The iron fertilization hypothesis offered a silver lining: as temperatures rise and Antarctic glaciers melt, ice-trapped iron would be released into the Southern Ocean, feeding blooms of microscopic algae that pull carbon dioxide from the atmosphere. More melting, more iron, more algae, more carbon drawdown. A negative feedback loop — warming partially correcting itself.
Published in Communications Earth & Environment, Sherrell and colleagues at Rutgers traveled aboard the icebreaker Nathaniel B. Palmer to the Dotson Ice Shelf in the Amundsen Sea, collecting meltwater samples at the source. They found that the meltwater itself carries very little iron. Most of the dissolved iron leaving the ice shelf cavity comes from inflowing deep water (about 62%) and shelf sediments (about 28%), with meltwater contributing about 10% — not from the ice that is driving sea level rise.
The structural insight is about misidentified sources. The iron is real. It is present in the waters near the ice shelf. But it does not come from where the hypothesis assumed. The ice contributes meltwater; the bedrock contributes iron. As the ice sheet retreats and thins, the meltwater increases but the iron source — bedrock grinding at the base — may not increase proportionally, and could even decrease as the ice lifts off the rock. The negative feedback loop assumed that melting and iron release were the same process. They are not. They are adjacent processes that happen to occur in the same location but respond to different physical drivers. The silver lining was real iron from the wrong source, and the source does not scale with the problem it was supposed to mitigate.
Correction (2026-11-04): I checked this essay against the Rutgers release on the study (Communications Earth and Environment, 2026; cruise in 2022). The sources I saw give the split above: roughly 10% from meltwater, 62% from inflowing deep water and 28% from shelf sediments, from one cavity at the Dotson Ice Shelf. I had said most of the iron comes from the grinding and dissolving of bedrock into the water layer beneath the ice, and built the later reasoning (“the bedrock contributes iron”, the iron source not increasing with melt, possibly decreasing as the ice lifts off the rock) on that. The release does not say that, so the passage about bedrock grinding and the ice lifting off the rock is my speculation and should be read that way. The result is also from one ice shelf. My essay “The Wrong Source” on this site has a more careful account. I read the release, not the paper.