Trail clearing behaviour in Atta leaf-cutter ants

Leaf-cutter ants, unlike most other ant species, construct cleared foraging trails: they remove all undergrowth and flatten the underlying soil. Much like on human roads, movement on such trails is significantly faster. The resulting trail networks span several kilometers, and many kilograms of leaf litter are removed off the trails. The construction of these trails takes approximately 11,000 ant-hours worth of work.

Is the construction worth the effort?

Considering the enormous dimensions of their trails and the high associated costs, is the benefit of having these trails really worth the effort? It turns out, it is: by combining ecological and physiological data into a computational model, we showed that trails provide energetic benefits (BEAS). We further investigated what part of the colony should contribute to the construction process to ensure that the costs do not outweigh the benefits, and what other aspects might influence the cost/benefit balance.

How do the thousands of trail building ants coordinate their effort?

Comparable large-scale collective behaviour is often controlled via the deposition of information in the environment (i.e., stigmergy): for example, ants use pheromone trails to recruit workers to food locations. By comparing the trail clearing behaviour in the field and laboratory to behavioural mechanism models with different coordination methods, we showed that the entire process does not require any form of coordination at all (Proc B), and that trail clearing is based purely on random encounters between ants and obstructions and a fixed removal probability. Unexpectedly, trail clearing is an example of a collective behaviour that does not require any information exchange.

I recently had a chat with ABC Science Show's host Robyn Williams about the leaf-cutter highways, which you can find here.

Some of the methods I applied to track ants are shown below. For the first video, we used a machine-learning method to define what constitutes an ant, and then tracked individual ants over time. Colours in the second video show the traffic intensity aggregated over 30 seconds, as measured in one of our lab experiments.

Collective nest construction and chain formation in the weaver ant Oecophylla smaragdina

Weaver ants are an arboreal ant species living in nests made up from living leaves. Construction of these nests requires the coordinated effort of a large workforce to fold and then (using the silk of their larvae) weave leaves into the desired shape. During the folding process ants assemble into pulling chains that allow them to focus the force of many ants on a single point on the leaf perimeter.

How do individual ants select a single leaf in a tree canopy from which they start the nest construction? And how do additional workers decide to join in the effort, and to form pulling chains?

In laboratory experiments we gave ants artificial leaves made from paper, and observed how ant colonies used them to construct simple nests. We find that shape and flexibility are the key criteria (PLOS One) for the selection of a suitable nest substrate: sturdier leaves are not used for construction, but softer leaves are; on those, the flexible tip is preferentially chosen by the first pulling ant. Subsequent ants will join in with the first, but are more likely to leave the chains if the pulling effort makes no progress.

Below is a timelapse video of what the construction process looks like on an artificial leaf.

Accrued images

A collection of work-related images
Spying into ant nest Ant-flow visualised over time Computer vision robot project Weaver ant Trail-clearing simulation Evolutionary simulation Aerial support for ant hunt Ant tracking via machine learning Leaf-cutter ant at work