When the Mediterranean Dried Up: The Salt Desert Under the Sea and the Flood That Refilled It

Stand on the Rock of Gibraltar and look south, and Africa is right there. The Strait of Gibraltar is only about 14 kilometers wide at its narrowest point, and every hour a river of Atlantic seawater pours through it into the Mediterranean. That inflow is not optional. The Mediterranean loses more water to evaporation than all of its rivers, rain and snow put back, so it survives on a constant top-up from the ocean next door.

About six million years ago, that top-up stopped. Plate movements in what is now southern Spain and Morocco closed the old seaways to the Atlantic, and the Mediterranean began to dry out. Over hundreds of thousands of years it shrank, refilled, shrank again and left behind more than a million cubic kilometers of salt. Rivers cut canyons as deep as the Grand Canyon to reach a sea floor that lay kilometers below the ocean outside. Then, about 5.33 million years ago, the Atlantic broke back in, very likely in the largest flood known to science.

Geologists call it the Messinian salinity crisis. The idea sounded absurd when it was first proposed, and it still has open questions. But the salt is real, the canyons are real, and recent work has added new evidence for both the drop and the flood. This is the story of how a drill ship found a desert under the sea, and of the architect who, decades earlier and without knowing any of it, wanted to drain the Mediterranean on purpose.

A strange layer under the sea floor

The first clue came from sound. In 1961, seismic surveys of the Mediterranean picked up a strong reflecting layer some 100 to 200 meters beneath the sea floor. It followed the shape of the floor above it closely, as if it had been laid down evenly across the whole basin. Geologists called it the M reflector, and many suspected it was salt.

Salt was not a crazy guess. The Messinian, the last age of the Miocene epoch, had been named back in 1867 by the Swiss stratigrapher Karl Mayer-Eymar after the city of Messina in Sicily, where evaporite rocks of that age are exposed. Sicily, southern Spain and other places around the Mediterranean are full of thick gypsum and salt beds from the late Miocene. Earlier suggestions by Georges Denizot in 1952 and Giuliano Ruggieri in 1967 had already placed the strange layer in that period, and it was Ruggieri who coined the phrase “Messinian salinity crisis.” The question was what the layer meant. Salt can form in a shallow lagoon on the edge of a sea. Finding it under the middle of the deep Mediterranean was a different matter.

What the Glomar Challenger pulled up

To find out, someone had to drill. In August 1970 the drill ship Glomar Challenger sailed from Lisbon on Leg 13 of the Deep Sea Drilling Project, the program that grew out of the failed attempt to reach Earth’s mantle described in our story of Project Mohole. Its co-chief scientists were the American marine geologist William B. F. Ryan and the Chinese-born geologist Kenneth J. Hsü, then a professor at ETH Zurich. The cruise ran until October and returned to Lisbon.

The cores that came up from the deep basins were not what anyone expected to find under two or three kilometers of water. There was gypsum and anhydrite. There was rock salt, and in a few places potash, the bitter salts that crystallize last when brine dries out completely. There were river gravels and red and green floodplain silts. There were stromatolites, layered mats built by microbes in very shallow, sunlit water. One core held something even stranger: deep-sea ooze made of the shells of tiny plankton that had apparently dried into dust and been blown into cross-bedded layers by wind, mixed with quartz sand from the surrounding continents.

Hsü, Ryan and Maria Bianca Cita laid out the conclusion in Nature in March 1973. The evidence, they wrote, showed “that the Mediterranean Sea was a desiccated deep basin some 6 million years ago.” In other words, the salt had not formed in a lagoon on the edge of a sea. It had formed at the bottom of an enormous pit, after the sea itself had largely evaporated away. Hsü later told the story for general readers in his book The Mediterranean Was a Desert. Scientific drilling has a habit of upending expectations, as the Soviet crews of the Kola Superdeep Borehole would later find on land.

Rivers that cut a mile below sea level

The strongest case for a dramatic drop in sea level does not actually come from the salt. A deep salt layer could, in principle, form from very heavy brine at the bottom of a sea that was still full. What a full sea cannot explain is the canyons.

In 1967, the geologist Ivan Chumakov reported marine fossils of Pliocene age, tiny plankton shells known as foraminifera, from deep below the Nile valley at Aswan in southern Egypt. They lay in a buried gorge that the Nile had cut down to about 200 meters below sea level, far up the river. Farther north the buried canyon goes much deeper. Just north of Cairo it reaches about 2,500 meters below sea level. A river can only cut its bed that deep if the water it flows into is that low. For the Nile to carve a gorge to 2,500 meters below the modern sea, the Mediterranean it drained into had to be lower still.

