Tidal dynamics and mangrove carbon sequestration during the Oligo–Miocene in the South China Sea

Modern mangroves are among the most carbon-rich biomes on Earth, but their long-term (≥106 years) impact on the global carbon cycle is unknown. The extent, productivity and preservation of mangroves are controlled by the interplay of tectonics, global sea level and sedimentation, including tide, wave and fluvial processes. The impact of these processes on mangrove-bearing successions in the Oligo–Miocene of the South China Sea (SCS) is evaluated herein. Palaeogeographic reconstructions, palaeotidal modelling and facies analysis suggest that elevated tidal range and bed shear stress optimized mangrove development along tide-influenced tropical coastlines. Preservation of mangrove organic carbon (OC) was promoted by high tectonic subsidence and fluvial sediment supply. Lithospheric storage of OC in peripheral SCS basins potentially exceeded 4,000 Gt (equivalent to 2,000 p.p.m. of atmospheric CO2). These results highlight the crucial impact of tectonic and oceanographic processes on mangrove OC sequestration within the global carbon cycle on geological timescales.

D espite accounting for only c. 0.04% (0.14 Â 10 6 km 2 ) of the present-day global ocean area 1 , mangroves are responsible for c. 4% (c. 24 Â 10 12 g organic carbon (OC) year À 1 ) of mean annual carbon burial in the global ocean ( Fig. 1a) [2][3][4][5][6] . High OC burial in mangroves on centennial timescales (c. 170 g OC m À 2 year À 1 ) 3,5 is due to high rates of primary productivity (1,110 g OC m À 2 year À 1 ) 3 and exceptionally efficient sediment trapping by complex roots 7 , which promotes rapid sediment accretion (c. 5 mm year À 1 ) 3,5 . Consequently, mangroves are a major oceanic 'hotspot' for OC burial (Fig. 1b) 6,8,9 . On geological timescales (Z10 6 years), the preservation potential of OC in coastal-shelf sediments depends on the relative rate and magnitude of sedimentation, subsidence and erosion (Fig. 1c). Sediment supply, accommodation (the space available for sediment accumulation) 10 and depositional process control the rate of sedimentation, sediment thickness and character of sedimentary layering (stratigraphic architecture) 11 . Subsidence is driven by tectonics, sediment loading and compaction. Erosion occurs by several mechanisms with the repeat time between successive erosion events varying across 12 orders of magnitude (Fig. 1c) 11 . The effectiveness of erosional (b) Area-normalized annual OC burial yields 3,6 . Deltaic and non-deltaic continental shelf environments are combined into 'continental shelf' due to poor constraints on deltaic area 6 . (c) The range and timescales of sedimentary erosion processes 11 and geological controls on sediment burial and lithospheric preservation in the context of mangrove systems. The shaded triangles indicate the main controls on preservation, including the importance of erosion processes.
processes is generally greater as the frequency between successive erosive events increases (Fig. 1c) 11 . Sediment is preserved on geological timescales when it is buried deeper than the depth of erosion and accumulates faster than the repeat time of each successive erosion event. Consequently, sediment preservation is enhanced by higher long-term rates of accommodation space creation, sediment supply and subsidence 10,11 . On short timescales (r10 3 years), the mangrove biome contributes to increased OC preservation by trapping and stabilizing sediment and reducing the magnitude of erosion induced by waves, tides, storms and extreme events (for example, tsunamis) 7,12 .
The geographical (10-10 3 km) distribution of modern mangroves is controlled by climate (temperature, precipitation and storms), salinity and sea-level fluctuations 12,13 . Mangroves are most extensive along tide-dominated, tropical-subtropical shorelines 12 because: (1) higher tidal range increases the intertidal area for mangrove colonization; (2) tidal inundation excludes colonization by salinity-intolerant flora; and (3) tidal action produces tidal channels and lagoons and generally increases coastline rugosity and protection from waves 7,12,13 . Mangroves also occur within abandoned areas of fluvial-dominated deltas, including inactive channel systems, around protected coastal embayments and lagoons, and along some wave-dominated shorelines [13][14][15] . At present, around 40% of the world's mangroves are in Southeast Asia 1 . However, this region is geologically dynamic and has undergone complex and significant plate tectonic and palaeogeographic changes during the Oligo-Miocene 16 . Hence, coastal processes and geomorphology,  intertidal vegetation and coastal-shelf OC burial are all likely to  have varied during the past 25-35 Myr. Furthermore, this region contains significant hydrocarbon accumulations that were principally sourced by terrestrial-and mangrove-dominated OC preserved in coastal-shelf and deep-water sediments 17,18 .
