Integrated pseudosection modeling and monazite petrochronology of paragneiss from the Kanchenjunga region of northeastern Nepal reveal the presence of cryptic tectonometamorphic discontinuities within the Himalayan metamorphic core. These new data outline a series of thrust-sense structures that juxtapose rocks that generally record a protracted history of early Eocene to latest Oligocene–early Miocene (ca. 41–23 Ma) prograde metamorphism and lateral extrusion against others that typically record short prograde (between 2 and 7 m.y.) and retrograde (between 3 and 6 m.y.) histories variably spanning the middle Oligocene to the middle Miocene (ca. 31–12 Ma). Retrograde metamorphism in the hanging wall of the thrust faults is typically coeval with prograde metamorphism in the footwall, indicating that overthrusting/underthrusting accommodated crustal shortening and drove metamorphic processes. The structures and juxtaposed panels were cut by early Miocene (ca. 20–18 Ma) out-of-sequence thrusting coincident with the previously mapped High Himal thrust. The resulting kinematic model for the evolution of the Himalayan metamorphic core in the Kanchenjunga area demonstrates that the Himalayan metamorphic core was dominated by underplating from at least the Oligocene through to present, and that the internal structure of the exhumed metamorphic core is significantly more complex than has been documented previously.
The ongoing India-Asia collision has often served as a template for understanding processes in continent-continent convergent margins. Because of its utility as a type example, the models developed in the Himalaya are often applied to other orogenic systems. The relevance and transferability of these models, however, are limited by how well they actually represent the Himalayan system.
Over the past ∼20 yr, there have been a variety of models put forth to explain the geologic evolution of the Himalayan front. Perhaps the most investigated of these have been the channel flow hypothesis (e.g., Beaumont et al., 2001) and wedge taper hypothesis (e.g., DeCelles et al., 2001). Numerous studies have been dedicated to comparing new data and geologic constraints to model predictions in an effort to determine which of the two mentioned models is a better representation of the processes that have built the mountain range (e.g., Law et al., 2006, and references therein; Kohn, 2008; McQuarrie et al., 2008; Robinson, 2008; Sachan et al., 2010; Corrie et al., 2012; Larson et al., 2010, 2011, 2013; Yakymchuk and Godin, 2012; Searle, 2013; Mukherjee, 2013). Recent reviews of this type of research in the Himalaya (Beaumont and Jamieson, 2010; Larson et al., 2013; Larson and Cottle, 2014), and more generally (Jamieson and Beaumont, 2013), indicate that both processes may have been active in spatially distinct regions at different times during the evolution of the orogen.
A wealth of detailed data has been collected by numerous researchers as a result of examination of the greenschist- to granulite-grade exhumed midcrust (the Himalayan metamorphic core) along the Himalaya in an attempt to distinguish between proposed models. From these data sets, which include monazite and zircon geochronology and metamorphic pressure-temperature-time (P-T-t) paths, researchers have documented a number of tectonometamorphic discontinuities (Fig. 1; e.g., Kohn et al., 2005; Carosi et al., 2010; Rubatto et al., 2013). Until the recognition of these structures, most models for burial and exhumation have focused on the two fault systems that bound the Himalayan metamorphic core: the South Tibetan detachment system above, and the Main Central thrust below (Beaumont et al., 2001, 2004; Jamieson et al., 2004; Webb et al., 2007, 2011). Increasing chronologic and thermobarometric data from across the Himalaya, however, indicate that significant amounts of horizontal shortening and vertical thickening were accommodated along structures within the Himalayan metamorphic core (e.g., Jain and Manickavasagam, 1993; Goscombe et al., 2006; Groppo et al., 2009; Carosi et al., 2010; Imayama et al., 2010; Larson et al., 2013; Larson and Cottle, 2014; Mottram et al., 2014; Warren et al., 2014). These structures have only a cryptic surface expression and are often recognized by abrupt spatial breaks in P-T-t(-deformation [D]) paths (e.g., overall shape, timing, peak conditions). Moreover, although such P-T-t(-D) discontinuities have been identified along the length of the Himalaya (Fig. 1; Montomoli et al., 2014), little is known about how they developed and their overall importance to the kinematic evolution of the orogen.
The Kanchenjunga region of far northeastern Nepal exposes a thick section of garnet + biotite–grade to migmatitic, K-feldspar + sillimanite–grade paragneiss of the Himalayan metamorphic core (Fig. 2). Goscombe et al. (2006) and Imayama et al. (2012) mapped what they refer to as the High Himal thrust through the area. As interpreted in those studies, the High Himal thrust is a tectonometamorphic discontinuity, as described already, and as such provides an opportunity to examine, in detail, the characteristics of such a structure and develop a model for how the midcrust may have evolved. Pseudosection modeling and in situ laser-ablation–split-stream U-Th/Pb monazite petrochronology methods were applied to specimens collected to bracket the location and timing of the High Himal thrust and identify any other cryptic structures within the exposed Himalayan metamorphic core. The resulting data confirm the existence of the High Himal thrust and reveal the presence of a number of other structures not previously unidentified.
The Kanchenjunga region of northeastern Nepal (Fig. 2) is underlain by rocks of the Himalayan metamorphic core and Lesser Himalayan Sequence (Shrestha et al., 1984; Schelling, 1992; Goscombe and Hand, 2000; Goscombe et al., 2006). The Lesser Himalayan Sequence is exposed in the Tamor window and consists of Chl + Bt + Ms (abbreviations throughout are after Whitney and Evans, 2010) phyllite and quartzite (Shrestha et al., 1984; Schelling, 1992), which have been intruded by Paleoproterozoic granite (now orthogneiss; Upreti et al., 2003; Sakai et al., 2013). The Himalayan metamorphic core, which has been thrust over the Lesser Himalayan Sequence along the Main Central thrust, is a dominantly metasedimentary assemblage of paragneiss and quartzite with subordinate calc-silicate, metabasite, and orthogneiss. The Himalayan metamorphic core is intruded locally by muscovite ± tourmaline ± biotite–bearing leucogranite dikes and sills, which typically increase in abundance toward higher structural levels (to the north; Goscombe et al., 2006).
Himalayan Metamorphic Core
Rocks of the Himalayan metamorphic core form an inverted metamorphic sequence from Ms + Bt + Grt grade at the base to Kfs + Sil grade in the upper portion within the field area (Goscombe et al., 2006; Imayama et al., 2010). In the studied area, the Himalayan metamorphic core is divided into five geologic units (Fig. 2) based on lithology and metamorphic observations made during this study. The structurally lowest unit consists of mylonitic orthogneiss and Grt + Chl + Ms + Bt ± Ky ± St phyllite and schist. This unit includes what has previously been mapped as the uppermost Seti Formation, the Ulleri and Kushma Formations (Shrestha et al., 1984), and the lowermost Junbesi Paragneiss (Schelling, 1992). The base of the structurally overlying units to the southwest and north is marked by the first appearance of migmatite, which contrasts sharply with the underlying rocks, which contain low (0%–10%) melt volume. To the southwest, these rocks consist of locally migmatitic Grt + Bt + Ms ± Ky schist, gneiss, and quartzite previously mapped as the Junbesi Paragneiss (Fig. 2; Schelling, 1992), while to the north, they consist of locally migmatitic Kfs + Sil + Grt + Bt + Ms ± Ky gneiss and quartzite (Fig. 2) previously mapped as the uppermost Junbesi Paragneiss and lowermost Kanchenjunga migmatites (Schelling, 1992). Metamorphic grade continues to increase to the north through the overlying, dominantly migmatitic unit previously mapped as the Kanchenjunga Paragneiss (Schelling, 1992). This unit has the same assemblage as both the underlying and overlying units but lacks muscovite (Fig. 2; Kfs + Sil + Grt + Bt). The base of the structurally highest unit observed in the field area is marked by the return of muscovite (Kfs + Sil + Grt + Bt +Ms). This uppermost unit was previously mapped as part of the Kanchenjunga migmatites (Schelling, 1992) and Yangma Paragneiss (Goscombe et al., 2006).
Previously Mapped Structures
There is a broad zone of deformation associated with the Main Central thrust, and, as such, interpreting where it should be mapped is difficult and has typically varied with the preference of the researcher along the orogen (for a summary of problems associated with mapping the Main Central thrust, see Searle et al., 2008). Previous studies have variably mapped the Main Central thrust at the base (Goscombe et al., 2006), top (Schelling, 1992), and just above (Imayama et al., 2010) the mylonitic orthogneiss in the lowest unit of the Himalayan metamorphic core. In this study, the Main Central thrust is mapped just below the mylonitic orthogneiss (Fig. 2) for two reasons, both of which are consistent with the definition of the structure by Searle et al. (2008). First, the mapped trace marks the base of closely spaced, inverted, metamorphic isograds that overlie very low-metamorphic-grade rock. Second, the mapped trace is located between rocks above, which record Himalayan-related metamorphism, and rocks below, which were relatively unaffected; a mylonitic orthogneiss unit, just up structural section from the garnet-in isograd, records Cenozoic cooling ages, whereas Paleoproterozoic granites, near the center of the Tamor window, record Paleoproterozoic cooling ages (Sakai et al., 2013).
