Open Research Developing an inverted Barrovian sequence; insights from monazite petrochronology

In the Himalayan region of Sikkim, the well-developed inverted metamorphic sequence of the Main Central Thrust (MCT) zone is folded, thus exposing several transects through the structure that reached similar metamorphic grades at different times. In-situ LA-ICP-MS U–Th–Pb monazite ages, linked to pressure–temperature conditions via trace-element reaction ﬁngerprints, allow key aspects of the evolution of the thrust zone to be understood for the ﬁrst time. The ages show that peak metamorphic conditions were reached earliest in the structurally highest part of the inverted metamorphic sequence, in the Greater Himalayan Sequence (GHS) in the hanging wall of the MCT. Monazite in this unit grew over a prolonged period between ∼ 37 and 16 Ma in the southerly leading-edge of the thrust zone and between ∼ 37 and 14.5 Ma in the northern rear-edge of the thrust zone, at peak metamorphic conditions of ∼ 790 ◦ C and 10 kbar. Monazite ages in Lesser Himalayan Sequence (LHS) footwall rocks show that identical metamorphic conditions were reached ∼ 4–6 Ma apart along the ∼ 60 km separating samples along the MCT transport direction. Upper LHS footwall rocks reached peak metamorphic conditions of ∼ 655 ◦ C and 9 kbar between ∼ 21 and 16 Ma in the more southerly-exposed transect and ∼ 14.5–12 Ma in the northern transect. Similarly, lower LHS footwall rocks reached peak metamorphic conditions of ∼ 580 ◦ C and 8.5 kbar at ∼ 16 Ma in the south, and 9–10 Ma in the north. In the southern transect, the timing of partial melting in the GHS hanging wall ( ∼ 23–19.5 Ma) overlaps with the timing of prograde metamorphism ( ∼ 21 Ma) in the LHS footwall, conﬁrming that the hanging wall may have provided the heat necessary for the metamorphism of the footwall. Overall, the data provide robust evidence for progressively downwards-penetrating deformation and accretion of original LHS footwall material to the GHS hanging wall over a period of ∼ 5 Ma. These processes appear to have occurred several times during the prolonged ductile evolution of the thrust. The preserved inverted metamorphic sequence therefore documents the formation of sequential ‘paleo-thrusts’ through time, cutting down from the original locus of MCT movement at the LHS–GHS protolith boundary and forming at successively lower pressure and temperature conditions. The petrochronologic methods applied here constrain a complex temporal and thermal deformation history, and demonstrate that inverted metamorphic sequences can preserve a rich record of the duration of progressive ductile thrusting. © 2014 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/3.0/). Such advances in and geochronological the complex tectonic and thermal evolution of thrust zones associated with inverted metamorphism. This study presents a comprehensive, 4D (pressure–tempera-ture–time–deformation) examination of the inverted Barrovian metamorphic sequence beneath the Himalayan Main Central Thrust in the Sikkim region of India. Petrochronology techniques that combine petrography, thermobarometric calculations, geochronological data, and trace-element geochemistry link the timing of prograde, peak and retrograde metamorphic conditions across three structural transects. Our data show that (1) the timing of peak metamorphism decreases structurally downwards from the initial thrust surface into the lower grade metamorphic rocks; (2) the timing of attainment of the same grade of peak metamorphism varies by ca. 4–6 Ma around the exposed MCT rocks in the Sikkim Himalaya and (3) the timing of melting in the hanging wall was, in some instances, contemporaneous with prograde or peak metamorphism in the proximal footwall. Overall, our data show that the MCT in Sikkim was active for at least 11 Ma during the Miocene.

In the Himalayan region of Sikkim, the well-developed inverted metamorphic sequence of the Main Central Thrust (MCT) zone is folded, thus exposing several transects through the structure that reached similar metamorphic grades at different times. In-situ LA-ICP-MS U-Th-Pb monazite ages, linked to pressure-temperature conditions via trace-element reaction fingerprints, allow key aspects of the evolution of the thrust zone to be understood for the first time. The ages show that peak metamorphic conditions were reached earliest in the structurally highest part of the inverted metamorphic sequence, in the Greater Himalayan Sequence (GHS) in the hanging wall of the MCT. Monazite in this unit grew over a prolonged period between ∼37 and 16 Ma in the southerly leading-edge of the thrust zone and between ∼37 and 14.5 Ma in the northern rear-edge of the thrust zone, at peak metamorphic conditions of ∼790 • C and 10 kbar. Monazite ages in Lesser Himalayan Sequence (LHS) footwall rocks show that identical metamorphic conditions were reached ∼4-6 Ma apart along the ∼60 km separating samples along the MCT transport direction. Upper LHS footwall rocks reached peak metamorphic conditions of ∼655 • C and 9 kbar between ∼21 and 16 Ma in the more southerly-exposed transect and ∼14.5-12 Ma in the northern transect. Similarly, lower LHS footwall rocks reached peak metamorphic conditions of ∼580 • C and 8.5 kbar at ∼16 Ma in the south, and 9-10 Ma in the north. In the southern transect, the timing of partial melting in the GHS hanging wall (∼23-19.5 Ma) overlaps with the timing of prograde metamorphism (∼21 Ma) in the LHS footwall, confirming that the hanging wall may have provided the heat necessary for the metamorphism of the footwall. Overall, the data provide robust evidence for progressively downwards-penetrating deformation and accretion of original LHS footwall material to the GHS hanging wall over a period of ∼5 Ma. These processes appear to have occurred several times during the prolonged ductile evolution of the thrust. The preserved inverted metamorphic sequence therefore documents the formation of sequential 'paleothrusts' through time, cutting down from the original locus of MCT movement at the LHS-GHS protolith boundary and forming at successively lower pressure and temperature conditions. The petrochronologic methods applied here constrain a complex temporal and thermal deformation history, and demonstrate that inverted metamorphic sequences can preserve a rich record of the duration of progressive ductile thrusting.

