Rapid Flooding of the Sunda Shelf: A Late-Glacial Sea-Level Record

See allHide authors and affiliations

Science  12 May 2000:
Vol. 288, Issue 5468, pp. 1033-1035
DOI: 10.1126/science.288.5468.1033


The increase in sea level from the last glacial maximum has been derived from a siliciclastic system on the tectonically stable Sunda Shelf in Southeast Asia. The time from 21 to 14 thousand calendar years before the present has been poorly covered in other records. The record generally confirms sea-level reconstructions from coral reefs. The rise of sea level during meltwater pulse 1A was as much as 16 meters within 300 years (14.6 to 14.3 thousand years ago).

Sea-level curves of the last deglaciation have been constructed from coral reefs by means of U/Th or radiocarbon dating of corals [e.g., Barbados (1,2), Tahiti (3, 4), New Guinea (5, 6)]. The Barbados record contains only a few clustered data points between larger intervals without data for the early phase of the late-glacial sea-level rise; the Tahiti record starts at 13.8 thousand years ago (ka) (all ages are in calendar kiloyears before present) (7) above a late Pleistocene reef unit; and the New Guinea record starts at 13.1 ka and is situated in an area with strong tectonic uplift. There is a prominent hiatus in most records corresponding to the meltwater pulse (MWP) 1A around 14 ka (2). Catastrophic sea-level rise for that time is interpreted in the Caribbean-Atlantic region from an abrupt deepening of the coral assemblages (8). From these and other records, two alternative models of the eustatic late- to postglacial sea-level rise have been proposed (9): (i) a continuous model in which sea level rose steadily with varying transgression rates and (ii) an episodic model in which sea level rose in steps with several pauses or even erosion in between.

Here we present a record of the late-glacial transgression on the Sunda Shelf, the largest shelf area outside the polar regions, covering an area of 1.8 × 106km2 between the Indonesian archipelago and Vietnam. During the last glacial maximum (LGM), the Sunda Shelf was widely subaerially exposed. The adjacent South China Sea was strongly reduced in size, forming a semi-enclosed marginal sea. This exposed a large low-gradient coastal plain on the Sundaland craton. The Singapore platform in its central part was tectonically stable during the Pliocene and Pleistocene, and the modern Malayan Peninsula and Borneo and Sumatra islands formed highlands to the south (10). Several river systems, originating here, drained the coastal lowlands during the LGM and the first phase of late-glacial flooding (11); the largest paleoriver system is known as the North Sunda River or Molengraaff River and flowed to the northeast (12).

More than 50 sediment cores along two transects were collected during cruise 115 of R/V Sonne: One transect was oriented northeast to southwest and extended 600 km from the upper continental slope to the middle shelf in the area of the former North Sunda River. The second transect ran east to west off the modern Mekong Delta for 200 km (13). The sedimentary records were used to determine position relative to sea level at the time of deposition by analysis of the depositional environment. We identified mangrove swamps, mud flats, and coarser grained siliciclastic deposits from the base of the transgressive system (14) that formed in a microtidal environment with a tidal range of less than 2 m (15). Analysis of terrestrial organic material (pollen, wood fragments, in situ mangrove roots, peaty layers) provided additional information on terrigenous sources of sediments.

We obtained accelerator mass spectrometry (AMS) radiocarbon dates of terrestrial organic material along several cores to follow the coastline migration in time and space (16). Because sediments can consist of a complex mixture of recent terrestrial and marine biological products and older reworked organic and inorganic carbon-containing terrestrial and marine material, clearly identifiable material is required for meaningful dates. We dated mangrove roots, larger wood fragments (centimeter sized), and coarse-grained peaty detritus in tidal flat deposits as direct sea-level indicators (17). In addition, we dated plant material from the landward delta plain as well as from the seaward bay lagoon and delta front environment and benthic foraminifera (Rotalia sp.) from the oldest purely marine part. Because transport and reworking of organic material may cause a bias in the determination of sea-level ages, we tried to estimate the effect on age determinations of mixing different terrigenous organic materials. Residues of the acid-alkali-acid extraction of peaty layers and organic macrofossils such as larger wood fragments and in situ root remnants yielded the youngest and most reliable ages. The differences in age between macro- and microplant remnants in sediments containing <1% Corg are up to 1500 years, and the larger biases are toward younger ages, which may be explained by increased reworking and admixture of older detritus as sea level rises. The dates from alkali-extracted organic material (humic acids) were generally somewhat older than the macrofossils and peat but younger than the organic-poor residues. An additional indicator for recognition of fresh organic material and a measure of rapid primary deposition is the ash content, which depends on microbial decay, reworking, and early diagenetic processes (18).

On the proximal part of the Sunda Shelf transect, the late-glacial transgression could be directly followed between a water depth of 70 and 126 m. Pedogenesis on the late Pleistocene land surface indicates a low sea-level stillstand in the LGM at 22 ka. This surface is overlaid by transgressive coastal sediments. These are partly cut by incised channels detected from shallow seismic surveying, which belong to the ultimate phase of the North Sunda River system before marine flooding. Rapid lateral and vertical facies changes, extreme sedimentation pulses, and/or hiatuses of several thousand years characterize the record between 21.0 and 13.4 ka in 37 cores (19). Cores from the Vietnamese part of the Sunda Shelf in similar depositional environments yielded ages between 13.0 and 11.0 Ka corresponding to depths between 56 and 48 m. Fully marine sands then cover the transgressive sediments after a hiatus. Ages fromRotalia sp. from this cover layer vary widely from 12.4 to 4.1 ka and indicate that these shelf sediments were reworked and mixed up to recent times (14, 20).

We present the first half of the late to postglacial sea-level curve from the Sunda Shelf (Fig. 1) as a probability band, taking into account the uncertainties (1σ) of relevant data in both age and position relative to present sea level (7, 21). The curve can be divided into four segments.

