A hidden Rodinian lithospheric keel beneath Zealandia, Earth's newly recognized continent

The Neoproterozoic amalgamation and subsequent breakup of the Rodinia supercontinent were significant events in Earth’s history. However, despite consensus that Rodinia was assembled at ca. 1300–900 Ma and rifted apart ca. 800–600 Ma, debate continues as to the internal configuration of continent-sized blocks (e.g., Li et al., 1995, 2008; Cawood et al., 2013). Central to the Rodinia debate is the location of cratonic blocks to the east of the Australia–East Antarctica margin, with arguments that the margin was adjacent to the western Canadian Laurentian margin (Dalziel, 1991; Moores, 1991) or the western United States Laurentian margin (Karlstrom et al., 1999). Alternatively, it is also hypothesized that the South China block was positioned between Australia–East Antarctica and Laurentia (Li et al., 1999).

Recently recognized as a distinct continent, Zealandia—of which 94% is currently underwater—formed following Late Cretaceous breakup of the Gondwana supercontinent (Mortimer et al., 2017). Prior to its separation, the basement rocks of continental Zealandia were created by multiple episodes of terrane accretion and arc-related magmatism along the paleo-Pacific Gondwana margin from the Cambrian to Early Cretaceous (Mortimer, 2004). No Precambrian rocks are exposed onshore in New Zealand. The oldest basement rocks of Zealandia are divided into two provinces: the early Paleozoic Western province, comprising metasedimentary rocks and Paleozoic–Mesozoic intrusions, and the late Paleozoic–Mesozoic Eastern province, a series of plutonic-metasedimentary terranes accreted to the Gondwanan margin (Fig. 1) (Mortimer, 2004). The Median batholith, a long-lived arc once part of the active Gondwanan margin, sutures the two provinces (Mortimer, 2004). The inferred eastern limit of Gondwanan Paleozoic upper- to mid-crustal metasedimentary rocks in Zealandia is well defined by linked major ductile shear zones, marking a major crustal boundary (Fig. 1B) (Allibone and Tulloch, 2004; Scott et al., 2011; Klepeis et al. 2019). Previous isotopic studies focused on this boundary demonstrate that Mesozoic Zealandia consisted of separate crustal blocks that are isotopically distinct (Schwartz et al., 2021). In this study, we present in situ O-Hf-U-Pb isotopic zircon data for Cambrian–Cretaceous plutonic rocks throughout Zealandia that enable us to determine the middle- to lower-crustal source(s) of these diverse isotopic domains through time. Our results reveal for the first time that continental Zealandia is underlain by a broad Precambrian lithospheric keel, which allows us to place Zealandia into the greater Rodinia supercontinent puzzle.

Zircon is the foremost deep-time recorder of Earth’s history, preserving a rich archive of isotope information that informs on magma parentage and crust-mantle evolution (Valley et al., 2005; Kemp et al., 2007). Unlike whole rocks, zircon is highly resistant to alteration and weathering, (Hoskin and Schaltegger, 2003), providing a robust record of the U-Pb age and O-Hf isotope composition of the melts from which it crystallized. Importantly, the δ¹⁸O and Lu-Hf isotope composition of zircon is particularly sensitive in evaluating the interaction between crust and mantle reservoirs (Valley et al., 2005; Kemp et al., 2007). Mantle-like zircon has an δ¹⁸O composition of +5.3‰ ± 0.8‰; deviation of zircon δ¹⁸O values above +6.1‰ reflects supracrustal recycling processes, whereas low zircon δ¹⁸O values (<+4.5‰) require a source that has undergone high-temperature alteration or melting of source rocks that have interacted with fractionated meteoric waters at high paleo-latitudes and/or paleo-elevations (Valley et al., 2005; Bindeman, 2008). The ε_Hf(t) composition of zircon can be used to assess the relative contributions of primitive (mantle-derived) and continental crustal sources to magma petrogenesis and to calculate episodes of crustal extraction (Hf model age = T_DM) from a depleted mantle source (Vervoort and Blichert-Toft, 1999; Kemp et al., 2007).

To test the evolving nature of Zealandia’s crustal architecture and the age and composition of the underlying lithosphere, we analyzed zircon from 169 samples that characterize the main episodes of plutonism in Zealandia’s Western province. This new data set includes >1500 oxygen-isotope analyses (using secondary ion mass spectrometry) and >3550 Lu-Hf-U-Pb analyses (collected simultaneously using split stream laser-ablation–inductively coupled plasma–mass spectrometry [U-Pb] and multi-collector–inductively coupled plasma–mass spectrometry [Lu-Hf]). A full description of sample preparation, analytical protocols, and results is provided in the Supplemental Material¹ and raw data for all zircon analyses from individual plutonic samples, sample IDs, and location information is provided in the Petlab database (https://pet.gns.cri.nz/; Strong et al., 2016). To ensure analysis of a single domain representative of magmatic crystallization, all analyses were targeted to a single location of the zircon, guided by cathodoluminescence images (Fig. S1 in the Supplemental Material).

