Mesozoic evolution of the eastern Pamir

We present field and analytical results from the Tashkurgan and Waqia valleys in the southeastern Pamir that shed new light on the tectonic evolution and terrane architecture of the region. Field mapping of metasedimentary and igneous units along the Tashkurgan and Waqia valleys in the Southeast Pamir, integrated with metamorphic petrology, garnet-biotite thermometry, and zircon U/Pb isotopic analysis, help identify major structures and terrane boundaries in the region, as well as compare structural units across the Miocene Muztaghata gneiss dome. South of the Muztaghata dome, the gently northwest-plunging synformal Torbashi thrust klippe juxtaposes amphibolite facies Triassic Karakul-Mazar terrane schist and gneiss structurally above (1) greenschist facies Triassic Karakul-Mazar terrane metasedimentary rock in the north, and (2) lower-amphibolite facies schist in the south that are interpreted to be Gondwanan-derived crust (Central or South Pamir terrane). Farther south, the Rouluke thrust fault imbricates the Gondwanan crust, placing early Paleozoic schists over Permian marble and slate. Exposure of the Torbashi thrust sheet terminates in the southeast, and with it the surface exposure of the Triassic Karakul-Mazar terrane, leaving the Paleozoic Kunlun terrane juxtaposed directly against Gondwanan terrane crust. Based on lithologic and isotopic similarities of units north and south of the Muztaghata gneiss dome, we document the existence of a regionally extensive thrust nappe that stretched across the northern and eastern Pamir, prior to being cut by Miocene exhumation of the Muztaghata dome. The thrust nappe links the Torbashi thrust in the southeast Pamir with the Tanymas thrust in the northern Pamir, and documents regionally extensive exposure of lithologically continuous units across the northeast Pamir. While timing of emplacement of the Torbashi thrust klippe and displacement on the Rouluke fault to the south is not well constrained, we interpret shortening to be Cretaceous in age based on previously published cooling ages. However, a component of Cenozoic shortening cannot be ruled out. A key observation from our mapping results is that the surface exposures of the Karakul–Mazar–Songpan Ganzi terrane are not continuous between western Tibet and the Pamir, which indicates tectonic and/or erosional removal, likely sometime in the Mesozoic. Furthermore, our documentation of the Jinsha suture in the southeast Pamir on the eastern side of the Karakoram fault shows deflections of terranes across the Himalayan-Tibetan orogen were not primarily accommodated along discrete, large displacement faults (>400 km) faults. Instead, oroclinal bending of the northern Pamir, and dextral shear along the Pamir margins, may be largely responsible for the northward deflection of terranes. the structural relationships and distribution of rock units in the southeast Pamir south of the Cenozoic Muztaghata gneiss dome. Our results show that the southeast Pamir consists of two distinct structural/lithologic domains. (1) To the north, Triassic amphibolite facies schist and gneiss intruded by Late Triassic–Early Jurassic granite lie structurally above Triassic greenschist facies phyllite, marble, and quartzite. These rocks constitute portions of the Triassic Karakul-Mazar terrane, which was imbricated during later crustal shortening. To the South, early Paleozoic greenschist facies metasedimentary and igneous rocks lie structurally above marbles slate. we along likely our results correlate portions Karakul-Mazar across northern and which have by Miocene suggesting the presence of a regionally thrust sheet across the northern which accommodated significant Cretaceous shortening.


INTRODUCTION
The Cenozoic Himalayan-Tibetan orogen is considered one of the type examples for understanding lithospheric deformation processes associated with continent-continent collision zones. One of the most pronounced aspects of the orogen is the strong northward deflection of topography and tectonic terranes at the western end of the orogen in the Pamir-Karakoram region (Fig. 1). However, it is unclear whether this deflection has been a result of distributed deformation, or displacement on discreet structures, notably the Karakoram fault, largely due to uncertainties in terrane boundary locations near to and west of the Karakoram fault. This has led to significantly different interpretations of how strain has been accommodated in the Cenozoic at the western end of the Himalayan-Tibetan orogen (Burtman and Molnar, 1993;Yin and Harrison, 2000;Lacassin et al., 2004;Schwab et al., 2004;Robinson et al., 2012). Estimates for the total right-lateral slip across the Karakoram fault range from ~160 to >500 km based on correlations of geologic units and interpreted sutures across the fault (Valli et al., 2007(Valli et al., , 2008Robinson, 2009b). A related challenge in understanding the evolution of the Himalayan-Tibetan orogen has been separating the effects of the Cenozoic collision between India and Asia from the protracted Mesozoic tectonic history consisting of the accretion of several discrete terranes to the southern margin of Asia.
Geologic age and lithologic correlations are generally not well defined for the northern and eastern Pamir. Older regional maps depict a wide range of age assignments that are not laterally continuous or easily correlated across the region, mostly due to the high metamorphic grade of the rocks and resulting lack of available biostratigraphic data (Yin and Bian, 1995;Strecker et al., 1995;Sobel and Dumitru, 1997;Schwab et al., 2004; Area of previously uncertain Terrane boundary relationships 38°7 6°80°3 6°0 200 km

T a n y m a s T h ru s t
? ?
? ?    Robinson et al., 2007). Recent detrital zircon studies in the northern Pamir have shed significant light on the matter, demonstrating that many of the units in the region are Triassic rather than Proterozoic to Paleozoic, and are therefore part of the Karakul-Mazar terrane, equivalent to the Songpan-Ganzi terrane to the east Yang et al., 2010;Robinson et al., 2012;Schwab et al., 2004). Further, detrital zircon and whole rock Nd isotopic analyses provide evidence for correlating units of the northern with the eastern portions of the Pamir across the Muztaghata gneiss dome (Robinson et al., 2012). However, recent studies of ages of intrusive igneous bodies have suggested that much of the eastern Pamir is part of a continuous Cambrian terrane, the Bulunkuole Group, which collided with Tarim basin in the early Paleozoic (Zhang et al., 2018a(Zhang et al., , 2018b. While progress has been made in understanding the correlative relationships among units in the northern and eastern Pamir, several questions remain, including the following.

R e d R iv e r F a u lt
(1) What is the structural and terrane architecture of the southeastern Pamir east of the Karakoram fault, and how does it compare with other portions of the Pamir? (2) How do metasedimentary rocks in the southeast Pamir correlate with metasedimentary rocks in the northern Pamir? (3) Where is the boundary between Gondwanan crustal fragments and the Paleozoic-early Mesozoic southern boundary of Asia (i.e., the Tanymas-Jinsha suture), in the eastern Pamir? Addressing these questions is crucial for better understanding the tectonic history of the Pamir region and the greater Himalayan-Tibetan orogen. We present results from detailed field mapping, integrated with petrologic and geochronologic data, which provide new insights into the structural relationships and distribution of rock units in the southeast Pamir south of the Cenozoic Muztaghata gneiss dome. Our results show that the southeast Pamir consists of two distinct structural/lithologic domains. (1) To the north, Triassic amphibolite facies schist and gneiss intruded by Late Triassic-Early Jurassic granite lie structurally above Triassic greenschist facies phyllite, marble, and quartzite. These rocks constitute portions of the Triassic Karakul-Mazar terrane, which was imbricated during later crustal shortening.
(2) To the South, early Paleozoic greenschist facies metasedimentary and igneous rocks lie structurally above low-grade Permian marbles and slate. These rocks constitute Gondwanan crust of the Central or southern Pamir that we interpret to have been imbricated along SW-vergent thrusts, likely during the Cretaceous. Further, our results correlate portions of the Karakul-Mazar terrane across the northern and eastern Pamir which have been cut by the Miocene Muztaghata gneiss dome, suggesting the presence of a regionally extensive (i.e., 100s of km) thrust sheet across the northern Pamir, which accommodated significant Cretaceous shortening.

