Subduction metamorphism in the Himalayan ultrahigh ...

Subduction metamorphism in the Himalayan1

ultrahigh-pressure Tso Morari massif: an integrated2

geodynamic and petrological modelling approach3

Richard M. Palin1,2*, Georg Reuber1, Richard W. White1, Boris J. P. Kaus1, and Owen M.4

Weller35

1Institute of Geosciences, Johannes-Gutenberg University of Mainz, 55128 Mainz, Germany6

2Department of Geology and Geological Engineering, Colorado School of Mines, Golden,7

80401, Colorado, USA8

3Department of Earth Sciences, University of Cambridge, Cambridge, CB2 3EQ, UK9

*Corresponding author: [email protected]

ABSTRACT11

The Tso Morari massif is one of only two regions where ultrahigh-pressure (UHP)12

metamorphism of subducted crust has been documented in the Himalayan Range. The13

tectonic evolution of the massif is enigmatic, as reported pressure estimates for peak14

metamorphism vary from ⇠2.4 GPa to ⇠4.8 GPa. This uncertainty is problematic for15

constructing large-scale numerical models of the early stages of India–Asia collision. To16

address this, we provide new constraints on the tectonothermal evolution of the massif via17

a combined geodynamic and petrological forward-modelling approach. A prograde-to-peak18

pressure–temperature–time (P–T–t) path has been derived from thermomechanical19

simulations tailored for Eocene subduction in the northwestern Himalaya. Phase20

equilibrium modelling performed along this P–T path has described the petrological21

evolution of felsic and mafic components of the massif crust, and shows that di↵erences in22

their fluid contents would have controlled the degree of metamorphic phase transformation23

in each during subduction. Our model predicts that peak P–T conditions of ⇠2.6–2.8 GPa24

and ⇠600–620 �C, representative of 90–100 km depth (assuming lithostatic pressure), could25

1

have been reached just ⇠3 Myr after the onset of subduction of continental crust. This26

P–T path and subduction duration correlate well with constraints reported for similar27

UHP eclogite in the Kaghan Valley, Pakistan Himalaya, suggesting that the northwest28

Himalaya contains dismembered remnants of what may have been a ⇠400-km long UHP29

terrane comparable in size to the Western Gneiss Region, Norway, and the Dabie-Sulu belt,30

China. A maximum overpressure of ⇠0.5 GPa was calculated in our simulations for a31

homogenous crust, although small-scale mechanical heterogeneities may produce32

overpressures that are larger in magnitude. Nonetheless, the extremely high pressures for33

peak metamorphism reported by some workers (up to 4.8 GPa) are unreliable owing to34

conventional thermobarometry having been performed on minerals that were likely not in35

equilibrium. Furthermore, diagnostic high-P mineral assemblages predicted to form in Tso36

Morari orthogneiss at peak metamorphism are absent from natural samples, which may37

reflect the widespread metastable preservation of lower-pressure assemblages in the felsic38

component of the crust during subduction. If common in such subducted continental39

terranes, this metastability calls into question the reliability of geodynamic simulations of40

orogenesis that are predicated on equilibrium metamorphism operating continuously41

throughout tectonic cycles.42

Keywords: Tso Morari massif; ultrahigh-pressure metamorphism; metastability; forward43

modelling; overpressure44

1 INTRODUCTION45

The Tso Morari massif, northwest India (Fig. 1), is one of only two examples of46

ultrahigh-pressure (UHP) metamorphism in the Himalayan Range and provides evidence47

for deep subduction of the Indian continental margin beneath Asia during the early Eocene48

(Epard and Steck, 2008). Constraining the petrological and metamorphic49

pressure–temperature–time (P–T–t) evolution of such eclogite-facies rocks is critical to50

understanding the geodynamic processes responsible for the burial and rapid exhumation51

of crustal material during collisional orogenesis (e.g. Warren et al., 2008). However, the52

2

tectonothermal evolution of the massif is controversial, with reported conditions of peak53

metamorphism varying from around the quartz–coesite transition (⇠2.4–2.8 GPa and54

⇠420–650 �C: St-Onge et al., 2013; Chatterjee and Jagoutz, 2015) to within the diamond55

stability field (⇠3.9–4.8 GPa and ⇠550–800 �C: Mukherjee et al., 2003; Wilke et al., 2015).56

This uncertainty hinders reliable large-scale tectonic reconstruction of the early stages of57

India–Asia collision.58

Thermomechanical modelling of heat transfer during plate-tectonic interactions allows59

calculation of theoretical P–T (–t) paths that a metamorphic rock may follow in a given60

geodynamic environment (Peacock, 1989). The P–T–t paths that these models produce61

may then be compared to the results of thermobarometric calculations performed on62

natural rocks. Forward modelling of slab-surface P–T conditions at subduction zones63

allows the influence of di↵erent geological variables on subduction zone thermal structures64

to be examined (Gerya et al., 2002), though the simulations are subject to significant65

uncertainties in the values of key parameters, including slab age, dip angle, and66

convergence velocity. The calculated evolutions of pressure and temperature with time67

provide tests of the geodynamic model within the uncertainty envelope.68

We have combined geodynamic and petrologic parameters specific to Eocene69

subduction of Indian-plate crust in the northwest Himalaya in simulations that provide70

new constraints on the prograde-to-peak P–T–t evolution of the Tso Morari massif.71

Petrological modelling was used to calculate bulk-rock properties for the main lithologies72

exposed in the massif at P–T conditions applicable to Phanerozoic subduction. These73

results provided inputs for a forward numerical model constrained by geodynamic criteria74

specific to India–Asia convergence, which predicted peak metamorphic conditions of75

⇠2.6–2.8 GPa and ⇠600–620 �C. Extremely high pressures up to 4.8 GPa inferred by some76

workers (Mukherjee et al., 2003; Wilke et al., 2015) appear to be geodynamically77

implausible. Further, diagnostic high-P metamorphic mineral assemblages predicted to78

form in felsic crust under equilibrium conditions are absent from the massif itself,79

suggesting widespread metastability during subduction. Our calculated P–T conditions are80

similar to those documented in the nearby UHP Kaghan Valley in Pakistan, which suggests81

3

that the northwestern Himalaya contains fragments of what may have been a coherent82

UHP terrane ⇠400 km long and ⇠150 km wide, comparable in size to the Western Gneiss83

Region, Norway, and the Dabie-Sulu belt, China.84

2 GEOLOGICAL BACKGROUND85

The Tso Morari massif, Ladakh Himalaya, is a structurally coherent block of thinned86

Indian continental margin (Mascle et al., 1994) exposed between Neotethyan sedimentary87

units of the Zanskar Range, and melange and ophiolite of the Indus Suture Zone (Fig. 1).88

The massif is ⇠100 km long, ⇠50 km wide, and ⇠7 km thick, and comprised primarily of89

quartzofeldspathic orthogneiss that hosts metre- to decametre-scale boudins of90

coesite-bearing mafic eclogite (De Sigoyer et al., 2004). Rare metasedimentary units also91

occur (Guillot et al., 1997). Field relations, major and trace element geochemistry, and92

U–Pb age data show that outcrops of Ordovician S-type granite exposed at Polokongka La,93

a low-strain portion of the massif (Fig. 1), represent relics of the orthogneiss’s protolith94

(Girard and Bussy, 1999). Eclogite boudins represent dismembered and metamorphosed95

mid-crustal doleritic dykes associated with Permian–Carboniferous Panjal Trap flood96

basalts exposed to the west (Fig. 1: Berthelsen, 1953; De Sigoyer et al., 2004).97

Geochronological and petrological work performed on both lithologies has identified98

multiple stages in the massif’s tectonothermal history. U–Pb zircon dating shows that the99

subduction of Indian continental crust initiated no later than c. 58–57 Ma (Guillot et al.,100

2003; Leech et al., 2005) and peak UHP conditions were reached at c. 51 Ma (St-Onge et101

al., 2013). These ages delimit a subduction duration of ⇠6–7 Myr, which correlates well102

with a ⇠7–9 Myr period inferred by Kaneko et al. (2003) for subduction of UHP eclogites103

in the nearby Kaghan Valley (Fig. 1). Initial and rapid exhumation to the base of the104

continental crust was followed by slower tectonic exhumation involving an105

amphibolite-facies metamorphic overprint at c. 45 Ma (U–Pb zircon, Leech et al., 2005;106

Th–Pb monazite, St-Onge et al., 2013). Ar–Ar data suggest exhumation through ⇠300–350107

�C at 30 Ma (De Sigoyer et al., 2000), although the timing of surface exposure is unknown.108

4

Despite the P–T conditions of retrograde amphibolite-facies overprinting in the crust109

being constrained at 1.0 ± 0.4 GPa and 650 ± 50 �C (e.g. De Sigoyer et al., 1997),110

estimates for peak pressure conditions fall into two distinct groups. Some workers suggest111

that subducted crust reached 80–100 km depth, inferred from P–T conditions of ⇠2.4–2.8112

GPa and ⇠420–650 �C around the quartz–coesite transition (St-Onge et al., 2013;113

Chatterjee and Jagoutz, 2015). However, others suggest subduction up to ⇠150 km depth,114

implied by putative P–T conditions of ⇠3.9–4.8 GPa and ⇠550–800 �C (Mukherjee et al.,115

2003; Wilke et al., 2015) well within the diamond stability field (Fig. 2). Notably, these116

discrepant P–T estimates were derived from rocks collected from the same roadside117

outcrop.118

Eclogite-facies metamorphism is recorded in mafic boudins by the assemblage garnet,119

omphacite, phengite, rutile, sodic-calcic/sodic amphibole, and quartz or coesite (De Sigoyer120

et al., 1997; Epard and Steck, 2008). Diamond has not been identified in the region. The121

host orthogneiss records negligible petrological evidence of UHP metamorphism, being122

characterised by the amphibolite-facies assemblage quartz, K-feldspar, plagioclase,123

muscovite, biotite, ilmenite, and garnet, with incipient partial melting (De Sigoyer et al.,124

2004; St-Onge et al., 2013). Nonetheless, Girard and Bussy (1999) and De Sigoyer et al.125

(2004) reported evidence for rare prograde reaction between magmatic plagioclase and126

biotite in the granite protolith to form Ca-rich garnet, kyanite, phengite, zoisite, and rutile.127

This assemblage occurs in some other high-P orthogneisses worldwide (e.g. in the Gran128

Paradiso and Monte Rosa Alpine terranes: Dal Piaz and Lombardo, 1986), and129

demonstrates that the Polokongka La granite protolith experienced metamorphic P–T130

conditions identical to the metabasites. Retrograde, mid-crustal overprinting of eclogite is131

recorded by a transition from garnet- and omphacite-dominated assemblages in boudin132

cores to an amphibolite-facies rim assemblage of quartz, plagioclase, calcic amphibole,133

epidote/clinozoisite, sphene, and biotite (De Sigoyer et al., 2000; Palin et al., 2014).134

5

3 INTEGRATED FORWARD MODELLING135

New constraints on the evolution of the Tso Morari massif have been obtained here via a136

three-stage forward-modelling approach that closely integrates geodynamic and137

petrological information. Firstly, phase diagram modelling was used to calculate the fluid138

contents and bulk-rock densities of stable assemblages that might have formed in139

metamorphosed felsic and mafic massif lithologies during subduction. Bulk-rock140

compositions for protoliths of the massif’s orthogneiss (Polokongka La granite) and eclogite141

