Active Nordic Seas deep-water formation during the last glacial maximum (original) (raw)

The Nordic Seas are the primary location where the warm waters of the North Atlantic Current densify to form North Atlantic Deep Water, which plays a key part in the modern Atlantic Meridional Overturning Circulation. The formation of dense water in the Nordic Seas and Arctic Ocean and resulting ocean circulation changes were likely driven by and contributed to the regional and global climate of the last glacial maximum (LGM). Here, we map the source and degree of mixing of deep-water in the Nordic Seas, and through the Arctic Gateway (Yermak Plateau) over the last 35 thousand years using neodymium isotopes (εNd) measured on authigenic phases in deep-sea sediments with a high spatial and temporal resolution. We find that a large-scale reorganisation of deep-water formation in the Nordic Seas took place between the LGM (23-18 thousand years ago) and the rapid climate shift that accompanied the subsequent deglaciation (18-10 thousand years ago). We show that homogeneous εNd signatures across a wide range of sites support LGM deepwater formation in the Nordic Seas. In contrast, during the deglaciation disparate and spatially variable εNd values are observed leading to the conclusion that deep-water formation may have been reduced during this time. Deep-water formation processes in the Nordic Seas regulate the global climate via the redistribution of heat by the surface ocean and the capacity of the deep ocean to store carbon 1. At present the Atlantic Meridional Overturning Circulation (AMOC) links polar and sub-polar climate with the formation of North Atlantic Deep Water (NADW), a major component of the global oceanic thermohaline circulation. The densest northern-sourced waters in the modern AMOC are formed in the Nordic Seas, primarily by deep convection and gradual transformation of North Atlantic surface waters 2. These dense waters formed in the modern Nordic Seas overflow the Greenland-Scotland Ridge (GSR), eventually contributing to NADW accumulating carbon and nutrients as it flows throughout the deep ocean 2 (Fig. 1). The extent, mechanism, and importance of deep-water formation in the Nordic Seas during glacial periods and periods of ice rafting during meltwater events (Heinrich Events/Heinrich Stadials) are still not adequately understood. The canonical view is that the glacial AMOC was displaced from the Nordic Seas to south of Iceland in the form of a fast and shallow overturning cell forming Glacial North Atlantic Intermediate Water and that there was Southern-sourced water in the deep (> 2.5 km) Atlantic e.g,3. Contrary to this several studies e.g.4,5 argue for the presence of glacial NADW and speculate that this dense water may have been sourced from the Nordic Seas. Keigwin and Swift 6 similarly suggest that a Northern-sourced water mass may have been present in the deep (~ 5000 m) Atlantic, which could plausibly have been sourced from the Nordic Seas 7. However, proposed scenarios of LGM deep-water formation in the Nordic Seas range from near-cessation to vigorous present-day-like deep-water formation 8-11. There is evidence supporting a continued or intermittent subsurface inflow of the North Atlantic Current (NAC) during the LGM 12,13 to the Norwegian Sea. Polynya formation proximal to ice-sheets has been inferred, ventilating parts of the LGM deep Nordic Seas 7,9,14. Several proxy studies indicate a persistent overflow from the Nordic Seas into the glacial Atlantic Ocean 8,15,16. However, warmer waters (~ 1 to 2 o C warmer than the modern) in the intermediate to deep Nordic Seas and Arctic 11,17,18 indicate a reduced heat release to the atmosphere and less net cooling of the NAC waters, which could be due to subsurface expansion and deepening of the NAC 13,19. This does not support widespread deep-water formation by brine rejection or modern-like open-ocean convection, which would produce cooler waters at depth. Nevertheless, periodic cooler bottom water temperatures have been observed in LGM-aged sediments near Svalbard and in the Lofoten Basin 18,20. Less efficient deep-water formation via convective processes, upwelling, slow modification and return of Atlantic waters, or small-scale brine formation at shelf edges could be consistent with studies to date. Ultimately, the question of Nordic Seas deep-water formation during the LGM and its geographical extent remains an open debate which has not yet been fully constrained by proxy studies. Resolving the location and extent of deep-water formation under glacial conditions is key to understanding the link between climate, the oceans, ice-sheets, heat transport and carbon cycling. In this study, therefore, we provide, at a high spatial and temporal resolution, a depiction of past Nordic Seas circulation under glacial conditions. Neodymium isotope tracing of ocean circulation Neodymium (Nd) isotopes measured on authigenic phases (authigenic εNd) are a powerful tool used to trace water mass