High‐Resolution P‐T‐Time Paths Across Himalayan Faults Exposed Along the Bhagirathi Transect NW India: Implications for the Construction of the Himalayan Orogen and Ongoing Deformation

Pressure‐temperature (P‐T) conditions and high‐resolution paths from 11 garnet‐bearing rocks collected across Himalayan fault systems exposed along the Bhagirathi River (Uttarakhand, NW India) reveal the tectonic conditions responsible for their growth. A garnet from the Tethyan metasedimentary unit has a 50.3 ± 0.6 Ma (Th‐Pb, ±1σ) monazite inclusion, suggesting that ductile mid‐crustal metamorphism occurred synchronously or soon after (<10 Myr) India‐Asia collision, depending on timing. High‐resolution garnet P‐T paths from the same rock show ∼1 kbar fluctuations in P as T increases over a ∼20°C interval, consistent with a period of erosion. We report garnets from the Main Central Thrust (MCT) hanging wall that have Eocene to Miocene monazite ages, and one garnet yields paths consistent with motion along the Main Himalayan Thrust (MHT) décollement. Most high‐resolution MCT footwall P‐T paths fluctuate in P (±1 kbar) as T increases, consistent with imbrication and paths from equivalent structural assemblages in central Nepal. Like those rocks, MCT footwall (Lesser Himalayan Formation, LHF) monazite ages are Early Miocene (9.3 ± 0.6 Ma) to Pliocene (3.0 ± 0.2 Ma). The results demonstrate the consistency in timing and conditions across the MCT at locations ∼650 km apart. If the present‐day Himalayan tectonic framework has not significantly changed since the Pliocene, the LHF duplex can be considered when attributing seismic events to particular faults. The MHT is undisputedly the significant factor in accommodating Himalayan seismic activity, but MCT footwall faults may explain some shallower hypocenters, without the need for unusual MHT geometries.

ed the earthquake is debated. The slip maxima for the Uttarkashi earthquake occurred 10 km west and 15 km southwest of the hypocenter (Cotton et al., 1996;Gupta et al., 2015). The earthquake may have occurred along the MCT (Jain & Chander, 1995;Kayal, 1996), or within the Lesser Himalayan Formations (LHF) duplex (Gupta et al., 2015;C. Thakur & Kumar, 1994). Alternatively, the MCT was not involved (Yu CATLOS ET AL.   (after Catlos et al., 2007). Location of significant fault systems after Bose and Mukherjee (2019), Metcalfe (1993), Mukherjee (2013), Searle et al. (1993), and our observations. Inset shows the location of the transect within the broader framework of the Himalayan fault systems. Th-Pb monazite ages are indicated (this contribution, Catlos et al., 2007). Location of the transect within the broader framework of the Himalayan fault systems is shown in the inset. The epicenter of the Uttarkashi earthquake and its aftershocks (numbered 2-7; USGS catalog) and other earthquake epicenters are also shown. See Table 1 for information about these events. Focal mechanism after Bondár and Storchak (2011), On-line Bulletin. (b) Cross-section along A-A' showing major fault systems and sample locations after Catlos et al. (2007). et al., 1995), and the earthquake may have occurred along the Main Himalayan Thrust (MHT) (Cotton et al., 1996).
The MHT ( Figure 2) is a pervasive décollement that separates the downgoing Indian plate from the Himalayan orogenic wedge (e.g., Bilham et al., 1997;Nelson et al., 1996;Subedi et al., 2018;Zhao et al., 1993). This structure is responsible for a significant component of the present-day seismicity of the Himalayan range and is considered one of the largest and fastest slipping continental megathrusts on Earth (e.g., Duputel et al., 2016;Rajendran et al., 2017;Searle et al., 2017). Understanding the geometry and history of the development of the MHT and the large-scale fault systems that splay into the structure has implications for assessing and predicting the hazard impacts of major event Himalayan earthquakes, including their initiation, propagation, and termination (e.g., Wang et al., 2017).
The focus of this paper is to report high-resolution garnet P-T paths and new monazite (REEPO 4 ) ages from samples collected from units within the Himalayan range exposed by the Bhagirathi River in NW India. The transect extends across fault systems that displace Himalayan rock formations, including the Tethyan Formation, Greater Himalayan Crystallines (GHC), and LHF (Figures 1 and 2). Samples from the base of the MCT along the Bhagirathi River in NW India contain monazite grains that are as young as 1-4 Ma ( Figure 2, Catlos et al., 2007). These ages suggest this may be an ideal location to understand better and apply the metamorphic history recorded by its assemblages to its more present-day tectonics. The approach to generate the P-T paths is similar to that applied to samples collected from the MCT shear zone ∼650 km further east in central Nepal by Catlos et al. (2018). Results from that study show that the MCT formed as individual rock packages moved at different times and suggested high erosion rates (12 mm/year since the Pliocene). We supplement the P-T paths here with new monazite ages and compare the results to those from that study. The possible geometry and slip dynamics along the MCT and within the LHF duplex in these portions of the Himalayas are presented. Results may be directly relevant to understanding the neotectonics of the Himalayan orogeny and the geometry of the MHT. CATLOS ET AL.  Figure 1 for epicenters of these events. All parameters were extracted from http://earthquake.usgs.gov/ earthquakes/search except for the starred entries. The first starred entry is from the Seismological Center Event Agency #317038, which includes the GCMT estimated focal depth. The following are summarized from Yu et al. (1995) and are the PDE monthly (October 1991), IMD, CMT (Harvard), and PDE weekly. Missing entries are not available. b All are body-wave magnitudes (Mb), except for the first two USGS and GCMT entries for event 1, which are the moment magnitudes (Mw). The PDE monthly is the surface-wave magnitude (Ms).

