Episodic reef growth in the Last Interglacial driven by competing influence of polar ice sheets to sea level rise

Abstract

Rapid, millennial-scale changes in sea level have been proposed for the beginning, middle, and/or end of the Last Interglacial (LIG) [~129 to 116 thousand years ago (ka)]. Each of these scenarios has different implications for polar ice sheet behavior in a warming world. Here, we present a suite of ²³⁰Th ages for fossil corals in the Seychelles within a detailed sedimentary and stratigraphic context to evaluate the evolution of sea level during this past warm period. The rise to peak sea level at ~122 to 123 ka was punctuated by two abrupt stratigraphic discontinuities, defining three distinct generations of reef growth. We attribute the evidence of episodic reef growth and ephemeral sea-level fall to the competing influence of Northern Hemisphere ice melt and Antarctic ice regrowth. Asynchronous ice sheet contributions would mask the full extent of retreat for individual ice sheets during the LIG and imply greater temperature sensitivity of ice sheets than previously inferred.

Fig. 1.. Outcrops with lithostratigraphic units, coral assemblages, sample locations and ages.

At (A) site 7, (B) site 11, and (C) site 33, episodes of reef growth defined by disconformities (dashed lines) capped by lateral encrustations of M. exaesa or coral rubble. White scale bars are 30 cm unless otherwise noted. Lithostratigraphic units are shown to the right of each photo where the italicized letters inside the box (A and B) refer to the coral assemblage (table S2). Panels (D) and (E) are examples of the two types of discontinuities observed in the outcrops. “Cor” denotes corals. Thin white line in (D) denotes top surface of M. exaesa. Photo in (A) is a splice of two separate photos. Photo credits: A. Dutton.

Fig. 2.. Age and elevation of fossil corals from the Seychelles.

Coral ages are shown with 2σ uncertainties [diamonds, this study and squares from (9)]. Elevation error bars are 1σ uncertainties, and paleowater depth range (translucent blue bars) is based on coral taxonomy, assemblage, stratigraphic context, and coral and outcrop morphology. Samples collected in caves represent a deeper coral facies (32). Open symbols indicate uncertain stratigraphic context for previously published data (33). Colored horizontal bars above the data correspond to site-specific hiatuses in reef growth associated with M. exaesa or rubble layers bound stratigraphically by dated corals. Symbols with a black asterisk were used to calculate the timing of the base and top of each reef unit (table S4). Horizontal gray arrows denote the uncertainty on the bounding ages on reef growth.

Fig. 3.. Comparison of sea level, North Atlantic proxies, speleothem, and ice core records.

Sea level data on left panel, top to bottom: Coral ²³⁰Th data for the shallowest corals from the Seychelles [this study and (9)], Western Australia and Barbados (84), and submerged speleothem ²³⁰Th from the Yucatan (69). The GIA correction introduces an additional vertical uncertainty, not shown here, to the datapoints from the Seychelles and Western Australia. Deep sea and speleothem data on middle panel, top to bottom: North Atlantic deep sea core MD03-2664 data (51, 60) on the Corchia cave timescale (52) for benthic δ¹⁸O and δ¹³C, % ice-rafted debris (IRD) and Neogloboquadrina pachyderma (sinistral) (Nps) coiling ratio, summer sea surface temperature (SST) estimate, Asian monsoon intensity (50), and speleothem δ¹⁸O for Corchia Cave denoting aridity (52). Timing of the red sediment layer (LILO) in the Labrador Sea denoted by red arrow. Ice core data on right panel, top to bottom: NEEM ice core δ¹⁸O (5), Greenland synthetic temperature (85), and NH insolation (40), atmospheric CO₂ (47) and CH₄ (48), Epica Dome C (EDC) δD (4), and South Hemisphere insolation (40). Red vertical bars denote timing of discontinuities in the Seychelles reef section, and gray vertical bars are defined by labels in the left panel.

Fig. 4.. Phasing of ice sheet retreat in each hemisphere from ice sheet models and ice sheet budgets for two time slices.

(A) Early retreat of the Antarctic ice sheet [AIS-1 scenario b from (45), AIS-2 from (14), AIS-3 from (46)], two models showing later maximum retreat of the Greenland ice sheet, GIS-1 from (37) and GIS-2 from (38). Stars denote the timing and elevation of GMSL by adding the two GIS scenarios to the higher AIS-1 scenario (solid stars) and to AIS-3 (open stars). The total sum GMSL over time reflects the sometimes-competing influence of ice decay versus ice growth. Evolution of the LIS is not depicted but is inferred to supply an additional source of meltwater at ~126 ka B.P. (see Discussion). (B) Implied magnitude of individual ice sheet contributions from these transient ice sheet models early in the LIG and near the time of peak GMSL. As shown, any residual ice in the LIS early in the LIG (magnitude unknown) would have to be compensated by meltwater from another source such as the AIS to match the observed GMSL. Hence, even if GMSL was the same as present at ~129 ka B.P. (54), if there was a residual LIS present, then this must have been compensated with meltwater from another source.