Tectonic Setting
The Pamir form an arcuate region of elevated topography at the northwest corner of the India-Eurasia collision zone (Fig. 1), bound to the north by the Tien Shan and Alai Basin, to the west by the Tajik Basin, to the east by the Tarim Basin, and to the south by the Karakoram Mountains. Significant crustal thickening during the Cenozoic, as revealed by Oligocene-Miocene prograde metamorphism recorded in Miocene gneiss domes across the region, has resulted in a present-day crustal thickness of ~70 ± 10 km (Negredo et al., 2007;Mechie et al., 2011;Schmidt et al., 2011;Mechie et al., 2012;Sippl et al., 2013, Stübner et al., 2013a., 2013bStearns et al., 2015;Rutte et al., 2017a;Hacker et al., 2017).
The Himalayan-Tibetan orogen consists of a series of distinct terranes accreted to the southern margin of Asia prior to the Cenozoic India-Asia collision (Yin and Harrison, 2000). In Tibet, the Jinsha suture separates terranes of Asian affinity from accreted terranes derived from Gondwana.
Terranes in the Pamir include, from north to south, the Paleozoic Kunlun and Permian-Triassic Karakul-Mazar terranes (often grouped together as the North Pamir), the Central and southern Pamir, and the Karakoram terrane (Fig. 1). The Paleozoic Kunlun terrane is bound in the north by the Main Pamir thrust and with the Karakul-Mazar terrane to the south (Fig. 1). The Permian-Triassic Karakul-Mazar terrane, correlative with the Songpan Ganzi-Hoh Xil terrane of Tibet (Schwab et al., 2004;Robinson et al., 2012;Xiao et al., 2002a), is exposed throughout the northern Pamir, but thins to the west (Schwab et al., 2004). The eastern continuation of the terrane toward the Western Kunlun is unclear. This terrane is bounded to the south by the north-dipping Tanymas-Torbashi thrust faults, which form the boundary between Asian affinity terranes to the north and Gondwanan affinity terranes to the south. The Gondwanan Central Pamir terrane is exposed between the Tanymas thrust fault and the Rushan-Pshart fault zone, and has been interpreted to be equivalent to the Qiangtang terrane of Tibet (Burtman and Molnar, 1993;Lacassin et al., 2004;Schwab et al., 2004;Robinson, 2015) or a distinct terrane (Robinson et al., 2009(Robinson et al., , 2012. The south Pamir and Karakoram terranes to the south are separated by the Wakhan-Tirich Boundary Zone, with both terranes interpreted to be equivalent to the Qiangtang terrane to the east (Robinson et al., 2012;Angiolini et al., 2013;Zanchi et al., 2000;Zanchi and Gaetani, 2011).
During the Cenozoic India-Asia collision, the Pamir has been interpreted to have been translated northward ~300 km via the Main Pamir thrust since ca. 25-16 Ma (Burtman and Molnar, 1993;Sobel and Dumitru, 1997), although several recent studies suggest more limited displacement (Chapman et al., 2017;Chen et al., 2018). Displacement was possibly due to rollback of the Alai slab in the north and west, and accommodated in the east by the right-slip Kashgar Yecheng Transfer System (Cowgill, 2010;Sobel et al., 2013) or driven by the arrival and underthrusting of Indan lithosphere (Kufner et al., 2016;Liao et al., 2017). While Pamir indention during the Cenozoic resulted in northward deflection of terranes and sutures relative to positions in Tibet ( Fig. 1), it is unclear the extent to which these deflections have been accommodated by displacement on discreet structures (i.e., the Karakoram fault) or distributed deformation between Tibet and the Pamir-Karakoram region.

Tanymas Thrust Fault
The Triassic Tanymas thrust fault strikes approximately east-west and dips to the north forming the boundary between the Karakul-Mazar terrane and the Central Pamir (Figs. 1, 2). Slip data long the fault trace yield top-to-the-south thrusting accompanied by dextral slip in the Paleogene (Rutte et al., 2017b). Along the central and eastern portion of the Tanymas thrust fault, the hanging wall consists of high-grade Triassic metasedimentary rocks intruded by Triassic granitoids (Schwab et al., 2004;Robinson et al., 2004Robinson et al., , 2012Rutte et al., 2017b). The footwall of the Tanymas thrust consists of siliciclastic and carbonate metasedimentary rocks of latest Proterozoic to Cretaceous age; units that are part of the Tuzguny-Terezki and Zartoshkol thrust sheets (Schwab et al., 2004;Rutte et al., 2017a). South of the Tuzguny-Terezki and Zartosckol thrust sheets in the Central Pamir are an antiformal stack of thrust fault blocks with the Central Pamir gneiss domes exposed in the core (Rutte et al., 2017b). Structural analysis of fault blocks in the footwall of the Tanymas thrust indicates Cenozoic shortening was at least 80 km (Burtman and Molnar, 1993) to >95 km (Rutte et al., 2017b) in the Central Pamir. The Tanymas thrust fault is cut by the North Muskol shear zone (NMSZ) at its eastern end. The NMSZ is the northern boundary of the Shatput and Muztaghata gneiss domes.

Baoziya Thrust Fault
The Baoziya thrust fault in the northeast Pamir is exposed in the hanging wall of the Kongur Shan Extensional System, northwest of the Muztaghata gneiss dome (Figs. 1, 2). The fault strikes approximately north-south with a moderate west-dip. Shear sense indicators show a top-to-the-south shear sense (Pan, 1992;Robinson et al., 2004), with the current west-dipping geometry interpreted to be a result of rotation of the structure during subsequent regional deformation. The Baoziya thrust fault juxtaposes Triassic amphibolite facies metasedimentary rocks of the Karakul Mazar terrane in the hanging wall against greenschist facies calcareous quartzite, slate, and marble in the footwall, which are also interpreted to be part of the Triassic Karakul-Mazar terrane (Robinson et al., 2004(Robinson et al., , 2012.

Torbashi Thrust Fault
The Torbashi thrust fault is located south of the Muztaghata gneiss dome (Figs. 2, 3). Identification of the Torbashi thrust fault in the field is based on amphibolite facies schist and gneiss structurally overlying greenschist facies phyllite, marble, and quartzite (Robinson et al., 2007). The thrust has been previously mapped as a synformal klippe, with a northwest-trending fold axis and a top-to-the west to southwest sense of shear. Biotite cooling ages of the hanging wall rocks suggests the age of the thrust is Late Cretaceous (Robinson et al., 2007).