A second, structurally higher ductile shear zone has been mapped within the Himalayan metamorphic core throughout eastern Nepal (Brunel and Kienast, 1986; Lombardo, 1993; Goscombe and Hand, 2000; Goscombe et al., 2006). To the west of the study area, in the Makalu and Everest regions, Brunel and Kienast (1986) and Lombardo et al. (1993) mapped this structure as the Main Central thrust. In the Kanchenjunga region, Goscombe and Hand (2000) also mapped this structure as the Main Central thrust; however, following more detailed work, Goscombe et al. (2006) remapped the same structure as the High Himal thrust and located the Main Central thrust structurally lower. They interpreted early Miocene moderate-pressure, high-temperature metamorphism in the hanging wall as being consistent with ductile extrusion of the Himalayan metamorphic core between the South Tibetan detachment system and High Himal thrust. In the footwall, relatively high-pressure, moderate-temperature, mid-to-late Miocene metamorphism was attributed to brittle extrusion processes (Goscombe et al., 2006). Recent thermochronology and zircon petrochronology in the immediate footwall of the High Himal thrust in the Kanchenjunga region have been interpreted to indicate thrust activity across the High Himal thrust in the late Oligocene to early Miocene (Imayama et al., 2012).
PETROGRAPHY AND MINERAL CHEMISTRY
In total, 64 specimens were collected along an ∼100 km transect between Basantapur and Yangma in the Kanchenjunga region of northeast Nepal (Fig. 2). Fifty polished thin sections were made for investigation with a petrographic microscope and, where necessary, a scanning electron microscope to identify mineral assemblages. Eight quartzo-feldspathic paragneisses were selected for phase equilibria modeling based on mineral assemblage, texture, and structural position. Each thin section was mapped for select major and trace elements using a Cameca SX100 electron microprobe (EMP) at the Saskatchewan Research Council, to create chemical distribution maps that could be used to determine the modal abundance of phases (Table 1). Of these, three specimens, one kyanite grade (KA007) and two Sil-Kfs grade (KA044 and KA064A), contained the partial-equilibria assemblages required to constrain P-T paths.
Spot analyses of minerals with solid solutions were performed at the Earth and Planetary Materials Analysis Laboratory at the University of Western Ontario with a JEOL 8530F Hyperprobe EMP using a 15 keV accelerating voltage and a 20 nA beam current. Twenty second counting times were used for both peak and background positions. Matrix corrections were performed using the ZAF correction procedures included in the JEOL software. As garnet was the primary mineral used for isopleth calculations, detailed profiles were measured across 2–3 grains per thin section to determine compositional zoning. In addition, analyses targeted solid solution minerals in different textural settings, including inclusions in garnet, rimming garnet, and distal to garnet. Representative spot analyses for each specimen are provided in Table 2. Complete mineral analyses for each specimen are included in the GSA Data Repository.1
Specimen KA007 is a Ky + Grt + Bt + Qz + Pl gneissic schist (Table 1; Fig. 3) collected near the southern end of the study area, ∼3700 m structurally above the Main Central thrust (Fig. 2). The specimen consists of garnet and kyanite porphyroblasts within a finer-grained matrix of quartz, plagioclase, and biotite (Fig. 4; Table 1). Accessory phases include rutile, monazite, zircon, apatite, and xenotime. At the outcrop scale, the rock is homogeneous, aside from minor, thin, concordant leucogranite segregations, which were not sampled. A foliation (142°/25°) is defined by weakly aligned biotite grains.
Garnet occurs as subhedral to anhedral porphyroblasts up to 1.7 mm in diameter (Fig. 4A). Inclusions of rutile and quartz in garnet range in abundance from rare to common. Profiles from two garnet grains were measured in specimen KA007. The larger garnet grain preserves compositional zoning; in the smaller grain, spessartine (0.07), grossular (0.035), and Fe# [0.84; Fe#(molar) = Fe2+/(Fe2+ + Mg2+)] are fairly homogeneous. In the larger grain (Fig. 5A), Fe# increases gradually from the core (0.83) to just inside of the rim (0.85), and sharply at the outermost rim (0.87); spessartine increases from the core (0.06) to the rim (0.095); and grossular decreases from the core (0.045) to the rim (0.025). These patterns are typical of diffusional homogenization at high temperature that has altered prograde growth zoning and resulted in profiles reflecting retrograde diffusion (Spear, 1991; Caddick et al., 2010). Additionally, increasing spessartine toward the rim is characteristic of garnet resorption; because Mn is not easily exchanged with other phases, it becomes concentrated in garnet rims during resorption (Woodsworth, 1977).
Kyanite occurs as nonaligned porphyroblasts up to 1.8 mm in length (Fig. 4A). Kyanite grains, which are commonly strongly embayed and skeletal, contain inclusions of, and are typically surrounded by, quartz (Figs. 4A and 4B). The matrix is dominated by plagioclase and quartz, while biotite occurs as both weakly aligned, fine-grained flakes within the matrix and fine-grained flakes replacing garnet (Fig. 4A), consistent with back reactions during melt crystallization (Le Breton and Thompson, 1988; Vielzeuf and Holloway, 1988; Kriegsman, 2001). The Ti concentration of biotite increases with decreasing Mg# (Mg# = 1 – Fe#) from Mg# = 0.41 and Ti = 0.35 apfu to Mg# = 0.34 and Ti = 0.48 apfu. Matrix biotite typically has higher Ti and a lower Mg# than biotite that has replaced garnet. Since Ti concentration and temperature typically exhibit a positive relationship (Henry and Guidotti, 2002; Tajčmanová et al., 2009), higher Ti concentrations in matrix biotite indicate that it grew at higher temperatures than the biotite surrounding garnet, consistent with garnet replacement during melt crystallization. The composition of plagioclase is homogeneous throughout (Ab76–80). Minor muscovite is present along biotite grain boundaries (Fig. 4C) and less commonly as inclusions in plagioclase.
Specimen KA044 is a Sil + Grt + Pl + Kfs + Qz + Bt gneiss (Table 1; Fig. 3) collected ∼3 km north of the village of Sekathum and ∼7050 m structurally above the Main Central thrust (Fig. 2). A well-developed migmatitic foliation (300°/36°) is defined by thin alternating melanocratic and leucocratic layers, while a stretching lineation (32° → 042°) is marked by biotite and quartz grain-shape and sillimanite alignment. The melanocratic layers are dominantly biotite and graphite, while the leucocratic layers are dominantly quartz, K-feldspar, plagioclase, and sillimanite. Garnet grains occur within both leucocratic and melanocratic layers. Accessory phases include ilmenite, rutile, zircon, monazite, graphite, and apatite.
Garnet occurs as subhedral to anhedral porphyroblasts that range in diameter from less than 0.5 mm up to 4 mm with a typical size of ∼3 mm (Figs. 6A and 6B). Garnet grains contain abundant inclusions of quartz, rutile, and less commonly biotite, ilmenite, and kyanite, which define an internal foliation that is not continuous with the matrix (Fig. 6B). Some garnet grains preserve inclusion-free rims along faces that are perpendicular to the foliation (Fig. 6B). Profiles from three garnet grains were measured in specimen KA044. The largest garnet (Fig. 6B) best preserves Fe-Mg zoning, with Fe# decreasing slightly from the core (0.78) through the mantle (0.77) and increasing sharply at the rim (0.84). In the same garnet, grossular is homogeneous in the core (0.11) and decreases moderately through the mantle and sharply at the rim (0.05). In the other two grains, grossular increases slightly from the core (0.05) to the mantle (0.06–0.09) before decreasing at the rim (0.05). Spessartine from each profile appears fairly homogeneous (0.01–0.02) aside from a minor increase at the rim. As with KA007, the composition has likely been modified by diffusion at high temperature. However, unlike KA007, while the mantle and rim profiles are typical of retrograde diffusion, the profile of the core appears to preserve prograde growth zoning.
Biotite occurs as aligned fine-grained flakes within the matrix, as fine-grained flakes replacing garnet, as inclusions within sillimanite, and as intergrowths with sillimanite (Figs. 6A, 6B, and 6D). Biotite is typically absent in garnet strain shadows. Biotite grains included in garnet have relatively low Ti (0.0–0.25 apfu) and high Mg# (0.51–0.67). Biotite in contact with and replacing garnet has intermediate Ti (0.17–0.25 apfu) and Mg# (0.46–0.56) values. Matrix biotite has the highest Ti content (0.25–0.29 apfu) and lowest Mg# (0.47–0.49). Although the absolute compositions have likely been modified by retrograde diffusion, the Ti concentrations indicate that matrix biotite crystallized at higher temperatures than biotite, which has replaced garnet, while biotite included in garnet grew at the lowest temperatures (Henry and Guidotti, 2002).
