Abstract The Main Mantle Thrust (MMT) represents the tectonic boundary between metamorphic shield and platform rock of the Indian plate hinterland, and dominantly mafic and ultramafic rock of the Kohistan-Ladakh arc complex in Pakistan. In some areas, this boundary is a sharp planar fault with development of mylonite; in other areas, it is a brittle-ductile imbricate zone; in still other areas, it contains large, discontinuous, slices of internally sheared and deformed ophiolitic mélange. The character of the MMT along its entire trace is discussed and it is concluded that there is no single continuous fault which marks the contact between the Indian plate and the Kohistan-Ladakh arc. On this basis, we propose a revised definition for the MMT that is consistent with both the original definition and with the usage of the term in literature. We suggest that the MMT fault contact be defined as the series of faults, of different age and tectonic history, that collectively define the northern margin of the Indian plate in Pakistan. On this basis, faults that define the MMT vary in age from Quaternary to possibly as old as Late Cretaceous. Discontinuous lenses of ophiolitic mélange that overlie the MMT fault contact, and which intervene between the Indian plate and the Kohistan-Ladakh arc, are considered to be part of an MMT zone that is equivalent with the Indus Suture Zone.

The Nanga Parbat–Haramosh (NPHM) massif is a unique structural and topographic high in the northwestern corner of the Himalayan convergence zone. Previously, the NPHM was thought to be bounded by the Main Mantle Thrust (MMT), a fault along which the Kohistan-Ladakh island arc was obducted onto the northern margin of India. This study presents field evidence that the recently active dextral reverse Raikot fault truncates the MMT and forms the western boundary of the NPHM. The Raikot fault separates medium-grade, Mesozoic to middle Cenozoic mafic metasedimentary and intrusive rocks of the Kohistan island arc (Kohistan Sequence) from high-grade Proterozoic metasedimentary rocks (Nanga Parbat Group) and orthogneisses of the Indian craton. The Kohistan Sequence rocks have experienced one tight to isoclinal folding event, probably associated with obduction of the island arc, and a second folding event associated with movement on the Raikot fault. The Nanga Parbat Group rocks were transposed by an early (possibly Proterozoic) isoclinal folding event and have subsequently been folded around east-trending axes in the early Cenozoic by the obduction of Kohistan, then around north-trending axes in late Cenozoic time in association with the uplift of the NPHM and initiation of the Raikot fault. The Raikot fault consists of both mylonite zones and numerous major and minor faults. Slickensides and mylonitic lineations both indicate dextral reverse slip. The Raikot fault and associated folds appear to have accommodated as much as 15 to 25 km of uplift during late Cenozoic time. The localization of the uplift and the involvement of the Moho suggest that the Raikot fault follows a major crustal structure, possibly a pre-collision Indian plate boundary. If this is the case, rotational underthrusting of greater India along the MMT would require dextral slip along the Raikot fault. It is proposed that the Raikot fault is a terminal tear fault on the MCT.

Gravity data along a north-south profile from Kohistan to the Punjab plain of Pakistan have been incorporated into recent interpretations of the gross structure of the foreland fold-and-thrust belt of the Himalaya. In northern Pakistan, large deviations from Airy Isostatic equilibrium are observed. An excess of mass characterizes the northern Kohistan arc, and a deficit of mass underlies a broad area extending from southern Kohistan to the Salt Range, while to the south a slight excess of mass seems to prevail in the region of the Sargodha high. This anomalous distribution of mass can be understood if the Indian elastic plate, which is assumed to overlie a buoyant “fluid,” is flexed down under the weight of both the overthrust mountains and the sediments eroded off the mountains and deposited in the foredeep basin. In many respects the intracontinental subduction of India beneath the Himalaya is similar to island arc formation, including the seismically active Sargodha high, a basement ridge analogous to the flexural bulge encountered seaward of oceanic trenches. Analysis of Bouguer gravity anomalies along a profile extending from the Sargodha high to the Main Mantle Thrust (MMT) shows that most of the negative-northward gravity gradient can be attributed to crustal thickening. In the Sargodha high area, an additional contribution of about 25 mgal appears to be due to excess of mass at lower crustal or upper mantle levels. The Moho discontinuity is interpreted to bulge up beneath the Sargodha high, then gradually increase in dip from 1° to 3° beneath the Salt Range and Potwar plateau (approximately equal to the change in dip of the basement surface). The Moho is interpreted to change from upwardly convex to upwardly concave beneath southern Kohistan. Finally, north of the Main Mantle Thrust it appears to bend down