Introduction
Inverted metamorphic sequences are commonly associated with fault zones, such as in the soles of ophiolites (Jamieson, 1986; metamorphic sequences are intimately related to major mountainbuilding structures, they preserve a record of ductile thrusting processes. These zones therefore provide an ideal natural laboratory to explore the thermal and temporal evolution of a fundamental geological mechanism in continental collisional events. In the Himalayan orogen, the duration and conditions of thrusting have been traditionally addressed using a combination of thermobarometric and geochronological data, through the use of monazite geochronology, pseudosection modelling, and less commonly, major and accessory phase geochemical analysis (e.g. Bollinger and Janots, 2006;Catlos et al., 2001;Crowley and Parrish, 1999;Gibson et al., 1999;Goscombe and Hand, 2000;Groppo et al., 2009;Kohn et al., 2001;Stephenson et al., 2000;Vannay and Grasemann, 2001;Yakymchuk and Godin, 2012). Monazite analysis frequently yields complex geochronological datasets (e.g. Buick et al., 2006;Gasser et al., 2012;Hermann and Rubatto, 2003), making it difficult to link the ages with the P -T conditions and crystallisation reactions of the phases. Emerging developments in linking 'age to stage', and particularly the use of combined petrography and traceelement data from major and accessory phases (petrochronology), allow high-spatial precision geochronological data to be more firmly linked to the P -T conditions of accessory phase crystallisation reactions (Foster et al., 2004;Gasser et al., 2012;Janots et al., 2008;Rubatto et al., 2013). This is achieved by detailed observation of textural relationships, geochemical analysis of coexisting accessory phases and systematic documentation of trace-element signatures in major phases that 'fingerprint' monazite-forming reactions (Foster et al., 2002(Foster et al., , 2000(Foster et al., , 2004Gasser et al., 2012;Hermann and Rubatto, 2003;Hoisch et al., 2008;Janots et al., 2006Janots et al., , 2007Janots et al., , 2008Janots et al., , 2009Kingsbury et al., 1993;Pyle and Spear, 2003;Rubatto et al., 2006;Smith and Barreiro, 1990;Spear, 2010;Wing et al., 2003). Such advances in metamorphic and geochronological analysis offer the potential to unravel the complex tectonic and thermal evolution of thrust zones associated with inverted metamorphism.
This study presents a comprehensive, 4D (pressure-temperature-time-deformation) examination of the inverted Barrovian metamorphic sequence beneath the Himalayan Main Central Thrust in the Sikkim region of India. Petrochronology techniques that combine petrography, thermobarometric calculations, geochronological data, and trace-element geochemistry link the timing of prograde, peak and retrograde metamorphic conditions across three structural transects. Our data show that (1) the timing of peak metamorphism decreases structurally downwards from the initial thrust surface into the lower grade metamorphic rocks; (2) the timing of attainment of the same grade of peak metamorphism varies by ca. 4-6 Ma around the exposed MCT rocks in the Sikkim Himalaya and (3) the timing of melting in the hanging wall was, in some instances, contemporaneous with prograde or peak metamorphism in the proximal footwall. Overall, our data show that the MCT in Sikkim was active for at least 11 Ma during the Miocene.