K-feldspar occurs as fine grains in the matrix and as coarser aggregates within garnet strain shadows (Figs. 6A and 6B). K-feldspar also occurs with plagioclase as two-feldspar aggregates interpreted to be the breakdown product of high-temperature ternary feldspar (Fig. 6C). The composition of plagioclase ranges between Ab58 and Ab64, whereas K-feldspar ranges between Or77 and Or89. The inclusion-free garnet rims adjacent to strain shadows, increased abundance of K-feldspar, and lack of biotite indicate that the strain shadows preserve a higher-grade assemblage that has been overprinted elsewhere in the thin section. Moreover, this apparent overprinting implies that much of the biotite in the matrix could be the result of retrograde breakdown of K-feldspar through a reaction such as L + Grt + Kfs = Bi + Als + Pl + Qz. Sillimanite occurs as aggregates of prismatic crystals that are associated with, and interpreted to have partially replaced, biotite (Fig. 6D), and as individual needles within quartz. Kyanite occurs as inclusions within garnet and as small anhedral grains within the matrix (Figs. 6A and 6B).
Specimen KA064A is a Ky + Grt + Sil + Kfs + Pl + Bt + Qz gneiss (Table 1; Fig. 3) collected above the village of Yangma, ∼9600 m structurally above the previous specimen (Fig. 2). Monazite, zircon, rutile, and ilmenite occur as accessory phases. At the outcrop scale, in addition to the sampled lithology, there are intercalated quartzites and large (1 m thick by 2 m long), concordant, coarse-grained, Tur + Ms + Fsp leucogranite lenses. Within the gneiss, the alignment of biotite and sillimanite and thin, alternating leucocratic and melanocratic layers (Fig. 3) define a foliation (306°/26°). Melanocratic layers consist mainly of biotite, while leucosome consists mainly of quartz, plagioclase, and K-feldspar.
Garnet porphyroblasts occur within a matrix of quartz, biotite, plagioclase, K-feldspar, and sillimanite. Garnet grains are subhedral to anhedral and range in diameter from less than 0.5 mm to over 3 mm with a typical size around 2 mm (Fig. 6E). They contain inclusions that are dominantly quartz and less commonly biotite and rutile. The density of inclusions in different garnet grains is variable, but rims are typically inclusion free. Some grains preserve an inclusion-free core, poikiloblastic mantle, and inclusion-free rim. Profiles measured from four garnet grains in specimen KA064A show similar trends but different absolute concentrations. Profiles from each garnet are available in the GSA Data repository material (see footnote 1), but only the largest grain, which is pervasively replaced by biotite along one face (Fig. 5C), is described here. In this grain, the Fe# increases slightly from the core (0.84) through the mantle (0.85), and sharply at the outermost rim (0.88–0.90), while spessartine is homogeneous in the core (0.05), increases slightly through the mantle (0.06) and sharply at the rim (0.08). Grossular is homogeneous through the core and mantle (0.02–0.04) but appears to decrease slightly at the rim. As with the previous specimens, these profiles are indicative of homogenization at high temperature and destruction of original prograde growth zoning. The trends through the mantle and rim are typical of retrograde diffusion and resorption.
Biotite occurs as well-aligned, large flakes and aggregates within the matrix (Figs. 6E and 6G), replaces garnet along strain caps (Fig. 6E), is found as inclusions in garnet (Fig. 6E), and is symplectically intergrown with quartz (Fig. 6F). Unlike the previous specimens, there is no clear relationship between Ti and Mg# in biotite (supplementary data [see footnote 1]). Different textural settings do, however, show a relationship with Ti. Biotite located in the matrix typically has the highest Ti content (0.38–0.44 apfu) and Mg# values between 0.34 and 0.35. Biotite included in and rimming garnet has a similar, relatively large range in Ti (0.01–0.4 apfu) and Mg# (0.33–0.39). The Ti content indicates that biotite in the matrix equilibrated at higher temperatures than biotite included in and rimming garnet.
Sillimanite occurs as fibrolitic pods and prismatic crystals that are typically interlayered with and have partially replaced biotite, and as needles within quartz (Fig. 6E and 6F), while kyanite occurs as skeletal grains (Fig. 6G). Plagioclase occurs throughout the matrix and as thin films that appear to have replaced garnet (Fig. 6E). K-feldspar is typically perthitic and occurs throughout the matrix associated with quartz and plagioclase (Figs. 6G and 6I). K-feldspar is located in garnet strain shadows, but it is not found in strain caps. The composition of plagioclase in contact with garnet (Ab26–32) overlaps with, but is generally less albitic than, plagioclase in the matrix (Ab22–31). The composition of K-feldspar located in the matrix ranges from Or72 to Or84.
Textural evidence indicates that a significant amount of the biotite crystallized during retrograde reactions. As with KA044, the preservation of K-feldspar in strain shadows and the replacement of garnet by biotite and plagioclase are indicative of retrograde melt crystallization. The intricate association of biotite and sillimanite (Fig. 6F) is consistent with melt crystallization in the sillimanite field. Moreover, Waters (2001) interpreted symplectic intergrowths of biotite and quartz (Fig. 6I) to be the result of subsolidus, retrograde replacement of K-feldspar by biotite and quartz. There is evidence, however, that prograde biotite remains in this specimen. Thin films of quartz between K-feldspar and corroded biotite (Fig. 6H) are interpreted as pseudomorphs of residual melt (Sawyer, 2008) that formed through the incomplete melting of biotite through a simplified reaction such as Bt + Qz + Ky/Sil = Kfs + L + Grt.
PHASE EQUILIBRIA MODELING
Phase diagrams were calculated using Theriak/Domino (de Capitani and Brown, 1987; de Capitani and Petrakakis, 2010) in the system NCKFMASHTO with data set55 of Holland and Powell (1998 [updated 2003]). Manganese is the only major element not used, as activity composition models for it are not reliable at the pressures and temperatures experienced by these specimens (White et al., 2007; Powell and Holland, 2010). Moreover, manganese is believed to only affect the stability of garnet at lower metamorphic grades (Spear and Cheney, 1989; White et al., 2007). The following activity-composition models were used: cordierite and staurolite (Holland and Powell, 1998); biotite, garnet, and melt (White et al., 2007); plagioclase and K-feldspar (Holland and Powell, 2003); ilmenite (White et al., 2000); muscovite (Coggon and Holland, 2002); and magnetite, spinel, and orthopyroxene (White et al., 2002). Water, quartz, kyanite, sillimanite, and rutile were entered as pure phases. The bulk compositions used to calculate the phase diagrams (Table 3) were measured by X-ray fluorescence (XRF; see supplementary data for detailed methodology [footnote 1]).
Phase Equilibria Modeling Results
The calculated phase diagrams all have a similar general topology (Fig. 7), which reflects the similarities in the rock types. The kyanite-sillimanite transition splits each diagram into an upper half, where kyanite is the stable aluminosilicate, and a lower half, where sillimanite is stable. The solidus, which ranges between 730 °C and 830 °C amongst the specimens, has a steeply negative slope in muscovite-absent fields and a steeply positive slope in muscovite-bearing fields. At lower pressures, the biotite-out lines parallel the solidi. Muscovite-bearing fields typically occur at high pressure and low temperature. The rutile-in and ilmenite-out lines are shown in gray, because they are dependent on the proportion of Fe3+, which, as discussed in the supplementary data (see footnote 1), is only qualitatively constrained in the specimens. As cordierite was not observed in any of the specimens, its appearance in each diagram is shown by a gray field that contains no details of other phases. Overlain on the topology are isopleths of grossular [Ca/(Ca + Mg + Fe)], almandine [Fe/(Fe + Mg + Ca)], Mg# [Mg/(Mg + Fe)] of biotite, and isomodes of garnet. The isopleths are shown only in the fields corresponding to core and rim intersections. A complete set of isopleths can be found in the GSA Data Repository (see footnote 1).
Melting and melt crystallization textures have been observed both at the macroscale (e.g., segregated leucosomes, leucogranite lenses) and microscale (e.g., evidence of muscovite and biotite dehydration melting in addition to evidence that garnet broke down during melt crystallization). As such, it is assumed that the rocks underwent peak metamorphism above their solidi, which is located at a minimum temperature of 730 °C in these specimens. At such temperatures, diffusion rates are expected to have been high enough to significantly modify original growth zoning and composition of garnet from the thermal peak (Florence and Spear, 1991; Spear and Florence, 1992; Caddick et al., 2010). Additionally, it is likely that the composition of garnet exteriors was further modified by diffusion during initial cooling (Spear, 1991). As such, temperatures approximated from garnet isopleths for peak conditions are minimum estimates. As discussed previously, although diffusion has modified the absolute values in garnet, it is anticipated that the general trends are preserved (Caddick et al., 2010).
