01 January 2009

Clone wars, or how evolution got a speed limit

The standard simplified narrative of evolutionary adaptation goes something like this. A population of organisms is exposed to a challenge of some kind. Perhaps a new predator has appeared on the scene, or the temperature of the environment has ticked up a degree or two, or the warm little pond is slowly accumulating a toxic chemical. Some of the organisms in the population harbor (or acquire) mutations – so-called beneficial mutations – and these individuals are more successful in the face of the challenge. The population evolves, then, as these beneficial mutations become more common until they are the new status quo. The change is brought about by selection, and the process is called adaptation.

These beneficial mutations, as one might suppose, are quite rare. Most mutations are either harmful to some degree or have little or no effect. Since the good stuff is so hard to come by, it follows that huge populations will be better able to adapt, and will do it faster, because they contain more of the good stuff.

It's a straightforward conclusion, and it's the basis of some recent challenges to evolutionary theory coming from the Intelligent Design movement. But it's mostly wrong. ResearchBlogging.org Here's the problem with the simple story.

In a very large population, many beneficial mutations will be present at the same time, in different individuals. When the challenge is presented, these beneficial mutants will compete against each other, and typically one will win. This means that most beneficial mutations – specifically those with small effects – will be erased from the population as it adapts. So, seemingly paradoxically, a very large population doesn't benefit from its bounty of beneficial mutations when it is subjected to an evolutionary challenge. It's as though adaptation has a built-in speed limit in large populations, and the effect has been clearly demonstrated experimentally. It's called clonal interference.

As geneticists examined this phenomenon, it became clear that any attempt to measure beneficial mutation rates would have been influenced, perhaps dramatically, by clonal interference. Such experiments were often done in bacteria, in the huge populations that can be so easily generated in the lab. Analyses in bacteria, published 6 or 7 years ago, had estimated the beneficial mutation rate to be about 10-8 per organism per generation. (That's 1 per 100 million genomes per generation.) Since the overall mutation rate is estimated to be about 10-3 per organism (a few per thousand genomes per generation), it was concluded that beneficial mutations are fantastically rare compared to harmful or irrelevant mutations.

Creationists have long emphasized the rarity of beneficial mutations, for obvious reasons. For their part, geneticists knew that clonal interference was obscuring the true rate, but no one knew just what that rate might be. That changed in the summer of 2007, when a group in Portugal (LĂ­lia Perfeito and colleagues) published the results of a study [abstract/full-text DOI] designed to directly address the effect of clonal interference on estimates of the beneficial mutation rate. Their cool bacterial system (based on good old E. coli) enabled them to genetically analyze the results of an evolutionary experiment, using techniques similar to those made famous by Richard Lenski and his colleagues at Michigan State University.

In short, Perfeito et al. took populations of bacteria and allowed them to adapt to a new environment for 1000 generations. Then they looked for evidence of a "selective sweep" in which one particular genetic variant (i.e., mutant) has taken over the population (their system was set up to facilitate the identification of these adaptive phenomena). The same system had been used before to estimate the beneficial mutation rate, and had arrived at the minuscule number I mentioned before.

The Portuguese group introduced one simple novelty: they studied adaptation in the typical large populations, but also in moderately-sized populations, and then compared the results. The difference was profound: the beneficial mutation rate in the smaller populations was 1000-fold greater than that in the very large populations. This means that clonal interference in the large populations led to the loss of 99.9% of the beneficial mutations that arose during experimental evolution. And that means that the actual beneficial mutation rate, at least in bacteria, is 1000 times greater than the typically-cited estimates.

Perfeito et al. further exploited their system to measure the fitness of all of the mutant clones that they recovered. They found that evolution in very large populations generally resulted in beneficial mutations with larger beneficial effects. This makes sense: the slightly-beneficial clones were eliminated by competition, so at the end of the process of adaptation, we're mostly left with the more-beneficial mutations.

Now some comments.

1. It might seem at first that the large populations are still better off during adaptation, since they do generate beneficial mutations, and selectively retain the more-beneficial ones. But the claim is not that large populations don't adapt; the point is that the vast majority of possible adaptive trajectories are lost due to competition, such that only the trajectories that begin with a relatively large first step are explored. That's a significant limitation, and quite the opposite of the simplistic models of design proponents like Michael Behe and Hugh Ross. Genetic models have shown that the only way for an asexual population to get around the barrier is to do what Michael Behe claims is almost impossible: to generate multiple mutations in the same organism. And recent experimental results show that this does indeed occur.

