Zealandia is a largely submerged, continental fragment in the southwest Pacific, generally considered to be derived from East Gondwana, but whose origins, age, structure, and relationships with other continental masses are poorly known. To explore the development of this microcontinent, a suite of mantle xenoliths was assembled from 12 localities throughout New Zealand, an emergent part of Zealandia. The 187Re-188Os isotopic systematics of the xenoliths yield model ages (TRD2) between 0 and 2.3 Ga. Six samples from the newly defined Waitaha domain, South Island, have a narrow range of TRD2 ages from 1.6 to 1.9 Ga, in agreement with an aluminochron model age for this mantle domain of ca. 1.95 Ga, and with a three-point Re-Os isochron age of 2.26 ± 0.10 Ga. These ages are >500 m.y. older than TRD2 ages preserved in other regions of mantle lithosphere from the eastern margin of Gondwana (e.g., southeastern Australia and Marie Byrd Land, Antarctica) and >1 b.y. older than the oldest crustal rocks exposed in New Zealand. Thus, the lithospheric mantle of Zealandia has a complex age structure, including a region of Paleoproterozoic cratonic mantle with a minimum extent of ∼45,000 km2. This ancient mantle resided at the margins of several supercontinents during the past ∼2 b.y., attesting to the durability of subcontinental lithospheric mantle domains, even when decoupled from overlying contemporaneous crust and in an oceanic setting distanced from stable cratonic nuclei.
The formation, preservation, and destruction of the lithospheric mantle and its role in craton growth and stabilization is a crucial, yet still poorly known component of the crust formation process. For more than 20 years, the Re-Os isotopic system has provided a useful tool for constraining the timing of lithosphere formation in continental settings. Early studies (e.g., Walker et al., 1989) showed that the ages of subcontinental lithospheric mantle (SCLM) underlying ancient cratons are equal to the oldest crustal ages, for the first time demonstrating the longevity of SCLM and implicating it in crust formation processes. Many subsequent studies have focused on determining the Re-Os characteristics of xenoliths from Precambrian continental regions. Here, we take an alternative approach to studying continent formation by examining the age structure of New Zealand, a young (<520 Ma) composite continental block that formed by stepwise accretion during the Phanerozoic.
Osmium data are presented for a suite of well-characterized mantle xenoliths from the SCLM throughout Zealandia emplaced over the past 85 m.y. Key questions addressed by this study focus on the comparison of the age structure of the SCLM with the oldest ages of the basement rocks. Is the underlying SCLM younger (e.g., tectonically replaced), much older (e.g., continental crust built on ancient mantle lithosphere), or the same age (e.g., felsic crust production and mantle lithosphere stabilization linked), and what are the implications of SCLM ages for plate reconstructions?
Zealandia, the broader New Zealand microcontinent, is a largely submerged (∼90%) continental ribbon (>3.5 × 106 km2) that formed on the eastern margin of Gondwana during the Phanerozoic (ca. 520–100 Ma). Prior to the fragmentation of Gondwana, proto–New Zealand was adjacent to Australia and Antarctica (e.g., Sutherland, 1999). Zealandia began to separate from Marie Byrd Land, West Antarctica, at the tip of the Chatham Rise at ca. 84 Ma (Eagles et al., 2004), and subsequently drifted ∼6000 km northwest. Present-day New Zealand comprises a complex collage of geological terranes, of which the Cambrian to Early Cretaceous basement is divided into two major provinces (Fig. 1), the older early Paleozoic Western Province and the younger (late Paleozoic to mid-Cretaceous) Eastern Province, which are separated by the Median batholith, a long-lived arc-root plutonic complex (Mortimer et al., 1999; Mortimer, 2004). The Dun Mountain–Matai terrane (Fig. 1) forms an important lineament in New Zealand geology, constraining an oceanic arc ophiolite accretion event in the Middle Permian (ca. 280 Ma) at the eastern margin of Gondwana (Kimbrough et al., 1992).
