More Recent Comments

Wednesday, December 09, 2015

Why doesn't natural selection reduce the mutation rate to zero?

All living organisms have developed highly accurate DNA replication complexes and sophisticated mechanisms for repairing DNA damage. The combination results in DNA replication errors that are about 1 per 10 billion base pairs (10-10 per bp). DNA damage due to other factors is effectively repaired with an error rate of 1 in 100 per base pair (10-2 per bp).

Mutations can be beneficial, deleterious, or neutral. In organisms with large genomes there are many more neutral mutations than the other two classes but in organisms with smaller genomes a higher percentage of mutations are either beneficial or deleterious. In all cases, there are more deleterious mutations than beneficial ones.

If deleterious mutations are harmful to the individual then natural selection should favor a low mutation rate in order to minimize that effect. This is especially true in large multicellular organisms where somatic cell mutations cause cancer and other problems. It seems logical that the optimal mutation rate should be zero in order to maximize the survival of the individual and its offspring.

Nothing in biology makes sense except in the light of population genetics.

Michael Lynch
But even though the number of beneficial mutations is low compared to those that are deleterious, this is the stuff of adaptive evolution. In the long run the population will become more fit if beneficial mutations occur and become fixed by natural selection. Eliminating mutations might provide a short-term advantage but eventually the population will go extinct if it can't adapt to new environments. (Neutral and deleterious mutations can also contribute to adaptation over the long term.)

The simplest explanation for this apparent paradox is that there's a trade-off between selection to minimize deleterious mutations and selection for long-term evolutionary advantage. The problem with that explanation is that it is very difficult to show how you can select for the future benefit of mutations to the species (population). It seems as though you have to invoke two bogeymen; group selection and teleology.

Maybe there's a better explanation?

Jerry Coyne recently thought about this problem and posted his analysis under the provocative title: The irony of natural selection. He concludes that there's some constraint that limits the ability of natural selection to achieve a zero mutation rate.
The most probable explanation is that evolution does not produce perfect adaptations. In the case of mutations, though natural selection favors individuals most able to repair any changes in DNA (although a small percentage of these might be adaptive), this level of perfection cannot be achieved because of constraints: the cost of achieving perfection, the fact that all errors are impossible to detect or remove, or that some cells (i.e., sperm or eggs) may not even have DNA-repair mechanisms because of genetic or physiological constraints.
I used to think that this was the best explanation. I taught my students that the accuracy of DNA replication, for example, comes at the cost of speed. The more accurate the polymerization process, the slower it takes. This makes a lot of sense and there's experimental support for the claim. Slowing down the time it takes to replicate the genome will affect the time it takes for cell divisions and that could be harmful ... or so the argument goes.

Unfortunately, I ran into Michael Lynch at an evolution meeting and he quickly destroyed that argument. There's no evidence that the speed of DNA replication is limiting the rate of cell divisions and, besides, there are easy ways for selection to get around such a limitation if it ever occurred. (This is a photo of Michael Lynch looking at me right after setting me straight. He's wondering how I could have been so stupid.)

When you think about it, there doesn't seem to be any biochemical or physiological constraints that could prevent the mutation rate from getting to zero ... or at least a lot closer than it is now.

Michael Lynch has a better answer and he explains it in a paper titled: "The Lower Bound to the Evolution of Mutation Rates" (Lynch, 2011).
As the mutation rate is driven to lower and lower levels by selection, a point must eventually be reached where the advantage of any further increase in replication fidelity is smaller than the power of random genetic drift (Lynch 2008, 2010). The goal here is to evaluate the extent to which such an intrinsic barrier can provide an adequate explanation for the patterns of mutation rates known to have evolved in natural populations.
The main "constraint" is the limited power of natural selection in the presence of random genetic drift. This will depend to some extent of the size of the population.

This idea is called the "drift-barrier hypothesis. It is described in Sung et al. (2012):
... the drift-barrier hypothesis predicts that the level of refinement of molecular attributes, including DNA replication fidelity and repair, that can be accomplished by natural selection will be negatively correlated with the effective population size (Ne) of a species. Under this hypothesis, as natural selection pushes a trait toward perfection, further improvements are expected to have diminishing fitness advantages. Once the point is reached beyond which the effects of subsequent beneficial mutations are unlikely to be large enough to overcome the power of random genetic drift, adaptive progress is expected to come to a standstill. Because selection is generally expected to favor lower mutation rates as a result of the associated load of deleterious mutations, and because the power of drift is inversely proportional to Ne, lower mutation rates are expected in species with larger Ne.
The Lynch lab has produced lots of evidence in support of the hypothesis although there may be some confounding factors in some populations.

