= Solution
At a <convergent margin>, the incoming <oceanic lithosphere> is a chemically and mechanically layered package: <sediments>, <oceanic crust> and <lithospheric mantle>. <Subduction> does not deliver this whole package unchanged to one destination. Solids can be accreted, underplated, exhumed or recycled deeply, while released fluids and melts carry selected components to the <overriding plate>. The outcome depends on incoming composition and hydration, <slab> age, convergence rate, thermal structure, interface coupling and subsequent tectonic history.
\Image[/past-exam-of-the-plate-tectonics-course-of-the-university-of-oxford/2025-subduction-fates.png]
{title=Diverging pathways of <sediments>, <oceanic crust> and <lithospheric mantle> at a <convergent margin>}
{description=Original process diagram. Solid-material pathways are separated from fluid and melt transfer. The arrows represent possible routes, not universal proportions or fixed reaction depths; temporary storage and later remobilization are possible.}
\b[The incoming material already contains a history.] <Sediments> include terrigenous debris, biogenic <carbonates>, siliceous deposits, <clay minerals>, pore <water> and <organic matter>. <Oceanic crust> consists mainly of <basalt>, sheeted dikes and <gabbro>; <hydrothermal alteration> can add <water>, <carbonates> and mobile elements before <subduction>. Beneath it, residual <peridotite> forms the <lithospheric mantle>. Faulting at a spreading environment, a transform boundary or an outer-rise bending region can let <seawater> enter and cause <serpentinization>. The <lithospheric mantle> is not uniformly hydrated to its full thickness. Consequently, sediment thickness alone does not measure the incoming volatile inventory.
\b[Sediments: separation starts near the trench.] Compaction and increasing effective <pressure> expel pore fluids, which can escape through the <forearc>, transport dissolved constituents and influence <pore pressure> on the plate interface. Some <sediments> are scraped off into an <accretionary prism>, shortened by thrusts and retained in the upper plate. Others pass beneath the leading edge, then detach and undergo <tectonic underplating> beneath the <forearc>. Deeper underplated material may return to the surface during <tectonic exhumation>, preserving a record of burial and deformation. At a sediment-starved erosive margin, little prism may form, and <tectonic erosion> can entrain material from the <overriding plate> as well. Thus net continental gain and net continental loss are both possible; they can even alternate at one margin.
<Sediments> that remain attached to the <slab> undergo <metamorphism>, <dehydration>, and potentially <decarbonation> or carbonate dissolution. The released fluids can transport <boron>, <barium>, <strontium>, <lead> and other soluble constituents. <Carbon> can leave as dissolved species, remain in <carbonate>, or occur as reduced <carbon> depending on <redox> conditions and reaction history. Along sufficiently hot paths, sediment-derived melts or mixtures of sediment and altered rock can transfer material into the <mantle wedge>; buoyant mixed material may rise as a <mélange diapir>. Cold paths can instead retain more solid material and some volatile-bearing phases to greater depths. These are alternative reaction and transport pathways, not a single compulsory sequence. A carbon-budget study explicitly includes fluid dissolution, sediment diapirism and storage in the upper plate, showing why volcanic <carbon dioxide> alone cannot close the whole <subduction> <carbon> budget.<Reevaluating carbon fluxes in subduction zones, what goes down, mostly comes up by Peter B. Kelemen and Craig E. Manning>[2015, proposed pathways and reservoir accounting]{ref}
\b[Oceanic crust: metamorphic transformation and chemical filtration.] Altered <basalt> and <gabbro> are progressively buried. Depending on their <pressure>-<temperature> paths and reaction kinetics, they can pass through <blueschist facies> or warmer assemblages and eventually form <eclogite>, dominated by <garnet> and <omphacite>. <Eclogitization> increases <density> and can strengthen negative <slab buoyancy>, but a kinetic delay can leave metastable rock. It is incorrect to assign every <slab> the same sequence or the same fixed depths for these changes.
During burial, breakdown of hydrous phases releases <water>-rich fluids; different <mineral> assemblages retain different quantities of <water>. The fluids migrate through or along the interface and into the <mantle wedge>, where they cause <metasomatism> and lower the <solidus> of <peridotite>. <Flux melting> of the <mantle wedge> generates much of the primary magma feeding a <volcanic arc>. This is different from saying that the entire descending <oceanic crust> melts beneath the arc. Most of its refractory mass can remain solid. <Partial melting> of <oceanic crust> is more plausible for unusually hot or young <slabs>, or other favorable thermal conditions, and produces a distinct crustal contribution.