Geological setting
The Himalayan orogen exposes one of the Earth's largest and best-preserved inverted Barrovian metamorphic sequences, which developed during thrusting along the 2500 km-long Main Central Thrust (MCT: Fig. 1; Catlos et al., 2001;Goscombe and Hand, 2000;Groppo et al., 2010;Kohn et al., 2001;Paudel and Arita, 2000;Sinha-Roy, 1982;Stephenson et al., 2000;Vannay and Grasemann, 2001;Vannay et al., 1999;Yakymchuk and Godin, 2012). The Darjeeling hills and Sikkim (collectively known as the Sikkim Himalaya) in the eastern Himalaya, represent the ideal location for studying the MCT (Fig. 1). During the Eocene-Recent collision of India and Asia, Indian crustal rocks were buried to crustal depths of 20-30 km before being exhumed by a combination of tectonic and surface processes. Miocene-aged movement along the MCT juxtaposed two isotopically distinct rock packages, the Greater and Lesser Himalayan Sequences, which originally formed the proximal and distal parts of the Indian margin (Mottram et al., 2014 and references therein). The MCT therefore represents both a protolith boundary and a wide zone of ductile shear. The ∼10 km thick thrust zone was folded after motion along it ceased, by an underlying late-stage duplex. This dome provides exposures of several different structural levels and 'time-windows' through the MCT zone (Bhattacharyya and Mitra, 2009;Fig. 1).
Samples of GHS, upper LHS and lower LHS material were collected from three transects across the deformed MCT 'zone': Takdah in the south-west, Mangan in the north and Rongli in the east (Fig. 1).

Electron Probe Micro-Analysis (EPMA) and scanning electron microscope (SEM)
Major elements in all minerals were analysed in polished thin section using the Open University Cameca SX100 EPMA. Mineral formulae were calculated on the basis of 12 oxygen for garnet, 8 for feldspars, 22 for micas and 46 for staurolite. Elemental X-ray maps of monazite, garnet and whole thin sections were produced using the EPMA and the Open University FEI Quanta 3D dual beam microscope SEM. Full operating conditions can be found in the supplementary material S.1.1-1.2.

Pressure-temperature calculations and modelling
Estimates of peak P -T were calculated using the Ti-in-biotite calculation of Henry et al. (2005), the garnet-biotite thermometer of Bhattacharya et al. (1992), the garnet-Al 2 SiO 5 -plagioclase (GASP) barometer of Powell and Holland (1988) and the Zr-in-rutile calibration of Tomkins et al. (2007), calculated at 9 kbar pressure.
Pseudosections of samples 16, 22, 60, 67 and 343 were constructed using Perple_X_6.6.8 (Connolly 1990(Connolly , 2009) using the internally consistent thermodynamic dataset and equation of state for H 2 O of Holland and Powell (2011). Samples were modelled in the system MnNKCFMASTH. Fe 3+ was ignored because there were no Fe 3+ -rich oxides in the assemblages and the allocated Fe 3+ in analysed phases was negligible. The addition of Fe 3+ to the bulk composition did not affect the overall topology of the final pseudosections (supplementary material S3.4). Pseudosections were calculated under fluid-saturated (aH 2 O = 1) conditions for samples that equilibrated under subsolidus conditions, and with H 2 O as a component for suprasolidus samples. The H 2 O content was calculated from the modal proportion of biotite, using the Ti-H substitution scheme of White et al. (2007). The bulk composition of each sample was calculated to represent the effective composition of the equilibrium volume either from an adapted XRF composition of the rock, or from calculating of the proportion of each mineral phase in the sample from analysis of thin section X-ray maps using Im-ageJ analysis software (supplementary material S3.4; Schneider et al., 2012).
The P -T conditions were constrained by comparing calculated weight% oxide isopleths of garnet (CaO, FeO and MgO) and plagioclase (CaO, Na 2 O) with observed compositions of those phases where present. As garnet zoning has a significant fractionation effect on the local effective bulk composition, it was necessary to consider two different effective bulk compositions in samples that showed significant garnet zoning, in order to calculate the conditions of garnet core and rim growth (Evans, 2004;Gaidies et al., 2006). The construction of garnet isopleths then allowed for the comparison of the corresponding garnet core or rim compositions. Full methods and solution models used are documented in the supplementary material S1.3.

U-Th-Pb monazite geochronology
U-Th-Pb concentrations in monazite were analysed at the NERC Isotope Geosciences Laboratories (NIGL), UK, using a Nu Attom sector-field single-collector inductively coupled plasma mass spectrometer (SC-ICP-MS) and New Wave Research UP193ss (193 nm) Nd:YAG laser ablation system. Monazites were commonly zoned in Y and Th, which were used to select suitable analysis points using laser conditions of 15 μm spot size at 5 Hz and ∼2.5 J cm −2 fluence (supplementary material S4.2).
The instrumental configuration and measurement procedures follow previous methods (Palin et al., 2013), and full analytical conditions are shown in the supplementary material S1.4. The monazite 'Stern' was used as the primary reference and 'Manangotry' and 'Moacyr' were used as secondary reference materials (Palin et al., 2013). Overall reproducibility achieved on the secondary reference material was <3% (2 SD) for U-Pb ages.
All data were processed using an in-house spreadsheet calculation routine. Full data tables of data can be found in the supplementary data tables. A 207 Pb-based common lead correction was applied to the U-Pb and Th-Pb ages. A further correction was made to the common lead corrected ages to account for excess 206 Pb derived from 230 Th, an intermediate daughter nuclide in the 238 U decay series. Uncorrected monazite data were plotted on Tera-Wasserburg plots, using Isoplot 4.14 (Ludwig, 2003), with intercept ages extrapolated from a 207 Pb/ 206 Pb ratio of 0.83 ± 0.02 (Stacey and Kramers, 1975). For samples with a sparse distribution of monazite spot ages, an MSWD is not quoted as it was not possible to constrict an intercept age through the array.
Both U-Pb and Th-Pb decay schemes can be used to date monazites (common-Pb ages shown in full data tables). Excess 206 Pb incorporated into monazite during crystallisation can cause the 238 U/ 206 Pb ages to appear older than the true crystallisation age, which is thought to be particularly significant in Cenozoic monazites. The common-Pb-corrected 232 Th/ 208 Pb ages have therefore been regarded as more accurate (Parrish, 1990). However in this study, due to the acquisition protocol, the 232 Th/ 208 Pb ages are typically less precise than the 238 U/ 206 Pb ages. We therefore quote the 238 U/ 206 Pb ages and uncertainties (±2σ ) when referring to the age of monazite populations.