Figure 1

Sea-level curve for the Sunda Shelf derived from shoreline facies. Colored points: most probable ages. Rectangles: vertical, maximum tidal extension; horizontal, 1σ probability. Gray band, range of 1σ and tidal influence; black line, most probable position of sea-level curve. 14C plateaus can cause multiple age points and error bars (rectangles) within the 1σ probability containing no age point (20). For explanation of segments, see text. mbss = meters below sea surface.

Between 21.0 and 19.0 ka, the terminal phase of the LGM, sea level rose slowly from −116 to −114 m below modern sea surface at a rate of 0.10 m per 100 years (1σ = 0.14 m per 100 years). Compared with the 2-m tidal range, this sea-level rise is statistically not significant and it is therefore poorly defined (segment 1).

Between 19.0 and 14.6 ka, sea level rose moderately from −114 to −96 m at a rate of 0.41 m per 100 years (1σ = ±0.08 m per 100 years) (segment 2).

Between 14.6 and 14.3 ka, sea level rose from −96 to −80 m at an accelerated rate of 5.33 m per 100 years. The large 1σ range and the impossibility of quoting the 1σ probability areas, which lead to a very asymmetric distribution of possible calibrated ages, results from 14C calibration over the Oldest Dryas/Bølling14C plateau. The possible rate estimated from probability distributions of calibrated ages ranges from 2.8 to 16 m per 100 years. A high value of 16 m is unrealistic because of the local sediment accumulation rate and the ability of mangroves to settle (segment 3).

Between 14.3 and 13.1 ka, sea level rose more gradually from −80 to −64 m at a rate of 1.33 m per 100 years (1σ = ±0.33 m per 100 years) (segment 4).

The younger sea level rise up to 11 ka is represented by only two data points from the Vietnam Shelf that indicate an average rise of 8 m in 700 years.

Segments 2 and 3 of our curve with control points covering the whole time range fill substantial time intervals without data points in the coral reef sea-level records. For segment 2, data from 16.6 to 15.4 ka are derived from three cores characterized by a moderate and continuous acceleration in sea-level rise filling a gap of 3 ka in the Barbados curve (1). Segment 3 is characterized by a rapid rise in sea level of 16 m. This rapid rise is resolved by 17 dates on plant fragments (14) and in situ preserved mangrove root remnants (3) from five cores. A 14C plateau extending from 15.3 to 14.4 ka is probably caused by large changes in oceanic circulation related to the strong MWP 1A (7). This 14C plateau causes ambiguities in the conversion of 14C ages to calibrated ages. Two age values in the lower half of the sudden rise are slightly younger than the end of the plateau, indicating that the youngest of the multiple age values of the plateau ages fit best to the curve. The start of segment 3 is defined as the average of the two calibration intercepts at 14.7 and 14.4 ka in core 5981; the intercept at 15.1 is impossible considering the age of the Oldest Dryas/Bølling transition at 14.65 ka in the GISP2 ice core (21). Similarly, at the end of this segment in 80-m water depth at 14.3 ka, the statistical 2σ probability allows a young age of maximal 14.08 ka in core 5609.

The Sunda Shelf sea-level curve is compared with other published curves plotted on the U/Th time scale or on the14C time scale newly calibrated using INTCAL98/CALIB 4.0 (7) (Fig. 2). For segments 2 and 3, we observe a remarkable agreement between the U/Th-dated Barbados sea-level curve and the 14C-dated Sunda Shelf sea-level curve. The recalibrated 14C ages of the Barbados curve not used for sea-level reconstruction, however, are 300 to 500 years younger. This indicates that the calibration based on U/Th data gives good results for 14C of terrigenous material but that discrepancies may occur for marine calibrations for a variety of reasons (22). The shape of MWP 1A calculated from the Sunda Shelf data is similar to MWP 1A calculated from Barbados. This is all the more remarkable, considering the larger uncertainty envelopes of both curves. It confirms that there is no evidence of a true sea-level jump but that the rise took place faster than has been suggested (1, 2, 8). The depth precision of the present sea-level record is determined by the tidal range of 2 m on the Sunda Shelf; this is narrower than the one resulting from the coral habitat zonation of 6 m in coral reefs.

Figure 2

Compilation of sea-level data in comparison. Solid line, Sunda Shelf sea-level curve; using Acropora palmata (1, 2), and using different coral (sub)species (3, 5,6). Note significant distance between 14C data points from Barbados and the Sunda Shelf, both based on 14C ages calibrated using INTCAL98/CALIB4. Dotted line below, rate of glacial meltwater discharge calculated from Sunda curve; solid gray line, rate of discharge from Barbados using U/Th ages. Discrepancies between both are within 1σ. mbss = meters below sea surface.

The phase of accelerated sea-level rise was attributed to MWP 1A, which started at 14.7 ka and ended at 14.1 ka in correspondence to the first cold reversal of the Bølling-Allerød observed in the GISP2 core (3, 4, 23). MWP 1A was previously described as a large sea-level jump before 13.8 ka (2), as a catastrophic rise event of 13.5 m in magnitude starting at 14.2 ka (8), and as a rapid sea-level rise of 24 m in 500 years (14.2 to 13.7 U/Th ka) [less than 1.014C ka around 12 14C ka corresponding to 14.5 to 13.6 ka (1, 2)]. Our record thus provides additional evidence for the existence and importance of MWP 1A and shows that sea level dramatically rose by as much as 16 m within 300 years. The sea-level curve shows a continuous sea-level rise before and after one accelerated short-lasting period due to a major melting event in the polar regions.

  • * To whom correspondence should be addressed. E-mail: th{at}


View Abstract

Navigate This Article