Here we integrate palaeogeographic reconstructions, palaeotidal modelling and sedimentary analysis and suggest that higher tidal range and stronger tidal currents caused widespread mangrove development along ancient shorelines in the Oligo-Miocene South China Sea (SCS). Furthermore, this has decreased through time due to tectonic-driven changes in tidal dynamics. We also show that preservation of mangrove OC was optimized along tide-dominated coastlines in the Oligo-Present SCS, which could be a significant component of the global carbon cycle on geological timescales.

Results
Oligo-Miocene palaeogeography of Southeast Asia. Southeast Asia comprises a complex mosaic of geological terranes assembled along a system of active and extinct subduction and collisional zones. Critically, the region is also located at the triple junction between three major tectonic plates: the Eurasian,   Indo-Australian, and Pacific plates 16 . Palaeogeographic modelling for the Oligo-Miocene in Southeast Asia synthesizes diverse published and unpublished sedimentological, stratigraphic and palaeogeographic data (see Methods section).
The Luzon Strait (LS) between Taiwan and the Philippines was wider in the past than the present day (Fig. 2). The LS decreased in width from c. 1,300 km in the Late Oligocene (Fig. 2a) to c. 500 km in the Late Miocene (Fig. 2g), due to northward movement of the Philippines relative to China and Taiwan. The modern LS is c. 350 km wide (Fig. 2h) and is critical for transferring oceanic flows, tides and waves from the Pacific Ocean into the SCS 19 . Furthermore, the Philippines presently have a significant blocking effect on tropical storms, which move westwards from the Pacific Ocean towards the SCS 20 . Therefore, a wider LS would have allowed more tide, wave and storm-wave energy to propagate into the SCS during the Oligo-Miocene. Furthermore, in the present day, a shallow (o100 m) Sunda Shelf permits throughflow of water from the Pacific, entering the SCS through the LS and exiting through several seaways into the Indian Ocean. However, the Sunda Shelf was emergent throughout the Oligo-Miocene 16 (Fig. 2), which prevented oceanic connection to the west; more tide, wave and storm-wave energy was trapped within the SCS.
Uncertainty in the palaeogeographic position of Palawan stems from a lack of conclusive data regarding Jurassic-Palaeogene plate reconstructions and the genesis of the Northwest Borneo-Palawan Trough 21,22 . As a result, two end-member interpretations of Palawan's position in Oligo-Miocene reconstructions are included as sensitivity studies. In our basecase palaeographic interpretations, Palawan is reconstructed to the northwest of the Late Oligocene-Early Miocene subduction zone along northwest Borneo 16 ( Fig. 2 and Supplementary  Fig. 1a-g). In contrast, Palawan has also been reconstructed to the southeast of the ancient subduction zone along northwest Borneo 21 . The sensitivity of palaeotidal modelling to Palawan's position was tested by removing an emergent Palawan in basecase palaeogeographic interpretations but the subduction zone position remained consistent ( Supplementary Fig. 1h-j).
Tectonics caused significant subsidence in shelf basins in the SCS during the Oligo-Miocene; extension in western and northern basins (Fig. 3a-basins 1-6) contributed to c. 44-5 km total (decompacted) subsidence 18 , whereas lithospheric flexure caused up to 8-12 km of subsidence in foreland basins along northwest Borneo 23 (Fig. 3a,b-basins 7-8). Several basins in the western and southern SCS were subject to inversion during the Middle Miocene-Present 16,18 . Variations in the magnitude of tectonic subsidence and inversion, thermal subsidence,  Tidal sediment preservation in the Oligo-Miocene SCS region. Tidal range and bed shear stress model output has been compared with sedimentological and biostratigraphic data from  the Oligo-Miocene SCS region. This indicates broad agreement between model results and previous palaeoenvironmental interpretations 18 , including the widespread occurrence of mangrove-related facies 24,25 . However, comparing and validating tidal bed shear stress output with sedimentological data 26 ( Fig. 6; see Methods section) depends on the grain size availability at the time of deposition, which controls the type and preservation potential of tidal signals in the sedimentary record.