Trace-element data
Monazite and garnet trace-element and Zr-in-rutile concentrations were acquired on the Open University Agilent 7500 quadrupole ICP-MS coupled to a New Wave Research UP213 (213 nm) Nd:YAG laser ablation system. The same monazites selected for U-Th-Pb dating were analysed in-situ in polished thin sections for REE concentrations. Laser spots for REE-analysis were sited either immediately next to, or as a larger spot engulfing, the pits formed during the U-Th-Pb analyses (see supplementary material S4.2.10 for images of pits). NIST-610 and 612 synthetic glasses were used as primary and secondary standards. Internal standardisation was to CeO for monazite and CaO for garnet using values measured by EPMA. Garnet analyses were screened for accessory phase interference (e.g. Zr for zircon and P for monazite). REE concentrations were normalised to chondritic values of McDonough and Sun (1995). Full data tables, running conditions and reproducibility of standards are documented in supplementary material S1.5.

Petrology and mineral chemistry
The structurally highest samples (343, from the Takdah section, 67, from the Mangan section, and 209, from the Rongli section), are Greater Himalayan Al 2 SiO 5 -bearing gneisses/migmatites. Sample 343 is described in detail here; other samples reached similar peak conditions and are described in supplementary material S2. Sample 343 contains garnet + kyanite + sillimanite + biotite + plagioclase + K-feldspar + quartz + rutile + apatite + monazite + xenotime + zircon. Garnets are unzoned in major elements (Fig. 2a), contain inclusions of monazite and yield an biotite, plagioclase and sillimanite that commonly surround garnet are interpreted as the localised products of a back-reaction between melt and garnet (Fig. 3a). The main population of biotite has a composition of X Mg = 0.43 and displays 'lath-shaped' domains of intergrown biotite and quartz (supplementary material S2.3). These features are all indicative that these phases coexisted with melt (Groppo et al., 2010;Waters, 2001).
The structurally lowest LHS samples (16, Takdah; 97, Mangan; and 24, Rongli) are (garnet-staurolite) mica schists. Sample 16 is described in detail but the other samples reached similar peak conditions and are described in detail in supplementary material S2. Sample 16 is a garnet-staurolite schist containing garnet + staurolite + biotite + muscovite + quartz + ilmenite + monazite + apatite + allanite + zircon. The sample has a welldeveloped schistosity forming an undulating fabric, with deformed quartz lenses (Fig. 3d). Garnets contain syn-kinematic inclusion trails of allanite, monazite and ilmenite, and are zoned in major elements ( Fig. 2g) with core compositions of X Ca = 0.06, X Mg = 0.09, X Fe = 0.78 and X Mn = 0.07 and rim compositions of X Ca = 0.02, X Mg = 0.11, X Fe = 0.85 and X Mn = 0.02. Biotite has an X Mg value of 0.44. The main mica-rich foliation wraps the garnets, and pressure shadows are well-developed. Pseudosection for sample 343, peak field is shaded in green (biotite, melt, plagioclase, K-feldspar, garnet, kyanite and rutile at 790 • C and 10 kbar). (B) Pseudosection for sample 22 has a peak field shaded in blue (biotite, staurolite, plagioclase, garnet, kyanite and rutile) at 655 • C and 9 kbar. (C) Pseudosection for sample 16, peak field is shown in red (biotite, staurolite, garnet and ilmenite) at ∼580 • C and 8.5 kbar. (D) Pseudosection for sample 60, drawn for a bulk composition representing the garnet core assemblage, this prograde assemblage is shaded in blue (chlorite, staurolite, garnet and ilmenite) at ∼550 • C and 9 kbar. The peak metamorphic assemblage for this sample was calculated in a separate pseudosection (supplementary material S3.4.4.4) with the bulk composition for the garnet rim and is shown as the dashed blue field (biotite, staurolite, garnet and kyanite). This is equivalent to the same field for sample 22 and supports the suggestion that the upper LHS samples had a similar prograde and peak metamorphic history. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

