Coevolution of craton margins and interiors during continental break-up

Cratons are commonly viewed as mechanically strong, long-lived, and characterized by extremely low erosion rates over geological timescales. The development of kilometre-high, laterally extensive great escarpments near shield edges and subsequent uplift and denudation of craton interiors are therefore geologically abrupt and enigmatic events. Although many studies associate escarpment formation with continental rifting and break-up, the mechanistic link between rifting, escarpment evolution, and subsequent hinterland plateau uplift remains uncertain. Competing hypotheses invoke flexural uplift at rift flanks due to unloading during extension, small-scale convection driven by lateral temperature gradients, coastal downgrading and base-level fall, or passive-margin rejuvenation, but these mechanisms struggle to explain exhumation far inland and tens of millions of years after rift cessation. The purpose of this study is to quantify the spatial and temporal relationships between rift systems, escarpments, and plateaus, and to use geodynamic and landscape-evolution modelling to understand how mantle processes during continental break-up drive coupled margin-interior evolution.

Prior work proposes several mechanisms for escarpment formation and margin uplift: (1) flexural isostatic uplift along rift flanks during extension; (2) small-scale, edge-driven mantle convection producing rift-shoulder uplift; and (3) coastal downgrading and inland base-level fall. Exhumation observed hundreds of kilometres from rifts and long after rifting challenges purely flexural or coastal processes. Alternative ideas include superswell-related dynamic support from deep mantle upwelling beneath moving plates; however, dynamic topography reconstructions indicate limited influence of large-scale mantle flow on these regions. Stochastic inversions suggest intermediate-scale dynamic mantle support contributes about 650 m to Southern African elevations, with an additional isostatic lithospheric contribution of roughly 670 m, in line with models indicating up to ~1 km of dynamic/static mantle support. Studies also discuss post-rift tectonic reactivation and passive-margin rejuvenation, though the latter is debated. Thermochronology and sediment flux records document protracted post-break-up exhumation in Southern Africa, Brazil, and India, hinting at mantle-lithosphere interactions (e.g., keel removal, delamination) affecting surface processes long after rifting.

The study integrates multiple approaches:

• Mapping and statistics: Escarpments along Southern Africa, Eastern Brazil, and the Western Ghats (India) were mapped using digital terrain models and ArcGIS; spatial statistics (e.g., nearest distances, orientations) relative to continent-ocean boundaries (COBs) were computed in R. Nearest distances and angular relationships between escarpments and COBs were analysed to test for rift-border fault origins.
• Lithospheric structure sampling: Present-day lithospheric thickness beneath escarpments was sampled from global reference models LITHO1.0 and LithoRef18 at 1.0° intervals, providing median LAB depths of ~177 km (LITHO1.0) and ~155 km (LithoRef18) along escarpments.
• Geodynamic simulations: 2D thermo-mechanical rift models build on prior work to simulate formation and cratonward migration of Rayleigh–Taylor instabilities at rifted cratonic margins. Key features include upward suction of low-viscosity asthenosphere beneath extending lithosphere, formation of lithospheric edges during necking that generate edge-driven convection, and sequential delamination of the thermal boundary layer (TBL). Model domain depths of 300–410 km, asymmetric and symmetric boundary conditions, and extension rates of 5, 10, and 20 mm/yr were explored. Instability migration rates of ~11–20 km/Myr and wavelengths of ~50–100 km were obtained, robust across tested conditions, and rift-border faults (proto-escarpments) emerged ~100–300 km from COBs.
• Isostatic-analytical estimates: Airy isostasy was used to relate keel removal thickness b and density contrasts to initial surface uplift s = b Δρ/ρa and to total denudation d = b Δρ/(ρa − ρc). Using representative densities, initial uplift for removal of an ~35 km TBL is ~50–100 m, and total denudation from erosion-driven rebound is ~0.5–1.6 km (isostatic factor from uplift to denudation ~0.03–0.04).
• Thermochronology compilation and analysis: AFT and AHe thermal history models across 47 sites on the Central Plateau of Southern Africa were compiled. From best-fit t–T paths and uncertainty envelopes (HeFTy-derived), the total temperature drop, maximum cooling rate and timing were extracted using a 2-Myr moving window. Monte Carlo sampling (20,000 samples per site) propagated distance and time uncertainties to produce spatiotemporal patterns of maximum cooling relative to South Atlantic break-up (135 Ma). Additional analysis removed cooling associated with kimberlite magmatism to test its contribution to observed trends.
• Landscape evolution modelling: The Fastscape model solved the stream power law and lithospheric flexure in plan-view. Uplift was imposed as a laterally migrating Gaussian wave (velocity ~20 km/Myr; half-width ~200 km) informed by geodynamics. Initial plateau elevation was 500 m; erodibility Kr = 1×10−5 m2/yr. Simulations over 50 Myr produced topography, erosion, and erosion rates every 5 Myr. A suite of 120 experiments varied Kr and initial plateau height h0; a misfit function identified parameter ranges matching three targets: plateau height (1,650 ± 250 m), total denudation (2,750 ± 500 m), and final drainage divide position (650 ± 100 km). Predicted AHe and AFT ages were computed under geothermal gradients of 23, 29, and 34 °C/km for comparison with observations.