again, but at a steeper angle of about 15°. Shorter wavelength anomalies, superimposed on the regional Bouguer gradient, are modeled in terms of upper crustal density changes, including those due to: (1) offsets of the basement surface; (2) variable thickness of the Eocambrian evaporite sequence that forms the basal décollement; (3) thrusting and folding of relatively high-density, older parts of the stratigraphic section to higher structural levels, particularly in the Salt Range and northern Potwar plateau; and (4) thickening of the low-density Neogene molasse sequence into the axis of the Soan Syncline, a structural depression between the Salt Range and northern Potwar plateau. Subsurface densities of the overthrust wedge, as well as the definition of the shape of the top surface of the Indian plate interpreted from seismic reflection and drilling data, place bounds on the flexural rigidity of such a plate and the forces that deform it. In northern Pakistan, a steeper Bouguer gravity gradient suggests that the flexural rigidity of the elastic plate (D = 4.0 [± 2.0] × 10 23 Nm) is a factor of 10 smaller than the current values interpreted for the central and eastern Himalaya. Moreover, the maximum flexural stresses are probably concentrated within the crust, which may account for the seismic activity of the Sargodha high and southern Kohistan. At the end of the Indian elastic plate (arbitrarily chosen at the MMT), a large positive vertical shear stress, 9.2 × 10 12 N/m < S 0 < 1.6 × 10 13 N/m, is applied to account for the topographic load north of the MMT. In addition, to fit the gravity constraints it was necessary to apply a large negative bending moment, −1.4 × 10 18 N < M 0 < −0.85 × 10 18 N, at the end of the plate. The negative bending moment can be explained by the combined effect of the northward migration of the Indian plate and the southward differential compressional force generated by the crustal rocks stacked at mid-upper crustal levels beneath the northern Kohistan arc. In addition, buoyancy of the crustal rocks at deeper levels beneath the Kohistan arc may contribute to the negative bending moment. Consequently, in southern Kohistan the surface of the Indian plate is concave up; compressional stresses in the upper part of the plate are probably the primary source of the Hazara seismic zone, where incipient reverse faulting seems to take place. In contrast, the pronounced upward convexity developed along the flexural bulge can account for (1) tensional stress in the upper part of the Indian plate, which is large enough to produce basement normal faults interpreted beneath the Salt Range and Sargodha high; and (2) compressional stress in the lower portion of the crust, which causes the excess of mass and seismicity beneath the Sargodha high.

Abstract There are several competing interpretations of the structure of the margins of the Nanga Parbat massif: that the massif is bounded by the original suture between the Indian continent and the Kohistan-Ladakh island arc―the Main Mantle Thrust; that the massif is entirely bounded by neotectonic faults; that it is bounded by a combination of early and late faults and shear zones. If the marginal structures of the massif are to be related to local and regional geo-tectonic evolution then their correct characterization is critical. The Raikhot Bridge area on the western margin of the massif is useful in this regard, as it provides accessible and near-continuous outcrops. This contact, sometimes called the Raikhot Fault, is composite. Sheared metagabbros of the Kohistan arc are juxtaposed tectonically against metasediments and orthogneisses of the Nanga Parbat massif along an early ductile shear contact, developed under amphibolite facies conditions. In this regard it may be a preserved segment of the Main Mantle Thrust. However, this ductile shear zone has been strongly modified, flattened and rotated, and is cut by younger shears and faults. The original kinematics of the shear zone have been largely overprinted by these subsequent deformations. The younger structures include NE-SW striking, dextral strike-slip faults and a major top-to-NW thrust and shear zone. A sequence of metamorphism, deformation and igneous emplacement may be used to study the history of structural evolution within the massif. The use of a single name (e.g. Raikhot Fault) for the present-day map contact between the Nanga Parbat massif and neighbouring Kohistan is misleading. The early contact (termed here the Phuparash Shear Zone, possibly the northeastern continuity of the Main Mantle Thrust) is modified by the Buldar Fault Zone (dextral strike-slip) and the Liachar Thrust Zone (top-to NW carriage of the Nanga Parbat massif across the Phuparash Shear Zone and onto Kohistan). The activity of the Buldar Fault and Liachar Thrust Zone continued during exhumation of the massif, through amphibolite facies to the Earth's surface. The interaction between these structures is at present unknown. However, establishing these and similar interactions within the Nanga Parbat area remain central to establishing the role of regional NE-SW dextral transpression in the modern structure of the massif.

