German researchers on Wednesday said they had evidence that sowing the ocean with iron particles sucks up and stores carbon dioxide (CO2), preventing the gas from stoking dangerous climate change.
But their work, touching on a fiercely controversial issue called geo-engineering, came under attack from other scientists and environmentalists.
These said a far bigger question -- whether such schemes could damage the marine biosphere -- remained unanswered.
Published in the science journal Nature, the paper is one of the biggest and most detailed probes into ocean fertilisation, a practice that is banned under international law although scientific research into it is permitted.
Its goal is to take CO2 from the atmosphere and store it in the deep sea so that it no longer adds to the greenhouse effect.
This would be done by scattering the ocean surface with iron dust, a nutrient for microscope marine vegetation called phytoplankton. As the plants gorge on the iron, they also suck up atmospheric CO2 thanks to natural photosynthesis.
In the next step, the phytoplankton die and sink to the deep ocean floor -- taking with them the CO2, which would lie in the sediment, possibly for centuries.
Critics, though, say geo-engineering schemes are riddled with unknowns, both in cost effectiveness and risks for the environment.
Scientists led by Victor Smetacek of the Alfred Wegener Institute for Polar and Marine Research in Bremerhaven took a research ship to the Southern Ocean off Antarctica in 2004.
There, they located a giant eddy -- a slowly-moving clockwise-rotating swirl 60 kilometres (37 miles across) that had relatively little interchange with the rest of the ocean -- and used it as a testbed for a five-week experiment.
They scattered seven tonnes of commercial iron sulphate particles, which within four weeks developed into a giant bloom of diatom plankton.
The diatoms then died, sinking in clumps of entangled cells, "far below" a depth of 1,000 metres (3,250 feet), according to samples measured with a gadget called a fluorometer.
They were probably deposited on the sea floor in a "fluff layer" that should remain for "many centuries and longer," Smetacek's team say.
Further work is needed to see what happens when sideways currents hit the diatom blooms, they add.
Other voices sounded a loud note of caution, saying the experiment took place in exceptional conditions and did not consider other environmental consequences.
Among them was Professor John Shepherd, who chaired a landmark report in 2009 by Britain's Royal Society into geo-engineering.
It concluded that ocean fertilisation would not suck up that much CO2 and could be harmful to the marine biosphere.
"Whilst the new research is an interesting and valuable contribution in this evolving field, it does not address the potential ecological side effects of such a technology in what is a poorly understood field," Shepherd said in an email to AFP.
The Canada-based ETC Group, an environmental NGO campaigning against geo-engineering, said the study "only focuses on a few narrow aspects and disregards or ignores others."
"The intended purpose of ocean fertilisation is to significantly disrupt marine ecosystems through drastic changes on phytoplankton, which is the base of the marine food web, so the effects would propagate throughout the ocean in unpredictable ways," it said.