A single-celled slime mould mindlessly foraging for food can create a network as efficient as the Tokyo rail system, researchers say.
A team of Japanese and British researchers say the behaviour of the amoeba-like mould could lead to better design of computer or communication networks.
The slime mould Physarum polycephalum grows to connect itself to food sources as part of its normal behaviour.
The mould "can find the shortest path through a maze or connect different arrays of food sources in an efficient manner," wrote Atsushi Tero of Hokkaido University and his colleagues in this week's issue of Science.
The researchers noticed that the slime mould spreading to gather scattered food sources organizes itself into a gelatinous network that interconnects the sources and looks somewhat like a railway system.
To emphasize this similarity, they placed oat flakes on a wet surface in locations corresponding to Tokyo and cities in the surrounding area and let the slime mould loose.
The resulting slime network bore a striking similarity to a simplified layout of the extensive railway network that exists around Greater Tokyo.
The resemblance to the real-world network was even stronger when the scientists prevented the mould from growing in areas corresponding to lakes and mountains — places the railway can't go — by bathing those areas in bright light, which the mould avoids.
Being a single-celled organism, the mould lacks a brain or any kind of nervous system, so the network wasn't planned but arose from very simple biological rules, which "have been honed by many cycles of evolutionary selection," the researchers wrote.
The researchers also compared the mould's network to an idealized network based on a mathematical model called the minimum spanning tree, a geometric concept used to create cost-efficient networks.
They broke down the mould's behaviour and distilled it into a set of simple mathematical rules and built a new computer model out of them.
The model could lead to more efficient transportation, computer and communications networks, the researchers said.
"The model captures the basic dynamics of network adaptability through interaction of local rules and produces networks with properties comparable to or better than those of real-world infrastructure networks," wrote Wolfgang Marwan of Germany's Otto von Guericke University in a related article in Science.