Convergence of atmospheric and North Atlantic carbon dioxide trends on multidecadal timescales

Oceanic uptake of carbon dioxide substantially reduces the rate at which anthropogenic carbon accumulates in the atmosphere1, slowing global climate change. Some studies suggest that the rate at which the oceans take up carbon has significantly decreased in recent years2–8. Others suggest that decadal variability confounds the detection of long-term trends9–11. Here, we examine trends in the partial pressure of carbon dioxide in the surface waters of three large biogeographic regions in the North Atlantic, using observational data collected between 1981 and 2009. We compare these oceanic observations with trends in atmospheric carbon dioxide levels, taken from a global observational network. We show that trends in oceanic carbon dioxide concentrations are variable on a decadal timescale, often diverging from trends in atmospheric carbon dioxide. However, when the entire 29-year period is considered, oceanic trends converge with atmospheric trends in all three regions; it takes 25 years for this long-term trend to emerge and overcome the influence of decadal-scale variability. Furthermore, in the southernmost biome, the data suggest that warming—driven by a multidecadal climate oscillation and anthropogenic forcing12,13—has started to reduce oceanic uptake of carbon in recent years. The ocean is the ultimate long-term sink for anthropogenic carbon, having taken up approximately 30% of anthropogenic emissions from pre-industrial times to 1994 (ref. 1). Anthropogenic climate change may drive physical and biogeochemical shifts in the ocean that result in reduced efficiency of this sink. Detection of such ‘climate-carbon feedbacks’ is of great interest, but is complicated by the influence of poorly quantified decadal timescale variability2–11,14,15. Previous studies have estimated trends in the North Atlantic carbon sink from oceanic pCO2 data and numerical model output for recent decades, but have not agreed as to its direction and magnitude2–11,16. Comparison of these studies is complicated by the different time periods, regions, and methodologies used. Distinct from previous studies, we determine trends in oceanic pCO2 from data across three large biogeographic regions (‘biomes’)17 that together occupy 87%of the total area of theNorthAtlantic (Fig. 1a). The northern seasonally stratified subpolar gyre (SP-SS) biome is cold and biologically productive, the southern permanently stratified subtropical gyre (ST-PS) biome is warm and has low productivity, and between these extremes is the seasonally stratified subtropical (ST-SS) biome. Our focus on biome-scale trends is motivated by relevance to the global scale partitioning of CO2 between the atmosphere and the ocean. Our methodology takes advantage of the strengths of both methods previously used to study trends in the ocean carbon uptake