Subduction dynamics, volatiles and melts: Investigations from surface to deep mantle


Subduction dynamics, volatiles and melts: Investigations from surface to deep mantle
Co-organized by GMPV2
Convener: Oğuz H Göğüş | Co-conveners: Taras Gerya, Ágnes Király, Wim Spakman, Jeroen van Hunen
vPICO presentations
| Wed, 28 Apr, 15:30–17:00 (CEST)

vPICO presentations: Wed, 28 Apr

Chairpersons: Oğuz H Göğüş, Ágnes Király, Jeroen van Hunen
Wouter P. Schellart and Vincent Strak
Flat slab subduction occurs when the subducted slab lies flat below the base of the overriding plate for up to several hundred kilometres before bending into the deeper mantle. It has been ascribed to a variety of causes, including subduction of buoyant ridges/plateaus and forced trench retreat. Ridge/plateau subduction, however, shows irregular spatial correlations with flat slabs, while forced trench retreat has required external forcing in geodynamic subduction models, which might be insufficient or absent in nature. Here we present buoyancy-driven numerical geodynamic models and aim to investigate flat slab subduction in the absence of external forcing. Furthermore, we test the influence of a variety of subduction zone parameters, including overriding plate strength and subducting plate thickness, on flat slab formation and its evolution. Flat slab subduction is reproduced during normal oceanic subduction in the absence of ridge/plateau subduction and without externally forced plate motion. Flat slab subduction only commences after a prolonged period of upper mantle slab dip angle reduction during lower mantle slab penetration. The flat slab is supported by mantle wedge suction, vertical compressive stresses at the base of the slab and upper mantle slab buckling stresses. Our models demonstrate three modes of flat slab subduction, namely short-lived (transient) flat slab subduction, long-lived flat slab subduction, and periodic flat slab subduction, which occur for different model parameter combinations. Most models demonstrate slab folding at the 660 km discontinuity, which produces periodic changes in the upper mantle slab dip angle. With relatively high overriding plate strength, such folding results in periodic changes in the dip angle of the flat slab segment, which can lead to periodic flat slab subduction, providing a potential explanation for periodic arc migration. Flat slab subduction ends due to the local overriding plate shortening and thickening it produces, which forces mantle wedge opening and a reduction in mantle wedge suction. As overriding plate strength controls the shortening rate, it has a strong control on the duration of flat slab subduction, which lasts only ~6 Myr for the weakest plate and exceeds 75 Myr for the strongest plate. Progressive overriding plate shortening during flat slab subduction might explain why flat slab subduction terminated in the Eocene in western North America and in the Jurassic in South China.

How to cite: Schellart, W. P. and Strak, V.: Short-lived, long-lived and periodic flat slab subduction, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-4004,, 2021.

Laurent Husson, Nicolas Riel, Sonny Aribowo, Christine Authemayou, Danny Hilman Natawidjaja, Boris Kaus, Gino de Gelder, and Kevin Pedoja

At the far end of the Tethyan realm, the Indo-Australian plate subducts in the Java and Banda trenches. Across the trench, a checkerboard-like distribution of continental and oceanic units sets the geodynamic stage since the Australian continent docked into the subduction zone a few Myr ago: to the East, the Australian continent now subducts and collides with the mostly oceanic Wallacea while to the West, the Indian oceanic plate subducts underneath continental Sundaland. We hypothesize that this fast and transient geodynamic regime explains many observations that characterize the region over the last few Myr: slab rollback and formation of the Banda arc, subsidence of the Weber superdeep seafloor to more than 7000 m, back-arc thrusting in Flores, dynamic subsidence in Sundaland and Sahul, and controversial slab tearing underneath Timor. We set out to model subduction dynamics accounting for the complex assemblage of plates in a real-Earth perspective, using the fast thermo-mechanical code LaMEM that allows dealing with complex setups. Our results predict the winding of the subduction zone around Papua, ultimately retreating into the Banda embayment, thereby causing the extreme dynamic subsidence of the Banda seafloor. Geometrical consistency imposes coeval slab tearing underneath Timor while the slab rolls back. The formation of the Flores backthrust quickly follows Australian collision with Wallacea and propagates westward in continental Sundaland. Shortening rates quickly drop tenfold while entering Sundaland, in Java, in agreement with kinematic and structural observations. In the geologically near future, the back-arc thrust is predicted to reverse the subduction polarity, Wallacea being on the brink to subduct southward underneath Australia. Last, transient mantle flow expectedly causes dynamic subsidence in Sahul and Sundaland, thereby profoundly remodeling the physiography of the entire region.

