September 2025

Oceanography | Vol. 38, No. 3

12

FEATURE ARTICLE

IS THERE ROBUST EVIDENCE FOR

FRESHWATER-DRIVEN AMOC CHANGES?

A SYNTHESIS OF DATA, MODELS, AND MECHANISMS

By Sophia K.V. Hines, Nicholas P. Foukal, Kassandra M. Costa, Delia W. Oppo,

Olivier Marchal, Lloyd D. Keigwin, and Alan Condron

INTRODUCTION

The Atlantic Meridional Overturning Circulation (AMOC) plays

a crucial role in regional and global climate. It transports mass

and heat to the Northern Hemisphere (e.g., Frajka-Williams et al.,

2019; Trenberth et al., 2019), is characterized by sinking at sev-

eral locations in the northern North Atlantic (e.g., Talley, 2013),

and thus provides a pathway for sequestering anthropogenic car-

bon for centuries to millennia (e.g., Gebbie and Huybers, 2012;

Brown et al., 2021). Here, we define the AMOC as the upper cell

of the meridional overturning circulation in the Atlantic Ocean. It

moves warm, saline waters northward where these waters lose heat

to the atmosphere, sink, and flow southward as colder and fresher

North Atlantic Deep Water (NADW). Due to positive feedbacks

involving the advection of salt by the northward-flowing branch,

the AMOC may be bistable, as suggested by simplified box models

of meridional overturning circulation (e.g., Stommel, 1961).

Paleoclimate data are consistent with the AMOC having more

than one equilibrium state, and they suggest that the AMOC has

abruptly changed in the past, sometimes in just a few decades. For

example, there is broad evidence from paleoclimate records that

AMOC existed for thousands of years in a reduced state during the

transition out of the last ice age (e.g., McManus et al., 2004; Lynch-

Stieglitz et al., 2014; Rafter et al., 2022), which may have driven

changes in atmospheric circulation, precipitation patterns, and

global surface temperature (e.g., Wang et al., 2001; Anderson et al.,

2009; Cheng et al., 2009; Clark et al., 2012). Some authors have

interpreted these intervals as times of AMOC collapse (McManus

et  al., 2004), but paleo data cannot quantitatively reconstruct

the strength of the AMOC, so there is a reluctance within the

paleoceanographic community to use this term. Nevertheless, a

popular schematic in paleoclimate research represents the AMOC

in either an “on” state or an “off” state (Figure 1; Rahmstorf, 2002).

A vigorous, or “on,” state of the AMOC would correspond to the

meridional circulation in the modern Atlantic, which is on the

order of 15–20 Sv (1 Sv = 106 m3 s–1; Frajka-Williams et al., 2019).

A “collapsed,” or “off,” state of the AMOC could occur when surface

waters are not dense enough to sink deeply in the North Atlantic.

Importantly, the upper cell volume flux during a “collapse” can-

not be quantified by paleo data. In this paper, we do not define an

AMOC “collapse” as a complete cessation of circulation but rather

a large and persistent reduction in upper cell volume flux relative

to that of the “on” state.

Global climate models from the International Panel on Climate

Change (IPCC) Coupled Model Intercomparison Project 6

(CMIP6) predict that AMOC will “very likely” decline over the

twenty-first century due to anthropogenic forcing, but it is less

likely that the AMOC will collapse (though the term “collapse”

is not precisely defined in this context; Fox-Kemper et al., 2021).

Some reconstructions of North Atlantic sea surface temperature

and other oceanographic properties during the past ~100 years

were interpreted to mean that the AMOC has weakened during

this period (Thornalley et al., 2018; Caesar et al., 2021), but there is

still significant uncertainty, as other North Atlantic records show

conflicting signals (Kilbourne et al., 2022; Terhaar et al., 2025).

Time series of direct AMOC observations are not long enough

to confidently detect trends in the magnitude of the overturning

ABSTRACT. The Atlantic Meridional Overturning Circulation (AMOC) transports heat to high latitudes and carbon to the deep

ocean. Paleoceanographic observations have led to the widely held view that the strength of the AMOC was significantly reduced at two

intervals during the most recent glacial-to-interglacial transition, with global climate impacts. Climate models predict that the AMOC may

decline in the future due to anthropogenic forcing, but the time periods for modern observations are too short to detect recent trends with

high confidence. To understand the likelihood of future changes in the AMOC, it is important to understand the mechanisms that drove

past changes in AMOC strength. In this paper we review (1) the paleoceanographic proxy data that have led to the widespread view that

the AMOC sharply decreased for periods of several thousand years during the last deglaciation, (2) climate model simulations of the last

deglaciation, with particular attention to their use of fresh water to alter the AMOC, (3) the physical mechanisms that could have driven

past changes in the AMOC, and (4) how insights from past ocean change can inform our understanding of what may happen in the future.