Composition and Significance of Mariana Trough Basalts Julian Pearce (Cardiff)

Composition and Significance of Mariana Trough Basalts with contributions from: Bob Stern (Dallas) Sherman Bloomer (Corvallis) Patty Fryer (Honolulu) Jon Woodhead (Melbourne)

The Mariana Trough Setting Unzipping of island arc with maximum extension in the centre and a N-S trend of rifting to drifting

Mariana Trough: A Key Locality in BABB Studies  Hart et al. (1972): First analyses of back-arc basalts  Gill (1976): First documentation of the differences between back-arc basin basalts and MORB  Hawkins (1977, 1978): First systematic sampling program for back-arc basins  Tarney et al. (1977): First discovery of island arc basalt from a back- arc basin  Garcia et al. (1979): First analysis of water in a back-arc basin glass  Fryer et al. (1981): First treatment of back-arc basin lavas as a distinct magma type  Sinton and Fryer (1987): First systematic petrogenetic interpretation of BABB  Stolper and Newman (1994): First use of BABB glasses to estimate the composition of the subduction component

Mariana Trough: Type Locality of Back-arc Basin Basalts (BABB) High Volatile Elements notably H (water) Vesicularity Large Ion Lithophile Elements (Ba, Sr, La, Th etc.) Al 2 O 3 Low High Field Strength Elements (Nb, Zr, Y etc.) FeO

How Distinctive are Mariana BABB? SZ

How Distinctive are Mariana BABB?

 14% of Mariana Trough lavas are MORB  6% of Mariana Trough lavas are IAB  80% of Mariana Trough lavas are BABB But it is not obvious whether BABB are simply transitional between MORB and IAB or a truly distinct magma type

What can the Mariana Trough Basalts tell us? 1.Mantle Input 2. Subduction Input 3. Mantle-Subduction interaction 4. Hydrous Ridge Crest Processes (melting, reaction & crystallization)

Mantle Input  Mantle Fertility  Mantle Flow  Mantle Provenance 1

Geochemical Tracing of Mantle Flow 1 Loss of melt fractions during flow to the arc front causes decrease in ratios of highly to moderately incompatible elements – as shown by McCulloch, Gamble, Woodhead and others in the early 1990s. Thus Nb/Yb is a good tracer for mantle flow (provided degrees of melting are high) – Pearce (2005) Nb/Yb decreases

Geochemical Tracing of Mantle Flow Izu: corner flow Lau: one-way ‘sideways’ flow Scotia: two- way sideways flow followed by corner flow Mariana: multi-centre upwelling followed by along-axis and corner flow Red – high Nb/Yb Blue – low Nb/Yb Pearce & Stern, 2006

Mariana Trough Mantle Flow Consistent with mantle upwelling in several centres within the basin and complex flow pattern Pearce et al., 2005

Mariana Trough Mantle Flow Pozgay et al. (2007)

Mantle Provenance Isotopes can be used to fingerprint the mantle domain that feeds the arc-basin systems. The Mariana system is dominated by ‘Indian’ mantle sources. But are these from a closing Indian Ocean, from contamination by lithosphere, or from a sub-Pacific plume of ‘Indian’ character? I I I P P P

Key Mantle Questions  What is the origin of the ‘Indian’ component of the Mariana Trough lavas?  How does the fertile mantle enter the Mariana Trough?  How has the mantle evolved with time since subduction started?

What can the Mariana Trough Basalts tell us? 1.Mantle Input 2. Subduction Input 3. Mantle-Subduction interaction 4. Ridge Crest Processes (melting, reaction & crystallization)

Subduction Input  Spatial Variations in the Subduction Component  Composition of the Subduction Components  Origin of the Subduction Components 2

Subduction Component Mapping I: Yb Normalization Normalizing to Yb, i.e. using ratios such as Nb/Yb, greatly reduces the effect of fractional crystallization and crystal cumulation

Subduction Component Mapping II: Splitting Patterns into Components The patterns can be broken up into components and ratios used as proxies of subduction processes

Mapping Total Subduction Input Each system has a different pattern with clear relationships to subduction zone proximity and mantle flow pattern Pearce & Stern (2006)

Mapping Total Subduction Input The subduction zone input in the back-arc basin is highest at the margins as the basin converges with the arc. Within the central part of the basin, there are three regions where the mantle is unaffected by subduction with subduction enrichments between them can also break this down into components Pearce et al., 2005

