Nature’s networks: biodiversity’s bottom-up approach
Challenge: more efficient networks
Natural inspiration: slime moulds, plants and ants
Networks are found in both nature and human
societies. They may be physical, like blood vessels
in the body and railway networks; or virtual, like information
flowing between animals in a population, or computers on the
internet. How well a network functions is a trade-off between
the cost of moving objects or information through the system, the
efficiency with which the network operates, and how well it copes
with problems. 1
Optimising the networks people use in everyday life
could bring a host of benefits: reduced journey times on
roads, faster internet connection and cheaper electricity supply to
name but a few. Improvements to networks would bring economic
gains. For example, in the UK road congestion is estimated to
cost the economy £20 billion per year, and this is
predicted to rise as road traffic grows faster than road capacity.
2 A better understanding of networks could enable
planners to improve the efficiency of our road system, allowing it
to carry more traffic without congestion. The mathematics
involved in designing networks is complex, and human designers do
not always come up with the optimum solution.
In nature networks are not designed, but develop
organically in a ‘bottom up’ fashion, the components of
the network finding their own place without any centralised
control. Evolution over the millennia has selected the best
solutions out of all the possible ways of organising systems.
1 This is the equivalent of running our computer models
and mathematical equations millions upon millions of times.
No wonder mathematicians and engineers are looking to diverse
biological systems for design inspiration. Answers have been
found so far in three very different areas of biodiversity:
slime moulds, plants and ants.
Slime moulds are
single-celled organisms that live on decaying matter and resemble
fungi, although they are unrelated. 3 Each
individual cell finds food by sending out a network of thin
tendrils. When one tendril finds a food source, it expands
while all the others slowly disappear 4.
Scientists have recently carried out a clever demonstration of the
efficiency of this method. They took a map of Japan and
placed food morsels on the major cities. They then introduced
a slime mould (Physarum polycephalum) at the point
representing Tokyo. The growing mould explored the map by
sending tendrils in all directions, strengthening those that found
food and terminating those that did not. After 23 hours, the
tendrils formed a network which closely followed Tokyo’s railway
system: the mould had arrived at the same solution as the railway
planners! 1 The railway planners set out to
connect Japan's cities as efficiently as possible; whereas the
mould was just growing towards food by a process of trial and
error. The qualities of self-organisation,
self-optimisation and self-repair exhibited by the mould are
exactly what is needed to build robust technological
Plants demonstrate a
variety of network patterns. Mathematicians have
shown that the most efficient network design resembles a
tree. A major route (like the trunk) splits off into smaller
routes (branches), which themselves split again into even smaller
routes (like the twigs), and so on. However, this type of
network does not cope well with problems: if one route is blocked,
everything ‘downstream’ of it is cut off. Imagine if the
vessels in one of a tree's branches are damaged: all of the twigs
growing from that branch will die. Physicists noticed that
the leaves of many plants have a different type of network: one
which contains loops as well as branches. They carried out
mathematical modelling to discover why this might be. The
results showed that, as well as coping better with ‘blockages’
(there is always another way round if one route is blocked), these
networks were more efficient at dealing with fluctuating loads.
6 This sort of design would clearly be best for
systems such as road networks, where the amount of traffic varies
greatly at different times, and parts of the system may be blocked
at any given time (traffic accidents, roadworks etc).
Ants are a highly successful family, being found
everywhere in the world except
Antarctica, although about 150 species of ant
are now listed as “vulnerable” or worse on the IUCN Red List.
7 People have long been fascinated by how ant
colonies function. These simple insects can perform incredible
feats of co-operation, achieving tasks far beyond any of the
individual ants. 8 Ants communicate with each
other by depositing pheromones (chemicals that other ants can
smell) on the ground. The pheromone is a sign to other ants
to follow the same route, leading to the familiar sight of ants
marching in a line between the nest and a food source.
Experiments have shown that this simple rule – choose the
route that most ants chose before you, by following the strongest
pheromone trail – allows ant colonies to solve a variety of
problems. For example, they can correctly identify the
shortest route to a food source 8, and they can even
gather up dead members of the colony into an ordered cemetery
9. No one ant is aware of the whole task: they do
not look around them or plan ahead. Each ant simply follows
the pheromone trail in front of it. 9 This method
has advantages and disadvantages over the systems usually adopted
by humans. A foreman directing a team of human workers could
probably get an equivalent job done more quickly than the ant
colony – but if the foreman leaves, or makes poor decisions, the
whole system collapses. If the task is very complex, the
foreman might not be able to plan out the most efficient way of
performing it; but an ant colony would eventually arrive at the
solution by trial and error. The ants' bottom-up
system is more robust because it does not rely on one central point
This concept has proved so useful
it has given rise to a branch of mathematics called “ant colony
optimisation” (ACO) , invented in the 1990s
10. In 2003, an academic paper was published that
demonstrated how ACO could be used to design an optimal water
distribution network. The new ant-based mathematics produced
a better solution than other methods. 11 Financial
analysts have since used ACO to solve congestion problems at
airports. At a busy airport, planes may have to queue for
airport gates, causing delays to flights. Computer software
designed using the ACO approach directs pilots to certain gates to
ensure the system as a whole operates as efficiently as
possible. It can successfully predict queues before they even
start to build up, and redirect pilots to avoid delays
12. The same airline company used ACO to
streamline its cargo operations, saving an estimated $10 million
per year. 13 The way ant colonies operate also has
lessons for how to manage personnel in a company: business
consultants are increasingly using ant-based models to advise
companies how to increase productivity. 13
Humans have only just begun to use the lessons we are
learning from biodiversity. The more we study
nature, at every scale from the microscopic veins in a leaf to the
'super-organism' of an ant colony, the more new ideas we find to
help human society become more productive and
Back to Natural Solutions front
- Tero, A. et al. (2010) Rules for Biologically Inspired
Adaptive Network Design. Science 22: 439 – 442
- Goodwin, P. (2004) The economic costs of road traffic
congestion. The Rail Freight Group, London, UK.
Article available in full
online. Accessed March 2010.
to the “Slime Molds”. University of California Museum of
Paleontology. Accessed March 2010.
Mold May Help Design Future Transportation Routes. Treehugger,
January 2010. Accessed March 2010.
- Marwan, W. (2010) Amoeba-Inspired Network Design.
Science 22: 419 - 420
- Katifori, E. et al. (2010) Damage and Fluctuations
Induce Loops in Optimal Transport Networks. Physical
Review Letters 104, 048704.
- IUCN Red List. Accessed
- Ant Colony Optimisation. Marco Dorigo and Thomas Stützle
(2004). Massechusetts Institute of Technology.
Sample pages available online. Accessed March 2010.
- Wokoma, I. et al. (2002). Biologically Inspired Models
for Sensor Network Design. Proceedings of LCS.
- Dorigo, M. el al. (2005) Ant colony optimization theory: a
survey. Theoretical Computer Science 344:
- Maier, H.R. et al. (2003) Ant Colony Optimization for Design of
Water Distribution Systems. Journal of Water Resources
Planning and Management 129: 200-209
Planes, trains and ant hills. Science Daily, April 2008.
Accessed March 2010.
Lessons from the Ant Farm. Chief Executive (US), January
2003. Accessed March 2010.