New middle Miocene to Pliocene (~14–3 Ma) apatite fission track (AFT) cooling ages combined with published K–Ar/Ar–Ar and zircon fission track (ZFT) ages from the Hazara and Swat regions of Pakistan are used to explain the Oligocene to Pliocene structural evolution in the Western Himalaya. The structural model explains the distribution of K–Ar/Ar–Ar ages in three distinct age groups (Proterozoic, Paleozoic-Mesozoic, and Eocene to Oligocene). The Proterozoic to Mesozoic sequence of northern Hazara and Swat experienced elevated temperature and pressure conditions, evident by reset Eocene to Oligocene K–Ar/Ar–Ar hornblende and Eocene to Miocene muscovite ages, caused by Kohistan overthrusting the Indian margin during and after the India–Asia collision. Samples from the Indus syntaxis with Paleo to Mesoproterozoic K–Ar/Ar–Ar hornblende ages and Eocene to Oligocene Ar–Ar muscovite ages show no signs of Cenozoic metamorphism; these samples were thermally imprinted up to the Ar–Ar muscovite closure temperature. Neoproterozoic to Lower Paleozoic rocks from the southern parts of Hazara and Swat show Mesozoic to Oligocene partially reset Ar–Ar muscovite ages and preservation of Ordovician metamorphism. The combined analysis of published K–Ar/Ar–Ar (muscovite), ZFT, and new AFT ages (~14–12 Ma) suggests that the Main Central thrust/Panjal thrust was active from Oligocene to early Miocene (~30–18 Ma), and the Nathia-Gali and Main Boundary thrusts were active from the middle to late Miocene (~14–9 Ma) in the Hazara area. New and published AFT ages (~6–3 Ma) from the Indus syntaxis suggest that early Pliocene tectonic thickening in the hinterland formed the N–S trending Indus anticline, creating an erosional half window in the Main Mantle thrust, forming the Indus syntaxis, and dividing the Main Central thrust sheet into the Hazara and Swat segments.
The Himalaya, one of the most tectonically active mountain ranges in the world, exposes thrust belts that record both in-sequence and, less frequently, out-of-sequence propagation of deformation over million to millennial and decadal time scales (Figure 1). Detailed geochronologic, thermochronologic, geomorphologic, and thermobarometric studies in different parts of the Himalayan Orogen have highlighted differences in the structural style, spatiotemporal development of the structures, and effects of Cenozoic Himalayan metamorphism on the subducting Indian plate (1, and references therein). The Himalaya is tectonostratigraphically subdivided into four zones (Tethys-, Greater-, Lesser-, and Sub-Himalaya), bounded by southward-younging thrusts originating from the sole thrust known as the Main Himalayan thrust (MHT). However, studies in the Northwestern Himalaya have suggested that deformation switched back to the hinterland, resulting in the formation of out-of-sequence thrusts, reactivation of older thrusts, and zones of high seismicity and exhumation driven by tectonic and climatic processes [2-4].
The Himalaya can be geographically subdivided into the Eastern, Central, Northwestern, and Western Himalaya (Figure 1). The Western Himalaya covers the region west of the Nanga-Parbat syntaxis and south of the Main Mantle thrust (MMT) in Pakistan  (Figure 1). Major differences between the Western Himalaya and the Northwestern and Central Himalaya are related to tectonostratigraphic subdivisions, timing of metamorphism, and timing of fold and thrust belt development [1, 5, 6].
Ar–Ar hornblende ages (~1970–32 Ma) from the Indus syntaxis and metamorphic monazite ages (464–482 Ma) from the northern part of the Hazara area were taken as evidence by Treloar et al. , suggesting that the peak metamorphic grade in this region directly reflects Paleoproterozoic and Ordovician orogenic events, and the metamorphic rocks were only thermally imprinted (reset Ar–Ar muscovite ages) during the Cenozoic [6, 7] (Figures 1 and 2). Treloar et al.  tried to explain the metamorphic evolution and the wide distribution of Ar–Ar hornblende ages through a structural model that requires delamination and deformation along multiple décollements to explain the structural evolution of the Western hinterland region (Figure 3). The only available published zircon fission track (ZFT) ages (26–17 Ma) and apatite fission track (AFT) ages (23–5 Ma) from the Indus syntaxis, Hazara, and Swat regions in the Western Himalayan tectonic wedge highlight the structural evolution of the region . The timing of major structural development in the fold and thrust belt [9, 10] was only constrained after the publication of these age data.
The new AFT ages reported in this study for the Hazara region are at least ~5 Ma younger than the previously published data by Zeitler  in the Hazara area. Furthermore, the only balanced cross section available for the region was constructed before the availability of age data to constrain the timing and sequence of structural evolution . The recent structural models also do not provide details about the structural evolution south of the MMT, in the Hazara and Swat regions [6, 7, 12]. In summary, previous studies have separately explained the metamorphic impacts and structural evolution in the region south of the MMT.
