Deglacial Meltwater Pulse 1B and Younger Dryas Sea Levels Revisited with Boreholes at Tahiti

See allHide authors and affiliations

Science  05 Mar 2010:
Vol. 327, Issue 5970, pp. 1235-1237
DOI: 10.1126/science.1180557


Reconstructing sea-level changes during the last deglaciation provides a way of understanding the ice dynamics that can perturb large continental ice sheets. The resolution of the few sea-level records covering the critical time interval between 14,000 and 9,000 calendar years before the present is still insufficient to draw conclusions about sea-level changes associated with the Younger Dryas cold event and the meltwater pulse 1B (MWP-1B). We used the uranium-thorium method to date shallow-living corals from three new cores drilled onshore in the Tahiti barrier reef. No significant discontinuity can be detected in the sea-level rise during the MWP-1B period. The new Tahiti sea-level record shows that the sea-level rise slowed down during the Younger Dryas before accelerating again during the Holocene.

Understanding the behavior and predicting the fate of large ice sheets can be done in parallel by studying recent and ongoing changes in the climate system (1, 2) and by studying the dramatic sea-level changes that occurred during the last deglaciation [21,000 to 5000 years before the present (yr B.P.)]. To date, the most complete record of deglacial sea level is based on reef cores drilled at Barbados, which have yielded ages from both 14C (35) and mass spectrometric U-Th methods (58). The Barbados record is characterized by two periods of sea-level acceleration [meltwater pulses (MWP) 1A and 1B] that occurred around 14,000 calendar yr B.P. (cal yr B.P.) and 11,300 cal yr B.P., respectively. During each MWP event, the sea level apparently rose by several meters per century, leading to a major hydrological perturbation that probably impacted the ocean circulation [e.g., (9, 10)]. Both the amplitude and the localization of injection into the ocean are crucial in understanding the climatic impact of a MWP event [e.g., (11)]. However, several first-order questions remain unresolved on the precise characterization of these events, despite the intensive research carried out over the last decade (see SOM text 1).

The precise timing and amplitude of MWP-1A and 1B are still open questions, because both of these events were originally detected as hiatuses between individual drill cores collected at different depths off Barbados (see SOM text 2 and map in fig. S2). Several other records have been interpreted later as direct or indirect evidence of the occurrence of MWP-1A (1216). One of the main goals of the recent Integrated Ocean Drilling Program (IODP) Leg 310 at Tahiti was specifically to collect an additional coral record over the MWP-1A time window. The new suite of coral samples collected during this IODP campaign successfully confirms the existence of MWP-1A and leads to a reassessment of its age and amplitude (17).

However, MWP-1B is even more controversial and still needs to be confirmed, both at Barbados and at other far-field sites. Indeed, subsequent coral studies at Huon Peninsula (18) and Tahiti (12) questioned the timing and amplitude of this freshwater pulse. Additional doubts were also raised about the existence of MWP-1B by a study of sea level in northwest Scotland based on the so-called “marginal basin isolation” technique (19). However, the interpretation of this Scottish record is complex due to its proximity to former ice sheets in a region where the postglacial rebound contribution is dominant, which explains why the local sea level continued to fall during most of the deglaciation. So far, the sample coverage and depth resolution of these different studies are still insufficient to reach a definitive conclusion about MWP-1B. Unfortunately, the new IODP sample collection from Tahiti is of little help in studying MWP-1B, because the depth range of the drill cores was targeted on MWP1-A and the earliest part of the deglaciation (i.e., 90 to 120 m). At these depths, only deep-living coral species persisted in the reef at the levels corresponding to the age of MWP-1B.

To settle the issue, we dated by U-Th 47 pristine coral samples from three new reef cores (P8, P9, and P10) drilled onshore of the Papeete barrier reef in Tahiti, close to the location of our previous study (23 U-Th dated corals from P6 and P7 cores) (12). P8 is at about the same position as P7 (12) but was drilled at an angle of 33° toward the sea, whereas P9 and P10 were collected in the inner part of the barrier reef toward the Papeete Pass (fig. S1).

