Rewind and Repeat

The final chapter of The Ancestor's Tale consider's a question that had occurred to me earlier in my comments: what would happen if we could rewind to some earlier era and let the whole process occur again? Would evolution retrace its steps? Closely? At all?

It turns out that Stuart Kauffman had posed just such questions some time ago. Physicists are addicted to this kind of gedanken experiment, but biologists, not so much. There is a pretty sharp divergence of opinion, it seems. Steven Jay Gould, and in a slightly different context, Ernst Mayr could be considered advocates of "radical contingency." In that view, the replay would be utterly different. Simon Conway Morris takes an almost opposite view, with Dawkins taking an intermediate position a bit closer to Morris. Put me down with Morris.

The core of the Gould-Mayr argument is that the detailed path of evolution depends on such an intricate set of contingencies that any replay would rapidly diverge from the original. The Morris argument depends on constraint and convergence - the idea that there are only so many ways to run a railroad, so that evolutionary detours tend to get herded back to the tracks. The space of possible evolutionary trajectories is very large, and nearby paths diverge rapidly, but the overall orbits are controlled by low dimensional attractors - if I can rephrase this in terms of dynamical systems theory.

There is some evidence. The separation of the continents can be considered natural experiments in this vein, and indeed convergent evolution seems to rule. The same sorts of evolutionary niches appeared on multiple continents, and were filled by outwardly similar creatures. A marsupial wolf differs in important respects from his canine counterparts, but the ecological space occupied was very similar, and a lot of convergent features evolved.

Certain adaptations appear again and again: echolocation, true flight, coasting flight, air breathing. Other clearly useful adaptations have failed to appear. No animal (except man) uses radar, and it was the result of intelligent design, not evolution. Ditto the transistor.

Some puzzles occur: nobody could ever explain to him, says Dawkins, why there were no dinosaur moles. It looks like a no brainer to me - that niche was already occupado, courtesy of our great^n - grandparents, our shrew-like mammal ancestors. Similarly disposed of is the question of why chimpanzees aren't evolving to our level of intelligence. Once again, the "occupied" sign is already up.