Eastern Mediterranean water outflow during the Younger Dryas was twice that of the present day

The Mediterranean Sea exhibits a negative net precipitation–evaporation balance that drives an anti-estuarine thermohaline circulation (Med-THC). Fresh Atlantic waters enter through the Strait of Gibraltar, become saltier, and form Levantine Intermediate Water (LIW) in the eastern basin, which, together with Eastern Mediterranean Deep Water (EMDW) formed in the Adriatic and Aegean seas, exits westward through the Strait of Sicily as Eastern Mediterranean Source Waters (EMSW). EMSW and Western Mediterranean Deep Water (WMDW) eventually exit to the Atlantic as Mediterranean Outflow Water (MOW). Because the Mediterranean and Atlantic are tightly coupled via Gibraltar, past climate-driven changes could modify MOW properties and potentially influence the Atlantic Meridional Overturning Circulation. During the last deglaciation, sea-level rise increased Atlantic freshwater inflow, enhancing stratification and weakening deep-water convection in the western Mediterranean, fostering organic-rich layer formation. The African Humid Period (~15–6 kyr BP) intensified North African monsoon rainfall and river runoff into the Eastern Mediterranean, strengthening surface stratification and contributing to Sapropel S1 formation (10.8–6.1 kyr BP). The Younger Dryas (12.95–11.65 kyr BP) involved AMOC slowdown, altered precipitation patterns, reduced freshwater supply to the Mediterranean, and overall cold/arid conditions in the Eastern Mediterranean, possibly reactivating intermediate/deep-water convection. Yet, quantitative estimates of east–west exchange through the Strait of Sicily across these events have been lacking. This study aims to quantify changes in EMSW outflow since the last deglaciation using εNd as a water-mass tracer.

Previous work indicates Med-THC sensitivity to climate oscillations, with deglacial sea-level rise weakening western deep-water formation and promoting organic-rich layers. The African Humid Period increased Nile and North African runoff, enhancing stratification and preconditioning for Sapropel S1, though the timing and mechanisms of ventilation changes remain debated. Evidence for Younger Dryas conditions in the Eastern Mediterranean suggests colder and drier climate, with some studies proposing partial reactivation of intermediate/deep convection. Sapropel S1 is widely attributed to freshwater-induced stratification and enhanced export productivity leading to anoxia, with indications that ventilation recovery was depth-dependent and asynchronous (intermediate waters re-ventilating between ~7.7 and 7 kyr BP). Prior studies explored potential impacts of S1 on east–west exchange but lacked quantitative estimates. Neodymium isotopes provide a robust tracer: present-day εNd is more radiogenic in EMSW (EMDW and LIW) than in WMDW, enabling reconstruction of mixing and circulation changes.

Study site and sampling: Gravity core NDT-6-2016 (~4 m) was recovered in the transition between the western Sicily Channel and the southern Tyrrhenian Sea (38°0′26.60″N, 11°47′44.84″E; 1066 m water depth), below the modern WMDW–EMSW interface, ideal for tracking intermediate–deep exchange. The core comprises homogeneous silty-clay; samples were taken every 1 cm; εNd analyses were every 2–6 cm between 1.20–1.62 m and every 10–30 cm elsewhere. Chronology: Twenty radiocarbon ages on monospecific planktic foraminifera (Globoconella inflata, >250 µm) were calibrated with MARINE20. Three additional tie points aligned the δ18O record of Globigerina bulloides to the NGRIP ice-core chronology. A Bayesian Bacon age model was constructed. Neodymium isotope analyses: For each εNd measurement, 20–30 mg of mixed planktic foraminifera (>212 µm) were handpicked (4–12 mg when scarce), ultrasonically cleaned with ultrapure water and methanol to remove clays/particles, then weakly leached/dissolved with ultrapure water and acetic acid. Solutions were centrifuged; Nd was purified via two-step column chemistry (Tru-Spec for REE separation, Ln-Spec for Nd). Five bulk sediment samples (<63 µm) underwent sequential leaching to isolate detrital fraction (organic removal, carbonate leaching, authigenic removal) and full digestion before Nd purification using an automated DGA-based protocol. εNd was measured on a Nu Plasma 3 MC-ICP-MS (CCiTUB). Instrumental mass bias was corrected with the exponential law; JNdi-1 standard bracketing (143Nd/144Nd = 0.512115) was applied. εNd is reported relative to CHUR (143Nd/144Nd = 0.512638). External reproducibility based on JNdi-1 was 0.3–0.5 εNd units (2σ). Procedural blanks (26–45 pg Nd) were negligible. Replicate analyses at 147 cm yielded mean εNd −7.8 ± 0.3 (n = 4; 2σ). Assessing non-seawater contributions: Detrital εNd (<63 µm) was measured to evaluate porewater/sediment exchange and dust/terrigenous inputs; values were much less radiogenic (−12.06 to −11.58) and stable through time, inconsistent with the foraminiferal εNd variability. The core shows no sapropel layers/low-oxygen sediments that might drive diagenetic Nd remobilization. The topmost foraminiferal εNd (−7.9) matches modern regional seawater εNd at comparable depths, supporting fidelity. Mixing model: A three-endmember isotopic mixing model estimated EMSW contribution at W-Sicily using time-varying εNd endmembers from published records: WMDW (SU92-33), EMDW (MS27PT; Levantine basin, representing mixed Adriatic+Aegean contributions), and LIW (BC07). The mass-balance incorporated Nd concentrations for each water mass (modern means ±1σ: WMDW 23.05 ± 1.05 pmol/kg; LIW 27.8 ± 1.6; EMDW 28.25 ± 3.35). Because past seawater Nd concentrations are unknown, a Monte Carlo approach (10,000 simulations) propagated analytical uncertainties and concentration ranges. To resolve underdetermination of LIW vs EMDW within EMSW, three scenarios were evaluated: equal LIW=EMDW, LIW=2×EMDW, and EMDW=2×LIW. Endmember εNd series were linearly interpolated to the NDT-6-2016 age points. Present-day EMSW contribution estimates derived from modern seawater εNd and Nd concentrations for regional water masses served as a check (~30%), consistent with independent T–S–O2 estimates (20–40%).