Evidence of melt loss has also been observed at the macro- and microscales. At the outcrop scale, segregated leucosomes and melanosomes aligned with the main fabric suggest that the associated strain may have led to the extraction of melt from the paleosome (Sawyer, 2008). Experiments have shown that melt can be efficiently extracted when the melt fraction rises above 7%–10% (Rosenberg and Handy, 2005). Accordingly, 7%–10% partial melting should be seen as a minimum in the most fertile lithologies. During field work, an effort was made to sample mesosome or paleosome, in order to avoid bulk composition modification (melt extraction) problems. However, even in these rocks, the preservation of K-feldspar, which did not completely revert to muscovite, is interpreted as evidence for melt extraction. This inference is supported by phase equilibria modeling, which predicts that the solidus of the investigated specimens is located at temperatures of at least 730 °C (Fig. 7), i.e., significantly above what would be expected from nonrestitic metapelites (around 650 °C; e.g., Johnson et al., 2008), indicating that a proportion of the original H2O was lost as dissolved water in the extracted melt. As such, it is possible that the bulk compositions used to generate the pseudosections do not represent the systems prior to the last melt-loss episode, which would have likely occurred during prograde heating. Nonetheless, Indares et al. (2008) and Guilmette et al. (2011) have shown that reintegrating melt in pelitic compositions does not significantly change the position of garnet isopleths at temperatures above the solidus, which allows some restrained modeling of prepeak and peak conditions from the observed restitic bulk composition.
The observed equilibrium assemblage for KA007 is shown with red text in Figure 7A. Garnet core isopleths intersect in a muscovite-bearing, kyanite-free field at ∼765 °C and ∼11.5 kbar (Fig. 7B, point C). While muscovite present along biotite grain boundaries is interpreted as retrograde, grains present as inclusions in plagioclase are thought to be prograde relics, consistent with garnet core isopleths intersecting in a muscovite-bearing field. Garnet rim isopleths intersect near the center of the observed assemblage field at ∼740 °C and ∼9.2 kbar (point R). This is in close agreement with matrix biotite Mg# isopleths, which indicate temperatures on the order of ∼745–755 °C for the same pressure. The interpreted P-T path is a clockwise, decompressional path, with minor cooling, in the kyanite field beginning at point C (Figs. 7A and 7B). If the path entered a K-feldspar–bearing field at slightly higher T, no evidence remains. Compared to the other two specimens in this study, the occurrence of euhedral garnet faces, an overall lower degree of garnet replacement by biotite and plagioclase, and the lack of segregated leucosome are consistent with a lower degree of partial melting. This can be explained by a decompressional cooling path through point R before crossing the solidus in the kyanite field. The entire length of the path crosses decreasing garnet isomodes, which is consistent with textural evidence of garnet replacement by biotite during melt crystallization as well as spessartine zoning that indicates garnet resorption.
The observed assemblage for KA044, split in two by the rutile-in line, is shown with red text in Figure 7C. Garnet core isopleths intersect in a kyanite- and muscovite-bearing subsolidus field (point C) at ∼715 °C and ∼12 kbar (Fig. 7D). Although no muscovite is preserved, the intersection location is consistent with the presence of kyanite inclusions within garnet. Almandine isopleths for the garnet rim do not intersect in the investigated P-T range, which may be reflective of a number of possibilities. First, the measured bulk composition may not be representative of the system. This issue was minimized, however, by measuring the bulk composition from offcuts from the thin section used for EMP analyses. Second, the composition of garnet rims may have been altered by retrograde diffusion. Spear (1991) showed that, depending on diameter, cooling rate, and ratio of garnet to biotite, diffusion can impact garnet rim compositions to temperatures as low as 550 °C. As the diffusion of Ca in garnet is orders of magnitude slower than Fe and Mg (Vielzeuf et al., 2007), and reaction rates are considerably slower below the solidus (White and Powell, 2002), the intersection of the grossular isopleths with the solidus (point C) is taken as the P-T conditions at which the rim last equilibrated. The interpreted prograde P-T path is a heating path with minor burial in the kyanite field (Figs. 7C and 7D). Such a trajectory across decreasing almandine and grossular isopleths is consistent with the measured garnet profiles (Fig. 5B). The P-T path crosses into a biotite-free field before crystallization of retrograde biotite, consistent with observed biotite textures. The path crosses the kyanite-sillimanite transition in the presence of biotite, as indicated by biotite inclusions in sillimanite, and sillimanite that terminates along biotite faces. This indicates decompression with minor cooling, at which point the solidus is intersected. A retrograde trajectory across decreasing garnet isomodes is consistent with textural evidence of garnet replacement by biotite.
The observed assemblage for KA064A, split in two by the rutile-in line, is shown with red text in Figure 7E. Garnet core isopleths intersect in the observed assemblage at ∼8.5 kbar and ∼780 °C (point C, Fig. 7D). Garnet rim isopleths intersect in the same field, near the solidus, at ∼6.5 kbar and ∼760 °C (point R). This is similar to temperatures of ∼775 °C estimated by biotite Mg# isopleths for the same pressure. The interpreted P-T path begins in the kyanite field, consistent with kyanite located in the matrix and as inclusions in garnet. Weak compositional zonation of garnet, consistent with homogenization at high temperatures, makes deciphering the prograde path difficult. The maximum grossular content measured in all garnet analyses (0.042), however, indicates that the maximum pressure reached by this specimen was only ∼8 kbar. This is consistent with a near-isobaric heating path from the kyanite field to the intersection of core isopleths (point C). The temperatures calculated from garnet cores, as discussed earlier, are minimum estimates. As such, the thermal peak of this specimen is likely higher than determined by almandine isopleths. Textural evidence of some prograde biotite (Fig. 6H), however, indicates it was stable at the thermal peak and hence that the P-T path did not cross the biotite-out line (∼820 °C). The resulting portion of the interpreted P-T path allows for various combinations of heating, cooling, and decompression in an otherwise overall clockwise path (Figs. 7C and 7D). Decreasing garnet isomodes during cooling is consistent, furthermore, with textural evidence of biotite having replaced garnet during melt crystallization.
To constrain the timing of metamorphism in the Himalayan metamorphic core, monazite grains in eight specimens (Fig. 2; including the three specimens used for phase equilibria modeling) were analyzed in situ by laser-ablation–split-stream (LASS) inductively coupled plasma–mass spectroscopy (ICP-MS) following the methods of Cottle et al. (2013) and Kylander-Clark et al. (2013). A full description of the methods used can be found in the GSA Data Repository (see footnote 1).
The full data set from all analyses (including omissions) and concordia plots are included in the GSA Data Repository (see footnote 1). The eight specimens analyzed are discussed next in order from south to north (Fig. 2). With the exception of KA007, for which the structural position is uncertain, the order discussed also corresponds to a progression from low to high structural level. All quoted dates were calculated from 238Th/208Pb ratios (for further discussion, see GSA Data Repository [see footnote 1]). In each specimen, Y concentrations and Gd/Yb values are typically inversely proportional. As such, while both Y and Gd/Yb values are presented in Figure 8, only Y is discussed in this section for brevity. A detailed examination of the usefulness of these elements follows with the interpretation.
Specimen KA007 is a Ky + Grt + Bt + Qz + Pl gneissic schist (Fig. 3) collected near the southern end of the study area (Fig. 2). It is the only specimen analyzed for monazite petrochronology south of the Tamor window. Monazite grains in this specimen occur solely within the matrix, are 30–60 µm across in thin section, are anhedral to subhedral and equant to elongate in shape, and are commonly spatially associated with quartz, plagioclase, and biotite (for monazite location in full thin-section element maps of all specimens analyzed, see GSA Data Repository [see footnote 1]). Element grain maps show irregular zoning of U and minor to no zoning of Th, while Y is typically low in core regions and high near rims (for monazite element maps of representative grains analyzed, see GSA Data Repository [see footnote 1]). Forty-four analyses on 10 grains yielded dates ranging from 30.7 ± 0.7 Ma to 18.6 ± 0.4 Ma. There is an overall negative relationship between date and Y concentrations in this specimen (Fig. 8A). From ca. 31 Ma to 24 Ma, Y concentrations range between 13,600 and 64,000 ppm but show no well-defined trend, while dates from ca. 24 Ma to 19 Ma have associated Y concentrations ranging between ∼34,600 and 170,000 ppm and increase sharply with decreasing date.
Specimen KA031B, a Sil + Ky + Grt + Bt + Qz + Pl + Ms gneiss (Fig. 3), is the structurally lowest specimen north of the Tamor window (Fig. 2). It is located ∼1550 m structurally above the Main Central thrust. Monazite grains in this specimen occur both in the matrix and as inclusions in garnet. Grains that occur in the matrix are anhedral to subhedral and equant to elongate in shape, 30–120 µm in size, and commonly spatially associated with muscovite. The three grains analyzed that occur as inclusions within garnet are anhedral and range from 25 to 50 µm in size. Element maps for all grains show irregular zoning in Th, U, and Y, although Y is typically present in lower concentrations in core regions and higher concentrations toward the rim. In total, 39 analyses on seven grains located in the matrix yielded dates spanning from 23.1 ± 0.5 Ma to 13.7 ± 0.3 Ma, while six analyses on three grains included within garnet yielded dates ranging from 24.1 ± 0.5 Ma to 18.4 ± 0.5 Ma. There is a negative relationship between dates and Y concentrations in this specimen for both matrix and included grains (Fig. 8B). From ca. 24 Ma to 20 Ma, Y concentrations range between ∼330 and 2800 ppm and show a slight increase with decreasing date, whereas from ca. 20 Ma to 14 Ma, concentrations are higher (∼3600−21,000 ppm) and increase sharply with decreasing date.