The Nile was not alone. The Rhône in France also cut a deep gorge, about 1,000 meters deep near its mouth, that was later filled with sediment. Around the Mediterranean, the steepest, driest phase of the crisis, roughly 5.6 to 5.5 million years ago, produced several huge canyon systems, some comparable to the Grand Canyon. When the sea came back, it drowned them. The Nile’s canyon became an arm of the sea reaching about 900 kilometers inland, as far as Aswan, which is exactly where Chumakov’s marine fossils came from.

How you empty a sea

It sounds impossible to dry out a sea the size of the Mediterranean, but the arithmetic is not complicated. Today the basin holds about 3.7 million cubic kilometers of water. If the Atlantic inflow were cut off, evaporation would take away around 3,300 cubic kilometers a year more than rivers and rain could replace. At that rate, the sea would mostly vanish in a little more than a thousand years, leaving a layer of salt tens of meters thick and raising sea level around the rest of the world by roughly 12 meters, because all that water would go somewhere.

One evaporation would not make a million cubic kilometers of salt, though. The amount of Messinian salt is about 50 times the salt dissolved in the Mediterranean today. To pile up that much, the basin had to receive seawater again and again, either through repeated cycles of drying and refilling or through a long period in which a trickle of Atlantic water flowed in and evaporated. The layers in the rocks show both kinds of behavior. Fine, even sediments laid down in deep or quiet water give way upward to wave-rippled layers, then stromatolites, then anhydrite formed in salty ground water under exposed flats. Then the sea returns and the cycle starts over.

The current timeline puts the start of the crisis at about 5.96 million years ago, when the gateways between the Atlantic and Mediterranean were being squeezed shut. Those gateways were not at Gibraltar. They ran through corridors in what are now southern Spain and northern Morocco, in a zone where the African and Eurasian plates grind together. As Africa pushed north, faulting, uplift and rotating blocks of crust opened and closed passages, and climate swings made the effect stronger or weaker. The basin was finally isolated for a longer stretch from about 5.59 to 5.33 million years ago.

How low did it go?

Here the agreement ends. Some geologists favor a “deep desiccation,” in which the basins dropped by kilometers and the deepest parts became salt pans and hypersaline lakes. Others argue that much of the salt formed under a deep layer of dense brine, with only the shallower margins drying out. A third camp suggests sea level fell everywhere at once but only the shallower basins dried enough to lay down salt. The debate often comes down to places like the Sorbas basin in southern Spain, where Messinian gypsum is exposed on land, and how precisely those beds can be matched in time with the salt buried far offshore.

A study published in Nature Communications in November 2024 took a new route into the problem. A team led by Giovanni Aloisi of France’s CNRS measured chlorine isotopes in halite from beneath the eastern Mediterranean. Because heavier chlorine is pulled into salt crystals slightly faster, the ratio shifts more when salt forms quickly, and very fast salt formation is what you would expect from a sea that is shrinking.

The isotopes point to two phases. In the first, which lasted about 35,000 years, salt formed only in the eastern Mediterranean, in a basin that was still full of brine. In the second, salt formed across the whole sea during a rapid drawdown that took less than 10,000 years. Sea level fell by about 1.7 to 2.1 kilometers in the east and about 0.85 kilometers in the west. The eastern basin lost up to 83 percent of its water, and the CNRS summed up the result as a loss of up to 70 percent of the Mediterranean’s water. The deepest pockets, the Ionian and Herodotus basins, still held more than a kilometer of brine. That is not a completely dry floor, but it is a sea reduced to a set of salty lakes at the bottom of a gigantic hole.

The hottest place on Earth

A floor kilometers below sea level would have been brutally hot. Air warms as it sinks and is compressed, which is one reason Death Valley, which dips below sea level, holds the world heat record. The Messinian basin had that effect on a vastly larger scale. The salt itself carries a clue: anhydrite, one of the minerals in the deep cores, forms from water warmer than about 35 °C.

Simple calculations that let air warm all the way down a dry basin four kilometers deep produce silly results, with summer highs near 80 °C. Climate models are more modest. Experiments published in 2009 by Murphy and colleagues found that a completely dry basin would warm by up to 15 °C in summer, while a basin with a lowered water surface would warm by only about 4 °C in summer. Either way, the floor of the Messinian Mediterranean would have been one of the most hostile landscapes on the planet, a place of salt flats, bitter lakes and wind-blown dust.

