One of the central features of plate tectonics is the formation of new crust at mid-ocean ridges. It was arguably the discovery of these ridges that drove widespread acceptance of plate tectonics as a theory. Thanks to decades of exploration, we now have a good picture of what the crust that forms at the site of spreading looks like. But we still have an incomplete idea of how its features are actually produced - like knowing the final score of a game but not how the plays unfolded.
That is starting to change. In 2024, a team of French scientists was able to remotely monitor a major event on the border between the Australian and Antarctic plates, only two months after they installed equipment on the ocean floor. Their data shows that most of the spreading occurred in a relatively short time window, and some key events happened without any obvious seismic activity - kind of like a tectonic ninja.
The site where the events took place is incredibly remote, about halfway between Australia and Madagascar, and far south of India. There’s a large seafloor feature called the Amsterdam - Saint Paul Plateau, interpreted as a rise driven by a deep ocean hotspot. The rift between the Antarctic and Australian plates runs right through the middle of this plateau.
Despite the indications of a tectonic hotspot, only two volcanic islands are present in the area, Amsterdam and St. Paul. The islands have a long history of failed colonization attempts, accidental strandings, and regular visits by fishermen and sealers. Initially claimed by France, they ended up so useless and remote that France dropped claim to them only a decade later. Forty years after that, the crew of a French ship reclaimed them on behalf of a country that didn’t seem certain whether it wanted the honor. Now, over a century later, the French government maintains research stations on the islands and sporadically sends ships to maintain equipment, deliver scientists, and perform supply duties.
The team behind the new work took advantage of one of these ships to deploy underwater monitoring stations along the spreading zone. These included hydrophones for locating seismic events and transmitters to track changes in distance between monitoring sites. Later visits from French supply ships performed three-dimensional mapping of the seafloor to determine the outcome of any events detected.
Earlier study of the region had shown that spreading occurs at an average rate of a bit over 60 millimeters a year, along a site with a roughly 2,000-meter depression flanked by rugged ridges.
All of the hardware was in place when the fault it was on rumbled to life in April 2024. The first cluster of events occurred progressively farther south along the main spreading area, with the last of them over 8 kilometers south of the first. That was followed by a series of events moving north, extending out over 9 kilometers. The researchers say this is typical of the formation of dykes - thin but long and tall structures formed by the intrusion of molten rock.
At the same time, sensors in the valley at the center of the spreading region started experiencing a drop. As dyke events continued, the drop accelerated until the sensors were sinking at about 5 centimeters a minute before slowing. But subsidence continued well after the initial events, with a total of 4.2 meters over a six-day period. The researchers interpret this as a magma reservoir beneath the ridge draining. Consistent with that, the temperature of water at nearby instruments started rising at the same time, suggesting magma was interacting with seawater.
While all of this was going on, instruments on opposite sides of the central valley started moving farther apart, in some cases by well over a meter. Sometime after the site had returned to background levels of activity, the next visit from a French research vessel occurred, and new imaging revealed sites over 90 meters higher than during previous mapping - well beyond potential instrument errors. One patch of material was over 4 kilometers long, and the researchers estimate the total amount of new material at about 150 million cubic meters.
The researchers performed modeling to figure out how all the events might be connected. They randomized different configurations of magma source, dyke extent, and fault geometries, sampling 10 million different ones to see if they could produce the changes the instruments picked up. Only 2,200 could, and they had common features: a collapse of a deep reservoir of molten material called a sill (a horizontally oriented version of a dyke), some magma going into a dyke to expand it, and faults spreading by anywhere from 2 to 4 meters.
The team estimates the total extension is the equivalent of 38 years of activity at the site’s average rate of spreading. They think this might be how mid-ocean spreading normally occurs: a buildup of strain and material followed by a series of rapid events that actually produce the new seafloor. The other striking thing is that some events occurred without obvious tectonic signals picked up by hydrophones. That suggests if we relied on seismic data alone, we might miss the full picture of how our planet renews its crust.
Nature, 2026. DOI: 10.1038/s41586-026-10785-0