Specimen KA034 is a Bt + Grt + Fsp + Qz gneiss (Fig. 3) located ∼2400 m structurally above KA031B (Fig. 2). Monazite grains in this specimen occur solely within the matrix, are anhedral to subhedral and equant to elongate in shape, are 50–120 µm in size, and are typically associated with quartz and less commonly biotite and plagioclase. Element maps demonstrate weak irregular zoning of Th and U, low-Y grain cores, and thin rims of high-Y concentrations. In total, 46 analyses on nine grains yielded dates ranging from 19.2 ± 0.5 Ma to 12.8 ± 0.4 Ma characterized by a negative relationship between dates and Y concentrations. From ca. 19 Ma to 16 Ma, Y concentrations range between ∼1500 and 4400 ppm and show a slight increase with younger dates (Fig. 8C). From ca. 16 Ma to 13 Ma, Y concentrations range between ∼4400 and ∼22,000 ppm and increase sharply with younger dates.
Specimen KA037 is a Grt + Ms + Bi + Fsp + Qz gneiss (Fig. 3) located ∼1750 m structurally above KA034 (Fig. 2). Monazite grains in this specimen occur solely within the matrix, range from to 400 µm in size, are anhedral to subhedral and equant to elongate in shape, and are commonly associated with biotite, quartz, muscovite, and plagioclase. Elemental maps show irregular zoning in Th and U, and, as in previous specimens described, Y is typically at lower concentrations in grain cores and higher concentrations toward grain rims. In total, 52 analyses on nine grains yielded dates spanning from 20.4 ± 5 Ma to 15.9 ± 0.4 Ma and show an overall negative relationship with Y concentrations. From ca. 20 Ma to 19 Ma, Y concentrations range between ∼570 and 5400 ppm with little discernible trend (Fig. 8D). From ca. 19 Ma to 16 Ma, however, Y concentrations range between ∼2000 and 12,000 ppm and increase sharply with decreasing date.
Specimen KA044 is a Gr+ Sil + Grt + Bt + Fsp + Qz gneiss (Fig. 3) located ∼1350 m structurally above KA037 (Fig. 2). Monazite grains occur within the matrix (with one exception), are anhedral to subhedral and equant to elongate in shape, are 50–550 µm in size, and are spatially associated with quartz, plagioclase, K-feldspar, and sillimanite. A single 20 µm subhedral monazite, grain 8, occurs as an inclusion in garnet. Elemental maps for all grains typically show weak to moderate irregular zoning in Th and U, with Y maps displaying weak zonation, where monazite rims are typically at higher Y concentrations than core regions. In total, 57 analyses on 10 grains located in the matrix yielded dates ranging from 39.3 ± 1.1 Ma to 16.1 ± 0.4 Ma, while the only analysis from the grain included in garnet yielded a date of 23.9 ± 0.7 Ma. The relationship between dates and Y concentrations is more complicated in this specimen than in those previously discussed; three different populations may be identified. From ca. 39 to 33 Ma, Y concentrations range between ∼400 and ∼2500 ppm with no obvious trend, while dates between ca. 31 Ma and 19 Ma have typically lower Y concentrations ranging between ∼100 and ∼1600 ppm (Fig. 8E). The youngest dates from this specimen (ca. 17–16 Ma) are associated with a sharp increase in Y with concentrations between ∼2300 and ∼6800 ppm.
Specimen KA055 is a Grt + Bt + Fsp + Qz gneiss (Fig. 3) located ∼3800 m structurally above KA044 (Fig. 2). Monazite grains in this specimen occur solely within matrix, are generally anhedral and equant to elongate in shape, are 60–120 µm in size, and are typically associated with biotite, quartz, and plagioclase. Elemental maps show weak, patchy to no zonation in Th and U; however, Y shows moderate irregular internal zoning with thin, high-concentration rims. In total, 48 analyses on nine grains yielded dates from 23.6 ± 0.6 Ma to 16.1 ± 0.4 Ma with associated Y concentrations that help define two populations (Fig. 8F). Between ca. 24 Ma and 20 Ma, Y concentrations range from ∼2700 to 5000 ppm and increase slightly with decreasing date, while from ca. 20 Ma to 16 Ma, Y concentrations range between 5400 and 27,000 ppm and continue the general trend of increasing with younger dates.
Specimen KA058B is a Grt + Bt + Sil + Qz + Fsp gneiss (Fig. 3) located ∼4250 m structurally above KA055 (Fig. 2). The monazite grains analyzed occur both within the matrix and as inclusions in garnet. Grains located in the matrix are anhedral and typically elongate in shape, range from 30 to 100 µm in size, and are commonly associated with feldspar, biotite, and quartz. Two grains that occur as inclusions in garnet are anhedral and oblate to elongate in shape and 25–40 µm is size. In total, 25 analyses on eight grains located in the matrix and five analyses on two grains located in garnet yielded dates that define two broad populations. An older population contains dates that span from 797.0 ± 29.6 Ma to 84.7 ± 7.5 Ma, while a younger population ranges between 41.0 ± 1.0 Ma and 18.6 ± 0.5 Ma. The older population yield ages older than the orogen itself and as such are not considered further. In the younger population, however, there appear to be three poorly defined growth events (Fig. 8G). Grains with the two oldest dates (ca. 41 and 37 Ma), representing the first event, have Y concentrations of ∼3500 and ∼5400 ppm respectively. Four analyses with dates between ca. 31 and 27 Ma define the second growth event and range in Y concentration from ∼1900 to ∼4200 ppm. The third growth event, from ca. 24 to 19 Ma, is associated with the largest range of Y concentrations, between ∼2800 and ∼8100, but it showed no obvious trend with date (Fig. 8G).
Specimen KA064A is a Grt + Sil + Bt + Fsp + Qz gneiss (Fig. 3) from the highest structural level examined during this study (Fig. 2). It is located ∼1550 m structurally above KA058B and ∼17 km structurally above the Main Central thrust (Fig. 2). The monazite grains analyzed occur both within the matrix and as inclusions in garnet. Matrix monazite grains are anhedral and oblate to elongate in shape, 50–150 µm in size, and commonly associated with biotite, K-feldspar, quartz, and/or sillimanite. Two grains that occur as inclusions in garnet are anhedral in shape and 17–50 µm in size. Elemental maps for all grains show weak to moderate patchy and irregular zoning of Th, U, and Y. Y maps, however, typically show lower concentrations in cores and higher concentrations toward the rim. In total, 69 spots on 10 grains define two broad populations: An older population (n = 3), of which two analyses are from grains included in garnet, spans between 496.8 ± 10.7 Ma and 105.6 ± 4.3 Ma, and a younger population (n = 66) ranges between 41.5 ± 0.9 Ma and 19.5 ± 0.6 Ma. In the younger population, there appears to be three growth events defined by a relationship between date and Y concentration. From ca. 42 Ma to 36 Ma, Y ranges between ∼5000 and ∼18,000 ppm and shows no obvious trend with date (Fig. 8H). Y is generally higher from ca. 35 to 26 Ma (between ∼20,000 and 28,000 ppm with the exception of two analyses) and typically increases with younger dates. Between ca. 24 Ma and 20 Ma, Y ranges between 8400 and 32,000 ppm and shows no discernible trend (Fig. 8H).
Previous investigations of monazite grain-scale age domains have demonstrated that they are commonly associated with distinct Y and heavy rare earth element (HREE) zones (e.g., Foster et al., 2000, 2002; Gibson et al., 2004; Cottle et al., 2009; Kellett et al., 2010; Larson et al., 2011; Langille et al., 2012). The variable composition of monazite grains can be interpreted as a function of interactions with and the growth/presence/consumption of garnet and other accessory phases that incorporate HREEs and Y (Pyle and Spear, 1999; Foster et al., 2000; Gibson et al., 2004; Buick et al., 2006; Rubatto, 2006). Garnet is the dominant HREE-bearing mineral phase present in the specimens examined in this study; other phases such as zircon and apatite were either not identified or are rare. Xenotime, YPO4, the presence or absence of which has been shown to exert control on the available HREE and Y budgets for mineral growth (Pyle and Spear, 1999, 2003; Pyle et al., 2001), was identified only in specimen KA007. However, only a few grains were recognized, and as such, it not considered to have a significant effect on the total HREE budget.
Given that garnet represents the largest volume percent of HREE-bearing phases in the specimens analyzed, it is interpreted to have exerted a primary control on the bulk HREE and Y budgets (e.g., Foster et al., 2000, 2002; Pyle et al., 2001; Gibson et al., 2004; Kohn et al., 2005). The HREE (and Y) concentrations within monazite are, therefore, interpreted to dominantly reflect changes in garnet modal abundance. Monazite domains that formed when garnet was stable/growing and abundant are expected to have relatively low concentrations of HREEs and Y, while monazite domains that grew during garnet breakdown and/or low modal abundance are expected to have relatively high concentrations of HREE and Y (Foster et al., 2000, 2002; Pyle et al., 2001; Gibson et al., 2004; Kohn et al., 2005). Similar interpretations between monazite growth zones and garnet stability can be made using HREE slope profiles (Gd/Yb ratios or similar). If garnet is abundant and stable or growing, the majority of the HREEs would be expected to reside therein (Pyle and Spear, 1999), despite a preferential partitioning of HREE into monazite (e.g., Buick et al., 2006; Rubatto, 2006). This would result in a steeply negative HREE slope or high Gd/Yb (or similar) ratios (Buick et al., 2006; Rubatto et al., 2006). On the other hand, monazite domains that grew coincident with either initial garnet growth (low modal abundance) or garnet breakdown are expected to be characterized by lower Gd/Yb (or similar) ratios. The following section is an interpretation of monazite growth domains based on Y concentrations and Gd/Yb ratios and their interpreted association with garnet. Dates older than the ca. 50 Ma initiation of the Himalayan orogen (n = 20; e.g., Green et al., 2008) are interpreted as being inherited and/or detrital, and hence are not discussed in the following.