The Lithospheric Subduction Component The Mariana region is complicated by the presence of a high Nb/Ta component that cannot easily be explained by subduction. They are globally characteristic of small degree melts, but here the degree of melting is high. We tentatively explain them in terms of enrichment of the lithosphere by small-degree melts. The lithosphere is reactivated by arc rifting. Where present, this component has to be numerically subtracted before other components can be studied

Mapping the Lithosphere Subduction Component

Possible Origin of the Lithosphere Component

Mapping the Deep Subduction Component

Mapping the Shallow Subduction Component

Summary of Geochemical Mapping

Subduction Component: HFSE Mobility? Pearce et al (1999): no significant Hf mobility Woodhead et al (2001): Hf and Nd both mobile HFSE immobility an important assumption for computing subduction components: so important to resolve

Subduction Component: HFSE Mobility? Pearce et al (1999): no significant Hf mobility Woodhead et al (2001): Hf and Nd both mobile Green circles = new data from Woodhead et al. (in prep.)

Subduction Component: HFSE mobility? Woodhead et al. (in prep.):still debated but Hf mobility likely limited.

Subduction Component: HFSE mobility? The Scotia system (Barry et al., 2006) is similar, but this time rear-arc and arc edge volcanoes demonstrate clear Hf mobility by addition of an inferred melt component with low Nd/Hf. This is not evident in the Mariana system – except for the high Nb/Ta samples (not shown)

Subduction Component: The Stolper and Newman Conundrum Stolper and Newman’s (1994) innovative use of water in Mariana Trough glasses to predict the subduction component, gives a component with Nd/Hf of c. 6 and Nb of ppm, as well as high Sc. This does not match observation. The question is why?

Subduction Component: The Stolper and Newman Conundrum One explanation is that least-squares methods break down when there are >2 components, some with variable composition.

Key Subduction Questions  Is the ‘lithosphere’ component really derived from the sub-arc lithsophere and, if so, when and how was it introduced?  Why does the Stolper and Newman study give a major HFSE component in the aqueous component: what do the data really show?  When and how were the subduction components added to the Mariana Trough mantle source?

Factors Controlling the Composition of Back-arc Basin Basalts 1.Mantle Input 2. Subduction Input 3. Mantle-Subduction interaction 4. Ridge Crest Processes (melting, reaction & crystallization)

Mantle-Subduction Interaction The back-arc subduction component could be by: Incorporation of previously-enriched lithosphere. Mixing of sub-arc and incoming mantle Direct addition from the subduction zone Via subduction- modified mantle melts 2 1 3

Subduction Zone Input: Timing Fretzdorff et al., 2003: Subduction component must have been added to the back-arc within 350ka. Isochron may be real but may also represent mixing of fluid and melt components E4 E8 fluid (same result as Peate et al. (2001) for Valu Fa Ridge, Lau Basin A similar study is needed on the Mariana Trough

Subduction Zone Input: Role of Mixing If the back-arc gets its subduction component by mixing with sub- arc mantle (the Martinez-Taylor model), why is it not simply a diluted version of the arc? It does, however, explain why back- arc basins are not more shallow

Factors Controlling the Composition of Back-arc Basin Basalts 1.Mantle Input 2. Subduction Input 3. Mantle-Subduction interaction 4. Ridge Crest Processes (melting, reaction & crystallization)

Trace Element Data Indicate Shallow Melting, and F increasing with X H20 Mariana Trough glasses plot on a spinel lherzolite melting trend, indicating shallow melting processes. The degree of melting varies (as Stolper and Newman first showed) as a function of water content. Exact calibration for F is difficult without knowing source composition, however

And an Increase in F from MORB through BABB to IAT Use of Nb/Yb (the flow tracer) as well as Yb deals with variations in source composition Pearce & Stern (2006)

Summary 1.Mantle Input Mantle rises diapirically at points within the basin; ‘Indian’ provenance, but cause not known 2. Subduction Input Lithospheric signals during arc rifting; deep subduction signals between ‘diapirs’; shallow subduction signal at centre and south of arc; cool subduction (no HFSE enrichment) 3. Mantle-subduction Interaction No evidence yet for mixing of incoming mantle with sub- arc mantle 4. Ridge Crest Processes Degree of melting increases from MORB through BABB to IAT (5-30%)