Figure 1B shows a comparison of the new U-Th data from P8, P9, and P10 cores with the previous Tahiti record (12). Unlike the Barbados cores, each of these Tahiti cores yields an uninterrupted record of the time window corresponding to MWP-1B. The new U-Th data (Fig. 1B and table S1) provide an unprecedented resolution and can be compared to the other sea-level records from Barbados (7, 8), Papua New Guinea (Huon Peninsula) (18, 20), and Vanuatu (Urelapa) (21) (Fig. 1, C and D). The North Greenland Ice Core Project (NorthGRIP) isotope record is also plotted in Fig. 1A using its most recent time scale (22). This is done to compare the sea-level records with climatic transitions such as the inception and the end of the Younger Dryas (YD), which marks the start of the Holocene period.

Fig. 1

Deglacial records over the 14,000 to 9000 yr B.P. time window. (A) δ18O record of the NorthGRIP Greenland core plotted on its most recent time scale (22). (B) Tahiti corals: depth in meters below present sea level versus U-Th ages in thousand years B.P. (core P6, dark blue; P7, light blue; P8, red; P9, orange; P10, green) (see table S1). (C) Pacific corals from Huon Peninsula, Papua New Guinea [brown dots (18), blue dots (20)] and from Urélapa, Vanuatu [light and dark green squares correspond to two different cores (21)]. (D) Barbados A. palmata corals [core #12, orange squares; core #7, red dots; core #16, green diamonds (8)]. All depths have been corrected for subsidence (Tahiti) and uplift (all other sites) as described in SOM text 2. Shaded time windows correspond to the YD [boundaries based on (A)] and to the MWP-1B event [boundaries based on (D)]. For Tahiti, the species and/or genus of the dated corals are provided in table S1. The bathymetric habitat of Acropora with Pocillopora [dots in (B)] is more restricted (about 6 m, a range shown by the short brown bar in the lower left corner) than those of Porites or Favidea corals [triangles in (B)], which can live in the top 10 to 20 m (a range shown by the longer brown bar in the lower left corner). Dark gray lines correspond to linear fits of sea-level data (SOM text 2 and listed in table S2). For Tahiti, the calculations exclude samples from the base of the P8 core deviated toward the shore (red dots below –65 m) and samples made of Porites or Favidea corals. Using other assumptions would not change significantly the calculated rates of sea-level changes and would not change our conclusions (SOM text 2 and table S2). For New Guinea, dashed lines are used when data are scarce (gap starting in the middle of the YD and period before YD).

The large number of data points derived from the four cores provides a very accurate constraint on the sea-level rise during this period, defined by the coherent upper envelope of the paleo-depths of the samples. The small scatter of the data reflects the inherent uncertainty linked to the paleo-bathymetry of corals and associated shallow-living biological assemblages (23, 24). Part of this overall scatter is also related to the different positions of the drill cores on the barrier reef (SOM text 2 and fig. S1). P9, and P10 record the upper reef crest on the inner part of the barrier. By contrast, P8 was drilled on the outer part of the barrier reef, with a deviation of 33° toward the sea. Therefore, in the lower sections of P8 below 65 m, the corals (red points in Fig. 1B) plot slightly lower than those from P7, P9 and P10, a difference that remains small (<6 m) but fairly systematic.

The rate of sea-level rise at Tahiti can be calculated by means of linear fits over the three specific climatic intervals: before, during, and after the YD event (thick lines in Fig. 1B) (see SOM text 2 and table S2 for details). Taken together, the Tahiti data define a relatively smooth sea-level rise, with no significant acceleration during the time interval corresponding to MWP-1B at Barbados (11,400 to 11,100 cal yr B.P.; area shaded green in Fig. 1). This conclusion is based on Acropora and Pocillopora samples from the four cores P7, P8, P9, and P10 (table S1), which exhibit a rather small scatter (<6 m) (see Fig. 1B). In contrast to the Tahiti record, the MWP-1B event appears as a prominent step of ~15 m between two drill cores at Barbados, implying an apparent rise of ~40 mm/year (Fig. 1D). In the Huon record, the time interval of MWP-1B falls within a time gap of about a millennium (12,100 to 11,100 cal yr B.P.). The Vanuatu record has only two coral samples in this interval, but even considering the few samples below (older) and above (younger), it remains difficult to pick out a step or pause during the 13,000 to 10,000 yr B.P. interval. Therefore, both records from Pacific far-field sites agree with the higher resolution record from Tahiti: None of these reconstructions shows an abrupt 15-m step around 11,300 cal yr B.P., in contrast to the Barbados record.

