Back-arc basins are extensional features formed behind subduction zones by progressive rifting of volcanic arcs or behind the arc fronts in pre-existing oceanic basins until passive mantle upwelling creates new oceanic crust (Karig, 1970). Their initiation is triggered by processes of either hinge-rollback (Chase, 1978; Scholz and Campos, 1995), and/or upper plate migration away from the trench associated with slab anchoring (Uyeda and Kanamori, 1979; Scholz and Campos, 1995; Heuret and Lallemand, 2005). Migration of the trench relative to the overriding plate leads to a diverse arrangement of plates and deformation styles within the near-arc and back-arc environments (Heuret and Lallemand, 2005). Edge-driven kinematics are common along the boundaries of microplates, where microplate rotation is driven by the motion of the two larger plates the microplate is pinched between (Schouten et al., 1993). These microplate rotations can cause block rotations, shearing, and further rift propagation (e.g., Easter Microplate: Neves et al., 2003). Hinge-rollback, seamount subduction, microplate interactions, and variations in trench geometry and/or subduction angle all influence the state of stress in the overriding plate, driving upper mantle flow and magmatic upwelling, and the formation of structures that provide pathways for magma to reach the surface. However, seismologic data alone cannot fully resolve the types of faulting, and therefore the stress regimes that lead to the emergence of these structures often remain enigmatic.

This paper focuses on the geological and structural evolution of the NE Lau Basin, which is one of the most volcanically- and tectonically-active places on Earth (Embley et al., 2009; Rubin et al., 2013; Embley and Rubin, 2018). New high-resolution multibeam bathymetry collected by the R/V Falkor expedition FK171110 in 2017 and R/V Sonne expedition SO-263 in 2018 are compiled with previously collected bathymetry and are used for morphotectonic analyses. Interpretations of this data, together with seafloor samples and ground truthing, are used to create the first remote-predictive geologic and structural maps of the region. The structures are further classified based on shallow seismicity (Centroid Moment Tensors) to provide insight into the recent kinematic evolution of the faulting. This analysis provides context for understanding the controls on the development of large magmatic-hydrothermal systems across the study area.

The Tonga-Kermadec subduction zone in the western Pacific extends over 2,000 km from New Zealand to Fiji, where the Pacific Plate subducts westward beneath the Indo-Australian Plate (Figure 1). Collision of the Louisville Seamount Chain with the trench has segmented the subduction zone into the Tonga Trench in the north and the Kermadec Trench in the south. This collision induced compression in the fore-arc and may have influenced where and when rifting in the back-arc occurred (Ruellan et al., 2003). Rifting of the Tonga Arc at 5.5–6 Ma formed the Lau Basin (Figure 2), which evolved to mature seafloor spreading along multiple spreading centers at ca. 4 Ma (Zellmer and Taylor, 2001; Taylor and Martinez, 2003). Arc volcanism shifted from the remnant Lau Ridge to form the new Tofua Arc at 3.5 Ma (Tappin et al., 1994). In the southern part of the subduction zone, rifting of the Kermadec Arc at ca. <2 Ma produced the Havre Trough (Wysoczanski et al., 2019).

The orientation of the modern Tonga Trench is north-easterly, with a sharp bend to a westerly direction at the northern apex where there is a transition from subduction to transform motion. This transition is associated with a Subduction-Transform-Edge-Propagator (STEP; Govers and Wortel, 2005). In the following, we refer to the east-west transform boundary as the “STEP fault,” which is associated with vertical tearing that causes a piece of the subducting plate to remain at surface (Govers and Wortel, 2005; Nijholt and Govers, 2015). Behind the arc, the modern Lau Basin displays a tapering V-shape with a width of ∼500 km in the north, narrowing to 200 km in the south where it merges with the Havre Trough at 26°S (Figure 2). In the south of the Lau Basin, two plates (Tonga and Australian) are separated by a single segmented spreading center. The number of spreading centers increases northward, indicating increasing tectonic complexity (Sleeper and Martinez, 2016). The Niuafo’ou Microplate in the north has been interpreted to occur between the Tonga and Australian plates, although it is bounded by and contains numerous small spreading centers and propagating rifts, and so structurally might be even more complex (i.e., multiple microplates). There are likely several other micro- or nano-plates in the northern basin with poorly defined (possibly diffuse) boundaries (Zellmer and Taylor, 2001; Phillips, 2003; Conder and Wiens, 2011; Baxter et al., 2020).