Muztaghata Gneiss Dome
The Muztaghata gneiss dome is the easternmost of a series of Miocene domes that crop out along the Central Pamir ( Fig. 2; Robinson et al., 2007;Schmidt et al., 2011;Cao et al., 2013;Smit et al., 2014). The core of the Muztaghata dome consists of Triassic orthogneiss and schist (Trgn) (Figs. 2, 3), the latter of which have detrital zircon signatures that are interpreted to be equivalent with the Karakul-Mazar terrane (Robinson et al., 2012). In the southern portion of the Muztaghata dome, the Kexi-Liqi thrust fault separates Triassic rocks in the footwall from early Paleozoic rocks (Pz 1 ) in the hanging wall (e.g., the Shen-Ti klippe of Rutte et al., 2017a). The Paleozoic rocks have detrital zircon signatures indicating a Gondwanan affinity with the Kexi-Liqi fault interpreted to represent the Paleo-Tethys suture zone (Yang et al., 2010;Robinson et al., 2012). Schist and gneiss in the Muztagh Ata dome record an Early Jurassic metamorphic event, overprinted by late Oligocene-early Miocene upper-amphibolite to granulite facies conditions (0.9-1.0 GPa, 700 °C−750 °C), with rapid cooling and exhumation in the middle to late Miocene (Robinson et al., 2007;Cao et al., 2013;Cai et al., 2017).
The Muztaghata dome is bound by the North Muskol shear zone on its northern margin, the Shen-Ti fault on its southern margin, and the Kuke fault on its eastern margin (Robinson et al., 2007;Sobel et al., 2011;Rutte et al., 2017b). The Shen-Ti fault generally strikes east-west, parallel with the axis of the dome. The fault separates migmatitic schist and quartzofeldspathic gneiss of the Muztaghata dome in the footwall (Tr gn and Pz 1 ) from greenschist facies metasedimentary rocks in the hanging wall (Tr sch2 ). Ductile shear sense indicators south of Muztaghata show top-to-the-south transport of the hanging wall (Robinson et al., 2007). The southern trace of the Shen-Ti fault connects with the Waqia fault (Imrecke, 2013) to the southeast, where it juxtaposes Ordovician meta-chert (O u ) and Paleozoic granite in the footwall (Pz i ) against Neogene sediments (Cz) with ~1 km thickness in the hanging wall (Imrecke, 2013).
The north-striking Kuke fault bounds the eastern margin of the Muztaghata dome, and forms a sub-vertical shear zone at its southern end (Sobel et al., 2011;Xinjiang Bureau of Geology and Mineral Resources, 1993). The steep fault dip and thermochronology data in the Muztaghata antiform suggest rotation of the fault to the present vertical orientation from a shallower orientation during its evolution from 12 to 6 Ma (Sobel et al., 2011). The Kuke fault trace is truncated by the Shen-Ti-Waqia fault at a high angle in the south (Robinson et al., 2007;Sobel et al., 2011). In the north, the Kuke fault is interpreted to link with the Ghez fault, which juxtaposes Kunlun terrane structurally above Karakul-Mazar terrane along the eastern Kongur Shan gneiss dome (Robinson et al., 2012).

Kongur Shan Extensional System
The prominent Kongur Shan Extensional System is an active 250-kmlong north-south-striking system of normal faults that accommodate eastwest extension along the eastern margin of the Pamir ( Fig. 2; Brunel et al., 1994;Arnaud et al., 1993;Robinson et al., 2004). A minimum of 34 km east-west extension is recorded in the area of Kongur Shan initiating at ca. 7-8 Ma (Robinson et al., 2004(Robinson et al., , 2010. Offset across the Kongur Shan Extensional System decreases toward the south, with ~8 km of eastwest extension near the anticlinal axis of Muztaghata initiating at 6-5 Ma, documenting south directed propagation of the system (Robinson et al., 2007;Robinson, 2010;Cao et al., 2013). South of Muztaghata, extension is transferred to the east-dipping Tashkurgan fault in the western portion of the study area (Robinson et al., 2007).

GEOLOGIC MAPPING
Structural field mapping was conducted in the Tashkurgan and Waqia valleys at 1:100,000 scale using ASTER images as map bases (Fig. 3). In the following sections, we describe the lithological and structural characteristics along north-to-south transects in the Tashkurgan and Waqia valleys.   Torbashi T a s h k u r g a n V a l l e y T a s h k u r g a n V a l l e y AR-9-10-03-7 AR-9-10-03-4 AR-4-28-00-11a AR-9-13-03-3 AR-9-13-03-3 T a s h k u r g a n f a u l t K u k e f a u l t T a s h k u r g a n f a u l t K u k e f a u l t Trend and Plunge of Minor Fold Axis

567
IMRECKE ET AL. | Mesozoic evolution of the eastern Pamir

Lithologic and Structural Observations in the Tashkurgan Valley
The northernmost section of the Tashkurgan valley is bound to the north by the south-dipping Shen-Ti fault, which forms the southern boundary of the Miocene Muztaghata gneiss dome (Figs. 2, 3). The Shen-Ti fault separates upper amphibolite facies (garnet + kyanite + K-feldspar) migmatitic schist and gneiss of the Muztaghata dome in the footwall from greenschist facies phyllite, quartzite, marble, and minor garnet-biotite schist (biotite + garnet + chlorite ± white mica) in the hanging wall (Fig. 4A). Garnets in hanging wall schist are subidioblastic and show strain shadows indicating deformation after peak metamorphism (Fig. 5A). Coexisting garnet and chlorite suggest maximum temperatures of 400-500 °C (Spear, 1995;Figs. 5A, 5B). Foliations in both the hanging wall and footwall strike northeast to northwest and dip moderately to the south (Fig. 3).
Farther south near the town of Gedeben (Fig. 3), metamorphic grade decreases over a short distance indicating a significant geologic contact. To the north, rocks are amphibolite-facies (Tr sch1 ), whereas to the south rocks are greenschist facies micaceous quartzite, marble, and schist (Pz 2 , Fig. 3) intruded by a large granitic sill. Peak metamorphic assemblages south of the contact are biotite + garnet (Figs. 5E, 5F, 5G). Quartz deformation consists of bulging and sub-grain rotation (SGR) deformation in deformed granites and the microlithons in quartz rich lithologies (Figs. DR1c-DR1e), indicating peak temperatures of at least ~450-500 °C (Stipp et al., 2002). Foliations south of the contact continue to strike northwest and dip northeast (Fig. 3). The consistent orientation of foliations above and below the contact and juxtaposition of high metamorphic grade rocks above low metamorphic grade suggests the contact is a thrust fault, which we interpret as the southern trace of the Torbashi thrust fault. The general trend of metamorphic fabrics can be distinguished in satellite imagery, which indicates the fault trace continues to the southeast. We note that our observations significantly shift the southern trace of the Torbashi thrust fault to ~20 km north of its previously interpreted position (compare to Robinson et al., 2007).
Continuing south, biotite-garnet schist and quartzite with northeast dipping foliations (Pz 2 ) structurally overlie low metamorphic grade slate and marble with north-dipping foliations (Pz 3 ) (Fig. 3). Top-to-the-southwest shear sense is recorded in S-C fabrics and σ-type porphyroblasts near the contact in Pz 2 , and tension gashes in the structurally lower Pz 3 (Fig. 4C). Lithologic relationships indicate this contact is a north-dipping thrust fault, which we name the Rouluke fault. The western continuation of the Rouluke fault is truncated by the younger Tashkurgan normal fault while the eastern continuation of the fault is unclear.
Rocks in the footwall of the Rouluke fault consist of slate, phyllite, quartzite, and marble (Pz 3 ) (Figs. 3, 5H, 5I). The marble is granoblastic and shows Type II twins (tabular and thick twins) indicating maximum metamorphic temperatures of 150-250 °C ( Fig. 5I; Groshong, 1988;Burkhard, 1993;Ferrill et al., 2004). Visible strain of quartz grains is limited to the presence of linearly oriented inclusions, which exhibit no evidence of dynamic recrystallization (e.g., bulging) or dislocation creep. Peak metamorphic mineral assemblages are limited to post-tectonic biotite. One sample, a biotite-and calcite-bearing quartzite (DI-7-15-10-4) has biotite growth across weakly defined foliation boundaries. Foliation consisting of slaty cleavage generally strikes northwest and dips northeast (Fig. 4F). Weak continuous foliation and cataclastic microtextures are present in the silicate-rich samples. As previously indicated, tension gashes in slate show top-to-the southwest sense of shear.