(Panjal Trap basalt) were used instead of those for the metamorphic products themselves,142

owing to (1) uncertainty surrounding the relative timing of mineralogical transformations143

within the subduction–exhumation cycle, and (2) to avoid the e↵ects of open-system144

amphibolite-facies retrograde metamorphic overprinting on peak UHP assemblages. This145

approach provides an alternative framework with which to predict the petrological changes146

that occur during subduction of crustal materials. These bulk-rock petrophysical estimates147

were incorporated into a two-dimensional thermomechanical model specific to Eocene148

subduction in the northwest Himalaya, which generated P–T–t paths tailored for the149

region. The phase diagrams calculated previously were then examined within the context of150

these calculated P–T paths in order to determine the changes in mineral assemblage that151

might have occurred during subduction and metamorphism of Tso Morari massif crust.152

Subduction-zone thermal structures are primarily controlled by plate convergence rate,153

and the age and dip angle of the descending slab (Kirby et al., 1991). Changes in the154

metamorphic mineral assemblages of constituent rocks determine subducted crust density.155

Most studies of UHP terranes worldwide focus on mafic components due to their retaining156

detailed P–T (–t) information about the subduction–exhumation cycle; however, their low157

volume proportion (⇠1–2%) in upper continental crust means that they make no158

significant contribution to bulk-terrane petrophysical properties (Peterman et al., 2009).159

While experiments show that felsic crust can transform to (U)HP minerals such as jadeite,160

kyanite, and/or coesite, natural examples are rare, implying that most such subducted161

material (1) does not transform, (2) transforms and reverts to low-P assemblages during162

exhumation or later metamorphic overprinting, or (3) transforms at depth but is not163

6

exhumed. Many studies of collisional terranes worldwide support the first option (Carswell164

et al., 1986), although high-P felsic granulite that represents overprinted eclogite has been165

documented from a number of regional metamorphic terranes worldwide (O’Brien et al.,166

2003). The limited ability of some crustal rocks to re-equilibrate during subduction and/or167

exhumation is attributed to the lack of a critical catalyst such as fluid or deformation (e.g.168

Rubie, 1986). As such, phase transformations cannot be assumed to occur a priori, even169

though geodynamic models of collisional orogenesis often consider this to be so (Gerya and170

Stockhert, 2006; Warren et al., 2008). Our thermomechanical simulations thus consider two171

end-member scenarios: non-transformation (metastability) and complete transformation172

(equilibrium) of Tso Morari massif lithologies.173

3.1 Metamorphic phase equilibrium modelling174

All phase diagrams were constructed using thermocalc v3.40i (Powell et al., 1988) and175

the internally consistent thermodynamic dataset ds55 (Holland and Powell, 1998; updated176

August 2004) in the Na2

O–CaO–K2

O–FeO–MgO–Al2

O3

–SiO2

–H2

O–TiO2

–O2

177

(NCKFMASHTO) compositional system. The activity–composition relations used for178

solid-solution phases are listed in the Supplementary Information. Bulk-rock compositions179

used for modelling were converted from weight-percent oxide values reported in each180

original study (Table S1) to molar proportions of oxides (Table 1), with the water content181

of each protolith derived from reported XRF loss-on-ignition contents. Fractionation of the182

Panjal Trap basalt/eclogite bulk composition owing to sequestration of cations into garnet183

porphyroblasts during prograde metamorphism was also implemented using the rbi code in184

thermocalc (see Supplementary Information). Parameter maps showing calculated185

bulk-rock densities and free H2

O contents for both protoliths (assuming complete186

metamorphic transformation) at 300–800 �C and 0.2–5.0 GPa are shown in Fig. 3. Full187

phase diagrams showing the equilibrium assemblages for each rock type are shown in Figs188

S1–4. Uncertainties on the position of calculated assemblage-field boundaries in P–T space189

are up to ±1 kbar and ±50 �C at 2� (Powell and Holland 2008; Palin et al., 2016).190

In order to e↵ectively integrate our petrological modelling results into the geodynamic191

7

simulations described below, we assumed that metamorphism of the massif as a whole192

occurred in a closed-system environment. Although fluid influx from rocks external to the193

massif crust may have occurred during subduction, the absence of constraints on the194

timing and/or amount of fluid infiltration precluded its consideration. Nonetheless, we195

qualitatively examine the e↵ects of open-system processes by examining how fluid released196

from each protolith during devolatilisation could interact with the other. This massif-wide197

closed-system scenario represents the simplest interpretation of its evolution, and so defines198

one end-member of a possible hydrodynamic spectrum that can be explored in greater199

detail if and when constraints on fluid migration patterns become available.200

3.2 Geodynamic numerical modelling201

The numerical code MVEP2 employed for geodynamic numerical modelling is described in202

detail in Kaus (2010) and Thielmann and Kaus (2012), and can be downloaded from203

https://bitbucket.org/bkaus/mvep2. A brief summary is given here, alongside a description204

of the model parameters. However, as P–T–t paths calculated via such modelling are205

complex functions of the input parameters, sensitivity testing allowed an assessment of the206

reliability of our main results according to geologically realistic variation in these values.207

3.2.1 Governing equations and numerical approach208

The incompressible Stokes equations are defined as follows:209

@vi

@xi= 0 (1)

�@P

@xi+

@⌧ij

@xj= ⇢gi (2)

⇢cp

✓@T

@t

+ vi@T

@xi

◆=

@

@xi

✓k

@T

@xi

◆+ ⌧ij ✏ij � ↵T⇢gvz (3)

where vi is velocity, P is pressure, ⌧ ij is deviatoric stress, ⇢ is density, gi is gravitational210

8

acceleration, cp is heat capacity, T is temperature, k is thermal conductivity, ✏ij is strain211

rate, and ↵ is thermal expansivity. Both shear heating and adiabatic (de)compression were212

incorporated into the energy equation. Density of the upper crust in models considering213

mineral transformation varied as a function of P and T according to the values calculated214

via petrological modelling (Fig. 3).215

A viscoplastic constitutive relationship was used:216

⌧ij = 2⌘e↵ ✏ij (4)

⌘e↵ = min (⌘disl , ⌘pl) (5)

where the e↵ective (e↵ ) viscosity is the minimum of the dislocation (disl) and plastic (pl)217

viscosities, which are defined as:218

⌘pl =�yield

2✏II(6)

⌘disl = A

� 1n✏II

(1�n)nexp

✓E + PV

nRT

◆(7)

Here, �yield [= P ·sin(�) + c·cos(�)] is the yield stress as a function of the friction219

coe�cient, pressure, and cohesion, which followed a Drucker–Prager relationship (Drucker220

and Prager, 1952), ✏II [= (0.5·✏ij2)0.5] represents the second invariant of the strain rate221

tensor, A is a pre-exponential material parameter, n the power law coe�cient, E the222

activation energy, and V the activation volume. These equations were solved with the223

MATLAB-based, two-dimensional, thermo-mechanical finite-element code MVEP2 using224

Q1

P0

elements. Additionally, we applied a marker-and-cell approach to advect markers225

with the material properties above a Lagrangian mesh. Remeshing was applied every226

timestep to prevent the elements from becoming too distorted.227

9

3.2.2 Model setup and extraction of results228

Simulations considered a 4000-km wide and 660-km deep domain that used 801 ⇥ 201229

elements along the horizontal and vertical axes, respectively (Fig. 4). This produced a230

resolution in the subduction zone area of approximately 1 km ⇥ 1.5 km. Boundary231

conditions comprised a free surface (no stress) at the top, and free slip at the bottom and232

sides. Compression of the crust used a pushing box at the right-hand side of the domain233

with rates defined by various plate velocity scenarios (see below). Thermal boundaries were234

isothermal at the top and the bottom, and flux-free at the sides. In addition to shear and235

adiabatic heating, a geotherm of 9 �C/km was used for the crust and lithospheric mantle,236

and 0.5 �C/km was used for the aesthenospheric mantle. This produced a maximum237

temperature of 1680 �C at 660 km and ensured that the lithosphere was always colder than238

1350 �C.239

A weak zone was incorporated into the lithosphere in order to initiate subduction,240

which had a lower friction coe�cient and viscosity than its surroundings. We applied a241

constant sedimentation rate of 1 mm/yr, and a constant erosion rate of 5 mm/yr if the242

topography exceeded 2 km in elevation. After each simulation, we implemented a 4th-order243

Runge–Kutta scheme (Press, 1992) to advect markers backwards in time and extract P–T244

information. This provided the opportunity to examine the total range and variability of245

any model parameter over any spatial area.246

3.2.3 Density247

Density changes were investigated using two end-member scenarios: full transformation248

(i.e. equilibrium metamorphism) and non-transformation (i.e. metastability) of the Tso249

Morari massif upper crust. For the former, bulk-rock densities for assemblages in each250

metamorphosed protolith were calculated between 0.2–5.0 GPa and 300–800 �C (Fig. 3);251

however, as field studies estimate that mafic eclogite represents no more than ⇠1–2% total252

volume of the Tso Morari massif, the terrane density at any P–T condition was considered253

as that of the metamorphosed Polokongka La granite protolith only. For254

10

non-transformation, the massif was assigned a fixed density of 2700 kg/m3.255

3.2.4 Geometry of the subducted slab256

Numerous possible slab geometries for subduction of Indian crust in the Ladakh Himalaya257

were outlined by Leech et al. (2005). We used geometric parameters from their preferred258

configuration (“Model 1”), which had a geologically realistic dip angle that increased with259

depth and a curved slab surface with a bending radius of 350 km. Under these conditions,260

the slab dips at ⇠6� at the trench and ⇠40� at ⇠100 km depth (the approximate HP–UHP261

transition: Fig. 2). Their model assumed an initial mid-crustal intrusive depth of 15 km262

for the massif’s mafic dyke and granite protoliths, which also we adopted here.263

3.2.5 Subduction velocity and duration264

Two subduction velocity profiles were considered in our geodynamic modelling: (1) a265

constant 7 cm/yr velocity, as proposed by Leech et al. (2005) in their preferred Model 1266

setup, and (2) a time-dependent velocity profile simulating the reported deceleration of the267

Indian indenter immediately following initial collision with Asia. This considered a 10268

cm/yr velocity for the first 2 Myr following the onset of crustal subduction and 4 cm/yr269

afterwards (modified from Guillot et al., 2003). These two profiles were initially combined270

with both mineralogical transformation scenarios described above to produce four271

end-member reference models (M1–4), as detailed in Table 2. All combinations used the272

above-described geometric setup and produced very similar prograde-to-peak P–T–t paths273

(Fig. S5); however, as a time-dependent deceleration velocity profile is more geologically274

realistic, we focus our discussion below on the results of these simulations (M3–M4).275

3.2.6 Sensitivity testing276

The sensitivity of our P–T–t results to the model setup conditions was investigated by277

re-running simulations M1–M4 whilst individually varying each of 12 key model278

parameters. These variables comprised: bulk-rock composition, convergence rate, upper279

11

and lower crustal flow laws, weak zone angle, sedimentation rate, erosion rate, friction280

angle of the upper and lower crust, heat production, heat capacity, and281

temperature-dependent conductivity (Table S4). The influence of bulk composition was282

investigated by recalculating densities for a fully-hydrated Polokongka La granite protolith.283

These were similar to those produced at low pressure (<15 kbar) using the actual284

XRF-derived water content, but lower at higher pressures owing to the stabilisation of285

white mica in place of K-feldspar and jadeite.286

3.3 Results287

3.3.1 Main simulations288

Figure 2 shows P–T–t paths obtained from simulations M3 and M4 (deceleration with or289

without phase transformations). Upper crustal elements tracking the leading edge of the290