• Escarpment-COB geometry: Escarpments are generally sub-parallel to adjacent COBs and lie a mean/median ~330–340 km inland globally (regional means range ~207–380 km). This matches predicted half-widths of rifted margins (250–600 km) and supports initiation near rift-border faults.
• Lithospheric context: Escarpments typically overlie thick lithosphere (median LAB depth along escarpments ~177 km in LITHO1.0; ~155 km in LithoRef18), consistent with cratonic roots susceptible to edge-driven convection and delamination.
• Geodynamic mechanism: Rayleigh–Taylor instabilities form at rifted craton edges and migrate cratonward along the keel at ~15–20 km/Myr (11–15 km/Myr in symmetric tests), sequentially removing the lithospheric TBL, inducing adiabatic upwelling, kimberlite magmatism, and isostatic uplift that propagates into the interior.
• Isostatic response and denudation: For removal of an ~35 km keel (TBL), initial uplift is ~50–100 m; erosion amplifies this to total denudation of ~0.5–1.6 km, given representative densities. Analytical factors imply erosion can be an order of magnitude larger than initial uplift (isostatic factor from uplift to denudation ~0.03–0.04).
• Thermochronology evidence: Southern African plateau records protracted exhumation with accelerated cooling at ~120–110 Ma. Peak denudation rates locally reached ~175 m/Myr during the Cretaceous, far above typical long-term cratonic rates (~2.5 m/Myr). Monte Carlo analysis across 47 sites shows inboard migration of maximum cooling after break-up at a median rate of 19.1 km/Myr (10th percentile 7.7; 90th percentile 55.2), overlapping with simulated instability migration and kimberlite migration. Total exhumation sampled from models has mean ~4.35 km and median ~4.06 km since 180 Ma.
• Spatial trends: AHe ages generally decrease cratonward across the Central Plateau and Eastern Brazil, consistent with a migrating uplift/erosion wave. Two domains of AHe resetting are predicted and observed: near the escarpment (ongoing retreat) and on the plateau (tracking inboard mantle forcing). AFT ages show limited resetting on the plateau due to higher closure temperatures.
• Landscape response: Fastscape models reproduce a pinned escarpment acting as a drainage divide while the broad locus of erosion migrates inward with the mantle-forced uplift wave. Parameter ranges exist that satisfy observed plateau height, total denudation, and drainage divide position.
• Broader applicability: Similar sequences inferred in Namibia and Eastern Brazil show initial margin-focused exhumation during rifting and break-up, followed by delayed interior exhumation tens of Myr later, indicating a widespread process linked to rifting and keel removal.

The results establish a mechanistic linkage between continental rifting, escarpment formation, and subsequent exhumation of craton interiors. Escarpments initiate at rift-border faults and initially retreat slowly, but the principal driver of interior plateau uplift and denudation is a cratonward-migrating sequence of mantle convective instabilities that delaminate the lithospheric keel. This process produces a long-wavelength uplift signal and a synchronized, migrating wave of denudation that persists for tens of millions of years after break-up. Thermochronology from Southern Africa demonstrates timing, magnitude, and migration rates of cooling consistent with the geodynamic model and kimberlite migration, indicating strong deep Earth forcing. The landscape-evolution simulations show how escarpments become pinned drainage divides while the erosion front continues to sweep across the interior, shaping plateau growth. Compared with alternative hypotheses (pure flexural uplift, coastal downgrading, or superswell-only dynamic support), the integrated evidence better explains inland, delayed exhumation patterns, magnitudes, and spatiotemporal trends observed across cratons. These findings revise the view of cratons as uniformly stable terrains and highlight the role of rift-driven mantle processes in reorganizing surface topography and erosion far into continental interiors.

Continental break-up triggers a coupled sequence of surface and deep Earth processes: (1) rifting and escarpment formation near rift-border faults; (2) escarpment pinning and limited retreat; and (3) sequential, cratonward delamination of the lithospheric keel that migrates at ~15–20 km/Myr, driving interior uplift, a migrating wave of denudation, and organized plateau growth. Observed distances of escarpments from COBs, lithospheric thickness patterns, simulated instability migration rates, and thermochronology-derived exhumation magnitudes and migration rates collectively support this model. This framework explains the rejuvenation and secular exhumation of otherwise stable cratonic interiors long after rifting and has implications for long-term weathering, sediment fluxes, and climate evolution. Future work should extend comprehensive thermochronology and sedimentary constraints to additional cratonic margins, refine geodynamic-rheological parameters and 3D effects, better quantify dynamic topography contributions, and integrate climatic and drainage evolution to resolve regional variability.

• Spatial resolution and assumptions: Lithospheric thickness sampling relies on low-resolution global models (1°), and present-day LAB depths are used as proxies for past conditions.
• Geodynamic model scope: The thermo-mechanical models do not directly predict surface topography amplitudes; they capture instability formation and migration but require coupling to surface-process models for erosion patterns.
• Analytical simplifications: Airy isostatic estimates neglect dynamic mantle support, density changes from metasomatism and melting, and longer-term thinning beyond immediate keel removal, potentially underestimating uplift/denudation contributions.
• Parameter uncertainty: Landscape model results depend on assumed uplift wave properties, erodibility, initial topography, and geothermal gradients; multiple parameter sets can fit observations.
• Data coverage: Thermochronology constraints are strongest in Southern Africa; other regions (e.g., Eastern Brazil, Western Ghats) have sparser data, and further work is needed to confirm spatiotemporal trends.
• Magmatic effects: Although analyses suggest kimberlite-related cooling cannot explain first-order trends, localized effects and preservation biases (e.g., partial preservation of Upper Cretaceous pipes) indicate uncertainty in absolute denudation magnitudes.
• Climatic and drainage variability: Regional climate and drainage reorganizations modulate landscape response times, introducing spatial variability not fully constrained here.