A tectonic reconstruction model for Aotearoa-New Zealand from the mid-Late Cretaceous to the present day

ABSTRACT

We present a mid-Late Cretaceous to present day tectonic reconstruction model for Aotearoa-New Zealand. Our GPlates model comprises 50 rigid crustal blocks grouped into regions with common deformation histories set within a well-defined Australia-Pacific-Antarctica plate circuit tied to a published global paleomagnetic absolute reference frame. Within the model, four distinct periods of deformation are recognised from both near- and far-field observations. A key model assumption is the continuity of basement terranes between North and South Zealandia prior to Middle Eocene rifting and Late Oligocene initiation of transform motion. To complement the rigid crustal block model, continuously closing polygons show a ∼25% decrease in plate boundary area since the Middle Eocene that has been compensated for by corresponding increases in crustal thickness. A kinematic fault-propagation fold model demonstrates the plausibility of post-Late Oligocene asymmetric oroclinal folding on both sides of the evolving transform boundary. ‘Missing’ areas of map section common to previous tectonic reconstructions can be reconciled through contraction, elongation and vertical-axis rotation of continental crust within the deformation zone ahead of northward propagation of the Alpine Fault. This tectonic reconstruction model provides an open, accessible, and testable foundation for current and future paleogeographic and tectonic studies across Zealandia.

KEYWORDS:

• Tectonic reconstruction
• Zealandia
• Aotearoa-New Zealand plate boundary
• Plate boundary deformation
• rigid-plate model
• GPlates

Introduction

Plate tectonic reconstructions are key for understanding the evolution of plate boundary zones by providing a spatial and temporal context for geological and geophysical observations. Plate tectonic reconstructions help constrain the timing and mechanisms of subduction and continental break-up, the relationships between crustal and mantle processes, and provides the context for paleo-environmental interpretation along with the underlying mechanisms associated with plate boundary deformation, mountain building and basin formation (e.g. Seton et al. 2012; Reyners 2013; Müller et al. 2019; Hines et al. 2022; Strogen et al. 2022). In the Southwest Pacific, the continent of Zealandia has experienced a complex tectonic history since the Late Cretaceous, including the break-up of Gondwana, subduction initiation and reactivation, and formation of the Alpine Fault transform between the Australian and Pacific plates (King 2000; Sutherland et al. 2000; Schellart et al. 2006; Herzer et al. 2011; Lamb 2011; Bache et al. 2014; Matthews et al. 2015; Mortimer et al. 2017; Strogen et al. 2017; Sutherland et al. 2017). As a result of this deformation the lithosphere of Aotearoa-New Zealand (A-NZ) has accommodated significant strain since its formation.

From the breakup of Gondwana in the mid-Late Cretaceous to the Middle Eocene there is little documented evidence for plate boundary scale structures separating North and South Zealandia into independent continental fragments (e.g. King 2000; Cande and Stock 2004a; Matthews et al. 2015). The separation of North and South Zealandia into independent continental fragments began in the Middle Eocene with rifting and seafloor spreading between the Challenger and Campbell plateaux (Sutherland 1995). Displacement of this Eocene rift margin since the Late Oligocene (Sutherland 1995; Sutherland et al. 2000; Keller 2004) records the majority of the ∼800 km oblique right lateral relative motion accrued between the Australian and Pacific plates since this time (Figure 1). The Alpine Fault also formed through the A-NZ plate boundary from the Late Oligocene onwards, and it is along this fault that distinctive basement strain markers record at least half of the total relative plate motion since this time (Figure 1) (Wellman 1956; Sutherland 1995). In the Quaternary ∼80% of relative plate motion has been accommodated by displacement on the Alpine Fault and Hikurangi subduction thrust (Norris and Cooper 2001; Nicol and Beavan 2003; Barth et al. 2014). The remainder of plate boundary deformation includes a combination of contraction, strike-slip, extension and vertical-axis rotations distributed across a zone 200–400 km wide (Figures 1 and 2) (e.g. Walcott 1987; 1998; Sutherland 1999; Wallace et al. 2004; Ghisetti and Sibson 2006; Nicol et al. 2007; Giba et al. 2010; Lamb 2011).

Figure 1. A, Present-day configuration of basement terranes through the Aotearoa-New Zealand plate boundary (after Mortimer 2014). Abbreviation, DMOB = Dun Mountain Ophiolite Belt. Displacements of key strain markers shown in brackets. B, Middle Eocene model for the distribution of gently curving contiguous strain markers and Mesozoic continental-ocean boundary prior to ∼800 km of right-lateral transpression along the Alpine Fault (e.g. Sutherland 1995; King 2000). Central deformed region (CDR) or the ‘missing area’ inferred by previous rigid plate reconstructions (Kamp 1987; King 2000; Wood and Stagpoole 2007) shown by grey triangle. C, Alternative model of strain markers offset by Cretaceous left-lateral motion (e.g. Bradshaw et al. 1996; Lamb et al. 2016).

Figure 2. The Aotearoa-New Zealand plate boundary zone. From the north, oblique westward subduction of the Pacific plate (Hikurangi Plateau) beneath the North Island transitions southwards to transpression along the Alpine Fault at the latitude of the Chatham Rise (Beavan et al. 2002; 2007). Toward the south, the Alpine Fault transitions to eastward subduction of Australian plate oceanic crust beneath the Puysegur-Fiordland margin. At present the entire Hikurangi margin fore arc rotates at rates of ∼3°/Myr facilitated by extension in the Havre Trough (Wallace et al. 2004). Selected neotectonics faults (solid thin black lines) show the distribution of Holocene deformation along with inactive major Neogene structures (thick dashed black lines) (Stagpoole and Nicol 2008; Herzer et al. 2011; Edbrooke et al. 2014; Seebeck et al. 2022). Regional clockwise vertical-axis rotations of up to 70–90° have occurred along the Hikurangi margin forearc and southeastern South Island about two hinge zones (dashed white lines) during the Miocene (Hall et al. 2004; Lamb 2011). A vertical-axis fold hinge (dashed white line) runs the length of the South Island parallel to the Alpine Fault (Mortimer 2014). The simplified distribution of basement terranes (coloured polygons; modified after (Mortimer 2004; Heron 2020) throughout the plate boundary zone provide passive strain markers for reconstruction, in particular the trend of the Dun Mountain – Maitai terrane and the boundary between the Rakaia and Pahau petrofacies (Esk Head Mélange) of the Torlesse composite terrane. Location of relative motions shown in Figure 3 shown by white circle at the intersection of the Wairau and Waimea-Flaxmore faults. Inset shows map-view curve terminology from Marshak 2004). Abbreviations: AF, Alpine Fault; HM, Hikurangi margin; MFS, Marlborough Fault System; P-FM, Puysegur-Fiordland margin; TF, Taranaki Fault; VMFZ, Vening Meinesz Fracture Zone; W-FF, Waimea-Flaxmore Fault; WF, Wairau Fault.