Previous Analyses
We collected rocks across the MCT and South Tibet Detachment System (STDS) along the Bhagirathi River transect in the NW Indian Himalayas (Catlos et al., 2007) (Figure 1). The MCT separates LHF lower-grade (chlorite-garnet-kyanite) schists from high-grade (kyanite-sillimanite) schists and gneisses of the GHC (e.g., Martin, 2017aMartin, , 2017b and is underlain by a 1-10 km-thick shear zone (e.g., Mukherjee, 2013). The MCT could be considered to be an expression of the MHT (i.e., the MHT-MCT fault system of Bollinger et al., 2004 andMahajan et al., 2010). Other surface expressions of the MHT include the Main Boundary Thrust (MBT) and Main Frontal Thrust (MFT) (Figure 1).
Although the transect has long been studied for understanding the tectonics and hazards in the region, the mapped surface expressions specific fault systems vary considerably (e.g., Bose & Mukherjee, 2019;Manickavasagam et al., 1998;Mukherjee, 2011Mukherjee, , 2013Searle et al., 1993;Sorkhabi et al., 1999;Stern et al., 1989;C. Thakur, 1980). The MCT shear zone itself has an upper and lower bound, with its lower structure in this region termed the Munsiari Thrust (MCT-I), and the upper is named the Vaikrita Thrust (MCT) (e.g., C. Thakur, 1980). MCT shear zone thickness varies from 1 to 10 km along its entire strike between LHF and Greater Himalayan rocks (e.g., Mukherjee, 2013). The Martoli Formation overlies the GHC, and, in this region, is made up of phyllites and quartzites (e.g., Searle et al., 1993;C. Thakur, 1980). This formation is separated from higher-grade GHC gneisses by the STDS and differs from the GHC in lower metamorphic grade.
In this study, we rely on the structures mapped by Metcalfe (1993), Searle et al. (1993), and our observations. Note that Mukherjee (2013) (Schelling & Arita, 1991) connected to a seismic reflection profile (Zhao et al., 1993). We also overlie the P-to-S receiver function migration imaging the crust after Subedi et al. (2018). The Moho and the intracrustal low-velocity zone are highlighted with a black dashed line. Green crosses show Moho depths computed from H-K stacking. The section illustrates the juxtaposition of tectonostratigraphic units across the major Himalayan faults and interprets the Lesser Himalaya as a hinterland-dipping duplex. The lowvelocity layer and hypocenter of the Mw 7.8 Gorkha earthquake (red circle) and some aftershocks (black circles) are also plotted after Arora et al. (2017). We include the Global Centroid Moment Tensor Project (GCMT) and USGS solutions of the Gorkha earthquake hypocenters and its aftershocks. The locations of the hypocenters for both datasets agree, except for two of the earthquakes selected by Arora et al. (2017).
However, they do note the presence of numerous smaller-scale detachment structures both within the MCT shear zone and coinciding with our placement of the STDS. In Figure 1, we include the placement of the lithological contacts between the GHC-Tethyan Formation and LHF-GHC based on mapping done by Mukherjee (2013).
Most previous work along the Bhagirathi River focuses on MCT hanging wall rocks (e.g., Mukherjee, 2013;Searle et al., 1993;Tripathi et al., 2019). Leucogranites from the GHC unit along the transect yield ages from the Paleocene (64 ± 11 Ma, Rb-Sr, Stern et al., 1989), Eocene (46 Ma, U-Pb zircon, Singh, 2019), and Miocene (20 Ma, Singh, 2019; 21.1 ± 0.9 Ma, Rb-Sr and 18.9 ± 1.3 Ma, K-Ar muscovite, Stern et al., 1989). Hornblende 40 Ar/ 39 Ar ages of 19.8 ± 2.6 Ma are considered the activation age of the MCT in the region (Metcalfe, 1993;Mukherjee, 2011). Catlos et al. (2007) report that GHC rocks have monazite ages that range from 38.0 ± 0.8 Ma (BR20) to 19.5 ± 0.3 Ma (BR18), consistent with the onset of metamorphism in the unit and timing of MCT slip. A single monazite inclusion in garnet from BR21 yields 21.1 ± 0.5 Ma (Th-Pb), indicating Miocene activity along the MCT. This rock yields high-grade conditions (730°C ± 25°C and >10 kbar), and we were unable to model its P-T path using the Theriak-Domino approach due to modifications of its primary zoning. The high-pressure may be overestimated due to albite plagioclase that is problematic for the Hoisch (1990) Figure 1 for sample locations. b Temperatures determined using garnet-biotite (GB) thermometry (Berman, 1990;Ferry & Spear, 1978) with uncertainty representing the range of conditions at the specified pressures. If no P was measured, these temperatures represent the conditions from 0 to 10,000 bars (from Catlos et al., 2001). c Pressures determined using garnet-plagioclase-biotite-muscovite (GPBM) barometry (Hoisch, 1990) (from Catlos et al., 2001). d Conditions estimated for core or rim of garnet using Theriak-Domino (De Capitani & Brown, 1987;De Capitani & Petrakakis, 2010). Error either estimated from the area of the polygon created by intersecting garnet compositional isopleths for single paths, or the average of final conditions obtained from multiple paths. These are likely underestimated (see discussion in text). e "-" = not estimated.

Table 2 Summary of P-T Conditions for Analyzed Samples
Sample BR14 collected just below the GHC (Figure 1) yields the lowest conventional P-T conditions, whereas sample BR10 from lower structural levels has the highest (Table 2). Sample BR14 also contains matrix monazite grains that are 6.5 ± 1.4 Ma-3.6 ± 0.8 Ma (Catlos et al., 2007). The youngest monazite ages of ∼1 Ma are from the base of the MCT shear zone near the Munsiari Thrust. 40 Ar/ 39 Ar mica ages reported along transect suggest activity along the MCT (Vaikrita Thrust) at 9-8 Ma, and MCT-I (Munsiari Thrust) at 5-4 Ma . We sought to further evaluate the significance of these results through a more recently developed thermodynamic modeling routine (De Capitani & Brown, 1987;De Capitani & Petrakakis, 2010) and explore for the presence of these monazite young ages in additional rocks.

Approach
New P-T data are reported from 10 assemblages (Table 2), and 40 new in situ (dated in rock thin section) Th-Pb ion microprobe monazite ages were generated from four rocks (Table 3). Bulk rock and mineral compositional data, modeling results, and parameters, and details regarding the monazite geochronology are available as supporting information. We briefly describe the process of generating the data here. Note that no effort was made to link the monazite ages to the P-T paths, which would require more extensive petrochronological and geochemical efforts. Most of the dated monazite grains occur within the matrix, and monazite was not found in some samples that produce P-T paths.

Geochronology
Monazite ages were obtained using the CAMECA ims1280-HR ion microprobe at Heidelberg University over 2 days (see supporting information). The standards used were monazite 554 (45.3 ± 1.4 Ma, ±2σ, Harrison et al., 1995) and 44069 (Delaware, 424.9 ± 0.4, Aleinikoff et al., 2006 was applied for all grains. To test for potential matrix effects that are known for monazite with different Th abundances (e.g., Stern & Berman, 2001;Fletcher et al., 2010), a series of eight reference monazites was cross-calibrated using 44069 monazite as the primary reference. A new analytical procedure using high-energy ions (−40 eV relative to the nominal accelerating voltage of +10 kV) minimized monazite matrix effects to within ±2% (relative) of reported Th-Pb ages for most reference monazites, except for very high-Th monazite. The approximately ten times lower abundance of high-energy secondary ions, however, precluded taking this approach for the comparatively young Himalayan monazite samples. Instead, a conventional (Harrison et al., 1995) setup was used, for which an additional age uncertainty of ∼5% (relative) is estimated because of the slightly higher Th abundances indicated by primary-beam normalized ThO 2 + count rates for the unknowns relative to the reference monazites. This bias is typically within the stated uncertainties based on propagating counting and calibration curve uncertainties for the young monazites studied here.
Although the use of monazite standards is sufficiently old to determine U-Pb and Th-Pb at reasonable age precision, there is also a caveat that for the comparatively young Himalayan monazites studied here, only Th-Pb ages are relevant due to high uncertainties of U-Pb ages, and the potential of U-Pb (in particular 206 Pb/ 238 U) ages being affected by initial disequilibrium (i.e., 230 Th excess). Himalayan monazites are notorious for excess 206 Pb in U-Pb ages (see Figure 2 in Harrison et al., 2002). The U-Pb calibration using monazite 554 results in geologically meaningless ages, and thus we do not report U-Pb ages nor Concordia plots from these grains.