In this study, by linking the Oligocene to Pliocene structural evolution with data on the Cenozoic exhumation, we can better explain the sequence of deformation, the effects of Eocene thermal imprinting on the subducted part of the Indian plate, and the preservation of different generations of metamorphism in the study area. We present 11 new AFT ages from granitic rocks of the Hazara and Indus syntaxis area (Figure 2,4). The new data are presented alongside published low-temperature thermochronologic and K–Ar/Ar–Ar age data from the adjoining areas of Swat, Kaghan, Neelum, and Nanga Parbat to constrain the timing of the structural evolution in the Western Himalaya hinterland. Based on the models and observations by Treloar et al. , DiPietro et al. , and results from this study, we derive a conceptual structural model that shows the Oligocene to Pliocene structural development and how this drives exhumation in the western hinterland region of Pakistan.
2. Tectonostratigraphic Framework of the Western Himalaya and Surrounding Regions
The Hindukush, Karakoram, and the Kohistan Arc, located in northern Pakistan, were formed during Paleozoic-Mesozoic times at the boundary of India and Asia (Figure 1). The opening of the Neo-Tethys in the late Paleozoic resulted in the separation of rifted blocks that migrated northward and accreted to Asia during the Late Triassic to Early Jurassic Cimmerian Orogeny that form the Karakoram and Hindukush [13, 14]. The Kohistan Arc is separated from the Eurasian Plate (Karakoram and Hindukush) by the Main Karakoram Thrust (MKT) in the north and from the Indian Plate by the Indus Suture, also called the MMT in the south [15-17]. The Kohistan Arc is regarded as an obducted island arc formed by intraoceanic subduction in the Tethys, which started in the Late Jurassic-Early Cretaceous time [17-19]. Late Cretaceous rifting within the Kohistan Arc resulted in the emplacement of extensive magmatic intrusions . The closure of the Tethys and collision of India with Asia was completed by the suturing of the Kohistan Arc to India and Asia during the Eocene time [10, 17, 20]. The Nanga Parbat syntaxis is a north–south-trending domal structure located between the Kohistan and Ladakh Arcs and bounded almost entirely from east to west by the MMT . The Nanga Parbat syntaxis comprises Indian Plate rocks that were metamorphosed to high pressure and temperature conditions after the collision and intruded by young granites . The core of the structure is characterized by high uplift and erosion rates [8, 22].
The collision of India and Asia resulted in the development of Himalaya and its fold and thrust belt since Eocene times. The Western Himalayan Orogen is a broad term that has been used to represent areas west of Nepal to the Arabian Sea, including the Northwestern Himalaya, Western Himalaya, and Himalayan axial belt  (Figure 1). Here, we will use the term Western Himalaya for the Himalayan region south of the MMT between the Nanga Parbat syntaxis and the Himalayan axial belt, as proposed by DiPietro and Pogue . The Western Himalaya is divided into three different tectonostratigraphic zones, from north to south: (1) the Western Hinterland (igneous and metamorphic zone), (2) the Lesser Himalaya, and (3) the Sub-Himalaya  (Figures 1 and 2). The Western Hinterland is bordered to the north by the MMT and to the south by the Main Central thrust/Panjal thrust (MCT/PT). This zone is predominantly composed of Lower Proterozoic to Mesozoic igneous and metamorphic rocks of the Indian plate [5, 23, 24]. Lower Proterozoic medium-grade metamorphic rocks are exposed in the Indus anticline, which divides the area into the Hazara and Swat regions . In these regions, the Neoproterozoic rocks are intruded by the Ordovician Mansehra and Swat Granites  (Figure 2). The northern part of the Lesser Himalaya comprises Proterozoic slates overlain by Mesozoic to Eocene sedimentary strata that were intensely deformed into stacks of thrust sheets exposed between the MCT/PT and the Nathia-Gali thrust (Figure 2). The hanging wall of the Main Boundary thrust (MBT) comprises an intensely deformed Mesozoic to Eocene sedimentary succession that is thrust onto the Neogene foreland strata of the Hazara–Kashmir syntaxis and the Sub-Himalaya [9, 27-29]. The Hazara–Kashmir syntaxis contains folded Neogene foreland strata and the Balakot–Bagh fault in the core (Figure 2). The western boundary of the Hazara–Kashmir syntaxis is marked by the active Balakot–Jhelum fault. The Sub-Himalaya is composed of Precambrian to Pliocene age sedimentary sequences that form the external part of the fold and thrust belt [28-30]. The Salt Range thrust, structurally equivalent to the Main Frontal thrust (MFT), forms the southern boundary of the Sub-Himalaya and also represents the range front of the Western Himalaya in Pakistan [30, 31].
