In recent decades, the Arctic sea-ice has dramatically decreased1, and the sea-ice decline causes freshening of the Arctic Ocean, especially in the oceanic Beaufort Gyre of the western Arctic Ocean (Fig. 1), where freshwater has been accumulated2,3. Pacific water (PW) inflow through the Bering Strait, which increased 70% from 2001 (0.7 Sv) to 2014 (1.2 Sv), also contributes to the freshening and warming in the western Arctic4. The anomalous PW influence is called Pacification5. The PW transports a large quantity of nutrients; however, the accumulation of freshwater within the Beaufort Gyre deepens the nutrient-rich PW6 and may decrease phytoplankton production7. In contrast, increased eddies due to the sea-ice loss could replenish nutrients and may increase the phytoplankton production8.
On the other side, recent changes in the eastern Arctic Ocean are frequently described in terms of Atlantification, which is related to the progression of anomalies carried by Atlantic water (AW) into the eastern Arctic9 and intensified with the Arctic sea-ice decline10. Atlantification accompanies salinification, shoaling of the AW layer, weakening of the ocean stratification, uplifts of nutrient-rich water, and a possible increase in phytoplankton biomass by the vertical nutrient supply5 and the phytoplankton transport by AW11. Thus, the responses of ocean conditions to the sea-ice loss are quite different between the western (Pacific) and eastern (Atlantic) Arctic12. However, due to a lack of data, we have little knowledge about the changes in ocean environments in a boundary region between the PW and AW influences.
Using fragmented data in the boundary region, it was suggested that the Pacific/Atlantic boundary shifted its position along the shelf-slope of the East Siberian Sea (ESS), depending on cyclonic or anticyclonic modes of Arctic atmospheric circulation13,14. In addition to the PW and AW, water derived from the ESS plays an important role in nutrient and phytoplankton distributions in the boundary region15. In this region, a cyclonic atmospheric circulation pattern in 2017 induced an eastward extension of Atlantic-origin water (colder than a core of AW) that pushed up the nutrient-rich PW and ESS water, resulting in anomalously high surface phytoplankton blooms16. Thus, there seem to be more complicated processes than Pacification and Atlantification in the boundary region. In 2020, a multiship sampling campaign covering a wide research area in the western Arctic Ocean, including the Pacific/Atlantic boundary region, was conducted under the Synoptic Arctic Survey (SAS)17. During the survey, anomalous water characteristics were observed in the boundary region. Here we examine the formation mechanism of this water by combining the data obtained by the R/V Araon (Korea), R/V Mirai (Japan), and CCGS Louis S. St-Laurent (Canada) in the SAS project.
Anomalous water found in the Pacific/Atlantic boundary region
We found extremely low dissolved oxygen (ExLowDO < 200 μmol kg−1; Fig. 2a) in water on a southern portion of the Chukchi Plateau (~75° N, 168° W) in 2020 that has never been observed there at least in the last two decades (Fig. 2b). The ExLowDO water also corresponded to acidified water, which was corrosive to aragonite (i.e., saturation state Ω < ~0.7; Fig 2c) and almost corrosive to calcite (Ω < ~1.1; Fig 2d). For simplicity, we call it corrosive to calcium carbonate (CorrCaCO3). This anomalous DO and Ω water may impact marine ecosystems around the Chukchi Plateau, which is expected to be a fishable area in the future ice-free Arctic Ocean18,19. Polar species are most sensitive to changes in DO20.
Origin of the anomalous water
The correspondence of ExLowDO and CorrCaCO3 in the water on the Chukchi Plateau (Fig. 2) suggests a high quantity of organic matter decomposition that used DO and produced CO2, resulting in large decreases in DO and Ω, respectively. However, the transmission of ExLowDO water on the Chukchi Plateau was much higher than that of the bottom water on the Chukchi shelf (Extended Data Fig. 1a, b), implying a smaller accumulation of organic particles on the plateau seafloor than on the shelf bottom. Thus, the organic matter decomposition characterizing the ExLowDO and CorrCaCO3 would occur in a remote region.
