The diminishing returns of beneficial mutations

Published: 7 July 2011 (GMT+10)
Photo by Eric Erbe, wikipedia Escherichia coli

Low-temperature electron micrograph of a cluster of E. coli bacteria. Each individual bacterium is oblong shaped.


Published: 7 July 2011(GMT+10)
Subsequently published in Journal of Creation 25(3)

Beneficial mutations are often seen as the engine of evolution (Mutations: evolution’s engine becomes evolution’s end!). However, beneficial mutations by themselves don’t solve the problem (see Beetle Bloopers). Mutations not only have to be beneficial, but they have to add biological information, i.e. specified complexity. However, practically all beneficial mutations observed have been losses of specified complexity (The evolution train’s a-comin’), with only a few disputable examples of mutations increasing information ever found (e.g. bacteria that digest nylon, citrate or xylitol).

Epistasis: how do mutations interact?

However, mutations need to be more than beneficial and information-increasing to produce new coordinated structures and systems, as microbes-to-man evolution requires. Mutations don’t act alone; the effect of a mutation on an organism’s phenotype depends on other genes, and mutations in those genes, in the genome. This is called epistasis; it is an important consideration for evolution because how mutations interact will determine if they could possibly build new structures in a stepwise manner.

… the cumulative effect of the ‘beneficial’ mutations together was smaller than it would be if the mutations were considered independently.

For microbes-to-man evolution to occur, mutations need to be not just (specified) information-increasing and beneficial, they also need to work together. This also has to be the main dominant trend in adaptive evolution so that the mutations can together produce new biological structures and systems. This phenomenon is called synergistic epistasis (SE), where the combined effect of mutations is greater together than the sum of their individual effects. This is obviously a good situation for beneficial mutations, but very bad for harmful mutations. In harmful mutations, SE can result in synthetic lethality1, where the combined effects of several harmful mutations are compounded by each other’s presence, resulting in such a bad effect that it kills the organism.2 So evolution needs SE to be common only in beneficial mutations; it works against evolution when it occurs in harmful mutations.

Antagonistic epistasis (AE) is the opposite of SE. It occurs when mutations have a negative influence on each other, such that their combined effect is less than the sum of the effect of the individual mutations. For harmful mutations, this is a good thing because it mutes the effect of individual mutations and stalls error catastrophe.3 This is obviously no help for evolution in the long run, since they are still harmful mutations. However, AE presents problems for evolution if it occurs in beneficial mutations. The benefits of individual mutations are muted by other beneficial mutations, resulting in a decreasing rate of fitness increase with every beneficial mutation added.

How not to work together

Two recent studies investigated the effects that beneficial mutations have on each other and came up with basically the same results. One study looked at the combined effect on fitness from some of the earliest beneficial mutations to occur in Richard Lenski’s “Long Term Evolution Experiment” on 12 Escherichia coli populations.4

This is the same experiment that produced an E. coli population with the ability to utilize citrate under aerobic conditions, whereas it couldn’t before. This was widely hailed as an example of ‘evolution’, but see our response, Bacteria ‘evolving in the lab’?. Another study, published in the same issue of Science, looked at the effect beneficial mutations have on each other in an engineered strain of Methylobacterium extorquens.5

Both studies found that beneficial mutations interacted under an overall trend of antagonistic epistasis. Khan et al., in comparing their study with Chou et al., pointed out the results of both studies were virtually identical:

“Note that similar trends were seen by Chou et al. … . That study, like ours, found that four mutations interacted to yield diminishing fitness returns, whereas one mutation had the opposite effect.”6

Therefore, the cumulative effect of the “beneficial” mutations together was smaller than it would be if the mutations were considered independently—i.e. they display an overall trend of antagonistic epistasis. Some individual mutations displayed synergistic epistasis, but they were a minority, and were not enough to reverse the overall trend.

Khan et al. explain this as a result of environmental adaptation:

“Mechanisms that may explain this deceleration include reductions in the number and effect-size of beneficial mutations as a population becomes better adapted to its environment … In other words, epistasis acts as a drag that reduces the contribution of later beneficial mutations.”

But is this it? No doubt that this is a fair assessment of these results as far as they go. These experiments were done in strictly controlled environmental conditions, so the range of questions that can be answered is limited. However, these results didn’t take into account environmental flexibility and change. Khan et al. observed examples of previous mutations that stymied the adaptive capabilities of some lines relative to others in the population.7 But the obvious question is this: what effect does antagonistic epistasis have when the environment changes? Are the organisms as robust to mutation and as adaptable as their ancestors?