Marked differences in δ¹⁸O in zircon from Zealandia plutonic rocks highlight distinct lithospheric domains (Figs. 1B and 2A). Almost all plutonic rocks emplaced east of the limit of Paleozoic metasedimentary rocks have consistently low δ¹⁸O values (Figs. 1B and 2A), with most within ±1‰ of the median δ¹⁸O value of +4.1‰ (a range from −8.1‰ to +8.9‰). Plutonic rocks are Carboniferous to Cretaceous, with whole-rock SiO₂ from 50 to 77 wt%. Low intrasample δ¹⁸O variability for most granitoid samples (Fig. 2A) supports isotopic homogenization in high-temperature melt-rich systems in the lower crust and/or upper mantle (Bindeman, 2008). To the west of the limit of Paleozoic metasedimentary rocks, Cambrian-Ordovician to Cretaceous plutonic rocks have δ¹⁸O values that range from −4.7‰ to +11.3‰ (median of +6.8‰; Figs. 1B and 2A) and whole-rock SiO₂ values between 47 and 78 wt%. I-type plutonic rocks from this western domain typically have mantle-like values (+5.3‰ ± 0.9‰) from melting of a sediment-modified mantle source and/or melting of subducted oceanic crust (i.e., Bolhar et al., 2008; Schwartz et al., 2021); conversely, S-type (peraluminous) and A-type (peralkaline) plutonic rocks have δ¹⁸O values >>+6.1‰, consistent with significant crustal recycling (Hiess et al., 2015). For all plutonic rocks, no correlation is observed between δ¹⁸O and fractionation indexes (i.e., SiO₂, Zr/Hf; Fig. S2), indicating that variability in O-Hf isotope compositions is not controlled by fractional crystallization. This new data set documents a widespread and internally homogeneous eastern isotope domain (EID) of plutonic rocks with anomalously low δ¹⁸O values that extends for >10,000 km² (Figs. 1B and 2A). This contrasts with plutonic rocks emplaced in the western isotope domain (WID), which have mantle and crustal δ¹⁸O values and almost no plutonic rocks with δ¹⁸O zircon values <4.5‰.

Insights into the source(s) for the low-δ¹⁸O EID come from considering zircon Hf isotope compositions. Plutonic rocks from the low-δ¹⁸O EID have more radiogenic ε_Hf(t) values (median ε_Hf(t) = +6.1) and are tightly clustered compared to those emplaced in the WID (median ε_Hf(t) = +1.9, broad range of values) (Fig. 2B). Coupled O-Hf zircon isotope compositions indicate that plutonic rock compositions from the low-δ¹⁸O EID were controlled by melting of a relatively isotopically homogeneous mafic lower-crustal source (radiogenic ε_Hf(t) values) that had experienced high-temperature hydrothermal alteration (responsible for the low-δ¹⁸O signature). Plutonic rock compositions from the WID are controlled by melting of a mafic lower-crust and/or mantle source mixed with variable amounts of a metasedimentary source.

Magmas and zircon with δ¹⁸O values lower than mantle are relatively rare in the geological record. Where present, they are primarily related to voluminous magmatism and elevated heat flux associated with hotspot and rift environments (Wang et al., 2011; Troch et al., 2020). Conversely, they are rarely reported from magmas emplaced within arc settings (Muñoz et al., 2012). Mesozoic plutonic rocks with low δ¹⁸O (<<+4.5‰) have previously been reported in Fiordland, New Zealand (Bolhar et al., 2008; Schwartz et al., 2021), and attributed to an underthrust low-δ¹⁸O source of unknown age. Our analysis of Cambrian-Ordovician and Carboniferous plutonic rocks with a low-δ¹⁸O signature demonstrates that a source for these rocks is at least Cambrian and likely older.

To assess the age of the lower-crustal source(s) in the EID, we calculated crustal residence ages (T_DM) (Fig. 3). A broad range of model ages is observed for low-δ¹⁸O plutonic rocks of the EID, with most T_DM ages between ca. 1300 and 500 Ma (Fig. 3). For WID plutonic rocks, T_DM ages range from >>2000 to 500 Ma (Fig. 3). We suggest that the T_DM age range and radiogenic ε_Hf(t) values for the EID are indicative of a primitive lithospheric mafic source produced by melting of the depleted mantle at different periods between 1300 and 500 Ma. The isotopically homogeneous low δ¹⁸O zircon values of the EID imply widespread high-temperature hydrothermal alteration of this primitive mafic source.

