Most vertebrates' early ancestors were a lot squashier than they are today- they had a primitive backbone called a notochord, but no vertebrae. Since the 1930s scientists have thought that vertebrae evolved to make bodies and tails stiffer, which would have helped these ancient predecessors swim more powerfully by attaching muscles to solid vertebrae to produce more force. John Long at Vassar College and his research team are interested in vertebrae evolution, so they came up with a novel way to test this idea using both robots and computer simulations(p. 4732).

First, the team built experimental robots, three one-eyed light-sensitive`tadpoles' called Tadros, complete with a minicomputerised brain that controls their movement. The team then modified Tom Koob's method of making artificial tendons out of gelatine to manufacture tails of different lengths and stiffnesses. Wondering how these tails would help the Tadros to swim, they entered the robots in a swimming gala. Placing the three Tadros on one side of a tank, they trained their cyclops eyes on the target: a light suspended above the tank on the other side. The team measured the Tadros' speed and time as they raced toward the light, and their ability to maintain position under it,to rank their overall swimming performance.

Next, the team wondered what would happen if the three Tadros could mate with each other and produce the next generation, complete with new tails;would better swimmers have offspring with stiffer tails? They turned to an algorithm in a computer to model the process of genetic mixing that occurs during mating, to produce the next generation of Tadro tails, allocating the most successful tail the greatest mating success and poorer swimmers less luck. The final product of the algorithm was the stiffness and length values for the second generation of tails, which the team made and attached to the Tadros for the next swimming gala, starting the evolutionary cycle all over again.

The team showed that over 10 generation cycles, stiffer tails evolved as swimming performances improved. In generations where one or two Tadros did very well and were successful in mating, there was a big improvement in the next generation's swimming performance. They swam faster to the light target,maintained position more reliably and wobbled less. Tail stiffness went up too, showing that a stiffer tail is important for improving swimming performance.

However, Long says, their analysis shows that increasing tail stiffness only accounts for 40% of the improvement in swimming performance, meaning that other factors are involved. The next step, he adds, will be to find out what these factors are. He's planning a nasty surprise for the next generation of Tadros - adding a predator to the tank during the swimming competition, to see how this affects their swimming performance, and their tail evolution.

Long, J. H., Jr, Koob, T. J., Irving, K., Combie, K., Engel, V.,Livingston, N., Lammert, A. and Schumacher, J. (
). Biomimetic evolutionary analysis: testing the adaptive value of vertebrate tail stiffness in autonomous swimming robots.
J. Exp. Biol.