The oldest U-Pb zircon age obtained from crustal rocks in New Zealand is 515 ± 7 Ma (T. Ireland, 2011, personal commun.) from the Cobb-Tunnel fault slice, Takaka terrane, Western Province. Despite intensive geochronological investigations, no Precambrian cratonic core or evidence for Precambrian ages from crustal xenoliths has been identified. Continental intraplate volcanism, unrelated to any plume-like feature, has occurred sporadically throughout Zealandia since the mid-Cretaceous (McCoy-West et al., 2010, and references therein), and has sampled the SCLM, providing entrained mantle xenoliths.
SAMPLES AND METHODS
Peridotite xenoliths are rare within the Cenozoic volcanic sequences of New Zealand. Here, through new fieldwork and sampling existing collections, a comprehensive suite of mantle xenoliths has been assembled from 12 localities, including the North Island, South Island, and Chatham Islands (Fig. 1). These xenoliths are exclusively from the spinel facies, range in size from 3 to 40 cm across, and exhibit predominantly protogranular textures with common kink-banding of olivine. Of 41 mantle xenoliths examined, lherzolites (n = 21) and harzburgites (n = 16) dominate, with rare dunites and wehrlites (n = 4). All xenoliths are from monogenetic vents, dikes, or small periphery flows of larger volcanoes that have been erupted over the past 85 m.y.
Rhenium and Os concentrations and Os isotopic compositions were determined for ∼1.5 g aliquots of sample powders at the University of Maryland using Carius tube digestion and isotope dilution protocols (e.g., Puchtel et al., 2009; Connolly et al., 2011). Detailed descriptions of the samples and methods, and complete data sets including whole-rock major element and mineral compositions, are presented in the GSA Data Repository1 (Tables DR1–DR9).
Xenoliths have olivine forsterite contents ranging from 88.9 to 92.2 (Table DR1) and spinel Cr-numbers generally between 10 and 50. Mineral compositions within a xenolith are homogeneous, plotting on the forsterite-rich side of the olivine-spinel mantle array (Arai, 1994) (Fig. DR1). Whole-rock Al2O3 contents are generally lower than estimates for primitive upper mantle (4.2 wt% Al2O3; Becker et al., 2006) and vary between 0.06 and 3.43 wt%, with the majority of samples having <2 wt% Al2O3 (Fig. 2). Measured 187Os/188Os for 27 xenoliths are from 0.115 to 0.133 (Table DR1), and Os concentrations range from ∼1 to 7 ppb, with the exception of two samples with low Os contents of only ∼0.020 ppb (e.g., P43153b and P80291). Rhenium contents are generally low (<0.05 ppb), although some samples have up to ∼0.3 ppb Re (e.g., MSI79C).
Re-Os Melt Extraction Ages
Model ages (Table DR1) for each sample were calculated using three approaches: TMA ages, TRD ages (Walker et al., 1989), and a modification of the standard Re depletion ages based on a two-stage model (TRD2) ages. TMA and TRD2 represent maximum and minimum ages, respectively. For TRD2 age calculations, the initial 187Os/188Os isotopic composition is first calculated for the time of the host magma eruption, using the measured 187Re/188Os, and then projected to intersection with the mantle evolution curve assuming Re/Os = 0. TRD2 ages thus take into account any Re addition that may have occurred during the entrainment and eruptive process. In almost all cases here, there is no significant difference between TRD2 and TRD ages, but for samples with high Re/Os (e.g., P43153b), or with old (>0.5 Ga) emplacement ages, this correction can have a large effect (e.g., Shirey and Walker, 1998).
Eleven of the 27 samples have TRD2 ages from 0.83 to 0.36 Ga (Fig. 3), comparable to the oldest U-Pb zircon ages from overlying crustal rocks; ten samples have TRD2 ages from 1.32 to 2.32 Ga. On the basis of geographic proximity and Proterozoic Re-Os model ages, we group together five localities to define the Waitaha domain (Fig. 1). Six samples with the oldest TRD2 ages from the Waitaha domain yield a narrow age range from 1.6 to 1.9 Ga; additionally three fertile samples from the Trig L locality have TMA ages of 1.6–1.9 Ga (Table DR1). Samples from the adjacent Fortification Peak locality are excluded from the grouping based on irregular Re-Os systematics (Fig. DR2).
