Sea-level stands from the Western Mediterranean over the past 6.5 million years

Abstract

Sea-level reconstructions are important for understanding past ice sheet variability and its response to past and future warming. Here we present Neogene and Quaternary sea-level snapshots using phreatic overgrowths on speleothems (POS) from caves on Mallorca, Spain. POS are excellent sea level index points because of their clear relationship to sea level and precise U–Pb chronology. We find that local sea-level before and at the onset of the Messinian Salinity Crisis was at 33.3 ± 0.25 m (6.54 ± 0.37 Ma) and 31.8 ± 0.25 m (5.86 ± 0.60 Ma) above present levels, respectively. We further present global mean sea level (GMSL) estimates, i.e. local sea level corrected for glacial isostatic adjustment and long-term uplift, for three other POS. The results show that GMSL during the Pliocene–Pleistocene Transition was 6.4 m (− 2.0–8.8 m) at 2.63 ± 0.11 Ma and during the beginning and the end of the Mid-Pleistocene Transition was − 1.1 m (− 5.6–2.4 m) and 5 m (1.5–8.1 m), respectively. These estimates provide important constraints for the past evolution of sea level and show that local sea level prior to the MSC was similar to the highest stand during the Pliocene, with markedly lower position afterwards.

Sea level and deep-sea temperature reconstructions suggest quasi-stable states and critical transitions over the past 40 million years

“…Compared with the Red Sea record ( 11 ), our new Δ SL records agree much better with the multiproxy assessment for MIS 11 ( 48 ), a further indication that the Red Sea record may underestimate sea level at that time. All Middle Pliocene sea level benchmarks shown ( 49 , 50 ) agree with our Δ SL , although the youngest does so only at its extreme age uncertainty. Our Δ SL may offer a means to refine age estimates for benchmarks with relatively large dating uncertainties at times of large Δ SL variations.…”

“…However, Middle Pliocene sea level benchmarks ( Figs. 3A and 4A ) ( 49 , 50 ) and Miocene best and maximum sea level amplitude estimates from New Jersey and the Delaware Coastal Plain ( Fig. 4A ) ( 53 ) indicate that the previous δ c -based Δ SL deconvolution ( 4 ) is anomalously flat and low between 3 and 13 Ma ago and that its amplitude variations between 13 and 22 Ma ago exceed independently established amplitude constraints ( 53 ).…”

“…Cyan boxes indicate maximum amplitude envelopes of New Zealand relative sea level fluctuations ( 52 ), vertically positioned for comparison. Green dots are western Mediterranean benchmarks ( 49 , 50 ), with highlighted Middle Pliocene range (yellow box). ( B ) T w changes derived using δ c and δ w , where δ w is calculated from Δ SL [this study; black and pink using Δ SL in (A); orange using Δ SL from the unconstrained regression; fig.…”

“…( D ) T w changes based on δ c residuals in(C) (black, pink, and orange). Regular italic numbers indicate positive validations: 1 , western Mediterranean benchmarks ( 49 , 50 ); 2 , best- and maximum-amplitude sea level ranges ( 53 ); 3a and 3b, first major iceberg calving in Nordic Seas ( 59 ) and LIS calving onset ( 60 ), respectively; 4 , AIS-volume variability range ( 61 ); 5 , onset partial/ephemeral northern ice ( 1 ); 6 , end of last intermittently ice-free period ( 3 ); 7 , AIS glaciation onset ( 1 , 3 , 4 , 64 , 66 – 70 ); 8 , MMCT δ w change of 0.35 ± 0.12‰ ( 62 ); 9a and 9b, loess fit through Mg/Ca T w with 95% confidence bounds; at 34.5 to 37.5 Ma ago, data were questioned ( 67 ). Roman numerals indicate discrepancies: i , amplitudes of sea level change in North West Australia ( 63 ); ii , EOT sea level drop of 70 to 80 m ( 64 ) versus ~30 m in this study (Discussion); iii , 2.5°C EOT cooling ( 66 ) versus 3.5°C cooling here (Discussion); iv , MMCT cooling of 1.5° ± 0.5°C ( 62 ) versus 2.5° ± 0.5°C here.…”

“…Next, we observe reasonable agreement between our new δ c regression–based Δ SL and independent multiproxy sea level reconstructions for interglacial Marine Isotope Stages (MIS) 5e and 11 ( 48 ), the Red Sea sea level record ( 11 ), and Middle Pliocene sea level benchmarks ( Figs. 2, A and C , and 3A ) ( 49 , 50 ). The δ c records that underpin our regression-based Δ SL do not fully resolve millennial-scale features, so our new (δ c regression–based) Δ SL records are likely to miss short-term extremes of several-meters amplitude that can be important in interglacials like MIS 5e and 11 [e.g., ( 51 )].…”