Plate reconstructions by Sleeper and Martinez (2016) suggest that non-rigid plate behavior may be important in this area, where plate boundaries can propagate and rotate, and experience intraplate deformation. Notably, the NE Lau Basin is dominated by strike-slip faulting, revealed by analysis of focal mechanisms of shallow crustal earthquakes (Hawkins, 1995; Baxter et al., 2020). Baxter et al. (2020) describe these focal mechanisms across the entire Lau Basin, and interpret strike-slip faulting to result from re-activation of normal faults produced along spreading centers. In this study, we focus on the structures of the NE Lau Basin in greater detail, complemented by morphotectonic analyses (remote-predictive geologic mapping), in order to better understand the origin and evolution of strike-slip faulting and the possible influence on back-arc magmatic-hydrothermal processes.

This study combines large hydro-acoustic datasets (multibeam bathymetry and backscatter) and seafloor sampling and observations from the past decade. Interpretation of this data (including morphotectonic analyses) is used to generate a remote-predictive geological map of the study area. This is critical for understanding the controls on the distribution of structural and geologic features. Publicly available Centroid Moment Tensor (CMT) data is interpreted together with the orientations of nearby structures to investigate fault kinematics.

Most of the prior work toward understanding the geology of the Lau Basin has been on regional-scale studies of the crustal evolution and shallow seismicity (e.g., Sleeper and Martinez, 2016; Baxter et al., 2020; Stewart et al., in press). Higher-resolution studies have focused on the southern basin along the Eastern and Central Lau Spreading Centers, where crustal accretion resembles mid-ocean ridges. This contrasts the processes of crustal accretion in the NE Lau Basin, where a diffuse system of back-arc extension, short-lived rifts, spreading centers, jumping ridge crests, and point-source volcanism dominate (Taylor et al., 1996; Embley et al., 2009). In the study area, there are five distinct crustal types: Lau back-arc crust, Lau rear-arc crust, Tofua arc crust, paleo-arc crust, and Pacific Plate crust (undifferentiated), as shown in Figure 4 and described in Supplementary Table S1. These crustal types are primarily based on lithological information (Supplementary Figure S3), relative locations, inferred ages from literature data, and seafloor morphology. Additional constraints are provided by vertical gravity gradient (VGG) anomalies from the CryoSat-2 and Jason-1 missions (Sandwell et al., 2014), and seafloor magnetization anomalies (0.5-arc-minute: Austin, 2012, and references therein). These data highlights areas of focused crustal accretion such as back-arc spreading centers or volcanoes, which appear as gravity highs in the VGG data and tend to correlate with positive magnetic anomalies. Finally, we also extrapolate multichannel seismic reflection data along the FRSC (Schmid et al., 2020) to distinguish the approximate transition from arc crust to back-arc crust.

Based on detailed morphotectonic analyses and additional ground-truthing data, we present new remote-predictive geological and structural maps of the NE Lau Basin at high-resolution (Figures 5–7; Supplementary Table S2), which highlight the wide variety of rock types, extensive volcanism, and complex structural fabrics resulting from the dynamic evolution of the area. We categorize our map into nine assemblages, which are characterized by distinct modes of formation (e.g., related to specific spreading centers or volcanic/sedimentary processes) and geochemical signatures. Each assemblage contains one or more physiographic units, which display internal uniformity in terms of seafloor morphology (e.g., smooth vs. hummocky) and interpreted mode of formation (e.g., lava flow vs. sediment), as well as relative age when determinable (e.g., through embayment relationships, acoustic backscatter signature).

