We live in a world that is deeply interconnected. Nowadays, transportation networks are fundamental to exchange resources and information from a point to another, from a person to another. Every day we travel on roads, we use electricity and water that are carried from distant plants, we connect to the internet to read about facts happened on the other side of the world. Ideally, we would like to be able to travel between cities on the shortest way possible, but sometimes we have to follow long detours. Almost everyone in his life has experienced a black out: sometimes the breakdown of a cable is enough to compromise the distribution of electricity in a whole suburb. However, we all know how expensive it is to install new cables at home, and we can imagine the cost of building a highway.
Thus network planners struggle to build transportation systems that are efficient and robust, but also not too expensive, trying to find the best compromise between competing design goals.
Searching for inspiration, researchers have turned towards nature, observing the spontaneous formation process of natural transportation networks, from ant trails to leaf veins.
One of the main reasons why ants are so interesting from a scientific perspective is that they perform their activities without any central planning. Unlike human systems, there is no single ant having the “big picture”: every ant has just a bit of information about what is going on in the colony. Still, the whole colony together manages to find optimal solutions to the tasks that are needed for its survival. Including moving a (dead) grasshopper for several meters or building a network of trails connecting their nest to the cookies’ cupboard!
So for most people ants mean troubles. Specially when you find a thin black line across the kitchen and wonder how, despite all your efforts, ants have managed to find the cookies for the third time in a single summer. However, some other people, typically kids and researchers, find ant behaviour extremely fascinating.
The colonies of the Australian meat ant, for example, are made of several nests distributed on a large territory. These ants feed by milking honeydew from aphids living on trees, rather than searching for cookies in your kitchen. Since they live in a wild environment and need to transport resources from one nest to the other, they do not limit themselves to follow pheromones traces but build physical trails by removing all the vegetation from the paths connecting nests and trees. Thus meat ant’s trails are likely expensive to build and maintain. Also, it is reasonable to assume that ants would like to have an efficient network that is also robust to predation.
Basically meat ants face the exact same problems of network planners!
A recent study has indeed pointed out that these ants are incredibly good at building transportation networks that satisfy environmental constraints, solving the difficult task of balancing efficiency, robustness and cost.
Understanding how ants achieve such a balance means to understand the basic building rules underlying the formation of their transport networks.
Based on prior studies of meat ant trails, we modeled how these networks evolve, identifying a general mechanism of local cost minimization as the basic rule that leads to a balance between competing design goals.
It turns out that the ants follow simple rules: "connect a new nest to the closest nest”, and “connect a new nest to a food tree if the resulting road is shorter than any existing nest-tree link”. These two rules together are enough to balance efficiency and cost and to achieve some robustness too. Sounds too complicated? Download a demo of the model here: MLM video
Once we have found what nature does, we have tried to apply the same simple rules to predict what would happen to man-made system, electric grids for example, if they were built by these ants.
When we scaled up the model to 2000 nodes, the networks were still efficient and cheap, but lost their robustness. This might be why meat ant colonies tend to stay small and splitting rather than sacrificing robustness. However, human transport networks cannot stay small and need to be robust and resilient. Allowing for a local rule of connection to tree-source nodes, the meat ant rules can fit more human-scale developments. So what ants taught us is that when building a new suburb, it is sufficient to connect it to the closest city area to ensure that the whole power network will be relatively cheap but also quite efficient on the long run. Then we figured out that robustness can be increased or decreased by changing the frequency with which new suburbs are connected to service centers, in this specific example to power plants.
Ça va sans dire, this procedure is very effective when the position of future nodes is not known in advance. If ants could plan ahead the number and positions of their nests, it might be the case that we can return the favour and give them some good advice!
The study is published on the Interface journal of The Royal Society.