Previous tectonic and palinspastic reconstructions have utilised different datasets across a wide range of spatial and temporal scales, making comparison and quantitative assessment of their predictions difficult (e.g. Kamp 1986; 1987; Walcott 1987; King 2000; Crampton et al. 2003; Nicol et al. 2007; Lamb 2011; Mortimer 2014; Lamb et al. 2016; Ghisetti 2021). Of these A-NZ plate boundary reconstructions, we consider King (2000) to be one of the most comprehensive syntheses of deformation, and sedimentary basin evolution. King (2000) presented a four-plate model reconstructing the plate boundary zone using a generally accepted assumption that oroclinal bending of Paleozoic-Mesozoic basement terranes mostly occurred after the Middle Eocene (Figure 1B) (see Mortimer 2014 for discussion). Although the majority of geologic constraints used to formulate this reconstruction are still valid, the finite rotations constraining relative motion between the Australian and Pacific plates were only schematic and semi-quantitative (King 2000). The limited number of crustal blocks used by King (2000) resulted in highly generalised and largely unconstrained deformation within the A-NZ plate boundary zone. This deformation is dominated by displacement on the Alpine Fault and clockwise vertical-axis rotations of crustal blocks across central parts of the plate boundary zone which resulted in the closing of a large triangular shaped region (e.g. Kamp 1987; King 2000), termed the central deformed region by Wood and Stagpoole (2007) (Figure 1B). A general summary for the evolution of the A-NZ plate boundary can be found within the Supplementary data, while an in depth paleogeographic synthesis can be found in Strogen et al. (2022).

Over the last two decades since the publication of King (2000), global apparent polar-wander paths (Torsvik et al. 2012) and finite rotations within the Zealandia plate circuit have been refined (Keller 2004; Cande and Stock 2004b; Whittaker et al. 2007; Croon et al. 2008; Granot et al. 2013; Whittaker et al. 2013; Wright et al. 2015; Choi et al. 2017), nationwide geological maps have been updated (Rattenbury and Isaac 2012; Edbrooke et al. 2014; Heron 2020) and plate boundary zone deformation better constrained (e.g. Nicol et al. 2007; Lamb 2011; Mortimer 2014; Ghisetti et al. 2016; Caratori Tontini et al. 2019). These refinements allow the update and improvement of King (2000) within a consistent and reproducible plate tectonic framework (e.g. Seton et al. 2012; Matthews et al. 2015). While there are alternative finite rotations for plate pairs within the Australia-Antarctic-Pacific plate circuit, not all of which are appropriate for reconstructing the A-NZ plate boundary (e.g. Seebeck et al. 2018), as well as different ways of configuring the plate circuit (e.g. Steinberger et al. 2004; Matthews et al. 2015) we use the most recent and well constrained finite rotations compatible with existing global plate reconstructions (e.g. Seton et al. 2012) for the model presented here.

This article describes the development of a rigid-plate reconstruction for the A-NZ sector of Zealandia from the mid-Late Cretaceous to the Holocene using GPlates (Boyden et al. 2011). We document the model boundary conditions, the geometry of its constituent crustal blocks, and the assumptions and constraints used for the reconstructions. The models are made available in digital form (see Supplementary data) and can be tested as new data become available. We use these models to examine the evolution of the A-NZ plate boundary zone and demonstrate that oroclinal bending of basement terranes from the Late Oligocene is feasible given the available data. These models provide a quantitative temporal and spatial framework for future geological studies of Aotearoa-New Zealand.

Data and methods

The A-NZ tectonic block model comprises three main components: an absolute reference frame, the relative motions between tectonic plates linked via plate circuits, and a plate model that describes the major and minor structural domains within the A-NZ plate boundary zone. We use GPlates plate tectonic reconstruction software (Boyden et al. 2011) and published finite rotations to provide the boundary conditions for relative motions between North and South Zealandia since the mid-Late Cretaceous in a paleomagnetic reference frame (Torsvik et al. 2012). Details of the paleomagnetic reference frame (Figure S2); the total reconstruction poles for the Zealandia plate circuit (Tables S1–S3) and location of stage poles constraining relative motions between North and South Zealandia from the Middle Eocene (Figure S4) are provided in Supplementary data. Rates of relative motion between the Australian and Pacific plates for a location within the central plate boundary zone are illustrated in Figure 3 (see Figure 2 for location). Reconstructions are presented for Australian Plate motion relative to the Pacific, however, within the supplied GPlates model the anchored plate can be specified to either of the major plates or the Earth’s spin axis. Crustal block rotations within the plate boundary zone are calculated interactively and are based on details summarised below and in the Supplementary data.

Figure 3. Relative plate motions since the Middle Eocene in central Aotearoa-New Zealand with respect to the trend of the Alpine Fault. See Figure 2 for location of point used to estimate rates of relative motion between North (Australian plate) and South (Pacific plate) Zealandia. Total reconstruction poles for the Australia, Antarctic, Pacific plate pairs used in this study can be found in Tables S1–S3. Chrons defining the stage pole rotations that constrain motion between North and South Zealandia shown by thin grey and black vertical lines. Chrons used as reconstruction times (Figures 8–11) shown by black vertical lines. Inset illustrates how the rate of relative motion (thick black line) is decomposed into normal (red line) and parallel (blue line) components with respect to the trend of the Alpine Fault (055°E).

Continuously deforming boundaries

To establish first-order estimates on the rates of plate boundary area change and crustal thickness through time, we define a continuously deforming polygon (Gurnis et al. 2012) representing the zone of A-NZ plate boundary deformation from the Middle Eocene (Figure 4). To assess crustal mass balance, rather than defining ‘missing areas’ between rigid blocks as with previous studies (Wood and Stagpoole 2007; Hines et al. 2022), a zone is defined that encompasses continental Middle Eocene–Holocene deformation zone from the southern Campbell Plateau to the northeastern continental margin (Figure 4). This approach considers the plate boundary zone a single deforming region bounded by rigid plates. The boundaries separating deforming from non-deforming regions can be represented as a set of line elements where each line moves with a single point of rotation (Gurnis et al. 2012). Together these lines form a closed polygon within which deformation occurs, either between a set of smaller internal rigid blocks undergoing differential motion, distributed deformation or a combination of the two (Gurnis et al. 2012). With the exception of the Hikurangi margin, these boundaries are controlled by the far-field motions of the Australian and Pacific plates. The rotation of the Hikurangi margin (which also includes the opening of the Havre Trough) is simply the result of joining two fixed points on either side of the plate boundary with the rate of rotation determined by far field plate motions (e.g. Lamb 2011). These long-term rotation rates are consistent with contemporary vertical-axis rotation rates (Wallace et al. 2004) and the geometry of the crustal block model described below. Further detail on geometry of the deforming zone can be found in the Supplementary data.

Figure 4. A continuously deforming polygon (red shaded region) representing the Aotearoa-New Zealand plate boundary zone from the Middle Eocene to the present-day shown relative to a fixed Pacific plate. With the exception of the Hikurangi margin, the deforming polygon is formed from boundaries (red lines) linked to Australia or Pacific plate motion. The extension of the boundary lines from the deforming region ensures a continuously closed polygon throughout the reconstruction (Gurnis et al. 2018). Refer to text and Supplementary data for further details. A rigid plate representation of the present-day coastline (blue line) and region defined by the 2000m isobath (grey polygon) are added for reference. Graph shows area (blue line) and rates of area change (red line) of the deforming region through time.

