Deformation across the active Hikurangi subduction margin, New Zealand, including shortening, extension, vertical‐axis rotations, and strike‐slip faulting in the upper plate, has been estimated for the last ∼24 Myr using margin‐normal seismic reflection lines and cross sections, strike‐slip fault displacements, paleomagnetic declinations, bending of Mesozoic terranes, and seafloor spreading information. Post‐Oligocene shortening in the upper plate increased southward, reaching a maximum rate of 3–8 mm/year in the southern North Island. Upper plate shortening is a small proportion of the rate of plate convergence, most of which (>80%) accrued on the subduction thrust. The uniformity of these shortening rates is consistent with the near‐constant rate of displacement transfer (averaged over ≥5 Myr) from the subduction thrust into the upper plate. In contrast, the rates of clockwise vertical‐axis rotations of the eastern Hikurangi Margin were temporally variable, with ∼3°/Myr since 10 Ma and ∼0°–1°/Myr prior to 10 Ma. Post 10 Ma, the rates of rotation decreased westward from the subduction thrust, which resulted in the bending of the North Island about an axis at the southern termination of subduction. With rotation of the margin and southward migration of the Pacific Plate Euler poles, the component of the margin‐parallel relative plate motion increased to the present. Plate convergence dominated the Hikurangi Margin before ca. 15 Ma, with the rate of margin‐parallel motion increasing markedly since 10 Ma. Vertical‐axis rotations could accommodate all margin‐parallel motion before 1–2 Ma, eliminating the requirement for large strike‐slip displacements (for example, >50 km) in the upper plate since the Oligocene.

We use seismic reflection and rock sample data to propose that the first‐order physiography of New Caledonia Trough and Norfolk Ridge formed in Eocene and Oligocene time and was associated with the onset of subduction and back‐arc spreading at the Australia‐Pacific plate boundary. Our tectonic model involves an initial Cretaceous rift that is strongly modified by Cenozoic subduction initiation. Hence, we are able to explain (1) complex sedimentary basins of inferred Mesozoic age; (2) a prominent unconformity and onlap surface of middle Eocene to early Miocene age at the base of flat‐lying sediments beneath the axis of New Caledonia Trough; (3) gently dipping, variable thickness, and locally deformed Late Cretaceous strata along the margins of the trough; (4) platform morphology and unconformities on either side of the trough that indicate a phase of late Eocene to early Miocene uplift to near sea level, followed by rapid Oligocene and Miocene subsidence of ∼1100–1800 m; and (5) seismic reflection facies tied to boreholes that suggest absolute tectonic subsidence at the southern end of New Caledonia Trough by 1800–2200 m since Eocene time. The Cenozoic part of the model involves delamination and subduction initiation followed by rapid foundering and rollback of the slab. This created a deep (>2 km) enclosed oceanic trough, ∼2000 km long and 200–300 km across, in Eocene and Oligocene time as the lower crust detached, with simultaneous uplift and local land development along basin flanks. Disruption of Late Cretaceous and Paleogene strata was minimal during this Cenozoic phase and involved only subtle tilting and local reverse faulting or folding. Basin formation was possible through the action of at least one detachment fault that allowed the lower crust to either be subducted into the mantle or exhumed eastward into Norfolk Basin. We suggest that delamination of the lithosphere, with possible mixing of the lower crust back into the mantle, is more widespread than previously thought and may be commonly associated with subduction initiation, such as Cenozoic events in the Mediterranean and western Pacific.

We use seismic reflection and refraction data to determine crustal structure, to map a fore‐arc basin containing 12 km of sediment, and to image the subduction thrust at 35 km depth. Seismic reflection megasequences within the basin are correlated with onshore geology: megasequence X, Late Cretaceous and Paleogene marine passive margin sediments; megasequence Y, a ∼10,000 km3 submarine landslide emplaced during subduction initiation at 22 Ma; and megasequence Z, a Neogene subduction margin megasequence. The Moho lies at 17 km beneath the basin center and at 35 km at the southern margin. Beneath the western basin margin, we interpret reflective units as deformed Gondwana fore‐arc sediment that was thrust in Cretaceous time over oceanic crust 7 km thick. Raukumara Basin has normal faults at its western margin and is uplifted along its eastern and southern margins. Raukumara Basin represents a rigid fore‐arc block >150 km long, which contrasts with widespread faulting and large Neogene vertical axis rotations farther south. Taper of the western edge of allochthonous unit Y and westward thickening and downlap of immediately overlying strata suggest westward or northwestward paleoslope and emplacement direction rather than southwestward, as proposed for the correlative onshore allochthon. Spatial correlation between rock uplift of the eastern and southern basin margins with the intersection between Moho and subduction thrust leads us to suggest that crustal underplating is modulated by fore‐arc crustal thickness. The trench slope has many small extensional faults and lacks coherent internal reflections, suggesting collapse of indurated rock, rather than accretion of >1 km of sediment from the downgoing plate. The lack of volcanic intrusion east of the active arc, and stratigraphic evidence for the broadening of East Cape Ridge with time, suggests net fore‐arc accretion since 22 Ma. We propose a cyclical fore‐arc kinematic: rock moves down a subduction channel to near the base of the crust, where underplating drives rock uplift, oversteepens the trench slope, and causes collapse toward the trench and subduction channel. Cyclical rock particle paths led to persistent trench slope subsidence during net accretion. Existing global estimates of fore‐arc loss are systematically too high because they assume vertical particle paths.

No abstract

We analyse results of dredge sampling, a high-resolution seismic reflection survey, and unpublished petroleum industry data from part of the overriding continental plate of the Fiordland subduction zone. Plate tectonic calculations show that convergence rates have progressively increased since Miocene time, when subduction-related processes started. An initial deformation phase was characterised by reverse throw on pre-existing structures during the interval 16-8 Ma. This is interpreted to be when a throughgoing subduction interface developed, as fracture zone linkage and spreading ridge extinction occurred south of New Zealand. A Pliocene-Quaternary deformation phase characterised by renewed folding and reverse faulting on structures subparallel to the plate boundary may be due to a regional change in plate motion or the inherited geometry of the Eocene continent/ocean transition that is obliquely colliding with the plate boundary, or both. Initial uplift of the region south of Fiordland to sea level was followed by c. 1800 m of subsidence at the Snares Depression. This subsidence may have been an isostatic response to tectonic erosion of a crustal root, or due to negative slab buoyancy associated with greater total convergence, or both. Persistence of Fiordland topographic elevations above sea level, and continued rock uplift of Fiordland, is likely to be related to greater initial crustal thickness of the overriding Pacific plate in the north, and the present southward transition from continental collision to intra-oceanic subduction. The locations of the first three subduction-related volcanoes have a relationship to pre-existing faults and raise the possibility that volcanism and deformation may be intimately associated during subduction initiation.