The bottom line is that the real irony of natural selection is that it's just not powerful enough to reduce the error rate of replication and repair below the values we currently see.

In a sense, it's the "error rate" of fixation by natural selection in the face of random genetic drift that allows evolution to occur.

The more we learn about biology the more we learn that it's messy and sloppy at every level. Evolution is not a watchmaker and it's not even a blind watchmaker. It's a tinkerer1 and the "watch" barely keeps time.


Image Credit: The Mendel's traits image is from Wikispaces Classroom.

1. Jacob, F. (1977) Evolution and tinkering. Science (New York, NY), 196:1161. [PDF]

Lynch, M. (2011) The lower bound to the evolution of mutation rates. Genome Biology and Evolution, 3:1107. [doi: 10.1093/gbe/evr066]

Sung, W., Ackerman, M.S., Miller, S.F., Doak, T.G., and Lynch, M. (2012) Drift-barrier hypothesis and mutation-rate evolution. Proc. Natl. Acad. Sci. (USA) 109:18488-18492. [doi: 10.1073/pnas.1216223109]

56 comments :

Stephen Matheson said...

It seems to me that the drift-barrier reasoning would apply to any trait, and he seems to say as much in the 2012 abstract. This means that the drift barrier should limit the achievement of "perfection" across the board, and maybe it does. But isn't it reasonable to consider the possibility that a zero mutation rate is -- at least in some populations some of the time -- evolutionary suicide, and therefore subject to selection at that level? I haven't read Coyne's piece, and read the Lynch paper a long time ago, so maybe they both dealt with that.

Petrushka said...

All kinds of optimizations seem to lead to extinction when environments change.

anonymous said...

I find it ironic (no that that is not the word I am looking for) rather I find it incongruous that panegyrics lauding Gould’s insights on Selection in the previous thread simultaneously ignore some of his greatest insights.

Selection encompasses a biological hierarchy ranging from genes and organisms to entire species. The notion of optimal mutation rate could be considered from the perspective of “species selection”, a notion that has been derisively dismissed in previous threads.

Rather than playing cat and mouse by allowing some present to paint themselves into predictable corners, I will simply take my leave after dropping off this “Christmas present”.

http://gbe.oxfordjournals.org/content/early/2015/08/06/gbe.evv152.full.pdf

Mikkel Rumraket Rasmussen said...

I used to think the solution to this problem was simply that it was unlikely at any point in time, for all extant species to simultaneously evolve a "perfect" replication system.

As such, even though some species might at some point evolve perfect replication, and subsequently go extinct, there'd invariably always have been multiple speciation events before fixation of the "perfect" replication alleles, so a lot of cousin lineages were still around without "perfect" replication systems. And so on ad infinitum.

In retrospect given the above, I guess my rationalization only makes sense if selection was overpowered by drift, since otherwise you'd still expect perfect replication to fix in all lineages eventually.

Steve said...

Extinctions of species has no impact on life.

Call me when a class of organisms goes extinct.

Then we can pop that cyanide cap over a beer.

The Other Jim said...

All kinds of optimizations seem to lead to extinction when environments change.

But if this was the mechanism, would you not expect to detect the occasional living species that went very close to zero (or much higher), if true? The would be the ones that went on this path, but have not yet faced the "environment change".

The universality of an approximately similar mutation rate would argue that something is keeping it there, not just extinction of those who move too far from the "comfort zone".

judmarc said...

Rather than playing cat and mouse by allowing some present to paint themselves into predictable corners, I will simply take my leave after dropping off this “Christmas present”.

http://gbe.oxfordjournals.org/content/early/2015/08/06/gbe.evv152.full.pdf


In general, I like the paper and its emphasis that we are not dealing with teleology. The idea that promotion of species-level selection via the presence of TEs is in principle testable is interesting.

Fair Witness said...

Steve said "Extinctions of species has no impact on life."

This is going in my list of favorite creationist quotes.

Stephen Matheson said...