Indian plate margin initiated at ⇠0.4–0.5 GPa and ⇠160–190 �C at t = 0 Myr,291

representative of an initial depth in the crust of ⇠10–15 km.292

The complete transformation of massif components to high-P assemblages during293

subduction (M4) predicts ⇠1.2 GPa and ⇠240 �C to be reached after ⇠1 Myr of294

subduction, with a significant P–T increase up to ⇠2.4 GPa and ⇠550 �C in the next 2295

Myr (Fig. 2). Continued subduction until t = 7 Myr—the upper limit of subduction296

duration—produced only a small increase in P–T conditions to ⇠2.9 GPa and ⇠620 �C,297

with the tracked elements representing the leading edge of the Indian plate terminating298

just above the quartz–coesite transition (Fig. 2).299

The modelled P–T path for non-transformation (M3) was very similar to that for300

transformation, though subduction was slightly slower (Fig. 2). Near-UHP conditions of301

⇠2.5 GPa and ⇠425–575 �C were reached after ⇠4 Myr of simulation time, with similar302

peak P–T conditions of ⇠2.9 GPa and ⇠600 �C reached at t = 5.5 Myr (Fig. 2).303

Continuation of this model until t = 7 Myr did not significantly increase this peak P–T304

condition, as the low density of the crust causes it to stall and thicken at ⇠100 km.305

Notably, both modelled paths lie within the range of modern-day subduction-zone slab-top306

12

P–T profiles reported by Syracuse et al. (2010).307

The equilibrium mineral assemblages in both protoliths along the full transformation308

(M4) P–T path are shown in Fig. 5. Low-grade (<1.7 GPa, <350 �C) Panjal Trap309

metabasalt assemblages are dominated by glaucophane, chlorite, and epidote, with lesser310

actinolite, omphacite, muscovite, lawsonite, and sphene (Fig. 5a). Both garnet and rutile311

stabilise at higher grade at the expense of lawsonite, actinolite, and glaucophane. No free312

H2

O occurs at low/medium-grade conditions, as any produced via the breakdown of313

chlorite and epidote would be incorporated into other newly-formed hydrous phases.314

Excess fluid is not generated until relatively late in the subduction history at ⇠2.3 GPa315

and ⇠520 �C (Figs 3 and 5a). The UHP eclogite-facies assemblage stable at peak P–T316

conditions (⇠2.8 GPa and ⇠620 �C) is dominated by garnet (⇠30%) and omphacite317

(⇠40%), with talc, coesite, muscovite, lawsonite, rutile, and H2

O comprising 1–7% each318

(Fig. 5a), which closely matches assemblages in relatively fresh mafic eclogite from the319

massif (Epard and Steck, 2008). Putative continued subduction to ⇠4.8 GPa (dashed320

arrow on Figs 2 and 3), as suggested by Mukherjee et al. (2003) and Wilke et al. (2015),321

would involve the loss of talc and lawsonite, a minor increase in the proportions of garnet,322

omphacite, and coesite, and continued devolatilisation (Fig. 5a).323

The Polokongka La granite exhibits a notably simpler evolution during subduction.324

Under equilibrium conditions, the protolith assemblage K-feldspar, quartz, plagioclase,325

biotite, and muscovite transforms to jadeite/omphacite (⇠18%), muscovite (⇠13%),326

K-feldspar (⇠23%), quartz (⇠45%), garnet (⇠1%), rutile (<1%), and kyanite (<1%) at327

relatively low-grade conditions (<350 �C), with virtually no further changes predicted up328

to UHP conditions except for the replacement of quartz by coesite (Fig. 5b). The low H2

O329

content of the protolith (0.57 wt%: Table 1) keeps the rock fluid-undersaturated for the330

entire prograde P–T path, with water-saturated conditions restricted to low-P331

amphibolite-facies conditions (Fig. 3). Thus, prograde metamorphism would have332

proceeded under fluid-absent conditions where reaction kinetics are sluggish (Rubie, 1986).333

13

3.3.2 Additional simulations334

Sensitivity testing of models M1–M4 was achieved by running thirty-eight additional335

simulations that allowed the main e↵ect of each parameter on the style of subduction to be336

individually assessed. These results are shown in Figure 6 alongside our four reference337

models (bold red arrows). The evolution of each sensitivity test as a function of simulation338

time is shown in Fig. S6 and each test’s parameter combination is listed in Table S5.339

A lower density calculated for a fully hydrated Polokongka La granite protolith340

produced an increased buoyancy force that resulted in lower overall peak pressure341

conditions than M1–M4, as shown in Fig. 6 by the calculated mean P–T conditions of the342

entire subducted upper crust (“Hydrated crust” path). A greater plate velocity for initial343

collision (15 cm/yr; cf. Jagoutz et al., 2015) resulted in faster and steeper subduction, and344

thus calculated conditions at any equivalent simulation time relative to M1–M4 were at345

higher pressures (⇠0.5–0.8 GPa greater), but similar temperatures (Fig. 6: “High plate346

velocity” path). Nevertheless, since the upper crust is typically not subducted deeper than347

the lithosphere–asthenosphere boundary (Marquardt and Miyagi, 2015), the mean P–T348

conditions for upper crustal units was similar to M1–M4, albeit reached 1–2 Myr earlier349

due to more rapid convergence (Fig. S6). An increased weak zone angle up to 40� at the350

trench produced exceptionally steep and cold subduction through the HP–UHP transition,351

with subducted crust consistently entering into the Forbidden Zone (Fig. 6: “High weak352

zone angle” path).353

Varying sedimentation rate had no e↵ect on the subduction style, though high erosion354

rates led to a slightly shallower subduction angle. A higher friction angle for the upper355

crust produced smaller peak pressures, while a higher friction angle for the lower crust356

produced similar results to M1–M4 (Table S4). Heat production did not influence the357

subduction style in our simulations, although a higher heat capacity caused slightly358

shallower subduction and thus lower peak pressures than our main models (Fig. 6).359

Finally, implementing temperature-dependent conductivity with a non-linear crustal360

geotherm (Table S5 and Fig. S8) had the main e↵ect of increasing the mean temperature361

of subducted upper crust by approximately ⇠80–100 �C (Table S4), although calculated362

14

pressures at any given simulation time were very similar to M1–M4, and the upper crust363

mostly failed to reach pressures above ⇠3 GPa (Fig. 6: “Temperature-dependent364

conductivity” path).365

4 DISCUSSION366

4.1 Reliability of calculated P–T–t paths367

We designed an integrated geodynamic and petrological model to constrain the368

prograde-to-peak P–T–t evolution of Tso Morari massif crust during the initial stages of369

India–Asia collision. All four of our main models (M1–M4: Table 2) exhibited similar370

P–T–t paths to one another, as did all but ten of our additional 38 simulations. Key371

deviations occurred when the initial angle of the weak zone was increased, leading to372

tracked continental crustal nodes consistently entering the Forbidden Zone (Fig. 6), which373

is unexpected in modern-day subduction (Liou et al., 2000). Such simulations produced374

maximum pressures >5 GPa for both upper and lower crustal components: however, these375

high weak zone angles are unrealistic for the uppermost portions of modern-day subduction376

systems (Syracuse et al., 2010), and the most deeply subducted continental crust in our377

simulations were small fragments dragged into the mantle during slab breako↵, which378

would never be exhumed. The P–T paths of reasonably exhumable felsic379

rocks—represented here by those that stagnated at the buoyancy-driven threshold for the380

subduction of the upper crust at around 100 km depth (Agard et al., 2009)—exhibited381

peak temperatures of 600–700 �C in all simulations (Fig. 6). Generally shallower382

subduction resulted from applying a ‘mafic granulite’ flow law to the upper and/or lower383

crust (Fig. 6, Table S4), although this is not a representative rheology for the felsic384

lithologies that comprise the bulk of the Tso Morari terrane.385

Considered together, these results indicate that the general shape of the P–T paths386

calculated in simulations M3–M4 (Fig. 2) is robust within the range of reasonable natural387

variation in model parameters. Modifications to our preferred Tso Morari-specific388

geometric and geological constraints (e.g. slab dip angle, convergence velocity) did cause389

15

large variations in calculated P–T–t conditions; however, these alternative paths mostly390

failed to intersect or terminate (reasonably close to) any of the calculated P–T conditions391

proposed for the evolution of the massif, in contrast to the results of our main simulations392

(Fig. 2). Furthermore, these extreme variants are unsupported by independent datasets393

(e.g. U–Pb geochronology), which must provide absolute constraints on our region-specific394

investigation.395

4.2 Implications for the Tso Morari massif396

4.2.1 Orogen-parallel correlation in the northwestern Himalaya397

The geological evolution of the Tso Morari massif is strongly debated, owing to wide398

variation in interpreted depths of subduction and thermal gradients of metamorphism.399

Such geodynamic criteria are important for formulating regional-scale numerical models of400

India–Asia collision, which are commonly benchmarked against P–T–t data reported for401

these Himalayan UHP eclogites (e.g. Warren et al., 2008). In our main models (M3–M4),402

peak eclogite-facies P–T conditions of ⇠2.6–2.8 GPa and ⇠620 �C are reached ⇠3 Myr403

after subduction of continental crust initiated (Fig. 2). This correlates well with ⇠2.4–3.2404

GPa and ⇠650–750 �C reported for eclogites in the UHP Kaghan (Kaneko et al., 2003) and405

HP Stak (Lanari et al., 2013) valleys located 400 km northwest along strike (Fig. 1, inset).406

If pressures up to 4.8 GPa (cf. Mukherjee et al., 2003; Wilke et al., 2015) for peak407

metamorphism at Tso Morari are true, the upper crusts in each locality must have behaved408

independently during subduction. A strong gradient in plate-convergence velocity would409

thus be required between both localities over a relatively small length scale, which we have410

shown produces a significantly di↵erent P–T path (Fig. 6). This seems unreasonable due411

to their current proximity, equivalent orogen-parallel tectonic setting, and similarities in the412

ages and durations of subduction and exhumation. As such, we suggest that both regions413

of the northwest Himalaya represent now-dismembered parts of a single, coherent crustal414

terrane that underwent deep subduction to UHP conditions of ⇠ 2.8 GPa, comparable in415

size to the Western Gneiss Region, Norway, and the Dabie-Sulu terrane, China.416

16

How then can these extreme pressure estimates be reconciled, given that all samples417

studied from the massif must have experienced the same tectonothermal evolution? We418

envisage two possibilities: tectonic overpressure or mineralogical disequilibrium. For the419

former, recent studies have shown that localised, non-lithostatic overpressure may be420

prevalent in some convergent tectonic settings (Petrini and Podladchikov, 2002; Li et al.,421

2010), and small-scale and rigid petrological elements may act as foci for its accumulation422

(Reuber et al., 2016). Importantly, our simulations considered a laterally homogeneous and423

mechanically weak crust, and the maximum modelled tectonic overpressure at the424

resolution of our model never exceeded 0.5 GPa (Fig. 7a–b), so cannot account for the425

discrepant pressure estimates. Although, smaller-scale heterogeneities may result in426

significantly higher localised tectonic pressures (Reuber et al., 2016), such heterogeneities427

in the Tso Morari massif would likely have been on the scale of individual boudins (i.e.428

tens of meters). Considering such features in a lithospheric-scale simulations is429

computationally challenging and was unable to be tested at this point in time, although430

indicates an avenue for future study.431

The latter possibility is mineralogical disequilibrium. We note that both studies432

claiming anomalously high peak pressures for the region (Mukherjee et al., 2003; Wilke et433

al., 2015) applied conventional thermobarometry to minerals that were not conclusively434

proven to have ever been in chemical equilibrium with one another, thus rendering the P–T435

results of questionable veracity. Indeed, final equilibration of Tso Morari crust at 80–100436

km depth, as suggested in this study, produces lithostatic pressures straddling the437