KA044, KA058B, and KA064A
Monazites from KA044, KA058B, and KA064A are discussed together because they record similar, relatively old dates and protracted histories (Figs. 8E, 8G, and 8H). Monazite from KA044 records three compositionally distinct growth events during the: (1) middle Eocene to early Oligocene, (2) early Oligocene to early Miocene, and (3) early to middle Miocene (Fig. 8E). Low to moderate Y concentrations and moderate to high Gd/Yb ratios in monazite domains associated with the earliest event are consistent with having grown during initial garnet growth, with REEs potentially buffered through the prograde breakdown of another REE phase such as xenotime (Pyle and Spear, 1999; Pyle et al., 2001). The low Y concentrations and moderate to high Gd/Yb ratios associated with early Oligocene to early Miocene monazite dates indicate that garnet was likely abundant and stable at this time, consistent with progressive garnet growth during protracted prograde metamorphism. Finally, a significant increase in Y concentrations and low Gd/Yb ratios in early to middle Miocene monazite domains indicate that growth occurred coeval with garnet breakdown.
Similar to KA044, monazites from KA058B and KA064A are interpreted to record major growth events during the: (1) middle to late Eocene, (2) late Eocene–early Oligocene to late Oligocene, and (3) late Oligocene to early Miocene (Figs. 8G and 8H). The Y concentrations and Gd/Yb ratios from spots analyzed in KA058B form less well-defined patterns than those in the previous specimen. The Y data are characterized by concentrations changing from moderate in the oldest event, to moderate to low in the middle event, and to moderate to high in the youngest event. Gd/Yb ratios range from moderate to low in the early and middle events and yield the highest and lowest values in the youngest event. While the Y and Gd/Yb data are broadly consistent with protracted garnet growth through the first two events, the spread in Y concentrations and Gd/Yb ratios in the youngest event is difficult to interpret. This subset of data contains the highest Y concentrations and the lowest Gd/Yb ratios, which are consistent with garnet breakdown. It is possible that the large range in values may reflect incongruent anatexis and crystallization, with local growth or breakdown of garnet.
Although KA064A records similar dates to KA058B, the Y and Gd/Yb data appear better behaved. Y concentrations range from low to moderate during the oldest event, while Gd/Yb ratios at the same time range from high to low. During the late Eocene–early Oligocene to late Oligocene, Y concentrations increase with younger dates, while Gd/Yb ratios are relatively stable. Y concentrations yield their highest values during the youngest growth event, although there is some downward spread. Gd/Yb data associated with the same dates range from low to moderate. The Y concentration and Gd/Yb ratio patterns are consistent with early monazite growth with garnet during the middle to late Eocene. This was followed by the growth of higher-Y (HREE) and lower-Gd/Yb monazite in the late Eocene–early Oligocene to late Oligocene. The REE chemistry during this stage may be indicative of garnet breakdown or, alternatively, the breakdown of xenotime to form monazite in the presence of relatively stable garnet. Finally, the high Y concentrations and low Gd/Yb ratios associated with the late Oligocene to early Miocene are consistent with garnet breakdown at this time; however, the large range in values may again reflect incongruent anatexis with local open-system behavior.
KA031B, KA034, KA037, and KA055
Unlike the previously discussed specimens, monazites from KA031B, KA034, KA037, and KA055 do not record protracted, episodic histories (Figs. 8B, 8C, 8D, and 8F). It is possible that an earlier metamorphic history has been destroyed, as has been interpreted in the Manaslu region of west-central Nepal (Larson et al., 2011), where in situ partial melting and the production of significant anatexite (>40% by volume) were thought to have resulted in the resorption of earlier monazite. Such a scenario for these specimens is considered unlikely, however, because they contain similar or lower total volume percent anatexite (10%–15%) as compared to the previously discussed specimens that preserve an earlier history (Fig. 3). If the monazite record is considered to be complete, then these specimens record a relatively short history that begins in the late Oligocene or early Miocene, broadly coincidental with the final growth events of KA044, KA058B, and KA064A. Y concentration in all specimens increases through time with younger age, while Gd/Yb ratios show the opposite trend (Figs. 8B, 8C, 8D, and 8F). This pattern is compatible with initial growth of monazite in the presence of growing or stable garnet in the late Oligocene and into the early Miocene, followed by further monazite growth during garnet breakdown in the early to mid-Miocene.
KA007 records a similar history to the specimens just discussed (Fig. 8A). KA007 contains relatively little anatexite and is characterized by a similar lithology to specimens KA044, KA058B, and KA064B, which record an early and protracted monazite growth history (Fig. 3). Therefore, as with KA031B, KA034, KA037, and KA055, the monazite record preserved in KA007 is interpreted to represent the entire metamorphic history of the specimen.
Metamorphism and associated monazite growth in KA007 do not appear to have occurred until ca. 31 Ma (Fig. 8A). The relatively low Y concentrations from monazite in this specimen, which generally increase with younger dates through the Oligocene, and the initially high Gd/Yb ratios, which decrease through the same time period, are consistent with garnet growth and progressive prograde metamorphism. This period is broadly coeval with the second monazite growth event recorded in KA044, KA058B, and KA064A (Figs. 8E, 8G, and 8H). Increasing Y concentrations and further decreasing Gd/Yb ratios from the latest Oligocene and into the early Miocene are compatible with garnet breakdown during this time. This later time span overlaps with the final episode of monazite growth in KA044, KA058B, and KA064A and the onset of monazite growth in KA031B, KA034, KA037, and KA055. As such, the petrochronologic data are interpreted as recording contemporaneous garnet breakdown in KA007, KA044, KA058B, and KA064A, and garnet growth in KA031B, KA034, KA037, and KA055.
P- T- t Paths
The integration of pseudosections and petrochronology can be used to constrain P-T-t paths and help unravel a specimen’s geologic history. In this study, P-T-t paths generated for three specimens from different strategic structural levels were used to constrain the geologic history and, in particular, the internal structure of the Himalayan metamorphic core. These specimens are discussed next from lower to higher structural position.
The composition of monazite grains in KA007 is consistent with garnet growth between ca. 31 and 24 Ma and garnet breakdown between ca. 24 and 19 Ma (Fig. 8A). It was not possible to determine the full prograde P-T path from this specimen; however, garnet growth from ca. 31 Ma to 24 would be consistent with prograde metamorphism (Figs. 7A and 7B). The decompressional cooling path outlined in the pseudosection for this specimen crosses closely spaced, decreasing garnet isomodes during melt crystallization (Figs. 7A and 7B). A decompressional cooling path would continue to cross decreasing isomodes below the solidus, where they would be more widely spaced. As such, it is interpreted that the majority of garnet breakdown, and resulting release of HREEs and Y available for incorporation into monazite (e.g., Kohn et al., 2005), likely occurred during melt crystallization from ca. 24 Ma to 19 Ma (Figs. 7A and 7B).
Monazite grains in KA044 record three growth events (Fig. 8E). The first two events are interpreted to reflect simultaneous garnet growth, while the final event is interpreted to record garnet breakdown. Moderate to heavy REE and Y concentrations in monazite domains associated with the first event range from ca. 39 Ma to 33 Ma, consistent with initial prograde garnet growth. Further monazite growth from ca. 31 to 19 Ma records even lower Y concentrations, consistent with continued garnet growth and HREE and Y sequestration. The P-T path for this specimen as interpreted from the pseudosection only reveals the high-metamorphic-grade history. Because garnet is stable at lower pressures and temperatures than what is recorded, it is likely that initial garnet and monazite growth occurred at lower metamorphic grades than those covered by the P-T path. The second monazite growth event, however, may correlate with the near-isothermal heating path interpreted on the pseudosection (Figs. 7C and 7D). As monazite is typically resorbed with increasing melting fraction (Kelsey et al., 2008; Spear and Pyle, 2010; Yakymchuk and Brown, 2014), the second event is interpreted to be the result of subsolidus growth coeval with garnet between ca. 31 and 19 Ma. In addition to helping add timing constraints, the Y concentrations in monazite also help to constrain the trajectory of the prograde path. Based on isomodes, garnet would be expected to breakdown along a heating path with minor decompression, thereby releasing HREEs and Y. The fact that HREEs and Y do not increase in monazite domains yielding dates between ca. 31 Ma and 19 Ma requires either a near-isobaric heating or a burial path during this time (Figs. 7C and 7D). The subsequent hiatus in monazite growth from ca. 19 Ma to 17 Ma is consistent with crossing the solidus and increasing melt fraction, and hence continued heating. The youngest monazite growth in the specimen records a sudden increase in HREEs and Y, consistent with garnet breakdown along the retrograde path. As with KA007, garnet isomodes are closely spaced above the solidus and widely spaced below (Fig. 7D), implying that the majority of garnet breakdown likely took place above the solidus. This is consistent with garnet breakdown during melt crystallization from ca. 17 to 16 Ma (Fig. 7C).