High-Resolution P-T Path Estimations
The approach to obtain P-T conditions and paths is the same as that outlined in Catlos et al. (2018) and Etzel et al. (2019). We describe the approach here in four steps. Each allows for an evaluation of sample suitability for the inherent thermodynamic assumptions involved in the process. At each step, an evaluation is made if the rock is appropriate for the next phase of modeling. We illustrate the process using Lesser Himalayan sample BR14, Greater Himalayan rock BR16B, and Martoli (Tethyan) Formation sample BR17 as examples (Figures 3-5  First, an isochemical phase diagram is created for each sample using rock bulk compositions, the software package Theriak-Domino (De Capitani & Brown, 1987;De Capitani & Petrakakis, 2010). The bulk rock compositions were obtained from rock chips using inductively coupled plasma spectrometry at Activation Laboratories. Note that no modifications in these compositions were made, and they are used as direct input for the development of the phase diagrams. For the phase diagram, we used the T. B. Holland and Powell (1998 with updates to solution models through 2010) thermodynamic dataset, and appropriate mixing models in the system MnO- Figure 3a). The specific solid solution models used are the same as in Catlos et al. (2018). We assume H 2 O saturation in these samples (i.e., H 2 O (100) in Theriak Domino), as is appropriate for these assemblages. We did not estimate Fe 3+ , which does not significantly affect results.
Compositional transects were made across garnet porphyroblasts using an electron microprobe. Isopleths of ±0.01 mole fraction spessartine, almandine, pyrope, grossular, and ±0.01 Mg-# (Mg/Fe + Mg) are plotted on the phase diagram and correspond with our closest approximation of the garnet central section (Figures 3a, 4a, and 5a). The core is defined as the portion of the garnet with the highest Mn Content. The range of ±0.01 mole fraction allows for some flexibility in searching for real core conditions. This initial P-T grid with intersecting garnet isopleths approximates the chemical system at the time garnet began growth. This diagram predicts a particular mineral assemblage and potential reaction in the region of intersecting core isopleths. If the mineral assemblage appeared unreasonable (i.e., no garnet in the intersecting stability field), or if the garnet core isopleths did not intersect, we do not continue the modeling process. In this case, the sample may have experienced open system behavior and/or modification of garnet compositions and cannot be used to make inferences regarding Himalayan tectonics.
In sample BR14, the topology of the assemblage in the region of the intersecting garnet isopleths appears along the garnet-in reaction line at 534°C ± 5°C, and 4.40 ± 0.40 kbar and suggest that garnet appears in the rock through the dehydration of chlorite ( Figure 3a; Table 2). In sample BR16B, the intersection isopleths are located outside of the garnet-in line along the 3% volume growth contour at higher-grade conditions (630°C ± 8°C, and 7.90 ± 0.40 kbar; Figure 4a, Table 2). This observation is consistent with conditions recorded by hanging wall assemblages and suggests the influence of diffusional modification. The BR17 garnet core isopleths intersect along the 1% garnet growth volume contour at 582°C ± 5°C and 5.67 ± 0.50 kbar ( Figure 5a, Table 2). The isochemical phase diagram also suggests a similar chlorite dehydration reaction is responsible for garnet growth in this sample. In all cases, the intersections are located in mineral stability fields that are consistent with their assemblages and anticipated conditions.
If the isopleths intersect and mineral assemblages are appropriate, the third step is to create the high-resolution P-T path (Moynihan & Pattison, 2013). In sample BR14, we obtained compositional data from three garnet transects using an electron microprobe ( Figure 3b). One transect was obtained across a single grain, and two transects were made from the central portion to the rim of two distinct grains. In GHC BR16B and Martoli Formation BR17, we obtained transects from core to rim across two distinct garnets. The ability to model P-T paths across more than one garnet and transect allow for an evaluation of the reproducibility of the paths and an estimate of uncertainty. The expectation is that the P-T paths from the same garnet and CATLOS ET AL.
10.1029/2020GC009353 8 of 29 Notation is analysis day\grain number\spot number. The first day's calibration curve is ThO 2 + /Th + = 0.1638Pb + / Th + -0.7400 ± 0.0035. This curve reproduced the standard ages to 45.0 ± 0.5 Ma (n = 7 spots) and 403.7 ± 4.3 Ma (±1σ) (n = 5), respectively. The second day's calibration curve is ThO 2 + /Th + = 0.0073Pb + /Th + +1.5465 ± 0.0035. This curve reproduced the standard ages to 45.3 ± 0.9 Ma (n = 15) and 421.0 ± 7.7 Ma (n = 3), respectively. b Percent radiogenic 208 Pb. c Value used in the age equation. d Measured ratio in the sample. Ideally, this value lies within the range of what was developed in the calibration curve. e Used in the common Pb age correction. f Monazite inclusion in garnet. multiple garnets in the same sample should record similar trajectories. However, the starting and endpoints may differ due to the proximity of the garnet core and extent of garnet rim preservation.
The garnet zoning profiles are smoothed using a Savitzky-Golay function, and data points were added based on the smooth profile to minimize the impact of missing analyses due to inclusions or cracks (Figures 3b, 4b, and 5b). A Matlab script applies the Theriak-Domino program (De Capitani & Brown, 1987;De Capitani & Petrakakis, 2010) to search a P-T grid for the smallest misfit between the modeled garnet core and measured composition, and then calculates the portion of the bulk composition sequestered in the first step of garnet growth (see Moynihan & Pattison, 2013  Isopleths intersect at the garnet-in reaction line, as indicated by the black polygon. This intersection represents the closest approximation to garnet core growth in and closed system. Garnet volume 0.5% contours are also overlaid on the diagram. (b) Compositional transect across three garnets in sample BR14. The data include raw electron microprobe, smoothed input into the model, and model predicted compositional zoning. (c) High-resolution P-T paths from sample BR14. Central section and rims are indicated. (d) Isochemical phase diagram for sample BR14 created using the final effective rock bulk composition generated by the Theriak-Domino approach. Isopleths of ±0.01 mole fraction spessartine, grossular, pyrope, almandine, and Mg-number of the compositional data point selected from the garnet's lowest Mn content are overlain on the diagram. We overlie the biotite Mg-number and plagioclase An-number isopleths, and conventional P-T conditions. The garnet and biotite isopleths intersect, as indicated by the black polygon. The intersection represents the closest approximation to conditions recorded by the garnet rim in a closed system. See supporting information compositions data and model outputs. See Table 2 for a summary of conditions. rock bulk composition to estimate an "effective" bulk composition for the next step of garnet growth. The process is repeated for all data points along a garnet zoning profile from the core to rim. Each step along a garnet traverse yields an estimate of the P-T conditions of incremental growth and a new effective bulk rock composition, ultimately culminating in a high-resolution P-T path (Figures 3c, 4c, and 5c).
The modeled P-T path predicts garnet compositional zoning (Figures 3b, 4b, and 5b). If the garnet behaved in an ideal system, the predicted and raw electron microprobe garnet compositions should be similar. This situation is seen in Figures 3b, 4b, and 5b, where the modeled, smoothed, and raw compositional data match within ±0.1 mole fraction for spessartine, pyrope, almandine, and grossular components. In most cases, the match is almost identical. The similarity in observed and P-T path predicted garnet compositions provides confidence that the approach is suitable for these garnets and their bulk compositions.
CATLOS ET AL.   Table 2 for a summary of conditions. P-T paths from multiple compositional transects across garnets from rim to rim, or from additional garnets in the same rock should also agree in overall shape and conditions. If they deviate significantly, this also suggests that the sample is inappropriate for the process. In sample BR14, we find the highest Mn content represents the lowest P-T conditions, as expected as the garnet grew with increasing T. All garnets grew over 530°C-565°C and 4.2-4.8 kbar intervals (Figure 4c). Likewise, the P-T paths from sample BR16B are nearly isothermal at 630°C ± 20°C and show the rocks grew over increasing P from ∼7.9 to 8.1 kbar. The paths for sample BR17 are nearly identical and show a decrease and then increase in P over 5.0-6.0 kbar over the same 580°C-600°C interval. Path 2 in this sample shows an increase from 5.5 to 6.0 kbars before decreasing in P. This fluctuation is likely due to the distribution of garnet grossular compositions. Minor changes in CATLOS ET AL.  Table 2 for a summary of conditions. P (±10-50 bars) are not due to changes in conditions but represent a range in uncertainty as the program searches for the best fit for the next parameter.
The fourth and final step is to create a phase diagram using the last estimated effective bulk composition and garnet rim composition (Figures 3d, 4d, and 5d). If available, isopleths of ±0.01 An content (Ca/ [Ca + Na + K]) from matrix plagioclase or ±0.01 Mg-# (Mg/Mg + Fe) from matrix biotite are overlain on the rim phase diagrams. This process allows an evaluation if the matrix mineral compositions from these rocks yield conditions that overlap or lie near those corresponding to the garnet rim, and was done for samples BR14 (Figure 3), BR07, BR10B, and BR26 (see figures in supporting information).
Rim P-T conditions are compared to those obtained using conventional GB thermometry, and GBMP barometry for Lesser Himalaya samples BR14, BR26, BR07, BR09, and BR10B (Table 2; Figure 4d, see supporting information). The goal for the comparison of the conventional and isopleth conditions is to identify in which rocks conditions overlap and ascertain the potential reasons for any disparity for others. Most Himalayan P-T conditions are reported using conventional approaches and have been used to develop sophisticated models for the development of the range, thus a comparison of these results with the isopleth method is warranted.