Larry Loeb's group did an experiment several years ago looking at how replication accuracy affects fitness. I haven't checked to see how/if Lynch discusses the results, and I'm sure that other papers have looked at this, too. But the findings in that paper suggest that (at least in bacteria competing in laboratory conditions) very high fidelity can lead to lower fitness. Interested in what others think. Paper here: http://depts.washington.edu/loeblabs/pdf/2010_Loh.pdf

Anonymous said...

There are sulfur bacteria that live in the anoxic sediments in the deep ocean - an environment that's barely changed ( if at all) in several billion years. I used to think they'd have the most accurate polymerases but I guess not.
It seems to me that the relevance of population size to population genetic models for bacteria only apply to the extent that bacteria are able to exchange DNA with each other, either through transformation or transduction.

Christian said...

Is the error rate of DNA replication tied to the generation time of a species? In humans for instance, maybe our 20-25 year generation time doesn’t provide enough selective pressure to get the error rate closer to 0. Or, in other words, the error rate is good enough to get us to this reproductive age before succumbing to disease? What if we didn’t reproduce until we were 1000 years old - presumably getting to this point would require pushing the error rate closer to 0?

Larry Moran said...

The complete reference is:

Brunet, T. D., and Doolittle, W. F. (2015) Multilevel selection theory and the evolutionary functions of transposable elements. Genome Biology and Evolution 7:2445-2457. [doi: 10.1093/gbe/evv152]

The authors argue that transposons (TE's) MAY affect the survival or extinction of species. If so, this might be an example of species selection (species sorting is a better term). TE's may have a "function" related to the survival of the species.

According to Brunet & Doolittle, it's better to invoke species sorting as a "higher level" process than to promote teleological explanations for the presence of TE's.

There's no evidence that a genome full of TE junk made our ancestors more successful than their close primate cousins who had less TE junk.

If you are looking for adaptationist ways of explaining a genome full of defective TE's then it may be more legitimate to make up stories about species selection rather than silly teleology but the bottom line doesn't change. Defective TE's are still junk.

Gary Gaulin said...

"Multilevel selection theory"

LOL!

https://sites.google.com/site/intelligencedesignlab/home/Origin3Cau3Gen1600.png

Petrushka said...

Is it relevant that all kinds of living things are thriving around Chernobyl, with elevated mutation rates?

whimple said...

Look at some really successful viruses, like the RNA based viruses HIV and hepatitis C. These are under enormous selective pressure and generally operate close to the ragged edge of as mutable as possible, but not so mutable that the population crashes from accumulation of deleterious mutations - sometimes you can push them over the edge therapeutically as with ribavirin. The notion that a zero mutation rate represents perfection seems to be a human perceived construction - as adaptable as possible while yet remaining within parameters of long-term viability seems a more reasonable strategy.

The Other Jim said...

Is there actually real evidence of increased mutation rates?

Jmac said...

Natural selection is kind of smart and very, very selective. It ALWAYS leaves enough for the Darwinists to work with but not in all species just to keep them guessing and come up with the ridiculous ideas to cover their story.

DAK said...

2015 lecture on YouTube, where Lynch discusses the drift barrier hypothesis.

Gary Gaulin said...

The new "Multilevel selection theory" seems to be an example of what you are describing. Its illustration does not even sort out the different kinds of speciation: Molecular (chemical), Genetic, Unicellular and Multicellular.

http://gbe.oxfordjournals.org/content/7/8/2445/F1.expansion.html

If this keeps up then there will soon be a scientifically useless "Theory of (Un)Intelligent Selection" being applauded as having defeated ID for good.

Petrushka said...

That seems to be a controversial topic.

Gary Gaulin said...

Tages Haruspex: I find it ironic (no that that is not the word I am looking for) rather I find it incongruous that panegyrics lauding Gould’s insights on Selection in the previous thread simultaneously ignore some of his greatest insights.

Selection encompasses a biological hierarchy ranging from genes and organisms to entire species. The notion of optimal mutation rate could be considered from the perspective of “species selection”, a notion that has been derisively dismissed in previous threads.


I have to add: that I agree there is a biological hierarchy ranging from genes and organisms. But where the goal has to fudge "selection" into the picture even the best of attempts are destined to lead to a confusing arrows that only makes it harder for everyone to conceptualize how unintelligent molecular (chemical species) relates to self-learning genetic species and other things.