We propose that Phanerozoic plutonic rocks emplaced within the low-δ¹⁸O EID of Zealandia were produced by partial melting of a hydrothermally altered Precambrian lower-crustal mafic source. This accounts for calculated crustal residence ages between ca. 1300 and 500 Ma, radiogenic ε_Hf(t), and low δ¹⁸O zircon values (Fig. 2; Figs. S3 and S4). A three-stage process is evoked to explain the formation and subsequent alteration of the lower-crustal Precambrian source. In the first stage, melting of depleted mantle between ca. 1300 and 900 Ma produced mafic melts that ponded at the base of the crust. Magmatism during this period occurred along an active oceanic arc margin prior to final suturing and accretion of the Rodinia supercontinent (Fig. 4A) (Li et al., 2008). During the second stage, melting of depleted mantle between ca. 800 and 500 Ma produced additional mafic melts that also ponded in the lower crust. Mantle melting during this period was associated with Rodinian rifting events in response to a mantle superplume focused beneath Australia–East Antarctica and Western Laurentia (Fig. 4B) (Li et al., 1999, 2008). In the final stage, widespread hydrothermal alteration of the lower-crustal mafic material was synchronous with Rodinian rifting (800–500 Ma) due to high-temperature water-magma interaction during plume-driven magmatism (Wang et al., 2011). This would have been enough to impart the low-δ¹⁸O signature (Wang et al., 2011). However, the involvement of glacier-derived waters along Rodinia rifting zones contributing to the low-δ¹⁸O signature of the source cannot be ruled out (Zheng et al., 2004). In either case, the O-Hf isotope composition of Paleozoic–Mesozoic plutonic rocks in the low-δ¹⁸O EID is directly tied to melting of this hydrothermally altered low-δ¹⁸O mafic Rodinian keel.

The spatial extent of the Rodinian keel beyond the EID is difficult to establish. In the WID, any low-δ¹⁸O signature is obscured by magmas contaminated by Phanerozoic continental crust. The lack of Phanerozoic plutonic rocks with low δ¹⁸O values along the formerly contiguous Gondwana margin of southeastern Australia and Antarctica (Fig. 4C) (Kemp et al., 2007; Yakymchuk et al., 2013) suggests either that these segments of Gondwana are not underlain by the same Rodinian lithospheric keel as Zealandia or that that any low-δ¹⁸O isotopic signal is also obscured by crustal contamination. Mantle xenoliths from the Waitaha domain (Fig. 4C) within Zealandia’s Eastern province have Re-Os melt extraction ages that support an underlying Paleoproterozoic cratonic mantle (McCoy-West et al., 2013) but not a crustal keel of Rodinian age as revealed in this study.

The continent of Zealandia was once adjacent to western Tasmania in the Cambrian (Münker and Crawford, 2000). Our data, for the first time, also permit a Precambrian correlation to Tasmania and consequently Australia–East Antarctica. We suggest that Zealandia was part of (or proximal to) the Proterozoic microcontinental block VanDieland (which includes Tasmania) (Fioretti et al., 2005; Li et al., 2008; Cayley, 2011); the location of Zealandia in this context has implications for the position of South China (Fig. 4). In fact, the position of the South China block within Rodinia is controversial, with two main models proposed: (1) the South China block occupied an external position along a convergent margin adjacent to Western Australia and northern India (Karlstrom et al., 1999; Wang et al., 2017; Cawood et al., 2018), or (2) the South China block was located in the center of Rodinia between eastern Australia and Laurentia (Li et al., 1995, 1999, 2008). One of the pieces of evidence linking the South China block with northwestern India is the correlation of diverse Precambrian rocks from both blocks with anomalously low δ¹⁸O values and the lack of any rocks with low δ¹⁸O values in Australia–East Antarctica (Wang et al., 2017). The low-δ¹⁸O EID in Zealandia, inherited from a low-δ¹⁸O Precambrian mafic source, provides an alternative solution (Fig. 4) and a possible link to the South China block. Along with the VanDieland microcontinent, Zealandia may be the “linkage” terrane between East Gondwana, the South China block, and Laurentia.

This research was funded by the New Zealand Government and a Marsden Fast-Start grant (GNS1701) to Turnbull and a U.S. National Science Foundation grant (EAR-1352021) to Schwartz. This study was enabled by AuScope (auscope.org.au), the Australian Government via the National Collaborative Research Infrastructure Strategy (NCRIS), and the Australian Research Council LIEF program (LE150100013). Discussions with Zheng-Xiang Li and constructive comments from two anonymous reviewers are much appreciated.