How to cite: Husson, L., Riel, N., Aribowo, S., Authemayou, C., Natawidjaja, D. H., Kaus, B., de Gelder, G., and Pedoja, K.: Subduction dynamics, tectonics, and dynamic topography in the Banda-Java subduction zone, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-9654,, 2021.

Valeria Turino, Valentina Magni, Hans Jørgen Kjøll, and Johannes Jakob

The transition between continental and oceanic lithosphere in rifted margins can display a wide range of characteristics, which primarily depend on the regional tectonic evolution. Rifted margins form when continents rift apart and are commonly characterized by a thinned transition zone between the continental crust and the oceanic crust. The velocity and duration of the rifting process influence the dimensions and geometry of the passive margin. Rifted (or passive) margins are often subdivided in a magma-rich type and a magma-poor type, where the magma-rich are characterized by large input of mafic melt, derived from the mantle, into the crust. Magma-poor rifted margins on the other hand are characterized by much less magma production during the rifting process. This causes high variability in the geometry and rheology of passive margins.

The aim of this work is to understand how different types of passive margins can influence the dynamics of continental collision. We modelled subduction using the finite element code Citcom and to describe the dynamics of continental collision we mainly focused on the time and position of the slab break-off after the collision and on the fate of the passive margin material.

We compared these models as a function of various parameters (e.g., margin length, density, and viscosity), in order to understand how the architecture of a passive margin affects the dynamics of continental collision. We find that passive margins have a noticeable impact on subduction, as we observe a large variability in slab break-off times (about 10–70 Myr after continental collision) and depth (about 200–450 km). Furthermore, the factor that shows the largest impact on subduction dynamics is the rheology of the passive margin. Our results show that for both magma-poor and magma-rich margins, part of the margin does not subduct but, instead, exhumes and accretes to the overriding plate. Importantly, the amount of accreted material to the overriding plate is much larger when the passive margin is magma-poor compared to the magma-rich case. This is consistent with geological observations that fossil magma-poor passive margins are preserved in many mountain ranges, such as the Alps and the Scandinavian Caledonides, whereas remnants of magma-rich rifted margins are scarce. Because, in our models, the slab break-off occurs inboard of the LCB, magma-rich rifted margin may only be preserved when the density of the LCB is similar to that of the rest of the continental plate. Therefore magma-rich rifted margins are prone to be subducted and recycled into the mantle. Importantly, our results show that rifted margin type controls the architecture of the subsequent collisional phase of the Wilson cycle.

How to cite: Turino, V., Magni, V., Kjøll, H. J., and Jakob, J.: The effect of magma poor and magma rich rifted margins on continental collision dynamics, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-6052,, 2021.

Valentina Magni, Manel Prada, John Naliboff, and Carmen Gaina

Back-arc basins often present multiple spreading centres that form one after the other (e.g. Mariana subduction zone), propagate and rotate (e.g., Lau Basin) following trench retreat. In some cases, rift jumps can create continental fragments or microcontinents (e.g., Coral Sea, Central Mediterranean, Scotia Sea). The processes controlling rift jumps and possible formation of continental fragments are still not fully understood, but they are certainly related to the dynamics of subduction.

In this work, we show how episodic trench retreat shapes the morphology of back-arc basins and can produce rift jumps. We use the finite element code ASPECT to model the rifting of continental lithosphere in 2D with boundary conditions that simulate the asymmetric type of extension caused by the trench retreat. We perform a parametric study in which we systematically vary the duration of different extensional phases, simulating episodes of trench retreat. Our results show that when extension is continuous, continental break-up occurs and a spreading centre develops. On the other hand, rift jump occurs in models with multiple extensional phases resulting in more complex morphologies that go from a hyperextend margin, to microcontinent formation, to spreading centre jumps within the newly formed oceanic lithosphere. In the first two cases (i.e., hyperextended margin and microcontinent), the length of the rift jump ranges from about 40 to 100 km and the timing varies from about 2 to 6 Myr. In the latter case (i.e., spreading centre jump within oceanic lithosphere) the length of the jump is significantly lower, 10-15 km, and the time needed for the ridge jump to occur is <2 Myr. These values depend on the rheological properties of the lithosphere, but, importantly, we show that the resulting scenario is controlled by the duration of the first extension stage and of the break before the next one.

