20th-Century Weakening of North Atlantic Meridional Heat Transport: Evidence from Global δ18O Proxies

The Atlantic Meridional Overturning Circulation (AMOC) plays a key role in regulating the global climate system by regulating meridional heat transport and facilitating deep-sea carbon uptake. Theoretical and modeling studies indicate that the AMOC may weaken—or even collapse—under anthropogenic warming, yet robust observational records remain limited. Continuous measurements at ~26.5° N (RAPID array) have been available only since 2004, restricting our knowledge of long-term AMOC variability and the associated regional and global feedback mechanisms. Here, we integrate outputs from the isotope-enabled Community Earth System Model Last Millennium Ensemble (iCESM-LME) to identify and isolate a primary mode of AMOC variability that strongly aligns with the RAPID array observations. To investigate potential teleconnections between the AMOC and globally distributed isotopic (δ¹⁸O) proxy records, we employ a suite of Proxy System Models to simulate four primary pseudoproxy archives—corals, speleothems, wood cellulose, and ice cores—forced by iCESM output. These pseudoproxies mimic the spatial distribution, seasonality, and time span of our compiled δ¹⁸O proxy database. Through a series of pseudoproxy experiments, we illustrate that the 'RAPID' mode manifests across several leading modes of variability in all four types of pseudoproxies. This coupling is further validated using the actual δ¹⁸O proxy records. Building on these findings, we apply a nested-PCA approach to extract the first globally distributed AMOC ‘RAPID’ mode signal from real isotopic proxy records spanning the past four centuries. The reconstructed signal reveals a gradual weakening of North Atlantic meridional heat transport strength beginning in the early 20th century—slightly lagging the onset of regional warming observed in temperature reconstructions. Notably, our results also share key temporal features (r = 0.6+) with previous AMOC reconstructions derived from individual proxies and ‘fingerprints’. Overall, this study introduces a novel method of reconstructing historical AMOC variability from globally distributed isotopic proxies by focusing on a single sub-component of the circulation that is directly linked to observational data.