Rigid plate models

Within the deforming plate boundary zone lithospheric deformation reflects the relative motions of crustal fault-bounded blocks that extend at least through the brittle upper crust with ductile flow occurring at greater depths (Molnar et al. 1999; Thatcher 2009). The kinematics of contemporary deformation through the A-NZ plate boundary zone have been described in terms of both continuous strain (Beavan et al. 2007; Haines and Wallace 2020) and crustal block rotation (Wallace et al. 2004). Because the upper crust is brittle and elastic with faults weaker than the surrounding crust, crustal deformation can result in rigid-body like behaviour above a continuously deforming mantle (Molnar et al. 1999).

Within the A-NZ plate boundary zone, paleomagnetic data from Late Cretaceous–Quaternary sedimentary strata indicate that crustal blocks with along strike dimensions of 100–200 km can have similar vertical-axis rotation histories independent of their associated tectonic plate (Lamb 2011). Therefore, as constrained by the rigid-plate method, the approach adopted here is to treat fault-bounded regions or domains as crustal blocks. One advantage of this approach is that discrete tectonic episodes, and associated changes in kinematics, can be represented within a single coherent and reproducible model. The disadvantage of this approach is that all fault-bounded domains or blocks are likely to be experiencing internal deformation which is not explicitly captured in rigid-plate models (although may be partly manifest as overlap or separation of adjacent blocks). To help address this issue at a plate boundary scale we use the velocity field of a simple fault-propagation fold algorithm to model spatially distributed deformation during the Miocene to guide the configuration of rigid crustal blocks within the plate boundary.

Distributed deformation model

One of the debates for the reconstruction of the A-NZ plate boundary is the Middle Eocene configuration of the Mesozoic continental-oceanic margin along what is now the Hikurangi subduction margin, prior to the development of the Alpine Fault (Figure 1). On both sides of the Alpine Fault, the regional trend of distinctive Paleozoic–Mesozoic basement terranes bend toward the fault, forming an asymmetric ‘Z-shaped’ map-view curve or orocline (Figures 1A and 2) (Wellman 1956; Mortimer 2014). Evidence for a proto-Alpine fault prior to the Middle Eocene is equivocal (King 2000; Matthews et al. 2015). This, coupled with the observation of only minor deformation occurring through the A-NZ plate boundary from the mid-Late Cretaceous to Middle Eocene, has led many to interpret the development of oroclinal bending, in some or all regions of the plate boundary, as a largely Neogene phenomena (e.g. Molnar et al. 1999; Sutherland 1999; King 2000 and references therein; Lamb 2011). In addition, the difference in finite displacement between piercing points outside the plate boundary zone (i.e. the Eocene Campbell Plateau and Resolution Ridge rift margins) and those within (Figure 1) indicate a significant proportion of strain is accommodated through distributed deformation (cf. Sutherland 1999). Following these interpretations, the Middle Eocene (and therefore the inferred mid-Late Cretaceous) configuration of the Hikurangi margin largely determines the magnitude and style of later deformation throughout the plate boundary zone.

Models relating to this Middle Eocene configuration primarily fall into two end members; gently curving and contiguous basement terranes with a curvilinear Mesozoic margin termed here the contiguous model (Figure 1B) (e.g. Kamp 1987; Walcott 1987; Sutherland 1999; King 2000 and references therein; Mortimer 2014), or an embayed Mesozoic margin with or without left-lateral offsets of basement terranes here termed the offset model (Figure 1C) (e.g. Bradshaw et al. 1996; Crampton et al. 2003; Nicol et al. 2007; Lamb et al. 2016). The offset model requires that oroclinal bending largely occurred prior to the mid-Late Cretaceous separation of Zealandia from Gondwana and that the majority of transform motion between the Australian and Pacific plates has been accommodated primarily on the Alpine Fault.

To investigate the assumptions associated with the Middle Eocene contiguous model, we use the velocity field derived from a fault-propagation fold algorithm (Trishear) within the MOVE^TM software suite to model distributed deformation from the onset of transform motion in the Late Oligocene (26.55 Ma) to the Late Miocene (6.04 Ma) (Figure 5). This simple algorithm is used to demonstrate that the crustal block model developed here represents a plausible, though non-unique, initial configuration prior to the development of distributed right-lateral shear across the A-NZ plate boundary.

Figure 5. Miocene fault-propagation folding through the Aotearoa-New Zealand plate boundary zone. The plate boundary zone contains asymmetric map-view curves including recesses (Marlborough) and salient (Hikurangi margin) curves, and a fault cutting through the common limb at high cut off angles relative to the folded basement terranes (Alpine-Wairau fault) (Marshak 2004) (Figure 2). This type of deformation, commonly observed in cross-section through fold and thrust belts, can be described by a fault-propagation fold algorithm (trishear) (Hardy and Ford 1997). Folding develops progressively in a triangular zone (red dashed line) which opens away from a propagating fault tip (solid black line; A and B) that eventually cuts through the fold pair (C and D) (Hardy and Allmendinger 2011). Passive strain circles (grey circles) deforming in the triangular zone highlight the progressive contraction and elongation of the central deformed region (through the conservation of area) over time. In this model a propagation/slip (P/S) ratio of 2.3 replicates the general shape and progressive displacement of terrane boundary strain markers in map-view. Inset shows the relationship between the fault-propagation fold model and the crustal block model at 6.04 Ma.

In both contractional and extensional settings, fault-propagation fold algorithms are generally applied in cross-section to replicate field observations of asymmetric fold pairs verging in the direction of slip (Hardy and Ford 1997; Hardy and Allmendinger 2011). In fold and thrust belts these fold pairs comprise a footwall syncline, a hanging-wall anticline, high strains in the forelimb and a fault cutting through the common limb at high cut-off angles relative to folded strata (Hardy and Ford 1997). In map-view, similar geometric relationships have formed throughout the A-NZ plate boundary zone (Figure 2). Map-view curves in basement terranes in the form of salient and recess curves (curves that are convex and concave in the direction of transport, respectively; Marshak 2004) have high-angle intersections with and are displaced by decreasing amounts to the north by the Alpine-Wairau fault (Figures 1 and 2). With respect to the Pacific Plate, the Hikurangi margin and northwestern South Island forms the salient curve while the zone ∼80 km to the east of the Alpine-Wairau fault forms the recess curve (Figure 2; Marshak 2004).