The new Tahiti record includes many samples covering the YD cold period, in particular from cores drilled in the inner part of the outer reef (P9 and P10). An important observation based on this data set is that the deglacial sea-level rise slowed down during the YD event and reaccelerated during the early Holocene (7.5 ± 1.1 mm/year during the YD, compared with 11.7 ± 0.4 mm/year after and 12.1 ± 0.6 mm/year before). Similar conclusions can be derived (see details in SOM text 2 and table S2) by considering the data obtained on the shallow-living corals (Acropora and Pocillopora, dots in Fig. 1B) or by looking at the entire data set, including those measured on more ubiquitous corals (Porites and Favidea, triangles in Fig. 1B). The Barbados record also suggests that the rate of sea-level rise was reduced to 5.6 ± 0.4 mm/year (Fig. 1D and table S2), in agreement with the slowing down observed for Tahiti. The case for a slower sea-level rise during the YD event than during the period immediately before (9.3 ± 0.4 mm/year) is particularly strong because all A. palmata samples come from the very same Barbados drill core #12 (in contrast to MWP-1B, which occurs between Barbados cores #12 and #7). The existence of a transient pause in the deglacial sea-level rise was also suggested by considering samples older and younger than the 12,100 to 11,100 cal yr B.P. data gap in the Huon record (18). Between 12,100 and 11,100 cal yr B.P., the sea-level rise clearly slowed down, although the data gap in the Huon record does not correspond exactly to the YD event and makes it difficult to compare with records from other sites.

In addition, the coral data plotted in Fig. 1 could suggest a small step (<6 m) in sea level near the onset of the YD event at around 13,000 yr B.P. (arrow in Fig. 1). This small step also corresponds to a rate change in both the Barbados and Tahiti records. At Huon, this period is covered by only a few corals that could possibly be fitted with a small step. However, the existence of such a structure is within the overall uncertainty of the approach (see details in SOM text 2) and thus remains speculative.

Relative sea level (RSL) differences between the records obtained for different sites should be interpreted with caution because isostatic effects are not the same everywhere on the planet. Therefore, a more complete comparison between these reconstructions of local sea-level records requires geophysical modeling. Milne and Mitrovica (25) drew up a comprehensive comparison by using a wide range of mantle viscosity and lithospheric models forced by different ice-sheet histories. These model simulations illustrate systematic differences in RSL between the different sites: In the time window of interest, the shallowest sea level should be observed at Huon and the deepest at Barbados, whereas Tahiti RSL falls in between. This is in good agreement with the observations: At 12,000 cal yr B.P., the RSL is situated at 62, 59, and 55 m below modern sea level for Barbados, Tahiti, and Huon, respectively.

Geophysical processes, such as gravitational and rotational effects, can also affect the relative amplitude of abrupt sea-level changes expressed as steps in RSL records (26). Clark et al. (11) even proposed that comparing such steps at different sites could serve to identify the ice sheet(s) that released large amount of icebergs or meltwater. The apparent discrepancy between the Barbados and Tahiti records over the MWP-1B period around 11,300 cal yr B.P. may be reconciled by assuming that the ice was released exclusively from the Pacific sector of the Antarctic ice sheet (11, 26). However, in this scenario, the model predicts that a small residual step should be expected at Tahiti. Our data suggest that this step would be masked by the inherent uncertainty linked to the coral approach (less than 6 m). Further postglacial rebound modeling simulations such as (11, 26) should be performed to investigate whether such a scenario could generate steps larger than 10 m at Barbados and less than 6 m at Tahiti, a relative gradient even larger than the one caused by a release from the West Antarctica Ice Sheet (11, 26). This hypothetical scenario would also require that the Antarctic ice sheet was much larger than today during the glacial period. This issue has been controversial, but recent numerical modeling of the Antarctic ice sheet (27) is compatible with a total loss of ice of 17.5 m of equivalent sea level since 15,000 yr B.P. (including both contributions of MWP-1A and MWP-1B). However, specific attempts of geophysical modeling focused on the specific contribution to MWP-1B (28, 29) failed to reconcile the observed contrast in RSL with a reasonable contribution from the Antarctic ice sheet [see figure S3 in (29)]. Another puzzling aspect of MWP-1B is the lack of clear signals in marine sediments, from the Southern Ocean or the North Atlantic, for a large freshwater release to the Ocean that should have been similar to the Heinrich events.