Lithologic and Structural Observations in the Waqia Valley
The northernmost section of the Waqia valley is bound to the north and northeast by the eastward trace of the south-dipping Shen-Ti fault (Figs. 2, 3), which connects with the southeast dipping Waqia normal fault bounding the northeastern side of the Waqia valley along the southern trace (Figs. 2, 3). Rocks south and west of the Shen-Ti and Waqia faults consist of schist, phyllite, quartzite, and marble, similar to the Tashkurgan valley (Tr sch2 , Fig. 3). Peak metamorphic assemblages consist of biotite + garnet + chlorite ± white mica (Figs. 5A, 5B). In biotite schist samples, quartz displays bulging and sub-grain rotation deformation along grain boundaries of quartz-rich microlithons. Foliations in Tr sch2 generally strike east-west along and near the Tashkurgan River, but change to a northeast strike parallel to the trend of the Waqia valley (Fig. 3), and generally dip southwest, although several map-scale upright folds are present in the unit. Shear-sense indicators (S-C fabrics and asymmetric boudins) along the Tashkurgan valley show top to the south shear-sense (Robinson et al., 2007). Farther south in the Waqia valley mineral elongation lineations in Tr sch2 are oriented northeast with top-to-the-south sense of shear recorded by asymmetric kink folds in phyllite (Fig. 5B). A notable characteristic of Tr sch2 is the absence of igneous intrusions, although a meta-rhyolite layer is present along the Tashkurgan River (age of 228 ± 2 Ma; Zhang et al., 2007).
The Torbashi thrust fault strikes east-west along the northern end of the Tashkurgan valley before changing to a southeast strike along the northern end of the Waqia valley (Fig. 3). Continuing to the south, the fault trace strikes parallel to the axis of the Waqia valley through most of the valley (Fig. 3); the fault contact is generally buried beneath Neogene and Quaternary sedimentary rock and sediment (Figs. 3, 4E). Field evidence of the approximate location of the contact is indicated by significantly different metamorphic grades on either side of the valley (Tr sch1 versus Tr sch2 in Fig. 3) and radiometric ages obtained from samples discussed in subsequent sections. On the northeast side of the Waqia valley, metasedimentary rocks consist of greenschist facies phyllite, described above, while along the southwest side of the valley, metasedimentary rocks consist of upper amphibolite facies schist and gneiss (Figs. 5G, 5H). As along the Tashkurgan valley, high-grade schists and gneisses contain garnet + biotite + sillimanite indicating upper amphibolite peak metamorphic conditions (>700 °C) ( Fig. 5D; Robinson et al., 2007;Yang et al., 2010;this study). Sillimanite displays a fibrolite habit (Fig. 5D)  IMRECKE ET AL. | Mesozoic evolution of the eastern Pamir RESEARCH pressure growth of sillimanite from muscovite, less than or equal to 0.65 GPa, which is consistent with a reasonable 35 °C/km geothermal gradient. Grain boundary migration deformation fabrics present in quartz-rich microlithons are overprinted by bulging fabrics in places. The hanging wall of the Torbashi thrust is intruded by multiple granitic sills and dikes, most of which are pervasively sheared. An exposure of high-grade schist lies structurally above the greenschist facies Tr sch2 along the eastern margin of the valley in the immediate hanging wall of the Waqia normal fault (Figs. 2, 3). The metamorphic rocks are identical to the high-grade rocks of the Torbashi fault hanging wall, and are interpreted as part of the Torbashi thrust hanging wall (Tr sch1 ).
Farther south, small exposures of quartzofeldspathic gneiss, amphibolite, and quartzite intruded by undeformed to weakly foliated granite are present (Pz aq ). However, the contact with the greenschist facies rocks to the north is buried under Neogene and Quaternary sediments (Fig. 3).
At the southern end of the Waqia valley the southern exposure of the Torbashi thrust fault trace continues southeast from the Tashkurgan valley. The footwall of the thrust consists of similar greenschist facies schists and quartzites with northeast-dipping foliations as observed in the Tashkurgan valley. At the southern end of the Waqia valley, the exposed width of the hanging wall of the Torbashi fault tapers to the southeast where we interpret it to end. The map pattern of the Torbashi fault hanging wall therefore defines a gently northwest-plunging synformal klippe whose exposure terminates to the southeast.

Methods
To quantify peak metamorphic temperatures, and possible gradients, along the southwestern footwall of the Torbashi thrust fault, two samples were analyzed using the garnet-biotite Fe-Mg exchange thermometer. Electron Microprobe analyses were conducted at the Texas A&M University microprobe facility using a 1 µm spot size with a 15 kV accelerating voltage, a 20 nA beam current, and a 30 second counting time for each element. Activities of garnet and biotite end-members and equilibrium Clapeyron slopes were calculated using the programs AX and Thermocalc 2.7 (Powell et al., 1998). Electron microprobe traverses were used to identify zoning patterns in garnets and interpret the temperature results.
Average activities of matrix biotite end-members (phlogopite and annite) were used because (1) there was little variation in matrix biotite analyses (Table DR2), (2) using the highest and lowest biotite activities resulted in temperature differences of <40 °C, and (3) biotite activity variations did not correlate with distance from garnet grains. Pyrope activities (a mg ) are not interpreted for compositional variation due to the low values in the analysis (Table DR2). Temperatures are reported based on a pressure of 0.4 GPa, based on reasonable geothermal gradient of 30 °C/km (Spear, 1995).

AR-940-09
Sample 940-09, a quartz-rich garnet-biotite schist (Fig. 5G), was collected from the northern hanging wall of the Rouluke fault (Figs. 3, 5G). The primary metamorphic mineral assemblage is garnet + biotite + quartz. Garnets are subidioblastic to xenoblastic and appear to be late syn-tectonic to post-tectonic, evidenced by consistent foliation internal and external to garnet grains and the lack of strain shadows.
Electron microprobe transects were completed across three garnet grains in AR 940-09. Garnet transect 1 shows weak zoning in Mn with increasing concentrations toward the center (inversely related to Ca) while Fe and Mg# is invariable across the garnet grain (Fig. 6A). Garnet transect 2 shows similar weak zoning (Fig. 6B), with a decrease in Mn within <500 μm from the rims, and no clear zoning in Fe, Ca, and Mg#. Garnet transect 3 shows no evidence of consistent compositional zoning (Fig. 6C). Garnet-biotite thermometer results from the garnet rim analyses from transect 1 yield a temperature of ~560 °C, which we interpret to represent the peak metamorphic temperature of this sample.

