Thomas Thompson Evidence for shallow dehydration of the subducting plate beneath the Mariana forearc: New insights into the water cycle at subduction zones Julia Ribeiro , Robert J. Stern , Katherine A. Kelley , Alison M. Shaw , Fernando Martinez , Yasuhiko Ohara 5, 6 2 4 3 1: University of Texas at Dallas; 2: GSO, University of Rhode Island; 3: Woods Hole Oceanographic Institution; 4: SOEST, University of Hawai'i at Manoa; 5: Hydrographic and Oceanographic Department of Japan, Tokyo, Japan; 6: JAMSTEC, Yokohama, Japan; author email: [email protected]. 1 1 Guam Eocene forearc ~3cm/yr Guam MGR Santa Rosa Bank WSRBF FNVC Today W. Mariana Ridge Eocene forearc ~3cm/yr Challenger deep Guam Eocene forearc ~3cm/yr 0 100 km Guam Eocene forearc ~3cm/yr f SEMFR SEMFR 143 E 144 E 145 E 12 N 13 N 14 N o o o o o o M a la g u a n a - G a d a o - R i d g e Active magmatic arc WSRBF FN VC Ba/Th <150 150 - 200 200 - 250 250 - 300 > 300 shallow subduction component a 2 FNVC 143 E 144 E 145 E o o o WSRBF b 50 km 100km 100km 150 km 150 km 200 km 200 km 50 km 100km 100km 150 km 150 km 200 km 200 km SE shallow subduction component Rb/Th <=10 10 - 15 15 - 20 20 - 25 >25 0 1 2 3 0 200 400 600 800 1000 1200 slab fluid seawater contamination 400 bars (~ seafloor) 500 bars 1000 bars 1500 bars 2000 bars 100 200 300 400 500 600 700 likely degassed minimally degassed likely degassed minimally degassed 0 100 200 300 400 500 600 700 0 1 2 3 0 200 400 600 800 1000 1200 H O (wt%) S (ppm) 2 2 CO ppm H O (wt%) 2 0 1 2 3 0 1 2 3 10 100 1000 .1 1 Cl/K Ba/Nb MGR glass, this study SE SEMFR glassy rind SE SEMFR melt inclusion NW SEMFR melt inclusion NW SEMFR glassy rind SE SEMFR glassy rind SE SEMFR melt inclusion NW SEMFR melt inclusion NW SEMFR glassy rind likely degassed glass : least degassed glass : fractionation 48 50 52 54 56 58 60 0 1 2 3 H O (wt%) SiO (wt%) 2 2 c- Opening of the Mariana Trough at ~5Ma d- Formation of SEMFR crust at 2.7 - 3.7 Ma e- Formation of SEMFR rift structures < 2.7 Ma a- Filtering for minimally degassed volatile contents b - Minimally degassed volatiles 200 μm d- Olivine-hosted melt inclusion enclosing a gas bubble Chromium spinel c- Water re-equilibration process e - Assimilation of Cl-rich material 85 86 87 88 89 90 91 92 93 94 95 0.10 0.15 0.20 0.25 0.30 CaO Fo groundmass xenocrysts b 85 86 87 88 89 90 91 92 93 94 95 0.10 0.15 0.20 0.25 0.30 SE SEMFR ol phenocryst NW SEMFR ol phenocryst SE SEMFR ol xenocryst NW SEMFR ol xenocryst OSMA SEMFR peridotites (Mischibayashi, 2009, G-3) SE. Mariana forearc peridotites (Ohara and Ishii, 1998, IAR) Mariana Trough peridotites (Ohara et al., 2002, Contrib. Min. & Petrol.) m a n t l e d e p l e t i o n 95 90 85 80 0 20 40 60 80 100 % Cr# spinel % Fo olivine 95 90 85 80 0 20 40 60 80 100 c 3 - Melt inclusions are hosted by olivine xenocrysts from the forearc mantle 2- Data filtering: degassing, water re-equilibration and assimilation of Cl-rich material 5- Other slab proxies also peak at 70-80 km depth to the subducting plate 1- SEMFR: a recently discovered site where extensionally induced magmatism occurred in Pliocene time unusually close to the trench Filtering the volatile contents of the glassy rinds and olivine-hosted melt inclusions. (a-b) Glasses with CO2>50 ppm and S>500 ppm are considered to be minimally degassed and they have retain most of their original water content. c) Effect of crystal fractionation on the water contents of the SEMFR glasses. Correlation between H2O and SiO2 (element