Sources of Uncertainty
Assumptions underlie any P-T estimate by thermodynamic modeling, including conventional or the approach we outline here. Sources of uncertainty include the thermodynamic properties inherent in the internally consistent T. B. Holland and Powell (1998) database (e.g., molar enthalpy of formation, molar entropy, molar volume, heat capacity, bulk modulus, Landau parameters, and Margules parameters, White et al., 2014). Thermobarometry requires the assumption of equilibrium, which has never been proven for any rock system (e.g., Spear & Peacock, 1989). We assume closed system behavior and that the primary composition of the mineral phases and the bulk rock have not changed significantly (e.g., Lanari & Engi, 2017). The compositional core may not coincide with the geometric garnet center (e.g., Spear & Daniel, 1998), and the conditions we report for the core may be too precise.
However, a significant value of the P-T path approach is that when systems stray from any of these assumptions, a user can detect the problem. The uncertainty we report in the conditions represents the average from the results of multiple paths or is the range of the area estimated via intersecting isopleths if we modeled only one garnet transect. The precise uncertainty is challenging to determine due to a large number of factors incorporated into the Theriak-Domino approach (see also discussion in Catlos et al., 2018;Etzel et al., 2019). The paths reported here are reproducible, and the uncertainty values are used to gauge the level of agreement between results.
We emphasize that the high-resolution P-T paths reported here approximate how a garnet with a specific type of compositional zoning would behave in a closed system of a known bulk composition as it evolves during increasing temperature. Rocks are open systems, but in what appears to be the case for the samples identified here, they can be defined such that they approach this ideal scenario. Ultimately, a most important check on the feasibility of the P-T path and isopleth conditions is if the results seem geologically reasonable and consistent with mineral assemblages. The strategy has proven successful in other studies (e.g., Catlos et al., 2018;Craddock Affinati et al., 2019;Etzel et al., 2019;Kelly et al., 2015;Moynihan & Pattison, 2013).

Rock Textures and Monazite Ages
We dated monazite grains from the GHC (BR20), and LHF (BR29 and BR27B), and Martoli Formation of the Tethyan metasediments (BR17) located just north of the unit's mapped contact with the GHC (Figure 2). Figures 6-9 show thin sections for each rock, which were scanned using BSE to identify the monazite grains and evaluate their microstructural locations. The Tethyan and GHC samples have the oldest ages, which range from 50.3 ± 0.6 Ma to 24.7 ± 0.8 Ma (Table 3). The oldest grain is an anhedral inclusion in a rounded Martoli Formation garnet in sample BR17 ( Figure 6). Randomly oriented biotite grains in sample BR17 become aligned near the zonal crenulation foliations, which are comprised of thicker amalgamations of biotite. These define shear bands (Figure 6a). The preferred orientations of biotite grains within microlithons in between the shear bands vary. Quartz grains in between the shear bands have sharp boundaries and do not show significant bulging or impinging textures. Staurolite is a minor component of this sample, but when it appears, it aligns with the crenulated biotite foliation. Younger monazites grains in sample BR17 are near or in contact with randomly oriented biotite, whereas the garnets are near or within the foliation. Garnets are ∼100 μm in diameter, are rounded or fragmented, and appear syntectonic. Most are inclusion-free or have small (∼10 μm in long direction), rounded quartz inclusions. Matrix monazite grains in this sample CATLOS ET AL.   Table 3 for monazite age information.
are small (∼10 μm in long direction) and rounded. The anhedral monazite inclusion is located near garnet cracks.
Like sample BR17, GHC sample BR20 is foliated, but the foliation is defined by both aligned biotite and plagioclase (Figure 7). The biotite foliation is spaced and disjunctive and forms thin discrete anastomosing layers between aligned plagioclase and quartz augens with irregular grain boundaries. Smaller, rare, and rounded inclusions of quartz (∼20-50 μm) are also present within larger plagioclase. Garnet is also rare, small (∼50-100 μm), and rounded, and did not contain monazite inclusions. Overall, monazite ages in sample BR20 range from 45.2 ± 1.3 Ma to 27.3 ± 0.8 Ma (Table 3). Monazites are found primarily along the grain boundaries of biotite, muscovite, or plagioclase, and one is within apatite (Figure 7d). We find no correlation between monazite age and textural location.   Table 3 for monazite age information.  Table 3 for monazite age information.  Table 3 for monazite age information.
Samples BR27B and BR29 were collected from the same outcrop, ∼10 km south of sample BR20 (Figure 2). Biotite in these rocks are thin, elongated, and spaced, or have coarser sizes (Figures 8 and 9). Coarser biotite grains exist as either distinct grains or elongated groups of clusters. The thinner biotite lathes comprise a weaker disjunctive foliation in sample BR27B. Chlorite in sample BR27B appears to be primary, as it is in similar textural relations are the biotite and is not primarily associated with garnet. Sample BR29 has a continuous foliation, defined mainly by thinner elongated biotite, muscovite, and ilmenite grains. Thicker biotite grains overprint the continuous foliation and elongated quartz veins. We did not find chlorite in sample BR29. Rounded to subhedral garnets in both rocks form in clusters.
The youngest monazite grains dated in this study are in LHF rocks BR27B and BR29 (Figures 8 and 9), and range from 9.3 ± 0.6 Ma to 3.0 ± 0.2 Ma. Monazites in both samples are small (∼30 m) rounded or anhedral, and several are found within or associated with the coarser biotite. These ages are significantly younger than those in the GHC rocks. All grains are found in the rock matrix in association with either chlorite or larger biotite grains. In sample BR29, we find the youngest monazite along the edges of garnet, but garnets are either inclusion-free or have quartz inclusions. In sample BR27B, garnets are inclusion free or have small ilmenite inclusions.