HP–UHP transition (Fig. 2). This result supports numerous petrographic observations438

that some eclogite boudins in the region lack coesite in the peak assemblage, containing439

monocrystalline quartz inclusions in garnet outer rims instead (De Sigoyer et al., 1997;440

St-Onge et al., 2013).441

4.2.2 Metastability during the subduction of felsic crust442

Our petrological modelling suggests that the Polokongka La granite protolith did not443

pervasively transform during subduction, as indicated by the absence (or evidence for the444

17

former existence) of diagnostic high-P phases preserved in the orthogneiss today. This445

non-transformation is attributed to a lack of free fluid, low temperatures associated with446

the early stages of subduction (<400 �C), and the short timescale involved for447

metamorphism (<7 Myr): all of which promote sluggish reaction kinetics (Rubie, 1986).448

By contrast, the Panjal Trap basalt must have pervasively transformed at UHP449

conditions, as shown by the occurrence of coesite in some eclogite boudins. Excess H2

O450

would be liberated from the basalt relatively late in the subduction history at ⇠2.3 GPa451

and ⇠520 �C (Figs 3 and 5a) following the breakdown of chlorite and amphibole. This fluid452

would then have locally infiltrated the enclosing granite and promoted minor mineralogical453

change in the host. As such, fluid-catalysed transformation in the granite may have only454

occurred in thin rinds along boudin–host contacts due to the volumetrically negligible455

proportion of metabasite (<1%) in the terrane as a whole. Any prograde or retrograde456

reactions occurring in the bulk of the granite would have been controlled by its own457

(sluggish) rate of internal dehydration. Jadeitic clinopyroxene, which should have458

comprised ⇠18% of the orthogneiss UHP assemblage under equilibrium conditions (Fig.459

5b), has never been reported from Tso Morari, nor have any indication of its former460

existence, such as relic inclusions or amphibole-bearing symplectites (Epard and Steck,461

2008). Nonetheless, evidence of any fluid-absent transformation that may have occurred462

during subduction would likely have been lost during retrograde low-P amphibolite-facies463

metamorphic overprinting and recrystallisation that occurred under H2

O-saturated and/or464

suprasolidus conditions (Figs 2, 3, and S2: Palin et al., 2014b).465

Similar large-scale metastability of subducted continental crust has been inferred by466

numerous studies of UHP terranes worldwide, including the Western Gneiss Region,467

Norway (Peterman et al., 2009; Young and Kylander-Clark, 2015). The common, rather468

than exceptional, non-transformation of felsic crust during subduction has important469

implications for plate tectonic processes and the geodynamic simulations describing them.470

For example, models of crustal strength based on feldspar-absent assemblages at high-P471

conditions (Stockhert and Renner, 1998), and those invoking fluid-fluxed melting472

(Labrousse et al., 2011), retrogression, or hydrolytic/transformational weakening (Warren,473

18

2013) to instigate exhumation of UHP terranes by a↵ecting the mechanical behaviour and474

rheology of high-P mineral assemblages may not be applicable if vast swathes of475

continental material remain metastably dry, solid, and undeformed at eclogite-facies476

depths. Our petrological modelling suggests that the majority of the Tso Morari terrane477

would have resisted transformation at UHP conditions owing to its fluid-undersaturated478

state, and so is unlikely to have pervasively weakened or strengthened either. Field479

observations in the massif (and from other large-scale continental terranes) provide480

first-hand evidence that internal deformation of UHP sheets is uncommon, with shear481

largely confined to the edges of internally low-strained blocks (e.g. De Sigoyer et al., 2004;482

Epard and Steck, 2008). As such, metamorphic reactions are not simply passive recorders483

of plate tectonic processes, but play an active role in orogenesis; a point that geodynamic484

models involving subduction should consider. Close integration with phase equilibrium485

forward modelling of suitable rock types o↵ers a robust way to achieve this.486

5 CONCLUSIONS487

Integrated petrological and geodynamic forward models of the tectonometamorphic488

evolution of the Himalayan UHP Tso Morari massif predicts peak P–T conditions of489

⇠2.6–2.8 GPa and ⇠600–620 �C. These conditions are consistent with observed490

metamorphic assemblages in mafic eclogite, and were simulated to have been achieved491

within timescales consistent with the reported duration of subduction. Significantly higher492

pressures suggested for peak metamorphism (up to ⇠4.8 GPa) are interpreted to be493

spurious and likely resulted from thermobarometry performed on minerals that were not in494

chemical equilibrium. Our models with homogeneous crustal properties suggest that495

large-scale tectonic overpressures are of insu�cient magnitude to account for this496

discrepancy. Small, boudin-scale variations in mechanical strength—not resolved in our497

simulations—may cause more significant overpressures (Reuber et al., 2016). Yet, the498

proximity, equivalent orogen-parallel tectonic setting, and similar P–T conditions and499

timing of peak metamorphism between Tso Morari and the Kaghan Valley terrane suggest500

that they acted as a coherent, ⇠400-km-long crustal domain that was subducted to UHP501

19

conditions during the earliest stages of India–Asia collision.502

Our petrological modelling shows that degree of fluid saturation of each massif503

protolith imparts a strong control on their transformation potential during subduction.504

The relatively high H2

O content of Panjal Trap basalt—the protolith for the mafic505

eclogite—allowed generation of free fluid and complete mineralogical transformation at506

eclogite-facies conditions. The relatively low H2

O content of Polokongka La granite—the507

protolith of the orthogneiss that now hosts mafic eclogite boudins—would have inhibited508

transformation. It therefore likely persisted as a metastable granite during much of the509

subduction–exhumation cycle. Widespread metamorphic change must have occurred510

during retrograde residence in the crust, as supported by its characteristic511

amphibolite-facies assemblages and lack of evidence for the former existence of diagnostic512

high-pressure phases predicted by petrological modelling (e.g. coesite, jadeitic513

clinopyroxene). The volumetric dominance of dry, felsic lithologies in continental (U)HP514

terranes worldwide indicates that the large-scale metastability of subducted crust may be515

common during collisional orogenesis. Geodynamic simulations of tectonic processes516

predicated on the operation of equilibrium metamorphism should consider the e↵ects of517

evolving fluid budgets in controlling reaction catalysis and the variable transformation of518

rocks at depth in the Earth.519

6 ACKNOWLEDGEMENTS520

Reviews by Stephan Guillot, Julia De Sigoyer, Mary Leech, Clare Warren, and an521

anonymous reviewer are gratefully acknowledged. Michael Bickle is thanked for editorial522

handling. This research did not receive any specific grant from funding agencies in the523

public, commercial, or not-for-profit sectors.524

20

References525

Agard, P., Yamato, P., Jolivet, L., Burov, E., 2009. Exhumation of oceanic blueschists and526

eclogites in subduction zones: Timing and mechanisms. Earth-Science Reviews 92,527

53–79.528

Berthelsen, A., 1953. On the geology of the Rupshu district, NW Himalaya. Meddelelser529

fra Dansk Geologisk Forening 12, 350–415.530

Carswell, D. A., Cuthbert, S. J., 1986. Eclogite-facies metamorphism in the lower531

continental crust. In: Dawson, J. B. et al. (Ed.), The Nature of the Lower Continental532

Crust. Vol. 24. Blackwell Scientific Publications, Oxford, UK, pp. 193–209.533

Chatterjee, N., Jagoutz, O., 2015. Exhumation of the UHP Tso Morari eclogite as a diapir534

rising through the mantle wedge. Contributions to Mineralogy and Petrology 169, 1–20.535

Dal Piaz, G. V., Lombardo, B., 1986. Early-alpine eclogite metamorphism in the Penninic536

Monte Rosa-Gran Paradiso basement nappes of the north-western Alps. GSA Memoirs537

164, 249–266.538

De Sigoyer, J., Chavagnac, V., Blichert-Toft, J., Villa, I. M., Luais, B., Guillot, S., Cosca,539

M., 2000. Dating the Indian continental subduction and collisional thickening in the540

north-west Himalaya: Multichronology of the Tso Morari eclogites. Geology 28, 487–490.541

De Sigoyer, J., Guillot, S., Dick, P., 2004. Exhumation of the ultrahigh-pressure Tso542

Morari unit in eastern Ladakh (NW Himalaya): A case study. Tectonics 23, TC3003.543

De Sigoyer, J., Guillot, S., Lardeaux, J. M., Mascle, G., 1997. Glaucophane-bearing544

eclogites in the Tso Morari dome (eastern Ladakh, NW Himalaya). European Journal of545

Mineralogy 9, 1073–1083.546

Drucker, D. C., Prager, W., 1952. Soil mechanics and plastic analysis for limit design.547

Quarterly of Applied Mathematics 10, 157–165.548

Epard, J. L., Steck, A., 2008. Structural development of the Tso Morari ultra-high pressure549

nappe of the Ladakh Himalaya. Tectonophysics 451, 242–264.550

21

Gerya, T., Stockhert, B., 2006. Two-dimensional numerical modeling of tectonic and551

metamorphic histories at active continental margins. International Journal of Earth552

Sciences 95, 250–274.553

Gerya, T., Stockhert, B., Perchuk, A. L., 2002. Exhumation of high-pressure metamorphic554

rocks in a subduction channel: a numerical simulation. Tectonics 21, 1056.555

Girard, M., Bussy, F., 1999. Late Pan-African magmatism in the Himalaya: new556

geochronological and geochemical data from the Ordovician Tso Morari metagranites557

(Ladakh, NW India). Schweizerische Mineralogische und Petrographische Mitteilungen558

79, 399–417.559

Guillot, S., De Sigoyer, J., Lardeaux, J. M., Mascle, G., 1997. Eclogitic metasediments from560

the Tso Morari area (Ladakh, Himalaya): evidence for continental subduction during561

India–Asia convergence. Contributions to Mineralogy and Petrology 128, 197–212.562

Guillot, S., Garzanti, E., Baratoux, D., Marquer, D., Maheo, G., De Sigoyer, J., 2003.563

Reconstructing the total shortening history of the NW Himalaya. Geochemistry,564

Geophysics, Geosystems 4, 1064.565

Holland, T. J. B., Powell, R., 1998. An internally consistent thermodynamic dataset for566

phases of petrological interest. Journal of Metamorphic Geology 16, 309–343.567

Jagoutz, O., Royden, L., Holt, A. F., Becker, T. W., 2015. Anomalously fast convergence568

of India and Eurasia caused by double subduction. Nature Geoscience 8, 475–478.569

Kaneko, Y., Katayama, I., Yamamoto, H., Misawa, K., Ishikawa, M., Rehman, H. U.,570

Kausar, A. B., Shiraishi, K., 2003. Timing of Himalayan ultrahigh-pressure571

metamorphism: sinking rate and subduction angle of the Indian continental crust572

beneath Asia. Journal of Metamorphic Geology 21, 589–599.573

Kaus, B. J. P., 2010. Factors that control the angle of shear bands in geodynamic574

numerical models of brittle deformation. Tectonophysics 484, 36–47.575

Kirby, S. H., Durham, W. B., Stern, L. A., 1991. Mantle phase changes and576

deep-earthquake faulting in subducting lithosphere. Science 252, 216–225.577

22

Labrousse, L., Prouteau, G., Ganzhorn, A. C., 2011. Continental exhumation triggered by578

partial melting at ultrahigh pressure. Geology 39, 1171–1174.579

Lanari, P. L., Riel, N., Guillot, S., Vidal, O., Schwartz, S., Pecher, A., Hattori, K. H., 2013.580

