Reconstructing the route, not gauging the flow
Late-Pliocene Agulhas Leakage was not measured by a flow gauge. It was reconstructed through a proxy chain left on the seabed. A Utrecht University-led team used dinoflagellate cysts and organic lipid biomarkers in cores from Site U1475 on the Agulhas Plateau to locate the Southern Ocean subtropical front between 3.6 million and 2.58 million years ago. The front determines the route by which Indian Ocean water can pass around Africa into the Atlantic, so its position becomes an inference about possible leakage. From roughly 3.6 million to 3.3 million years ago, a northward migration of the front marks a gradual reduction in leakage volume and salt supply.[1]
The conventional account would expect less Indian Ocean salt to weaken the Atlantic Meridional Overturning Circulation. The comparison data moved the other way. Surface and bottom measurements along the AMOC pathway, Gulf of Mexico temperatures, published Atlantic and Caribbean data, and climate simulations indicate that overturning strengthened and North Atlantic Deep Water formation intensified during the same interval. That does not prove Agulhas salt has no role. A salt anomaly takes time to travel north, and a lag between the data series could conceal a weaker but real link. Still, opposite movements over the same interval make it harder to treat the leakage as the first-order control on circulation at this timescale.[1]
The boundary of the counterexample
The counterexample can easily say too much when carried into today's Atlantic. Although continental positions were broadly familiar, late-Pliocene ice volume, Arctic freshwater supply, temperature patterns and background climate differed from present conditions. The glacial event examined here accompanied cooling of roughly 3 degrees Celsius. The researchers state the boundary directly: the result belongs to a specific ancient geography and climate, and its applicability to current or future warming is uncertain. The study therefore does not say that today's AMOC is safe. It says something narrower and more useful: any model that assigns circulation strength to one salt gateway must be tested against the other freshwater and density flows in the system.[1]
The honest bridge to the present is a model ranking, not a headline. AMOC sensitivity comparisons published over the next two years can test the Agulhas salt term beside other controls under matched conditions. If that term falls from first-order to second-order across models, the constraint from the sediment reconstruction will have entered the machinery of prediction. If it remains dominant, the late-Pliocene disconnection will need an explanation in lag, proxy choice or another freshwater source peculiar to that period. Applying the same proxy chain to a glacial-interglacial interval closer to the present would also show whether this was a one-period counterexample or a recurring property of the system.[1]