Otherwise, we should consider the alternative possibility that MWP-1B might have been overestimated at Barbados. Mapping site locations on the Barbados south coast shows that cores defining MWP-1B are drilled in different environments: submerged fossil barrier reef for the deepest cores #12 and 16, and fringing reef for the shallow core #7 (map in fig. S2). A systematic bathymetric difference between both environments could be invoked, but this hypothesis would imply that, during the sea-level transition, the depth tolerance of A. palmata exceeded its 5-m limit, observed during the modern period characterized by sea-level stability. An additional cause of the bias might be the local tectonics of Barbados. Due to its location on the accretionary prism at the convergent boundary between the Caribbean and South American plates, Barbados is characterized not only by a general uplift of varying amplitude around the island but also by several faults and tectonic flexures (30). It is thus probably an oversimplification to apply a constant uplift rate to all samples. Indeed, the position of cores #12 (#16) and #7 suggests that they may belong to different neotectonic segments (map in fig. S2) and thus were affected by different uplift rates. Nevertheless, it seems very unlikely that this tectonic factor could explain the full amplitude of the jump observed in the coral record, since even tripling the differential uplift correction between the cores would only contribute to half of the 15-m sea-level step observed at Barbados (see SOM text 2 for a full discussion of this issue). In principle, the different explanations invoked to explain the Barbados step (East Antarctic release and local biases at Barbados) are not mutually exclusive and could have been superimposed, but it seems rather unlikely that all these independent processes occurred within the same time window of a few centuries.

In addition to the absence of a detectable MWP-1B step at Tahiti, the other conclusion of our study, that the rate of sea-level rise was reduced during the YD period and reaccelerated during the early Holocene, has often been overlooked. This scenario would resolve the long-standing controversy between the observation of a slower rate of sea-level rise during the YD and the hypothesis that this millennium-long cold event was triggered by a meltwater pulse that slowed the Atlantic meridionaloverturning circulation. Our new record is compatible with previous modeling work (10) mentioning a reduced sea-level rise during the YD and briefly discussing the climatic implications with respect to freshwater forcing. The detection of a small sea-level change just before the start of YD at ~13,000 yr B.P. is tempting but remains difficult to prove. The reduced rate of sea-level rise observed during the following millennium (i.e., the YD event) would then correspond to a return of glacial conditions that interrupted the deglaciation process and, in some cases, even favored glacier readvances in Europe (3133).

Our results on the final stages of the last deglaciation illustrate the complexity of the melting of ice sheets that once covered a large fraction of the northern hemisphere continents. Modeling this retreat, together with the associated icebergs and freshwater drainage history, will help in quantifying the complex impact of ice-sheet melting on ocean circulation and, more generally, Earth’s climate over the first half of the Holocene period (34, 35). In addition, the observed long-term sea-level changes will allow geophysicists to extract the isostatic “memory” component from modern satellite data to quantify recent processes such as oceanic thermal expansion, melting of mountain glaciers, and loss of ice from the Greenland and West Antarctica ice sheets (36).

Supporting Online Material

SOM Text

Figs. S1 and S2

Tables S1 and S2


References and Notes

  1. Materials and methods are available as supporting material on Science Online.
  2. The authors thank G. Cabioch for help with coring and sampling; G. Ménot and W. Barthelemy for help with U-Th analyses; D. Borschneck for help with x-ray diffraction analyses; P. Deschamps for age calculations and drafting Fig. 1; P. Dussouillez and J.-J. Motte for help with maps; and F. Antonioli, G. Cabioch, P. Clark, P. Deschamps, and F. Taylor for useful technical and scientific discussions. Numerical data are available in table S1 and also at the National Oceanic and Atmospheric Administration’s National Geophysical Data Center public data repository. Paleoclimate work at CEREGE is supported by grants from the Gary Comer Foundation for Science and Education, the European Science Foundation (EuroMARC), the CNRS, and the Collège de France.
View Abstract

Navigate This Article