How to cite: Magni, V., Prada, M., Naliboff, J., and Gaina, C.: Rift jump and microcontinent formation in back-arc settings, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-7256,, 2021.

Arianna Secchiari, Alessandra Montanini, Dominique Cluzel, and Elisa Ferrari

The New Caledonia ophiolite hosts one of most complete sections of a nascent arc, representing an excellent natural laboratory for investigating the origin and the evolution of subduction systems. The sequence, originated during the onset of the Eocene subduction [1, 2], is composed of ultra-depleted forearc harzburgites [3] overlain by a dunite-dominated transition zone (500m thick), in turn overtopped by mafic-ultramafic cumulate lenses. The ultramafic rocks of the transition zone (mainly dunites and wehrlites) most likely resulted from melt-peridotite reactions involving primitive arc tholeiites and boninitic magmas [2]. By contrast, dunite-pyroxenite and gabbronorite cumulates derive from H2O-poor depleted melts transitional between boninites and arc-tholeiites [2, 4].

In this contribution, we report the first occurrence of amphibole-bearing ultramafic lithologies in the New Caledonia arc sequence. Our study deals with a petrological and geochemical characterisation of a 2.5km x 5km composite, roughly snowball-shaped, intrusive body, consisting of pyroxenite, dunite, wehrlite, hornblendite and associated mafic-ultramafic, locally sheared, dikes from the Plum area (Massif du Sud).  The pyroxenite component, which forms the core of the intrusion, consists of coarse-grained websterites, mainly composed of weakly oriented orthopyroxene (~ 30-75 vol.%) and clinopyroxene (~ 20-40 vol.%), with variable amounts of edenitic amphibole (~ 2-30 vol.%). The amphibole generally occurs as poikilitic crystals or develops as coronas on pyroxenes. Highly calcic plagioclase (An= 83-96 mol %) is scarce in the pyroxenite body (~ 2 vol. %), but more abundant in the associated dikes (~ 10-50 vol.%). Clinopyroxene shows variable Mg# (0.82-0.92) and low Al2O3 (0.99-2.00 wt%). Likewise, amphibole yields high Mg# (0.712-0.865). Estimated equilibrium temperatures based on conventional pyroxene thermometry range between 870-970°C, whereas amphibole-plagioclase pairs provide slightly lower values (800-910 °C).

Whole rocks have moderately high Mg# (71-82) and REE concentrations one to five times chondritic values. The websterites of the main body show LREE-depleted (LaN/NdN = 0.5-0.8), weakly concave downward patterns, with flat HREE segments (LuN/TmN = 1.1-1.3). The mafic-ultramafic dikes display similar patterns, bearing depleted to flat LREE segments (LaN/NdN = 0.6-1) and positive Eu anomalies. For all the investigated lithologies, extended trace element diagrams indicate enrichments for FME (i.e. Rb, Ba, U) and Th, coupled to Zr-Hf depletion. Strong Sr positive spikes are only observed for the crosscutting dikes, while the pyroxenite body yields Sr negative anomalies.

Geochemical modelling shows that the putative liquids in equilibrium with the websterites have intermediate Mg# (57–63) and incompatible trace element patterns sharing remarkable similarities with the New Caledonia CE-boninites [5]. However, they significantly differ from the equilibrium melts reported for the other intrusive rocks of the sequence [1, 4], suggesting greater compositional variability for the liquids ascending into the Moho transition zone and lower crust. Our results support the notion that petrological and geochemical heterogeneity of magmatic products may be distinctive features of subduction infancy.



[1] Marchesi et al., Chem. Geol., 2009, 266, 171-186.

[2] Pirard et al., J. Petrol., 2013, 54, 1759–1792.

[3] Secchiari et al., Geosc. Front., 2020, 11(1), 37–55.

[4] Secchiari et al., Contrib. Mineral. Petrol., 2018, 173(8), 66.