A total of 174 shallow CMTs (≤25) km are reported in the centroid moment tensor catalog within the map area. Of these, we interpreted the focal planes of 125 CMTs according to their location and proximity to known structural and tectonic features, and classified them according to orientation in Figures 7, 8 and Supplementary Figure S3. The focal planes of 47 CMTs trend 351–10°, 35 CMTs trend 11–30°, 6 CMTs trend 31–50°, 5 CMTs trend 51–70°, 6 CMTs trend 71–90°, 10 CMTs trend 271–290°, 9 CMTs trend 291–310°, and 7 CMTs trend 331–350°. Fault planes were plotted as stereonets to check for consistency in the groupings (Supplementary Figure S5). The fault plane solutions of the remaining 49 CMTs could not be determined (Figure 8).

Normal earthquakes comprise only 10% of the CMTs with known focal plane solutions, and these events are restricted to the FRSC in the south of the map area (n = 8; Figure 7A), as well as the northern tip of the NELSC (n = 4; Figures 7B,C). However, extensional rift-parallel horst-and-graben fault patterns are common across the map area, including within neovolcanic zones and off-axis terrains along the main spreading centers (Figure 7A). Extensional structural fabrics in the NE Lau basin are characterized by two main structural trends. The first trend is parallel to sub-parallel with the NELSC and MTJ-N (NNE- to NE-trending; Figures 7B,C). This trend follows the orientation of the Tofua arc between 15°30ʹS and 17°35ʹS (southern part of the map area; Figures 2, 5). The second trend is parallel to sub-parallel with the MTJ-S and FRSC (N-trending; Figure 7A). This trend follows the orientation of the Tofua arc north of 15°30ʹS (northern part of the map area; Figures 2, 5). Subordinate structures are orthogonal to these trends but also display normal faulting and extensional features, including the NE- and ENE-trending large volcanic ridges near the southern NELSC and within the Mata volcanic group (Figures 5, 7C,D). The lack of normal faults in the CMT data likely indicate that extension in the basin is simply not producing high-magnitude (Mw > 5) earthquakes. This may indicate that crustal accretion is dominated by magmatic extension with strain accommodated by dike injection rather than tectonic extension with strain accommodated by brittle faulting (e.g., Buck 2004; Buck et al., 2005; Buck, 2006; Ito and Behn, 2008; Wright et al., 2012; Ebinger et al., 2013; Anderson et al., 2017). Alternatively, the lack of CMTs with normal fault plane solutions around the spreading centers may be related to high spreading rates, which are associated with lower magnitude earthquakes (<5 Mw; Macdonald, 1982). Notably, thousands of lower magnitude (<5 Mw) seismic events around the ELSC, CLSC, and FRSC have been recorded by ocean bottom seismometers deployed in these regions (Eguchi et al., 1989; Conder and Wiens, 2011; Schmid et al., 2021), supporting these suggestions.

The CMT data indicate that strike-slip and oblique-slip faulting is widespread across the study area comprising 90% of the interpreted events. This has been described by earlier authors (Hawkins, 1995; Baxter et al., 2020), but here we provide higher-resolution interpretations of the orientation and shear sense interpreted from these CMTs, supported by data of offset features (described in Shallow Seismicity and Fault Kinematics). The main N- and NNE-trending structures (along with minor NNW-trending structures) display a right-lateral shear sense (Figures 7A,B,I), accounting for 73% of shallow strike-slip CMT solutions in the area. The N-trending structures are parallel to the FRSC and the MTJ-S, while the NNE-trending structures are parallel to the NELSC. In contrast, NE-trending structures do not display strike-slip motion; instead, these structures may be associated with pure low-magnitude normal faulting with no oblique motion. Subordinate NE-, ENE-, E-, W-, NNW-trending structures display a left-lateral shear sense (Figures 7D–G), accounting for 27% of the shallow strike-slip CMT solutions. The left-lateral CMTs occur near the northern plate boundary (n = 7), likely caused by slip between the northern microplates and the Pacific Plate, and farther to the south along the W Rift, MTJ-W, and northern FRSC (n = 19). Dip-slip faults also occur along the STEP fault east of the NELSC, trending NE, E, W, and WNW (n = 12; Figures 7C,E–G).

