The deep 'overturning' circulation in the North Atlantic, in which northward-flowing surface water sinks at high latitudes and flows back at depth, may not be fully understood, but its existence has long been public knowledge. But only recently has there been general awareness that two shallow circulation cells lie on either side of the Equator in the Pacific. As they describe elsewhere in this issue (Nature 415, 603–608; 2002), M. J. McPhaden and D. Zhang have now produced an analysis of historical data to show that both cells have been slowing since the 1970s.

The circulation loops occur because winds produce surface currents away from the Equator. Water masses then flow along surfaces of constant water density (the figure shows the 25 kg m−3 surface as an example) from subtropical regions towards the Equator, at depths of 100–400 m. Near the Equator, upwelling water replaces that lost at the surface to wind action.

McPhaden and Zhang have looked at data for the Pacific between 20° S and 50° N for the years 1950 to 1999. The circulation loops (depicted here as asymmetric, though they may not be) carried much the same volumes of water in 1956–65 and 1970–77. But in the period 1990–99, the volumes were 25% less than at the earlier times.

This slowdown could help account for other changes that occurred in the past couple of decades. For instance, sea surface temperatures in the tropical Pacific Ocean have risen by about 0.8 °C since the 1970s, despite increasing cloudiness. And the amount of carbon dioxide released into the atmosphere from that part of the ocean has declined. Both observations can be explained by a reduction in the delivery of relatively cool, CO2-rich subtropical water to the equatorial ocean.

Perhaps the most intriguing question is whether there is any connection here with the El Niño/Southern Oscillation (ENSO), which occurs on shorter timescales. A shift towards stronger and more frequent El Niño events occurred in 1976–77, just when the decadal decline in the circulation began. But it is not clear whether the longer-term variation in circulation is a consequence of the increasing number of El Niño years, or whether the circulation slowdown provided the background for the change in ENSO cycles.

In the North Atlantic, changes in atmospheric conditions (the so-called North Atlantic Oscillation index) have brought milder and stormier winters to Northern Europe. As with the events documented by McPhaden and Zhang, this is a change that occurs over a period of several decades. There may in both cases be a connection with human-induced climate change, but for the moment that possibility remains speculative.