In shoreline-shelf depositional systems, grain size availability is controlled by the grain size distribution of sediment supplied to the shoreline through river mouths. Tidal signals will be absent if the available sediment was coarser than the maximum sediment calibre capable of being reworked by the tides. Conversely, if the available sediment was finer than the maximum sediment calibre capable of tidal reworking, then an increase in the size and frequency of tidal bedforms could be expected. A comparison between the model results and the actual stratigraphic preservation for peripheral SCS basins (numbered 1-8 in Fig. 6) is discussed below (see Methods section).
In the Nam Con Son Basin, the Late Oligocene-Early Miocene Dua Formation comprises paralic (coastal-deltaic to shallow marine) coals and mudrocks, commonly containing mangrove palynomorphs 27 , and fine-to-medium-grained sandstones ( Fig. 6-basin 3), which are interpreted to have been deposited under tidal influence 28 . Modelled tides capable of reworking gravel (Fig. 4e) are consistent with tide-influenced deposition and preservation of decimetre-to metre-scale bedforms in fine-to-medium-grained sandstones ( Fig. 6-basin 3).
In the Pattani Basin, Early-Late Miocene mudstone/coal source rocks with common mangrove pollen are interbedded with medium-grained sandstones ( Fig. 6-basin 4); the interpreted tide-influenced environments (fluvio-deltaic and mangrove-vegetated lower coastal plain/lagoons) 29 is consistent with modelled tides capable of entraining coarse sand ( Fig. 4e-g) and formation of decimetre-to metre-scale bedforms ( Fig. 6basin 4). In Late Oligocene-Middle Miocene strata in the Malay Basin, paralic mudstones and coals contain abundant mangrove (and freshwater flora) pollen and represent the dominant hydrocarbon source rocks ( Fig. 6-basin 5) 17 . Modelled tides were capable of reworking coarse sand ( Fig. 4e-g) and preservation of decimetre-to metre-scale bedforms within interbedded medium-grained sandstones (Fig. 6-basin 5). Furthermore, mangrove pollen acmes also occur in Late Oligocene-Middle Miocene strata of the West Natuna and Cuu Long basins 24,27 (Fig. 6-basins 2 and 6), consistent with macrotidal tides capable of reworking sand (Fig. 4a,d). Therefore, the palaeo-Gulf of Thailand, a tectonically controlled regional-scale (100 s of km) embayment, facilitated mangrove colonization throughout the Mio-Pliocene. This was enhanced by a combination of sheltering from direct wave approach and tidal amplification, which compares closely to the modern Gulf of Thailand and Brunei Bay, northwest Borneo.
Impact on OC burial in the Oligo-Miocene SCS. Total organic carbon (TOC) burial in the Oligo-Miocene SCS, which contains a significant component of mangrove OC 17 , is evaluated through a series of assumptions regarding the volume of sediment preserved, the volume of hydrocarbons generated, the dominant source rock type and TOC of preserved sediment. We first estimate the amount of OC burial in the BDP (Fig. 3b) since the mid-Miocene (15 Ma) using two approaches: Method 1-volume of hydrocarbons in place (that is, oil and gas that has been proven to be trapped following secondary migration from source rocks); and Method 2-average TOC for the total preserved sediment volume. Detailed calculations are made in the BDP because there is published information on subsidence history 23 (Fig. 3a-basin 8), sediment thickness 37 (Fig. 3b), TOC values of source rocks 38 and hydrocarbon volumes in place 23,39 . Second, we compare the sediment volume in the BDP to the total sediment volume in peripheral Borneo 37 (Fig. 3b) and SCS 40 basins ( Fig. 6  ARTICLE sediment volume in these basins was comparable to that within the BDP since 15 Ma (calculated using Methods 1 and 2) ( Table 1). This is assumed because deposition in all these basins occurred in similar tropical, wave-and tide-influenced, coastal-deltaic to deep marine systems, with abundant evidence for mangrove vegetation (Fig. 6) (Fig. 3b). This corresponds to an average sedimentation rate of c. 0.5 mm year -1 , although at times during the last 15 Ma sedimentation was even higher, keeping pace with tectonic subsidence rates of up to 3 mm year À 1 (ref. 23). Mangrove OC in coastal-deltaic, shallow-marine and deep-marine sedimentary rocks 23,38 is the dominant source material for substantial volumes of hydrocarbons in the BDP (Brunei 23 and Sarawak 39 ); an estimated 7 billion barrels of oil and 19 trillion cubic feet of gas initially in place (before production) (Supplementary Table 2) suggests a minimum OC burial (as trapped oil and gas; Method 1) of c. 1.9 Gt in the BDP since 15 Ma, equivalent to 0.1 p.p.m. atmospheric equivalent CO 2 per Myr (Table 1). This represents a minimum estimate because not all of the accumulated carbon would have generated hydrocarbons, not all hydrocarbons have been found, some hydrocarbons have migrated to the surface and shallow occurrences have undergone biodegradation due to freshwater flushing 23 .
