The Arctic Ocean has experienced rapid warming and sea ice loss in recent decades, becoming the first open-ocean basin to experience widespread aragonite undersaturation [saturation state of aragonite (Ω _arag ) < 1]. However, its trend toward long-term ocean acidification and the underlying mechanisms remain undocumented. Here, we report rapid acidification there, with rates three to four times higher than in other ocean basins, and attribute it to changing sea ice coverage on a decadal time scale. Sea ice melt exposes seawater to the atmosphere and promotes rapid uptake of atmospheric carbon dioxide, lowering its alkalinity and buffer capacity and thus leading to sharp declines in pH and Ω _arag . We predict a further decrease in pH, particularly at higher latitudes where sea ice retreat is active, whereas Arctic warming may counteract decreases in Ω _arag in the future.

The Arctic Ocean has turned from a perennial ice‐covered ocean into a seasonally ice‐free ocean in recent decades. Such a shift in the air‐ice‐sea interface has resulted in substantial changes in the Arctic carbon cycle and related biogeochemical processes. To quantitatively evaluate how the oceanic CO₂sink responds to rapid sea ice loss and to provide a mechanistic explanation, here we examined the air‐sea CO₂flux and the regional CO₂sink in the western Arctic Ocean from 1994 to 2019 by two complementary approaches: observation‐based estimation and a data‐driven box model evaluation. ThepCO₂observations and model results showed that summer CO₂uptake significantly increased by about 1.4 ± 0.6 Tg C decade⁻¹in the Chukchi Sea, primarily due to a longer ice‐free period, a larger open area, and an increased primary production. However, no statistically significant increase in CO₂sink was found in the Canada Basin and the Beaufort Sea based on both observations and modeled results. The reduced sea ice coverage in summer in the Canada Basin and the enhanced wind speed in the Beaufort Sea potentially promoted CO₂uptake, which was, however, counteracted by a rapidly decreased air‐seapCO₂gradient therein. Therefore, the current and future Arctic Ocean CO₂uptake trends cannot be sufficiently reflected by the air‐seapCO₂gradient alone because of the sea ice variations and other environmental factors.

The distributions of dissolved O₂and CO₂have not previously been systematically compared across the global surface ocean, despite their significance for life and climate. Here we analyze carbon dioxide and oxygen concentrations relative to saturation (equilibrium with the atmosphere) in surface waters, using two large datasets (ship-collected and float-collected data). When applied to a high-quality global ship-collected dataset, CO₂and O₂concentrations relative to saturation exhibit large seasonal and geographic variations. However, linear fits of CO₂and O₂deviations from saturation (ΔCO₂against ΔO₂) yield y-intercepts close to zero, which suggests a requirement for data validity. We utilize this finding to investigate the accuracy of carbonate system data from biogeochemical-Argo floats. We find significant discrepancies in ΔCO₂-ΔO₂y-intercepts compared to the global reference, implying overestimations of float-based CO₂release in the Southern Ocean. We conclude that this technique can be applied to data from autonomous platforms for quality assessment.

To examine seasonal and regional variabilities in metabolic status and the coupling of net community production (NCP) and air‐sea CO₂ fluxes in the western Arctic Ocean, we collected underway measurements of surface O₂/Ar and partial pressure of CO₂ (pCO₂) in the summers of 2016 and 2018. With a box‐model, we demonstrate that accounting for local sea ice history (in addition to wind history) is important in estimating NCP from biological oxygen saturation (Δ(O₂/Ar)) in polar regions. Incorporating this sea ice history correction, we found that most of the western Arctic exhibited positive Δ(O₂/Ar) and negative pCO₂ saturation, Δ(pCO₂), indicative of net autotrophy but with the relationship between the two parameters varying regionally. In the heavy ice‐covered areas, where air‐sea gas exchange was suppressed, even minor NCP resulted in relatively high Δ(O₂/Ar) and low pCO₂ in water due to limited gas exchange. Within the marginal ice zone, NCP and CO₂ flux magnitudes were strongly inversely correlated, suggesting an air to sea CO₂ flux induced primarily by biological CO₂ removal from surface waters. Within ice‐free waters, the coupling of NCP and CO₂ flux varied according to nutrient supply. In the oligotrophic Canada Basin, NCP and CO₂ flux were both small, controlled mainly by air‐sea gas exchange. On the nutrient‐rich Chukchi Shelf, NCP was strong, resulting in great O₂ release and CO₂ uptake. This regional overview of NCP and CO₂ flux in the western Arctic Ocean, in its various stages of ice‐melt and nutrient status, provides useful insight into the possible biogeochemical evolution of rapidly changing polar oceans.