Goats that walked to Mallorca

For sea life, the crisis was a catastrophe. Much of the Mediterranean’s native marine fauna died out. The sea’s present pattern, where the number of species falls as you travel east, took shape after the refill, as life came back in from the Atlantic. Whales, seals and other marine animals had to colonize the basin all over again.

For land animals, the dry basin opened new routes. Africa and the Iberian Peninsula were joined, and species crossed between them. Some animals reached places that would soon become islands again. The most famous is Myotragus, the “mouse-goat” of Mallorca and Menorca. Its ancestors are thought to have walked onto the Balearics from mainland Europe during the crisis. When the sea came back, they were cut off.

They stayed cut off for more than five million years. With no predators and few other mammals for company, Myotragus evolved into a very odd animal: smaller over time, with short, stiff legs, forward-facing eyes and an ever-growing lower incisor. The paleontologist Dorothea Bate described it in 1909. The last species, Myotragus balearicus, died out roughly 4,300 years ago, after people had reached the islands. Its whole history began with a sea that had dried up.

The flood that refilled the Mediterranean

About 5.33 million years ago, at the boundary between the Miocene and the Pliocene, the Atlantic came back. The ancients had a story for the strait, which Pliny the Elder repeats: Hercules dug it, letting the ocean into the Mediterranean. The modern version has no demigod, but it does have a breach. Tectonic sinking of the Gibraltar region may have lowered the rocky barrier until the Atlantic spilled over it into the empty basin below, though the exact trigger is still not known.

In 2009 a team led by Daniel Garcia-Castellanos of the Spanish research council CSIC published in Nature an analysis of a channel about 200 kilometers long that runs through the Strait of Gibraltar, cut more than 250 meters deep into the rock on both sides. Earlier researchers had put it down to river erosion during the dry period. Garcia-Castellanos and his colleagues argued it was carved by the incoming sea itself. Once water started to pour through, it cut the channel deeper, which let more water through, which cut faster still.

Their numbers are hard to picture. The early trickle may have lasted up to several thousand years. But at its peak, the flood ran at about 100 million cubic meters per second, hundreds of times the flow of the Amazon River, and the channel deepened by more than 40 centimeters a day. According to the model, 90 percent of the water arrived in a period between a few months and two years. At times the level of the Mediterranean may have risen by more than 10 meters a day.

The popular picture of this event is a waterfall, the Atlantic tumbling off a cliff kilometers high. The studies of the channel suggest something less cinematic and more powerful. The water seems to have roared down a long, relatively gentle ramp toward the basin floor, more like a colossal rapids than a single falls, with flow speeds that one estimate puts at more than 40 meters per second. The channel the flood left behind continues across the Alboran Sea for hundreds of kilometers, with giant underwater bars of debris along the way. As the Mediterranean filled, the world’s other oceans dropped by about 10 meters to make up the difference.

New evidence in Sicily

Gibraltar was not the only place where water had to pour over a rim. The Mediterranean is really two basins, west and east, separated by a shallow sill between Sicily and North Africa. If the flood came in from the Atlantic, the western basin would have filled first and then overflowed eastward across that sill into the much deeper Ionian Sea. On the eastern edge lies the Malta Escarpment, an underwater cliff about three kilometers high.

In December 2024, a team led by Aaron Micallef published in Communications Earth & Environment what they called “a clear line of onshore-to-offshore evidence” for that overflow. On land in southeastern Sicily, once a shallow sea corridor, they mapped more than 300 streamlined ridges, shaped and aligned like the hills sculpted by huge floods elsewhere on Earth. Between the ridges and the sea lies a buried channel 20 kilometers wide, leading to the Noto Canyon, a theater-shaped notch cut into the Malta Escarpment. On top of the ridges sits a jumble of poorly sorted rock fragments wedged between the Messinian layers and the first Pliocene deposits. Offshore, east of the canyon, a chaotic sediment deposit of about 1,500 cubic kilometers lies buried on the floor of the Ionian Sea.

Their estimates for the overall flood, drawn from earlier studies, run from 68 to 100 million cubic meters per second, lasting between 2 and 16 years. The authors concluded that the crisis “was terminated through a cataclysmic flood.” Not everyone agrees. Some geologists argue the refill was gradual, or that the strait was partly open all along, and some evidence from the southern margin of the Alboran Sea fits a gentler reconnection better. The flood is still a hypothesis, though it now has physical evidence on land.