P-T Conditions and Paths
We obtained P-T conditions from a rock from the Martoli Formation (BR17), one GHC samples (BR16B), and eight LHF rocks (BR07, BR09, BR10B, BR14, BR26, BR27B, BR29B, and BR30B) (Figure 1). Table 2 lists the samples and their conditions estimated using conventional and Theriak-Domino approaches, whereas Figure 10 summarizes the paths from 10 of the rocks.
Based on the isochemical phase diagrams, garnet most likely appears in all rocks through dehydration of chlorite. Overall, the Martoli Formation and GHC samples record higher P-T conditions compared to LHF garnets (Table 2, Figures 3-5). GHC and Martoli Formation garnet core isopleth intersections overlap the 1%-3% volume growth contours indicating that we likely did not intersect the real garnet core, and the samples experienced modification (Figures 4 and 5). Core conditions for all LHF garnets lie along the garnet-in reaction curve, except for samples BR09, BR26, BR27B, and BR29B, in which isopleth overlaps along the 1% garnet volume growth curves (see supporting information). P-T paths from rocks above the MCT were only obtained for BR16B and BR17 (Figures 4, 5, and 10). Two paths from GHC sample BR16B are isobaric, and average 8.0 ± 0.1 kbar from 630°C ± 8°C at the core to 642°C ± 10°C at the rim. Two paths from Martoli Formation sample BR17 show "N-shaped" P-T path at conditions much lower than sample BR16B (Figure 10a). The N shape develops as the garnet experiences a pressure increase-decrease-increase trajectory. The BR17 garnet grew from 582°C ± 5°C to 600°C ± 5°C, which is ∼40° lower than the T recorded by sample BR16B. This rock also experienced more moderate CATLOS ET AL.  baric conditions, from 5.7 to 5.9 kbar. Within that pressure range, however, the garnet experienced ∼1 kbar fluctuations in P.
Most of the LHF samples also show fluctuations, except for samples BR07, BR09, path two from BR26, and path one from BR29. As stated previously, the minor variations in P (±10-50 bars) are not due to changes in conditions but instead represent a range in uncertainty as the program searches for the best fit for the next parameter. Most LHF garnets show simple increases in P and T, consistent with burial. Although all LHF garnets grew over a narrow T interval, with the majority falling within the 540°C-580°C, the rocks show a wide range of baric conditions, from a low of 3.9 to 5.7 kbar (Figure 10).

Comments on the Theriak-Domino P-T Modeling Approach
We report conventional rim T from five of the samples and P from three of the rocks (Table 2). Most P constraints were not obtained due to inappropriate mineral assemblages or An contents using the Hoisch (1990) calibration. Most of the conventional rim T agrees with results produced by the Theriak-Domino rim approach, except sample BR10. This garnet's high GB T (720°C ± 50°C) is at odds with its prograde zoning profile (see figures in supporting information). Thus, we have more confidence in the T-D rim conditions for this rock. The conventional rim P is 2-4 kbar higher than those provided by Theriak-Domino (Table 2), which is a common observation (Catlos et al., 2018;Etzel et al., 2019). Garnet grossular, Mg-number, pyrope, and almandine isopleths are typically found to be parallel, with spessartine following the garnet-in reaction line. P estimates, therefore, may be underestimated using the Theriak-Domino approach-most rim P for the footwall BR samples are 5 ± 1 kbar.
Overall, we have confidence that the high-resolution P-T paths and Theriak-Domino conditions are an accurate representation of what the sample may have experienced during metamorphism. These are consistent with mineral assemblages and rock textures, the P-T paths accurately predict observed garnet zoning, and isopleth intersections occur for garnet core and rim compositions within ±0.01 mole fraction grossular, almandine, pyrope, and spessartine, and ±0.01 Mg/(Mg + Fe) compositions. Garnet rim isopleths intersect matrix mineral compositions in some cases (samples BR07 and BR26). Garnet P-T paths from the same sample show similar trajectories, including complex fluctuations in P. Although precise uncertainty cannot be determined, the abundance of evidence indicates that this value is likely well within levels attributed to conventional approaches. We speculate that the error that should be attributed to the Theriak-Domino generated conditions is similar to conventional thermobarometry (±25°C and ±1 kbar, Palin et al., 2016;Spear & Peacock, 1989).
Most P-T paths record conditions that span a T interval of 20°C-40°C, regardless of the unit from which the sample was collected. Some P-T paths from the same garnet or garnets from the same sample do not precisely match (e.g., Figures 3 and 4). The recognized uncertainty in the absolute values of the P-T paths resolves most of the issues regarding the path agreement. Other differences may be due to the choice of garnet transect or error inherent in the parameters of the solution models as the Theriak-Domino program searches for the most appropriate condition. The uncertainty in path conditions should be considered when comparing path shapes. We attempt to model seven other rocks from the transect, but these were not considered reliable. Failures occur because assumptions in the Theriak-Domino approach are inappropriate for the rock's history. These samples may have experienced alteration since initial garnet crystallization. Multiple episodes of garnet growth may have occurred, as well as changes in bulk compositions. These changes in bulk rock can take place due to extensive fluid interaction, extreme retrogression, or the introduction of new material. Garnets in some samples were consumed by chlorite, and in the case of hanging wall rocks, were affected by high T (>650°C). The higher-grade conditions led to the modification of initial garnet compositions. Only one hanging wall rock was able to be modeled, and this sample experienced T that would have predicted diffusional zoning (e.g., Spear & Peacock, 1989). The garnet is small (∼100 μm in diameter) but was able to preserve its composition. Preservation may have occurred due to faster exhumation that hindered modification. Another option for failure is the use of an inappropriate bulk rock composition. Most samples in which the Theriak-Domino approach worked well are MCT footwall assemblages.