Strictly speaking, many of the criteria determining whether the use of a fault-propagation algorithm to model map-view curves is appropriate (Hardy and Ford 1997; Hardy and Allmendinger 2011), such as steeply dipping boundaries between adjacent terranes, are not met (e.g. Mortimer et al. 2002). However, due to their varying compositions, thicknesses and structural relationships, rheological contrasts and layer-parallel flexural slip are inferred within and between terranes (e.g. Little and Roberts 1997; Little and Mortimer 2001; Upton et al. 2009; Turner et al. 2012; Edbrooke et al. 2014) supporting the use of this type of kinematic algorithm. In addition, from the Late Oligocene to Late Miocene, relative plate motion was predominantly strike-slip (Figure 3) inferring steeply dipping plate boundary structures that would be analogous to a reverse (or normal) fault in cross section. As gravity is not explicitly incorporated into fault propagation folding, the velocity field generated by the algorithm provides a simple kinematic method for the development of folding ahead of a propagating fault tip regardless of map-view or cross-section orientation. In this case, utilising the fault propagation fold algorithm makes little difference to the configuration of North Zealandia crust blocks within the deformation zone that are linearly interpolated by GPlates from their inferred Late Oligocene configuration to that determined in the Late Miocene. Utilising this type of algorithm does, however, provide insight into the formation of map-view curves either side of the Alpine Fault, the northward sequential offset of strain markers, and the change in shape rigid blocks may have undergone, all within a simple conceptual framework.

Two distinct strain markers, the Dun Mountain Ophiolite Belt (DMOB) and a mélange zone between the Rakaia and Pahau components of the Torlesse composite terrane, known as the Esk Head Mélange, provide constraint on reconstructions (Figures 1 and 2) (e.g. Wellman 1956; Kamp 1987; Mazengarb 1994; Sutherland 1999; King 2000; Nicol et al. 2007; Mortimer 2014; Ghisetti 2021). These key strain markers are representative of the trends of basement terranes either side of the Alpine-Wairau fault (Edbrooke et al. 2014; Heron 2020). The compilation and synthesis of existing data along the Hikurangi margin indicate clockwise vertical axis rotations typically between 70–90° at rates of 3–3.5°/Myr relative to the Australian Plate has occurred between two vertical-axis hinge zones situated in Marlborough (southern hinge) and Raukumara Peninsula (northern hinge) (Lamb 2011) (Figure 2). Clockwise vertical-axis rotation about the northern hinge initiated in the Miocene (Rowan and Roberts 2008) and most likely ceased with the development of intra-arc rifting in the Havre Trough and Taupō Volcanic Zone during the Pleistocene (Nicol and Wallace 2007). Similarly, the southern hinge in the northeastern South Island may record ≥90° clockwise deflection of basement terrane boundaries and structural fabrics during the Miocene (Little and Roberts 1997; Hall et al. 2004; Lamb 2011).

In summary, the use of a fault propagation fold algorithm to model map-view curves is supported by; (1) the propagation of the Alpine-Wairau fault inferred by the northward decreases in the finite displacements of key piercing points from the Late Oligocene to Late Miocene, (2) the high angle intersection of key strain markers with the Alpine-Wairau fault, and (3) paleomagnetic data indicating asymmetric vertical-axis rotation of a fold limb about a pair of vertical-axis hinges along the Hikurangi margin (Figure 2). Rigid plate reconstruction provides minimum ages of offset for the key strain markers where restoration of Esk Head Mélange piercing points between the northeastern South Island and southern North Island occurs around 4.24 Ma while DMOB piercing points could be considered contiguous across the Alpine-Wairau fault prior to 10.95 Ma (Figure S6). Further details of fault-propagation folding can be found in Supplementary data (Figure 6).

Figure 6. Central Aotearoa-New Zealand plate boundary. The similarity of structural trends between the southern North Island and northeastern South Island suggests a common deformation history prior to offset along the Wairau Fault (WF). Steeply dipping bedding in the Esk Head MélangeMélange (yellow line; strike in °E) in the Marlborough region are comparable to those north of the Wairau Fault (Rattenbury and Isaac 2012; Heron 2020). To the northwest of the Alpine-Wairau fault the Buller-Takaka terrane boundary (solid and dashed red line) has undergone post-Late Oligocene apparent right-lateral shear of a similar magnitude to strike-slip displacements of the Esk Head Mélange (EHM) across the Marlborough Fault System (∼55 km). In addition, Murchison Basin (MB) has experienced ∼50% shortening since the Middle Miocene (Lihou 1993). Paleomagnetic declination anomalies indicate vertical-axis rotations of >70° have occurred in the Marlborough and lower North Island during the Neogene (Hall et al. 2004; Lamb 2011). Axial trace of the Goulter syncline from Little and Mortimer (2001). Inset shows relative vector orientations scaled to slip rate derived from Figure 3 from the Late Oligocene to Late Miocene. Underlying geological map after Edbrooke et al. (2014).

Crustal block reconstruction

Within the A-NZ plate boundary zone we have developed a crustal block model (Figure 7) primarily delineated by geological and structural considerations such as major basement-penetrating crustal-scale faults, terrane boundaries, and vertical-axis rotation boundaries defined by paleomagnetic declination anomalies (Figure 2). Crustal block boundaries were preferentially defined by faults that have demonstrated evidence of Miocene or pre-Neogene displacement (e.g. Edbrooke et al. 2014 and Heron 2020 and data summarised therein). In most cases, crustal block boundaries are well-defined by discrete structures, however, in a few cases, boundaries are considered diffuse zones of faulting and deformation (e.g. Taupō Volcanic Zone). Although our division of crustal blocks along crustal-scale faults share some similarities with the structural sub-divisions of previous local and regional models (e.g. Moore 1988; King 2000; Wallace et al. 2004; Lamb 2011), the details of our crustal block boundaries differ from these previous reconstructions. The wider Zealandia region is divided into 50 crustal blocks, 40 of which comprise the A-NZ plate boundary zone encompassing the Raukumara, Hikurangi margin, Murchison-Westland, Marlborough, and Fiordland regions (Figure 7). Further information relating to the definition of crustal blocks can be found in Supplementary data and Hines et al. (2022).

Figure 7. Crustal block model for Aotearoa-New Zealand. Coloured regions represent crustal blocks that rotate relative to the primary plates of North and South Zealandia. Refer to Supplementary data text for crustal block boundary rationale. Inset shows the full extent of the crustal block model which incorporates the South Fiji and Norfolk basins adjacent to the north of Aotearoa-New Zealand.

To determine the relative motions of crustal blocks within the plate boundary zone, this reconstruction model has been developed iteratively in four time periods from the present to the mid-Late Cretaceous. Firstly, between c. 0 and 6 Ma, displacements on all faults except the Alpine Fault were progressively restored to determine the Late Miocene configuration, primarily of the Hikurangi margin and Marlborough regions (see section Holocene–Late Miocene below). Between c. 6 and 26 Ma the fault-propagation fold model guides the unbending of passive markers through the plate boundary zone to a contiguous model configuration (cf. Kamp 1987; Sutherland 1999; King 2000; Mortimer 2014) (see section Late Miocene–Early Oligocene below). Thirdly, between c. 26–44 Ma the relative positions of North Zealandia crustal blocks within the A-NZ deformation zone are fixed and move with the Australian Plate (see section Early Oligocene–Middle Eocene below). Finally, as only minor intraplate deformation is documented between c. 44–83 Ma (King 2000; Matthews et al. 2015) the relative position of the Australian and Pacific plates (and associated crustal blocks) at c. 44 Ma are fixed and carried back to c. 83 Ma to the onset of seafloor spreading between Australia, Antarctica and Zealandia (cf. Cande and Stock 2004a; Matthews et al. 2015) (see section Middle Eocene–mid-Late Cretaceous below).