[5] Cluzel et al., Lithos, 2016, 260, 429–442.

How to cite: Secchiari, A., Montanini, A., Cluzel, D., and Ferrari, E.: Hydrous mafic-ultramafic intrusives in a nascent arc (Massif du Sud, New Caledonia ophiolite)., EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-10634,, 2021.

Cees-Jan De Hoog, Keiko Hattori, and Eleri Clarke

Boron provides an efficient tracer of fluids in subduction zones, due to its high concentration in surface reservoirs, low concentration in the mantle, and large isotope fractionation. The Higashi-akaishi peridotite body in Sanbagawa UHP belt, Japan, is composed of partially serpentinised dunites and harzburgites, which are interpreted to be exhumed mantle wedge peridotites. Compositions of olivine (Fo90-94, NiO 0.28-0.48 wt%, MnO 0.10-0.16 wt%) and chromite (Cr# >0.7, TiO2 <0.4 wt%) confirm its origin as highly refractory fore-arc mantle. Several generations of olivine and serpentine can be recognised in the samples, and were analysed in-situ for their B content and B isotopic composition by SIMS. Coarse-grained primary mantle olivine has low [B] (1-3 µg/g), but is still significantly B-enriched compared to typical mantle olivine, and has δ11B of -10 to -3 ‰. Lower B contents in olivine cores compared to rims suggests diffusive incorporation of B from slab-derived fluids at high temperature.  Later fine-grained olivine neoblasts, products of dynamic recrystallization, have higher [B] (3-11 µg/g) and higher δ11B (-7 to +2‰). Platy antigorite associated with the olivine neoblasts have similar [B] (4-12 µg/g) but higher δ11B (-4 to +6‰). Late-stage greenschist-facies overprint resulted in lizardite veining with high [B] (18-52 µg/g) and a narrow range of δ11B (-2 to -1‰).

We envisage the following scenario. Coarse-grained mantle olivine acquired B from slab-derived fluids when the peridotites were dragged down by mantle corner flow and positioned near the slab-mantle interface. The values of δ11B (-10 to -3‰) are consistent with fluids from dehydrating slab at ca. 110-150 km depth, but are potentially affected by diffusion-controlled kinetic isotope fractionation. High temperatures (> 650-700°C) prevented the peridotites from serpentinisation. Subsequently the rocks were down-dragged in a subduction channel where olivine neoblasts formed first and platy antigorite crystallized later when temperature dropped below 650°C. Both phases show heavier δ11B than coarse-grained olivine; the values are consistent with fluids from dehydrating slab at ca. 70-100 km depth. Finally, the peridotites were exposed to crust-derived B-rich fluids with low δ11B during exhumation and amalgamation with crustal units, forming lizardite veining during greenschist-facies overprint.

This study shows that mantle olivine may scavenge significant amounts of B from percolating fluids by diffusive re-equilibration or dynamic recrystallisation, lowering the B content of such fluids and potentially modifying their B isotopic composition.

How to cite: De Hoog, C.-J., Hattori, K., and Clarke, E.: Boron isotope systematics of Higashi-akaishi mantle wedge peridotites (Sanbagawa belt, Japan): implications for fluid recycling in subduction zones, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-12019,, 2021.

Serge Lallemand and Diane Arcay
To address the question of the triggers and mechanisms involved in the process of subduction zones formation, we have explored all available clues attesting for subduction initiation (SI) during the Cenozoic. We have defined several stages starting from incipient-diffuse, incipient-localized, achieved to self-sustained subduction. We have also included prematurely extinct, i.e., aborted, subduction attempts in order to better understand the reasons for the stoppage of subduction process, and thence to specify the conditions of success. This comprehensive study led us to observe that new subductions regularly initiate at a mean rate of about once per million years. Two third of those initiated during the Cenozoic are still active. A majority initiated at the transition between an ocean and a continent, a plateau or a volcanic arc. Lithospheric forces are needed for SI with the help of mantle forces in one third of the cases. Multiple triggers, like a collision followed by a slab breakoff, are common. The stress at SI is compressional in most cases if not all and oriented oblique to the nascent plate boundary in more than half of the cases. The nascent plate boundary generally reactivates a former lithospheric fault, most often with a change in its kinematics (conversion of spreading center, normal or detachment fault or transform fault) or using the same kinematics when reactivating former subduction faults. There is no rule regarding the age of the subducting plate which varies from 0 to 140 Ma in the studied examples. In the same vein, the subducting plate is not necessarily older than the overriding plate. Both situations are equally observed.

