interpret the PLP to include the entire basement to the central and western Canning Basin, which, together with the Musgrave, Madura, and Coompana provinces forms a continuous, 1700-km-long section of mid-Proterozoic (ca. 1.9–1.3 Ga) lithosphere between the WAC, NAC, and SAC (Fig. 3).

Isotope data for magmatic zircons and most xenocrystic zircons from felsic igneous rocks in the PLP fall entirely within the evolution array of the Musgrave province (Fig. 2A). The six zircons below the Musgrave array from sample 214946 (Fig. 2A) are likely xenocrystic grains assimilated from metasedimentary rock because they have similar Th/U and ages to zircons from wall-rock psammitic gneiss (Fig. S1) and hence do not reflect the composition of the PLP. We interpret the 655–505 Ma granitic rocks in the PLP to have been derived through reworking of Musgrave-like crustal basement.

The Musgrave, Madura, and Coompana provinces record juvenile crustal formation at ca. 1.6 and 1.4 Ga by construction of successive oceanic arcs (Fig. 2A; Kirkland et al., 2017). The similarity of Hf isotope data between the PLP and the other three provinces indicates that the PLP was originally Mesoproterozoic oceanic lithosphere. The least-radiogenic magmatic zircons and six xenocrystic zircons from the PLP within the Musgrave array lie along a Hf array that intersects model depleted mantle at 1.93–1.90 Ga, similar to data for the Musgrave, Madura, and Coompana provinces for which cryptic juvenile mantle input was inferred at 1.95–1.90 Ga and interpreted to reflect formation of oceanic lithosphere (Kirkland et al., 2015, 2017). The differentiated remnant or felsic crust formed by subduction of the putative Mirning Ocean crust might have contributed to PLP, explaining the trend to more negative zircon εHf(i) values (Fig. 2A).

Zircons in granitic rocks in the PLP are dominated by high δ18O (6.1‰–9.3‰; Fig. 2B; Fig. S2), indicating the sources of these granitic rocks have undergone chemical weathering and O isotope exchange at low temperatures (<300 °C) prior to melting (Valley et al., 2005; Troch et al., 2020). This high-δ18O feature indicates the PLP must have been exposed at or near Earth's surface prior to formation of the 655–505 Ma granitic rocks.

A 10-cm-wide metagabbro vein dated at ca. 826 Ma represents the youngest juvenile mantle input into the PLP (sample 199495; Fig. 2A). Its sub-mantle-to mantle-like oxygen isotope composition suggests derivation from mantle-derived mafic magmas that assimilated rocks previously hydrothermally altered at high temperature (>300 °C), as is commonly observed in rift settings (Valley et al., 2005; Troch et al., 2020). This interpretation is consistent with intrusion of the gabbro being coincident with initial breakup of Rodinia ca. 825–800 Ma and the onset of deposition in the Centralian Super-basin (Wingate et al., 1998). However, the ca. 826 Ma gabbro is unlikely to have been a major source component for the more widely distributed 655–505 Ma granitic rocks (Fig. 1).

The precise timing of final assembly of Proterozoic Australia is controversial. For example, although the Paterson-Petermann orogen between the WAC and NAC has been considered intracratonic (Cawood and Korsch, 2008; Maidment, 2017), some argue that this orogen was driven by subduction from ca. 680 to 530 Ma, with suturing of the NAC and WAC at 550–530 Ma, based on the interpretation that ca. 630 Ma granitic rocks in the Paterson orogen are subduction related (Martin et al., 2017). However, our new zircon Hf-O isotope data do not support the hypothesis of juvenile oceanic arc magmatism at 650–500 Ma, which would be expected to have produced granitic rocks with predominantly superchondritic zircon Hf (Pastor-Galán et al., 2021). Although Lu-Hf data alone do not preclude continental arc magmatism, the assumed 650–500 Ma arc magmatism is inconsistent with continuous sedimentation from ca. 870 to 530 Ma in the intracratonic Centralian Superbasin (Haines and Allen, 2017) and the widespread emplacement of the ca. 1075 Ma Warakurna large igneous province (Wingate et al., 2004). We propose reworking of a Mesoproterozoic Musgrave-like crustal substrate in an intracontinental setting produced granitic rocks at 650–500 Ma during the intracratonic Paterson orogeny (Fig. 2A). The Lu-Hf isotope map clearly reveals the presence of a continuous Proterozoic lithosphere extending from the PLP to the Musgrave, Madura, and Coompana provinces, where arc magmatism started at least from ca. 1.6 Ga (Fig. 3). Therefore, assembly of Proterozoic Australia must have occurred after ca. 1.6 Ga.