Using Method 2, with an average TOC of 0.05%, OC burial in the BDP since c. 15 Ma increases to 90 Gt; this is equivalent to sequestration of c. 2.8 p.p.m. atmospheric equivalent CO 2 per Myr (Table 1). An average TOC of 0.05% is significantly lower than the average 2.4% TOC (n ¼ 20) measured in the western BDP 38 ; however, the western BDP sample set was biased towards carbonaceous facies. The paucity of published TOC data in the BDP prevents calculation of a more accurate, average TOC value. However, the high average sedimentation rate in the BDP since c. 15 Ma (c. 0.5 mm year À 1 ) 23 suggests a relatively high efficiency for OC preservation due to a decrease in the exposure time to oxygen during burial 41,42 . Since the depth of oxygen penetration in marine sediments is typically 1-10 mm 4 , sediments deposited in the BDP were probably exposed to O 2 for only a few decades. A maximum average TOC of 1% for sediments deposited in the BDP since 15 Ma may therefore be plausible, especially with an expanded and more productive coastal mangrove biome. An average TOC of 1% would correspond to c. 1,800 Gt OC burial, equivalent to sequestration of 48 p.p.m. atmospheric equivalent CO 2 per Myr ( Table 1).
The estimated sediment volume deposited in present-day shelf basins of the SCS (Fig. 6 GtC) 44 . These estimations highlight the significant impact of geological sequestration of mangrove OC to the global carbon cycle during the Neogene, although the precise size of the expanded mangrove biome is uncertain.

Discussion
Oligo-Miocene tides in the SCS were strongly influenced by rapid and substantial palaeogeographic changes (Fig. 2), including the emergence of the Sunda Shelf, variations in the width and depth of the LS and the position and bathymetry of the IBM arc ( Fig. 4 and Supplementary Figs 3,4 and 7). The emergence of the Sunda Shelf was linked to the plate tectonic evolution of Southeast Asia, which created a 'blind' gulf-like basin geometry with uplifted areas on three sides of the SCS and an open ocean connection to the northeast (Fig. 2). Furthermore, during the Oligo-Miocene, the wider and deeper Luzon Strait and ocean connection between the IBM Arc and Japan (Fig. 2) enhanced inflow of tidal energy from the Pacific Ocean compared with the present day (Fig. 4).
A decrease in tidal range and tidal bed shear stress through the Miocene reflects a decrease in tidal energy entering the SCS, caused by: (1) the Luzon Strait becoming narrower and shallower due to northward migration of the Philippines and associated volcanic activity 16 , and (2) northward movement of the IBM arc. Thermal subsidence and rising sea levels increased shelf width in the western and southern SCS during the Mio-Pliocene to present day 18 . This would have increased frictional damping of shoaling tides, contributing to decreasing tidal range and strength 26,45 . However, funnelling of tides in palaeobathymetric constrictions produced elevated tidal range and bed shear stresses throughout the Miocene, most notably during the Early Miocene in northwest Borneo (Fig. 6-basin 7) and during the Middle-to-Late Miocene in the Gulf of Thailand (Fig. 6-basins 4-6).