Previous authors have suggested that the strike-slip motion in the NE Lau Basin results from strain induced by the transform motion along the northern STEP fault and asymmetric slab rollback (e.g., Hawkins, 1995; Baxter et al., 2020). This interpretation is supported by the close association of strike-slip CMTs along the boundary, extending southward to ∼16°S where strike-slip CMT solutions terminate abruptly (Figure 8; see also Figure 5 in Baxter et al., 2020). This model suggests that there is a major, regional-scale diffuse sinistral shear zone in the NE Lau Basin with a width of approximately 120 km. Notably, there is no evidence for slip along the inferred southern boundary of the shear zone, which instead is dominated by a diffuse zone of seismicity. This is consistent with other edge-driven microplate rotation models (e.g., Schouten, 1993). On land, regional shear zones referred to as “megashears” have been interpreted to be induced by the relative motion between two plates (e.g., Arthaud and Matte, 1977; Neev et al., 1982; Campbell and Anderson, 2003) or in intra-plate settings where pre-existing weaknesses in the basement localize shearing (e.g., Frisicale et al., 2010). In these strike-slip fault systems, the fracture patterns typically follow specific geometries referred to as a Riedel shear zone, which can occur at different scales.

Here, we relate the distribution and orientation of right-lateral and left-lateral strike-slip faults within the rigid block boundaries to typical Riedel megashear mechanisms, revealing two distinct sets of structures (Figures 9, 10). The first set of Riedel megashears, which we group under the name “Riedel System 1,” is dominated by NNE-trending Rʹ-shears, with minor E-trending R-shears and NNW-trending P-shears (Figure 9) The R’-shears are oriented at a high angle (∼75°) counter-clockwise to the boundaries of the megashear zone, which trend ∼95° following the orientation of the northern plate boundary. The R′-shears follow the main trends along the NELSC and the W Rift zone and are sub-parallel to MTJ-N. Typically, R′-shears may develop with or after R-shears (e.g., Atmaoui et al., 2006), which are oriented ∼10–15° counter-clockwise and synthetic to the megashear zone boundaries. In the study area, R-shears are not widespread, but are sub-parallel to MTJ-W, and closely align with the structural trend of the northern Mata volcanoes and a well-developed strike-slip fault zone in the northwest map area. The P-shears are more difficult to identify but may be manifest as the WNW-trending fabrics in the southern and northern parts of the map area. Within this configuration, compression associated with σ₁ may produce the large anticline(?) associated with the NW Ridges Assemblage (Supplementary Figure S41) and may also result in bulging morphology of the paleo-arc crust. Extension associated with σ₃ is associated with the large ENE-trending volcanic ridges at the southern end of the NELSC, as well as West and East Mata. The orientation of these structures is oblique to the direction of hinge-rollback, suggesting that normal back-arc spreading processes cannot account for all the extension in this area. These extensional features are offset by the R-shears, creating an extensional duplex (Figure 9). North of the FRSC, there appears to be a cluster of NNE-trending seismicity associated with left-lateral fault motion, contrasting the right-lateral motion of faults along other NNE-trending CMTs in the region (Figure 9B). Sleeper and Martinez (2016) suggests that these strike-slip events are due to a zone of transferred lithosphere, although the regional kinematic control on this reversal is unclear. We therefore interpret the southernmost diffuse boundary of the megashear zone to occur near Escarpment B (Figures 5, 9B).