The collision of India with Asia in northern Pakistan has been estimated to have occurred between 55 and 40 Ma [20, 32, 33]. Deformation propagated into the interior of the Indian plate after this collision along a major sole thrust known as the MHT that branches out temporally into southward younging thrust systems such as the MCT, MBT, and MFT. The movement on the MMT is reported to be almost synchronous with the collision in the Eocene; however, the fault was reactivated later as a normal fault in Oligocene and Pliocene times [7, 10, 34]. There are no direct age constraints available for the activation of the MCT/PT in Pakistan; however, it is accepted that motion was synchronous with the early Miocene development of the MCT in the north of the Hazara–Kashmir syntaxis and in the Central Himalaya of India and Nepal [27, 35-37]. The MBT is a major structure that developed synchronously along the Western and Northwestern Himalaya of Pakistan and India around ~10 Ma [9, 27, 29] and at 5 Ma in Nepal . The Salt Range thrust in Pakistan developed at 4–5 Ma [30, 39]. A similar age of 4 Ma is reported for the MFT in the northwestern part of India and 2 Ma in central Nepal [4, 40].
3. Published Structural Models of Western Hinterland in Pakistan
The structural model by Palin et al.  suggested that the Lesser Himalaya in the Hazara region does not show signs of Himalayan metamorphism, in contrast to Cenozoic Himalayan metamorphism that is typically related to the Greater Himalayan sequence in the Central and Northwestern Himalaya (Figure 3(a)). The structural model by Treloar et al.  attempts to solve the metamorphic history enigma of the Hazara and Swat region; they suggest that Paleoproterozoic basement (Besham Group) exposed in the Indus syntaxis, Neoproterozoic-Lower Paleozoic rocks of Hazara, and Upper Paleozoic-Mesozoic rocks of Swat were subjected to different temperature–pressure conditions that resulted from downward heat diffusion from Kohistan to the underthrusting Indian strata (Figure 3(b)). The structural evolution suggests that the stratigraphic contacts between the Paleoproterozoic basement, Neoproterozoic-Lower Paleozoic, and Upper Paleozoic-Mesozoic rocks were activated as décollements that facilitated the accretion of these three rock packages with different temperature, pressure, and time histories beneath Kohistan. Furthermore, the normal movement on the MMT has exposed basement rocks in the Indus syntaxis.
DiPietro et al.  for the first time compiled all available zircon U-Pb geochronology, Ar–Ar, and ZFT thermochronology data for the Hazara, Swat, and Indus syntaxis. Their resulting model suggests that by ca 50 Ma, the Kohistan Arc and associated melange units had reached their southernmost extent along the MMT onto the underthrusted Indian plate, and the subducted rocks of the Indian plate had reached their maximum burial depth (Figure 3(c)). Their interpretation and model further suggest that rocks in the Indus syntaxis were already being exhumed by 50 Ma; however, exhumation occurred later in the domal structures of Swat around 39 Ma. Previous work and this model have argued about the dominant dextral strike-slip motion of the Kohistan Arc after a collision along the northernmost trace of the MMT known as the Kohistan fault, which became inactive between 29 and 20 Ma [5, 12].
The latest model for the northern Swat area, proposed by Soret et al. , suggests that rocks in northern Swat were subjected to amphibolite-facies condition from ≤47 to ~39 Ma, synchronously with the formation and partial exhumation of ultra-high-pressure eclogites (Figure 3(d)). Rapid exhumation in the northern Swat region started from 38 Ma due to crustal thickening and foreland formation when thick Indian continental crust was involved in the collision.
The available models explain the structural and metamorphic evolution of only limited areas. For example, the model by Soret et al.  explains the metamorphism and structural evolution of the Swat region and does not mention details about the nearby Indus syntaxis and Hazara region. The models by Palin et al.  point out that the metamorphism in the Hazara area is related to the Bhimphedian Orogeny of the Ordovician age  but does not explain the metamorphism of the Swat region. This issue was partly addressed by Treloar et al. , which clearly pointed out the enigma; however, the proposed structural model is very complicated. Their model suggested that the stratigraphic units from the Indian passive margin were delaminated from each other after subduction and metamorphosed to different temperature–pressure conditions and then thrusted on the top of each other during the development of the fold and thrust belt. However, the structural map shows that the structural style in the hinterland simply consists of south-vergent thrusting over the MCT/PT. So far, no other major fault has been mapped that could have accommodated the shortening prior to the development of the MCT/PT in this region. The model by Treloar et al.  focused primarily on the pre-Oligocene metamorphic evolution; therefore, the subsequent structural evolution of the region was only briefly discussed. The structural model by DiPietro et al.  does not explain the structural evolution in the Hazara and Swat areas.
4. Fission Track Dating Method
Radioactive fission decay of 238U produces spontaneous fission tracks that accumulate in apatite and zircon as a direct function of the uranium concentration and time. The AFT and ZFT methods are based on quantifying the parent (uranium concentration represented by induced tracks produced by irradiating the sample with thermal neutrons) and daughter (spontaneous tracks) products to estimate the time since the crystals passed through ~120°C (closure temperature [Tc] for AFT) and ~240°C (Tc for ZFT) isotherm [43, 44].