Along tropical coastlines in the Early-Middle Miocene of the SCS, assuming a consistent coastal plain gradient, higher tidal ranges would have increased the intertidal area for mangrove colonization (Fig. 8). Stronger tides would have promoted development of embayments, tidal channels and lagoons, thereby providing greater wave protection for optimum mangrove growth (Fig. 8). Higher tidal range would have also increased tidal flow and salinity in the lower reaches of marine-connected river channels and adjacent tidal channels. This, combined with stronger tides, would have increased the supply of sediment, oxygen and nutrients throughout tide-influenced areas of the lower coastal plain. Together, this would have enhanced mangrove development, productivity and sediment trapping and accretion, leading to increased intertidal OC burial (Fig. 8c) 7,46,47 . Stronger tides would have also increased erosion of tidal channels and maintained their flow capacity. Although modern mangroves are typically found along low wave and tide energy coastlines 12,13 , they also occur on open coastlines subject to relatively high tidal energy (for example, Ganges-Brahmaputra 48 and Fly River 49 deltas) and mixed tidal-wave energy (for example, Mekong Delta 13 ). If subjected to high coastal energy and degradation, mangroves can rapidly migrate and regenerate 50 . However, higher tidal ranges can promote the stability of tidal channel networks and intertidal vegetation, making mangroves less susceptible to erosion and destruction 46 . This is augmented by the exceptional sedimenttrapping capacity of mangrove roots, which dramatically stabilizes sediment in the intertidal zone 7,12 .
Preservation of mangrove sediment depends on the interplay between the rates of accommodation space creation and sedimentation 13 . During the Miocene, an overall rise in sea level in the western SCS (Fig. 2) contributed to a long-term (Z10 6 years) increase in accommodation space for mangrove accumulation 13 and may have increased OC sequestration independent of changes in tidal dynamics. However, this long-term trend includes many shorter-term fluctuations in relative sea level (RSL), the causes of which include glacio-eustasy 51 , sediment supply variations and tectonics. SCS basins display multiple phases of relative shoreline transgression and regression 18 but with distinctive stratigraphic patterns reflecting different tectonic and basin-fill histories (Fig. 3) 18,22 . Shoreline process models 32,52 suggest that tideinfluenced coastlines are more common during transgressive phases than regressive phases because shelves are wider (possibly enhancing tidal shoaling and resonance) and estuaries are more abundant (increasing local tidal amplification). However, extensive mangroves and OC burial have occurred along both transgressive 36 and regressive deltaic shorelines 53 .
Vertical stacking of tide-and wave-dominated stratigraphic units (facies successions) in the Balingian and Baram Delta provinces (Fig. 7a,b) may have been caused by internal (autogenic) or external (allogenic) forcing. Tide-dominated stratigraphic units may have been preferentially deposited in coastal embayments formed by tectonics or during transgressive phases of allogenic-driven RSL cycles 54   ARTICLE changes in river-mouth position along a tide-and waveinfluenced deltaic shoreline. Axial to river mouths, fluvial and tidal processes dominate; tidal currents are funnelled and amplified as they propagate into distributary channels and ebb tides are augmented by fluvial outflow and form elongate tidal bars (Fig. 7e). Lateral to river mouths, fluvial power and tidal amplification is diminished and waves are relatively more important, but tides may still dominate if they are stronger than wave and fluvial processes (Fig. 7e). Lateral changes in process dominance and mangrove distribution are observed in the present-day Ganges-Brahmaputra Delta 32,48 . The eastern Ganges-Brahmaputra Delta is fluvial dominated 48 ; however, diminished fluvial sediment supply to the western Ganges-Brahmaputra Delta has meant that tides, supplemented by wave and storm processes, exhibit the dominant control on sediment redistribution across the lower delta plain and delta front 48 . This has formed a dense tidal channel network within which mangroves have proliferated and stabilized intertidal substrates 48 . High sediment supply to the adjacent fluvial-dominated area and effective sediment reworking by tides and storms enhances mangrove sedimentation and optimizes OC burial in tidedominated areas of the delta.