Similar to reconstruction snapshots depicted in the literature, our plate boundary crustal block configurations were formed at selected ages defining total reconstruction poles for Australia-Pacific relative motion. The absolute ages selected for reconstruction are: 2.58, 4.24, 6.04, 10.95, 20.13, 26.55, 40.13 and 83.5 Ma corresponding to seafloor magnetic anomalies 2Ay, 3y, 3Ay, 5o, 6o, 8o, 18o and 34y, respectively (Figures 8–12).

Figure 8. Retro-deformation of the Aotearoa-New Zealand plate boundary zone. Crustal block reconstruction presented for the A, Holocene and B, Pleistocene. See Figure 7 for the crustal block configuration and Tables S1–S3 for the far-field plate motions constraining the model. The present distribution of basement terranes (Figure 2) allows the tracking of strain markers back through time. The trends of the rigid crustal blocks within the plate boundary zone are consistent with the Miocene fault-propagation fold model (Figure 5). Subducted horizontal extent of Pacific plate Benioff zone (red dashed lines) derived from Reyners et al. (2011) and Syracuse and Abers (2006). Subducted extent of the Hikurangi Plateau (blue dashed lines) from Reyners et al. (2011) and Riefstahl et al. (2020). Volcanic centres from Seebeck et al. (2014) and references therein.

Figure 9. Retro-deformation of the Aotearoa-New Zealand plate boundary zone. Crustal block reconstruction presented for the A, Pliocene and B, Late Miocene. See Figure 8 caption for more details.

Figure 10. Retro-deformation of the Aotearoa-New Zealand plate boundary zone. Crustal block reconstruction presented for the A, Middle and B, Early Miocene. See Figure 8 for more details. The trends of the rigid crustal blocks within the plate boundary zone are consistent with the Miocene fault-propagation fold model (Figure 5).

Figure 11. Retro-deformation of the Aotearoa-New Zealand plate boundary zone. Crustal block reconstruction presented for the A, Late Oligocene and B, Middle Eocene. The trends of the rigid crustal blocks within the plate boundary zone are based on the assumption of contiguous basement terranes between North and South Zealandia prior to transform motion (Figures 1 and 5).

Figure 12. Late Cretaceous reconstruction of Zealandia with Australia and Antarctica from Strogen et al. (2022) using finite rotations and crustal block configurations described in this study.

Reconstructions

Holocene–Late Miocene (0–6.04 Ma)

The Holocene to Late Miocene kinematics of the Hikurangi margin are intimately related to subduction and rollback processes along the Kermadec subduction margin to the north and continental collision of North and South Zealandia along the Alpine Fault to the south (Nicol et al. 2007; Wallace et al. 2009). Large-scale changes in the relative motions of the Australian and Pacific plates during the Late Miocene (Figure 3) (Croon et al. 2008; Austermann et al. 2011) changed the predominant tectonic regime along the Hikurangi margin from one associated with contraction and vertical-axis folding to one of bulk vertical-axis rotation facilitated by the opening of the Havre Trough (e.g. King 2000; Nicol et al. 2007; Lamb 2011).

For our reconstruction, the bulk vertical-axis rotation of the Hikurangi margin from the present to c. 6 Ma is restored (Figures 8 and 9). Due to the absence of documented relative motions between the Kermadec Ridge, Raukumara Basin and Raukumara Peninsula, restoration of back-arc spreading and rifting within the Havre Trough determines the location of the northern Hikurangi margin prior to c. 6 Ma. Bathymetric, magnetic and gravity features within and bounding the Havre Trough provide piercing points from which to reconstruct the separation of the Colville and Kermadec Ridges since c. 6 Ma (Wysoczanski et al. 2010; Caratori Tontini et al. 2019). The rates of bulk rotation of the Hikurangi margin crustal blocks used in this model are consistent with extrapolated contemporary GPS-derived rates of rotation back to c. 2 Ma (Wallace et al. 2004; Nicol and Wallace 2007) where closure of the Taupō Volcanic Zone is completed by 2.58 Ma (Figure 8B) (cf. Wilson et al. 1995). By maintaining line lengths along the Hikurangi margin, the near complete closure of the Havre Trough by 4.24 Ma (Figure 9A) resolves an overlap (∼100 km) generated by a rigid plate model between the continental crust of the northeastern South and southern North Islands (Figure S6). Margin-perpendicular contraction across the southern Hikurangi margin is restored using contraction estimates from the balanced cross-sections of Nicol et al. (2007). The full closure of the Havre Trough by c. 6 Ma (Figure 9B) results in the northern hinge zone south of the Raukumara Peninsula positioned along a relative plate vector trend to the east of prominent deflections of the DMOB in the western North Island (cf. Nicol et al. 2007). This northern hinge zone is coincident with a change in the thickness of the subducting plate (i.e. from thickened crust of the Hikurangi Plateau in the south to Cretaceous-age oceanic crust to the north; Figure 9B) (cf. Reyners et al. 2011; Reyners 2013).

In northeastern South Island by 6.04 Ma, 55 km of right-lateral displacement of the Esk Head Mélange within the Marlborough Fault System is restored (Figures 6 and 9B) (cf. Little and Roberts 1997). The restoration of displacements across the Marlborough Fault System results in the Esk Head Mélange having a broadly north–south trend approaching the plate boundary (Wairau Fault), similar in trend to correlative units in the southern North Island (Figures 6 and 9B). This interpretation differs from that of Ghisetti (2021), for example, in that the Alpine-Wairau fault is treated as a single continuous northward propagating fault rather than two faults propagating from the south and north, respectively. Vertical-axis rotations of crustal blocks within the northern Marlborough region (after Lamb 2011) unbend the inferred extension of the plate boundary structure through Cook Strait (Figure 9B) to approximate a small circle about the 3Ay-3o stage pole (Figure S4).

Late Miocene–Late Oligocene (6.04–26.55 Ma)

Crustal block reconstruction describing the restoration of key strain markers across the A-NZ plate boundary by the Late Oligocene is guided by the fault-propagation fold model (Figure 5). Here the progressive offsets of the Esk Head Mélange and DMOB along with the distributed deformation inferred ahead of a propagating Alpine Fault tip are approximated by crustal block rotations prior to 6.04 Ma (Figures 10 and 11A).

By 10.95 Ma terrane boundary trends along the Hikurangi margin through to the Marlborough region are sub-parallel and contiguous implying retro-deformation of Eastern province terranes (Figure 10A). Timing for offset of the Esk Head Mélange across the Wairau Fault is dependent on the magnitude of contraction across the southern North Island and is estimated here to occur between 6.04 and 10.95 Ma (cf. Little and Roberts 1997). As a conservative estimate, restoration of the Esk Head Mélange as a contiguous strain marker across the Wairau Fault occurs by 10.95 Ma (see discussion for further details). Similarly, displacement of the DMOB across the Alpine Fault commences around 18 Ma. Generalised contraction and strike-slip deformation in the Murchison-Westland region to the west of the Alpine-Wairau fault (e.g. Figure 6) are restored using estimates from balanced cross-sections (Nicol et al. 2007; Ghisetti et al. 2016) and strain markers. The restored eastward extent of the highly simplified crustal blocks of the Murchison-Westland region place an important constraint on the relative location of North and South Zealandia in the Middle Eocene (Figure 11B). Here, the restored map-view extent of the Australian Plate crust in this region constrains the western-most location of Pacific Plate during Eocene rifting and sea-floor spreading south of New Zealand (Figure S3). Fiordland remains static in the reconstruction.

