More than One Deleterious Mutation per Genome per Generation

Mutations are the fuel that drives the engine of evolution. Without mutations there would be no variation upon which natural selection and other evolutionary forces could act. Furthermore, much of the theoretical results regarding evolutionary genetics depend on estimates of mutation rates. For example, Kimura showed that the rate of fixation of neutral mutations is equal to the neutral mutation rate. Additionally, many models to explain the evolution of sex and recombination depend on the amount of deleterious mutations per genome per generation (U).

A group led by Peter Keightly (DOI) have used three inbred lines of Drosophila melanogaster to estimate the per nucleotide mutation rate (u). They identified 37 new mutations out of the over 20,000,000 nucleotides screened from each of the lines. This gives them an estimate of 8.4x10-9 for u averaged across all three lines, but the lines did have significantly different mutation rates.

To estimate the deleterious mutation rate, Keightly and colleagues used the following formula:

U = 2uGC

Where 2G is the diploid genome size (236 Mb), and C is the fraction of nucleotides under selective constraint. Their estimate of C (0.58) came from this paper comparing the D. melanogaster and D. simulans genomes. Plugging in all the values, one calculates U = 1.15. That means each individual, on average, inherits a new deleterious mutation from one parent.

These estimates are possible because the authors studied an inbred line. In nature, many of these deleterious mutations would be purged by natural selection. But the inbred line is not susceptible to these selection pressures and can accumulate deleterious alleles at much closer to the neutral rate.

The evolution of sexual reproduction and recombination comes at a cost for organisms that employ this mechanism of reproduction. To maximize one's genetic input to the next generation, one should reproduce asexually and pass on all of one's alleles to one's progeny -- rather than the 50% that results from asexual reproduction. Researchers who have modeled the evolution of sexual reproduction have concluded that the benefits of recombination in removing deleterious mutations make up for the costs of sex if the deleterious mutation rate is high enough. The estimated deleterious mutation rate from this study, however, is too low to explain the recombination rate observed in Drosophila. This excludes a purely functional explanation for recombination -- proper segregation of homologous chromosomes -- which some people prefer to the population geneticists' model.

I'd like to close by pointing to an assumption made by the authors regarding the neutrality of synonymous mutations.

Second, our estimate of the nucleotide site mutation rate is about 5-fold (95% confidence limits 2-fold and 12-fold) higher than a phylogenetic estimate from synonymous site divergence, assuming that wild flies undergo ten generations per year. This could be partly due to inaccurate estimates of species divergence times or to differences in generation times between laboratory flies and wild flies.

Or it could be due to the selective constraint on synonymous sites due to codon bias. The experimental populations are not under the same selective constraint as the wild populations so they may not be as effective at purging non-optimal (and only slightly deleterious) synonymous changes.


Haag-Liautard C, Dorris M, Maside X, Macaskil S, Halligan DL, Charlesworth B, Keightley PD. 2007. Direct estimation of per nucleotide and genomic deleterious mutation rates in Drosophila. Nature 445: 82-85. doi:10.1038/nature05388

More like this

The overall mutation rate during DNA replication is about 10-10 per site per replication. In order to convert this to a mutation rate per generation we need to know how many DNA replications there are between zygote and gamete.

In most mutlicellular animals this number is about 100--that's the average of a high value in males and a lower one in females.

This means that the average mutation rate per generation should be about 10-8 which is pretty damn close to the value that Haag-Liautard et al. obtained. Many of us have been using u = 10-8 per generation for years and years. It's nice to see that experimental measurements agree with theory.

Is that per site mutation rate constant across all domains? Do the bacterial and eukaryotic DNA polymerases have similar error rates?

As far as I know, all prokaryotes and all eukaryotes utilize a DNA replication complex that has an accuracy of at least 10-8 per nucleotide. They all possess repair enzymes that repair 99% of the damage caused by DNA polymerase errors, giving an overall mutation rate of 10-10.

I know it's commonly believed that different species have different mutation rates but I don't think there's any solid evidence to support that notion. (Exception: radiation -resistant bacteria.)