Deciphering high-pressure metamorphism in collisional context using microprobe581

mapping methods: Application to the Stak eclogitic massif (northwest Himalaya).582

Geology 41, 111–114.583

Leech, M., Singh, S., Jain, A. K., Klemperer, S. L., Manickavasagam, R. M., 2005. The584

onset of India–Asia continental collision: Early, steep subduction required by the timing585

of UHP metamorphism in the western Himalaya. Earth and Planetary Science Letters586

234, 83–97.587

Li, Z. H., Gerya, T. V., and Burg, J. P., 2010. Influence of tectonic overpressure on P–T588

paths of HP–UHP rocks in continental collision zones: thermomechanical modelling.589

Journal of Metamorphic Geology 28, 227–247.590

Liou, J. G., Hacker, B. R., Zhang, R. Y., 2000. Into the Forbidden Zone. Science 287,591

1215–1216.592

Marquardt, H., Miyagi, L., 2015. Slab stagnation in the shallow lower mantle linked to an593

increase in mantle viscosity. Nature Geoscience 8, 311–314.594

Maruyama, S., Liou, J. G., Terabayashi, M., 1996. Blueschists and eclogites of the world595

and their exhumation. International Geology Review 38, 485–594.596

Mascle, G., Colchen, M., Guillot, S., Delaygue, G., 1994. Preliminary results on the597

metamorphic evolution of the Tso Morari dome: Consequences for the geodynamic598

evolution of the Indian margin during Himalayan collision. Nepal Geological Society599

Journal 10, 85–86.600

Mukherjee, B. K., Sachan, H. K., Ogasawara, Y., Muko, A., Yoshioka, N., 2003.601

Carbonate-bearing UHPM rocks from the Tso Morari region, Ladakh, India: petrological602

implications. International Geology Review 45, 49–69.603

23

O’Brien, P. J., Rotzler, J., 2003. High-pressure granulites: formation, recovery of peak604

conditions and implications for tectonics. Journal of Metamorphic Geology 21, 3–20.605

Palin, R. M., St-Onge, M. R., Waters, D. J., Searle, M. P., Dyck, B., 2014. Phase equilibria606

modelling of retrograde amphibole and clinozoisite in mafic eclogite from the Tso Morari607

massif, northwest India: constraining the P–T–M(H2

O) conditions of exhumation.608

Journal of Metamorphic Geology 32, 675–693.609

Palin, R. M., Weller, O. M., Waters, D. J., Dyck, B., 2016. Quantifying geological610

uncertainty in metamorphic phase equilibria modelling; a Monte Carlo assessment and611

implications for tectonic interpretations. Geoscience Frontiers 7, 591–607.612

Peacock, S. M., 1989. Thermal modeling of metamorphic pressure–temperature–time613

paths: A forward approach. In: Spear, F. S., Peacock, S. M. (Eds.), Metamorphic614

Pressure–Temperature–Time Paths. Vol. 7. American Geophysical Union Short Course615

Series, Washington, DC, pp. 57–102.616

Peterman, E. M., Hacker, B. R., Baxter, E. F., 2009. Phase transformations of continental617

crust during subduction and exhumation: Western Gneiss Region, Norway. European618

Journal of Mineralogy 21, 1097–1118.619

Petrini, K., Podladchikov, Y., 2002. Lithospheric pressure–depth relationship in620

compressive regions of thickened crust. Journal of Metamorphic Geology 18, 67–77.621

Powell, R., Holland, T. J. B., 1988. An internally consistent dataset with uncertainties and622

correlations: 3. Applications to geobarometry, worked examples, and a computer623

program. Journal of Metamorphic Geology 6, 173–204.624

Powell, R., Holland, T. J. B., 2008. On thermobarometry. Journal of Metamorphic625

Geology 26, 155–179.626

Press, W. H., Flannery, B. P., Teukolsky, S. A., Vetterling, W. T., 1992. Runge-Kutta627

Method. Numerical Recipes in FORTRAN: The Art of Scientific Computing. Cambridge628

University Press, Ch. 16.1–16.2, pp. 704–716.629

24

Reuber, G., Kaus, B. J. P., Schmalholz, S. M., White, R. W., 2016. Nonlithostatic pressure630

during subduction and collision and the formation of (ultra)high-pressure rocks. Geology631

44, 343–346.632

Rubie, D. C., 1986. The catalysis of mineral reactions by water and restrictions on the633

presence of aqueous fluid during metamorphism. Mineralogical Magazine 50, 399–415.634

Shellnutt, J. G., Bhat, G. M., Wang, K. L., Brookfield, M. E., Jahn, B. M., Dostal, J.,635

2014. Petrogenesis of the flood basalts from the Early Permian Panjal Traps, Kashmir,636

India: Geochemical evidence for shallow melting of the mantle. Lithos 204, 159–171.637

St-Onge, M. R., Rayner, N., Palin, R. M., Searle, M. P., Waters, D. J., 2013. Integrated638

pressure–temperature–time constraints for the Tso Morari dome (NW India):639

Implications for the burial and exhumation path of UHP units in the western Himalaya.640

Journal of Metamorphic Geology 31, 469–504.641

Stockhert, B., Renner, J., 1998. Rheology of crustal rocks at ultrahigh pressure. In: Hacker,642

B. R., Liou, J. G. (Eds.), When Continents Collide: Geodynamics and Geochemistry of643

Ultrahigh-Pressure Rocks. Kluwer Academic Publishers, Dordrecht, pp. 57–95.644

Syracuse, E. M., van Keken, P. E., Abers, G. A., 2010. The global range of subduction645

zone thermal models. Physics of the Earth and Planetary Interiors 183, 73–90.646

Thielmann, M., Kaus, B. J. P., 2012. Shear heating induced lithospheric-scale localization:647

does it result in subduction? Earth and Planetary Science Letters 359–360, 1–13.648

Warren, C. J., 2013. Exhumation of (ultra-)high-pressure terranes: concepts and649

mechanisms. Solid Earth 4, 75–92.650

Warren, C. J., Beaumont, C., Jamieson, R. A., 2008. Modelling tectonic styles and651

ultra-high pressure (UHP) rock exhumation during the transition from oceanic652

subduction to continental collision. Earth and Planetary Science Letters 267, 129–145.653

Whitney, D., Evans, B., 2010. Abbreviations for names of rock-forming minerals. American654

Mineralogist 95, 185–187.655

25

Wilke, F. D. H., O’Brien, P. J., Schmidt, A., Ziemann, M. A., 2015. Subduction, peak and656

multi-stage exhumation metamorphism: Traces from one coesite-bearing eclogite, Tso657

Morari, western Himalaya. Lithos 231, 77–91.658

Young, D. J., Kylander-Clark, A. R. C., 2015. Does continental crust transform during659

eclogite facies metamorphism? Journal of Metamorphic Geology 33, 331–357.660

26

Figures661

Ordovician granite

GHS

Thrust fault

Normal fault

Neotethyan shelfTso Morari

massif (eclogite, orthogneiss,

metasediment)

Mélange and ophiolite

TownRiver

Karakoram Range

Trans-Himalayan batholith

HIG

HH

IMAL

AYA

LehLeh

PadumPadum

lakelake

Ladakh batholith

Ladakh batholith

Zanzkar Range (Tethyan Series)

Zanzkar Range (Tethyan Series)

Shyok suture zone

Shyok suture zone

Indus suture zone

Indus suture zone

77°E77°E

78°E78°E

34°N34°N

77°E77°E 78°E78°E

NN

40 km40 km

Panjal Trapbasalt

Panjal TrapbasaltZanskar

River

ZanskarRiver

Tso Morari

Tso Morari

30°N

KG ST

INDIA

TIBET30°N

75°E 80°E

HIMALAYAN RANGEMFT

ISZ

PolokongkaLa

PolokongkaLa

lakelake

Figure 1: Simplified geological map of the Tso Morari region, northwest India, modified after

Epard and Steck (2008), and Shellnutt et al. (2014). Inset shows its location within the Himalayan–

Tibetan orogen. GHS = Greater Himalayan Series, ISZ = Indus Suture Zone, KG = Kaghan Valley,

MFT = Main Frontal Thrust, ST = Stak Valley.

27

Temperature (°C)600 800700100 300 400 500 1000900200

Pres

sure

(GPa

)

3.5

0.5

1.0

1.5

2.0

2.5

3.0

0

5.0

4.0

4.5

Approximate depth assum

ing lithostatic pressure (km)

48

112

32

16

64

144

160

128

0

96

80

150 °

C/GPa

Jd + Qtz

Ab

CoeCoe

Qtz

GrDia

Syra

cuse et al. (2010)

Zo-EclZo-EclLawso

niteec

logite

Lawso

niteec

logite

Dry eclogiteDry eclogite

Amp-EclAmp-Ecl

GranuliteGranuliteAmphiboliteAmphibolite

Ep-Amp

Ep-Amp

GsGs

PPPPZeoZeo

Diamondeclogite

Diamondeclogite

UH

TU

HT

Blueschist

Blueschist

High-Pgranulite

High-Pgranulite

W15

M03

S13

G97D97

G97 D97S13 W15

CJ15

0 Myr0 Myr

2 Myr2 Myr

3–7 Myr3–7 Myr

1

2

3

0

200 400 600

T (°C)

P (G

Pa)

QtzQtzCoeCoe

1 Myr1 Myr

CJ15

Full t

ransfo

rmat

ion

Full t

ransfo

rmat

ion

Non-transf

orm

ation

Non-transf

orm

ation

Full t

rans

form

atio

n

Full t

rans

form

atio

n

Non-tran

s

Non-tran

s

Figure 2: Pressure–temperature (P–T ) plot of reported metamorphic conditions for Tso Morari

units alongside our thermomechanically modelled prograde P–T paths M3 (non-transformation)

and M4 (full transformation). Inset shows elapsed simulation time at 1-Myr intervals along each

path. Dashed black arrow continuing to higher pressures represents a non-calculated theoretical

extension of these paths. Thick dashed white arrow represents an estimated exhumation path.

Boxes indicate estimated P–T conditions of metamorphism during subduction and exhumation:

D97 = De Sigoyer et al. (1997), G97 = Guillot et al. (1997), M03 = Mukherjee et al. (2003), S13

= St-Onge et al. (2013), CJ15 = Chatterjee and Jagoutz (2015), and W15 = Wilke et al. (2015).

Facies grid is modified after Maruyama et al. (1996): Zeo = zeolite, PP = prehnite–pumpellyite,

Gs = greenschist, Ep-Amp = epidote-amphibolite, Amp-Ecl = amphibole-eclogite, and Zo-Ecl =

zoisite-eclogite. Band labeled Syracuse et al. (2010) represents P–T profiles for descending slab

surfaces in modern-day subduction zones. Geotherm marked 150 �C/GPa denotes the limit of the

Forbidden Zone (Liou et al., 2000).

28

2010 0000

3.00

2.80

2.60

2.952.95

2.652.65

3.05

CoeCoeQtzQtz

70

0

70

60609090

202010103.503.50

3.40

3.153.15

3.553.55

3.35

3.503.50

CoeCoeQtzQtz

3.003.002.952.95

Temperature (°C)500 600 800700300 400

3.7

2.7

2.9

3.1

3.3

3.5

2.5

2.8

3.0

3.2

3.4

3.6

2.6Bulk

-rock

den

sity

(0.0

01 ×

kg/

m3 )

100

0

>0

10

20

30

40

50

60

70

80

90

Bulk

-rock

H2O

as a

free

pha

se (%

)

1.0

2.0

3.0

0.2

5.0

4.0

Pres

sure

(GPa

)

POLOKONGKA LA GRANITE PANJAL TRAP BASALT

1.0

2.0

3.0

5.0

4.0

Pres

sure

(GPa

)

Temperature (°C)500 600 800700300 400

Figure 3: Variations in bulk-rock water contents and densities for metamorphosed Polokongka

La granite and Panjal Trap basalt protoliths. Fully labeled phase diagrams are presented in Figs

S1–4. Stippled regions represent the global range of P–T profiles for descending slab surfaces

in modern-day subduction zones (Syracuse et al., 2010). Thick, dashed, white arrows represent

an estimated exhumation path reproduced from De Sigoyer et al. (2000). Grey boxes represent

previously reported P–T conditions of peak metamorphism, as outlined in Fig. 2.

