The Atlantic meridional overturning circulation (AMOC) is a system of ocean currents that has an essential role in Earth's climate, redistributing heat and influencing the carbon cycle. The AMOC has been shown to be weakening in recent years ; this decline may reflect decadal-scale variability in convection in the Labrador Sea, but short observational datasets preclude a longer-term perspective on the modern state and variability of Labrador Sea convection and the AMOC. Here we provide several lines of palaeo-oceanographic evidence that Labrador Sea deep convection and the AMOC have been anomalously weak over the past 150 years or so (since the end of the Little Ice Age, LIA, approximately AD 1850) compared with the preceding 1,500 years. Our palaeoclimate reconstructions indicate that the transition occurred either as a predominantly abrupt shift towards the end of the LIA, or as a more gradual, continued decline over the past 150 years; this ambiguity probably arises from non-AMOC influences on the various proxies or from the different sensitivities of these proxies to individual components of the AMOC. We suggest that enhanced freshwater fluxes from the Arctic and Nordic seas towards the end of the LIA-sourced from melting glaciers and thickened sea ice that developed earlier in the LIA-weakened Labrador Sea convection and the AMOC. The lack of a subsequent recovery may have resulted from hysteresis or from twentieth-century melting of the Greenland Ice Sheet . Our results suggest that recent decadal variability in Labrador Sea convection and the AMOC has occurred during an atypical, weak background state. Future work should aim to constrain the roles of internal climate variability and early anthropogenic forcing in the AMOC weakening described here.

In the mid-1990s the North Atlantic subpolar gyre warmed rapidly (1), which had important climate impacts, such as increased hurricane numbers (2), and changes to rainfall over Africa, Europe and North America (3; 4). Ev

Quantifying signals and uncertainties in climate models is essential for climate change detection, attribution, prediction and projection [1][2][3] . Although inter-model agreement is high for large-scale temperature signals, dynamical changes in atmospheric circulation are very uncertain 4 , leading to low confidence in regional projections especially for precipitation over the coming decades 5, 6 . Furthermore, model simulations with tiny differences in initial conditions suggest that uncertainties may be largely irreducible due to the chaotic nature of the climate system 7-9 . However, climate projections are difficult to verify until further observations become available. Here we assess retrospective climate predictions of the last six decades project (GA 776613). FJDR, LPC, SW and RB also acknowledge the support from the EUCP project (GA 776613) and from the Ministerio de Economía y Competitividad (MINECO) as part of the CLINSA project (Grant No. CGL2017-85791-R). SW received funding from the innovation programme under the Marie Skĺodowska-Curie grant agreement H2020-MSCA-COFUND-2016-754433 and PO from the Ramon y Cajal senior tenure programme of MINECO. The EC-Earth simulations were performed on Marenostrum 4 (hosted by the Barcelona Supercomputing Center, Spain) using Auto-Submit through computing hours