Ghost-arc geochemical anomaly at a spreading ridge caused by supersized flat subduction

Mid-ocean ridges produce MORB from variably depleted mantle sources. Some MOR segments exhibit subtle geochemical fingerprints of past subduction influence (ghost-arc signatures), reflecting slab-derived components (sediments, altered oceanic crust, hydrous fluids) introduced into the mantle. Competing hypotheses invoke advection of refractory arc mantle relicts by convection versus preservation of long-lived, near-stationary subduction-modified mantle anomalies (e.g., mélange diapirs that become neutrally buoyant). The SASWIR exhibits a conspicuous, spatially restricted ghost-arc signature whose origin and transport to the ridge remain unresolved. The study aims to determine whether the SASWIR subduction-modified mantle is a long-lived, near-stationary feature linked to Jurassic subduction beneath southwestern Gondwana and to assess the geodynamic mechanism capable of positioning this anomaly beneath the present ridges. The authors combine plate kinematic reconstructions, global seismic tomography, and 2-D thermomechanical modeling to test the role of a supersized Late Triassic–Early Jurassic flat slab in forming and positioning the anomaly.

Prior work documents ghost-arc-like geochemical signals in MORB, attributed to residual subduction inputs in the convecting mantle. Mechanisms proposed include entrainment and removal of refractory arc mantle by convection, and long-lived near-stationary anomalies preserved as mélange diapirs that become neutrally buoyant after arc activity ceases. Interactions between MORs and mantle plumes or delaminated lithosphere can also modify sources. Seismic tomography suggests whole-mantle convection without strict layering, and prior studies proposed that subduction-modified mantle beneath the SASWIR existed since at least the Early Jurassic, with onshore Karoo magmatism showing subduction-related signatures. However, simple convection-driven transport struggles to explain the restricted nature and location of the SASWIR anomaly. Large flat-slab episodes (e.g., Laramide, Eastern China) have been linked to inland migration of arcs and lithospheric modification, suggesting flat subduction can redistribute mantle domains.

• Geochemical comparison: Compiled MORB from SASWIR and Pacific MOR to assess ghost-arc-sensitive ratios (Ce/Pb, Ba/La, Th/La, Ba/Nb, etc.). Noted zircon xenocrysts (2.8 Ga) in SW Indian Ridge MORB supporting sediment involvement.
• Seismic tomography and active mantle mapping: Used SA2019S S-wave tomography to map low-velocity dv/v anomalies (−1% and −2%) at 100 km depth as proxies for active mantle beneath SASWIR. Inspected UU-P07 P-wave tomography for lower-mantle structure.
• Vote maps: Constructed positive wave speed vote maps by stacking 26 global P- and S-wave models to identify robust high-velocity anomalies (slab remnants) and their relation to SASWIR.
• Plate reconstructions: Reconstructed positions of the subduction-related Southern Karoo magmatic province at 183 Ma using multiple mantle reference frames (Müller 2019, Müller 2022, van der Meer 2010, Torsvik & Cocks 2016), and coupled trench positions with tomographic slabs assuming average whole-mantle sinking rates of 1.1 cm/yr to correlate lower-mantle high-velocity anomalies (Georgia Islands slab) with Jurassic trench locations.
• Flat slab kinematics: Estimated flat-slab tip migration rates from arc distance-to-trench through time, using trench–slab-wall distances (2280–2600 km) and a 35 Myr development interval, yielding 5.08–6.0 cm/yr.
• Numerical modeling: Performed 2-D viscoplastic thermomechanical modeling with Underworld2 to simulate development of large-scale flat subduction and associated upper mantle advection. Included temperature- and strain-rate-dependent rheology, partial melt weakening, Drucker–Prager plasticity, and tracked Lagrangian markers to quantify frontal translation of upper mantle. Model domain 5000 km × 660 km; viscosity limited to 1e19–1e24 Pa s; maximum velocities ~8.4 cm/yr. Simplifications include no trench-parallel flow and no explicit melt extraction.
• Kernel density analysis: Built a global kernel density map of MORB with ghost-arc signatures (after filtering out active backarc settings) to assess spatial clustering (SASWIR vs global).