The WAC and NAC were certainly amalgamated prior to emplacement of the ca. 1075 Ma Warakurna large igneous province, which extends across both cratons (Wingate et al., 2004). High-pressure metamorphism and magmatism during the 1377–1275 Ma Parnngurr orogeny in the Rudall province was contemporaneous with tectonism and magmatism of the 1330–1260 Ma stage I of the Albany-Fraser orogeny and the 1345–1293 Ma Mount West orogeny in the Musgrave province (Johnson, 2021; Payne et al., 2021). These synchronous ca. 1.3 Ga orogenic events surrounding Proterozoic lithosphere suggest terminal WAC-NAC-SAC assembly by ca. 1.3 Ga (Fig. 3). These data indicate that the accreted arcs and oceanic crust associated with the Mesoproterozoic ocean were reworked into the <700 Ma granitic rocks in the PLP, consistent with their Lu-Hf isotope compositions (Fig. 2A). We interpret convergence of the WAC and NAC to have commenced by consumption of the putative Mirning Ocean and terminated by ca. 1.3 Ga, with the present configuration broadly established prior to ca. 1075 Ma. If so, WAC-NAC convergence was younger than the 2.1–1.8 Ga amalgamation of Columbia (Zhao et al., 2002), and Proterozoic Australia formed at ca. 1.3 Ga during the earliest phases of Rodinia amalgamation. Moreover, the formation of Proterozoic Australia may not have been linked directly to Rodinia amalgamation but was related to convergence of small continental blocks into larger ones, which may be the megacontinent precursor stage to Rodinia (Wang et al., 2021).

The lack of inherited Archean zircons and model ages from the PLP as well as its seismic structures, which are distinct from neighboring Archean cratons, suggest the absence of Archean lithosphere in the region. Our results indicate that large tracts of Proterozoic continental lithosphere may not necessarily have required reworking of Archean lithosphere (cf. Begg et al., 2009) but could have been extracted from Earth's mantle in the Proterozoic and preserved during craton amalgamation. The Lu-Hf isotope map implies that >20% of Western Australia (including the Percival Lakes, Musgrave, Madura, and Coompana provinces) consists of Proterozoic lithosphere and that the WAC, NAC, and SAC never came into direct contact during assembly of Proterozoic Australia (Fig. 3).

The reason for this failed collision is unknown. One possibility for apparent failed, or “soft”, collision is that complete collision of entire lithospheric sections of thick Archean blocks might have been prevented by mutual impingement of cratonic keels that were laterally more extensive than the overlying crust, as has been proposed to explain the Mesoproterozoic geodynamic evolution of the Musgrave province above the triple junction of the WAC, NAC, and SAC lithospheres (Smithies et al., 2011; Spaggiari et al., 2018).

We are grateful to Brendan Murphy, Erin Martin, and Chris Spencer for constructive reviews and to Rob Strachan for editorial handling. Y. Lu, M. Wingate, R. Smithies, K. Gessner, S. Johnson, D. Kelsey, P. Haines, and D. Martin publish with the permission of the Executive Director, Geological Survey of Western Australia. Zircon U-Pb analyses were conducted at the John de Laeter Centre, Curtin University (Perth, Australia), using facilities operated with financial support of the Australian Research Council (ARC) and AuScope (Australian National Collaborative Research Infrastructure Strategy). Oxygen and Lu-Hf analyses employed the facilities and scientific and technical assistance of Microscopy Australia at the Centre for Microscopy, Characterization and Analysis, University of Western Australia, and were supported in part by ARC grants LE100100203 and LE150100013. M. Aleshin is thanked for assistance with oxygen analyses. We acknowledge funding from the Government of Western Australia Exploration Incentive Scheme.

1Supplemental Material. Detailed descriptions of methods and sample results, Figures S1–S3, and Tables S1 and S2. Please visit https://doi.org/10.1130/GEOL.S.20247603 to access the supplemental material, and contact [email protected] with any questions.