Abstract This paper summarizes some 30 years of more intense recent work and almost 100 years of geological observations in Kohistan. The paper is divided into two section: an earlier factual-based section with minimal interpretation, and a later section summarizing a range of ideas based on the data as well as presenting new thoughts and interpretations. Kohistan is a c. 30 000 km 2 terrane situated in northern Pakistan. The great bulk of Kohistan represents growth and crustal accretion during the Cretaceous at an intra-oceanic island arc dating from c. 134 Ma to c. 90 Ma (Early to Late Cretaceous). This period saw the extrusion of c. 15–20 km of arc volcanic and related sedimentary rocks as well as the intrusion of the oldest parts of the Kohistan batholith, lower crustal pluton intrusion, crustal melting and the accretion of an ultramafic mantle–lower crust sequence. The crust had thickened sufficiently by c. 95 Ma to allow widespread granulite-facies metamorphism to take place within the lower arc. At around 90 Ma Kohistan underwent a c. 5 Ma high-intensity deformation caused by the collision with Eurasia. The collision created crustal-scale folds and shears in the ductile zone and large-scale faults and thrusts in the brittle zone. The whole terrane acquired a strong penetrative foliation fabric. Kohistan, now an Andean margin, was extended and intruded by a diapiric-generated crustal-scale mafic–ultramafic intrusion (the Chilas Complex) with a volume of 0.2×10 6 km 3 that now occupies much of the mid–lower crust of Kohistan and had a profound impact on its thermal structure. The Andean–post-collisonal ( c. 90–26 Ma) period also saw the intrusion of the stage 2 and 3 components of the batholith and the extrusion of the Dir Group and Shamran/Teru volcanic rocks. Collision with India at c. 55–45 Ma saw the rotation, upturning, underplating and whole-scale preservation of the terrane. The seismic structure of Kohistan has some similarities to that of mature arcs such as the Lesser and Greater Antilles and Japan, although Kohistan has a higher proportion of high-velocity granulites in the lower crust. The chemical composition of Kohistan is very different from that of average continental crust, although it is similar to an analogue obducted arc within Alaska (Talkeetna), suggesting that ‘mature’ continental crust undergoes a series of geochemical processes and reworking to transform an initial stage 1 ‘primitive arc crust’. Most of Kohistan is gabbroic in composition, particularly within the lower and middle crust. A high proportion of the ‘basement’ volcanic units is also basaltic to basaltic andesite with smaller proportions of boninite, andesite to rhyolite, ignimbrite and volcaniclastic material. Post Eurasian-collision ‘cover’ volcanic rocks are highly evolved, comprising predominant rhyolites, ignimbrites and related volcaniclastic rocks. Most lithological units throughout the crustal section have an arc-like geochemical composition (e.g. high LREE/HREE and LFSE/HFSE ratios) although some have oceanic (main ocean and back-arc) characteristics. Isotopic compositions indicate that the great bulk of igneous rocks have an ultimate sub-arc mantle source. In broad terms the Kohistan terrane represents a juvenile mantle extract addition to the Phanerozoic continental crust with a total volume of c. 1.2×10 6 km 3 (equivalent to c. 1/50 the volume of the Ontong–Java Plateau or Alaska).