Such ExLowDO and CorrCaCO3 waters were found on the ESS shelf with salinity (S) = ~30–32, and on the shelf-slope where S of the water is 34–34.5 (Extended Data Fig. 2a, b) (ref. 21). The ExLowDO and CorrCaCO3 water on the shelf-slope would be formed by the brine rejection in the ESS shelf22 and/or a long-time contact of the water to the shelf-slope sediments23. The ExLowDO and CorrCaCO3 water in the ESS is also characterized by low NO, where NO is defined as 9[NO3-] + [O2] (µmol kg−1) and used as a quasi-conservative tracer that is independent of biological processes24. The low NO in the ESS is likely caused by an input of river water that contains low preformed nitrate, sedimentary denitrification, and decomposition of terrestrial organic matters, which are relatively nitrogen-poor compared to marine organic matters25. In addition, a ratio of NO to PO (NO/PO), where PO is another quasi-conservative tracer defined as 135[PO43-] + [O2] (µmol kg−1)24, likely has a value less than 0.75 in the ESS, which is the lowest value among pan-Arctic shelves26. The ESS shelf-slope water had values of low NO (200–300 µmol kg−1; Extended Data Fig. 2c) and NO/PO < 0.75 (Extended Data Fig. 2d). The anomalous DO and Ω water on the Chukchi Plateau was also low NO (~230 µmol kg−1; Extended Data Fig. 1c) and NO/PO = ~0.66 (not shown). Therefore, the ESS shelf-slope water with S = 34–34.5, NO = 200–300 µmol kg−1, and NO/PO < 0.75 was probably transported to the Chukchi Plateau, on which we found the similar water characteristics (ExLowDO and CorrCaCO3 water with S ~ 34.5, NO = ~230 µmol kg−1, and NO/PO = ~0.66).
Historical data indicate that the bottom water of ExLowDO (<200 μmol kg−1; Extended Data Fig. 3a) occupied the shelf-slope of ESS between 170° E and 170° W with S = ~33–34.5 (Extended Data Fig. 3b). West of 170° E, AW would penetrate from the west along the shelf-slope, and PW occupied east of 170° W and south of 73° N in the Chukchi shelf. Both the AW and PW were likely well ventilated in winter27, so DO concentrations could be high. The shallow ESS shelf might also be ventilated in winter and influenced by the PW penetration from the east28. Consequently, the ExLowDO zone along the ESS shelf-slope is thought to be a region where AW and PW influences were minimal, and thus, recognized as a shadow (less-ventilated) zone of the AW and PW inflows. The shadow zone is consistent with the lowest concentrations of an artificial inert gas, sulfur hexafluoride (SF6), in the ESS shelf-slope23.
Changes in PW and AW circulation
We examined ocean circulation that determines water distributions, including the ExLowDO water (Fig. 3). The circulation pattern in 2020 was compared with that in 2008–2016, which was selected as a period when the Beaufort Gyre extent was large under an anticyclonic mode of atmospheric circulation16,29. These circulation patterns and DO distributions were also compared with that obtained from historical data in a heavy sea-ice era before 2000 (during 1948–2000 for DO30 and 1948–1993 for temperature and S data31). The circulation pattern in each case was calculated as the dynamic height at a depth of 100 m, which is a suitable depth to detect the Pacific/Atlantic boundary in the previous study25 and as described below.
In 2020 (Fig. 3a), an anticyclonic circulation, the Beaufort Gyre, occupied east of the Chukchi Plateau (~168° W), and northward flows were predominant west of the plateau. However, in 2008–2016 (Fig. 3b), the Beaufort Gyre extended toward the west, and the westward flow of the southern branch of the gyre overshot the Chukchi Plateau. As a result, the northward flow over the Chukchi Plateau was very weak compared with that in 2020 (>2.5 cm s−1 at ~100 m in 2020, but <1 cm s−1 in 2008; Extended Data Fig. 4). Before 2000 (Fig. 3c), the Beaufort Gyre was diminished to the east of Chukchi Plateau, similar to 2020, but the flows were very sluggish. This is probably because heavy and less mobile sea-ice before 2000 inhibited the input of wind-stress curl to the ocean32. The atmospheric circulation generally controls the extent of the Beaufort Gyre, and the gyre extent is likely to oscillate in a decadal timescale33,34. However, an ongoing trend of decreasing sea-ice in the eastern Arctic and the resultant intrusion of North Atlantic cyclones into the western Arctic eventually collapsed the atmospheric Beaufort High in 2017 for the first time35, resulting in a shrink of the oceanic Beaufort Gyre in 201716 and beyond including 2020 (Fig. 3a). In the case of small Beaufort Gyre extents (2020 and before 2000), its northward flow could be along the Chukchi Plateau as a western boundary current of the gyre36.
The circulation patterns would be associated with temperature and S distributions along 75° N across the Chukchi Plateau (Fig. 3d–f). The Beaufort Gyre transports cold PW that has entered the Arctic Ocean through the Bering Strait in winter and is characterized by a temperature minimum around S = ~3337. Another cold water from ESS (S = 32–33) occupied west of the Beaufort Gyre15. Layers below the cold waters are occupied by lower halocline water (LHW) derived from the eastern Arctic (S = 34–34.5)37,38 and warm AW (temperature > 0 °C, S > 34.5)39. In 2020 (Fig. 3d), a strong front between the cold PW and eastern Arctic-derived LHW was found at a depth of 100 m just west of the Chukchi Plateau with a shoaling of the AW. This front caused a significant northward flow (>2.5 cm s−1) there (Fig. 3a and Extended Data Fig. 4a). Such a frontal structure was not found in 2008–2016 (Fig. 3e and Extended Data Fig. 4b) because the cold PW penetrated further west of the Chukchi Plateau with the extension of the Beaufort Gyre (Fig. 3b). Before 2000, the cold PW of S = ~33 was only found east of the Chukchi Plateau (Fig 3f), consistent with the shrunk Beaufort Gyre (Fig. 3c). On the other hand, west of the Chukchi Plateau, the cold water of S = 32–33 might be derived from ESS, and the influences of LHW and AW were likely small.