The Buller-Takaka terrane boundary in northwestern South Island, along with the distinctive plutonic rocks of the Median Batholith, have no apparent vertical-axis rotation relative to the Australian Plate (Figure 6) (Turner et al. 2012; Lamb et al. 2016). The Buller-Takaka terrane boundary (Anatoki Thrust), however, can be tracked to a piercing point along the Alpine Fault (Edbrooke et al. 2014; Heron 2020) indicating that this strain marker has been deflected southwards (with the equivalent of ∼55 km of right-lateral displacement) across an ∼70 km-wide zone relative to northwestern South Island (Figure 6). While deformation in this zone appears largely ductile, individual oblique strike-slip faults within this deformation zone record post-Oligocene right-lateral apparent displacements of 6–9 km across fault blocks ranging between 10–50 km in width (Edbrooke et al. 2014; Heron 2020). At a regional scale, the apparent shearing of the Buller-Takaka terrane boundary as it approaches the Alpine Fault represents an ∼20° clockwise deflection of this strain marker (Figure 6). In addition, this 70 km wide deformation zone to the north of the Wairau Fault also encompasses a clockwise vertical-axis rotation (25–30°) of a synform displaced across the Alpine-Wairau fault (Figure 6) (Little and Mortimer 2001). The crustal block containing the Goulter synform also contains the DMOB piercing point on the western-side of the Alpine-Wairau fault along with basement fabrics that are sub-parallel with the Esk Head Mélange in both North and South Islands (Figure 6). These observations indicate that right-lateral shear and crustal block rotation has occurred to the west of the Alpine-Wairau fault. The paleomagnetic declination anomalies associated with the Australian Plate in this region (Figure 6) are discussed further below.

The maximum Neogene-Quaternary contraction perpendicular to the Alpine-Wairau fault across northwestern South Island is estimated at 17–30 km (Ghisetti and Sibson 2006; Nicol et al. 2007). Exhumation rates in this region increase significantly from 6 Ma (Jiao et al. 2017) suggesting much of the contraction has accrued since the Late Miocene. Within the 70 km wide deformation zone to the northwest of the Alpine-Wairau fault, the north–south-trending Murchison and Maruia basins have experienced contraction of up to 50% since the Middle Miocene (Lihou 1993; Ghisetti and Sibson 2006) parallel to the relative plate vector (Figure 6). In this deformation zone northwest of the Alpine-Wairau fault it appears strike-slip shear zones and faults trending sub-parallel to the Alpine-Wairau fault link north–south trending reverse faults and folds resulting in a right-stepping en echelon distribution of basins (Figure 6). The width of this deformation zone north of the Alpine-Wairau faults has similar dimensions and trend to that of the Marlborough Fault System (∼80 km width) that developed from 7 to 6 Ma to the south (Figure 6) (Little and Roberts 1997).

To the north of A-NZ during the Early to Middle Miocene, the restoration of seafloor generated at the Minerva Triple Junction in the South Fiji Basin prior to 15 Ma results in the westward motion of the Three Kings and Colville-Lau ridges relative to Northland (Figure 10B) (Herzer et al. 2011). At the same time, closure of the Norfolk Basin results in a similar westward motion of the Three Kings Ridge along the Vening Meinesz and Cook fracture zones (Herzer et al. 2011). The location of the Raukumara Basin pre-Middle Miocene is uncertain. The basin is inferred to overlie Cretaceous age oceanic crust (Sutherland et al. 2009) which is largely outboard of the continental margin in a location adjacent to the Kermadec-Colville arc.

Late Oligocene-Middle Eocene (26.55–43.79 Ma)

Stage poles determining relative motion between North and South Zealandia (Australia and Pacific plates, respectively) between the Middle Eocene and Late Oligocene are located within or in close proximity to the A-NZ plate boundary zone (Figure S4) (Sutherland 1995; Wood et al. 1996; Keller 2004; Cande and Stock 2004a; Granot et al. 2013). Crustal blocks of the Murchison-Westland region and their juxtaposition with the Campbell Plateau rift margin provide constraint for the closure of the northern end of the Emerald Basin through the minimisation of continental crust overlap (Figure 11B). Fiordland is situated in a position whereby rocks of the Median Batholith occur either side of the location occupied by the future Alpine Fault. Minor crustal block rotations within the plate boundary zone maintain the contiguous terrane boundaries between North and South Zealandia. The location of the stage pole at this time (Figure S3(d)) indicates very low relative motions for crustal blocks of the Hikurangi margin and Marlborough regions, and as such, these regions are fixed to their respective plates. Closure of the South Fiji and Norfolk basins is complete by the Early Oligocene (c. 34 Ma) restoring the continuity of the Three Kings and Loyalty ridges along the eastern margin of North Zealandia (not shown) (Herzer et al. 2011).

Middle Eocene-Mid-Late Cretaceous (43.79–83.5 Ma)

In this model, the geometry of crustal blocks in the Middle Eocene is used to define the geometry of Zealandia as far back as the mid-Late Cretaceous breakup of eastern Gondwana (Figures 11B and 12). Opening of the Emerald Basin occurs from anomaly 20o (43.79 Ma), with magnetic anomaly picks indicating the majority of seafloor spreading accumulating from 40.13 Ma (Keller 2004; Granot et al. 2013). From the Middle Eocene, crustal blocks within the plate boundary zone are fixed to their respective plates along with the relative position of North and South Zealandia and are carried back via the plate circuit to the mid-Late Cretaceous initiation of seafloor spreading between Zealandia, Antarctica, and Australia at c. 83 Ma (Figure 12) (cf. Cande and Stock 2004a; Matthews et al. 2015). The assumption of no relative motion between the North New Zealand-Challenger Plateau and South New Zealand-Pacific crustal blocks during this time results in unconstrained motion between East and West Antarctica during this period (cf. Cande and Stock 2004a; Matthews et al. 2015) and the overlap of the Campbell Plateau with West Antarctica from c. 79 Ma. Resolving the uncertainties of the plate circuit propagating through East and West Antarctica (e.g. Davey et al. 2016; 2022) is beyond the scope of this present model. 