ABSTRACT Plate tectonic reconstructions play a pivotal role in unravelling the complexity of the evolution of the Indian plate during its longest voyage and provide a platform to address long-standing geologic and paleobiogeographic questions in a geodynamic context. The northward drift of the Indian plate from its original Gondwana home in the late Paleozoic to its current position in Asia since the early Cenozoic provides a unique natural laboratory for tracking its changing geography, climate, tectonics, and vertebrate evolution for the past 300 m.y. Lithologic indicators of climate indicate a progressive amelioration of the climate of India with time, from Early Permian Ice Age through cooler temperate to warmer temperate climate in the Late Permian. By the Triassic, a more tropical monsoon-type climate prevailed, and India continued within the subtropical to tropical climates during its northward journey until its collision with Asia. In the Pangean world from the Late Permian to the Late Jurassic, India exchanged tetrapod fauna both with Gondwana and Laurasia without any physical barrier. During that time, India was pivotal in the emergence of major groups of tetrapods such as squamates, sauropods, and early mammals. For nearly 100 m.y., the Indian plate drifted from Gondwana until its collision with Asia ca. 55 Ma, during which time there were shifting roles of dispersal and vicariance that shaped the Indian paleobiogeography in time and space. With the breakup of Gondwana in the Late Jurassic, India began to disintegrate into a smaller plate, becoming partially isolated during the Early Cretaceous Period but possibly retained a biotic link with Africa via Madagascar. Circa 80 Ma as the Indian plate collided with the Kohistan-Ladakh (KL) arc, Arabia collided with the Oman arc; the dual collision formed a continuous accreted terrane—the Oman-Kohistan-Ladakh (OKL) arc, which served as a biotic filter bridge between India and Africa. India also established another circumpolar filter bridge during the Late Cretaceous via emergent Ninetyeast Ridge and Antarctica to exchange tetrapod fauna with South America. The biotic connectivity with Africa and South America resolved one of the greatest conundrums of Indian paleobiogeography, namely the lack of endemism among Late Cretaceous Indian tetrapods. The northward motion of the Indian plate is recorded from ocean magnetic anomalies and two spectacular linear-hotspot trails left by the Réunion and Kerguelen plumes, respectively, in the Indian Ocean since the Cretaceous. After the accretion of the OKL arc, the active subduction shifted farther north from the Indus suture to the Shyok suture. At the Cretaceous–Paleogene (K/PG) boundary, India was ground zero for two catastrophic events—the Shiva impact and the Deccan volcanism, which have been linked to the dinosaur extinction. At the same time, Seychelles was separated from India. During the Late Cretaceous (ca. 67 Ma), the Indian plate suddenly accelerated its motion to 20 cm/yr between two transform faults that facilitated the northward movement like the parallel tracks of a rail line—the Owen-Chaman fault on the west and the Wharton Ridge–Sagaing fault on the east. As a result, the Neotethyan plate, bordered by these two transform faults, became a separate oceanic plate called the Kshiroda plate. India continued acceleration during the Paleocene as a passenger ship with a mobile gangplank of the OKL arc, carrying its impoverished Gondwana fauna. During this time India exchanged tetrapod fauna with northern Africa and Europe via the Spain-Morocco corridor. Despite several decades of investigations, inferences of the timing and nature of collisions between India and Asia remain controversial. A pronounced global warming took place during the Paleocene–Eocene thermal maximum (PETM), when India collided with Asia; this warming caused significant tectonic changes and tetrapod radiation. India slowed down dramatically to 5 cm/yr during its initial collision, and its tetrapods underwent an explosive evolution in response to a new ecological opportunity resulting in the Great Indo-Eurasian Interchange. As India joined with Eurasia, Indian tetrapod fauna became highly diverse and acquired European heritage. The earliest clades of frogs, agamids, and several clades of placental mammals such as bats, artiodactyls, whales, perissodactyls, primates, and lagomorphs appeared abruptly on the Indian subcontinent in the paleoequatorial region during the Early Eocene and dispersed rapidly in the Holarctica province, thus strengthening the “out-of-India” hypothesis. Our results suggest that terrestrial faunas could have dispersed to or from Europe during the initial collision via the Kohistan-Ladakh arc corridor. The postcollisional tectonics during the Neohimalayan stage created the world’s highest, youngest, and most tectonically active mountain belt on Earth—the Himalayan Mountains–Tibetan Plateau. As India converged with Asia, the Nanga Parbat syntaxis (NPS) and Namcha Barwa syntaxis (NBS) functioned like two prongs of a rigid, V-shaped indentor that produced the uniform curvature of the Himalayan arc, squeezed the Tibetan block, and resulted in the formation of the Altyn Tagh and the Karakoram strike-slip faults. The channel-flow model explains a genetic relationship between the uplift of the Tibetan Plateau and the unusual metamorphic rocks of the Higher Himalaya. Such a drastic change in topography has fundamentally influenced regional and global climate. During the Neohimalayan tectonic uplift, the intensity of monsoon increased with exhumation of the Himalaya. A foreland basin developed in front of the Lesser Himalaya, where rich Siwalik vertebrates thrived in the floodplains of the Siwalik River from Miocene to Pleistocene mimicking the Serengeti ecosystem. The Siwalik megafauna suffered greatly during the Late Pleistocene extinction. The antecedent Himalayan rivers began to emerge along the Indus-Tsangpo suture zone in the early collision stage and modified in concert with the rise of the Himalaya. The present-day drainage systems of the Himalayan river systems were reorganized, fragmented, and rerouted from the ancient Siwalik River because of tectonic forces.

