Error Catastrophe, when phenotypes evolve to go extinct

This is the Wiki Entry on Error Catastrophe:

The wiki entry is cryptic. It appears to relate to viruses, but one might imagine it could be extensible to humans or other kinds of populations.

My reading of the wiki doesn’t mean viruses totally disappear, only that the original strain of virus disappears, which of course would be of interest in treating diseases caused by viruses. Thus if we extend the idea to humans, the error catastrophe model doesn’t say that humans will necessarily go extinct, only that the idealized fittest human stops being the fittest and that phenotype goes extinct.

I could of course be reading this wiki entry totally wrong…

I did text search in Joe Felsensteins book on Theoretical Evolutionary Genetics and could not find the term explicitly mentioned, although there is a reference to Mutational Load and the ENCODE project on page 161.

Everyone is invited to post on what they find or think on the topic of error catastrophe.

1 Like

Thought-wise, it’s just math. The math describes (in a much simplified form) that we think happens in populations.

I once worked out similar math for Dawkins’ WEASEL example. Depending on the the choice of parameters the search will converge to the target with certainty, with probability, or not at all. Error Catastrophe is more complex, but the operating principles are the same.

In other words, they evolve, or drift.

1 Like

After reading thru the “Eigen’s Paradox” link, I think we should be careful about context: The paradox refers to self-replicating molecules, and the human genome (or most any other critter) is well beyond this level of mathematical modeling. I think we can consider Error Catastrophe in terms of continuation or extinction of phenotypes, not of whole genomes.

WikiPedia is usually a good first port of call (though investigating further is often wise), but this doesn’t strike me as a particularly good instance of a WikiPedia article. I think that it ought to refer to populations, rather than organisms.

It seems to me that at it’s core the concept is simple - above a certain mutation rate mutation will overwhelm the ability of stabilising selection to maintain the genotype of the population. Looking a little further I discover that the concept is generally applied in the fields of molecular evolution and virology (but I see no a priori reason why it couldn’t apply to a mutator phenotype of a cellular organism), and that there is a phase transition between a quasispecies maintained by selection and a population with randomised genomes.

I hypothesise that the model assumes soft selection. With harder selection the population would decline to extinction in a mutational meltdown.

I presume that cellular genomes operate well short of the error catastrophe threshold. (Any species that crossed it wouldn’t be around for use to see.) There’s a lineage of the plant genus Plantago were mitomic evolution has gone into overdrive; a mutator phenotype presumably got fixed by drift. Within a sub-lineage evolution has accelerated even faster, as a further defect in DNA repair/proof reading has been incorporated. Since these lineages haven’t crossed the error catastrophe threshold, I infer that normal genomes must be well away from the threshold.


Error catastrophe is usually applied to changes within an individual, particularly within a cell. It is the idea that a mutation could degrade the accuracy of reproduction of the genotype, enough that each mutation would end up causing more than one further mutation. As the process runs away with itself, there is then an explosion of mutations. How it applies within a population is not clear. Within a cell, the main observation seems to be that effects of mutations on further rates of mutation are not great enough to be in the realm of the “error catastrophe”.


Is this related to Mullers Ratchet? If so, wouldn’t recombination severely mitigate its effects?

Thanks Joe, that was helpful.

Yup, that’s what it looks like to me with the original genotype/phenotype or whatever one wishes to call it, it’s just gone. BUT it doesn’t mean the descendants are gone.

The way I interpret this is that a particularly virulent pandemic strain is gone out of the hosts (like humans), but it doesn’t mean the descendants of that original strain are completely gone.

1 Like

I have read a little more and find at least 3 definitions of “error catastrophe”. The one I described here is one I learned long ago from Jon Gallant, long a colleague in my department who wrote a review article on it with Chuck Kurland. But I also see people using an “EC” argument for explaining aging by accumulated mutations, and another that is like the Wikipedia page description here – an argument about mutational load depressing fitness just enough to make a population of single-celled organisms unsustainable in the long run. These are three very different, quite incompatible definitions. When I get a chance I need to ask Jon Gallant and Claus Wilke which of them is the real error catastrophe. So I withdraw my explanation as it is OK for one of these concepts, not for the other two.


I didn’t mean to imply that Gallant and Wilke were themselves error catastrophes. Make that “I need to ask Jon Gallant and Claus Wilke which of these definitions is the real error catastrophe”.