Discussion

Crustal mass balance and contraction

Estimates of crustal mass balance provide information about changes in crustal thickness in response to regional contraction or extension. To examine first-order changes in crustal mass balance from the Middle Eocene we utilise crustal thickness estimates from Grobys et al. (2008) that encompasses the deforming A-NZ plate boundary zone and the adjacent Challenger and Campbell plateaux. The Middle Eocene configuration (Figure 11B) shows that the Campbell and Challenger plateaux were juxtaposed and have comparable crustal thicknesses of 22.1 ± 2.4 km and 20.1 ± 2.4 km, respectively (Figure S8) (Grobys et al. 2008). Offshore of the eastern South Island, deep seismic-reflection profiles and gravity modelling indicate crustal thicknesses of 20–23 km (Mortimer et al. 2002) that are comparable to the Chatham Rise (Grobys et al. 2008; Riefstahl et al. 2020). Crustal thickness estimates in the northern North Island, away from the influence of the Hikurangi subduction margin, indicate crustal thicknesses of 20–25 km (Wood and Woodward 2002; Stern et al. 2006) and within the uncertainty of the estimates for the Challenger Plateau presented here. The average crustal thicknesses within the initial configuration of deformation zone in the Middle Eocene is therefore assumed to be similar to those of the Challenger and Campbell plateaux. Using the higher of these two crustal thickness estimates combined with the area of the initial configuration of Middle Eocene deformation zone results in a first-order crustal volume of 1.44 × 10⁷ km³. Considering the 1-σ uncertainty for the crustal thickness estimate, crustal volumes for the initial configuration of deformation zone in the Middle Eocene range from 1.2 × 10⁷ km³ to 1.49 × 10⁷ km³. These estimates are equivalent to the crustal volume within the present-day configuration of the deforming zone of 1.29–1.33 × 10⁷ km³ derived from a volume calculation (using a 5 km³ sample cell size) or an average crustal thickness (26.9 ± 8.9 km; Figure S8) (Grobys et al. 2008).

The evolution of the continental deformation zone represented by continuously closing boundaries shows decreasing area from the Middle Eocene (Figure 4). With respect to the Alpine Fault, the motion of the boundary representing the Challenger Plateau rift margin indicates that prior to the Late Miocene motion was predominantly strike-slip (Figure 3) with only minor convergence (represented by overlap) of these two boundaries (Figure 4). This contrasts with thermochronology along the Alpine Fault indicating crustal thicknesses >35 km had been reached by the end of Middle Miocene (Ring et al. 2019). By the end of the Middle Miocene, the area of the deforming zone had decreased by ∼12% from its initial Middle Eocene size. The conservation of area constraining the fault-propagation folding algorithm is consistent with this small reduction in deformation zone area to this time. Convergence of the Challenger Plateau rift margin along the proto-Alpine Fault begins in earnest during the Late Miocene (Figure S7). The rate at which the deforming zone was decreasing in area increases significantly in the latest Miocene (Figure 4) due to short duration changes in relative motion (Figure 3). Since the Pliocene, the rate of area decrease has been broadly constant with minor variations associated with the opening of the Havre Trough and Taupō Volcanic Zone. At present, the deformation zone has reduced in size from the initial Middle Eocene area by 24% (Figure 4).

Spatially distributed shear and fault propagation

The northward propagation of the Alpine-Wairau fault is inferred to be driven by a combination of the weakening and thinning of continental crust between the Campbell and Challenger plateaux through rifting and seafloor spreading to the south (Sutherland 1995) and the roll-back of the Pacific Plate to the north (Herzer et al. 2011) (Figure 11). Stage poles describing Australia-Pacific relative plate motion between the Middle Eocene-Late Oligocene are generally located in proximity to the Chatham Rise close to the southern termination of subduction (Figures S3 and S4). The location of Australia-Pacific stage poles in the Late Oligocene resulted in northward decreasing rates of relative motion during the initiation of transform motion (Figure 11A) requiring the northward propagation of the transform boundary at least into the Early Miocene.

During subduction initiation, or the subduction of thickened oceanic crust (such as the Hikurangi Plateau), frictional coupling between the over-riding and subducting plates is inferred to be high, resulting in contraction and uplift of the over-riding plate (House et al. 2002; Nikolaeva et al. 2010; Tetreault and Buiter 2012; Vogt and Gerya 2014). Strong and potentially spatially varying plate coupling along the Hikurangi margin in the Early Miocene coupled with strong lithospheric contrasts between North and South Zealandia continental fragments within the deformation zone (e.g. Reyners et al. 2011; Reyners 2013), in combination with a propagating transform boundary, provide all the necessary conditions for the development of oroclinal folding (cf. Marshak 2004).

The crustal mass balance estimates above indicate that a model with a contiguous and gently curved Mesozoic continental margin (Figures 1B and 5A) is a plausible Middle Eocene configuration regardless of the style and distribution of subsequent deformation. The kinematic fault-propagation fold model (Figure 5) accounts for both the asymmetric orocline on both sides of the plate boundary and the ‘missing’ areas commonly identified in rigid plate reconstructions (e.g. Wood and Stagpoole 2007; Hines et al. 2022). The conservation of area within the fault-propagation fold model suggests that much of the ‘missing’ area represented in the model for the central deformed region (Wood and Stagpoole 2007; Hines et al. 2022) can be accounted for by distributed shear within a triangular deformation zone ahead of a propagating fault tip (Figure 5). This distributed shear results in elongation and contraction of crustal blocks toward the present which when restored result in ‘missing’ areas common to rigid-plate models (e.g. King 2000).

Models with pre-Middle Eocene offsets along the Alpine-Wairau fault require the majority of relative plate motion to be localised along the Alpine-Wairau fault and Hikurangi subduction thrust from the Late Oligocene. For models containing prior offsets of the Mesozoic continental boundary, the problem of the ‘missing’ area of the central deformed region in rigid plate models is simply moved to the incipient Hikurangi margin resulting in the earlier initiation (i.e. Late Oligocene rather than Early Miocene), with potentially larger amounts of subduction than presently documented (e.g. Reyners et al. 2011; Bland et al. 2022; Strogen et al. 2022).

Oroclinal bending and vertical-axis rotations

Evidence for distributed shear within the North Island implicit in the model presented here is equivocal. The fault-propagation fold model provides a mechanism for the bending of basement terrane strain markers and tightening of map-view curves proximal to the Alpine-Wairau fault (Little and Mortimer 2001). The deflection of the DMOB in the southern South Island toward the trend of the Alpine Fault occurs along a fold hinge that runs the length of the South Island (Mortimer 2014), incorporating the southern hinge zone in the Marlborough region (Figure 2). The magnitude of clockwise deflection of the DMOB is equivalent to that observed for the Esk Head Mélange. The fold hinge joining these two strain markers lies sub-parallel to, and 60–80 km east of, the Alpine Fault. Moderately dipping strata (40–50°) steepen (70–90°) westward across this fold hinge toward the Alpine Fault (Little and Mortimer 2001; Heron 2020) which forms the eastern boundary to the triangular deformation zone in the fault-propagation folding model. In the model presented here, the proto-Alpine Fault initiates in the south along the Challenger Plateau rift margin (cf. Sutherland 1995) propagating northwards from the Late Oligocene (Figures 5 and 8) (cf. Ghisetti 2021). This northward propagation rotates strain markers ahead of the fault tip, resulting in the sequential bending and offset of the basement terranes along the South Island. Due to the short wavelength of the oroclinal folding east of the Alpine Fault, the crustal block model does not yet include this kind of small-scale deformation south of the Marlborough region.

