Sensory Worlds · Chapter 11

The Pedometer in the Sand

Rüdiger Wehner, Saharan Ants, and Dead-Reckoning Path Integration

Across the flat Chott el-Djerid salt pans of southern Tunisia at solar noon, ground surface temperatures reach 60 degrees Celsius (140 degrees Fahrenheit). Every spider, lizard, and beetle retreats underground to escape thermal coagulation of its proteins. At that hour, when the desert surface is empty of predators, the Saharan silver ant (Cataglyphis fortis) erupts from its pinhole burrow entrance and sprints across the salt crust at speeds approaching one meter per second—fifty body lengths per second—lifting its body high on elongated legs to keep its vital organs in the slightly cooler air layer millimeters above the scorching salt.

The silver ant has less than ten minutes to locate the heat-stricken corpse of an insect before its own body temperature crosses its critical thermal maximum of 53.6 degrees Celsius. In the search for food, a foraging ant zigzags outward along a winding search path stretching 100 to 200 meters across featureless white salt flats where 60-degree surface heat evaporates chemical pheromone trails in seconds and no stones or plants exist as visual landmarks. Yet the instant the ant seizes a desiccated arthropod fragment in its mandibles, it pivots on the spot and runs in a dead-straight line directly back toward its invisible two-millimeter nest hole.

Starting in 1968 at a field station in Maharès, Tunisia, Swiss neuroethologist Rüdiger Wehner spent four decades turning the North African salt pan into an outdoor psychophysics laboratory. By painting grid lines across the desert floor and tracking marked ants with wheeled recording trolleys, Wehner proved that Cataglyphis navigates by continuous path integration (dead reckoning). At every step $k$ along its outbound trajectory, the ant couples a directional compass angle $\theta_{k}$—derived from its dorsal-rim ultraviolet polarization analyzer reading the Rayleigh scattering pattern of the sky—with a distance step $s_{k}$, continuously updating a two-dimensional homebound vector $\mathbf{H}$ in the central complex of its sub-milligram brain:

$$\mathbf{H} = -\sum\nolimits_{k=1}^{N} s_{k} \begin{pmatrix} \cos\theta_{k} \\ \sin\theta_{k} \end{pmatrix}$$

When Wehner captured a homebound ant just as it picked up food at a feeder 50 meters north of its nest and displaced it in a dark vial to an unfamiliar test flat two kilometers away, the released ant ran due south for 50 meters—and at 49.5 meters broke off its straight run into tight systematic search loops around the nonexistent nest entrance.

How does a walking insect measure the scalar distance $s_{k}$ of its journey? For decades, three competing hypotheses divided sensory physiologists. First, the ant might measure metabolic energy expenditure, which Harald Wolf and Wehner disproved in 1992 by loading ants with lead weights four times their body mass without altering their return distance. Second, like Karl von Frisch's honeybees, the ant might integrate the speed of visual patterns scrolling across the ventral retina of its compound eyes (optic flow). In 1995, Bernhard Ronacher and Wehner tested this by training ants to walk inside Perspex channels whose floors scrolled forward or backward on motorized belts; while ventral optic flow contributed less than 10 percent correction, ants whose ventral eyes were painted over with opaque black lacquer still measured exact return distances in total visual uniformity.

That left the third hypothesis, first suggested by Henri Piéron in 1904: an internal biomechanical step counter (pedometer) that multiplies leg strides by stride length. In 2006, Matthias Wittlinger, Harald Wolf, and Rüdiger Wehner tested the pedometer model in Maharès with an experiment of classical mechanical simplicity—putting desert ants on stilts and stumps.

The researchers trained free-ranging Cataglyphis fortis workers to walk from their nest along a linear 10.0-meter aluminum channel open to the polarized blue sky to a feeder containing cookie crumbs. Once an ant reached the feeder at 10.0 meters, Wittlinger captured her, anesthetized her briefly by cooling, and altered the length of all six legs before releasing her in a parallel test channel to run home:

When those same stilted and stumped ants were returned to their nest and allowed to make a second complete trip to the 10.0-meter feeder on their modified legs the entire way out and back, both groups stopped and searched at 10.0 meters with zero error, proving that femoral chordotonal proprioceptors and thoracic step-integrators count actual stride executions out and count the identical quota of strides back down to zero.