29

0 4000Width (km)

No stress

Free slip

Free

slip

Free slip

Dep

th (k

m)

0

200

400

660

Upper crustLower crust

Lithospheric mantle

Asthenospheric mantle

Weak channel

100 °C

500 °C

900 °C

1300 °C

1700 1800 1900 2000 2100

Figure 4: Initial setup for geodynamic modelling. Lithospheric convergence was controlled by a

pushing force directed from right-to-left (red box and arrow). A weak channel was included in order

to initialise subduction. 1 and 2 = upper crust, 3 and 4 = lower crust, 5 = lithospheric mantle, 6 =

aesthenospheric mantle, and 7 = weak channel. Detailed information about rheological parameters

is given in Tables S2 and S3 in the Supplementary Information.

30

Temperature (°C)470 520 570370 420 620320270 670 720 770

Mod

al p

ropo

rtio

n

0.4

0.6

0.8

1.0

0.2

0

Density (ρ) (0.001 × kg/m

3)

3.3

3.4

3.8

3.2

3.1

3.5

3.6

3.7

2.42.0 2.82.61.4 1.6 1.8 4.83.2 3.6 4.0 4.4Pressure (GPa)

sub-

met

amor

phic

/rap

idly

dec

reas

ing

relia

bilit

ysu

b-m

etam

orph

ic/r

apid

ly d

ecre

asin

g re

liabi

lity

A

Grt rim

Ctd

GlnAct

Omp

Tlc

Ms

BtChl

ChlEp

Rt H2OSpnLws

Ms

Coe

Brs–Wnc

H2O

Grt rimGrt core Grt core

ρbulk-rockFR

AC

sub-

met

amor

phic

/rap

idly

dec

reas

ing

relia

bilit

ysu

b-m

etam

orph

ic/r

apid

ly d

ecre

asin

g re

liabi

lity

Mod

al p

ropo

rtio

n

0.4

0.6

0.8

1.0

0.2

0

Temperature (°C)470 520 570370 420 620320270 670 720 770

Density (ρ) (0.001 × kg/m

3)

2.9

3.0

3.1

2.8

2.6

2.7

2.42.0 2.8Pressure (GPa) 2.61.4 1.6 1.8 4.83.2 3.6 4.0 4.4B

Omp

CoeQtz

Ms

Kfs

Grt

Jd

Ky

Rt

Jd

Ms

Kfs

Ky

Grt

ρbulk-rock

Figure 5: Changes in calculated mineral proportions for both protoliths during metamorphism

along the calculated M4 simulation P–T path (up to ⇠2.8 GPa and ⇠620 �C) and its theoretical

extension up to 4.8 GPa and ⇠770 �C shown in Fig. 2. A: Panjal Trap basalt. B: Polokongka

La granite. Bulk-rock densities (dashed lines labeled ⇢bulk-rock

) assume complete transformations.

Mineral abbreviations are after Whitney and Evans (2010): Act, actinolite; Brs, barroisite; Bt,

biotite; Chl, chlorite; Coe, coesite; Ctd, chloritoid; Ep, epidote; Gln, glaucophane; Grt, garnet;

Jd, jadeite; Kfs, K-feldspar; Ky, kyanite; Lws, lawsonite; Ms, muscovite; Omp, omphacite; Qtz,

quartz; Rt, rutile; Spn, sphene; Tlc, talc; Wnc, winchite. FRAC denotes the P–T conditions at

which garnet cores were fractionated out of the eclogite bulk composition.

31

Temperature (°C)

600 800700100 300 400 500 9002000

Pres

sure

(GPa

)

1.0

2.0

3.0

0

5.0

4.0

6.0

Mean P–T path for entiresubducted upper crustin each simulation

Jd + Qtz

150 °C/G

Pa

Qtz

DiaGr

Coe

Ab

Forbidden zoneSy

racu

se et

al. (2

010)

High weakzone angle

Temperature-dependent conductivity

Hydratedcrust

Temperature (°C)

600 800700100 300 400 500 9002000

Pres

sure

(GPa

)

1.0

2.0

3.0

0

5.0

4.0

Jd + Qtz

150 °

C/GPa

Ab

P–T paths for reasonablyexhumable upper-crustal rocks in eachsimulation

Syra

cuse

et al

. (2010)

High platevelocity

High weakzone angle

Mafic granulitecrustal flow law

Temperature-dependent conductivity

T-dependentconductivity

GrDia

Temperature (°C)

200 600 800 1000 12004000

Pres

sure

(GPa

)

2.0

3.0

4.0

5.0

6.0

0

10.0

7.0

1.0

8.0

9.0

Jd + Qtz

150 °C/G

Pa

Coe

DiaGr

Qtz

Ab

P–T path for highest-pressure upper-crustalnode in each simulation

Syra

cuse

et a

l. (2

010)

High weakzone angle

Forbiddenzone

Temperature-dependent

conductivity

Temperature (°C)

200 600 800 1000 12004000

Pres

sure

(GPa

)

2.0

3.0

4.0

5.0

6.0

0

8.0

7.0

1.0

Jd + Qtz

150 °C/G

Pa

Coe

DiaGr

Qtz

Ab

Forbiddenzone

P–T path for highest-temperature upper-crustalnode in each simulation

Syra

cuse

et

al. (

2010

)

High weakzone angle

Temperature-dependent

conductivity

Figure 6: Results of numerical modelling sensitivity tests. Each arrow represents a P–T path

calculated for a di↵erent set of model parameters. Red arrows represent P–T paths calculated for

simulations M1–M4. Band labelled Syracuse et al. (2010) represents the range of P–T profiles for

descending slab surfaces in modern-day subduction zones, and gray boxes show P–T metamorphic

conditions reported for the Tso Morari massif by various studies (cf. Fig. 2). Top left: Mean

P–T paths for the whole subducted upper crust. Blue squares represent one standard deviation

in temperature and pressure. Top right: P–T paths of reasonably exhumable rocks. Bottom row:

P–T paths for nodes that reached the maximum pressure within the subducted upper crust (left)

and lower crust (right) in each simulation.

32

CRUST

coesitequartz

quartz

coesite

CRUST

MANTLE

MOHO

Maximum overpressure (GPa)

0 0.2 0.4 0.6 0.8 1.0−1.0 −0.8 −0.6 −0.4 −0.2

1650 1700 1750 1800 1850 1900 1950 2000 2050 2100 2150 2200

Dep

th (k

m)

0

50

100

150

200

7 Myr

ASIAN PLATE INDIAN PLATE

SUTUREZONE

TSO MORARIMASSIF UNITS

Width (km)1650 1700 1750 1800 1850 1900 1950 2000 2050 2100 2150 2200

MANTLE

Dep

th (k

m)

0

50

100

150

200

1.2−1.2

A

BTSO MORARI

MASSIF UNITS

Figure 7: Results of integrated petrological–geodynamic modelling for simulation M3. A: Geo-

dynamic model components after 7 Myr of subduction. Green square represents the locus of the

Tso Morari massif units under simulation M3 conditions after this time period, and blue hexagon,

shown for reference, represents these units for simulation M4. B: Maximum overpressure recorded

by each individual node. Solid black lines demarcate the crust and dashed black line represents

a P–T locus for the quartz–coesite transition. Note that the lower crust in these simulations is

mechanically homogeneous and internal small-scale heterogeneities are ignored, which may result

in additional overpressures (Reuber et al., 2016).

33

Tables662

Table 1: Bulk-rock compositions used in phase diagram construction for the Polokongka La granite

(Girard and Bussy, 1999) and the Panjal Trap basalt (Shellnutt et al., 2014) given as molar pro-

portions of oxides: aunfractionated composition (no sequestration of elements into garnet cores),

bfractionated bulk composition (following sequestration of elements into garnet cores). FeOtot =

all iron as FeO, such that mol.% O = mol.% Fe2

O3

.

Lithology H2

O SiO2

Al2

O3

CaO MgO FeOtotal K2

O Na2

O TiO2

O

Polokongka La granite 2.04 80.00 8.35 0.68 0.48 1.55 3.88 2.46 0.15 0.40

Panjal Trap basalta 13.11 44.43 9.36 8.74 10.42 8.60 0.36 3.06 1.49 0.43

Panjal Trap basaltb 13.95 44.54 9.07 8.01 11.01 7.76 0.38 3.25 1.59 0.44

Table 2: Summary characteristics of the reference geodynamic models presented in this work.

aprofile after Leech et al. (2005), bprofile modified from Guillot et al. (2003).

Model no. Convergence velocity Phase transformation

M1 Constant (7 cm/yr)a No: constant terrane density of 2700 kg/m3

M2 Constant (7 cm/yr)a Yes: P–T -dependent terrane density (cf. Fig. 3)

M3 Deceleration (10 cm/yr for first No: constant terrane density of 2700 kg/m3

2 Myr and 4 cm/yr afterwards)b

M4 Deceleration (10 cm/yr for first Yes: P–T -dependent terrane density (cf. Fig. 3)

2 Myr and 4 cm/yr afterwards)b

34

1

SUPPLEMENTARY INFORMATION for:

Subduction metamorphism in the Himalayan ultrahigh-pressure Tso

Morari massif: an integrated geodynamic and petrological modelling

approach

Richard M. Palin*1,2, Georg Reuber1, Richard W. White1, Boris J.P. Kaus1 and Owen M.

Weller3 1Institute of Geosciences, Johannes-Gutenberg University of Mainz, 55128 Mainz, Germany 2Department of Geology and Geological Engineering, Colorado School of Mines, Golden,

80401, Colorado, USA 3Department of Earth Sciences, University of Cambridge, Cambridge, CB2 3EQ, UK

*corresponding author: [email protected]

Contents

• Petrological modelling parameters, methodology, and bulk compositions

• Tables S1–S5

• Figures S1–S8

• References

BULK COMPOSITIONS AND ACTIVITY–COMPOSITION RELATIONS USED FOR

METAMORPHIC PHASE EQUILIBRIA MODELLING

Petrological phase diagram modeling utilized the following activity–composition relations for

solid-solution phases: glaucophane, actinolite, hornblende, gedrite, diopside, omphacite, and

jadeite (Diener and Powell, 2012), muscovite and paragonite (Coggon and Holland, 2002), talc

and epidote (Holland and Powell, 1998), chlorite (Holland et al., 1998), cordierite, biotite and

garnet (White et al., 2007), plagioclase and K-feldspar (Holland and Powell, 2003), ilmenite

and hematite (White et al., 2000). Pure phases comprised albite, zoisite, lawsonite, rutile,

sphene, quartz, sillimanite, andalusite, kyanite, and fluid (H2O).