P-T evolution
Calculated pseudosections are shown in detail in Fig. 4 and summarised in Fig. 5. Further detail is provided in supplementary material S3.4. The pseudosection results are supported by average P -T , Ti-in-biotite and Zr-in-rutile calculations (Fig. 5, and supple-mentary data S3.1-3.3), which demonstrate that samples collected from the same structural levels in the three transects experienced similar P -T conditions. The GHS rocks equilibrated at peak conditions around the kyanite-sillimanite transition in the presence of melt. This is shown in the pseudosection of sample 343, which shows that the peak assemblage of biotite, garnet, K-feldspar, kyanite, melt, plagioclase and rutile was stable at ∼790 • C and ∼ 10 kbar, above the solidus (Fig. 4). Sample 67, from the Mangan section, equilibrated at similar conditions (∼800 • C and ∼9.5 kbar) represented by the stability field of biotite, melt, garnet, plagioclase, K-feldspar, sillimanite and quartz (supplementary material S3.4.4.2). The presence of spinel (hercynite composition) in a garnet gneiss collected at the same locality as 67 and comparable in lithology (sample 72 in supplementary data S2), suggests that the rocks decompressed isothermally to relatively low pressures, where spinel formed (Baldwin et al., 2007).
The prograde history of the upper LHS rocks is calculated from sample 60, in which garnet preserves significant prograde growth zoning (Fig. 2e). The pseudosection calculated for the garnet core composition (Fig. 4), demonstrates that garnet equilibrated at conditions of ∼550 • C and ∼9 kbar. The garnet rim grew in association with biotite, staurolite, kyanite, garnet, quartz and white mica at ∼675 • C and ∼9 kbar (supplementary material S3.4.4.4).
Sample 22, with the peak assemblage biotite, staurolite, plagioclase, garnet, kyanite and rutile, equilibrated at similar conditions of ∼655 • C and ∼9 kbar (Fig. 4). The peak estimates are supported by Zr-in-rutile thermometer calculations, which yield temperatures of 675 ± 9 • C (supplementary material S3.3). The retrograde history of the upper LHS rocks is constrained by the appearance of fibrolite needles in samples 22 and 30, evidence that samples passed through the sillimanite field during exhumation.
The lower LHS samples record a tight P -T loop as evidenced by sample 16. Despite significant garnet zoning in this sample (Fig. 2g), as the modal abundance of garnet is only ∼2.5%, garnet growth probably caused limited fractionation of the bulk composition. The average garnet composition was therefore used to represent the composition of the peak assemblage of garnet, staurolite, biotite and ilmenite, stable at ∼580 • C and ∼8.5 kbar (Fig. 4).

U-Th-Pb monazite ages
Whereas the P -T data in this study indicate that the metamorphic history of samples from the same structural level across the MCT was similar, monazite ages yielded from the samples around the MCT exposure are different ( Fig. 6; supplementary material S4 and supplementary data table).

Monazite geochemistry
The monazite and garnet trace-element data for samples from the Takdah transect are summarised in Fig. 8  in contrast to the flat profiles recorded for the major elements (Fig. 7d). The garnet cores are relatively enriched in Y and HREE and depleted in MREE (∼600-900 ppm Y, Dy N /Yb N = 1.2) in comparison with the rims (∼200-300 ppm Y, Dy N /Yb N = 3.8, Fig. 8).
Monazites in lower LHS sample 16 contain <6000 ppm Y (Fig. 2h), display a small Eu anomaly of ∼0.5 and a Dy N /Yb N ratio ranging from 16 to 171. Garnets show bell-shaped Y zoning and concentrations of up to ∼2000 ppm.