Tages Haruspex: Rather than playing cat and mouse by allowing some present to paint themselves into predictable corners, I will simply take my leave after dropping off this “Christmas present”.

I love the way the authors at least got as far as they did with that thought. But what resulted does not sort out all in biology into a programmable model that gets rid of an ambiguous generalization that can apply to anything influencing something else. The only way around that problem is take "natural selection" completely out of both the model and theory but then it's not Darwinian anymore, and "evo" from from "devo" must be operationally defined all over again, in which case thousands of hours and millions of dollars later the need for what has already been around and being experimented with for at least half a decade will finally be "discovered by scientists".

The "Christmas present" indicates some progress along that rather self-punishing path has been made. So thanks from me!" It's just that the obligatory variable is destined to cause a problem that is only solved by doing the almost unthinkable to the "evolution by natural selection" paradigm by throwing it out of serious models for the origin of species. Complex virtual environments where some of the critters end up getting isolated from others can be pointed at as being "selection" of one kind or another but that's an outside view generalization, not a variable in the mechanism causing their speciation.

Unknown said...

@Larry: "species sorting is a better term"
In this case it's not. For a trait to be under selection it has to be heritable and differences in the trait must lead to differential survival and/or fecundity. Sorting is more general, because it relaxes the condition that the trait must be heritable - there is a higher extinction rate for highly endemic species, but if a species is highly endemic it is not clear whether potential daughter species would be as well. Since we can't tell whether the trait is heritable we would say there's sorting. But rather obviously TEs are heritable, so the chance that there is sorting but not selection is nil.

The whole truth said...

I just want to say that the text here is either too big or too small now.

Petrushka said...

It's just right for me, and it's sleeping in my bed.

Petrushka said...

It's just right for me, and it's sleeping in my bed.

Larry Moran said...

Are you referring to the text in the comments? Is it too small?

Larry Moran said...

Some species might exhibit more speciation (or less extinction) than others. They will contain heritable traits that may or may not contribute to their success with respect to other species. We are interested in knowing whether species level effects are playing a role.

The effect we see is species sorting. The mechanism could be due to selection or the species-level equivalent of random genetic drift. You should not refer to the effect as "species selection" until you've got evidence that the species level phenomenon is actually due to selection and not drift.

It's the species-level version of the adaptationist fallacy where every heritable change is incorrectly assumed to be due to selection.

judmarc said...

If you are looking for adaptationist ways of explaining a genome full of defective TE's then it may be more legitimate to make up stories about species selection rather than silly teleology but the bottom line doesn't change. Defective TE's are still junk.

I'm not even sure I'd characterize a fair amount of the argument in the paper as adaptationist. Much of the discussion seems to be about the possibility that a genome full of TEs would more readily undergo changes that would prevent interbreeding and thus result in a larger number of "offspring species."

Unknown said...

@Larry: The species level equivalent to drift is species drift. It is emphatically not sorting.
In both species drift and species selection you have heritable traits. In both species selection and non-selective sorting you have differential expected survival or fecundity.
To put that into a simple graphic form: http://post-neo.com/evo/sorting.png

Larry Moran said...

@Simon Gunkel,

I see the problem. You are referring to a different kind of "species sorting."

I'm referring to the one that is relevant to a discussion about species selection and hierarchical theory. See:

Vrba, E.S., and Gould, S.J. (1986) The hierarchical expansion of sorting and selection: sorting and selection cannot be equated. Paleobiology, 217-228.

In a nonhierarchical world, where selection on organisms regulated all nonrandom evolutionary change, the traditional equation of selection (a cause of sorting) with sorting itself (differential birth and death among varying organisms within a population) would rarely lead to error, even though the phenomena are logically distinct (for sorting is a simple description of differential "success," and selection a causal process). But in a hierarchical world, with entities acting as evolutionary individuals (genes, organisms, and species among them) at several levels of ascending inclusion, sorting among entities at one level has a great range of potential causes. Direct selection upon entities themselves is but one possibility among many. This paper discusses why hierarchy demands that sorting and selection be disentangled. It then presents and illustrates an expanded taxonomy of sorting for a hierarchical world. For each of three levels (genes, organisms, and species), we show how sorting can arise from selection at the focal level itself, and as a consequence either of downward causation from processes acting on individuals at higher levels or upward causation from lower levels. We then discuss how hierarchy might illuminate a range of evolutionary questions based on both the logical structure of hierarchy and the historical pathways of its construction-for hierarchy is a property of nature, not only a conceptual scheme for organization.