AR-9-10-03-1
Sample AR-9-10-03-1, a garnet-biotite schist (Fig. 5F) was collected from the southern hanging wall of the Rouluke thrust fault (Fig. 3). Petrographic analysis of garnet grains demonstrates syn-tectonic growth characteristics, indicated by rotated inclusion trails. An electron microprobe transect across a single large garnet grain displays strong compositional zoning in Mn and Fe (Fig. 6D) with a bell-shaped curve in the Mn profile from core to rim, which is characteristic of preserved growth zoning during prograde metamorphism (Hollister, 1966). The Fe and Mg# profiles show an inverse relationship to Mn across the grain. Using the garnet-biotite thermometer from rim analyses of garnet and an average of matrix biotite analyses yields a temperature of ~480 °C, which we interpret to represent the peak metamorphic temperature of this sample.

Summary of Results
Peak temperatures in the footwall of the Torbashi thrust fault are significantly lower (upper greenschist to lower amphibolite facies) than peak temperatures indicated by metamorphic assemblages in its hanging wall (upper amphibolite to granulite facies; Yang et al., 2010), consistent with our mapped location of the southern trace of the Torbashi fault. The garnetbiotite thermometry results from the lithologies between the Torbashi and Rouluke faults shows a gradient in peak metamorphic temperatures from lower amphibolite to upper greenschist facies from north to south. Temperatures in the northern samples are slightly higher than would be expected given the quartz deformation fabrics (i.e., Stipp et al., 2002), suggesting that peak metamorphic temperatures post-dated deformation, consistent with the syn-to post-tectonic garnet inclusion patterns of AR-940-09.

Methods
Zircon grains from igneous and metasedimentary rock samples were separated using conventional techniques, mounted in epoxy, and polished to expose grain interiors. U-Pb analyses were performed at the University of Houston using a Varian 810 quadrupole inductively coupled plasma− mass spectrometer (ICP-MS). Laser ablation analyses were performed with a Cetac LSX-213 laser with a 50-20 nm beam diameter at 15 Hz and a Photon Machines Analyte 193 excimer laser with a 20 μm beam diameter at 10 Hz. Isotope abundances of 201 Hg, 202 Hg, 204 (Hg + Pb), 206 Pb, 207 Pb, 208 Pb, 232 Th, and 238 U were measured. On-peak backgrounds were measured for 20 s with the laser off prior to each 20 s of sample ablation. The first two seconds of sample and standard ablation are not included in the measured sample data to avoid surficial contamination. Instrumental fractionations of U/Pb and 206 Pb/ 207 Pb during analysis were calibrated by analyzing fragments of zircon standard Plesovice (all age uncertainties are expressed as 2σ) in between 5 consecutive unknown analyses following methods outlined in Shaulis et al. (2010). Analyses of zircon standard FC5z (1096.2 ± 1 Ma) or Peixe (564 ± 4 Ma) (Shaulis et al., 2010) were conducted after every 10 unknown analyses to monitor 571 IMRECKE ET AL. | Mesozoic evolution of the eastern Pamir RESEARCH the calibration. Analyses were not corrected for common Pb because the 204 Pb signal after 204 Hg background subtraction was below instrument detection. Age probability plots were created using Isoplot 4.1 (Ludwig, 2009). Detrital zircon probability plots exclude isotopic ratios with a discordance of >30% and < −5%. Ages >1.0 Ga are reported as 206 Pb/ 207 Pb ages and zircon ages <1.0 Ga are reported as 206 Pb/ 238 U ages. Maximum depositional ages are reported as the calculation of the weighted mean age of the youngest three overlapping zircon ages in the sample. Igneous ages are reported from the calculation of 206 Pb/ 238 U ages using the Isoplot TuffZirc age extractor (Ludwig, 2009).

Sample AR-9-10-03-4
Sample AR-9-10-03-4 is from a weakly mylonitic granite that intrudes upper greenschist to lower amphibolite facies schist (Pz 2 ) in the southern footwall of the Torbashi thrust fault along the eastern Tashkurgan valley (Figs. 2, 3). The granite forms an ~1-km-thick sill, sub-parallel to the regional foliation and continues to the southeast for at least 5 km. Thirty zircons were analyzed, which yielded 24 concordant analyses. The concordant analyses yield an Early Ordovician TuffZirc age of 479.0 +14.2 / −2.9 Ma from a coherent group of 20 grains (Fig. 7A).

Sample DI-7-25-10-17
Sample DI-7-25-10-17 is from a biotite hornblende granite (Pzaq) that lies in the southeastern footwall of the Torbashi thrust fault in the Waqia valley (Figs. 2, 3). 30 zircons were analyzed, with 13 yielding concordant analyses. Zircon grains from this sample were relatively U-poor (average of ~60 ppm). The concordant analyses have a spread in age from 303 to 334 Ma, and yields a Carboniferous weighted mean age of 315.9 ± 6.0 Ma from a coherent group of 10 analyses (Fig. 7B).

Sample AR-9-10-03-7
Sample AR-9-10-03-7 is from a garnet-bearing, weakly to non-foliated biotite granite sill that intrudes upper amphibolite facies schist and gneiss (TRsch1) in the hanging wall of the Torbashi thrust fault in the Tashkurgan 572 IMRECKE ET AL. | Mesozoic evolution of the eastern Pamir RESEARCH valley (Fig. 2, 3). 45 zircons were analyzed, with 14 yielding concordant analyses. The concordant analyses have a spread in ages from 188 to 220 Ma, and yields an Early Jurassic TuffZirc age of 197.6 +3.3 / -4.8 Ma from a coherent group of 11 analyses (Fig. 7C).

Figure 7. Results from zircon U-Pb analyses of igneous rocks sampled within the study area. (A) Concordia diagram for zircons from a weakly foliated granite in Pz 2 (Rouluke fault hanging wall). Inset is a U/Pb cumulative probability diagram of concordant zircon ages showing early Paleozoic peaks. (B) Concordia diagram for zircons from a granite in Pz aq (northeastern Torbashi fault footwall). Inset is a U/Pb cumulative probability diagram of concordant zircon ages showing Triassic and Jurassic peaks. (C) Concordia diagram for zircons from a granite in Trsch 1 (Torbashi fault hanging wall). Inset is a U/Pb cumulative probability diagram of concordant zircon ages showing Triassic and Jurassic peaks. (D) Concordia diagram for zircons from a granite in Trsch 1 (Torbashi fault hanging wall). Inset is a U/Pb cumulative probability diagram of concordant zircon ages showing Triassic and Jurassic peaks. Sample ages and locations are on Figures 2 and 3.
573 IMRECKE ET AL. | Mesozoic evolution of the eastern Pamir RESEARCH zircons overlapping in age yield an Early Devonian maximum depositional age of 413 ± 17 Ma (Fig. 8) Gehrels et al., 2011), and the previously published Permian age for the phyllite is likely correct (e.g., Yin and Bian, 1995).

Sample AR-4-27-00-9
Sample AR-4-27-00-9 is a chlorite-bearing quartz-rich schist from the footwall of the Baoziya thrust fault north of Muztaghata (Fig. 2). 136 zircons were analyzed with 73 yielding concordant ages. Conservative analysis of the maximum depositional age using the three youngest zircons overlapping in age yields a Permian maximum depositional age of 268 ± 11 Ma (Fig. 8). However, two concordant grains in this sample with 218 ± 5.4 and 234 ± 6.2 Ma ages suggest the sample may be Late Triassic. Prominent Proterozoic age populations occur at ca. 790, ca. 1600-2100, and ca. 2500 Ma. Significant Phanerozoic age populations occur at ca. 270, ca. 335, and ca. 410 Ma.