2. Since the early days of evolutionary genetics, the genetic benefits of sex have been postulated to include the bringing together of beneficial mutations to create more-fit genetic combinations expeditiously. In 2002, an experimental study validated this conjecture, showing that sexual reproduction circumvents the "speed limit" imposed by clonal interference in large populations, and in 2005 another experimental analysis showed that sex speeds up adaptation in yeast but confers no other obvious advantage. Perfeito et al. identified this connection as a major implication of their own work:
...if there is a chance for recombination, clonal interference will be much lower and organisms will adapt faster. [...] Given our results, we anticipate that clonal interference is important in maintaining sexual reproduction in eukaryotes.
(One of the hallmarks of sexual reproduction, besides fun, is recombination – the active shuffling of genetic material that generates offspring with wholly unique mixtures of genes from mom and dad.) In other words, one of the most important benefits of sexual reproduction – and especially of genetic recombination – is negation of the evolutionary drag of clonal interference.

3. All of the examples I've mentioned here are bacterial or viral. If clonal interference arises merely as a result of large population sizes, then it should be an issue for other populations too. And it is: in last month's issue of Nature Genetics, Kao and Sherlock present a tour de force of experimental evolution in a eukaryote, demonstrating the importance of clonal interference and multiple mutations in yeast cells growing asexually. In their study, they identified each beneficial mutation by sequencing the affected gene. Wow.

Why does all of this matter? Well, because it's cool, that's why. And it does mean that our biological enemies have a lot more adaptive resources than we used to think. Here are the closing comments of Perfeito and colleagues:
...our estimate of Ua implies that 1 in 150 newly arising mutations is beneficial and that 1 in 10 fitness-affecting mutations increases the fitness of the individual carrying it. Hence, an enterobacterium has an enormous potential for adaptation and [this] may help explain how antibiotic resistance and virulence evolve so quickly.
But also: keep clonal interference in mind when you encounter any simple story about evolution and genetics. Evolution isn't impossibly difficult to comprehend, but getting it straight requires just a little more effort (and a whole lot more integrity) than has been demonstrated in recent work by those who just can't believe that it could be true.

Article(s) discussed in this post:
L. Perfeito, L. Fernandes, C. Mota, I. Gordo (2007). Adaptive Mutations in Bacteria: High Rate and Small Effects. Science, 317 (5839), 813-815 DOI: 10.1126/science.1142284
K.C. Kao and G. Sherlock (2008). Molecular characterization of clonal interference during adaptive evolution in asexual populations of Saccharomyces cerevisiae. Nature Genetics, 40(12), 1499-1504. DOI: 10.1038/ng.280

12 comments:

John Farrell said...

Superb post, Steve. Here's to a great 2009!

RBH said...

And that just generated a project for my evolutionary modeling lab this coming semester. Thanks on behalf of the students in that lab!

Stephen Matheson said...

Thanks John! Yeah, I'm going to focus a lot more on these reviews of peer-reviewed research. They're a lot more fun to write than smackdowns of creationist ludicrosity.

RBH, wow that's cool. Can you share more?

RBH said...

RBH, wow that's cool. Can you share more?

Since the inspiration just now struck me, there's not much more to share. :)

We'll be mainly using Avida as the lab platform, and I'm sure one can replicate the core methodology in the paper you described in Avida -- in fact, a fast Google suggests it's been done; see here (pdf). I'm sure I can work that into an exercise for the kids in the lab.

Wedge said...

How might this offset (or not) the effect of Muller's ratchet on small populations?

Regarding the rate of beneficial mutations, it sounds like the estimates are based on a single measurement taken some x generations after introducing a new pressure. From an ID perspective, it would be more interesting to know the derivative of this value. How is it changing as the organism adapts? If the frequency of beneficial mutations remains relatively constant as the organism continues to adapt to a certain selective pressure, that is consistent with evolution and a capacity for continuous improvement. If the frequency drops off over time and the organism asymptotically approaches some sort of maximum fitness, that is consistent with ID.

Also, a small correction: Behe has never claimed that it is impossible to generate multiple [beneficial] mutations in the same organism. He's more concerned with the ability of evolution to produce changes requiring two or more mutations which are neutral by themselves but beneficial together.