Apatites were separated from rock samples at the University of Potsdam using standard magnetic and heavy liquid separation techniques. Spontaneous fission tracks on internal surfaces of apatite were revealed by etching the fission track mounts with 5.5 M HNO3 solution for 20 seconds at 21°C. Mica sheets were attached to the fission track mounts, which were then irradiated at the FRMII reactor at Garching (Munich, Germany). During irradiation with thermal neutrons, induced tracks are produced by fission of 235U; some of these tracks penetrate the mica sheets. After irradiation, induced tracks are revealed by etching the mica sheets in 40% HF at 21°C for 45 minutes. AFT ages were calculated using the external detector method  with a personal zeta value of 353 ± 7 (Humaad Ghani).
5. AFT Samples and Results
5.1. New AFT Thermochronologic Data
All AFT ages presented in this study pass the Chi squared test P(χ2), suggesting that the single-grain ages are consistent with a single population . The AFT ages are reported as pooled ages with ±1σ error  (Table 1; additional details are provided in the online supplementary material). Samples M2–M5, from the southwestern part of the Mansehra Granite in the Hazara region, yield AFT ages between 12.0 ± 0.6 and 14.2 ± 0.6 Ma (Figure 4). Single sample M1, collected from the hanging wall of the MCT/PT (the surface trace of the MCT/PT and MBT are very close together around the Hazara–Kashmir syntaxis) in the southeastern part of Hazara, shows a comparatively young age of 9.0 ± 0.6 Ma. Samples M6–M11, collected at similar elevations, show a northward-younging pattern of AFT ages ranging from 6.0 ± 0.6 Ma at the flank to 3.3 ± 0.4 Ma in the core of the Indus anticline. None of the samples yielded sufficient confined track lengths for measurement that could be used for inverse thermal modeling. Generally, the AFT ages are middle Miocene in the southern part of the Hazara except one sample that yielded a late Miocene age. Ages reported from the Indus syntaxis/Indus anticline are late Miocene to middle Pliocene.
5.2. Published Ar–Ar Age Data and Thermochronologic Data
5.2.1. Published K–Ar/Ar–Ar data
We have shown all seventy-seven K–Ar and Ar–Ar ages (thirty-five hornblende, ten biotite, and thirty-two muscovite) reported in Zeitler , Treloar et al. , Maluski and Matte , Palmer-Rosenberg , Baig , Treloar and Rex , and Anczkiewicz et al. , from the original data compilation of DiPietro et al.  (Figure 4, online supplementary information). The hornblende Ar–Ar and K–Ar age data from the core of the Indus anticline show Proterozoic ages, and the flanks show early Oligocene to Eocene ages, with some Mesozoic to Paleozoic ages to the east and west of the limbs in adjacent areas (Figures 4(a) and 4(b)). The hornblende Ar–Ar data from the northern part of Swat ranges from Paleocene to early Oligocene, with a single Paleozoic age from the southern part of Swat. The biotite Ar–Ar and K–Ar age data show a wider distribution in ages, with Mesozoic to Paleocene ages in the northern part of Hazara, Proterozoic to Mesozoic ages in the core of the Indus anticline, and Mesozoic to Oligocene ages along the western flank of the Indus anticline. The majority of muscovite Ar–Ar and K–Ar age data show younger ages, ranging from Eocene to late Oligocene, except for two Proterozoic ages from the core of the Indus anticline. The muscovite Ar–Ar ages from Swat range from early Oligocene to early Miocene. All of the muscovite Ar–Ar data from the MMT zone and the Kohistan arc are Mesozoic in age. In summary, the majority of Ar–Ar and K–Ar ages from hornblende, as well as one biotite and two muscovite ages from the core of the Indus anticline, are Proterozoic in age; ages get young toward the flank and get older again toward the northern part of Hazara, whereas ages are younger westward toward Swat. Although biotite Ar–Ar and K–Ar ages show a wide distribution, a younging pattern in ages could be observed from the core toward the flanks. The distribution of muscovite Ar–Ar and K–Ar ages shows a much clearer pattern in age versus structure relationship: Proterozoic ages in the core of the Indus anticline get sharply younger toward the flanks. DiPietro et al.  have noted that the wide distribution in ages is suspected to be the result of partial to full resetting of mineral closure systems after rocks of the Indian Plate were underthrusted beneath Kohistan. We agree with their observation; however, no such structural model was constructed to show the overall thermal structural evolution of this region. We address this point in section 6, where we show how different age groups are distributed in the hinterland region of the Western Himalaya in Pakistan.