not affected by re-equili- bration process) suggests that the water content of the melt inclusion did not re-equilibrate with the olivine host. e) Cl/K vs Ba/Nb diagram of Kent et al. (2002, EPSL) used to discriminate the effect of seawater alteration vs slab-derived fluids in SEMFR glasses. The lack of clear positive correlation suggests that the composition of SEMFR glasses may be affected by Cl assimilation. Summary: T53A-4653 a) Location map of the southernmost Mariana convergent margin. The blue box highlights the area in B. b) Bathymetric map of the S. Mariana intraoceanic arc, with location of the dives - dredges performed during Yokosuka and Thomas Thompson cruises (TN273 dredges have a "D"). Panels c-f sketch the geodynamic evolution of SEMFR since the opening of the Mariana backarc Trough (the Malaguana-Gadao Ridge: MGR) ~ 5 Ma ago, resulting in stretching of the forearc lithosphere and forming the SEMFR ~ 3.7 - 2.7 Ma ago. Since < 2.7 Ma, magmatic activity stopped and SEMFR is dominated by post-magmatic rifting. FNVC: Fina-Nagu Volcanic Chain (extinct arc chain); WSRB: W. Santa Rosa Bank; WSRBF: Fault separating WSRB and SEMFR. Water is efficiently recycled at subduction zones. It is fluxed from the surface into the mantle by the subducting plate and back to the surface or crust through explosive arc volcanism and degassing. Fluids released from dehydrating the subducting plate are transfer agents of water. Recent thermal models along with estimates of the dehydration efficiency of global subduction zones suggest that the slab in cold subduction systems, such as the Mariana system, can retain mineral bound-water to depths > 80 km before mostly dehydrating beneath the arc volcanoes (Van Keken et al., 2011, JGR; Wada et al., 2008, JGR). Shallow dehydration of the subducting plate is central in assessing the fluid budget; yet, we can rarely sample such shallow subduction fluids released beneath forearcs despite their importance. Here, we investigate the Southernmost Mariana Forearc Rift (SEMFR) where extensionally induced magmatism occurred unusually close to the trench (Ribeiro et al., 2013, IAR; Fig.1), enabling examinantion of the shallow slab-derived aqueous fluids released at ~ 30 to 100 km depth from the subducted plate. Examining the trace element and water contents of olivine-hosted melt inclusions and glassy rinds from the young (< 4Ma) and fresh SEMFR pillowed basalts provide new insights into the global water cycle. SEMFR glassy rinds and olivine-hosted melt inclusions contain ~2 wt % (Fig. 2a-b), which represents minimum water estimates. The olivine-hosted melt inclusions (MI) have the highest subduction-related H2O/Ce (H2O/Ce = 6000-19000; Fig. 4) yet recorded in arc magmas (H2O/Ce < 10600 and global averaged H2O/Ce < 3000). Our findings show that (i) slab-derived fluids released beneath forearcs are water-rich as compared to the deeper fluids released beneath the arc system; and (ii) cold downgoing plates lose most of their water at shallow depths (~ 70-80 km slab depth), suggesting that water is efficiently recycled beneath forearcs. Our results also demonstrate that the downgoing minerals carrying water (i.e., serpentinite, amphibole, chlorite, barite, phengite) mostly break down at depths of ~70 - 80 km to release their water-rich aqueous fluids beneath the forearc (e.g., Bebout et al., 2007, Chem. Geol.; Hattori and Guillot, 2003, Geology; Schmidt and Poli, 1998, EPSL). Using geochemical proxies to map the shallow subduction inputs (Ribeiro et al., 2013, G-3) along the South Mariana Intraoceanic arc. Arrows point toward increasing subduction component. Cs/Th shows the same compositional gradient as Rb/Th. WSRBF is the expression in surface of a slab tear (Fryer et al., 2003, EPSL). The red dashed lines approximates the slab depth (Becker,2005). 