Is mutation nonrandom? If so, in what sense?

I was hoping to write this post a week ago, but war intervened, and I spent much of the time since then watching news about the invasion of Ukraine by Russia, the incredible courage of large numbers of Ukrainians who are resisting it, and the incredible courage of those Russians who are protesting their government’s aggression. This post takes a short break from that.

Two months ago, an article in Nature argued, from observations on the plant Arabidopsis thaliana that mutation is not random. This was the occasion for loud declarations from creationists and ID advocates that the foundations of the Modern Synthesis had been undermined.

What I would like to ask is, whether modern evolutionary theory, rests on the assumption that mutation is “random”. And if so, in what sense does it need to be random for the theory to be tenable? I’ll make some suggestions below, but they are tentative, and your thoughts are welcome.

Some questions:

  1. Is there evidence that the mixture of mutations that happen shifts in response to environmental changes, in such a way that among those mutations that do occur, a larger fraction of them are adaptive in those environments?

For example, when one could have mutations that helped the organism adapt to higher temperatures, do more of those mutations show up? And when the temperatures are getting colder, do the mutations that occur become more likely to help adaptation to lower temperatures and less likely to help adaptation to high temperatures? I don’t know of any evidence that shifts like this happen.

  1. Does modern evolutionary biology have as a fundamental tenet that mutation rates in all parts of the genome are equal and that mutations that change bases are equally likely to result in each of the other three possible bases?

There, the answer is clearly no, since many models of base substitutions in neutral mutation have for a long time allowed for mutation rates to differ in different parts of the genome, and for the pattern of changes to depart from simple symmetrical models such as the Jukes-Cantor model. This includes allowing models of regional mutation rate variation, such as Hidden Markov Models.

Let's discuss ...