5.2.2. Published Thermochronology Data
The amount of thermochronologic data obtained from the Nanga Parbat syntaxis, Indus syntaxis, and intervening areas is limited [8, 10, 27, 55, 56] (Figure 4). The youngest ZFT ages of ~2 Ma are reported from the core of the Nanga Parbat syntaxis . Pliocene (<5 Ma) AFT and apatite (U-Th-Sm)/He (AHe) ages reported from the Neelum valley of the Kashmir region suggest that the zone of tectonically driven exhumation is parallel to the hanging wall of major thrusts . AFT and AHe age assemblages from the Kaghan valley range from ~4 to ~16 Ma [8, 55], except 21–29 Ma ages from samples that were collected in the footwall close to the trace of the MMT . A 49-Ma ZFT age from an Oligocene age sandstone collected from the Hazara–Kashmir syntaxis reflects the detrital cooling age of the source region . Late Oligocene to early Miocene (26.1 ± 1.2 to 19.7 ± 1.2 Ma) ZFT ages and late Oligocene to early middle Miocene AFT ages (22.8 ± 4.3 to 15.7 ± 4.1 Ma) are reported from the hanging wall of the MCT/PT in the Hazara and Swat regions . An early Oligocene AFT age of 30 ± 13 Ma is reported from the footwall of the MCT/PT in the southern Hazara area . Two ZFT (28.2 ± 3.4 and 20.5 ± 3.6 Ma) and two AFT (15.1 ± 3.4 and 11.2 ± 2.2 Ma) ages are reported from the immediate footwall of the MMT in the Swat area . Three ZFT (22.6 ± 2.6 to 18.8 ± 2.1 Ma) and five AFT (5.6 ± 2.3 to 3.7 ± 1.1 Ma) ages are reported from the Indus anticline and syntaxis [8, 10].
5.2.3. Comparison of Published and New Thermochronology Data
Our AFT ages are similar to those reported in the Indus syntaxis [8, 10]. However, our five AFT ages in the Hazara region (9.0 ± 0.6 to 14.2 ± 0.6) are younger than the two previously reported AFT ages (19.3 ± 6.2 and 22.8 ± 4.3) from this area. The latter two published AFT samples were collected from the core of the syncline in the Hazara area; these were dated using the population method and were reported with 2σ uncertainty by Zeitler . This method is less precise and may be less accurate; therefore, it is now rarely used for AFT analyses . In comparison to the previous work, our new data provide an east-to-west structural profile of the syncline in the Hazara area. Ages are overlapping, show less dispersion, and are aligned with the structural geometry of the syncline, showing comparatively older ages in the core and younger ages on the flanks. In summary, the younger ZFT and AFT ages are reported from the northernmost part of the hinterland, especially the Indus syntaxis region, compared with the older ages from the southern part of Hazara and Swat.
6. Discussion and Implications
6.1. Implications of Eocene Himalayan Metamorphism and Thermal Imprinting on the Western Hinterland
The subduction of the leading edge of India beneath the Kohistan Arc started the Himalayan phase of metamorphism, which peaked at 47 Ma, as revealed by ultra-high-pressure eclogite rocks from Kaghan and kyanite-sillimanite-grade rocks of northern Swat exposed in the footwall of the MMT [7, 58, 59]. However, Ordovician to Neoproterozoic hornblende Ar–Ar ages (c. 1920 Ma, 550 Ma) from Paleoproterozoic rocks exposed in the core of the Indus anticline do not show any signs of Cenozoic Himalayan metamorphism (7 and references therein; Figure 4). Furthermore, based on the hornblende and mica K–Ar and Ar–Ar ages (c. 33–45 Ma, closure temperature (Tc) of ~500°C for hornblende, ~425°C for muscovite, and ~300°C for biotite) [60-62] reported from chlorite to sillimanite-grade metamorphic rocks in the Hazara area, metamorphism in this area was previously thought to have occurred during the Eocene [23, 53]. However, new U-Pb monazite ages (Tc ~900°C ) from the sequence yield Ordovician metamorphic ages (c. 470 Ma), suggesting that the peak metamorphic sequence in the Hazara region exposed in the hanging wall of the MCT/PT is related to the Ordovician Bhimphedian Orogeny  (Figure 3). Therefore, Treloar et al.  suggested that the late Eocene to Oligocene K–Ar and Ar–Ar cooling ages imply that rocks of the Hazara region were thermally imprinted by heating to ~500°C for a “short time”; however, they did not observe new metamorphic assemblages formed during Eocene Himalayan metamorphism.
The impact of Eocene Himalayan tectonothermal imprinting has been demarcated in the Kel area of Neelum valley (roughly parallel to and east of the Kaghan area) using reset versus unreset U-Pb apatite ages, assuming a 450°C ± 75°C Tc  (Figure 4). The U-Pb apatite age ranges between 43 and 18 Ma from rocks exposed in the hanging wall of the MCT, suggesting that these rocks were completely to partially reset (with respect to the Tc) by Himalayan tectonism compared with Proterozoic U-Pb apatite ages of rocks exposed between the MCT and the MBT . In contrast, thermobarometric and petrologic modeling of the Upper Paleozoic to Mesozoic metamorphic rocks (Alpuri Group) of Swat suggests that these rocks were subjected to pressure and temperature conditions of ~10–11 kbar and 700°C due to subduction beneath the Kohistan Arc during Eocene Himalayan metamorphism .