White cliffs from the new sea

You can see the moment the sea came back without a drill ship. On the south coast of Sicily near Realmonte, the Scala dei Turchi, the “Stair of the Turks,” rises in bright white steps above the water. The rock is marl of the Trubi Formation, built from the shells of countless planktonic foraminifera. It was laid down on the floor of the refilled sea in the early Pliocene, right after the Mediterranean came back. Across the region, the switch from the salts and brackish “Lago Mare” lake beds of the late crisis to deep open-ocean sediments above them is sharp, the clearest mark the flood left in the rocks.

The architect who wanted to do it again

Decades before the Glomar Challenger sailed, a German architect had already worked out that the Mediterranean would sink if you blocked Gibraltar. Herman Sörgel of Munich came across the idea in 1927 in a book by the geographer Otto Jessen, who described the Mediterranean as an “evaporation sea” that would lose water if the Atlantic inflow were cut off. Sörgel decided this was not a danger but an opportunity.

His plan, first called Panropa and then Atlantropa, centered on a gigantic dam across the Strait of Gibraltar. With the inflow cut, evaporation would slowly lower the sea, and the water allowed back in through turbines would generate enormous amounts of hydroelectric power. More dams would close the Dardanelles and, later, the gap between Sicily and Tunisia. By one account of the final plans, the western Mediterranean would drop by 100 meters and the eastern by 200, exposing about 660,000 square kilometers of new land, an area larger than France. Further dams on the Congo River were meant to create inland seas in Africa.

Sörgel promoted the scheme from 1928 until his death in 1952. He recruited well-known architects, including Peter Behrens, who sketched a tower about 400 meters tall rising from the Gibraltar dam. He spread the idea through newspapers, radio, films, exhibitions, poetry and even an Atlantropa symphony. He believed a shared continental power grid could keep the peace, with an independent authority able to switch off the electricity of any country that threatened war.

The vision also had an ugly side. Sörgel saw Africa as land and resources at Europe’s disposal, a view that fit the colonial and racist attitudes of his era, and his 1938 book carried a quotation from Hitler on its flyleaf to show the plan was compatible with Nazi ideology. He gave little thought to what Mediterranean nations or the people of Africa might want, or to the coastal cities that would be stranded inland as the sea retreated. Interest revived briefly after the Second World War, but nuclear power, the cost of rebuilding and the end of colonial empires made the project pointless. Sörgel died on December 25, 1952, after a car hit him while he was cycling to a lecture in Munich. The Atlantropa Institute lasted until 1960.

The irony is hard to miss. Sörgel imagined lowering the sea by a couple of hundred meters as the ultimate act of engineering. Nature had already lowered it by kilometers, and nobody knew it until the cores came up in 1970.

Will it happen again?

Today the Mediterranean is still saltier than the Atlantic beside it. Dense, salty Mediterranean water slides out along the bottom of the Strait of Gibraltar, while lighter Atlantic water flows in on top. The whole sea depends on that exchange through a gap a few hundred meters deep at its shallowest sill. If the strait were ever cut off again, the arithmetic is the same as six million years ago: most of the Mediterranean could be gone in about a thousand years.

The plate boundary that closed the old gateways has not gone quiet. In February 2024, João Duarte and colleagues published a computer model in Geology suggesting that the subduction zone under the Gibraltar Arc is still active, just moving very slowly. In their simulations, the slow phase lasts about another 20 million years before the subduction zone pushes out into the Atlantic, building a new subduction system there roughly 50 million years from now. The researchers also note that during the quiet phase, large earthquakes like the one that devastated Lisbon in 1755 remain possible.

None of this is a forecast anyone alive needs to plan around. But the Messinian salinity crisis is a reminder that seas are not permanent features. The Mediterranean, the sea of Homer and the Phoenicians, Rome and Venice, is a basin that once nearly emptied, sat for hundreds of thousands of years as a salt desert, and then was refilled in a matter of years by an ocean pouring through a gap at the edge of Europe. Like the Laki eruption or the methane-laden depths of Lake Kivu, it is one of those places where the planet shows how fast it can change once a threshold is crossed. Under the blue water of every Mediterranean holiday photo, more than a million cubic kilometers of that salt is still lying in the dark.

Further reading