Quaternary landscape dynamics boosted species dispersal across Southeast Asia

The study investigates how Quaternary landscape dynamics, beyond eustatic sea-level fluctuations, influenced biodiversity and species dispersal across Sundaland over the last ~500 kyr. Traditional views posit that sea-level oscillations acted as a species pump, alternately isolating and connecting landmasses. However, recent evidence indicates long-term subsidence and predominantly terrestrial conditions before ~400 ka, challenging sea-level dominance. The authors aim to quantify how geomorphological evolution—drainage reorganization, river captures, and changing physiography—modified connectivity and dispersal routes, thereby affecting diversification patterns across Southeast Asia.

Prior research emphasized eustatic control on Sundaland’s biogeography, proposing that glacial lowstands facilitated dispersal while interglacials promoted isolation and speciation. New work suggests persistent subsidence and extensive subaerial exposure before ~400 ka, implying that some biogeographic patterns predate the Quaternary or reflect additional drivers such as edaphic variation, forest fragmentation, and paleoclimate change. Paleo-drainage networks have been used to explain distributions of freshwater taxa, and theory and case studies highlight landscape morphology as a key driver of dispersal and endemism. Yet quantitative assessments of how geomorphology reconfigured dispersal pathways across Sundaland during the Late Pleistocene have been limited.

The authors coupled long-term landscape evolution modeling with structural connectivity analyses. Landscape evolution was simulated with Badlands over the last 1 Myr at 5 km resolution, including fluvial incision, sediment transport/deposition, and hillslope diffusion. Five scenarios tested combinations of forcings: (1) no vertical deformation; (2) uniform subsidence (−0.25 mm/yr); (3) spatially variable vertical motion; (4) variable tectonics with uniform rainfall; and (5) variable tectonics with an alternative sea-level curve. Two sea-level reconstructions were used (Bintanja & van de Wal, 2008; Spratt & Lisiecki, 2016). Precipitation forcings were taken from PaleoClim snapshots and held constant between intervals; a uniform rainfall sensitivity run was included. A regional tectonic vertical motion map was compiled from uplift/subsidence observations. Calibration used: (i) paleo-river patterns from phylogenetic studies and seismic surveys; (ii) flooding episodes from Malay Basin boreholes; and (iii) sediment accumulation patterns across the shelf. Model outputs included evolving physiography, drainage basin extents, river captures, and cumulative erosion/deposition. Connectivity modeling targeted lowland evergreen rainforest taxa using species-agnostic cost surfaces derived from three geomorphic features: (1) normalized landscape elevational connectivity (LEC), representing the energetic cost to move outside an elevational niche; (2) distance to rivers, tested as corridors or as barriers; and (3) local slope. Costs were equally weighted and discretized; impermeable areas (e.g., ocean, steep slopes, low LEC) had high costs, and permeable areas low costs. Circuit theory (Circuitscape, pairwise mode) computed current density maps across 5 km grids connecting perimeter source nodes around Sundaland to identify preferential pathways. Temporal standardization used z-scores of current flow. Spatial clustering of connectivity was evaluated with Getis-Ord Gi hotspot analysis across multiple time slices (periods with ≥50% shelf exposure), then stacked to identify persistent migratory corridors through time.