Linking monazite age to metamorphic stage
Monazite geochronology commonly yields complex datasets where a continuous spread of ages is seen (e.g. sample 22 in this study). In the past, these data types of dataset have been presented as probability density plots (e.g. Catlos et al., 2001;Corrie and Kohn, 2008;Rubatto et al., 2013), or have been interpreted as two end-member populations with analytical mixing in between (e.g. Dahl et al., 2005). Both interpretations provide limited insight into the causes of the yielded spread, and the interpretation of individual ages as documenting 'events' in geological time. The successful interpretation of metamorphic monazite ages requires the determination of the reactions which produced the monazite and the P -T conditions at which these reactions took place. Insight into both is provided by analysis of changes in diagnostic trace-element concentrations and ratios in monazite and co-crystallising major phases. The approach is outlined here for samples from the Takdah transect in detail; similar interpretations can be made for comparable samples from the other transects (supplementary material S5).
Overall the dataset suggests that the GHS rocks experienced a prolonged metamorphic history during which monazite grew as a product of several reactions. Monazite grains yield progressively younger ages, and record crystallisation over successively shorter timescales, at lower structural levels (Fig. 9).
GHS sample 343 yields monazites that grew over an extensive period of time, from ∼37 to 16 Ma. The earliest reliable monazite population is represented by cores of matrix grains that yield an age of 26.2 ± 0.8 Ma (MSWD = 2.9). These cores are depleted in HREE and Y and have a minor Eu anomaly (Fig. 8a). Monazites included in garnet and matrix monazite rims, however, yield younger ages of 22.7 ± 0.8 Ma (MSWD = 3) and 19.5 ± 0.4 Ma (MSWD = 1.7). Grains yielding these latter populations have a more pronounced Eu anomaly and show greater enrichment in HREE and Y (Fig. 8a, d). As Eu is preferentially incorporated into feldspars, particularly K-feldspar, during melt crystallisation (Buick et al., 2010;Fig. 8b), we interpret the monazite included in garnet (interpreted as a peritectic phase) and the rims of the matrix grains to constrain the time at which the rock experienced suprasolidus conditions (∼790 • C and >10 kbar) to between ∼23 and 19.5 Ma. Two further monazite analyses suggest continued monazite crystallisation until ∼16 Ma.
Monazites in upper LHS sample 22 record ages from 20.8 ± 0.2 Ma (MSWD = 1.2) to 15.6 ± 0.2 Ma (MSWD = 2.1), indicating a prolonged metamorphic growth history. Allanite and xenotime inclusions in the HREE-enriched garnet cores indicate these phases were stable during prograde metamorphism. Their absence in the current matrix assemblage suggests they became unstable during garnet growth at higher grades (Fitzsimons et al., 2005;Janots et al., 2007;Smith and Barreiro, 1990). We therefore interpret the monazite cores, which are depleted in HREE, to have formed from the breakdown of allanite and xenotime at 20.8 ± 0.2 Ma (MSWD = 1.2), on the prograde path.
Monazite rims, in contrast, are enriched in HREE and Y. When normalised to garnet rim REE abundances, the monazite rims show trends that closely match experimental (supra-solidus) monazitegarnet partitioning data (Fig. 8c, Buick et al., 2006;Hermann and Rubatto, 2003;Rubatto et al., 2006). This suggests that the monazite rims grew in equilibrium with the garnet rims, and that the available HREE and Y were preferentially incorporated into monazite. Additionally, the close match between natural and experimental data suggests that equilibrium partitioning of HREE and Y between garnet and monazite is similar under both sub-solidus and supra-solidus conditions. The relationship between the HREE-Y concentrations in the two minerals may therefore be used to test for equilibrium.
Additional evidence for equilibrium between monazite rims and garnet rims comes from textural relationships. Monazite is in-cluded in garnet rims (grains were too small to analyse by laser ablation) and matrix grains are also found in close proximity to garnet. The matrix monazite rims yield an age of 15.6 ± 0.2 Ma (MSWD = 2.1). This age probably constrains the time of garnet rim growth during peak metamorphic conditions of ∼650 • C and 8-10 kbar. Rutile inclusions in the garnet rim, yield Zr-in-rutile temperatures of 674.8 ± 9.4 • C, supporting the P -T conditions determined from the pseudosection. The MREE enrichment in the garnet rim (Fig. 8a) suggests the breakdown of a MREE-enriched phase such as apatite or xenotime during garnet rim growth where the release of P may have assisted the growth of the monazite rims (Pyle and Spear, 2003;Simpson et al., 2000).
In many studies that document Himalayan monazite geochronology (e.g. Kohn et al., 2001;McQuarrie et al., 2014) the formation of high-Y monazite rims are linked to melt crystallisation or to garnet breakdown. Our data suggests that although high Y rims can be linked to melt crystallisation (for example matrix monazite rims in sample 343), high Y rims are also present in samples which equilibrated under subsolidus conditions. Furthermore, the garnetmonazite partitioning data suggests that the high Y monazite rims in sample 22 could have been growing in equilibrium with garnet rims, in association with the breakdown of another accessory phase such as apatite or xenotime, and therefore not during garnet breakdown. This demonstrates that although the assumption that high Y monazite forms during the breakdown of garnet may be valid for some samples, the trace-element budget of the rock is governed by complex relationships between the major and accessory phases, which can only be revealed by detailed use of petrochronology and petrological analysis.