RBH said...

Wedge wrote

If the frequency of beneficial mutations remains relatively constant as the organism continues to adapt to a certain selective pressure, that is consistent with evolution and a capacity for continuous improvement.

That's not the case. Recall that 'beneficial' is not a property of a mutation in isolation, but is a property of a mutation in a particular selective context. And bear in mind that mutations vary in their effects -- some have very small effects, some larger.

Given that, as a population becomes better adapted to a stable environment, the proportion of mutations that are 'beneficial' -- that can generate still better adaptation -- will tail off. JBS Haldane's analogy is still valid: If a microscope is far out of focus, as many as 50% of changes will bring it into better focus. If it is very near a good focus, most changes will drive it away from a good focus, some by moving it away from focus and some by over-shooting.

Evolutionary theory doesn't entail "continuous improvement."

Wedge said...

RBH,

Recall that 'beneficial' is not a property of a mutation in isolation, but is a property of a mutation in a particular selective context.

Right. This makes me skeptical about the usefulness of suggesting a frequency for positive mutations.

as a population becomes better adapted to a stable environment, the proportion of mutations that are 'beneficial' -- that can generate still better adaptation -- will tail off.

Then just keep increasing the selective pressure to keep evolution going - remove more of nutrient x from the medium, etc.

Evolutionary theory doesn't entail "continuous improvement."
No, of course not. Evolution says nothing about the long-term viability of any particular trajectory. It could be that every time we test evolution in the lab, we wind up with dead-end trajectories that are incapable of adapting beyond some more-or-less fixed amount. But is this what you would expect from evolution?

RBH said...

Wedge wrote

Regarding the rate of beneficial mutations, it sounds like the estimates are based on a single measurement taken some x generations after introducing a new pressure.

Actually, no, the observations were made at intervals over 1,000 generations of adaption to a new selective environment.

How about this? Get on eof those highly productive ID researchers -- Behe, Minnich, Axe -- to do it.

Oh, and provide a rationale, based on ID theory, for the prediction you provided:

If the frequency drops off over time and the organism asymptotically approaches some sort of maximum fitness, that is consistent with ID.

Why? Does the designer get tired? Does it (or they) know when the population is approaching a local maximum in the fitness space? How? You want to do science you have to do more than pull a prediction out of thin air.

Wedge said...

RBH,

Actually, no, the observations were made at intervals over 1,000 generations of adaption to a new selective environment.

Thanks, I'll have to read the paper. Where does the beneficial mutation rate come from, though? Is it an average of the measurements?

Does the designer get tired? Does it (or they) know when the population is approaching a local maximum in the fitness space?

My experiment was designed to test the potential of existing organisms to evolve. If ID is true, you would expect an organism's ability to adapt to a new environment via random mutation to be bounded (though perhaps very great). This sounds almost like a tautology to me. If adaptation potential isn't bounded, then (at least in the tested case) the ID hypothesis is superfluous.

I don't know why you're talking about the designer tinkering during the experiment.

Anonymous said...

"My experiment was designed to test the potential of existing organisms to evolve. If ID is true, you would expect an organism's ability to adapt to a new environment via random mutation to be bounded (though perhaps very great). "

Why would ID's predictions differ from evolutionary predictions? Does ID use a different model of fitness space? Why? Stuart Kauffman and I think others (Wright) have always said that natural selection could take a species up a fitness peak and then leave it there.

It was Fisher, btw, who said that tiny changes in fitness had a 50 percent chance of being beneficial and I think (but am not sure) he used the microscope analogy.

Donald

Anonymous said...

CLonal interference occurs in asexual homogeneous populations only. This mean the individuals are clones or nearly so and there is no recombination. This makes the statement of "large populations cannot benefit from their bounty of beneficial mutations" wrong, since it does not apply to all or even most of populations, which are sexual (and then recombine).

Stephen Matheson said...

Anonymous, you're right that clonal interference -- strictly defined -- only occurs in asexual populations. But the same process occurs in sexually-reproducing organisms, traveling under the name of the Hill-Robertson effect. You can't call it clonal interference, because it's not clonal, but some authors do call it interference, and it is otherwise defined in the same way. So in fact, my point is correct, and merely states what is widely known in population genetics regarding the fates of mildly beneficial alleles during adaptation. And as I noted, recombination in sexual populations is a major solution to what would otherwise be a major impediment to adaptation.