4- Evidences for shallow dehydration of the slab: H2O/Ce ratio peaks at 70-80 km depth to the slab NW 6- Water is efficiently recycled beneath forearcs MORB a b forearc peridotites Forearc Arc volcanoes 2 H O/Ce 0 20 40 60 80 100 120 140 160 180 200 0 Slab depth (km) 0 5000 10000 15000 20000 0 5 10 15 20 2 Antilles Centram Marianas Mexico Kamchatka Alaska-Aleutians 0 20 40 60 80 100 120 140 160 180 200 0 5000 10000 15000 20000 0 5000 10000 15000 20000 0 5 10 15 20 Cs/Th H O/Ce 0 5000 10000 15000 20000 0 5 10 15 20 Marianas Alaska-Aleutians Mexico MORB glass average SE SEMFR melt inclusion NW SEMFR melt inclusion MGR glassy rind NW SEMFR glassy rind SE SEMFR glassy rind Depth (km) 0 100 200 300 400 Distance from trench (km) 0 100 150 50 Plate motions slab-derived fluids Oceanic crust subducting plate Backarc basin spreading center Trench Arc volcano forearc serpentinized forearc mantle 2 1 convecting mantle convecting mantle hydrous mantle melting H O-rich fluids less H O-rich fluids solute-rich fluids 3 2 2 peak in H O/Ce, Rb/Th, Cs/Th at ~70-80 km depth 2 H O/Ce, Rb/Th & Cs/Th decrease 2 0.5 mm olivine xenocryst olivine phenocryst spinel a (a) SEMFR melt inclusions are fully enclo- sed by large (> 1 mm) Mg-rich olivines (ol) (Fo90-93; Ribeiro et al., 2013, IAR) that have a Fe-rich reaction rim indicating dis- equilibrium with their host melt. They also hosts chromian spinel (Cr# >= 50) and (b) the Fo and CaO contents of their disequi- librium rim overlap the Fo-CaO contents of the groundmass phenocrysts (< 1 mm), suggesting that the olivine hosts are xenocrysts. Olivine - spinel assemblages plot in the mantle array (Arai, 1994, Chem. Geol.) and in the compositional field of Mariana forearc mantle peridotites (c), suggesting that the olivine xenocrysts are derived from harzburgitic forearc mantle (Ribeiro et al., 2013, IAR). SEMFR melt inclusions record distinctly higher subduction-related H2O/Ce and Cs/Th ratios relative to their host basaltic glassy rinds and to the averaged arc lavas, demonstrating that the melt inclusions captured water-rich fluids released from the shallow part of the subducting plate. H2O/Ce peaks at ~70- 80 km slab depth, demonstrating that the subducting plate mostly dehydrates at depths shallower than beneath the arc magmatic front in cold subduction zones. Also, the high H2O/Ce ratios observed in SEMFR lavas indicates that the slab surface temperature beneath forearcs is likely sub-solidus (< 700°C; Cooper et al., 2012, G-3; Plank et al., 2009, Nature), suggesting that such shallow outfluxes are solute- poor aqueous fluids. Centram. : Central America. We demonstrate that cold downgoing plates mostly dehydrate beneath the forearc at ~ 70 - 80 km depth (1). The fluids progressively become less enriched in water and more enriched in dissolved solutes with increasing slab depth (2-3). Our findings provi- de new constraints for our understanding of depths at which the downgoing plate dehydrates. They support the idea that the thermal state of the downgoing plate con- trols the depth of mineral breakdown and hence, the chemical composition of the fluids. ol core ol rim