Metamorphic evolution of the MCT zone
The detailed petrochronological dataset presented here elucidates a specific and complex history of burial, thrusting and exhumation. Lowermost GHS samples 343 and 67 preserve evidence of an early (subsolidus) prograde history between ∼37 and 26 Ma.
Petrographic and trace-element chemistry evidence from sample 343 suggests that garnet and monazite crystallised coevally in the kyanite field and in the presence of melt between ∼22.7 and 19.5 Ma. The rocks were then decompressed, and/or heated into the sillimanite field where samples 67 and 209 equilibrated (at ∼17-14 Ma). Evidence of garnet/kyanite breakdown to hercynite in sample 72, adjacent to sample 67, suggests that the rocks decompressed into (and out of) the spinel field on the retrograde path.
At the same time the upper LHS rocks (initially in the footwall of the MCT) were buried and heated. This shows that although the GHS has an early history of deformation, temporally unrelated to the underlying footwall rocks, there is a later shared history of the two units. The early, prograde, monazite growth in all upper LHS samples overlaps with the final stages of monazite growth in the GHS samples; in the Takdah section, partial melting at peak conditions in the GHS coincided with prograde metamorphic conditions in the upper LHS samples at around 21-19.5 Ma. The upper LHS samples record protracted monazite growth over several million years. Monazite probably grew continuously: multiple resolvable populations of monazite are preserved within different index minerals in sample 60, demonstrating continuously evolving mineralogy from ∼14.6 to 11.8 Ma at moderately high P -T conditions ( Fig. 9).
Monazite in the lower LHS samples began to grow during the last stages of monazite growth in the upper LHS samples (e.g. at 16 Ma in sample 16) and continued to crystallise until ∼9 Ma in the northern-most samples.
This study therefore documents the diachronous attainment of peak metamorphic conditions through the inverted metamorphic sequence, with the timing, duration and grade of metamorphism decreasing with increasing structural depth through the thrust zone. Ages determined from samples from different exposures through the domed Sikkim MCT thrust zone also vary from north to south: spanning ∼14.5-9 Ma in the northern (rear-ward) section and ∼21-16 Ma in the southern (leading edge) section. Our dataset thus covers a >10 Ma 'time window' into the ductile thrusting history of the MCT.
Other studies have recorded monazite ages decreasing with depth below major thrust zones, both in the MCT zone, where monazite ages generally span between ∼21 and 11 Ma throughout the central to eastern Himalaya (Bollinger and Janots, 2006;Catlos et al., 2004Catlos et al., , 2001Harrison et al., 1997;Kohn et al., 2001;Larson and Cottle, 2014;Larson et al., 2013;McQuarrie et al., 2014), and the Canadian Cordillera (Crowley and Parrish, 1999;Gibson et al., 1999). Monazite ages collected from this study are therefore in agreement with the age range yielded from previous studies. However recent analytical advances that allow higher precision data from smaller volumes of material and developments in understanding monazite petrochronology have provided new insight into the duration of ductile thrusting processes via different 'time windows' exposed in the Sikkim region.
Monazite grains yielding ages between 9 and 3 Ma reported from the MCT zone in the Nepal Himalaya (Bollinger and Janots, 2006;Harrison et al., 1997) have been cited as evidence for outof-sequence thrusting along the MCT. In contrast to these previous studies, the comprehensive monazite dataset presented here implies that prolonged monazite growth was the product of several different reactions in multiple MCT transects, suggesting a model of continuous thrusting throughout the Miocene. The apparent lack of young (<9 Ma) monazites from the Sikkim Himalaya could be explained by differences in erosion horizon along the Himalaya, as demonstrated by the divergent ages from the leading and rear edge of the thrust zone presented in this study.
The P -T evolution of samples from different structural levels presented in this study are also broadly in accord with previous estimates of the P -T conditions from the MCT zone, both in Sikkim (Dasgupta et al., 2009(Dasgupta et al., , 2004Dubey et al., 2005;Rubatto et al., 2013;Sorcar et al., 2014), and from similar structural levels elsewhere in the Himalaya (Caddick et al., 2007;Goscombe and Hand, 2000;Groppo et al., 2009;Kohn et al., 2001;Searle and Rex, 1989;Stephenson et al., 2000;Vannay and Grasemann, 1998;Yakymchuk and Godin, 2012). The constant pressures recorded throughout the thrust zone in the Sikkim Himalaya suggest that Indian crustal material was continuously buried to similar depths before exhumation. However the uncertainties on the pressure calculations, both in pseudosection and average P -T analysis, make accurate interpretations difficult.