As Re may be susceptible to late mobilization, elements of similar compatibility (e.g., Al2O3) can be used as an index of melt depletion and combined with Os isotopic compositions to estimate melt depletion ages (e.g., “aluminochrons”) for suites of related rocks, constraining the formation age of the SCLM (e.g., Reisberg and Lorand, 1995). An aluminochron for the Waitaha domain gives an age of 1.94 ± 0.31 Ga (Fig. 2B). The 187Re-187Os data for three xenoliths from the Trig L locality yield an isochron age of 2.26 ± 0.10 Ga. As would be expected for a true isochron, rather than a fortuitous array, the 187Os/188Os isotopic composition determined by the intercept also yields a Paleoproterozoic TRD model age (1.95 Ga). An aluminochron for Trig L samples defines an age of ca. 2.10 Ga, consistent with the age defined by the Re-Os data.
The ages of the domain, as determined by the Re-Os isochron and the aluminochron, are indistinguishable. Considered together, with the minimum 1.6–1.9 Ga TRD2 ages of xenoliths from several localities, these ages indicate a widespread mantle melting event ∼2 b.y. ago. The Waitaha domain underlies ∼30% of the South Island, extending from Banks Peninsula to Otago Peninsula, with a conservative estimate of its size being ∼45,000 km2. Additionally, two low-Re/Os xenoliths from the Chatham Islands give model ages of ca. 1.0 Ga. This suggests that ancient SCLM could extend under a considerable portion (>200,000 km2) of Zealandia. In contrast, relative probability plots indicate a model age peak at ca. 0.5 Ga for the remaining New Zealand xenolith population (Fig. 3A), which is comparable to the oldest exposed crustal rocks of New Zealand (i.e., the Middle Cambrian, Takaka terrane, Western Province; Münker and Cooper, 1999). However, within the Waitaha domain, a much older age peak of ca. 1.7 Ga is dominant.
Distribution of Ancient Lithospheric Ages
A fundamental observation from the earliest applications of the Re-Os isotopic system is that model ages of lithospheric mantle underlying cratons are typically coeval with the oldest overlying continental basement (e.g., southern Africa, Siberia; Pearson et al., 2002). Subsequently, Os model ages for peridotite xenoliths that are much older than overlying crust have been reported in some off-cratonic regions (e.g., in southeastern Australia, Proterozoic mantle underlies Phanerozoic continental crust; Handler et al., 1997). Our results, however, differ substantially from other reported occurrences of older lithospheric mantle both in the relative uniformity of whole-rock ages from a single region, and in that this ancient lithospheric mantle is overlain by some of New Zealand’s youngest crustal basement comprising a Permian to Late Triassic metagraywacke sequence (e.g., Rakaia terrane; Fig. 1; Wandres et al., 2004). With an age disparity of >1.5 b.y. between the crust and the underlying mantle, this is the largest age decoupling yet observed.
Furthermore, in other localities with large age differences between the SCLM and crust, there is typically crust of similar age in proximal cratonic regions (e.g., Canadian Cordillera; Peslier et al., 2000). Thus, the occurrence of older lithospheric mantle has been attributed to processes of lithospheric thinning and displacement during rifting, with more ductile mantle pulled from beneath the crust, causing displacement of the mantle by tens to hundreds of kilometers (e.g., Handler et al., 1997). The occurrence of Paleoproterozoic mantle underlying eastern New Zealand is unique in that the closest occurrences of Paleoproterozoic crust are presently >3000 km away in cratonic Australia (e.g., Gawler Craton, South Australia). Although the convecting mantle has increasingly been shown to possess long-lived Os isotopic heterogeneities, particularly at the single-grain scale (e.g., Meibom et al., 2002), the old ages are generally outliers of unclear origin with no evidence for the persistence of ancient regional mantle domains having been documented.