Modeling LHF P-T Paths and Monazite Ages
The Himalayas has been the focus of a large number of geodynamic models that make predictions regarding the paths that rocks follow during collision (Beaumont et al., 2001;Burg & Podladchikov, 2000;Carosi et al., 2018;Guillot & Allemand, 2002;Imayama et al., 2010;Larson et al., 2013;Li, 2014;Montemagni et al., 2020;Rolfo et al., 2014;Searle et al., 2017;Waters, 2019;Yin, 2006). However, the primary focus of many models of Himalayan development has been the GHC, and the majority of the P-T paths we obtained are from the MCT footwall. To interpret the results and compare to high-resolution P-T paths from the MCT footwall from the Marsyangdi transect in central Nepal, we use the Harrison et al. (1998) model modified in Catlos et al. (2018) (Figure 11). This is a two-dimensional numerical heat flow model of thrust fault systems with simple geometries. The diffusion-advection equation is solved by finite-difference methods using an explicit direct method to simulate the fault displacement, topographic changes, and internal radiogenic heat, followed by an alternating-direction implicit method to calculate heat diffusion of the crust at each time step. Zero-flux lateral boundaries are imposed, as are constant surface temperatures (25°C) and constant flux at the bottom of the grid. The internal heating is given by a depth-exponential decaying function, and a consistent initial crust geothermal gradient is calculated following Turcotte and Schubert (1982). The initial surface, basal heat flux, and internal heating depth are prescribed to yield a desired initial average geotherm consistent with the studied geological framework. The surface location on the grid is either constant (i.e., uplift = denudation) (Harrison et al., 1998), or changes in topography can be simulated by slowly varying the height of the surface (Catlos et al., 2018). We choose this model because it is a quantitative framework that can predict P-T path trajectories for multiple samples. The location of significant thrust systems, including the MHT, MCT, and MBT, is constrained by seismic and structural interpretations. Parameters appropriate to crustal rocks and flat ramp geometry of the faults are the same as in Catlos et al. (2018) (Table 4, Figure 11). We also stress that the model includes a general link between the age and monazite appearance with the P-T conditions of the sample. The ages are meant to constrain the timing of fault systems within the model that are responsible for moving particles and developing the P-T paths. The paper does not seek to directly link any monazite ages with particular P-T conditions, as this requires detailed geochemistry and numerous assumptions about the rock chemistry, fluid history, and monazite protolith. Instead, the ages are only used to provide broad constraints on when faults within the Himalayas were moving.
All samples are displaced by motion along the MCT, MCT-I, MBT, and MHT during specified periods. Table 4 indicates the times that these structures are active and the amount of slip that they accommodate. The model measures heat flow at locations due to varying conditions. Garnets can only grow as T increases (e.g., % volume growth lines in Figures 3-5), whereas advection cools and halts garnet growth. The P that garnet records are due to emplacement or erosion of a topographic overburden.
The model predicts that the P-T paths recorded by MCT footwall garnets result from thermal advection combined with alteration of topography. The model-predicted P-T paths have solid blue lines that represent prograde portions of rock trajectories (Figures 11c-11e). Dashed lines correspond with periods when samples experience a T decrease. We use the model to generate P-T paths for rocks at equivalent structural positions as those collected in this study and compare them to the shapes and conditions produced using the Theriak-Domino approach.
Samples move over a thrust flat-ramp geometry at different times along the MCT, MHT, or MBT (Figures 11a and 11b). Slip is accommodated along the MHT-MCT ramp between 25 and 15 Ma and along the MBT-MHT ramp from 15 to 8 Ma at times, rates, and distances listed in Table 4. At 8 to 2 Ma, fault ramps within the LHF become active, labeled as the MCT from 8 to 6 Ma, and MCT-I from 6 to 2 Ma (Figure 11b). These time frames are consistent with the monazite ages obtained here (Table 3; Figures 6-9) but were initially constrained by ages from within the MCT shear zone elsewhere (see Harrison et al., 1998). The MCT dip angle is higher (30°) from 8 to 2 Ma compared to its position from 25 and 18 Ma (7°). The flat ramp of the MCT-I (labeled as the MHT in Figure 9b) is below that of the MCT.
In the model, the MCT footwall speed rate, the rate at which a sample would travel along the MCT within the footwall is 5 km/Ma from 25 to 18 Ma. This rate differs from the hanging wall speed rate, which is maintained at 10 km/Ma until topography progressively accumulates to a maximum height of 3.5 km. An increase in topography translates to higher P changes recorded by the garnets while allowing them to grow as T increases. Emplacement of overburden or burial increases P. Erosion decreases P. Note that minor P CATLOS ET AL. fluctuations (50-100 bars), like those seen in the P-T paths of samples BR26 and BR29, can develop as the Theriak-Domino program searches for the best fit of model parameters. These are not considered significant in determining the tectonic history of the rock.
In the model, the 3.5 km of topography is achieved at 18 Ma (Figure 11a). At this time, a period of nonslip is introduced and lasts until 15 Ma. This time frame allows heat to advect across the MCT and provides for a period of denudation to occur at a rate of 1.5 km/Ma. This is reflected by the lack of topography seen in Figure 11b. Although the model transfers motion from the MCT zone to the MBT-MHT ramp from 15 to 8 Ma (e.g., Burbank et al., 1996), other fault systems south of the MCT, such as those present in the LHF (e.g., Bose & Mukherjee, 2019) may fulfill a similar role without changes in results. Motion along the MBT sensu stricto is not required. The model reactivates the MCT from 8 to 6 Ma, and the MCT-I accommodates slip within the LHF from 6 to 2 Ma (Figure 11b). Processes that occur within the MCT and its footwall can be envisioned as the development of the LHF duplex. We use the model to explore if the Bhagirathi River P-T paths are consistent with the results generated 650 km further east along the Marsyangdi River in central Nepal (Catlos et al., 2018). The model traces the P-T trajectories of each sample inside the grid. The simulation of the topography and the resulting isotherms, especially near the surface, are probably quite rough, but a precise description of the thermal state of the crust is not the goal. The model is only intended to provide an estimate of the P-T evolution of samples within the grid. We note that there is no information about the timing or location of the samples at the time when the P-T values were recorded. Only the actual relative location of the sample with respect to the present-day thrust fault is known, but this could be quite different from its locations during the Late Miocene.
Still, we believe that it is quite significant that there are zones on the grid that are consistent with the P-T measurements, given the restrictions imposed by the overall system. As Figure 11 demonstrates, P-T paths from both the Bhagirathi and Marsyangdi transects across the MCT can be reproduced with the model parameters. All LHF high-resolution P-T paths (n = 7) show a reasonable fit with the modeling parameters and are similar to those obtained from the Marsyangdi River drainage in central Nepal. Note CATLOS ET AL.  that Early Miocene to Pliocene monazite ages are also found in footwall samples in central Nepal (Catlos et al., 2001).
Monazite ages from LHF rocks are significantly younger than the Miocene phase of ascribed to motion along the MCT. If these ages are the result of Pb loss from initially 20 Ma age grains, they would have had to experience >50% Pb loss through a process that does not require tectonic motion. Monazite has a high closure T (>800°C; Cherniak et al. 2004;Gardes et al. 2006). The footwall samples with the Late Miocene to Pliocene monazite grains do not show evidence of high-grade conditions. These would include significant diffusional modification of garnet zoning. The rocks also contain no textural evidence of fluid retrogression that would drive wholesale monazite recrystallization (Figures 8 and 9). We do not know the protolith of monazite in these or the GHC samples, but we speculate that they likely formed via common reactions that include allanite or preexisting monazite. These minerals are found in other samples at lower metamorphic grades (Catlos et al., 2007).
The P-T paths themselves require footwall imbrication, and the model suggests the outlined parameters are appropriate for rocks from both the Marsyangdi and Bhagirathi river transects. Other models anticipate similar behavior in the MCT shear zone (e.g., Braden et al., 2018;Caddick et al., 2007;Groppo et al., 2009;Herman et al., 2010;Imayama et al., 2020;Larson et al., 2015;Montemagni et al., 2020;Mosca et al., 2012;Mottram et al., 2014). The Theriak-Domino P-T paths from both the Bhagirathi and Marsyangdi regions cannot develop if only a single phase of Miocene MCT motion is imposed. From 8 to 2 Ma, the model continues the imbrication process with the activation of the MCT-I ( Figure 11). This time frame corresponds to the phase of monazite crystallization and the development of a duplex in the LHF. Most of the Bhagirathi River LHF rocks fit the P-T paths suggested during this time and is consistent with the distribution of monazite ages. We did not find evidence of extreme exhumation (12 mm/year since the Pliocene) in the Bhagirathi River footwall samples, as we did with the one sample from the Marsyangdi River region (Figure 11e; Catlos et al., 2018). The highest P footwall samples are BR26 and BR07, which record ∼5.6 kbar. The baric conditions suggest a maximum erosional exhumation rate of ∼7 mm/year (3.7 kbar/km baric gradient, 20-22 km depths over 3 Myr). Samples that experienced these higher exhumation rates may be present in NW India, as evidenced by P-T paths and conditions obtained from LHF rocks in the adjacent Alaknanda and Dhauli Ganga valleys that suggest exhumation from greater depths (6.3-7.5 kbar, Iaccarino et al., 2020;S. S. Thakur et al., 2015S. S. Thakur et al., , 2018.