• Flooding history and exposure: Except for the no-subsidence scenario, Sundaland was predominantly terrestrial prior to ~400 ka (MIS 11), consistent with subsidence studies. Since MIS 11, pervasive flooding (>50% of shelf) occurs for <20% of the time (~80 kyr). Even with high-amplitude sea-level forcing, most of the shelf remains subaerial for much of the last 500 kyr, with persistent mainland Asia–Java terrestrial connections up to the Holocene in many scenarios.
• Sedimentation: Simulations predict up to ~105 m of deposition in the Gulf of Thailand (scenario 1) over 1 Myr and ~300 m on the continental slope facing the SE South China Sea (scenarios 2,3,5). Deposition correlates positively with sea-level rise periods.
• Drainage reorganization and captures: Strong correlations exist between sea-level fluctuations and catchment area changes. Siam and Johore basins experienced 30–50% area reductions during major flooding. Simulated captures include: Johore capturing Siam headwaters ~330 ka; post-flooding (~240 ka) re-splitting; Siam and West Borneo captures lasting ~60 kyr and ~15 kyr since ~200 ka; and capture of Siam headwaters by the East Sunda Basin over the last ~30 kyr until full flooding ~10 ka.
• Connectivity structure: During full exposure with rivers as corridors, narrow high z-score pathways align with Johore and Siam paleo-drainages, forming continuous trans-Sundaland corridors (e.g., at 250 ka). If rivers act as barriers, connectivity is patchy, with fragmentation across major streams. Partial to near-full flooding constrains corridors to limited high-connectivity belts (e.g., between the Titiwangsa Mountains and marine incursions) and funnels movement through narrow pinch points.
• Persistent corridors: Hotspot analysis shows connectivity hotspots occupy ~24% of Sundaland, with the hottest class (normalized Gi* > 0.75) comprising ~8%. Hotspots cluster across the paleo Johore (Gulf of Thailand) and Siam basins, typically at depths now 40–80 m below sea level. Migratory corridors inferred from geomorphology account for ~8% of the exposed shelf and span biogeographic provinces.
• Biotic timing: Divergence times compiled for multiple taxa over the last 500 kyr are spread across many intervals and are inconsistent with a single eustatic driver, aligning instead with prolonged, dynamic physiographic and climatic controls that enabled both fragmentation and dispersal.

The results demonstrate that Sundaland’s rapidly evolving physiography, including drainage reorganizations and river captures, exerted a strong control on species movement pathways and habitat fragmentation. This framework explains the broad distribution of divergence times among taxa and challenges the view that eustatic oscillations alone drove Quaternary biodiversification. When exposed, rivers could create continuous corridors that facilitated gene flow; under alternative behavioral/ecological regimes (rivers as barriers), they fragmented habitats, encouraging isolation and endemism. During transgressions, constrained terrestrial areas served as critical pinch points that likely mediated dispersal and secondary contact. Thus, integrating tectonics, sea level, and climate-driven surface processes yields a more complete mechanism for observed biogeographic patterns in Southeast Asia, reconciling both dispersal and vicariance signatures in the Late Pleistocene.

By integrating landscape evolution with structural connectivity modeling, the study quantifies Sundaland’s flooding history, documents multiple phases of drainage reorganization, and maps persistent, geomorphology-controlled dispersal corridors. Findings support that fast-changing physiography fostered Quaternary biodiversification by both fragmenting habitats and enabling migration across biogeographic boundaries. The approach reframes species evolution dynamics by coupling Earth surface processes to biotic connectivity over geological timescales. Future research should refine climate forcing, incorporate vegetation and temperature into resistance surfaces, adopt higher spatial resolution in identified hotspot regions (e.g., Gulf of Thailand and Siam basin), evaluate species-specific responses, and test alternative flow-routing and erosion laws to reduce numerical biases.

• Hydrological routing: Single-flow-direction algorithm may induce numerical diffusion, degrade knickpoints, and underestimate captures in flat areas; multiple-flow-direction methods could improve realism.
• Erosion laws and parameters: Uniform, time-invariant erodibility and a detachment-limited law were assumed; real soils vary among islands and through flooding cycles. Transport-limited laws and spatially variable erodibility could alter outcomes.
• Climate forcing: PaleoClim snapshots do not fully represent an emerged shelf for older intervals; precipitation fields include uncertainties and were held constant between snapshots, creating stepwise climate changes. Lack of million-year transient climate with an exposed shelf introduces uncertainty.
• Species-agnostic costs: Equal weighting of LEC, slope, and river distance excludes other key variables (e.g., temperature, vegetation, solar radiation) and species-specific behavior; rivers alternately treated as corridor or barrier but without taxon-specific calibration.
• Spatial resolution and domain: Coarse 5 km grids may miss microhabitat and microclimate effects that influence many species; hotspot regions would benefit from finer-scale modeling.
• Circuit theory setup: Limited number and placement of perimeter nodes, buffer-size choices, and memory constraints may bias current density patterns; tiling and sensitivity analyses could reduce bias.