The Shigar Valley is crossed by a large, southwest-verging reverse fault containing pods of serpentinized ultramafic rock; the fault is correlated with the Northern Suture, which separates Paleozoic shelf-type sedimentary rocks of the Asian plate from Cretaceous volcanic rocks of the Ladakh-Kohistan arc. In the Shigar valley, the Asian plate is represented by a series of folded schists and marbles (the Daltumbore Formation) that is faulted southward over metasedimentary rocks and volcaniclastics (the Bauma-Harel Formation) belonging to the volcanic arc. Cretaceous turritellid gastropod fossils were found in the Bauma-Harel Formation. Metamorphism on both sides of the suture occurred in a regime of high temperature but only low to moderate pressure. Metamorphic isograds are cut by the suture, so metamorphism must have occurred before faulting along the suture. Two main phases of igneous intrusion are exposed in the arc terrane: a pre-tectonic, possibly tholeiitic phase about 100 m.y. old, and a post-tectonic, calc-alkaline to subalkaline phase 40 to 60 m.y. old. The Northern Suture does not have the appearance of a major suture, but the Ladakh-Kohistan Arc seems to have been a separate plate from the Asian continent. The suture probably marks the closure of a small ocean basin in late Cretaceous to early Eocene time.

The gneisses of the Nanga Parbat–Haramosh Massif (NPHM), Pakistan, experienced peak metamorphic temperatures in the interval from 25 to 30 Ma, as revealed by 40 Ar/ 39 Ar cooling ages of hornblende and the ages of the youngest intrusions of the Kohistan batholith located immediately adjacent to the NPHM. 40 Ar/ 39 Ar and fission-track mineral ages indicate that the postmetamorphic cooling history of the NPHM has been controlled over the past 5 to 10 m.y. by active tectonism associated with the Raikhot Fault, although passive uplift and erosion in response to overthrusting of the NPHM by the Kohistan Arc has been underway as well. Net cooling rates for NPHM gneisses exposed today along the Indus River at low elevations have accelerated, from 20°C/m.y. at ∼ 20 Ma to 300°C/m.y. at 0 to 0.4 Ma. Following emplacement of aplite dikes at about 30 to 35 Ma, portions of the Kohistan Batholith adjacent to the NPHM experienced cooling rates similar to the NPHM of about 20°C/m.y. over the period 25 to 10 Ma, but the net cooling rates for the batholith of ∼30°C/m.y. over the past 10 m.y. have been much lower than those experienced within the NPHM. Ion microprobe and conventional U/Pb analyses of zircon show that the protoliths for the Iskere Gneiss and the structurally lower Shengus Gneiss of the NPHM are, respectively, ∼1850 Ma and 400 to 500 Ma in age. Zircons from the Iskere Gneiss have thin, relatively high U rims that yield ages from 2.3 to 11 Ma. These rims indicate that metamorphism of the NPHM gneisses is Tertiary, not Precambrian, in age. The ages and Concordia systematics of analyses of Shengus Gneiss zircons suggest that this gneiss may be a metamorphosed equivalent of the Mansehra Granite and other Paleozoic S-type granites found throughout the Himalaya.

