The Arctic region is undergoing rapid climatic and environmental change1, so knowledge of its past variability is crucial for understanding modern trends and predicting future ones. Ancient climate conditions and ocean behaviour are often reconstructed by analysing marine sediments. But Arctic sediments can be difficult to interpret, and much is still unknown about how the Arctic Ocean changed during specific glacial and interglacial periods over the past few million years2,3. Writing in Nature, Geibert et al.4 report analyses of an isotope of the element thorium in sea-floor sediments, which suggest that the Arctic Ocean swung between being filled with salt water and fresh water during periods of the two most recent glacials.
The authors base their argument on records of thorium-230, produced from the decay of dissolved uranium that is naturally present in seawater. Thorium is highly insoluble and sticks to solid particles such as dust grains or biological material, which sink to the sea floor and become buried in sediments5. Thorium that derives from the water column in this way is known as excess thorium-230 (230Thex). It is typically present in sediments deposited during the past 450,000 years and is often measured to determine sediment-deposition rates5,6. Geibert and colleagues’ innovation is instead to use these measurements to reconstruct how much 230Thex was produced in the Arctic Ocean over time, and thereby to determine how the salinity has changed.
The authors examined sediment cores from across the Arctic and Nordic seas, and found that 230Thex is absent in several layers of sediment deposited during the past 200,000 years. The cores suggest that no 230Th was produced in the water above the study sites between about 150,000 and 131,000 years ago (during the next-to-last glacial), 70,000 and 62,000 years ago (during early parts of the last glacial) and perhaps even as recently as about 15,000 years ago (at the end of the last glacial).
Thorium-230 produced in seawater is removed so rapidly by sinking particles that its net horizontal transport across the ocean is typically low5, even in the particle-poor Arctic. The intervals of absent 230Thex in the sediment cores therefore imply that the uranium concentration was low to non-existent in the water above the study sites when those sediments were deposited. This, in turn, implies that the entire water column was essentially fresh down to the sea floor — there were no dissolved salts of any type.
Thick ice shelves covered regions of the Arctic during previous glacials7. Geibert et al. posit that such ice shelves could have extended into the Nordic seas, possibly grounding on the Greenland–Scotland Ridge — the tall underwater feature that separates the Nordic seas from the rest of the Atlantic basin (Fig. 1). The ice shelves might, in effect, have dammed the Arctic and Nordic seas, isolating them from salty inflows from the Atlantic. The low sea levels at that time blocked the exchange of water with the Pacific Ocean through the Bering Strait. Fresh water from melting land ice and precipitation could therefore have entered and eventually filled the isolated northern basins.
An advantage of Geibert and co-workers’ 230Thex method is that, unlike many other techniques used in palaeoceanography, no biologically produced material is needed for the analysis. It can therefore be used to probe environments that would have had low to no biological productivity, such as a freshwater Arctic Ocean beneath ice shelves. Indeed, microfossils in the 230Thex-free sediment layers are absent or extremely rare, or are derived from older deposits rather than being contemporaneous with the 230Thex minima.
This new interpretation of 230Thex might also provide an intriguing means of reconciling contrasting results previously obtained from different methods of estimating past sea levels. The relative abundances of oxygen isotopes in global seawater are recorded in microfossils, and, in part, reflect the sequestration of evaporated ocean water into ice sheets or other freshwater reservoirs, which can affect sea level. For certain times during recent ice ages, sea-level records obtained from isotopic analyses of microfossils are inconsistent with records derived from corals8. Geibert et al. suggest that these inconsistencies could be explained by the proposed storage of large volumes of fresh water in the Arctic Ocean.
Various complications in the analysis will no doubt raise questions. Arctic sediments are notoriously hard to date owing to the lack of microfossils, and because sedimentation rates varied2,3. It is therefore uncertain whether the 230Thex-deficient intervals in the cores were produced at exactly the same times at all sites across the ocean basins. Moreover, the authors had to correct their data to account for 230Th that was produced from uranium decay in sediment grains, rather than in the water column5, and this contributed to the uncertainty in measured 230Thex. These corrections become proportionally more important for older sediments because 230Thex itself decays away; thorium decay also limits the time span over which the method can be used to investigate Arctic salinity. Finally, no freshwater fauna have been identified in the sediments concerned, so direct evidence of freshwater intrusion into deep Arctic basins remains to be found.
However, the various absences — of 230Thex, of microfossils and biological productivity, and of elements such as sulfur, which partly derive from salinity in marine sediments9 — suggest exciting avenues for future research. Computational modelling of Arctic Ocean circulation and ice-sheet behaviour will be needed to determine realistic estimates of the circulation and freshwater run-off from land that could produce a basin filled with fresh water. Further geochemical and fossil analyses might help to support or challenge the assertion that the Arctic Ocean could have been fresh. Geibert and colleagues’ innovative use of 230Th might spur a re-evaluation of what is possible in the Arctic Ocean, and of how dramatically this region can change.
Nature 590, 37-38 (2021)