Model Fits for Rocks Above the MCT
Some monazite ages from the GHC and Martoli Formation samples are older than the Miocene framework for the model as described in Figure 11. The Eocene to Oligocene monazite ages from the Tethyan and GHC samples are ascribed to the Eocene Eo-Himalayan event when the GHC collided with and subducted beneath Asia (see review in Carosi et al., 2016;Catlos et al., 2002;Thoeni et al., 2012;Waters, 2019). As convergence continued into the Miocene, the GHC exhumed along the MCT (e.g., Carosi et al., 2018;Martin, 2017aMartin, , 2017b. The Miocene monazite ages we obtained from the samples, including as inclusion in garnet from sample BR21, are consistent with the timing of this event. A single monazite age population should not be expected for GHC rocks based on its multistage history and is not commonly reported for the unit elsewhere (e.g., Catlos et al., 2001;Kohn et al., 2005;Martin et al., 2007). As with the LHF monazite ages, we do not ascribe the range to Pb loss because of the high monazite closure T (e.g., Cherniak et al., 2004;Gardes et al., 2006). Although matrix grains and inclusions in garnet near microcracks can be affected by fluid reactions, the ages we report here from GHC samples are found elsewhere across the range and are consistent with events affecting the formation of the unit.
The Martoli Formation is a pelitic lens within the broader context of Tethyan rocks and is described along the Bhagirathi and Dhauliganga rivers (Sachan et al., 2010;Searle et al., 1993). The oldest age from the monazite inclusion in garnet from Martoli Formation sample BR17 (50.3 ± 0.6 Ma). Monazite inclusions in garnet can be shielded from Pb loss and interactions with fluids that lead to dissolution-reprecipitation reactions (e.g., Catlos, 2013). The age is consistent with broad estimates for the timing of Indo-Asia collision (e.g., Guillot et al., 2003;Hu et al., 2016;Najman et al., 2017;Rowley, 1996;Tong et al., 2017) and suggests that metamorphism and ductile mid-crustal processes are recorded at the same time or soon after (<10 my) collision, depending on timing estimates.
In many models for Himalayan development, the GHC is a wedge in which material extrudes toward the south between the MCT and STDS (e.g., Beaumont et al., 2001;Webb et al., 2011;see review in Mukherjee, 2015;Searle et al., 2017). Extrusion of the GHC could be driven in part by large-scale erosion that changes boundary conditions resulting in its exhumation and material flow (Kohn, 2008;Mukherjee, 2015). In this scenario, the GHC transforms into a lithospheric channel. Evaluating models for GHC extrusion requires a detailed understanding of the P-T-t paths of its rocks as they experienced the transition from convergence and subduction to their final exhumation (e.g., Caddick et al., 2007;Catlos et al., 2001Catlos et al., , 2018Corrie et al., 2010;Kohn, 2008). However, high-grade conditions work to erase primary garnet zoning through diffusion, which leads to P-T inconsistent with mineral assemblages and eliminates the possibility of generating P-T paths.
Due to diffusional homogenization, only one GHC sample from the Bhagirathi River transect yields interpretable paths (Figures 10 and 11). Catlos et al. (2018) also estimated one GHC P-T path from the Marsyangdi River drainage. In the model (Figure 11), three possible depths (26.6, 27.5, and 28.5 km) for the isobaric trajectory of GHC rocks are proposed for the samples as they move along the MHT flat. Motion along a décollement or other shallowly dipping structure within the MCT hanging wall allows P to remain constant. This motion is best reproduced by sample BR16 and the Marsyangdi River GHC garnet, which are now found just north of the MCT exposure ( Figure 2; see Catlos et al., 2018). A ramp-flat geometry of the GHC best fits their high-resolution P-T paths, in terms of both shape and conditions.
The high-resolution P-T path for sample BR17 is the first obtained from a sample from the Tethyan Formation. We do not place sample BR17 in the context of the model in Figures 11a and 11b, as it is far north of the MCT and its monazite inclusions in garnet, and thus our interpretation of its P-T paths, are older than the model run time frame. We did, however, evaluate the rock within the Etzel et al. (2019) thermal model, which describes the potential geological history appropriate for similar N-shaped P-T paths. Although the present-day tectonic setting of western Turkey is characterized by extension, the conditions for garnet-growth was in a collisional regime (main Menderes metamorphism). The Etzel et al. (2019) model is similar to that in Figure 11 but is modified to reproduce P-T paths using a slightly hotter initial geotherm (30°C/km average between 0 and 20 km) and deeper placement of sample BR17 as an initial condition (∼19 km depth using a 3.7 kbar/km baric gradient). These conditions are needed for the model to reproduce segments of the P-T path. Other parameters (e.g., Table 4) remain the same.
One possible scenario to reproduce the BR17 P-T paths is that the rock experiences topographic denudation to account for the garnet's P decrease from ∼5.9 to 5.2 kbar. This situation is followed by activation of a thrust fault located ∼10 km north, which places ∼2 km of overburden and contributes to garnet growth with increasing T. Although the specific structure that metamorphosed the sample is unknown, the GHC and Martoli Formation are significantly imbricated units (e.g., Searle et al., 1993). Fault movement would need to occur quickly, as the garnet records only a small T change of ∼18°C. Overall, the P-T paths are consistent with the scenario in which BR17 was initially at ∼19 km depth and suffered an episode of denudation of ∼2 km, followed by a burial of about the same magnitude. The P-T paths record surface effects, as opposed to displacement, because T changes are small (∼18°C), and T monotonically increases. Note that this is only one option for a possible tectonic scenario, as additional Martoli Formation samples and age data, and more structural context is required to evaluate the significance and better model the BR17 P-T path. According to Searle et al. (1993), this sample would be located in the footwall of the upper splay of the STDS.
The estimation of the shallower focal mechanisms for both the Gorkha and Uttarkashi earthquakes, or at least some of their aftershocks, may have nucleated above the MHT within the LHF duplex system (Bai et al., 2016;Robinson & Martin, 2014;Robinson et al., 2003;Whipple et al., 2016). Figure 2 shows the approximate location of the Gorkha earthquake overlain onto one of the first geological cross-sections drawn across the Himalayan range (Schelling & Arita, 1991). More recent cross-sections are available, but most are consistent with the general outline and images of the subsurface (e.g., DeCelles et al., 2001;Murphy, 2007;Robert et al., 2011;Robinson & Martin, 2014;Sorkhabi, 2010;Subedi et al., 2018). The Gorkha earthquake hypocenter lies along a low-velocity layer using the GCMT solution, whereas the epicenter is within the LHF duplex using the USGS data (Arora et al., 2017).
An aftershock of the Gorkha earthquake lies along the MCT, which defines the northern limit of the LHF duplex (2015-04-25, 17:42:53, 28.2380°, 85.8290°, Mw 5.1, USGS catalog). The idea of a presently seismogenic MCT and the potential for activity within the duplex has been proposed several years before the Gorkha event based on focal plane solutions for earthquake events in western Nepal and India (Baranowski et al., 1984;De & Kayal, 2004;Mukhopadhyay, 2011;C. Thakur and Kumar, 1994). The LHF duplex has long been known to be a ∼50-km-wide seismogenic zone of predominately moderate earthquakes (Bai et al., 2016;Cattin & Avouac, 2000;Khattri & Tyagi, 1983;Mahajan et al., 2010). Bai et al. (2016) suggest that a thrust system within the LHF is the most seismically active region in the Himalayas and accommodates most of its elastic strain accumulation. Alternatively, the shallower events are explained by a segmented MHT that includes a ramp (He et al., 2018;Hubbard et al., 2016;Pandey et al., 1995). McNamara et al. (2017) suggest the Gorkha earthquake and its aftershocks define a broad, 10-15 km thick subduction channel between the Eurasian and Indian plates. The rupture behavior of the Gorkha earthquake resembles a megathrust event at a subduction interface (Yue et al., 2017).
The applicability of the high-resolution P-T paths reported here to deciphering the present-day dynamics and possible architecture of the LHF duplex system requires that the geometric constraints of Himalayan architecture remained constant over the last few million years (e.g., Hubbard et al., 2016). Because Pliocene-age and younger monazite grains exist within the MCT shear zone in central Nepal and over 650 km further west in NW India (1-4 Ma, Catlos et al., 2001Catlos et al., , 2007, these locations may be helpful in this regard. Monazite can grow in equilibrium with garnet in pelitic assemblages (Catlos, 2013;Catlos et al., 2001;Foster et al., 2000;Pyle et al., 2001;Yang & Pattison, 2006;Zhu & O'Nions, 1999). If these monazite grains time Pliocene or younger garnet growth, the P-T conditions and paths of the garnet associated with them lend considerable insight into the present-day framework and dynamics of the Himalaya. Rocks sampled from similar structural levels within the MCT shear zone at disparate locations display a remarkable consistency in P-T path and conditions and are consistent with a model in which the MCT shear zone develops individual rock packages move at different times ( Figure 11). The results indicate the MCT and associated fault systems within the LHF duplex can be considered when identifying accommodation structures of Himalayan seismic events and remove the need for the placement of additional fault segments associated with MHT.