Monazite grains in KA064A record three growth events (Fig. 8H). Low HREEs and Y associated with the first growth event are consistent with coeval garnet growth between ca. 42 Ma and 36 Ma. The interpreted P-T path for KA064 crosses increasing garnet isomodes below and above the solidus during heating (Figs. 7E and 7F). Monazite, however, is likely to resorb with increasing melt fraction above the solidus (e.g., Kelsey et al., 2008; Spear and Pyle, 2010; Yakymchuk and Brown, 2014), and, therefore, growth between ca. 42 Ma and 36 Ma is interpreted to coincide with subsolidus garnet growth. As with KA044, initial garnet growth likely began at lower pressures and temperatures than estimated by the P-T path; dates from the oldest monazite domains may correspond to initial prograde garnet growth. In contrast to early monazite growth, the later two growth events are both interpreted to record evidence of garnet breakdown. HREE and Y concentrations increase significantly from the first monazite growth event to the second, indicating that garnet was unstable between ca. 36 Ma and 26 Ma. This could occur along a decompressional heating path that crosses decreasing garnet isomodes (Fig. 7F). The subsolidus heating path shown in Figures 8E and 8F, however, crosses widely spaced increasing garnet isomodes. There are two possibilities that may explain garnet breakdown along a similar path. First, prior to melt loss, the muscovite-in line would have occurred at lower pressures and temperatures (e.g., Guilmette et al., 2011; Groppo et al., 2012), resulting in closely spaced, near-horizontal garnet isomodes at the pressures and temperatures of the estimated P-T path. An isobaric or decompressional heating path through a muscovite-bearing field would result in significant subsolidus garnet breakdown. Such a path, however, would also be expected to cross decreasing grossular isopleths (see GSA Data Repository material [see footnote 1]). This is not consistent with measured garnet core and mantle compositions in this specimen, which record constant grossular content. A second explanation is that pressures are underestimated. If the initial trajectory of the P-T path shown in Figures 8E and 8F were shifted to slightly higher pressures (so that the path crossed through the muscovite-bearing field), then subsolidus decompressional heating consistent with garnet breakdown would result. Again, this explanation is not supported by measured garnet compositions. It is possible, however, that substantial residence time at high temperatures has destroyed prograde garnet zoning (Caddick et al., 2010). This would imply that either explanation, or a combination of both, could explain the breakdown of garnet and uptake of Y and HREEs in monazite growth between ca. 35 and 26 Ma (Fig. 7E). High HREE and Y concentrations during the final monazite growth event are consistent with garnet breakdown along a decompressional retrograde path (Figs. 7E and 7F). As with specimens discussed previously, isomodes are more widely spaced below the solidus, consistent with the majority of garnet breakdown occurring during melt crystallization between ca. 24 Ma and 20 Ma (Fig. 7E).
P- T- t Discontinuities
The P-T-t paths and petrochronology reveal distinct metamorphic histories for specimens collected at different structural levels across the Himalayan metamorphic core (Fig. 9). Whereas specimens KA044, KA058, and KA064 record protracted middle Eocene to early Miocene monazite growth and metamorphic histories, specimens KA031B, KA034, KA037, and KA055, in contrast, record relatively short, late Oligocene to middle Miocene monazite growth/metamorphic histories (Fig. 9). Similar to KA031B, KA034, and KA037, specimen KA007 also records a relatively short metamorphic history (Fig. 9). Unlike those other specimens, however, KA007 records an early Oligocene to early Miocene history as well. The present spatial distribution of these specimens with varying geologic histories is not only consistent with the position of the previously mapped High Himal Thrust (Goscombe et al., 2006; Imayama et al., 2012), but it also reveals the presence of additional tectonometamorphic discontinuities. A total of five tectonometamorphic breaks can be interpreted (Fig. 10) and are discussed next in order from higher to lower structural levels, which roughly corresponds to decreasing age.
KA058B over KA055
Whereas specimens KA058B and KA064A record middle Eocene to early Miocene prograde metamorphism followed by late Oligocene to early Miocene retrograde metamorphism (Fig. 9), the structurally lower specimen, KA055, records late Oligocene to early Miocene prograde metamorphism followed by early Miocene retrograde metamorphism. Although no structure was identified in the field that constrained a discontinuity, KA055 and KA058B/KA064A are separated by the reappearance of muscovite (up structural section; Figs, 2 and 10). The muscovite-in line may indicate the location of a structural contact juxtaposing rocks with distinct metamorphic histories.
KA055 over KA044 (High Himal Thrust)
Specimens KA044 and KA055 (Fig. 2) consist of similar, Kfs + Sil–grade, muscovite-free assemblages. As discussed previously, KA055 records late Oligocene to early Miocene prograde metamorphism followed by early Miocene retrograde metamorphism (Fig. 9). The structurally lower specimen, KA044, in contrast, records protracted middle Eocene to early Miocene prograde metamorphism followed by early Miocene retrograde metamorphism (Fig. 9). The break between these specimens coincides with the High Himal thrust of Goscombe et al. (2006; Figs. 2 and 10 herein).
KA044 over KA037, KA034, KA031
Whereas specimen KA044 records protracted middle Eocene to early Miocene prograde metamorphism, the structurally lower specimens, KA037, KA034, and KA031B, record late Oligocene–early Miocene prograde metamorphism followed by early to middle Miocene retrograde metamorphism (Fig. 9). Although no structure was identified in the field to precisely constrain the location of such a discontinuity, the specimens fall on either side of the muscovite-out line (up structural section; Fig. 2). The disappearance of muscovite is, therefore, interpreted to approximately coincide with a structural contact (Fig. 10).
Base of Migmatites
Specimen KA029, situated just north of and structurally above the Tamor window (Fig. 2), marks the approximate location of the structurally lowest migmatites observed, while KA031B, located ∼2000 m to the northeast and ∼350 m up structural section, marks the first definitive appearance of sillimanite. It is possible that sillimanite occurs closer to, or at the location of, KA029; however, it was not identified in outcrop, and no specimens were collected for more detailed petrography. Rocks immediately structurally below KA029 are dominantly schists and phyllites that have very low anatexite volume percentages. Moreover, the location of KA029 roughly coincides with a sharp up-structural-section drop in metamorphic pressures and a change in garnet mineral chemistry (between specimens H1206 and H1205 from Imayama et al., 2010). Below KA029, rocks record pressures of ∼11 kbar, and garnet grains preserve prograde growth zoning (Imayama et al., 2010). Rocks up structural section, in contrast, record pressures of ∼7 kbar, and garnet grains exhibit chemical profiles typical of homogenization at high temperatures and retrograde rims (Imayama et al., 2010). The juxtaposition of the structurally lowest migmatites (e.g., KA029) against subjacent rocks with significantly different lithologies and garnet chemistry is interpreted to reflect a structural contact (Fig. 10).
KA031B over KA007
Similar to KA029, north of the Tamor window, the location of the KA011 marks the structurally lowest occurrence of migmatites observed south of the Tamor window. Although these migmatites have been previously mapped as belonging to the same continuous unit as KA029, KA031B, etc. (Shrestha et al., 1984; Schelling, 1992; Goscombe and Hand, 2000; Goscombe et al., 2006), the data from this study show that they record distinct geochronologic and metamorphic histories. Specimens KA031B, KA034, and KA037, north of the Tamor window (Fig. 2), record late Oligocene–early Miocene prograde metamorphism followed by early to middle Miocene retrograde metamorphism, whereas specimen KA007, located southwest of the Tamor window, records early to late Oligocene prograde metamorphism followed by late Oligocene to early Miocene retrograde metamorphism (Fig. 9). In addition to variable timing, the migmatites also record different metamorphic grades; rocks to the north of the Tamor window were metamorphosed at sillimanite-grade conditions, whereas those to the southwest only reached kyanite grade (Figs. 2 and 10). These data are consistent with a discontinuity separating the locally migmatitic rocks to the southwest of the Tamor window from the locally migmatitic rocks to the north. Such a discontinuity is interpreted as being completely eroded away above the Tamor window (Fig. 10).