Neogene doming in the north-central Klamath Mountains, California, tilted the Rattlesnake Creek terrane, chiefly an ophiolitic mélange, exposing an oblique cross section through disrupted and metamorphosed oceanic crust and mantle. The deepest section of the tilted terrane, in the Kangaroo Mountain area near Seiad Valley, contains tectonic slices of ultramafic, mafic, and sedimentary rocks that were penetratively deformed and metamorphosed under upper-amphibolite- to granulite-facies conditions. This section, called the Seiad complex, is the ophiolitic basement of an accreted Mesozoic island arc, and its polygenetic history reflects the magmatic and tectonic processes that occur during island-arc construction and evolution. The presence of metarodingite and metaserpentinite, and the concordance of structural elements and metamorphic grade among all units of the Seiad complex, indicate that initial tectonic disruption of the ophiolitic suite occurred in the upper crust and subsequent penetrative deformation and metamorphism occurred under high-temperature conditions in the deep crust. Crustal granulite-facies metamorphism is indicated by two-pyroxene metagabbroic bodies and two-pyroxene metasedimentary paragneiss. Geothermobarometric data from garnet amphibolite and granulite-facies metagabbro within the ophiolitic suite yielded pressure and temperature conditions of ~5–7 kb and ~650–750 °C. Geochemical data from samples of granulite, amphibolite, and leucotrondhjemite suggest a supra-subduction origin, although there is significant variation among the amphibolite samples, indicating multiple magma types. Crosscutting, radiometrically dated plutons and the regional geologic context suggest that high-grade metamorphism and deformation of these disrupted ophiolitic rocks occurred in the Middle Jurassic (ca. 172–167 Ma). This time interval broadly corresponds with contraction along several regional thrust faults in the Klamath Mountains province and juxtaposition of the Rattlesnake Creek terrane with terranes to the east. A U-Pb zircon age of 152.7 ± 1.8 Ma on a sample of a crosscutting leucotrondhjemitic dike swarm and published 40 Ar/ 39 Ar hornblende age spectra of ca. 150 ± 2 Ma from amphibolite indicate that magmatism and an accompanying thermal flux continued to affect this region into the Late Jurassic. Compared with the deep-crustal sections of the well-studied Kohistan and Tal-keetna arc complexes, the widespread mélange character of the Rattlesnake Creek terrane (including the Seiad complex) is unique. However, ophiolitic rocks, including mantle ultramafic rocks, are common components in the basal parts of these classic arc crustal sections. Hornblende gabbro/diorite and clinopyroxenite in the Seiad complex may be small-scale melt conduits that fed middle- and upper-crustal components of the arc, analogous to the relationship seen in Kohistan between deep-crustal ultramafic-mafic bodies and mid-crustal magma chambers.

The Chilas Complex is a large mafic-ultramafic body closely associated with the Kohistan Arc sequence in the western Himalaya of northern Pakistan. The arc and the Chilas Complex occupy an area of 36,000 km 2 , bounded on the north and south by major sutures. The arc formed close to the margin of Eurasia in response to the northward subduction of neo-Tethyan ocean lithosphere in Late Jurassic to middle Cretaceous time, and consists of intra-arc sediments, calc-alkaline volcanics, and diorite-tonalite-granite plutons. At its base is the Chilas Complex, which extends for more than 300 km and which has a maximum width of 40 km. Most of the complex consists of massive (although locally layered) gabbro-norites, which comprise variable amounts of plagioclase (An 64-40 ), orthopyroxene (En 76-48 ), clinopyroxene (mg = 75-55), magnetite, ilmenite, ±quartz, ±K-feldspar, ±hornblende, ±biotite, ±rare scapolite. In the central part of the complex, near the base, there are minor discordant dikes and intrusive bodies as large as 5 km 2 of a dunite-peridotite-troctolite-gabbronorite-pyroxenite-anorthosite association that displays excellent layering, graded bedding, slump breccias, and syndepositional faults. These rocks contain olivine (Fo 94-71 ), relatively Mg-rich orthopyroxene (En 91-65 ), clinopyroxene (mg = 85-67), and calcic plagioclase (An 98-83 ), ±hornblende, ±chrome spinel, and ±pleonaste, and represent a more primitive magma batch emplaced into the base of the gabbro-norite magma chamber. The mafic complex is not an ophiolite. Rocks of the complex have more petrographic and compositional similarities with plutonic blocks from island arcs and with other major mafic complexes such as the Border Ranges Complex of Alaska and those from the Ivrea Zone in the Alps. Trace-element patterns of the gabbro-norites have marked negative Nb anomalies, positive Sr, Ba, and P anomalies, and high K/Rb ratios, features consistent with melting of a hornblende-bearing sub-arc mantle source. The Chilas Complex either represents the root zone magma chamber of the Kohistan island arc, or magma generated by diapirism in the early stages of intra-arc rifting during formation of a back-arc basin.