Summary of Results
Rocks in the hanging wall of the Torbashi thrust yield Late Triassic to Early Jurassic ages based on: (1) crystallization age of igneous rocks (AR-9-10-03-7 is 197.6 +3.3 / -4.8 Ma and AR-4-28-00-11a is 205.6 +3.3 / -4.8 Ma), and (2) a Late Triassic maximum depositional age of 228 Ma for the intruded metasedimentary rocks (Yang et al., 2010). Our results support previous interpretations that the hanging wall of the Torbashi thrust are part of the Karakul-Mazar terrane (Robinson et al., 2012). These igneous ages are some of the younger ages within the Karakul Mazar terrane, but are consistent with magmatism associated with the closure of the Paleo-Tethys Ocean.
Farther south near the town of Gedeben in the southern Tashkurgan valley, zircon U-Pb ages are different than those in the north. The crosscutting Early Ordovician granite (AR-9-10-3-4) indicates that the southern footwall metasedimentary rocks of the Torbashi thrust fault are early Paleozoic (Fig. 3), consistent with recently published Cambrian ages from gabbros and meta-rhyolites within the formation (Zhang et al., 2018a;2018b) where the lithologies are part of the previously named Bulunkuole Group. This age is common in sediments from Gondwanan terranes in the Tibetan plateau, and we therefore interpret the block to be Gondwanan in origin, although others have interpreted these lithologies as part of the Western Kunlun Orogenic Belt (e.g., Zhang et al., 2018a). Detrital zircon results from Permian phyllites south of the Rouluke fault (AR-6-26-00-7) also yield a typical Gondwanan signature, as compared with detrital zircon results from Central Pamir metasedimentary rocks and the Qiangtang terrane (Figs. 2, 9, and 10C;Robinson et al., 2012;Rutte et al., 2017b).
Carboniferous granite intrudes gneisses to the east of the Torbashi klippe in the southern Waqia valley. This age is characteristic of the Carboniferous-Early Permian arc magmatism that is prevalent in the Western ? ?

RESEARCH
Kunlun terrane, and is interpreted to be part of the Sailiyak arc, which developed prior to the Mazar subduction complex of northwestern Tibet (Xiao et al., 2002a(Xiao et al., , 2002b. Detrital zircon results from similar lithologies in the northern footwall of the Torbashi thrust fault and footwall of the Baoziya thrust north of Muztaghata are strikingly similar (AR4-27-00-9 and AR4-28-00-7). We compared the two samples using aKolmogorov-Smirnov test, which yielded a P-value of 0.539 (Fig. 10A). As P-values greater than 0.05 are considered to represent populations that are highly similar, we interpret these results to indicate that the units on either side of the Muztaghata dome are equivalent, which is consistent with previous results based on Nd isotopic values (Robinson et al., 2012). These units are interpreted to be Late Triassic, based on the maximum depositional age of AR4-28-00-7, and the similarity in age distributions to the Triassic Karakul-Mazar terrane (Robinson et al., 2007), with strong Permian-Triassic and early Paleozoic age populations (Fig. 10B). We interpret these results to indicate these samples are part of the Triassic Karakul-Mazar terrane.

Tectonic Terrane Boundaries and Structural Relationships in Tashkurgan and Waqia Valleys
Geologic structures, zircon geochronology, and metamorphic petrology yield key information regarding the architecture, structural relationships, and deformation history of the southeastern Pamir. In the study area of the Tashkurgan and Waqia valleys, the rocks can be divided into three groups based on age, degree of metamorphism, and interpreted terrane affinity. Metamorphic and igneous rocks in the study area are separated by major fault systems in the study area.
The first group of rocks lies along the northern section of the study area south of the Shen-Ti fault and Muztaghata dome and consists of two separate Triassic metasedimentary units separated by the Torbashi thrust fault: (1) a lower greenschist facies unit in the Torbashi thrust footwall, Tr sch2 , with a Late Triassic maximum depositional age (ca. 212 Ma, Fig.  8) and a pronounced lack of intrusive igneous bodies; and (2) and an structurally higher upper amphibolite facies unit in the Torbashi thrust hanging wall, Tr sch1 , with a Triassic maximum depositional age (Yang et al., 2010;Zhang et al., 2007) intruded by Late Triassic-Early Jurassic igneous rocks (Fig. 8). Our zircon results, integrated with previous work, indicate these units are part of the late Permian-Triassic Karakul-Mazar terrane that formed as part of an arc-accretionary complex system along the southern margin of Asia during subduction of the Paleo-Tethys Ocean (Schwab et al., 2004, Robinson et al., 2012Xiao et al., 2002a), equivalent to the Songpan Ganzi-Hoh Xil terrane of Tibet (Fig. 10B). The Torbashi thrust between the two units suggests imbrication of the terrane, with previous thermochronology results from the hanging wall of the Torbashi thrust fault suggesting a poorly defined Cretaceous age for exhumation and juxtaposition (Robinson et al., 2007). We interpret the Torbashi fault to have originally dipped north, but was subsequently rotated to a southward dip during Miocene exhumation of the Muztaghata dome (Robinson et al., 2007;Sobel et al., 2011) resulting in folding of the hanging wall of the Torbashi fault into a northwest plunging synform, the Torbashi klippe (Figs. 3,9). Robinson et al. (2007) previously interpreted the southern contact of the Torbashi thrust in the location of the Rouluke fault in this study. However, lithologic relationships  and results from zircon U-Pb analyses (Figs. 2,7,8) show that the Rouluke fault is an independent structure from the Torbashi fault (Fig. 9).
The second group of rocks consists of low metamorphic grade rocks (Pz aq ) that are intruded by Late Carboniferous granite in the southeastern section of the Waqia valley in the footwall of the Torbashi thrust fault (Figs. 7, 8). Based on the age of the intrusive body (ca. 316 Ma, Fig. 7B), we interpret this unit to be part of the South Kunlun terrane, which records a significant period of Late Carboniferous-Permian arc magmatism. Unfortunately, the boundary between the Karakul-Mazar terrane and Kunlun terrane is covered by Neogene to Quaternary basin fill (Fig. 3), so we could not determine if the contact is depositional or tectonic (Fig. 3). However, both units are clearly in fault contact with the overlying Torbashi klippe.
The third group of rocks in the Tashkurgan and Waqia valleys is exposed south-southwest of the southern trace of the Torbashi fault, and consists of two main units: (1) upper-greenschist facies metamorphic