2

Whereas the whole-rock composition for Polokongka La granite sample AS9660 reported by

Girard and Bussy (1999) specified both FeO and Fe2O3 contents, that reported for Panjal Trap

basalt sample PJ2-010 by Shellnutt et al. (2014) presented all iron as Fe2O3. As such, a bulk-

rock XFe3+ = Fe3+(Fe2++ Fe3+) = 2 × O/FeOtotal ratio of 0.1 was applied for our modeling (Table

1 in the main manuscript), in keeping with the observation that unaltered mafic igneous rocks

generally have a low oxidation state (Schilling et al., 1983). The bulk-rock XFe3+ ratio for

Polokongka La granite sample AS9660 was not modified from its reported value.

Finally, fractionation of the eclogite bulk-rock composition owing to sequestration of cations

into growing garnet porphyroblasts during prograde metamorphism was considered in order to

examine the effect of an evolving bulk composition on the sensitivity of calculated phase

equilibria. Distinct compositional breaks across core–rim boundaries in garnet (e.g. St-Onge et

al., 2013; Wilke et al., 2015) were chosen as a key datum point for this procedure. Core

domains were assumed to represent ~20 vol.% of entire grains based on examination of

compositional line profiles in samples of fresh eclogite and assuming a spherical porphyroblast

geometry. Garnet constitutes ~30 vol.% of peak UHP eclogite-facies parageneses (Wilke et al.,

2015), and so a proportion of ~6 vol.% was used to represent the criterion at which core growth

was superceded by rim growth during subduction. Sample PJ2-010 was calculated to form 6%

garnet along the prograde P–T path calculated via geodynamic modelling at ~18.75 kbar (Figs

S3 and S4), and fractionation was performed at this point.

TABLE S1

Lithology SiO2 TiO2 Al2O3 Fe2O3 FeO MnO MgO CaO Na2O K2O LOIe

Polokongka La granitea

74.46 0.19 13.19 1.00 0.83 0.02 0.30 0.59 2.36 5.66 0.57

Panjal Trap basaltb 45.78 2.04 16.37 11.78c N.R.d 0.16 7.20 8.41 3.25 0.58 4.05

Protolith bulk-rock compositions (wt% oxide). aGirard and Bussy (1999), sample AS9660, bShellnutt et al. (2014), sample PJ2-010, cgiven in original study as total iron, dN.R. = not

reported, eLOI = loss on ignition.

3

TABLE S2

Component C (MPa)a Φb Cp (J/K)c k [W/(mK)]d Q (W/m³)e ρ0 (kg/m3)f

Upper crust 1 5.00 1000 2.339 1.750 × 10−6 2700g

Lower crust 1 5.00 1000 1.97 0.250 × 10−6 2900

Lithosphere 1 28.68 1000 1.99 0.022 × 10−6 3300

Asthenosphere 1 28.68 1000 1.99 0.022 × 10−6 3300

Weak channel 1 3.44 1000 2.00 0 3300

Material properties used for the main (reference) numerical simulation. Overall rheological

parameters are assembled from Turcotte and Schubert (2014), Li et al. (2010), and Bittner and

Schmeling (1995). aCohesion, bfriction angle, cheat capacity, dthermal conductivity, eradioactive heat generation, freference density (ρ = f(ρ0, T)), gdensities varied in the case of a

transformation simulation.

TABLE S3

Component Flow lawa A (MPa−n s−1)b nc E (kJ/mol)d V (m³/mol)e

Upper crust Quartzitef 6.7 × 10−6 2.4 156 0

Lower crust Plagioclaseg 3.3 × 10−4 3.2 238 0

Lithosphere Dry olivineh 2.5 × 104 3.5 532 17 × 10−6

Asthenosphere Dry olivineh 2.5 × 104 3.5 532 17 × 10−6

Weak channel Wet olivinei 2.0 × 103 4.0 471 0

Rheological parameters used for the main (standard-state) numerical simulation. aDislocation

creep flow law, bviscosity prefactor, cpower law exponent, dactivation energy, eactivation

volume, fRanalli and Murphy (1987), gShelton and Tullis (1981), hKirby (1983), iRanalli

(1995).

4

TABLE S4

Phas

e tra

nsiti

on

Con

verg

ence

rate

Dec

eler

atio

n

Flow

law

(U

C)

Flow

law

(L

C)

Wea

k zo

ne

angl

e

Eros

ion

rate

Sedi

men

tat

ion

rate

Fric

tion

angl

e (U

C)

Fric

tion

angl

e (L

C)

Q

Cp

Mea

n P

Mea

n T

Peak

P

Peak

T

Comments

(cm/a) (°) (mm/a) (mm/a) (W/m3) (J/K) (GPa) (°C) (GPa) (°C)

Ref_7 n 7 n Quartzite Plagioclase (An75) 15 5 1 5 5 v 1000 0.7 157 5 1030 M1

Ref_7_P y * * * * * * * * * * * 0.6 154 5.1 913 M2 Ref_10 * 10 y * * * * * * * * * 0.6 170 4 623 M3 Ref_10_P y 10 y * * * * * * * * * 0.7 165 4.2 832 M4

Ref_7_Hyd * * * * * * * * * * * * 0.4 120 0.8 220 Similar to reference; hydrated rock

Ref_10_Hyd * 10 * * * * * * * * * * 0.4 135 0.9 240 Similar to reference; hydrated rock

Ref_Cr_5 * 5 * * * * * * * * * * 0.5 154 4 690 Similar to reference Ref_Cr_5_P y 5 * * * * * * * * * * 0.6 152 4 773 Similar to reference Ref_Cr_10 * 10 * * * * * * * * * * 0.9 185 7 1020 Faster subduction Ref_Cr_10_P y 10 * * * * * * * * * * 0.9 178 7.1 989 Faster subduction

Ref_Cr_15 * 15 * * * * * * * * * * 1 211 9.5 1080 Faster subduction, thus higher P

Ref_Cr_15_P y 15 * * * * * * * * * * 1 188 8.2 1080 Faster subduction, thus higher P

Ref_UC_G * * * Mafic granulite * * * * * * * * 0.7 174 4.4 794 Early underthrusting

Ref_UC_G_P y * * Mafic granulite * * * * * * * * 0.7 177 4.6 846 Early underthrusting

Ref_LC_G * * * * Mafic granulite * * * * * * * 0.6 171 3.6 672 Shallow subduction

Ref_LC_G_P y * * * Mafic granulite * * * * * * * 0.6 168 4 764 Shallow subduction

Ref_WA_20 * * * * * 20 * * * * * * 0.7 177 4.4 655 Steep subduction Ref_WA_20_P y * * * * 20 * * * * * * 0.8 176 6.3 1030 Steep subduction

5

Phas

e tra

nsiti

on

Con

verg

ence

rate

Dec

eler

atio

n Fl

ow la

w

(UC

)

Flow

law

(L

C)

Wea

k zo

ne

angl

e

Eros

ion

rate

Sedi

men

tat

ion

rate

Fric

tion

angl

e (U

C)

Fric

tion

angl

e (L

C)

Q

Cp

Mea

n P

Mea

n T

Peak

P

Peak

T

Comments

(cm/a) (°) (mm/a) (mm/a) (W/m3) (J/K) (GPa) (°C) (GPa) (°C) Ref_WA_30 * * * * * 30 * * * * * * 0.6 176 4.2 800 Steep subduction Ref_WA_30_P y * * * * 30 * * * * * * 0.7 173 7 965 Steep subduction

Ref_WA_40 * * * * * 40 * * * * * * 0.7 189 4.4 936 Massively steep subduction

Ref_WA_40_P y * * * * 40 * * * * * * 0.9 198 9.7 1080 Massively steep subduction

Ref_SE_11 * * * * * * 1 * * * * * 0.7 177 4 600 Similar to reference Ref_SE_11_P y * * * * * 1 * * * * * 0.7 170 4.1 601 Similar to reference Ref_SE_15 * * * * * * 1 5 * * * * 0.8 151 4.6 730 Shallow subduction Ref_SE_15_P y * * * * * 1 5 * * * * 0.8 141 4.8 824 Shallow subduction Ref_fa_UC * * * * * * * * 15 * * * 0.6 178 2.3 540 Similar to reference Ref_fa_UC_P y * * * * * * * 15 * * * 0.5 167 2.2 513 Similar to reference Ref_fa_LC * * * * * * * * * 15 * * 0.7 165 4.2 667 Similar to reference Ref_fa_LC_P y * * * * * * * * 15 * * 0.7 160 4.3 628 Similar to reference Ref_Q2 * * * * * * * * * * 2 * 0.5 152 3.1 620 Shallow subduction Ref_Q2_P y * * * * * * * * * 2 * 0.5 148 3.2 621 Shallow subduction Ref_Q0 * * * * * * * * * * 0 * 0.7 142 4.3 650 Similar to reference Ref_Q0_P y * * * * * * * * * 0 * 0.7 140 4.4 860 Similar to reference Ref_CP_1 * * * * * * * * * * * 750 0.7 186 4 684 Similar to reference Ref_CP_1_P y * * * * * * * * * * 750 0.7 182 5.4 987 Similar to reference Ref_CP_2 * * * * * * * * * * * 1250 0.5 136 2.7 584 Shallow subduction Ref_CP_2_P y * * * * * * * * * * 1250 0.5 133 2.7 573 Shallow subduction Ref_7_kT * * * * * * * * * * * * 0.6 248 6.6 1070 Similar to reference

Ref_7_kT_P y * * * * * * * * * * * 0.6 236 6.8 1063 Similar to reference

Ref_10_kT * 10 y * * * * * * * * * 0.7 242 6.2 976 Similar to reference

Ref_10_kT_P y 10 y * * * * * * * * * 0.7 232 6.4 1030 Similar to reference

6

Results of sensitivity testing. Reading from left to right: Phase transition: ‘y’ if phase transitions are applied, ‘n’ if not. Convergence rate (cm/a):

background velocity. Deceleration: ‘y’ if the background velocity is varied with time, ‘n’ if not. Flow law (UC): rheological flow law for the upper

crust. Flow law (LC): rheological flow law for the lower crust. Weak zone angle (°): angle of the weak zone. Erosion rate (mm/a): erosion rate.

Sedimentation rate (mm/a): sedimentation rate. Friction angle (UC): applied friction angle for the upper crust. Friction angle (LC): applied friction

angle for the lower crust. Q (W/m3): heat production for all phases, ‘v’ if it is variable. Cp (J/K): Heat capacity of all phases. Mean pressure: mean

pressure achieved by the subducted upper crust. Mean temperature: Mean temperature achieved by the subducted upper crust. Peak pressure: highest

traced pressure achieved by the subducted upper crust. Peak temperature: highest traced temperature achieved by the subducted upper crust.

Comments: Comments on the evolution of the simulation. All rheological flow laws are taken from Kirby (1983) and Ranalli (1995). The

nomenclature for the simulations is written in the format: “Ref” “Modified parameter”, alongside “P” if phase transitions were applied.

Abbreviations: *: same value as the reference simulation, UC: upper crust, LC: lower crust, P: pressure, T: temperature. Simulations Ref_7,

Ref_7_P, Ref_10, and Ref_10_P (red text) provided the P–T paths discussed in the paper.