The depth of <dehydration> is controlled by thermal structure and composition. Global thermal-petrological calculations, for example, predict more efficient deep <water> transport in old, fast, cold <slabs> than in hot ones, with substantial differences between volcanic <crust>, <gabbro> and hydrated <peridotite>. Such a calculation supports conditional deep retention; its modeled proportions should not be imposed on every margin.<Subduction factory: 4. Depth-dependent flux of H2O from subducting slabs worldwide by Peter E. van Keken and others>[2011, abstract and sections 3–4]{ref}
Mechanical detachment adds more branches. Upper-crustal slices, seamounts or oceanic plateaus may be accreted or underplated instead of continuing down the <slab>. A complete or partial <ophiolite> can be emplaced onto another plate by <obduction> in suitable convergence histories. Most intact <oceanic crust> is not saved this way. High-<pressure> fragments may be entrained in an interface channel or <mélange> and exhumed; their small preserved volume should not be mistaken for the dominant global mass flux. Magma generated after slab-derived transfer may also pond and crystallize beneath the upper <crust> by <magmatic underplating>, which is a different process from mechanical <tectonic underplating> of solid slices.
\b[Lithospheric mantle: residual rock and a variable water carrier.] Most incoming <lithospheric mantle> is depleted <peridotite> left after extraction of ridge melts. Its cold, dense, mechanically coherent portion contributes to <slab pull> and generally sinks with the plate. Hydrated portions contain <serpentine> and possibly other hydrous <minerals>. Breakdown of these phases releases <water> at conditions that depend on the thermal path, feeding further <metasomatism> of the <mantle wedge> and sometimes <dehydration embrittlement>. The latter is a proposed contributor to intermediate-depth <earthquakes>, not an exclusive explanation for all such events. Very cold paths may transfer some <water> onward in high-<pressure> hydrous phases or as hydrogen incorporated in nominally anhydrous <minerals>.
Do not confuse the incoming hydrated <lithospheric mantle> with newly serpentinized <forearc mantle> in the <overriding plate>. The latter receives slab-derived fluids and can temporarily store their <water>. Hydrated, relatively buoyant interface material can facilitate return flow and <tectonic exhumation>, while the colder deeper bulk of the <lithospheric mantle> continues down. Some <peridotite> may also be accreted or obducted with an <ophiolite>. These exchanges mean that even an exposed <serpentinite> requires structural and chemical context before it can be assigned to its original plate.
\b[Deep recycling and eventual return.] The residual <eclogite>, remaining <sediments> and <lithospheric mantle> can enter the deeper <mantle>. A <slab> can temporarily stagnate around the <mantle transition zone> or penetrate into the lower <mantle>, depending on <buoyancy>, phase transformations, geometry and <mantle viscosity>. Crustal material can remain as compositionally distinct domains, mix, or react with surrounding <peridotite>; it need not retain the near-surface <eclogite> assemblage indefinitely as <pressure> rises. Subsequent <mantle convection> can return recycled components toward spreading regions or <mantle plumes>, where <partial melting> incorporates them into new <basalt>. A primary study of Hawaiian <basalt> illustrates a specific model in which melts derived from recycled crust react with <peridotite> to form a <pyroxenite> source. It is an interpretation of combined chemical evidence, not a unique label attached to every enriched <basalt>.<An olivine-free mantle source of Hawaiian shield basalts by Alexander V. Sobolev and others>[2005, recycled-crust reaction and melt-source model]{ref}
\b[How the pathways can be observed.] Seismic reflection profiles, drilling, wedge geometry and balanced structural sections constrain accretion, fluid escape and <tectonic erosion>. <High-pressure metamorphism>, mineral zoning, <radiometric dating> and <pressure-temperature-time paths> in exhumed rocks reveal burial, reaction and return. Their reconstructable histories are especially useful when deformation and fluid inclusions are linked to the dated <mineral> growth. <Earthquake> distributions, <seismic tomography>, <seismic attenuation> and <electrical conductivity> constrain the geometry and physical state of the <slab> and <mantle wedge>, although fluids, melts and <temperature> can produce nonunique signals.
<Arc magmas> carry complementary chemical evidence: relatively mobile <barium>, <boron>, <lead> or <strontium> can trace fluid-mediated transfer, whereas sediment-sensitive <thorium> and <isotope ratios> can help identify other contributions. Element mobility changes with fluid composition and <pressure>-<temperature> conditions, so no one element is an infallible tracer. <Boron isotope ratios>, radiogenic <strontium isotope ratios> and <lead isotope ratios> should be interpreted with measured incoming <sediments> and altered <oceanic crust>. Arc gases and <forearc> fluid discharge constrain volatile return; <ocean-island basalts>, <mantle xenoliths> and <diamond inclusions> provide possible evidence of deeper recycling. Each observation samples only part of the system.
For any element $X$, a useful control-volume statement is
$$
F_X^{\rm incoming}+F_X^{\rm upper\ plate\ erosion}
=F_X^{\rm accretion}+F_X^{\rm return}+F_X^{\rm deep}
+\frac{dM_X^{\rm stored}}{dt}.
$$
Here permanent accretion is counted outside the chosen reactive control volume, return includes fluid, melt and exhumed-solid exports, and transient underplating or reservoir growth contributes to storage until exported. For a different boundary choice, the terms must be regrouped consistently. The three incoming layers can have very different transfer efficiencies for bulk rock, <water>, <carbon> and a particular <trace element>. A small mass of escaping fluid can dominate an element's return while most of the solid <slab> continues into the deep <mantle>.
Back to article page
