Charles Darwin was one of the first biologists to wonder how migrating green turtles find Ascension Island, a 10-kilometre-long speck of land in the vast mid-Atlantic Ocean (Nature 7, 360; 1873). These turtles arrive at Ascension Island between December and March, having travelled a daunting 2,200 kilometres or so eastwards from their feeding grounds off Brazil. Graeme C. Hays and colleagues now suggest a role for wind-borne information in this remarkable island-finding ability (Proc. R. Soc. Lond. B doi:10.1098/rspb.2002.2308; 2003).

The navigational feats of marine turtles make them good subjects for researchers interested in migratory behaviour. Early theories proposed that green turtles (Chelonia mydas; pictured) pinpoint Ascension Island by using water-borne odours, carried towards Brazil by the South Atlantic Equatorial Current. More recent laboratory studies have shown that hatchling loggerhead turtles can detect the inclination and intensity of the Earth's magnetic field, perhaps using these features as 'magnetic markers'. However, these promising ideas have yet to be tested on sea turtles in their natural environment.

Hays et al. have now performed just such a test of another theory: that green turtles use wind-borne cues to find their way. The authors found six female turtles that had just nested on Ascension Island, and then 'displaced' them. During the nesting period, persistent trade-winds blow from the southeast. So three of the turtles were transported 50 kilometres upwind of the island, and the other three 50 kilometres downwind. The animals' movements were then tracked by satellite. Female C. mydas come ashore to lay eggs several times in a season, so they had a strong urge to return to lay their remaining eggs.

Incredibly, Hays et al. found that the three downwind-displaced turtles returned to the island within 1, 2 and 4 days. Two of those moved upwind eventually returned after 10 and 27 days — but only after passing downwind of the island. A third upwind-displaced turtle was unable to locate Ascension Island after 59 days, and was heading back to Brazil when satellite transmissions ceased.

It seems, then, that wind-borne cues emanating from the island — perhaps odours or sounds — might direct turtles. The authors concede that such cues will not work accurately over very long distances, so might not on their own account for island finding. But perhaps a combination of mechanisms is at work, with a geomagnetic sense aiding navigation over large scales (hundreds to thousands of kilometres), and smell or hearing operating on scales of tens of kilometres. Other long-distance movers might also use such a scale-dependent battery of sensory 'channels' for migration. Identifying the role of particular channels, and when they kick in, will be a challenge, but one that will surely benefit from comparisons of turtles with other vertebrates.