The way fungus-like slime moulds grow could help engineers design wireless communication networks.
Scientists drew this conclusion after observing a slime mould as it grew into a network that was almost identical to the Tokyo rail system.
The scientists describe their ideas for "biologically inspired networks" in the journal Science.
They have incorporated the slime mould's efficient strategy into a mathematical formula.
This "slime formula" could help engineers develop better, more efficient designs.
The single amoeboid cells of slime moulds fuse and spread into a network as they feed and grow.
"These biological networks have been honed by many cycles of evolutionary selection pressure," wrote the researchers in their article.
The research team, led by Dr Atsushi Tero from Hokkaido University, Japan, wanted to capture this evolved efficiency, which they say could be used to inform human engineering decisions.
The scientists put the slime to the test by allowing it to grow on a wet surface on which they placed oat flakes in locations that corresponded to the cities surrounding Tokyo.
They placed the slime mould, Physarum polycephalum, in the centre.
As it grew outwards, it organised itself into a network around the food that closely resembled the train network connecting Tokyo to its surrounding cities.
The researchers then converted this growth "strategy" into a mathematical formula.
The researchers say that this model could provide a starting point for improving the efficiency and even decreasing the cost of "self-organised networks", such as computer and mobile communication networks that are not centrally controlled.
Mould solves maze
One of the researchers, Dr Mark Fricker from Oxford University, UK, told BBC News that the whole idea of using slime moulds in this way came from Toshiyuki Nakagaki, a scientist also based at Hokkaido University.
A decade ago, Dr Nakagaki showed that the slime could find the most efficient way through a maze.
"Toshi has been working on getting them to solve all sorts of problems," said Dr Fricker, "and extending that work to show they form robust networks."
Professor Wolfgang Marwan of Otto von Guericke University, Germany - who was not connected to the research - described the significance of the findings.
The researcher, writing in the journal, said: "The work provides a fascinating and convincing example that biologically inspired mathematical models can lead to completely new, highly efficient algorithms... for applications in such areas as computer science."