576
IMRECKE ET AL. | Mesozoic evolution of the eastern Pamir RESEARCH rocks (Pz 2 ) intruded by an Early Ordovician granite and Cambrian gabbros (Zhang et al., 2018b) in the hanging wall of the Rouluke fault, which are part of the Cambrian Bulunkuole group (Zhang et al., 2018a), and (2) late Paleozoic (Carboniferous-Permian) limestone and slates in the footwall of the Rouluke fault that yield detrital zircon ages typical of Late Paleozoic Gondwana derived stratigraphy Robinson et al., 2012). Detrital zircon ages from the slates are consistent with those in the Qiangtang terrane in northern Tibet and in Central Pamir terrane rocks along the western portion of the Muztaghata gneiss dome (Robinson et al., 2012;Rutte et al., 2017a; Fig. 1), while late Proterozoic-Cambrian metamorphic rocks have been identified in the core of the Central Pamir gneiss domes (Rutte et al., 2017a), which are likely equivalent to the Bulunkuole group. Based on the above results, we interpret these units south of the Torbashi thrust fault to be Gondwanan terranes equivalent to the Central or southern Pamir to the west, and the Qiangtang terrane to the east. The Tanymas suture of the Pamir, equivalent with the Jinsha suture of Tibet, lies between lithologies of Karakul-Mazar Terrain and the Central/ southern Pamir terranes in the study area (Rutte et al., 2017b). In the Waqia valley, the suture is buried beneath Neogene sedimentary rocks (Nf) (Fig.  3) where it is cut on the eastern end by the younger Waqia fault and is defined by the Torbashi fault to the west (Fig. 3). Southeast of the Torbashi klippe, our mapping and zircon geochronology suggest that there is no exposure of the Karakul-Mazar terrane, and that in this region the Kunlun and Central/southern Pamir terranes are directly juxtaposed across the Paleo-Tethyan suture zone (the Tanymas suture of the Pamir and Jinsha suture of Tibet). The observation that the Karakul-Mazar terrane tapers out is consistent with regional geologic maps that show Triassic Songpan Ganzi-Hoh Xil terrane terminating in NW Tibet near the southeastern syntaxis of the Pamir Salient, and is not continuously exposed between northwestern Tibet and the Pamir (Fig. 11) (e.g., Yin and Bian, 1992;Xiao et al., 2002aXiao et al., , 2002bXiao et al., , 2005. This is similar to the situation on the western side of the Pamir, where the Karakul-Mazar terrane pinches out between the northern Pamir and Central Pamir near the border with Afghanistan (Schwab et al., 2004). We propose two possibilities for the absence of the Karakul-Mazar-Songpan Ganzi-Hoh Xil in this region: (1) the Karakul-Mazar-Songpan Ganzi-Hoh Xil has been tectonically removed via uplift along thrust faults and erosion; or (2) the Karakul-Mazar and Songpan Ganzi-Hoh Xil are two distinct terranes within the orogen. Although we cannot rule out the latter scenario, we prefer the interpretation that the Karakul-Mazar-Songpan Ganzi-Hoh Xil was tectonically removed, which is consistent with results from Cao et al. (2015), who documented significant Mesozoic exhumation of the West Kunlun Shan from detrital zircon fission track analyses from the Tarim basin. We suggest that in the West Kunlun Shan, the Karakul-Mazar-Songpan Ganzi-Hoh Xil terrane was tectonically and erosionally removed during documented Cretaceous shortening (Robinson et al., 2004;Cao et al., 2015). This is consistent with metamorphic grade of the Karakul-Mazar terrane generally increasing to the east in the Pamir where deeper structural levels should be exposed (Robinson et al., 2004), with the Torbashi klippe preserving the highest grade portions of the terrane. Finally, our results show that this fundamental terrane boundary is located farther north than previously interpreted (Robinson et al., 2012;Lacassin et al., 2004), showing there is no deflection of terranes between the Pamir and Tibet along a discreet fault, and that large-scale (>400 km) offsets along the Karakoram fault are not feasible (Valli et al., 2008).

Implications for the Cambrian Bulunkuole Group
Recently published work has suggested that much of the eastern Pamir consists of a structurally contiguous late Proterozoic-Cambrian terrane, the Bulunkuole Group, which is interpreted to have accreted to the southern margin of Tarim in the early Paleozoic and be part of the South Kunlun terrane of the Western Kunlun Orogenic Belt (Zhang et al., 2018a). Several key observations show that this is not the case. (1) The core of the Kongur Shan and Muztaghata gneiss domes are composed almost entirely of Triassic orthogneisses and granitic bodies, with no documented exposures of Cambrian or older lithologies except along the western portion of the Muztaghata gneiss dome (Robinson et al., 2012).
(2) Cambrian igneous rocks are only found south of the Torbashi klippe, and in the western Muztaghata gneiss dome where they are continuous with late Proterozoic-early Paleozoic lithologies in the core of the Shatput gneiss dome (Zhang et al., 2018a(Zhang et al., , 2018bRutte et al., 2017a). (3) As documented in this study and previous work Yang et al., 2010), the Torbashi klippe and its northern footwall are Triassic in age, and are continuous with the Karakul-Mazar terrane of the northern Pamir.
Based on these observations, we suggest that the Bulunkuole Group is exposed in a relatively narrow structural block in the southeast Pamir, located between the Torbashi and Rouluke thrust faults (Pz 2 ; Figs. 2, 3). Further, while we agree with the interpretation that the Bulunkuole Group likely records early Paleozoic subduction and collision between Tarim and Gondwana (Zhang et al., 2018a;Metcalfe, 2013), we suggest that it remained part of Gondwana during subsequent Devonian rifting. The Bulunkuole Group is therefore interpreted as part of the Qiangtang equivalent terranes of the Pamir, which collided with Asia during the Late Triassic closure of the Paleo-Tethys along the Tanymas-Jinsha suture.

Correlating Units across the NE Pamir
We have identified several characteristics of the geologic units in the northern and southeastern Pamir on either side of the Late Cenozoic Muztaghata dome that suggest the presence of regionally correlative rock units and a regionally extensive thrust nappe in the northern Pamir (Fig. 9).
To the south and north of the Muztaghata dome, the hanging wall and footwall rocks of the Torbashi and Baoziya thrust faults, respectively, share similar lithologic and geochemical characteristics. As reported above, south of the Muztaghata gneiss dome the hanging wall of the Torbashi thrust fault (Tr sch1 ) consists of Triassic upper amphibolite facies pelitic schist and quartzite intruded by Late Triassic-Early Jurassic granite (Figs. 2, 3, 7D) structurally overlying Triassic greenschist facies marble, phyllite, and quartzite. Similarly, north of the Muztaghata gneiss dome, the hanging wall of the Baoziya thrust consists of Triassic amphibolite facies pelitic schist and quartzite intruded by Late Triassic granite, structurally overlying Triassic greenschist facies marble, phyllite, schist, and quartzite (Figs. 2, 3). The footwalls of the Torbashi and Baoziya thrust faults have similar εNd and Sr isotope values (Robinson et al., 2012). Further, maximum depositional ages for the hanging walls of both thrusts are comparable (Triassic in age), with similar detrital age populations (Fig.  10A) (Robinson et al., 2012;Yang et al., 2010;Zhang et al., 2007), and our detrital zircon age populations between the footwalls of the Torbashi and Baoziya thrust faults (Tr sch2 ) are statistically indistinguishable indicating the protoliths were part of the same sedimentary basin (Figs. 10A, 10B).
The lithologic and isotopic similarities between the hanging wall and footwall rocks of the Baoziya and Torbashi faults indicate that these units and thrust fault separating them were contiguous across the eastern margin of the Pamir. The hanging wall was part of a regionally continuous thrust nappe across the northern Pamir. This thrust sheet emplaced higher-grade portions (T R sch 1 ) of the Karakul-Mazar terrane over lower-grade portions (T R sch 2 ), as well as the Central Pamir terrane to the south (Fig. 9). Robinson et al. (2007) reported muscovite and biotite cooling ages (105-125 Ma) in the Torbashi thrust hanging wall, which have been interpreted to 577 IMRECKE ET AL. | Mesozoic evolution of the eastern Pamir RESEARCH record exhumation during thrust displacement. These results indicate the once-continuous Baoziya-Torbashi thrust likely records Cretaceous retroarc shortening, which has been previously documented in the northeastern Pamir and South Pamir terrane (Robinson et al., 2012;Angiolini et al., 2013;Robinson, 2015;Chapman et al., 2018). The thrust fault, and the hanging wall and footwall lithologies, were then crosscut in the Miocene by the Muztaghata gneiss dome (Robinson et al., 2007;Thiede et al., 2013). As a final note, our results document a more coherent, and laterally continuous, series of lithologic units across the northern and eastern margin of the Pamir than previously recognized.