During RSL changes, mangroves must re-adjust to the changing intertidal area to maintain their capacity to store and bury OC 13,47 . However, previously buried OC may undergo erosion, reworking and oxidation during these periods 5 . This could offset any reduction in atmospheric carbon resulting from preceding OC burial unless reworked mangrove carbon is re-buried before oxidation. RSL rise will increase the accommodation space for mangroves 13 and generally facilitate increased OC burial 47,55 , especially when the rates of RSL rise and vegetation growth are balanced 56 . During RSL fall, erosion by fluvial channels may be substantial and can form large (10s of metres deep and 100 s of metres wide) incised valleys 54 . Mangrove-related sediments within incised valleys will mostly be eroded and transported downstream but a component may be re-deposited and stored within the incised valley 57 , especially during the subsequent period of RSL rise and valley-fill 54 . Mangrove sediment adjacent to the incised valley may still be preserved during transgressive phases, though oxidation will destroy a proportion of OC in long-lived soil layers.
During both RSL rise and fall, the depth to which preceding mangrove deposits are eroded may be insignificant compared to the thickness of preceding sedimentation. Many peripheral SCS basins have undergone 45 km subsidence since the Oligocene (Fig. 3a,b): (1) 16 . High fluvial sediment supply also contributes to mangrove nourishment by maintaining substrate depth during RSL rise and high OC burial 13,55 . The combination of high subsidence, tidal range, sediment supply and strong tides would have optimized: (1) the development of a potentially vast mangrove biome along tideinfluenced coastlines; (2) accretion and burial of OC-rich sediment, both within mangroves and adjacent environments; and (3) lithospheric storage of OC.
Mangroves are one of the most efficient links connecting the atmospheric, biospheric and lithospheric reservoirs of carbon 3,8,9 . During the Oligo-Miocene in the SCS, tidal dynamics optimized the development of carbon-rich mangroves, and high fluvial sediment supply and tectonic subsidence enhanced preservation of mangrove OC. The scale of this mangrove OC burial was a significant component of the global carbon cycle on geological timescales.

Methods
Palaeogeographic reconstruction. Highstand palaeogeographic reconstructions for Southeast Asia at seven timeslices during the Late Oligocene-Late Miocene ( Fig. 2 and Supplementary Fig. 1) were generated using the Getech plate model. Reconstructions synthesize diverse published and unpublished sedimentological, stratigraphic and palaeogeographic data 16,18,21,24 . Gross depositional environments are depth-delineated ( Supplementary Fig. 1) and the boundaries between these environments are extracted as palaeobathymetric contours and interpolated to form a grid (0.1°resolution) using ArcGIS 59 . Shelf-to-slope (o2,000 m) bathymetry is interpreted from available seismic, sedimentological, lithological, biostratigraphic and palaeogeographic information. Ocean crust bathymetry is calculated by applying an age-depth relationship 60 to a rotated ocean age data set 61 , with corrections for sediment cover, sea-level changes and the intrusion history of oceanic seamounts 59 . The detailed reconstructions for Southeast Asia were included in global palaeogeographic reconstructions for each ancient timeslice generated by the Getech Globe project.
Subsidence curves for 10 peripheral SCS and Borneo basins (labelled 1-10 in The maximum variation in global eustatic sea level during the Late Oligocene-Miocene is c. 50 m 51 . Therefore, to test the sensitivity of tidal modelling to palaeobathymetry, we also built tidal models for palaeobathymetric reconstructions for the seven ancient timeslices with sea level 50 m lower than our base-case, highstand palaeobathymetric interpretation (Supplementary Figs 4 and 5). Furthermore, additional tidal modelling was performed to test the sensitivity of model results to two major areas of palaeobathymetric uncertainty. First, to test the sensitivity to the positioning of Palawan relative to the proto-SCS subduction zone along northwest Borneo 22 , base-case interpretations place Palawan to the northwest of the subduction zone ( Supplementary Fig. 1a-g), whereas we also ran tidal models for palaeogeographic interpretations where Palawan is removed as an emergent feature ( Supplementary Fig. 1h-j). Second, the IBM arc is modelled as emergent throughout the Late Oligocene-Miocene ( Fig. 1a-g), despite being predominantly submerged in the present day (Fig. 1h). To investigate the blocking effect of an emergent IBM arc on tides entering the Philippine Sea and SCS from the Pacific Ocean (Fig. 1), we ran a tidal model for a Messinian (6 Ma) palaeobathymetric interpretation with the IBM arc submerged to a shallow depth of 10 m (Supplementary Fig. 8). Tidal modelling. Fluidity is a finite element hydrodynamic model that uses unstructured, tetrahedral meshes to maximize computational accuracy and efficiency 45 . Multi-scale, global computational meshes were generated with a finest mesh resolution of c. 10 km in areas of complex bathymetry (for example, steep topography and coastlines). Fluidity does not permit large-scale flooding/drying, therefore, there must be a minimum depth along the coastline to prevent the free surface from intersecting the bottom surface as it propagates 45 . Simulations represent full astronomical tidal forcing for 3 months of simulation time with a spin-up period of five days. Outputs are the amplitude of constituent tidal components, tidal range and the magnitude and direction of average and maximum tidal bed shear stress. Tidal range is calculated as the difference between the maximum and minimum free surface heights, which approximately equates to the maximum spring tidal range.
