Pacific icebergs may be quietly sapping the Atlantic's great conveyor

A pile of meltwater in the northeastern Pacific, dumped by icebergs calved off Alaska and the Bering Sea, can weaken the Atlantic Ocean's overturning circulation thousands of kilometres away. That is the conclusion of a University of California, Davis modelling study published in Nature on 13 July 2026, and it complicates the usual map of where the so-called Atlantic conveyor is most at risk. The system in question, the Atlantic Meridional Overturning Circulation, moves warm tropical water northward at the surface and sends cold, dense water back south at depth. It is the engine of Europe's mild winters, the lever for Sahel rainfall, and one of the most consequential pieces of climate plumbing on the planet. The new paper argues that the public conversation has been looking in the wrong ocean.

The standard framing treats the AMOC as a victim of local Atlantic inputs: Greenland melt, Arctic sea-ice retreat, and changes in North Atlantic precipitation. Pacific ice have entered the story as scenery, not as a player. The UC Davis team, led by oceanographer Till Wagner, asks a different question. If freshwater pulses are released far upstream, on the other side of the Americas, can the ocean transmit that signal through the tropical Pacific, around the tip of South America, and into the Atlantic on a timescale that matters? Their answer, in short: yes, sometimes, and the trigger does not have to be a one-time dump.

What the model actually does

The paper runs an ocean general circulation model under controlled freshwater hosing experiments. Rather than pouring meltwater into the North Atlantic, as most AMOC-tipping studies do, the authors release pulses in the northeastern Pacific, off the coast of Alaska and along the Aleutian arc. The volume is calibrated against observed iceberg discharges from Alaskan tidewater glaciers and from the Bering Sea margin, not against a worst-case Greenland collapse.

The mechanism runs through the tropical Pacific. Freshwater at high northern latitudes lightens the surface layer of the subarctic Pacific. The model responds by reorganising tropical Pacific wind patterns, which in turn change the salt and heat flux across the equator into the Atlantic via the Panama gateway. A weaker, fresher tropical Atlantic means a saltier, denser North Atlantic is harder to maintain, and the overturning cell that depends on that density contrast loses some of its drive. The effect in the runs is partial, not catastrophic: the AMOC weakens rather than collapses. But the signal is consistent across multiple ensemble members and is robust to plausible parameter choices, which is what gives the result its weight.

This matters because the Pacific ice reservoir is not a hypothetical. Alaska's glacier discharge has roughly doubled since the 1980s, and the rate of berg calving along the Bering coast has been climbing in tandem with regional ocean warming. The freshwater load the model uses is not a future worst case. It is, in the authors' framing, a present-day scale.

Why this is awkward for the existing debate

The AMOC question has been dominated for two decades by Greenland. The discourse runs roughly as follows: Greenland's ice sheet loses mass at an accelerating rate, the North Atlantic gets fresher, the overturning weakens, the conveyor slows, and at some point a tipping threshold is crossed. That framing has produced a research programme, a sub-literature of paleoclimate analogues (the Younger Dryas, Heinrich events), and a stream of high-profile model papers.

The Pacific-berg result does not refute any of that. It does push back on the assumption that the relevant freshwater is local. If a meltwater pulse released near Alaska can, after a transoceanic detour, weaken the AMOC by a measurable fraction, then the North Atlantic's effective freshwater budget has inputs that have not been fully accounted for. The implication is that the AMOC may be closer to a vulnerable state than North Atlantic-only mass-balance calculations suggest. The sources do not specify how close, and the modelling does not yet include interactive ice-sheet coupling. The honest reading is that the buffer is thinner than the Greenland-centric picture implies.

There is also a counterpoint worth naming. Some oceanographers argue that the tropical Pacific gateway is too narrow and too dominated by wind-driven processes for a subarctic freshwater anomaly to survive the transit intact. In that reading, the model's response is an artefact of coarse resolution or of an ensemble member that overstates tropical sensitivity. The Pacific-berg camp's reply, implicit in the paper's design, is that the response shows up across parameter perturbations rather than in a single tuned run. The disagreement is not yet settled, and the sources do not adjudicate it.

What it would take to break the system

The paper's framing is careful: no individual iceberg discharge flips the AMOC. The risk is cumulative and combinatorial. Pacific meltwater pulses of the size modelled, arriving alongside the more familiar North Atlantic forcing, would together push the system toward a state in which recovery from a single bad year becomes harder. The Atlantic's overturning is not a switch; it is a regulator with hysteresis, and the work of this paper is to add a previously uncounted input to the regulator's budget.

Practically, that means three things. First, Alaskan tidewater glacier monitoring deserves more weight in AMOC risk assessments than it has historically received. Second, the tropical Pacific, often treated as a separate climate system for the purposes of Atlantic forecasting, has to be modelled as coupled. Third, the long lead time implied by the transoceanic pathway is a planning problem in itself: by the time a Pacific-berg pulse shows up as an Atlantic signal, the originating ice is already gone.

What to watch next

The next round of evidence will come from observation rather than models. The Bering Sea and Gulf of Alaska freshwater flux is now measurable to within a few percent of the climatology, and the AMOC itself is monitored at 26.5°N by the RAPID array. The test is whether the modelled relationship shows up in the joint record over the next five to ten years. If it does, the Greenland-only framing of AMOC risk will need to be quietly retired, and the climate-risk literature will have to treat the Pacific as a first-order Atlantic forcing rather than a parallel story.

The paper does not settle the debate over whether the AMOC is approaching a tipping point. It does narrow the conditions under which that debate can be conducted in good faith. The freshwater pulse that breaks the Atlantic does not have to start in the Atlantic. That alone reshapes the map.

This piece treats the AMOC as a coupled ocean system and gives roughly equal weight to the North Atlantic and Pacific forcing pathways, a framing the wire coverage has so far only partially adopted.

Wire provenance

This editorial synthesis draws on the following public wire/social posts:

• https://en.wikipedia.org/wiki/Atlantic_meridional_overturning_circulation
• https://en.wikipedia.org/wiki/Panama_Basin