Going backwards, or, devolution?

Carl Zimmer has another one of his excellent summary articles, this time about the problems encountered by a research group that tried to make a protein that had evolved into one form,  evolve back to the starting point. This is being touted as a molecular version of “Dollo’s Law” (which is not a law), that evolution is irreversible. It isn’t, at least in theory, but it is, in practice. Why?

Consider the following: a single gene changes one base pair (one of the paired “letters” of the genetic “code”) with a probability of some kind. Let’s say it is one in a thousand million (one in 10 to the power of 9) copying events will lead to that exact mutation. Suppose it is viable, and so gets passed on. What is the chance that it will reverse? Presumably, if all else is equal, the very same: one in 10 to the power of nine. That is supposing that nothing else occurs that changes the viability of organisms that have that mutation. So the probability of the mutation and its reversal is one in 10 to the 18th power! This is not impossible, given the sizes of some populations, especially of bacteria and other single celled organisms, but it is highly unlikely.

Now suppose that because the first mutant has occurred, another mutant is going to be more likely to get passed on because it makes the first mutant more effective; say it acts to catalyse some biochemical reaction that is useful. That, too, has a certain chance. But now, the chance that the first mutant will be reversed is even higher, because now the initial gene is likely to be less effective in the presence of the second mutant’s product.

Iterate over several hundred trillion generations, both simultaneous and sequential. Every time a new mutant occurs, and some other gene now changes to make the new gene work better, the changes become entrenched in the developmental process of that species and lineage of organisms. Now, it gets less and less likely that reversals will be efficacious. Hence it is less and less likely that organisms that carry reversals will be viable.

Moreover, with the new genes, the secondary genes that rely on the new genes will over time tend to displace the old “helper” genes and products, which are now less useful. Eventually, if you revert a gene to the ancestral state, it will simply be unfunctional, as is the case in the reported research.

Evolution is a process of taking unlikely events and making them tie in with other unlikely events. To reverse evolution in every detail, is nigh on impossible, because there simply is not enough time or numbers of organisms or gene copies to be able to “find” the way back. You may get a similar outcome, say, body size or skin colour, but the pathways used will be different, because of the chances.

And that is why “devolution” is a nonsense term. No lineage of organisms or genes “devolves”; it merely evolves in novel ways.