How to cite: Lallemand, S. and Arcay, D.: Lessons learned from the study of 68 Cenozoic occurrences of subduction initiation, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-12194,, 2021.

Ben Harris, Cees-Jan de Hoog, and Ralf Halama

Nitrogen recycling from the Earth’s surface to the mantle through subduction zones is a key component of the long term global nitrogen cycle. Data on the nitrogen contents of formerly subducted rocks is key to constraining this flux and to understanding nitrogen behaviour during subduction dehydration. Studies have so far been restricted to analyses of whole rocks or mineral separates, which masks textural controls and mineral heterogeneity. Here we present the first in situ SIMS analyses of nitrogen contents in white micas and other minerals from a suite of subduction-related crustal rocks. We determine the nitrogen distribution in these rocks and explore the behaviour of nitrogen, compared to other fluid-mobile elements, during subduction and fluid-rock interaction. Samples from three localities were investigated: blueschist and eclogite from the Raspas Complex, Ecuador; blueschist and eclogite from the Franciscan mélange (Jenner, California); eclogite and garnet-phengite quartzite from Lago di Cignana, Italy.

Our data confirm that white mica (phengite, paragonite) is the primary host for nitrogen across all samples. Both phengite and paragonite contain substantial amounts of nitrogen (up to 320 ppm), but the concentrations vary widely across different samples. Chlorite replacing garnet in eclogites and blueschists contains little nitrogen. In contrast, chlorite occurring with garnet, phengite (108 - 270 ppm N), glaucophane and titanite in the matrix of a blueschist from Jenner contains measurable quantities of nitrogen (10 - 83 ppm). Other minerals (clinopyroxene, amphibole, epidote, titanite, garnet) contain little nitrogen (<5 ppm) in all samples.

A blueschist from Raspas contains coexisting phengite and paragonite, in addition to garnet, glaucophane, epidote, and accessory albite and carbonate. Nitrogen preferentially partitions into phengite (117 - 243 ppm) over paragonite (31 - 118 ppm). Albite also contains some nitrogen (15 ppm). Silicon contents of phengite vary from 3.32 – 3.40 a.f.u. Decrease in silicon is correlated with decrease in nitrogen and boron, and increase in lithium. These trends can be explained by growth of paragonite during retrograde fluid-rock interaction and redistribution of these elements between phengite, paragonite and glaucophane.

Variability in nitrogen concentrations in other samples which have undergone peak or retrograde fluid-rock interaction, and contain only phengite as a nitrogen-bearing phase, cannot be explained by redistribution. Different samples display either no change in nitrogen, or addition of nitrogen during fluid-rock interaction, as recorded by different generations of phengite. No correlation between nitrogen contents of the samples and P-T conditions was observed, but this was likely due to the large range of protoliths in this study.

Our results demonstrate that nitrogen behaviour during fluid-rock interaction is complex and can be variable between samples, and that in situ data can inform understanding of the processes controlling N distribution.

How to cite: Harris, B., de Hoog, C.-J., and Halama, R.: In situ measurements of nitrogen contents in formerly subducted rocks reveal variable behaviour of nitrogen during fluid-rock interaction., EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-12745,, 2021.

Işıl Nur Güraslan and Şafak Altunkaynak

South Shetland Islands in Western Antarctica is dominated by a widespread magmatism through Meso-Cenozoic due to the magmatic arc created by the subduction of Phoenix plate along the South Shetland trench. Within the scope of 4th Turkish Antarctic Expedition (TAE-IV) and Turkey-Poland Bilateral cooperation, field studies were conducted in Admiralty Bay (King George Island) that host various magmatic units  in order to understand the magmatic evolution beneath Meso-Cenozoic Western Antarctica.