The P-T-t paths and distinct petrochronology from this study indicate that the evolution and internal structure of the Himalayan metamorphic core are considerably more complicated than previously thought. During the collision between the Indian and Eurasian plates, specimens KA064, KA058B, and KA044 were buried to depths sufficient for garnet growth by the middle Eocene. A thickened continental crust during this time is consistent with: (1) U-Pb zircon ages of ca. 44 Ma from granites in southeast Tibet that are interpreted to postdate the majority of crustal thickening in the Tethyan Sedimentary Sequence (Aikman et al., 2008); (2) Lu-Hf garnet ages as old as ca. 54 Ma and 51 Ma from the Mabja and Kangmar gneiss domes, respectively (Smit et al., 2014); (3) Lu-Hf dating of garnet in the Ama Drime massif just to the northwest of the study area that record eclogite-facies metamorphism as early as ca. 38 Ma (Kellett et al., 2014); (4) titanite dating with paired Zr thermometry that indicates the midcrust in central Nepal was >700 °C at 37 Ma (Kohn and Corrie, 2011); and (5) illite 40Ar/39Ar ages of ca. 42 Ma interpreted to record initial crustal thickening in the northwest Himalaya of India (Wiesmayr and Grasemann, 2002).
By the latest Eocene or earliest Oligocene, these rocks, which now formed at least part of the Himalayan metamorphic core, were translated toward the Himalayan foreland (Fig. 11A). Extrusion during this period is consistent with the initiation of movement across the South Tibetan detachment system, as exposed in the Mabja dome to the north in southern Tibet, during the late Eocene (Lee and Whitehouse, 2007). The exact mechanism responsible for this movement is not clear. It may reflect lateral flow, as in the thermo-mechanical models of Beaumont et al. (2004); however, the time line for the initiation and melt volume expected are significantly different that the examples published therein. Alternatively, the movement may reflect translation/extrusion southward as part of a tectonic wedge (e.g., Webb et al., 2007, 2011; He et al., 2015).
The continued convergence that drove the Himalayan metamorphic core toward the foreland in the hanging wall of the basal detachment simultaneously buried the footwall material; KA007 was buried to garnet-grade depths by ca. 31 Ma (Fig. 11B). Continued burial of KA007 during the late Oligocene was contemporaneous with the onset of melting in portions of the Himalayan metamorphic core (Fig. 9). While KA064A began melting at ca. 26 Ma, the structurally lower specimen, KA044, did not begin to melt until ∼5 m.y. later. The earlier melting recorded by KA064A can be explained by its interpreted P-T path (Figs. 7E and 7F), which crosses the solidus at ∼750 °C, whereas KA044 does not cross the solidus until ∼800 °C (Figs. 7C and 7D). This is consistent with an isothermal depth profile between these specimens. Although KA044 was deeper, it reached approximately the same temperature, consistent with the findings of Imayama et al. (2010). The higher temperatures required to cross the solidus in specimen KA044 may reflect a combination of a different bulk composition and higher pressures.
Decompression of specimen KA007 at ca. 24 Ma coincides with melt crystallization in specimens KA064A and KA058B (Figs. 7 and 9). This is consistent with a rock package containing specimen KA007 being accreted to the extruding and exhuming Himalayan metamorphic core (Fig. 11C). The onset of metamorphism recorded by specimens KA055 and KA031B at approximately the same time is consistent with underthrusting below, and/or overriding by, the Himalayan metamorphic core during this time (Fig. 11C). From ca. 24 until 20 Ma, as shown in Figure 11D, specimens KA055 and KA031B continued to be underthrust/overrode, while specimens KA007, KA044, and KA064A were part of the hanging wall.
At ca. 20 Ma, specimen KA055 began to be exhumed, while the record of metamorphism in KA058B and KA064A ceased (Fig. 9). Meanwhile, specimens KA034 and KA037 record initial metamorphism, and KA044 reached peak temperature and underwent partial melting (Fig. 9). The contemporaneous burial of specimens KA034 and KA037, continued metamorphism of KA044, and exhumation of specimens KA055, KA058, and KA064A are consistent with the development of an out-of-sequence thrust between ca. 20 Ma and 18 Ma (Figs. 11E and 11F). Such an out-of-sequence thrust would coincide with the previously mapped High Himal thrust of Goscombe et al. (2006), which they interpreted as a primary structure that controlled the evolution of the Himalayan metamorphic core, as discussed in section 2.3 (Previously Mapped Structures). The data from this study indicate that the High Himal thrust was relatively short-lived. This interpretation is in line with Brunel (1986) and Brunel and Kienast (1986), who interpreted displacement along the High Himal thrust (which they mapped as the Main Central thrust) in the Everest and Makalu regions as postdating metamorphism. Out-of-sequence thrusting associated with the High Himal thrust is also consistent with interpretations of similar motion, though at slightly different times, along the Laya-Kakhtang thrust in Bhutan (Daniel et al., 2003; Warren et al., 2011; Grujic et al., 2002, 2011) and equivalents in northeast India (Warren et al., 2014).
Movement across the High Himal thrust as an out-of-sequence thrust at ca. 20 Ma is inconsistent with the interpretation of Imayama et al. (2012), and subsequent reinterpretation of that same data set, from the adjacent valley to the east in the Kanchenjunga region, by Montomoli et al. (2014). In both studies, movement across the High Himal thrust was interpreted to initiate at ca. 27 Ma and continue until ca. 18 Ma. These constraints came from zircon petrochronologic data that outlined two distinct geologic histories within the Himalayan metamorphic core. One specimen recorded prograde metamorphism in between 33 Ma and 28 Ma in the early Oligocene, followed by retrograde metamorphism in the late Oligocene–early Miocene (27–23 Ma), while a second, structurally lower specimen recorded prograde metamorphism in the early Miocene (21–18 Ma) and retrograde metamorphism in the early to mid-Miocene (18–16 Ma). The previous authors attributed the juxtaposition of rocks with distinct histories to juxtaposition by the High Himal thrust; however, their own mapping indicates that both specimens occur in the footwall of the fault. Alternatively, their data can be reinterpreted such that the break in histories noted between their specimens represents movement across the same structure that juxtaposed KA044 with the panel of rocks containing KA037, KA034, and KA031B (Fig. 10) in this study. The minor differences in timing of hanging-wall and footwall histories recorded between studies may reflect differences in the timing of monazite and zircon growth along the specimens’ respective P-T path (e.g., Yakymchuk and Brown, 2014).
Following out-of-sequence thrusting across the High Himal thrust, deformation migrated back toward the foreland, resulting in the crystallization/cooling/exhumation of the remaining specimens by ca. 12 Ma. The continuation of this processes and continued underplating of progressively lower-metamorphic-grade rock led to the formation of the Lesser Himalayan duplex (Fig. 11G; e.g., Bollinger et al., 2006; Webb, 2013).
Although interpreted herein to reflect duplexing processes and underplating of footwall material (compatible with He et al., 2015), the geologic history outlined herein and juxtaposition of rock units with variable histories are broadly consistent with first-order geometric predictions of dynamic models of the Himalaya (e.g., Beaumont et al., 2001; Jamieson et al., 2006). Specifically, in model HT111 of Jamieson et al. (2006), the Himalayan metamorphic core is shown to consist of rock packages that were laterally separated and subsequently juxtaposed following variable exhumation histories. This same model also predicts that the most foreland-ward unit in the final geometry may not have been so initially, which is consistent with the position and history of KA007 in this study. There are differences, however, between the present study and dynamic models in the time scales over which metamorphism and exhumation have occurred and the processes driving both. As published, HT111 predicts all units were at midcrustal depths at 24 Ma, and they were juxtaposed and exhumed between 16 and 6 Ma (Jamieson et al., 2006), largely reflecting lateral translation with erosion-focused exhumation. The data from this study, however, argue for initial juxtaposition as early as 31 Ma and exhumation as early as 24 Ma, both driven by underplating and duplexing.
This study demonstrates that the internal structure of the Himalayan metamorphic core includes a number of thrust-sense discontinuities reflective of its kinematic history. The results herein indicate the following:
(1) The protoliths of the rocks forming at least part of the Himalayan metamorphic core were buried to garnet-grade depths by at least the middle Eocene, indicating the presence of a thickened continental crust by this time.
(2) Metamorphism in the Himalayan metamorphic core was protracted, spanning more than 30 m.y.
(3) The Himalayan metamorphic core evolved and was assembled through a combination of lateral extrusion, underplating, and out-of-sequence thrusting as recorded across multiple P-T-t discontinuities.
(4) The High Himal thrust is an out-of-sequence thrust that was active between 20 and 18 Ma.
This research was supported by a Natural Sciences and Engineering Research Council of Canada (NSERC) Graduate Scholarship, a Geological Society of America Graduate Research Grant, a Gem and Mineral Federation of Canada David Barclay scholarship, and a University of British Columbia University Graduate Fellowship to T. Ambrose. Additional support came from a NSERC Discovery grant to K. Larson and a National Science Foundation grant (EAR-1119380) to J. Cottle. Field work was made possible by P. Tamang, S. Tamang, S.D. Tamang, T Tamang, and everyone else with Pike Peak Trekking. This work benefited greatly from discussions with G. Lederer. S. Creighton, M. Beauchamp, and D. Arkinstall are thanked for their assistance with scanning electron microscope/electron microprobe analyses. D. Tinkham is thanked for his assistance with Theriak/Domino. This contribution benefited greatly from excellent and thoughtful reviews by T. Imayama, M. Williams, and C. Beaumont. Many thanks to A.B. Weil for his excellent editorial work.