Humanity’s own long-term ‘experiments’ in artificial selection would probably present the clearest answer, and it would be “no”. For example, Dogs have been artificially selected for all sorts of traits for centuries, and the typical experience is that purebred dogs are weaker, have more congenital problems, and live shorter lives than ‘mongrel’ dogs (“A Parade of Mutants”—Pedigree Dogs and Artificial Selection).

What is a beneficial mutation?

Both studies stated they were studying beneficial mutations. But what do they mean by beneficial? Are these mutations universally beneficial, or only within a certain environmental context? These may seem like trite questions, but they become immensely important when we consider the context of these studies. As I stated above, these are laboratory studies conducted in strictly controlled environments, so the mutations observed are only known to be ‘beneficial’ within a strict environmental context.

Moreover, Chou et al. conducted their experiments on an engineered bacterial strain that, even without mutations, grew three times slower than the wild-type in the same environment.8

In the engineered strain Chou et al. eliminated an essential metabolic pathway and replaced it with another from a different species. All the ‘beneficial’ mutations in the engineered strain were merely compensating for the loss of the native metabolic pathway. The same mutations in the wild type would most likely be harmful.

Evolution thus has three strikes against it: most mutations are not beneficial, practically all mutations destroy specified complexity, and, now, even ‘beneficial’ mutations work against each other.

Finally, a beneficial mutation is not necessarily a mutation that increases specified complexity. Something is beneficial if it confers an advantage, not necessarily if it adds information. This points to an important issue: mutations not only have to add information to support evolution, but they also have to be selectable. Since mutations (apart from a few trivial examples) are universally losses of specified complexity, and natural selection is incredibly slow and weak, beneficial mutations are ultimately no help to evolution.

Genetic entropy and the mystery of epistasis

These studies reflect a universally consistent trend in lab experiments on adaptation:

“The most consistent finding across studies of laboratory-evolved populations has been a rapid deceleration of the rate of fitness increase.”

The two scientific reports discussed above are in line with those consistent results, and serve as further confirmation of Dr John Sanford’s landmark book: Genetic Entropy and the Mystery of the Genome.9 Dr Sanford pointed out that the genome is in a state of inexorable decay because of mutations. If beneficial mutations generally get in the way of each other, their combined effects cannot stop this process of decay in the genome. Evolution thus has three strikes against it: most mutations are not beneficial, practically all mutations destroy specified complexity, and, now, even ‘beneficial’ mutations work against each other. While mutations may be of limited benefit to a single organism in a limited context (e.g., sickle cell anemia can protect against malaria even though the sickle cell trait is harmful), mutations seem to be no benefit whatsoever for microbes-to-man evolution, whether individually or together.


  1. ‘Synthetic’ is used in this term as the adjectival form of ‘synthesis’ rather than as a synonym for ‘artificial’—i.e. the synthesis of multiple minor mutations is lethal to the cell or organism. Therefore, ‘synthetic’ is meant to be roughly synonymous with ‘synergistic’. Return to text.
  2. Kaelin Jr, W.G., The concept of synthetic lethality in the context of anticancer therapy, Nature Reviews Cancer 5:689–698, 2005. Return to text.
  3. Error catastrophe occurs when harmful mutations accumulate faster than natural selection can weed them out. It leads to a continuous fitness decline in every generation. If not reversed, it will result in the extinction of the population. Return to text.
  4. Khan, A.I., Dinh, D.M., Schneider, D., Lenski, R.E. and Cooper, T.F., Negative epistasis between beneficial mutations in an evolving bacterial population, Science 332:1193–1196, 2011. Return to text.
  5. Chou, H.-H., Chiu, H.-C., Delaney, N.F., Segrè, D. and Marx, C.J., Diminishing returns epistasis among beneficial mutations decelerates adaptation, Science 332:1190–1192, 2011. Return to text.
  6. Khan et al., ref. 3, p. 1195. Return to text.
  7. Khan et al., ref. 3, p. 1193. Return to text.
  8. Chou et al., ref. 4, p. 1191. Return to text.
  9. Sanford, J., Genetic Entropy and the Mystery of the Genome, 3rd edn, FMS Publications, New York, 2008. Return to text.

Readers’ comments

Comments are automatically closed 14 days after publication.