circulation 21. εNd is a proxy of the source of the Nd in the water mass. Spatial records can be related to each other to trace the flow path of water masses, and deduce their mixing with other water masses, so long as new sources of Nd are not added. However, the local input of Nd to water masses 22,23 can also alter the dissolved seawater εNd composition. This is likely to be of greater relative importance in controlling spatial patterns of εNd during times of reduced advection. High resolution εNd We measured authigenic εNd in 17 core sites to map out the geographical extent and consistency, or inconsistency, of Nordic Sea water-mass compositions from the last glacial period (~ 35 ka) to the late Holocene (< 5 ka). Individual core records from selected high-resolution core sites are shown in Fig. 2. Data are compared to similar records from the NE Atlantic and central Arctic Ocean as records in Fig. 2 and are also presented as time-slice cross sections in Fig. 3. Due to the high resolution of this data set and to understand the larger scale pattern of εNd, the data are temporally averaged. The data group naturally into 3 equal sized (5 ka) time intervals (Fig. 4), comprising the late Holocene (5-0 ka), the deglaciation (18-13 ka) and the LGM (23-18 ka). This compilation is focused on the Nordic Seas and Arctic Gateway (Yermak Plateau). The data are shown as probability density plots and histograms (Fig. 4). We compare the late Holocene with seawater values from the same depth and latitudinal range (Fig. 4), as well as to the LGM and deglacial values. These datasets were also compared using statistical tests (details are given in the Methods and Extended Data Fig. 5 and Extended Data Tables 1 and 2). The late Holocene compositions (0-5 ka) observed in the Nordic Seas and Yermak Plateau show the same spatial patterns (Fig 3A,B) and variability (Fig 4) as the modern across a range of core sites from the shallowest (at 488 m water depth, today bathed in warm northward flowing Atlantic waters on the Yermak Plateau) to the deepest in the Greenland Sea (3050 m water depth). This demonstrates that for the late Holocene, a seawater-derived signature is recorded with εNd on authigenic phases. In addition, the similar spatial pattern to the modern implies a hydrographic link between the inflow and deeper regions in the Nordic Seas and the Yermak Plateau. The homogeneity of the modern and late Holocene dataset relative to the large Nd isotope range of potential sources (which span almost the entire crustal array of εNd, ~-40 on Greenland, to ~ 10 on Iceland) also indicates vigorous circulation acting to homogenize compositions in the Nordic Sea, with a value of ~-10 (Fig. 4). The Yermak Plateau provides the ideal locality to test for changes in the NAC composition over time, because this is where warm high salinity Atlantic-derived subsurface waters are in contact with the sediment-water interface 26,27. A shallow sediment core from the Yermak Plateau (Fig. 2c, at 488 m) is used to monitor past changes in this Atlantic-derived endmember. Many studies indicate the continued strong influence of warm Atlantic waters in this region at the LGM, including at this core site 12,26. We compare this shallow core site to two other cores at different water depths (798 m and 2531 m). These Yermak Plateau cores sites have differing sedimentation rates (4, 6 and 10 cm/kyr) and likely distinctive sediment provenances 28. All the sites show similar changes through time (Fig. 2c). The three core sites are hydrographically linked in the modern ocean by vigorous circulation and the influence of Atlantic-derived waters across the Arctic Ocean 27. The Holocene εNd at these three core sites are within error of modern compositions 29,30. During the LGM the Yermak Plateau εNd averages to-13.1±0.9 (2σ (standard deviation) with a similar homogeneity and stability to the late Holocene but systematically offset in composition. The strong co-variation between these sites and the homogeneous LGM and Holocene compositions (Fig. 2C) indicates that these sites are recording an advected seawater signature resulting from Atlantic inflow and deep-water mixing and not localised sediment inputs or pore fluid processes. While some mixing with fresh and intermediate waters and localized inputs of Nd from sediment may change the NAC εNd along its pathway, the homogeneity of the signature at the LGM supports deep-water mixing that dominates over any local process. The LGM εNd at the Yermak Plateau is within error of modern composition of the NAC as it enters the Nordic Seas (-12.9±1.1 (2σ (standard deviation) 31), suggesting that this signature might be derived entirely from Atlantic waters. However, the past NAC endmember composition is unknown, and it may well have changed. The similar standard deviation of LGM εNd (Fig 2c) as the modern, regardless of the absolute value, over several residence times of Nd in the ocean, implies that at least the shallow site records the LGM NAC composition entering the...