In summary, the rocks exposed in the Kaghan and northern Swat area show signs of Himalayan metamorphism, rocks of Hazara show signs of the Ordovician Bhimphedian orogeny, and rocks from the Indus syntaxis show no sign of either the Himalayan or the Bhimpedian orogeny. Therefore, despite their close spatial occurrence, the rocks exposed south of the MMT preserve different metamorphic histories (Figures 1 and 2).
In earlier models constructed before the recognition of Ordovician-age metamorphism, metamorphic thrust stacking and subsequent cooling of the overburden were suggested as a mechanism for explaining how deformation propagated from the MMT southward into the Hazara region after 30 Ma [25, 49, 53]. In the revised model, Treloar et al.  suggested that cover rocks (Neoproterozoic-Mesozoic) above Paleoproterozoic basement with preexisting Ordovician age metamorphism were detached along unconformable stratigraphic contacts, metamorphosed, and thermally imprinted by the downward diffusion of heat from the overriding Kohistan Arc (Figure 3(b)). Subsequently, an internally imbricated cover sequence accreted stepwise to the base of the Kohistan arc such that higher metamorphic-grade rocks were stacked above lower-grade rocks prior to southward thrusting during the Oligocene to Miocene. The revised model by Treloar et al.  provided new insights to explain the metamorphic enigma through structural deformation (Figure 3(b)). However, the major foreland structure is the MCT/PT, which developed during the late Oligocene–early Miocene time, after the metamorphic stacking was completed; therefore, it is difficult to understand and not shown in their proposed model how deformation along different décollements in the hinterland was accommodated and “displayed” by deformational structures in the foreland. It has not yet been established whether Eocene to middle Oligocene cooling suggested by K–Ar/Ar–Ar muscovite ages was due to exhumation along the active foreland structures or to focused erosion above deeper structures. The latter scenario suggests that regional thickening of the tectonic wedge caused rock uplift above a roof thrust of duplexes formed by translation of older thrust sheets on top of younger thrust ramps. For example, if hinterland deformation was not concentrated along major thrusts, it should be kinematically distributed to the foreland structures, a mechanism suggested in the Kashmir Himalaya .
Here, we propose a new structural model that is based on the observations by Treloar et al.  for the diffusion of heat from the Kohistan Arc downward into the subducting Indian Plate. We suggest a different way that structural and exhumation processes could interact in order to explain the metamorphic enigma and the distribution of Ar–Ar, ZFT, and AFT ages observed south of the MMT (Figure 5). The original model of heat diffusion is slightly modified, as shown by a perturbation in the isotherms (Figure 5(a)). Our model does not require a complex assumption of delamination and later accretion; instead, we propose that the variable metamorphic grading in the Hazara, Swat, and the core of the Indus anticline is the result of the stratigraphic architecture of the subducting wedge. In our model, we show northward thickening Neoproterozoic to Mesozoic igneous intrusions and metasedimentary and metamorphic successions  in the Hazara and Swat regions that were subducted beneath the Kohistan Arc (Figure 5(a)). The proposed structural model suggests that thick Upper Paleozoic to Mesozoic strata were subducted beneath the Kohistan Arc during the Eocene after collision and then experienced elevated metamorphic conditions in the footwall of the MMT due to downward heat diffusion, as suggested by thermobarometric studies from the Swat and Kaghan [7, 58]. The higher temperature–pressure conditions resulted in the formation of Cenozoic Himalayan metamorphic assemblages in the Neoproterozoic to Mesozoic succession and fully reset K–Ar/Ar–Ar (hornblende, biotite, and muscovite) ages for the northern part of the Hazara and Swat regions (Figure 5(a)). In contrast, the Paleoproterozoic and Neoproterozoic rocks presently exposed in the Indus syntaxis and the southern part of Hazara and Swat were only thermally imprinted during the same time period because these units were located outside of the elevated temperature–pressure zone (Figure 5(a)), leading to unreset to partially reset K–Ar/Ar–Ar hornblende ages and full reset of Ar–Ar muscovite ages. The subsequent late Oligocene to middle Pliocene rock uplift and exhumation processes, active in the internal zones of the Western Himalaya, resulted in the exposure of these different metamorphic assemblages and distribution of different Ar–Ar age groups in the hinterland region south of the MMT. These steps are further described in the following sections.