In addition to the Beaufort Gyre extent, Atlantification would be another important factor characterizing the temperature and S distributions. The shoaling of AW in 2020 (Fig. 3d) is thought to be a direct signal of Atlantification, which has been significant in the eastern Arctic since the mid-2010s9. Considering the timescale of AW circulation (<ten years)40, the signal of Atlantification is likely to have reached the western Arctic. Atlantification may further influence the water column above the AW. It would cause the declining sea-ice in the northern Barents Sea, which is one of the formation sites of the LHW that initially has a near-freezing temperature41, resulting in a weakening of stratification above the AW that could induce a well-mixing between the LHW and AW39,42. These processes would make the LHW warmer, saltier, and less stratified. Such a less-stratified water column (temperature = ~−1 to 0 °C, S = 34 to >34.5) penetrated near the Chukchi Plateau in 2020 (Fig. 3d). The eastward penetrations of LHW and AW were presumably associated with the collapse of the atmospheric Beaufort High35. The LHW and AW could further flow along isobaths and spread into the Canada Basin41,43. However, the spreading, especially the LHW, was likely hampered by the Beaufort Gyre, whose western boundary appeared over the Chukchi Plateau when the gyre shrunk. Thus, a strong front between the PW and LHW was found in 2020. On the other hand, the front was not found before 2000 because the penetrations of LHW and AW were small (i.e., Atlantification had not occurred yet).
DO distribution determined by ocean circulation
The penetration of LHW associated with Atlantification and the PW/LHW front would determine the fate of ExLowDO water on the ESS shelf-slope. We examined the DO distributions on isohaline (i.e., approximately equal density) surfaces of S = 34.5 in 2020 and 2008–2016 (Fig. 3g, h) and S = 33 in the era before 2000 (Fig 3i). Each isohaline surface is located in a layer including the ExLowDO water in each case (Extended Data Figs. 1a, 2a, 3a). The lower S (~33) of the ExLowDO water before 2000 on the ESS shelf-slope was attributed to a deeper level of LHW (S = 34–34.5; Fig. 3f) compared with that in 2020 and 2008–2016 (Fig. 3d, e). The DO distributions in 2008–2016 (Fig. 3h) and before 2000 (Fig. 3i) show that the lowest DO area on the isohaline surface appeared along the ESS shelf-slope. This area is thought to be a shadow zone. However, in 2020 the lowest DO area moved to the Chukchi Plateau (Fig. 3g). It further extended toward the north with increasing DO. The migration of the lowest DO area suggests that the ExLowDO water in the shadow zone was washed by an eastward penetration of LHW (i.e., higher DO water) along the ESS shelf-slope. Thus, the shadow zone there was no longer maintained. The washed-out ExLowDO water was likely transported to the north by the strong frontal flow between the PW and LHW over the Chukchi Plateau.
The ExLowDO water on the Chukchi Plateau could further spread into the deep Canada Basin by the frontal northward flow and the adjacent Beaufort Gyre. To examine the ExLowDO water spreading, we compared distributions of temperature, S, and DO along 150° W, which crosses the Beaufort Gyre in the Canada Basin, between 2020 and 2008–2016 (Fig. 4). The temperature and S distributions were similar between 2020 and 2008–2016 (Fig. 4a, b); the cold PW of S = ~33 occupied around 150–200 m, and the LHW (S = 34–34.5) and warm AW lied below the cold PW. On the other hand, DO at depths of 200–300 m was low in 2020 compared with 2008–2016 (Fig. 4c, d). A significant decrease in DO (−8 μmol kg−1) was found in the central basin in the layer of S = 34–34.5 (Fig. 4e), where the ExLowDO water with the same S range could contribute to reducing the DO concentration. Moreover, a large decrease in DO (−12 μmol kg−1) occurred at the southern rim of the basin (Fig. 4e). This DO decrease might be resulted from an extension of the ExLowDO water to the east of Chukchi Plateau along the Chukchi shelf-slope, being associated with the eastward transport of AW43. The DO decrease at the basin's southern rim may also be related to sedimentation processes such as increased export production and coastal erosion due to the recent sea-ice loss.