Conclusions
Despite being studied for decades, the geometry of the deeper portions of the Himalayan range is unclear and likely varies along strike (see Hazarika et al., 2017). The shapes of high-resolution P-T paths from central Nepal indicate the potential for garnets to record the imbrication process within the LHF duplex, which accommodates a significant amount of convergence within the range (Catlos et al., 2018). We report here additional high-resolution P-T paths from nine rocks collected from a transect 650 km further west across the MCT in NW India. The paths include one from the GHC and one from a pelitic assemblage from the Tethyan metasediments. We supplement the results with monazite geochronology and compare the ages and P-T paths to those in central Nepal. In that study, only one GHC garnet was able to be modeled. That sample and the one from similar structural levels from NW India record isobaric P as T increased, likely due to growth along the MHT flat during convergence. The Tethyan garnet records ∼1 kbar fluctuations in P, consistent with a period of erosion followed by ongoing collision (Etzel et al., 2019). The oldest monazite inclusion in garnet from the Tethyan sample is 50.3 ± 0.6 Ma, which is consistent with the onset of Himalayan convergence at this time. GHC monazite ages are compatible with metamorphism during the Eocene Eo-Himalayan event and Miocene MCT motion. MCT footwall monazite grains are significantly younger than those found in the hanging wall and are Late Miocene to Pliocene. The ages, as young as 3.0 ± 0.2 Ma, are consistent with a model developed for central Nepal that includes imbrication in the MCT footwall.
If the present-day geometry of the Himalayas has likely not changed significantly since the Pliocene, the results suggest that fault systems within the MCT footwall and the LHF duplex should be considered when evaluating the hazards and attributing seismic events to particular fault systems in the range. Although the MHT is undisputedly a significant factor in accommodating Himalayan seismic activity across the entire range, MCT footwall fault systems that were operating during the more recent geological past may explain some of the shallower hypocenters reported for both the Gorkha and Uttarkashi earthquake and their aftershocks.

Data Availability Statement
Data supporting the conclusions of this paper is publically available from Texas Data Repository Dataverse (https://doi.org/10.18738/T8/K0NKVT).