Model for inverted metamorphism development in the Sikkim Himalaya
The data from the Sikkim Himalaya point to a model where motion along the MCT caused successive burial and accretion of material from the footwall into the hanging wall. Accreted material was then transported southwards and upwards to progressively shallower levels (Fig. 10). This model is supported by evidence of tectonic interleaving of GHS and LHS protolith material in the Sikkim MCT zone (Mottram et al., 2014). We suggest that the thrust originally nucleated at the LHS-GHS protolith boundary. The plane of deformation migrated over time away from this 'parent thrust' down into its footwall over a period of ∼5 Ma in each transect. The inverted metamorphic zone within the Sikkim Himalaya therefore records a series of 'paleothrusts', with each successive level representing a snapshot of MCT thrusting through time as material was accreted into the thrust zone. Accretion of material from the footwall into the hanging wall, decreasing peak metamorphic temperatures and progressively younger attainment of peak conditions are consequences of this model.
Our study provides petrochronological context to support previous models, based on more limited datasets, which are used to explain how material is accreted into thrust zones (e.g. Barr et al., 1986;Bollinger et al., 2004Catlos et al., 2001;Harrison et al., 1998Harrison et al., , 1997Kohn et al., 2001). A similar accretionmechanism to that found in the Sikkim Himalaya, is also predicted by both the channel flow and wedge-type Himalayan tectonic models (e.g. Beaumont et al., 2001;Jamieson et al., 2004;Kohn, 2008).
Heat sources for the development of inverted metamorphism have been widely discussed in the Himalayan literature (e.g. Burg et al., 1984;Duprat-Oualid et al., 2013;Harrison et al., 1998;Hubbard, 1996;Johnson and Strachan, 2006;Le Fort, 1975;Molnar and England, 1990;Pitra et al., 2010;Searle et al., 1999;Vannay and Grasemann, 2001). Although the dataset presented here from does not uniquely identify the heat sources for the inverted metamorphism, our data does provide some new insight into likely contributing sources.
It is firstly important to determine whether metamorphism and deformation were contemporaneous in the MCT zone. Our dataset includes ages from monazite inclusions within and associated with syn-kinematic metamorphic index minerals that are aligned and elongated along the main MCT sheared foliation (details in the supplementary material S2.3). This evidence suggests that metamorphism was intimately related, both texturally and temporally, to thrusting along the MCT.
Where hot rocks are thrust upon cold rocks (the hot iron model; Le Fort, 1975), the hanging wall is the heat source for metamorphism in the footwall. The data presented here clearly show that (decompressional) melting in the GHS hanging wall was contemporaneous with metamorphism in the upper LHS footwall, for at least the early period of thrusting. As anatexis is a consequence, rather than a cause, of heating and decompression (Harris et al., 1995), this suggests that the lower part of the GHS was at least hot enough to have melted and thus in bulk, could have provided the heat to drive the metamorphism of the footwall. In this way, heat was continuously advected to the footwall during progressive ductile thrusting. Accretion of footwall material to the hanging wall through time allowed the inverted metamorphic isotherms to be 'frozen' into the transect.
The data from the Sikkim Himalaya indicate that monazite continued to crystallise in the lower LHS footwall up to 5 Ma after the cessation of monazite crystallisation in the GHS hanging wall. It is important to note that although the hanging wall and footwall rocks are juxtaposed today, it is difficult to reconstruct their relative positions through the thrusting history (Jamieson et al., 1996). Despite this, it is possible that additional heat sources, other than a hot overriding GHS, are required in order to reproduce the temporal and thermal evolution of the LHS footwall rocks. It is therefore possible to suggest that heat advected from depth with the exhuming GHS was augmented by unusually high radiogenic heat flux generated from the K, U, Th-rich pelitic lithologies (England and Molnar, 1993;Inger and Harris, 1992;Johnson and Strachan, 2006;Ruppel and Hodges, 1994). Furthermore, the prolonged deformation and shearing that occurred during the down-cutting of the active thrust plane and the progressive accretion of the upper footwall to the hanging wall provides an additional possible heat source from shear heating (Arita, 1983;Bird, 1978;Duprat-Oualid et al., 2013;Kidder et al., 2013;Le Fort, 1975).

Conclusions
The MCT zone in the Sikkim Himalaya represents a zone of inverted metamorphism in which the age and grade of metamorphism decreases with depth within the thrust zone (by ∼5 Ma and 250 • C respectively). Doming of the thrust after motion ceased permits unique insight into differences in timing of metamorphism along the transport direction. Our data show a 4-6 Ma difference in the timing of attainment of similar metamorphic conditions between the more northerly (rear-ward) and more southerly (leading edge) exposures through the structure. The highest structural levels of the MCT zone (lowermost GHS rocks) experienced peak metamorphic conditions of ∼790 • C and 10 kbar at ∼22.5-19.5 Ma in the southern exposure of the thrust zone and between ∼30 and 20 Ma in the northern exposure. The upper LHS rocks reached peak metamorphic conditions of ∼655 • C and 9 kbar at ∼16 Ma in the south and ∼12 Ma in the north. The lower LHS samples reached peak metamorphic conditions of ∼580 • C and 8.5 kbar at ∼16 Ma in the south, and between 10 and 9 Ma in the north. The inverted metamorphic zone within the Sikkim Himalaya therefore preserves a record of >10 Ma of the MCT thrusting history. During this time, material became progressively incorporated into the hanging wall from the footwall as the thrust zone continuously cut down-section, over a period of ∼5 Ma. Thrusting is thought to have initiated from the original protolith boundary which represents the oldest active structure at the highest structural level of the thrust zone. Each slice of material that was incorporated into the thrust zone was metamorphosed during a consecutively younger metamorphic stage, by heating from the overthrust sheet, enhanced by radioactive heat production in the thrust zone and possibly by shear heating.
This study demonstrates the importance of combining spatiallyresolved geochronology, trace-element geochemistry and P -T modelling to link the monazite-forming reactions to the P -T evolution recorded in the major rock-forming minerals. It is through such detailed petrochronology that we can fully interpret the temporal evolution of the rocks and demonstrate that inverted metamorphic sequences play an important role in preserving the duration of movement in progressively deforming ductile thrust zones.