Army ant self-assembly
I study self-assembled structures created by army ants of the genus Eciton. These responsive and dynamic structures are formed when the ants join their living bodies together.
Army ants are nomadic predators that live in large colonies consisting of hundreds of thousands of ants, and they live a fast-paced life that requires a high rate of prey delivery. They deploy these self-assemblages to manage and organize the flow of traffic along their daily raiding trails. Therefore I often refer to these structures as a kind of responsive infrastructure. There is no one ant coordinating the construction process, but rather these structures form by self-organization, where the individual ants have access to only locally-sensed information. How such collective order can emerge from these local interactions is one of the main questions underlying this research.
Eciton burchellii and Eciton hamatum army ants build bridges out of their own bodies to cross over small gaps in the leaf litter of the rain forest where they forage.
In a set of experiments performed along with collaborators Simon Garnier and Iain Couzin, we found that these bridges are highly responsive to the flow of traffic, and adjust their size based on the amount of traffic they need to support. They also break apart quickly if the flow of traffic is stopped entirely, and are quickly restored to their original size if disturbed.
To understand the dynamics of these structures better, I co-designed and conducted another set of field experiments, in close collaboration with Chris Reid, on Barro Colorado Island in Panama. We wanted to know if the ants were using these bridges to create shortcuts, as we suspected from observing them in the field. In addition to Chris, Simon, and Iain, we also enlisted the help of Scott Powell, who had previously studied the behavior of “pothole plugging,” when individual ants use their bodies to fill holes.
We designed an apparatus that essentially forced the ants to deviate from their path. If they were creating shortcuts, we expected to see a small bridge forming at the corner of the apparatus, and then moving down over time. And this is exactly what we observed. However, to our surprise, the bridges did not move down the same amount when we changed the angle of the apparatus. At small angles, the bridges would move quite far, almost to the location of the original trail, but at larger angles, the bridges would only grow and move a short distance from where they originally formed.
We had an intuition about why this might be: Thinking about the geometry of our experiment, we realized that bridges at larger angles would grow much faster in area as they moved down. A bridge with a larger area would require more ants to form, and this could become quite costly if many ants were required to stop foraging to form a structure. To analyze this in more detail, we worked with Albert Kao, who developed a mathematical model of this cost-benefit trade-off. Ultimately we found that bridges move down to a point at which the cost of locking up additional workers in the structure begins to outweigh the benefit of the shortcut created. Thus, these ants are making a collective computation, to determine exactly how far to move their bridges. This computation results entirely from individual sensing, without any oversight or complex communication between individuals.
You can download and read the paper we published in PNAS here.
I am currently working on analyzing data from a few other experiments in which we examined the formation of different kinds of structures. I plan to update this page with more once that work is published. Check back soon!