Pamir Terrane Structure
The three-dimensional structural architecture between terranes in the northern Pamir and southeastern Pamir is complex, most notably the observation that the terranes are tectonically interleaved (Fig. 9). The Kexi-Liqi fault is the oldest structure, juxtaposing Pz 1 lithologies above Triassic lithologies during closure of the Paleo-Tethys. Next, the southdirected Baoziya-Torbashi thrust emplaced Triassic lithologies above

RESEARCH
Gondwanan Paleozoic lithologies, with the south directed Rouluke fault imbricating the Gondwanan terranes likely active at this time. These structures were then cross-cut in the Miocene by the Shen-Ti fault and North Muskol shear zone during tectonic exhumation of the Muztaghata gneiss dome (Robinson et al., 2007;Rutte et al., 2017a), and then subsequently cross-cut again by the late Miocene to recent Kongur Shan normal fault (Robinson et al., 2007). This study documents that in both the northern Pamir and southeastern Pamir, Triassic lithologies of the Karakul-Mazar terrane are structurally emplaced above Gondwanan terranes (Central and southern Pamir) with the Torbashi thrust an along-strike equivalent to the Tanymas thrust (Figs. 2,9).
Our results in the Waqia valley add another aspect of complexity as we document that the Karakul-Mazar terrane has been completely removed either structurally or erosionally in the southeast portion of the study area, with the Western Kunlun terrane in direct contact with Gondwanan terranes (Fig. 1). However, the polarity of this contact is currently not known, although relationships within the Muztaghata dome suggest the possibility of a south-dipping contact. Finally, our documentation of the location of the Paleo-Tethys suture, and the lack of documented strikeslip faults east of the Karakoram fault, shows that deflection of terranes between the Western Kunlun and Pamir was not primarily accommodated along discrete, large displacement faults. Instead, our results suggest that Cenozoic northward displacement of terranes was accommodated by oroclinal bending of the northern Pamir/Kunlun terranes and dextral shear along the Pamir margins. We note, however, that recent results indicating limited Cenozoic northward displacement of the Pamir suggest the northward deflection of terranes may be an inherited feature (Chapman et al., 2017;Chen et al., 2018).

Model for the Southeastern Pamir
Here, we propose an updated tectonic model for the evolution of the southeastern Pamir. North-directed subduction of the Paleo-Tethys Ocean beneath the Karakul-Mazar terrane in the late Triassic produced subduction-related magmatism and built an arc-accretionary complex, the Karakul-Mazar terrane ( Fig. 11A; Schwab et al., 2004;Robinson et al., 2012). During this time, drainages sourced north of the magmatic arc provided sediment from the southern margin of Asia, i.e., Kunlun terrane, as well as the active arc, evidenced by the occurrence of Paleozoic and Precambrian zircon age populations (Yang et al., 2010;Robinson et al., 2012). During subduction, the Karakul-Mazar terrane experienced highgrade metamorphism and intrusion of igneous bodies lasting into the earliest Jurassic, similar to what has been documented in the western Pamir in the Kurgovat metamorphic complex (Figs. 11A, 11B) (Robinson et al., 2004;Schmidt et al., 2011;this study). Initial closure of the Paleo-Tethys Ocean resulted in the Central Pamir/Qiangtang terrane being thrust over the Karakul-Mazar terrane (Figs. 9, 11A, 11B) (Robinson et al., 2012).
Retroarc shortening related to subduction of the Neo-Tethys Ocean during the Cretaceous is interpreted to have resulted in development of the Tanymas-Baoziya-Torbashi thrust as a regionally extensive thrust nappe. This period of deformation resulted in imbrication of the Karakul-Mazar terrane and emplaced it above the Central Pamir to the south (Fig. 11C) (Robinson et al., 2007(Robinson et al., , 2012Rutte et al., 2017b). Cooling ages in the hanging wall from Robinson et al. (2007) suggest shortening occurred at ca. 120-100 Ma. This slightly pre-dates documented Cretaceous shortening in the South Pamir terrane from ca. 110-80 Ma (Chapman et al., 2018), suggesting southward migration of deformation during retroarc shortening.
Finally, in the Cenozoic, Miocene north-south-directed extension and exhumation of the Central Pamir gneiss domes cut across the western end of this regionally extensive thrust sheet (Figs. 11D, 11E). Further, while our results do not document Cenozoic shortening along the eastern Pamir, it is possible that shortening on the Rouluke fault is related to that observed further west in the Central Pamir (Rutte et al., 2017a(Rutte et al., , 2017b.

CONCLUSIONS
Geologic mapping integrated with metamorphic petrology and igneous and detrital zircon geochronology, document the presence of several discreet thrust sheets in the region south of the Muztaghata gneiss dome. The Torbashi thrust sheet, made up of upper amphibolite facies Karakul-Mazar terrane, is folded into a broad gently northwest-plunging syncline, and overlies greenschist facies Karakul-Mazar terrane in the north, and Gondwanan terrane crust (Central or southern Pamir) to the south. We document the southern contact of the Torbashi thrust (which represents the local equivalent of the Jinsha suture zone) as farther north than previously proposed. Farther south, Gondwanan crust is imbricated by the southsouthwest-vergent Rouluke thrust fault, which places upper greenschist facies Cambrian rocks of the Bulunkuole group over sub-greenschist facies Permian slate and marble.
Along the southern end of the Waqia valley, the northern and southern traces of the Torbashi thrust connect and the Kunlun terrane crust is juxtaposed directly against Gondwanan crust. This shows that exposure of the Karakul-Mazar terrane are not contiguous with the Songpan Ganzi-Hoh Xil terrane of northern Tibet. We suggest this discontinuity is the result of structural and erosional removal the terrane in the Western Kunlun and southeastern Pamir, possibly during the Cretaceous. Further, our documented position of the Jinsha suture shows that deflections of northern terranes in the Pamir have not been accommodated by displacement on discreet, high-slip faults, and oroclinal bending within the eastern northern Pamir may account for northward deflection of terranes.
Lithologic similarities and concordant isotopic and detrital zircon age signatures document that the footwall and hanging wall of the Baoziya and Torbashi thrusts are directly correlative and represent a once-continuous structural-lithologic package, which was subsequently cut by the Miocene Muztaghata gneiss dome. Furthermore, our results indicate that the Torbashi thrust fault is correlative to the Tanymas thrust fault of the northern Pamir, and that both may have been active in the Early Cretaceous. These results suggest (1) that lithologic units are more contiguous across the northern Pamir than previously recognized; and (2) the presence of a regionally extensive thrust nappe, continuous across much of the northern Pamir and into the southeastern Pamir, which juxtaposed Karakul-Mazar terrane above Gondwanan terranes during Early Cretaceous retroarc shortening.