The second set of Riedel megashears are similar but rotated ∼8–12° counter-clockwise relative to the first set of shears, which we group under the name “Riedel System 2” (Figure 10). These shears are dominated by N-trending Rʹ-shears, with minor ENE-trending R-shears and W-trending P-shears. The Rʹ-shears follow the orientation of MTJ-S and FRSC, the N-trending fissure system extending north of the Niuatahi volcano, and other N-trending features in the area, including the area of extension labeled “Rifting” on Figures 5, 10A. The R-shears follow the trend of the West Mata and East Mata volcanoes and the large volcanic ridge to the southwest of Niuatahi, apparently re-activating previous extensional fabric. The megashear zone boundaries trend ∼80°, following the previous orientation of R-shears in System 1 (Figure 9B), and sub-parallel to the orientation of Escarpment B in the south (dashed lines in Figure 10). P-shears are also poorly defined for this configuration, but match strike-slip faulting near the northern and southern boundaries of the megashear zone. In this configuration, extension along σ₃ is associated with NE-trending volcanic ridges across the map area, such as the western Mata volcanoes, which are offset by R-shears. This extensional fabric is also parallel to the MTJ-N, the West Rift zone, and the northern and southern terminations of the NELSC and may be enhancing spreading associated with hinge-rollback.

The differing distribution of fault populations from a typical Riedel shear zone (dominated by R-shears) is likely because Rʹ-shears occur at an angle similar to that of faults produced by earlier back-arc extension. It is easier to reactivate these pre-existing faults as opposed to creating new ones as the stress required for frictional sliding along pre-existing faults is much less than the fracture strength (e.g., Byerlee, 1978). Similarly, as the stress-field rotated, pre-existing extensional faults were re-activated as R-shears. This effect has been observed in other back-arc basins (e.g., Manus Basin: Martinez and Taylor, 1996; Morley et al., 2004; Maestro-González et al., 2008). Faults that have undergone strike-slip reactivation tend to have zig-zag geometries with little throw. These morphologies are common throughout the map area but are particularly well-defined in the western part of the map area in the West Rift zone and the MTJ-W arm. Therefore, the observed strike-slip kinematics in the study area support a rigid block model of lithospheric-scale Riedel (mega)shearing, where shearing reactivates pre-existing extensional faults. This indicates that structures in megashear zones may be predisposed to align themselves to the pre-existing fabrics in back-arc settings. Since the tensional stresses in this region are accommodated by re-activation of normal faults as strike-slip faults, other large-scale strike-slip features (such as large-scale drag features) do not manifest in the map view of the area.

Structures associated with both Riedel systems 1 and 2 are distributed across the map area; however, cross-cutting relationships indicate that System 1 structures only cross-cut themselves, while System 2 structures cross-cut both themselves and System 1 structures (Table 1; Figures 6C,D), indicating that the System 2 structures formed later. In addition, the distribution of modern seismicity revealed by CMTs provides some insight into the timing of Riedel megashear formation. Active seismicity associated with System 1 megashears occurs mainly in the western portion of the map area (light blue beach balls in Figure 9B), while seismicity associated System 2 megashears occurs in the central part of the map area closest to the volcanic arc (green and pink beach balls in Figure 10B). Therefore, we propose that System 1 megashear structures formed first, encompassing the entire map area. This was followed by System 2 megashear structures. Currently, both configurations remain active in the different regions of the study area. Therefore, we propose that occurrence of two different Riedel megashear geometries in the NE Lau Basin (Systems 1 and 2; Figures 9B, 10B) is associated with a counterclockwise rotation of the stress field over time. In general, the geometry and kinematics of the back-arc structures support the idea of a megashear zone, driven by the relative motion between the plates. Additional work is needed to understand the geodynamic controls on the changing stress field throughout the map area.