6.2. Late Oligocene to Late Miocene Exhumation Associated with the MCT/PT and MBT
AFT and ZHe ages reported from the Kaghan valley and the northern part of the Swat, south of the MMT, range from late Oligocene to middle Miocene [8, 10, 55]. Units in the Kaghan valley and northern Swat were subjected to metamorphism soon after the collision until 47 Ma [58, 59] and subsequently cooled through higher temperatures due to exhumation from beneath the Kohistan Arc . How far did the Kohistan Arc overthrust regions of the Swat and Hazara along the MMT, resulting in the heating of the MMT footwall? The reset Ar–Ar muscovite and biotite ages suggest that the northern part of Hazara and Swat was overthrust by the Kohistan Arc after collision [7, 8]. DiPietro et al.  suggested that the northern part of Hazara experienced ongoing cooling through Ar–Ar hornblende Tc around ~50 Ma, while initial cooling in the northern Swat was delayed until ~39 Ma. However, the northern parts of both Hazara and Swat cooled at similar times through the Ar–Ar Tc of muscovite around the early Oligocene (~29 Ma). We interpret the Neoproterozoic to Mesozoic strata to have been about ~10 km thick, sufficient to reset ZFT and AFT ages in the Hazara area. The southernmost parts of the Hazara and Swat may not have been underthrusted beneath the Kohistan Arc (Figure 5(a)) because ZFT and AFT ages ranging from late Oligocene to Pliocene in the Hazara and Indus syntaxis show a northward younging trend (Figure 4). The only exception is an early Miocene ZFT age and late Miocene AFT age (sample M1) in the southeastern part near Hazara–Kashmir syntaxis (Figure 4).
In our interpretation, the cooling of Hazara and Swat region rocks through K–Ar/Ar–Ar muscovite Tc at ~29 Ma and Late Oligocene to early Miocene ZFT cooling ages from the Hazara and Swat area suggest that cooling was related to uplift and exhumation caused by activity on the MCT/PT in this area (Figure 5(b)). The MCT is suggested to have been active around ~27 Ma in the Neelum valley and in the early Miocene in the Kashmir Himalaya [4, 27]. This timing of MCT activity is at least similar on the eastern and western sides of the Hazara–Kashmir syntaxis in the Northwestern and Western Himalaya. It is unclear whether late Miocene to middle Miocene AFT ages from our samples M1–M5 in the southern part of the Hazara are associated with late-stage movement on the MCT/PT or an early stage of Indus anticline development or movement on younger foreland-vergent thrusts.
The middle Miocene ages (M2–5; ~14–12 Ma) in our study lie between the late stage of activity of the MCT/PT at ~14 Ma in the Northwestern Himalaya  and initial activity on the MBT at ~12 Ma in the Kohat area of Pakistan . In our interpretation, the ages from our samples M2–5 are associated with the initial phases of the Nathia-Gali thrust and the MBT activity in the foreland (Figure 5(c)). The MBT is temporally constrained in the Kashmir and Kohat regions to be active around ~12–10 Ma [9, 27, 29]. Samples M2–5 most probably remained hotter than the AFT Tc (~120°C) during the phase of translation on the MCT/PT (27–18 Ma), represented by the youngest ZFT ages in the Indus syntaxis. Middle Miocene AFT cooling ages result from cooling and exhumation on the MCT/PT thrust sheet, which is passively transported and rotated due to translation on the Nathia-Gali thrust and MBT (Figure 5(c)). The same kinematic model fits well with the exhumation history of sample M1, which yields a late Miocene (~9 Ma) age related to the uplift in the southeastern part due to late-stage activity on the MBT (active between ~12 and~8 Ma). The older ZFT and AFT ages are present along the N–S axis in the core of the syncline in the Hazara, and the younger ZFT and AFT ages are present in the eastern and western limbs of the syncline, suggesting structural control on exhumation and distribution of the AFT ages (Figure 4; cross section GH). The eastern limb of the syncline has experienced comparatively higher magnitudes of exhumation due to rock uplift during later activity on the MBT compared with the core and western limb of the syncline (Figures 4(d) and 4(e)). ZFT and AFT contours published in Zeitler  provide an impression of northward progressive cooling (Figures 4(d) and 4(e)). Our new AFT ages are younger in the Hazara region compared with Zeitler , and the AFT isochrons are modified to show the AFT age relationship with geological structures in the Indus anticline and syncline in the Hazara areas (Figures 5(c) and 5(d)), suggesting continued cooling and exhumation above the MCT/PT thrust sheet since Oligocene times.
6.3. Kinematics of the MMT and Development of the Indus Syntaxis
The development of the Indus syntaxis is related to two stages of normal faulting, starting at ~23 Ma (backsliding on the MMT) followed by a second stage of normal faulting in the Pliocene . In our model, we propose that Pliocene development and exhumation of the Indus syntaxis are caused by continued thickening in the hinterland, as is commonly reported in tectonic wedges [4, 29] (Figure 5(d)). In our interpretation, younger AFT ages from the Indus syntaxis represent late-stage activity in the footwall of the MMT, probably due to either underplating, where thrust slices detached from down-going Indian basement were antiformally stacked, or the presence of an NNE-SSW trending preexisting normal fault formed in the Late Paleozoic [12, 24] that facilitated the uplift, resulting in the development of the anticline since at least ~6 Ma (Figure 5(d)).