Raft History of New Zealand’s Lithospheric Mantle
The ca. 2 Ga Waitaha domain is older than SCLM from other potentially correlative East Gondwanan regions (Fig. 3). For example, peridotite xenoliths from Marie Byrd Land and southeastern Australia have younger TRD2 age peaks at 1.1 and 1.0 Ga, respectively. Although ca. 1.8 Ga crustal basement rocks do occur in central Australia (in the Arunta Inlier; Collins and Shaw, 1995), it is problematic for these to be the complement of the Waitaha domain. This would firstly require early partial or total decoupling of this SCLM through unroofing of the mantle during extension-related lithospheric thinning, analogous to that seen at the Galicia margin, Iberian Peninsula (Boillot et al., 1989; Brun and Beslier, 1996). Secondly, the mantle domain would then need to be preserved at the continental margin during >400 m.y. of subduction and numerous accretionary orogenies that resulted in >1000 km of new crust formation in the intervening region. Alternatively, we speculate that the Waitaha domain is exotic to Gondwana and may have originated on the margins of a Paleoproterozoic continent during a period of extensive global crust production from 1.8 to 2.0 Ga (Condie et al., 2009), possibly being sourced from western North America where abundant juvenile crust formed at ca. 1.9 Ga (e.g., Hoffman, 1988). Evidence that the Waitaha domain was a late addition to the Gondwana margin and sourced from the east includes: (1) its position to the east of the Dun Mountain–Matai terrane (i.e., a major oceanic closure event at ca. 280 Ma), and (2) it underlies the last terranes accreted to East Gondwana.
Major age peaks at ca. 1.0 Ga for Gondwanan xenoliths (Fig. 3), including the TRD2 ages of the oldest Chatham Islands samples in this study, suggest a major episode of crustal formation at this time, now preserved in the SCLM throughout the southwest Pacific. The age peak of ca. 0.5 Ga of the remaining New Zealand xenoliths, and its similarity with the oldest basement ages, show that regions of the SCLM underlying Zealandia are consistent with the typical chronologically coupled crust and SCLM formation processes.
The long-lived persistence of the Waitaha domain provides credence to the idea that ancient microcontinental blocks (e.g., Hoffman, 1988; Griffin et al., 2011) are required to stabilize continental crust. The effects of cratonic lithosphere rafts, such as the Waitaha domain, on mantle dynamics deserves further investigation; for example, this rigid mantle block may have controlled the tectonic development of New Zealand and the propagation of the Australian-Pacific plate boundary through southern Zealandia, with the Alpine fault appearing to dogleg around the western border of the Waitaha mantle block (Fig. 1).
Low 187Os/188Os measured in whole-rock peridotite xenoliths sampling a large region of southern Zealandia demonstrate an ancient melt depletion event at ca. 2 Ga. The terrane of old lithospheric ages, here named the Waitaha domain, underlies a significant portion (≥45,000 km2) of the youngest region of New Zealand’s crustal basement and exhibits the largest age decoupling (>1.5 b.y.) yet observed between the crust and the underlying mantle lithosphere. This SCLM likely originated in a near-cratonic setting, at the margin of a Paleoproterozoic continent (possibly western North America), and has subsequently been separated from its original overlying crust for >1 b.y. while surviving through multiple supercontinent cycles. This finding extends our knowledge of the persistence of large domains of lithospheric mantle, providing an extreme example of a continental “life raft” (Carlson, 1995) as a long-lived, major tectonic feature and a mechanism through which new continental crust can become stabilized.
Analytical work at the University of Maryland was supported by U.S. National Science Foundation grant EAR-0946629 to Puchtel. We thank T. Ireland, N. Mortimer, J. Hermann, H. Campbell, C. Timm, I. Smith, and A. Raey for providing samples and/or discussions. B. Rapp and F. Brink provided analytical assistance. W. Collins, S. Aulbach, and an anonymous reviewer are thanked for their constructive comments.