Fluidity has been extensively validated for tidal modelling in modern and ancient settings 26,45 , including the SCS (Supplementary Fig. 2). The pattern and magnitude of M 2 and K 1 tides in the SCS compare favourably to models that include data assimilation: (1) global tidal models ATLAS TPXO8 and FES2012, which both used structured meshes with c. 20 km (1/6°) and c. 7 km (1/16°) resolution, respectively 19 ; and (2) a regional tidal model OTIS 19 that used a structured mesh with c. 4 km (1/30°) resolution. For the M 2 tide, Fluidity underpredicts the amplitude in the Philippine Sea ( Supplementary Fig. 2), most likely due to a coarser mesh resolution. For the K 1 tide, Fluidity overpredicts the amplitude slightly (o0.1 m) in the western SCS ( Supplementary Fig. 2), most likely because insufficient energy is damped by frictional drag along the bottom surface, due to a coarser mesh resolution and smoothed bathymetry, and lack of internal drag 19 . For ancient simulations, internal drag is not included because the buoyancy frequency (a function of water column density gradient) is unknown.
Carbon burial estimates. Estimates of the total amount and rate of OC burial, and equivalent concentration of CO 2 , in the BDP since the Middle Miocene (c. 15 Ma), are derived by two methods: Method 1-using estimates of total in-place petroleum resources in hydrocarbon fields in Brunei 23 and Sarawak, Malaysia 39 (Supplementary Table 2); and Method 2-by estimating the sediment volume deposited since c. 15 Ma, using the area of the BDP defined in ref. 23 and sediment thickness map in ref. 37, and assuming average TOC values for the total sediment volume of 0.05 and 1% (Table 1). To estimate the total amount and rate of OC burial, and equivalent concentration of CO 2 , in peripheral Borneo and SCS basins throughout periods of the Oligocene-Present, we first assumed an equivalent amount of OC burial per unit sediment volume to that calculated in the BDP since the Mid-Miocene (using Methods 1 and 2), and second calculated various estimates of total sediment volume across the region during the Late Oligocene-Present (Table 1). These include the sediment volume preserved in peripheral, present-day shelf basins in the SCS (Fig. 6) since the Middle Miocene (c. 15 Ma) and Oligocene (c. 34 Ma) and the entire SCS since the Oligocene (c. 34 Ma); these are based on Table 1 in ref. 40, assuming a consistent average sediment density of 2,060 kg m À 3 and that 82% of sediment mass deposited in the SCS since the Oligocene (c. 34 Ma) was deposited in present-day shelf basins 40 . Estimated sediment volume deposited in peripheral Borneo basins (Table 1) during the Neogene is based on ref. 37.
The rate of OC burial in modern mangroves bordering the SCS (c. 1,000 Gt OC per Myr) was calculated using estimates of mangrove area in countries bordering the SCS 1 and assuming the mean global burial rate for soil carbon in mangroves 3 .
The proportion of hydrocarbons dominantly sourced by terrestrial-and mangrove-dominated OC (paralic source rocks) in basins in the SCS region was estimated based on the recoverable hydrocarbon volumes and dominant source rocks in ref. 17.
Data availability. The palaeogeographic and tidal modelling data sets generated during and/or analysed during the current study are not publicly available due to confidentiality restrictions but are available from the corresponding author on reasonable request and with the permission of Getech. The sedimentary and OC data sets are available from the corresponding author on reasonable request.