Magmatic products consists of Paleocene-Eocene aged volcanic and subvolcanic rocks in Admiralty Bay. Volcanic rocks are represented by terrestrial lavas and pyroclastic rocks (agglomerates, tuffs and volcanic breccias) while subvolcanic rocks consist of dykes and stocks. According to the petrographic investigations, volcanic and subvolcanic rocks in the area mostly display disequilibrium textures such as sieve textures and embayments in plagioclase and pyroxenes, patchy and oscillatory zoning in different generations of plagioclases and the existence of K-Feldspar xenocrysts with reaction rims along the borders.

Geochemically, the compositions of the magmatic rocks in the study area range from dacite to basalt. Volcanic and subvolcanic rocks show similar geochemical signatures. The samples show mostly calc-alkaline affinities. There are two predominant compositional variations, felsic and intermediate-mafic. Their MgO content ranges within 0.28-1.20 wt. % for the more felsic lavas and 2.78-5.24 wt. % for intermediate-mafic lavas. Their Al2O3 contents are relatively high (14.91-24.29 wt. %). The samples are slightly enriched in large ion lithophile elements (LILE) and light rare earth elements (LREE) compared to HFSE and HREE. The samples display high Th/Yb ratios ranging from 3.78 to 0.69. Strong depletions in Nb and Ti elements are observed as typical indicators for subduction zone magmatism. Although most of the samples show similar patterns in spider diagrams, a strong discrepancy is seen in immobile elements such as Hf and Zr, resulting in positive anomalies in felsic and negative anomalies in intermediate-mafic rocks. Similarly, negative Eu anomalies observed only in the felsic rocks. Eu/Eu* ratios varies within 0.59-0.71 for felsic rocks, and 0.85-1.12 for intermediate-mafic rocks. These different patterns in different compositions suggest an open system differentiation for the melt evolution. Petrographic and geochemical evaluations indicate that the magma beneath Meso-Cenozoic Western Antarctica is originated from lithospheric mantle metasomatized by subduction components, and fractional crystallization/assimilation fractional crystallization contributed to the magmatic evolution.




How to cite: Güraslan, I. N. and Altunkaynak, Ş.: Petrography and geochemistry of magmatic rocks from Admiralty Bay, King George Island (South Shetland Islands, Antarctica): Preliminary results, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-13594,, 2021.

Songqiao Shawn Wei, Peter Shearer, Carolina Lithgow-Bertelloni, Lars Stixrude, and Dongdong Tian

The Hawaiian-Emperor seamount chain that includes the Hawaiian volcanoes is created by the Hawaiian mantle plume. Although the mantle plume hypothesis predicts an oceanic plateau produced by massive decompression melting during the initiation stage of the Hawaiian hotspot, the fate of this plateau is unclear. We discovered a megameter-scale portion of thickened oceanic crust in the uppermost lower mantle west of the Sea of Okhotsk by stacking seismic waveforms of SS precursors. We propose that this thick crust represents a major part of the oceanic plateau that was created by the Hawaiian plume head about 100 Ma ago and subducted 20–30 Ma ago. Our discovery provides temporal and spatial clues of the early history of the Hawaiian plume for future plate reconstructions.

How to cite: Wei, S. S., Shearer, P., Lithgow-Bertelloni, C., Stixrude, L., and Tian, D.: Oceanic plateau of the Hawaiian mantle plume head subducted to the uppermost lower mantle, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-13874,, 2021.

Jiaqi Li, Min Chen, and Thomas P. Ferrand

At the top of the mantle transition zone, it is commonly accepted that olivine (α) transforms to wadsleyite (β) at about 410 km depth under equilibrium conditions, i.e., a pressure of ~ 14 GPa and a temperature of ~ 1350 °C. The subsequent wave speed increase upon the α-β phase transition leads to the discovery of the 410-km discontinuity as a global feature seismologically. However, the complex topography of the “410-km discontinuity” is unclear within cold subducted oceanic lithospheres sinking into the lower mantle, partly due to the sparsity of seismic waves sampling the pertaining complex 3-D structures.

This study uses triplicated P waves (~ 2 seconds), most sensitive to the 410-km discontinuity, to invert for the characteristic parameters of its depth and radial wave speed gradients near the discontinuity. Six distinct wave propagation directions are investigated for a carefully chosen earthquake. These directions are sub-parallel to the slab depth contours in the Kuril subduction zone to guarantee a simplified layered earth modeling. Our results show azimuthal variations of the discontinuity depth either above or within the slab.