Paleomagnetic declination anomalies in the northwestern South Island and western North Island (e.g. Figure 6) are commonly used to demonstrate the absence of vertical-axis rotations relative to the Australian Plate since the Middle Eocene (Turner et al. 2012; Lamb et al. 2016). Turner et al. (2012) and Lamb et al. (2016) determine the rotation of the Australian Plate through time relative to paleomagnetic north from the Australian apparent polar wander path of Veevers and Li (1991). In the global paleomagnetic reference frame of Torsvik et al. (2012) the Australian Plate rotates at rates ∼30% slower than those predicted by Veevers and Li (1991) (Figure S2). These lower rates of Australian Plate rotation predicted by Torsvik et al. (2012), supported by recent paleomagnetic studies in North Zealandia (Dallanave et al. 2022), would result in clockwise paleomagnetic declination anomalies for Late Eocene sites in the western North Island and northwestern South Island of ∼15° for Late Eocene strata (e.g. Figure 6). If confirmed, these clockwise vertical axis rotations would support the argument for longwave length oroclinal folding of basement terranes across the North Island since the Late Eocene (e.g. Figure 5).

In the reviews of Lamb (2011) and Mortimer (2014), the ∼90° clockwise vertical-axis rotation of the Esk Head Mélange in the northeastern South Island is interpreted to have occurred predominantly in the Neogene, ceasing between 8 and 4 Ma (Hall et al. 2004; Lamb 2011). Paleomagnetic observations also indicate that the eastern North Island (and the Esk Head Mélange) also rotated through the Neogene (Nicol et al. 2007; Lamb 2011). In a fault-propagation fold model, the tight map-view curve associated with the oroclinal-bending hinge east of the Alpine Fault would be continuous until a time when a through-going fault develops. In the model presented here, vertical-axis folding in the southern hinge zone mainly ceases when distributed shear localises on the Alpine-Wairau fault preserving the high-angle map-view trends of strain markers, such as the Esk Head Mélange, approaching the fault (Figure 6). The timing for the offset of the Esk Head Mélange along the Wairau Fault most likely occurred between 7–6 Ma during a significant change in the motion vector of the Pacific Plate (Austermann et al. 2011), increasing contraction across the Alpine Fault and southern North Island (Tippett and Kamp 1993; Jiao et al. 2015; 2017), and the development of the Marlborough Fault System to the east (Little and Roberts 1997).

Fiordland block deformation

Within the Median Batholith, a distinctive ∼20 km-wide belt of Late Jurassic to earliest Cretaceous plutons (Darran Suite) forms the eastern half of the Fiordland crustal block (Mortimer 2014). As with other strain markers along the plate boundary, the trend of these distinctive plutons is deflected toward the Alpine Fault, away from their regional trend further to the east (Edbrooke et al. 2014; Heron 2020). As discussed by Mortimer (2014), previous depictions of Fiordland in Eocene paleogeographic reconstructions fall into three categories; (1) a rigid block in its current position relative to Stewart Island (i.e. fixed to South Zealandia), (2) a rigid block translated south of Stewart Island (i.e. once part of North Zealandia), or (3) it is ignored in reconstructions.

While the shape of Fiordland has remained unchanged in the crustal block model presented here, an implication of the fault-propagation folding model is that Fiordland has undergone distributed shear resulting from the northward propagation of the transform plate boundary. We suggest that this style of folding has deflected the trend of the Median Batholith clockwise within ∼80 km of the Alpine Fault. This mechanism does not require large-scale vertical-axis rotation of the crustal block, for which there is a lack of evidence (Lamb et al. 2016) and inherent space problems in rigid plate models. The parallelism of Median Batholith trends in Fiordland and the northwestern South Island can be reconciled by folding succeeded by dissection of the fold limb by the Alpine Fault in a similar manner to that inferred for the Esk Head Mélange. This line of reasoning suggests that the Fiordland crustal block has always been a part of South Zealandia and has deformed in a non-rigid manner within ∼100 km of the plate boundary, similar to the trends of Late Eocene–Early Oligocene fracture zones in the Emerald Basin further to the south (Figure 1) (Hayes et al. 2009).

Concluding remarks

A new reconstruction for Aotearoa-New Zealand is presented based on a model with contiguous, gently curving Paleozoic-Mesozoic basement terrane configuration in the Middle Eocene. The A-NZ plate boundary zone has decreased in area by ∼25% since the Middle Eocene resulting in crustal thickening along the plate boundary. The decreasing northward displacement of basement strain markers, the tightening of map-view curves about two vertical-axis hinge zones and an asymmetric triangular region of deformation support the concept of a distributed fault-propagation folding style of deformation to describe the evolution of the Late Oligocene to Late Miocene A-NZ plate boundary zone. The ‘missing’ triangular area common to many map-view rigid-plate reconstructions of the A-NZ plate boundary can be reconciled through contraction, elongation and vertical-axis rotation of crustal blocks within the deformation zone ahead of a propagating Alpine Fault.

Whilst we recognise that the model presented in this paper over-simplifies deformation within the A-NZ plate boundary, the complementary use of rigid-plate, continuous deforming polygons, and simple kinematic models provide insight into the evolution of deformation along the plate boundary. As such, we consider the models presented here to be appropriately scaled to fit the resolution, simplification and assumptions used in the creation of the model. While the current crustal block model is unable to accurately represent some of the smaller wavelength features discussed above, future iterations of the model should incorporate changes in crustal block shape (e.g. Fiordland) to reflect a non-rigid style of deformation. To that end, we concur with Mortimer (2014) in that realistic reconstructions of Zealandia from at least the Middle Eocene will need to allow for internal crustal block deformation in order to reconcile constraints from plate motions, paleomagnetism, and basement geological markers. The crustal block model does, however, capture the first-order structure of the plate boundary and provides the foundation for recent regional paleogeographic studies (Hines et al. 2022; Strogen et al. 2022).

Acknowledgements

This work has benefitted from discussions with many colleagues both at GNS Science and throughout the wider New Zealand geological community. Special thanks go to our colleagues Peter King, Nick Mortimer, Cornel de Ronde, Matt Sagar and James Crampton among others. The EarthByte Group are thanked for the development and continuing improvement of GPlates software. We are grateful to Petroleum Experts for providing support for the MOVE^TM structural modelling software suite. Detailed comments by Francesca Ghisetti, the senior editor, and two anonymous reviewers greatly helped improve the manuscript.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The tectonic reconstructions presented in this study have utilised open-file and published datasets, which are cited through the paper. Supplementary data provide more detailed information on the evolution of the Aotearoa-New Zealand plate boundary, the absolute reference frame and finite rotations used for the model, crustal block definitions, fault-propagation folding parameters, description of continuously deforming boundaries and crustal volume estimates presented in this paper along with shapefiles and the GPlates rotation files used to construct the model. Supplementary data are available at figshare via https://doi.org/10.6084/m9.figshare.22105817.

Additional information

Funding

This work was supported by the New Zealand Ministry of Business, Innovation and Employment to GNS Science via the Strategic Science Investment Fund (contract C05X1702) as part of the Sedimentary Basins Research and Te Riu-a-Māui/Understanding Zealandia programmes.