Imagine you're an ant. You've run a virtual marathon from your nest in search of food and now you have to find your way home. Oh yes, and you're in a featureless desert. So, how do you do it? The answer is far from simple.

Desert ants are old favourites in studies of navigation because they can navigate with amazing precision over long distances in virtually landmark-free habitats. To a large extent the ants rely on a form of navigation called path integration (PI) to find their way around when there are no landmarks to refer to. Basically, the ant keeps a running tally of all the twists and turns it makes whilst searching for food and it quickly tots up a homeward vector when it's time to return to the nest. Ants also use remembered landmarks to guide them along familiar routes. Landmark navigation tends to be more precise than path integration, so ants often use it in preference to PI when available.

Matthew and Tom Collett, Stephane Chameron and Rüdiger Wehner wanted to know whether path integration and landmark-based information might interact(p. 877). Specifically they were interested in whether familiar landmarks, seen in unfamiliar positions,might reset the path integrator. This would allow the ant to recall previous PI information associated with a particular arrangement of landmarks and use it to recalibrate its path-finding system. As a result, the ant could correct any errors in its path rather like a human navigator looking at a map. Collett and his co-workers already knew that landmarks did not reset the integrator as the ant scurries home, but it might still be affected by landmarks seen as the ant searches for tasty titbits. Testing this required a bit of experimental`trickery'.

Studying the ants in their natural habitat meant a trip to Tunisia, where Rüdiger Wehner and his collaborators have studied these masters of navigation for several decades. First, they trained the ants by enticing them along an L-shaped route, sunk into channels, with pieces of ripe melon. The ants couldn't see any landmarks from the first leg of the L-shaped path, but they were helped to become familiar with the second part of the route by several plastic bucket landmarks placed around the channel. Once the ants were familiar with the training route, the team altered the length of the featureless leg of the path and allowed the ants to perform a test run to the feeder.

After the test run, the ants were captured when they arrived at the melon and, taken in darkness, to a bare sand test area where they were released to see which way they chose to run. If the ants were resetting their path integrator using the familiar landmarks surrounding their feeder, they would follow their usual direction to the nest. But if they calculated their homeward direction solely on the basis of the trip they had just made, using path integration alone, their heading would lead them home from the altered route.

The ant's bearing matched the one required to get home from their altered test route so their path integrator was unaffected, despite the scientists attempts to deceive them with the familiar outward landmarks. The researchers concluded that the ants are blessed with two largely separate navigational systems to guide them... and by using global PI the ant can always find its way home, no matter where its search for food takes it.

Collett, M., Collett, T. S., Chameron, S. and Wehner, R.(
). Do familiar landmarks reset the global path integration system of desert ants?
J. Exp. Biol.