• Foraminiferal εNd at W-Sicily varies strongly over the last ~15 kyr. Highest (more radiogenic) values occur during the Younger Dryas (−5.6 ± 0.4 to −6.3 ± 0.3), while the lowest occur during Sapropel S1 (−8.4 ± 0.3 to −7.8 ± 0.3). Pre- and post-S1 intervals show intermediate values (pre-S1: −7.0 ± 0.4 to −7.4 ± 0.6; post-S1: −7.3 ± 0.3 to −7.5 ± 0.3). • Detrital εNd (<63 µm) is stable and much less radiogenic (−12.06 to −11.58), indicating the foraminiferal εNd shifts reflect water-mass changes rather than variable sediment inputs or diagenesis. • Monte Carlo three-endmember mixing indicates enhanced EMSW at W-Sicily during the Younger Dryas: average EMSW contribution 57 ± 5% (n = 9), compared to pre-YD 23 ± 4% (n = 9) and post-YD 33 ± 5% (n = 15). Present-day EMSW at this site is ~30%. Thus, YD outflow was about twice present-day. • During S1, EMSW contribution is minimal: 16 ± 6% (n = 33), about half of pre-/post-S1 values (pre-S1 33 ± 5%, n = 15; post-S1 29 ± 2%, n = 9) and roughly one-third of YD levels. • Transition at the end of S1 shows EMSW increasing from 4 ± 8% to 28 ± 8% at W-Sicily, signaling re-establishment of east–west intermediate exchange between ~7.5 and 7 kyr BP, earlier than the canonical S1 end (~6.1 kyr BP). • The YD εNd peak coincides with independent evidence of stronger intermediate-depth currents in the western Mediterranean (grain-size maxima near Corsica and in the Alboran Sea) and intensified MOW speeds at the Strait of Gibraltar/Gulf of Cadiz, linking enhanced EMSW to stronger MOW. • Elevated εNd in the EMDW during the deglaciation/early Holocene in the Levantine Sea suggests radiogenic endmember influence (e.g., Nile inputs, Aegean contribution), but the confinement of the W-Sicily radiogenic anomaly to the YD indicates both endmember shifts and increased EMSW volume were required.

The findings quantitatively demonstrate that EMSW outflow into the western Mediterranean was maximized during the Younger Dryas, approximately doubling present-day levels and tripling those during Sapropel S1. The radiogenic εNd anomaly at W-Sicily in the YD reflects a combined signal: more radiogenic EMSW endmembers (due to enhanced contributions from sources such as the Aegean and/or Nile-influenced Levant) and increased EMSW volume relative to WMDW. Cold, arid YD conditions likely intensified Aegean deep-water convection, while contemporaneous weakening of western deep-water formation permitted deeper penetration of EMSW into the western basin. This enhanced intermediate exchange aligns with stronger intermediate currents and more oxygenated intermediate waters in the Alboran Sea and with intensified MOW at Gibraltar, implying basin-scale coupling between Eastern Mediterranean convection and Atlantic exchange. In contrast, during S1, freshwater-driven stratification in the Eastern Mediterranean curtailed intermediate/deep convection, reducing EMSW outflow and allowing more WMDW to occupy intermediate depths in the west. The observed increase in EMSW between ~7.5 and 7 kyr BP at W-Sicily indicates earlier re-ventilation of intermediate waters and resumption of east–west exchange before the end of S1, consistent with evidence for asynchronous, depth-dependent ventilation recovery in the Eastern Mediterranean. Overall, the results underscore the sensitivity of EMSW outflow to climatic forcing and its downstream impact on MOW dynamics and potentially on the Atlantic circulation.

Neodymium isotope records from core NDT-6-2016 provide a quantitative reconstruction of Eastern Mediterranean Source Water export to the western basin over the last deglaciation and Holocene. EMSW contribution peaked during the Younger Dryas (57 ± 5%), roughly twice modern and about three times higher than during Sapropel S1 (16 ± 6%). The YD enhancement likely arose from intensified Aegean/Levantine convection under arid, cold conditions combined with weaker western deep-water formation, facilitating deeper EMSW penetration. During S1, strong stratification suppressed EMSW export, with exchange re-intensifying between ~7.5 and 7 kyr BP, signaling early re-ventilation at intermediate depths. The temporal alignment between EMSW variability and MOW strength in the Gulf of Cadiz highlights the pivotal role of Eastern Mediterranean intermediate/deep convection in modulating Mediterranean–Atlantic exchange.

• Potential diagenetic contributions of reactive/labile Nd from sediments cannot be entirely excluded, although multiple lines of evidence (stable, unradiogenic detrital εNd; absence of local sapropel/low-oxygen sediments; agreement of core-top εNd with modern seawater) argue against significant overprinting. • The mixing model relies on time-evolving εNd endmembers from separate cores and assumes modern Nd concentrations for water masses; past seawater Nd concentrations are unknown, addressed via a Monte Carlo approach but still a source of uncertainty. • Endmember selection for EMDW (using Levantine data to represent mixed Adriatic/Aegean contributions) and assumed LIW:EMDW proportion scenarios introduce structural uncertainty. • Temporal resolution discrepancies among compared records (e.g., Corsica Channel) may obscure short-lived events. • Analytical uncertainties of εNd (typically 0.3–0.5 units, 2σ) and limited sample sizes at some depths constrain precision of estimates.