• SASWIR MORB exhibit ghost-arc signatures compared to Pacific MOR: Ce/Pb lower (SASWIR averages 21.99 and 22.768 vs Pacific 26); Ba/La higher (SASWIR 7.86 and 7.27 vs 3.57); Th/La higher (SASWIR 0.099 and 0.074 vs 0.052); Ba/Nb higher (SASWIR 6.09 and 8.24 vs 4.05). Presence of 2.8 Ga zircon xenocrysts in 53–35 Ma MORB from Shaka fracture zone supports sediment contribution to the mantle source.
• Spatial overlap: Reconstruction of the subduction-influenced Southern Karoo magmatic province at 183 Ma coincides with present SASWIR subduction-modified mantle zone (mapped by low-velocity SA2019S anomalies at 100 km), indicating long-lived mantle source beneath the region.
• Tomography correlation: The SASWIR mantle anomaly lies immediately east of a robust lower-mantle high-velocity wall (Georgia Islands slab) at ~1700–2800 km depth, interpreted as fossil slabs (Early Jurassic at upper part; Permian at base). Vote maps across 26 tomography models confirm this spatial relation.
• Coupled reconstructions (with 1.1 cm/yr sinking) align Jurassic trench positions with lower-mantle slab anomalies, but show a landward offset (inboard position) for the Georgia Islands slab consistent with a supersized South Gondwana flat slab and subsequent slab break-off at ~185–180 Ma.
• Flat slab extent and kinematics: Inferred paleo trench–slab-wall distances 2280–2600 km imply flat-slab tip migration rates of 5.08–6.0 cm/yr over ~35 Myr, exceeding known large flat slabs (Laramide, Eastern China ~1500–1900 km).
• Numerical modeling reproduces development of a ~2600–2700 km flat slab in ~35–40 Myr and demonstrates large-scale frontal translation (bulldozing) of upper mantle during inland growth; maximum modeled velocities ~8.4 cm/yr.
• Ridge sampling history: After flat-slab demise and slab break-off, the subduction-modified mantle remained near-stationary, was locally sampled by Karoo magmatism (186–176 Ma), and later by the migrating SASWIR since ~50–35 Ma. The SASWIR anomaly extends ~4400 km along the ridge and remains spatially restricted relative to broader Atlantic MOR.
• Alternative mechanisms (upper mantle convection delivering subduction-modified mantle from active margins, delaminated metasomatized lithosphere blocks, or plume-transported cratonic fragments) are inconsistent with the restricted distribution, tomography, or required scales in SASWIR.
• Introduced concept: Asthenospheric anomaly telescoping—flat subduction frontally shifts subduction-modified asthenosphere far inland, creating a near-stationary intraplate mantle geochemical anomaly later sampled by ridges or plumes.

The findings resolve the origin of the SASWIR ghost-arc signature by linking it to a long-lived, near-stationary subduction-modified mantle anomaly emplaced far inboard during an anomalously large Late Triassic–Early Jurassic flat-slab episode beneath southwestern Gondwana. The spatial correlation between the SASWIR mantle anomaly and the Georgia Islands lower-mantle slab, supported by multi-model vote maps and plate reconstructions with realistic slab sinking rates, indicates that flat-slab inland growth bulldozed subduction-enriched asthenosphere beneath the continental interior. Numerical models demonstrate the dynamical feasibility of developing a ~2.6–2.7 Mm flat slab over ~35–40 Myr and the associated frontal translation of upper mantle. After slab break-off, the anomaly remained quasi-stationary and was sampled first by Karoo plume-related magmatism and later by the east-migrating SASWIR, explaining both the restricted spatial distribution and the persistence of the ghost-arc signal. This mechanism, asthenospheric anomaly telescoping, offers a coherent alternative to broad convection-driven transport or delaminated lithosphere melting, reconciles geochemical data with geophysical structure, and has implications for interpretations of South Atlantic hotspot stability and South American plate motion histories.

The study identifies a previously unrecognized geodynamic effect of flat subduction—asthenospheric anomaly telescoping—wherein supersized flat slabs can frontally translate subduction-modified asthenosphere thousands of kilometers inland, creating near-stationary intraplate mantle anomalies. For SASWIR, a Jurassic flat slab (Georgia Islands slab) of exceptional extent (~2280–2600 km) emplaced a subduction-modified mantle domain later sampled by Karoo magmatism and the migrating Southern Atlantic–Southwest Indian ridges since ~50–35 Ma. Multi-proxy evidence (geochemistry, plate reconstructions, tomography vote maps) and 2-D thermomechanical modeling support this mechanism and its scale, exceeding those of well-known flat-slab events (Laramide, Eastern China). This process likely operated throughout Earth history, contributing to upper mantle heterogeneity beneath continents and potentially explaining other localized ghost-arc MORB occurrences. Future work should target: acquiring off-axis geochemical transects in SASWIR to document temporal sampling since 50–35 Ma; refining mantle reference frames and slab sinking rate constraints; 3-D geodynamic simulations including trench-parallel flow and melt extraction; and global searches for similar ridge–fossil slab–ghost-arc alignments.

• Data coverage: Sparse off-axis geochemical sampling in the South Atlantic and Southwest Indian oceans limits reconstruction of the anomaly’s complete sampling history since 50–35 Ma.
• Modeling simplifications: 2-D models neglect trench-parallel mantle flow, complex mantle wedge processes (e.g., mélange diapirs’ buoyant dynamics), and explicit melt extraction/transport; rheological and parameter choices are generalized.
• Kinematic assumptions: Reconstructions assume average whole-mantle slab sinking rates (1.1 cm/yr) and largely vertical sinking with minor lateral drift; uncertainties in mantle reference frames (e.g., Torsvik & Cocks) can yield incompatible geometries.
• Tomography resolution: Inherent limitations and variability across tomography models, despite vote mapping, can affect precise slab geometry and correlation.
• Affinities of localized isotopic anomalies (e.g., extremely low 187Os/188Os) may reflect additional localized processes (e.g., lithospheric erosion) not fully resolved regionally.