Deep oceanic overturning circulation in the Atlantic (Atlantic Meridional Overturning Circulation (AMOC)) is projected to decrease in the future in response to anthropogenic warming. Caesar et al.¹ argue that an AMOC slowdown started in the nineteenth century and intensified during the mid-twentieth century. Although the argument and selected evidence proposed have some merits, we find that their conclusions might be different if a more complete array of data available in the North Atlantic region is considered. We argue that the strength of AMOC over recent centuries is still poorly constrained and the expected slowdown may not have started yet.

Recently, Moffa-Sánchez et al.² compiled a comprehensive set of palaeoclimate proxy data from the North Atlantic and Arctic regions using objective criteria to identify high-quality datasets of ocean conditions that span the past two millennia (Fig. 1). Although no direct (singular) proxy for AMOC exists, the palaeoceanographic proxy data compiled by Moffa-Sánchez et al.² highlight the spatial and temporal complexities of the ocean state in modern times and the recent past. When all the available proxy records potentially related to AMOC variability and twentieth century observational datasets are considered, the time history of the AMOC system becomes less certain. In contrast, selecting only a subset of proxy records that share similar trends, as performed by Caesar et al.¹, provides an incomplete perspective on AMOC changes through time.