For example, the 410-km discontinuity is uplifted by 5-10 km at a depth of about 100 km above the slab upper interface. The uplift increases up to 15-20 km when the 410-km discontinuity is closer to, i.e., only 50 km above, the cold slab. This observation is consistent with the expected phase transition in equilibrium with temperatures greater than 1000°C. In contrast, within the cold slab (< 1000°C), the α-β transition exhibits drastic variations of P-wave speed. Our non-gradient-based inversion results show optimal models that place the following unique seismic constraints: 1) a significant P-wave speed increase within the slab (+5.5 ±1.5 %) compared to the ambient mantle; 2) a zone of extremely low wave speed (LVZ) within the slab with a P-wave speed reduction of -14 ±4 %. The observed LVZ is located near a depth of 350 km with an apparent thickness of 15-30 km, which can be much thinner in the direction normal to the slab upper interface.

These observations indicate a layer of destabilized olivine (LVZ) exists inside the slab. The α-β transition involves atomic diffusion highly dependent on temperature. Once olivine becomes unstable within a cold wedge, it cannot directly transform into wadsleyite. The drastic P-wave speed reduction is likely caused by the sudden grain-size reduction induced by the phase transition, and possibly also by the transient (meta)stability of an intermediate phase, the ω-olivine, under substantial shear stress during the transformation within the cold wedge of the sinking slab.

How to cite: Li, J., Chen, M., and Ferrand, T. P.: Olivine-Wadsleyite Transformation within the Subducting Pacific Slab in Kuril, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-13913,, 2021.

Ana M. Negredo, Carlos Clemente, Eugenio Carminati, Ivone Jiménez-Munt, Jaume Vergés, Javier Fullea, and Montserrat Torné

A number or previous studies indicate the possibility of post-collisional continental delamination in the northern Apennines. In this study we investigate by means of thermo-mechanical modelling the conditions for, and consequences of, delamination postdating continental subduction in this region. The modelled cross-section strikes approximately from Corsica to the Adriatic Sea. The initial model setup simulates the scenario at ca 20 Ma, where the oceanic lithosphere of the westward-subducting Adria plate was entirely consumed and some amount of continental subduction also occurred. The negative buoyancy of the slab remnant, together with the low viscosity of the dragged down lower continental crust, promote lithospheric mantle sinking into the mantle and asthenospheric upwelling and its lateral expansion along the lower crust. Consistent with geological data, the compressional front produced by delamination migrates about 260 km eastwards, causing a similar migrating pattern of extension from the northern Tyrrhenian Sea, to Tuscany and the seismogenically active Apennines backbone. The topographic response is computed by means of a true free-surface approach, and reflects the same eastward migrating pattern of uplift caused by asthenospheric inflow in the internal part of the system and crustal thickening; and subsidence at the front caused by the negative buoyancy of the sinking Adria slab. The conditions for the occurrence of magmatism and high heat flow beneath Tuscany are also explored. Simulations resulting in fast migration of the delamination front predict slab necking and breakoff, which could be consistent with the slab window observed beneath the central Apennines. Subcrustal seismicity beneath the Northern Apennines can be interpreted as the result to this incipient slab necking. This is a GeoCAM contribution (PGC2018-095154-B-I00)

How to cite: Negredo, A. M., Clemente, C., Carminati, E., Jiménez-Munt, I., Vergés, J., Fullea, J., and Torné, M.: Thermo-mechanical modelling of subducting plate delamination in the northern Apennines, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-15085,, 2021.

Sudeshna Basu and Adrian Jones

Nitrogen in the mantle exists in various speciation depending on oxidation conditions. Based on thermodynamic calculations, it predominantly occurs as N2 under relatively oxidized conditions and as NH4+ when conditions are reducing (Mikhail and Sverjensky, 2014). The speciation has an effect on its compatibility behaviour, being more soluble in melts when in reduced form, while the reverse is true for fluids (Mysen, 2019).  Carbonatites are very important to constrain the nitrogen composition of the mantle with important implications for the subduction history of the Earth. Carbonatites entrain components from different reservoirs including the deep Earth near the core-mantle boundary and, temporally encompass a wide range in age (Dauphas and Marty., 1999; Basu and Murty, 2015).  Studies from some young carbonatites in India from Sung Valley (107 Ma) and Ambadongar (65 Ma) indicate that the nitrogen is present as more than one component in the source, unhomogenised and hence identifiable, that can be related to their occurrence in more than one chemical form (Basu and Murty, 2015).