Gould also discusses this problem in The Structure of Evolutionary Theory. In that book he describes a higher level version of random genetic drift that can lead to species sorting.

anonymous said...

Professor Moran’s contention regarding “species sorting” is well taken.

However, Professor Moran also states: “Defective TE's are still junk.“

Hmmm... as long as we are quoting Gould & Vrba, I would suggest we follow Tyler D.P. Brunet’s and W. Ford Doolittle’s advice and revisit the great classic:

Exaptation-A Missing Term in the Science of Form - by Stephen Jay Gould and Elisabeth S. Vrba

The whole truth said...

At my usual browser magnification setting (125%) the text of your articles and on the left side of the page is now much larger than it was before, so I decreased the magnification to 110% but then in the comments the part below the person's username, date, and time is very small (I'm having a hard time reading it). I'm using Google Chrome.

DAK said...

I'm on 125%. Maybe I need glasses? Text is certainly small in the comments.

Anonymous said...

Text of comments works wel for me, and is a little larger than the comment headers. I'm not sure of the magnification but it's a little stronger than "normal." Using Firefox.

Larry Moran said...

You can only call something an exaptation after the fact. It's true that some parts of what is now junk DNA could evolve a function but we don't know which parts, maybe none. Until then, they're still junk. They have no function and they can be removed without harming the organism.

I discussed the Gould & Vrba paper in my post on Selfish genes and transposons.

Jmac said...

Interesting. Are there more examples of papers that you know about Gary? Or is this fairly new?

judmarc said...

It's true that some parts of what is now junk DNA could evolve a function but we don't know which parts, maybe none. Until then, they're still junk. They have no function and they can be removed without harming the organism.

That's pretty much what I understood the Brunet-Doolittle paper to say about the matter.

Unknown said...

TBH I think that a lot of the work on hierarchial theory has created unnecessary complications. Gould thought that the original version of species selection by Stanley had to be amended in some way, and there's a lot of discussion on emergent traits vs emergent fitness and the like.
I'm also not a big fan of sorting in the sense of V&G, because it is synonymous with what Darwin called selection, or what in modern parlance would be selection and drift combined - resampling in short.
On the other hand effects that bias a population in terms of non-heritable traits needs a term and sorting works well as a descriptor there.

AllanMiller said...

One factor I don't see mentioned much is that the process of refining mutation rate is itself dependent on mutation. The greater the fidelity of replication, the more slowly fitter variants will be produced, even if they are there to be located. Refinement grinds to a halt due to the consequences of refinement, not simply through drift or diminishing selective returns.

Gary Gaulin said...

There was first this hoopla from a 2012 event that became ID news:
http://www.uncommondescent.com/intelligent-design/video-the-dennis-noble-lecture-in-suzhou-china-on-physiology-and-neo-darwinian-evolutionary-biology/
(currently working) British Biologist Denis Noble Debunks Neo Darwinism
https://www.youtube.com/watch?v=QMVfafAYTMg

Denis explained some of the weaknesses of current evolutionary theory that are solved by the systems biology model (and included ID theory) I had online at Planet Source Code since 2011. I did not know whether he knew about the theory I have been developing, what mattered is that systems biology was none the less going in that direction.

This year has been the most exciting of them all. The best was a paper published in August by the Royal Society titled "The extended evolutionary synthesis: its structure, assumptions and predictions" that resulted in my writing code now in the ID Lab #5 to similarly illustrate what the multi-level model looks like:

http://www.antievolution.org/cgi-bin/ikonboard/ikonboard.cgi?act=ST;f=14;t=7420;st=14970#entry246562

Now I learned of the paper published in August by the Oxford Journals titled "Multilevel Selection Theory" that likewise (even where true for a generalization like "Natural Selection") is not useful to someone who has to computer model all in biology. Generalizations do not work where the experimenter is forced to in-silico demonstrate complex and sometimes intelligent behaviors.