TABLE S5

Component k1 k2

Sediments 0.64 807

Upper crust 0.64 807

Lower crust 1.18 474

Lithosphere 0.73 1293

Asthenosphere 0.73 1293

Weak channel 0.73 1293

Parameter values used in simulations that considered temperature-

dependent conductivity (Ref_7_kT, Ref_7_kT_P, Ref_10_kT, and

Ref_10_kT_P: Table S4), which used a conductivity–temperature

relationship k = k1 + k2/(T + 77) from Clauser and Huenges (1995),

where k1 and k2 are empirically derived prefactors and T is the

temperature in Kelvin. These simulations also used the nonlinear

crustal geotherm of McKenzie et al. (2005), which is shown in Fig. S8

7

FIGURE S1

Results of metamorphic mineral equilibria modeling for the bulk-rock composition of Polokongka La granite sample AS9660 (Table 1 in main manuscript text) from Girard and Bussy (1999). (a) Pressure–temperature (P–T) pseudosection. Some small, minor fields are unlabeled for clarity. Assemblage field boundaries marking the appearance or disappearance of magnetite, sphene, ilmenite, hematite, or rutile, are shown as grey dashed lines, which are omitted from the labeled assemblages due to their negligible calculated modal proportions (<1 mol.%). Inset at the top left corner of the diagram shows the P–T relations between these phases. Dotted region represents the global range of descending slab-surface P–T profiles in modern-day subduction zones (Syracuse et al., 2010), and dashed line labeled 150 °C/GPa represents the limit of the Forbidden Zone (Liou et al., 2000). Numbered fields are as follows: 1 – Grt Ms Omp Jd Gln Kfs Ky Qtz, 2 – Grt Ms Omp Jd Kfs Ky Ab Qtz, 3 – Bt Ms Jd Ep Kfs Ab Qtz, 4 – Grt Bt Ms Jd Ep Kfs Ab Qtz, 5 – Bt Ms Pl Kfs Ep Qtz H2O, 6 – Bt Ms Pl Kfs Ab Qtz H2O, 7 – Bt Ms Jd Pl Kfs Qtz, 8 – Grt Ms Jd Pl Kfs Ab Qtz, 9 – Liq Bt Ms Pl Kfs Qtz H2O, 10 – Grt Ms Jd Pl Kfs Qtz, 11 – Liq Grt Bt Ms Pl Kfs Qtz, 12 – Liq Bt Pl Kfs Ky Qtz, 13 – Ms Jd Kfs Ky Qtz.

8

FIGURE S2

Overlay of metamorphic P–T conditions reported for units of the Tso Morari massif (gray and white boxes, and white dashed arrow) and prograde P–T paths calculated via geodynamic modeling (green and yellow lines, and black dashed arrow) onto phase diagrams calculated for Polokongka La granite sample AS9660 (Fig. S1). Acronyms for studies are the same as those listed in the caption to Fig. 2 in the main manuscript. Blue P–T path ending with a hexagon represents that calculated for a time-dependent decreasing subduction velocity (‘deceleration’) and full mineralogical transformation, and green P–T path ending with a square represents that calculated for a time-dependent decreasing subduction velocity (‘deceleration’) with no mineralogical transformation. See the “Geodynamic numerical modeling” section of the main manuscript for more information. All other labels are equivalent to Fig. S1.

9

FIGURE S3

Pressure–temperature (P–T) pseudosection calculated for Panjal Trap basalt sample PJ2-010 from Shellnutt et al. (2014), considering the effects of cation sequestration into garnet cores (Table 1 in the main manuscript text). Some small, minor fields are unlabeled for clarity. Line colors and shadings as for Fig. S1. Bold red line marks the limit of garnet-bearing assemblage fields, and sub-parallel red dashed line marks where garnet comprises 6% of the calculated assemblage using an unfractionated bulk composition. As such, assemblage fields occurring to the high-P side of this contour represent equilibria calculated following fractionation of the effective bulk composition due to removal of garnet core domains. Numbered fields are as follows: 1 – Grt Bt Chl Ctd Tlc Omp Gln Lws, 2 – Grt Bt Chl Ctd Omp Act Gln Lws, 3 – Bt Chl Ep Ctd Omp Act Gln Lws, 4 – Bt Chl Ctd Omp Act Gln Lws, 5 – Chl Ms Ep Omp Gln Act Lws, 6 – Chl Ms Ep Omp Gln Act Lws Qtz, 7 – Bt Chl Ms Ep Omp Qtz Ab, 8 – Bt Chl Ep Act Gln Qtz Ab, 9 – Bt Chl Ep Act Hbl Qtz Ab, 10 – Bt Chl Ms Ep Gln Qtz, 11 – Chl Ms Ep Omp Gln Qtz, 12 – Bt Chl Ms Ep Gln Qtz Zo, 13 – Bt Chl Hbl Pl Qtz, 14 – Bt Chl Act Hbl Pl, 15 – Bt Hbl Kfs Pl H2O, 16 – Bt Opx Hbl Pl H2O, 17 – Bt Ms Hbl Qtz Zo H2O, 18 – Grt Bt Di Hbl Pl Qtz Zo H2O, 19 – Grt Ms Omp Hbl Qtz Zo H2O, 20 – Grt Ms Pg Omp Hbl Qtz Zo H2O, 21 – Chl Ms Ep Omp Gln Qtz Zo, 22 – Grt Chl Ms Ep Omp Gln Act Lws, 23 – Grt Chl Ms Omp Act Gln Lws, 24 – Grt Chl Ms Omp Gln Lws H2O, 25 – Grt Chl Ms Omp Gln Zo H2O, 26 – Grt Ms Omp Gln Zo H2O, 27 – Grt Ms Omp Gln Ky H2O, 28 – Grt Ms

10

Omp Hbl Ky H2O, 29 – Grt Ms Omp Coe Lws Ky H2O, 30 – Grt Ms Omp Coe Lws Tlc H2O, 31 – Grt Ms Tlc Omp Gln Lws H2O, 32 – Grt Bt Ms Tlc Omp Gln Lws H2O, 33 – Grt Bt Chl Tlc Omp Gln Lws H2O.

FIGURE S4

Overlay of metamorphic P–T conditions reported for units of the Tso Morari massif (gray and white boxes, and white dashed arrow) and prograde P–T paths calculated via geodynamic modeling (green and yellow lines, and black dashed arrow) onto phase diagrams calculated for Panjal Trap basalt sample PJ2-010 (Fig. S2). Acronyms for studies are the same as those listed in the caption to Fig. 2 in the main manuscript. Blue P–T path ending with a hexagon represents that calculated for a time-dependent decreasing subduction velocity (‘deceleration’) and full mineralogical transformation, and green P–T path ending with a square represents that calculated for a time-dependent decreasing subduction velocity (‘deceleration’) with no mineralogical transformation. See the “Geodynamic numerical modeling” section of the main manuscript text for more information. All other labels are equivalent to Fig. S2.

11

FIGURE S5

Results of geodynamic numerical modeling for the four end-member scenarios discussed in the

main text: A: Constant velocity with non-transformation (M1), B: Constant velocity with full

transformation (M2), C: Deceleration with non-transformation (M3), D: Deceleration with full

transformation (M4). Insets for each figure part show the modeled evolution of P–T conditions at 1-

Myr time intervals.

12

FIGURE S6

Results of numerical modelling sensitivity tests, color-coded for simulation time (Myr). The data

shown are identical to those in Fig. 6 of the main manuscript, with additional 1-Myr-interval

shading. Note that the upper crustal nodes that reached the highest pressures in each simulation

(bottom left panel) were not necessarily the same as those nodes that reached the highest

temperatures (bottom right panel).

13

FIGURE S7

Evolution of the four simulations at an approximately equal stage of subduction. The upper crust did not subduct to greater depths in any of the

cases. The numerical times of the simulations are highlighted in white within each subfigure. Markers representing the Tso Morari massif units have

the same shape and color as they appear in the main paper and elsewhere in the Supplementary Information.

14

FIGURE S8

Geotherms utilized in geodynamic simulations. Red line represents the non-linear geotherm of

McKenzie et al. (2005), which was utilized in simulations that considered temperature-dependent

conductivity. Blue line represents the constant 9 ºC/km geotherm employed for all other models,

including the reference (preferred) simulations.

15

REFERENCES CITED

Bittner, D., and Schmeling, H., 1995. Numerical modelling of melting processes and induced

diapirism in the lower crust. Geophysical Journal International, 123, 59–70.

Clauser, C., and Huenges, E., 1995. Thermal conductivity of rocks and minerals. Rock physics &

phase relations: A handbook of physical constants, 105–126.

Coggon, R., and Holland, T.J.B., 2002. Mixing properties of phengitic micas and revised garnet-

phengite thermobarometers. Journal of Metamorphic Geology, 20, 683–696.

Diener, J.F.A., and Powell, R., 2012. Revised activity–composition models for clinopyroxene and

amphibole. Journal of Metamorphic Geology, 30, 131–142.

Girard, M., and Bussy, F., 1999. Late Pan-African magmatism in the Himalaya: new

geochronological and geochemical data from the Ordovician Tso Morari metagranites

(Ladakh, NW India). Schweizerische Mineralogische und Petrographische Mitteilungen, 79,

399–417.

Holland, T.J.B., and Powell, R., 1998. An internally-consistent thermodynamic dataset for phases of

petrological interest. Journal of Metamorphic Geology, 16, 309–344.

Holland, T.J.B., and Powell, R., 2003. Activity–composition relations for phases in petrological

calculations: an asymmetric multicomponent formulation. Contributions to Mineralogy and

Petrology, 145, 492–501.

Holland, T.J.B., Baker, J.M., and Powell, R., 1998. Mixing properties and activity–composition

relationships of chlorites in the system MgO–FeO–Al2O3–SiO2–H2O: European Journal of

Mineralogy, 10, 395–406.

Kirby, S.H., 1983. Rheology of the lithosphere. Reviews of Geophysics, 21, 1458–1487.

Li, Z.H., Gerya, T.V., and Burg, J.P., 2010. Influence of tectonic overpressure on P–T paths of HP–

UHP rocks in continental collision zones: thermomechanical modelling. Journal of

Metamorphic Geology, 28, 227–247.

16

Liou, J.G., Hacker, B.R., and Zhang, R.Y., 2000. Into the Forbidden Zone. Science 287, 1215–

1216.

McKenzie, D., Jackson, J., and Priestley, K., 2005. Thermal structure of oceanic and continental

lithosphere. Earth and Planetary Science Letters, 233, 337–349.

Ranalli, G., 1995. Rheology of the Earth. Chapman & Hall, London.

Ranalli, G., and Murphy, D.C., 1987. Rheological stratification of the lithosphere. Tectonophysics,

132, 281–295.

Schilling, J.-G., Zajec, M., Evans, R., Johnston, T., White, W., Devine, J. D., and Kingsley, R.,

1983. Petrologic and geochemical variations along the Mid-Atlantic Ridge from 27ºN to

73ºN. American Journal of Science 283, 510–586.

Shellnutt, J.G., Bhat, G.M., Wang, K-L., Brookfield, M.E., Jahn, B.M., and Dostal, J., 2014.

Petrogenesis of the flood basalts from the Early Permian Panjal Traps, Kashmir, India:

Geochemical evidence for shallow melting of the mantle. Lithos, 204, 159–171.

Shelton, G., and Tullis, J., 1981. Experimental flow laws for crustal rocks. Eos Transactions,

American Geophysical Union, 62, 396.

Syracuse, E.M., van Keken, P.E., and Abers, G.A., 2010. The global range of subduction zone

thermal models. Physics of the Earth and Planetary Interiors, 183, 73–90.

Turcotte, D.L., and Schubert, G., 2014. Geodynamics: Cambridge University Press.

White, R.W., Powell, R., Holland, T.J.B., and Worley, B.A., 2000. The effect of TiO2 and Fe2O3 on

metapelitic assemblages at greenschist and amphibolite facies conditions: mineral equilibria

calculations in the system K2O–FeO–MgO–Al2O3–SiO2–H2O–TiO2–Fe2O3. Journal of

Metamorphic Geology, 18, 497–511.

White, R.W., Powell, R., and Holland, T.J.B., 2007. Progress relating to calculation of partial

melting equilibria for metapelites. Journal of Metamorphic Geology, 25, 511–527.