The axial trend of the Indus anticline is almost parallel to the western limb of the Hazara–Kashmir syntaxis and normal to the MMT, suggesting a pre-Cenozoic rigid basement control on the orientation of the Cenozoic structures; the older normal faults might have been exploited as structural ramps during the late Cenozoic shortening of the wedge. The gradual younging of AFT ages toward the core of the anticline suggests structural control on the AFT ages related to the exhumation of eroding anticlines . The termination of ZFT isochrons of 45–35 Ma across the curve of the Indus syntaxis suggests erosion due to late-stage uplift in the footwall (Figures 4 and 5).
A similar style of thick-skinned tectonic processes associated with late-stage exhumation of roof thrust rocks is reported from Indian basement rocks that form the Lesser Himalayan duplex beneath the Kishtwar window in the northwest Himalaya [2, 4] and a reactivated basement thrust beneath the Digne thrust sheet that forms an erosional half window in the western Alps .
Pliocene anticline development and foreland propagating shortening most probably uplifted the Upper Paleozoic to Mesozoic strata that were metamorphosed beneath the overthrusting Kohistan arc in the Cenozoic. These structurally shallow units were eroded from the core of the Indus syntaxis and northern part of Hazara, thus exposing unmetamorphosed Paleoproterozoic rocks in the core of the Indus anticline (Figure 5(d)). However, part of the wedge in the Swat and Kaghan was inactive during Pliocene times; therefore, due to comparatively slow uplift and related exhumation, the Cenozoic metamorphosed Mesozoic rocks were not eroded from Swat compared with the area of the Hazara and Indus syntaxis (Figure 5(c)).
The interior of the wedge is still active, as is evident by frequent earthquakes along the Indus-Kohistan seismic zone [69, 70]. Foreland propagating shortening is presently accommodated from the Indus syntaxis to the Balakot–Jhelum fault on the northeastern side of the wedge. On the western side of the wedge, deformation to the south of Swat is accommodated along the Salt Range thrust. The N–S trending Indus anticlinal structure extends southward to the Peshawar basin such that it has divided the MCT/PT thrust sheet into two zones of the Hazara and Swat; these were part of a single thrust sheet before the Pliocene.
6.4. Open Questions for Future Research
The model presented in this study provides a conceptual framework for the development of the fold and thrust belt south of the MMT in Pakistan. However, a new balanced cross-section across the entire region with additional low-temperature thermochronometric data is needed to constrain the spatiotemporal structural evolution. The metamorphic zoning in the southern part of the Swat area should be examined, keeping in view the study by Palin et al. .
The proposed conceptual structural evolution model for the Western Himalaya explains the effects and spatial distribution of Cenozoic Himalayan metamorphism and the coexistence of temporally different metamorphic-grade rocks in the area south of the MMT. New middle Miocene to Pliocene AFT cooling ages interpreted alongside published thermochronologic data constrain the spatiotemporal development of the major structures in the Hazara area of the Pakistan Western Himalayas. Our results suggest that the MCT/PT was active in the Hazara region in the late Oligocene to early Miocene. Early displacement on the Nathia-Gali thrust and MBT in the Lesser Himalaya has caused a passive uplift of the MCT/PT thrust sheet during the middle to late Miocene. The late Miocene age for the movement of the MBT is confirmed in the Balakot region, suggesting that it developed as a major structure in the Northwestern and Western Himalaya. Out-of-sequence anticline development due to underplating in the MMT footwall is responsible for the curved trace of the MMT in the Indus syntaxis and the division of the MCT/PT thrust sheet into the discrete Hazara and Swat areas since Pliocene times.
We thank Alexander Robinson, Syed Ali Turab, Paul R. Eizenhöfer, and anonymous reviewers for their detailed reviews and suggestions for improving the article. Eugene Grosch is acknowledged for handling the manuscript. We thank Muhammad Sajid, Fawad Ghani, and Christine Fischer for their assistance during field and lab work. Klaus Wemmer is thanked for discussion related to Ar–Ar data. The initial results of this study were presented in the 19th Symposium on Tectonics, Structural Geology, and Crystalline Geology (TSK 19; 9-11. March 2022) at Halle (Saale), Germany. This research work was supported by the Ministry of Science and Research, Brandenburg, and Potsdam Graduate School doctoral scholarship to Irum. The publication of the article is funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) – Projektnummer 491466077 regulated by University of Potsdam Open Access Publication Fund.
Conflicts of Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
The AFT age data (TRACKKEY files) are available from the corresponding author upon request.