The "evolution by natural selection" way of thinking only works as an outside-in view of the molecular based cause. Since it only sees in one direction it completely fails as theory for the inside-out view that modern biology now has to electronically model from all the molecular data now being generated. The situation is causing a scientific revolution, where the future is indeed ID. And even though what the Theory of Intelligent Design became is not what was at first imagined would be possible it's actually way better, though it may take some time for others to understand why.

Larry Moran said...

judmarc says,

That's pretty much what I understood the Brunet-Doolittle paper to say about the matter.

The main point of the paper seems to be that TE's may contribute to the evolvability of a species. If they do, then it's better to think of this in terms of species selection than in teleological terms.

In either case, the authors seem to imply that the presence of TEs (defective or active) has some advantage to the species, otherwise why bother writing the article?

If they do, indeed, confer some direct selective advantage for the species, do they still count as junk? I say, no. If that's true then they have a selected function.

This is not the same as saying that some of the sequences may accidentally be co-opted for a function (exaptation). That's not selected at any level.

AllanMiller said...

... also, the process of mutation refinement should have a depressive effect on the rate of divergence, and hence of speciation. Lineages that speciate less will go extinct more often, for stochastic reasons alone. Over time, variation at species level would appear to be biased (by something other than this effect) in favour of the more relaxed mutation rate, when it was simply a consequence of mutation rate's effect on cladogenesis and consequent patterns of preservation.

John Harshman said...

Lineages that speciate less will go extinct more often, for stochastic reasons alone.

...all things being equal. But they aren't. This assumes that species are interchangeable units. It seems to me, however, that range size is a factor in extinction probability, and that more frequent speciation should be correlated with smaller range. And this is not the only factor that would make species non-interchangeable.

AllanMiller said...

@John

more frequent speciation should be correlated with smaller range.

For each individual species. But from the perspective of their common ancestor, its multi-species descendant line would be (all else being equal!) at least as wide-ranging as a non-speciating line in toto. And, these species are likely to cover more niches, giving variation at species level, a hedge against loss.

So comparing Lineage A (zero mutation; no speciation) with Lineage B (appreciable mutation; speciation), I would still expect to find more Lineage B species than Lineage A. And hence, a continuous bias across species towards appreciable mutation, from this cause alone, given its (potential) effect on species-level variation and a kind of species-level 'offspring number'.

judmarc said...

In either case, the authors seem to imply that the presence of TEs (defective or active) has some advantage to the species, otherwise why bother writing the article?

If they do, indeed, confer some direct selective advantage for the species, do they still count as junk? I say, no. If that's true then they have a selected function.


I thought at first those must be the lines along which the authors were thinking. But upon going back and re-reading, what I concluded was, as I noted above:

Much of the discussion seems to be about the possibility that a genome full of TEs would more readily undergo changes that would prevent interbreeding and thus result in a larger number of "offspring species."

So perhaps the gist is that a larger number of offspring species may be at the interspecies level what a larger number of offspring is at the intraspecies level. I also think the authors are saying this thesis should in principle be testable.

But in any case: Is the susceptibility of a genome to undergoing changes that prevent interbreeding "adaptive" to the extent we ought not to call it junk on the intraspecies level? I wouldn't think so, but am ready to hear where I'm wrong.

judmarc said...

I have no idea if there is any practical way to test the proposition that lineages which speciate less will go extinct more often, which I think is directly raised by the Brunet-Doolittle article.

Some historical examples are interesting to think about. How close did modern humans come to going extinct? There were apparently many species of humans within the past few million years, and they all went extinct other than homo sapiens. How close did we come to going extinct as well? Would it have been a less narrow thing if there had been one or two dominant human species with greater numbers and range?

AllanMiller said...

Probably pretty close. But I might (pursuing my theme) argue that the production of multiple species from the LCA immediately after the Human-Chimp split may have provided it with a hedge against total loss anyway. It was harder for the forces of extinction to eliminate all of its descendants, and we just feel privileged to be the fortunate remainder. (This ignores the possible effect of H. sapiens itself on diversity in its clade, of course).

judmarc said...

It was harder for the forces of extinction to eliminate all of its descendants, and we just feel privileged to be the fortunate remainder. (This ignores the possible effect of H. sapiens itself on diversity in its clade, of course).

So if we don't ignore that, do we wind up with a big difference in ultimate outcome (in terms of number of descendant individuals) between competition among closely related species and competition among closely related individuals?