The emergence of efficient and long-lived plate tectonics is thought to be as early as Late Archean, based on nitrogen isotopic composition of placer diamonds from Witwatersand from the Kapvaal craton (Smart et al., 2016). While this may represent a global occurrence, we have undertaken a more robust study with carbonatites ranging in age from 2500 to 770 Myrs, with the goal of identification of the initiation of subduction in a global scale and investigation of any change in the nitrogen stored in the mantle with time. The carbonatites studied are from Khambamettu (2.5 Ga), Hogenakal (2.4 Ga), and Sevattur (770 Ma), located in the southern part of India. Calcites and apatites separated from the host rocks were analysed by vacuum crushing. The apatites were also analysed by stepwise pyrolysis to release and decouple different components at different temperatures. In the carbonates, the nitrogen contents vary from 140 to 1507 ppb with accompanying δ15N ranging from 4.7±0.4 to 11.7±1.3 ‰. The nitrogen in the apatites from Hogenakal and Khambamettu show depleted signatures with δ15N as low as ~ -22 ‰, accompanied by low nitrogen content of ~ 60 to 140 ppb. The apatite from the younger Sevattur complex is comparable to the carbonates in terms of both concentration and isotopic composition. This can be related to increase in nitrogen input via subduction with time during Earth’s history since the Proterozoic, transported to the deep mantle, consequently overprinting any primordial signatures inherited from precursor building material such as the echondrites.


Basu, S. and Murty, S.V.S. Journal of Asian Earth Sciences 107: 53-61 (2015).

Dauphas, N. and Marty, B. Science 286(5449): 2488-2490 (1999).

Mikhail, S. and Sverjensky, D.A. Nature Geoscience 7(11): 816-819 (2014).

Mysen, B. Prog Earth Planet Sci 6, 38 (2019).

Smart, K.A., Tappe, S., Stern, R.A., Webb, S.J. and Ashwal, L.D.  Nature Geoscience 9(3): 255-259 (2016). 10.1038/NGEO2628

How to cite: Basu, S. and Jones, A.: Subduction in early Proterozoic mantle: Implications from nitrogen in carbonatites, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-15953,, 2021.

Iris van Zelst, Timothy J. Craig, and Cedric Thieulot

The thermal structure of subduction zones plays an important role in the seismicity that occurs there with e.g., the downdip limit of the seismogenic zone associated with particular isotherms (350 °C - 450 °C) and intermediate-depth seismicity linked to dehydration reactions that occur at specific temperatures and pressures. Therefore, accurate thermal models of subduction zones that include the complexities found in laboratory studies are necessary. One of the often-ignored effects in models is the temperature-dependence of the thermal parameters such as the thermal conductivity, heat capacity, and density. 

Here, we build upon the model setup presented by Van Keken et al., 2008 by including temperature-dependent thermal parameters to an otherwise clearly constrained, simple model setup of a subducting plate. We consider a fixed kinematic slab dipping at 45° and a stationary overriding plate with a dynamic mantle wedge. Such a simple setup allows us to isolate the effect of temperature-dependent thermal parameters. We add a more complex plate cooling model for the oceanic plate for consistency with the thermal parameters. 

We test the effect of temperature-dependent thermal parameters on models with different rheologies, such as an isoviscous wedge, diffusion and dislocation creep. We find that slab temperatures can change by up to 65 °C which affects the location of isotherm depths. The downdip limit of the seismogenic zone defined by e.g., the 350 °C isotherm shifts by approximately 4 km, thereby increasing the maximum possible rupture area of the seismogenic zone. Similarly, the 600 °C isotherm is shifted approximately 30 km deeper, affecting the depth at which dehydration reactions and hence intermediate-depth seismicity occurs. Our results therefore show that temperature-dependent thermal parameters in thermal models of subduction zones cannot be ignored when studying subduction-related seismicity. 

How to cite: van Zelst, I., Craig, T. J., and Thieulot, C.: The effect of temperature-dependent thermal parameters in thermal models of subduction zones, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-120,, 2021.