In the NE Lau Basin, the counterclockwise rotation of the stress field is associated with a change in the orientation of the extensional axis along a new northward trend. As the extensional axis changes, there is a tendency for rifts to propagate outwards from the tips along the new extensional axis (e.g., CLSC: Parson et al., 1990). This process may be quicker when propagation occurs in thinner crust and where back-arc extension is the greatest associated with fast rates of trench retreat (Parson et al., 1990). This structural reconfiguration may have induced the propagation of new rifts, such as FRSC segments, which are interpreted to be the youngest features in the NE Lau Basin (Schmid et al., 2020). Microseismicity at the southern tip of the FRSC indicates ongoing southward propagation of this rift toward the volcanic arc (Schmid et al., 2021). There may also be simultaneous northward propagation of the FRSC and southward propagation of the MTJ-S (Figure 5). In the area where these two spreading centers overlap, we observe oblique and curved structural patterns and active seismicity (e.g., North Fiji Basin: Ruellan et al., 1994; Conder and Wiens, 2011). In addition, within the map area, N-trending rifting is occurring along the paleo-arc crust in the northern part of the map area (labeled “Rifting” in Figure 5). These structures may reflect the early stages of rift propagation, but higher-resolution magnetic data is needed to investigate the nature of these features. Ongoing rift propagation is interpreted along the northern and southern tips of the NELSC. The northern end is associated with a NE-trending V-shaped rift tip and active normal faulting (Figures 5, 7C), while the southern NELSC displays a SW-trending V-shaped rift tip bound by steep-walled escarpments (Escarpment B and C in Figure 5). The propagation of these NE- and SW-trending rift tips may be enhanced by σ₃ extension in the reconfigured (System 2) stress field, in addition to slab rollback (Figure 10B).

The re-orientation of the stress field has mainly affected the area closest to the arc, preserving the original megashear zone configuration in the west where it remains seismically active (Figure 9B). This could indicate that the eastern portion of the NE Lau Basin may be in the nascent stages of nanoplate emergence, proposed by Conder and Wiens (2011) as the separate “Niuatoputapu” plate. This is supported by a drastic change in the structural fabric of the basin from west to east across Escarpment C (Figures 5–7). The fabric to the west of this escarpment is dominated by zig-zagging normal-faulted terrane of the West Rift assemblage, and to the east of this escarpment are the N- and NNE-trending volcanic ridges of the NW Ridges and NELSC assemblages. Like other back-arc micro- and nanoplates, the precise boundaries of Niuatoputapu are difficult to define, and may remain only partially separated, as is the case in the southern diffuse boundary of the Niuafo’ou microplate (Figure 2; Sleeper and Martinez, 2016). This work demonstrates an important mechanism for the emergence of new micro- or nano-plates in back-arc environments; namely, the propagation of new rifts in response to a rotating stress field.

Compared to mid-ocean ridges, back-arc spreading centers are ephemeral features that evolve dynamically in response to convergence, subduction-zone collisions, and microplate interactions, among others. The NE Lau Basin is characterized by extreme tectonic complexity, associated with fast convergence rates, high upper mantle temperatures, and thin oceanic crust (Bevis, et al., 1995; Conder and Wiens, 2011; Embley et al., 2018), producing diverse lithologies and eruption styles in the rear-arc and back-arc regions (e.g., Embley and Rubin, 2018). This complexity is manifest as seafloor fabrics with variably oriented structures. The observed structural patterns reflect the interplay between extension associated with slab-rollback and strike-slip tectonics along the northern STEP fault, with shallow-crustal strike-slip faulting extending over ∼120 km from the STEP fault. Two distinct sets of structures associated with Riedel shear mechanisms are described for the first time, indicating a recent counterclockwise rotation of the stress field that primarily affects the region closest to the arc. The structural configurations highlight the importance of re-activation of earlier-formed structures during rotation of the stress-field. The structures that formed and were re-activated in this megashear zone account for the orientations of many of the previously-enigmatic seafloor features observed, such as the Mata volcanoes and large off-axis volcanic ridges. This study provides important constraints on the geologic and structural evolution of the NE Lau Basin, including the tectonic controls on enhanced magmatism in rear-arc and off-axis regions, as well as the distribution of hydrothermal vent sites in the region.