John Harshman said...

Alan,

You may be right. I wasn't trying to make a rigorous statement about outcomes, just pointing out that you can't naively assume that more species is more better and that counting species is the same as counting evolutionary success. At the very least, you have to validate that assumption. At the most, that assumption isn't true.

AllanMiller said...

[...] just pointing out that you can't naively assume that more species is more better and that counting species is the same as counting evolutionary success

No, I'm counting 'evolutionary success' as the continued existence of the clade, by virtue of at least one species remaining at the point where we sample mutation rates. I don't think it 'better', but may make it more likely for us to come along and pick a species at random with a nonzero mutation rate, even if the starting conditions had an equal number of 'zeros' and 'non-zeros'. It is naive, because of course there are nonzero rates that would be worse than zero. But surviving species will be impoverished in these, too.

With a 'just right' level of mutation, they would both speciate more and last longer (on the average) than zero-rate cases. The species would be expected to be more diverse between and varied within. So sampling species would be susceptible to a systematic bias away from zero-mutation cases.

I have no idea how to validate these assumptions!

anonymous said...

@ Allan Miller
Very Impressive! Bravo! You nailed it.

There still remains a nagging after-thought:

Substitute your conjectures on "mutations rates" and substitute instead "TEs" (and Brunet-Doolittle’s assessment of their impact at what Gould already deemed "a higher hierarchical level")

They propose that the presence of TEs (defective or active) has some advantage at a higher hierarchical level of selection.

If such direct selective advantage at a higher hierarchical level indeed accrues, do TEs still count as "junk"?

Not anymore. We need to expand our vocabulary on that score.

Let us consider this direct selective advantage of TEs at some higher hierarchical level? In your assessment, how would that work exactly? Perhaps by facilitating your suggestion of a 'just right' level of mutation? Let's try a different angle: How about by providing a greater potential repertoire of exaptation?

Athel Cornish-Bowden said...

Maybe I missing something important here, but it seems to me that both the question in your title

Why doesn't natural selection reduce the mutation rate to zero?

and this proposed comment on it:

When you think about it, there doesn't seem to be any biochemical or physiological constraints that could prevent the mutation rate from getting to zero ... or at least a lot closer than it is now.

very weird, because when _I_ think about it I find that there is an overwhelming biochemical constraint that prevents it from happening. There may be subsidiary reasons along the lines of your discussion, but the primary reason is that it doesn’t reduce it to zero because it can’t.

The numbers of groups involved in base-pairing are not infinite, and the thermodynamic differences between correct and incorrect pairs are not infinite either. Errors are inevitable, and must arise not only in the replication of DNA but in every stage from DNA to protein. Their incidence can be decreased by proof-reading, etc., but not to zero. However, all the error-correcting mechanisms consume ATP, and there is bound to be a point at which economics ensures that adding another round of checking for errors will cost more (in time and ATP) than it saves.

Larry Moran said...

There are antimutator mutations that reduce the error rate so we know it's possible to improve the error rate of DNA replication and the error rate of repair.

I agree that getting to zero is impossible in practice but I really wasn't interested in discussing a zero mutation rate, only one that was better that the current one.

The error rate of DNA repair seems to be a lot higher than any reasonable biochemical limits. This is why population genetics provides the correct answer to the question.

AllanMiller said...

Not anymore. We need to expand our vocabulary on that score.

Let us consider this direct selective advantage of TEs at some higher hierarchical level? In your assessment, how would that work exactly? Perhaps by facilitating your suggestion of a 'just right' level of mutation? Let's try a different angle: How about by providing a greater potential repertoire of exaptation?


I think you are right about the issue of 'vocabulary' - I am a bit lukewarm on the idea of differential persistence of species being termed 'selection' at another level.

I think an essential component of selection is some kind of ecological arena in which differential reproductive rates matter. That is, subpopulations under selection compete fairly closely, differentially. We could literally compare the differential reproduction of apples and oranges, but don't. Where there has been speciation, some species may have some internal factor that leads to greater rates of cladogenesis, but if (as is generally the case) these species are ecologically and geographically distinct, I dont see that as 'selection' in the same way as factors affecting the intra-population contest for continued representation.

Of course, this is maybe a mere semantic nicety. But when we consider whether TEs offer some 'advantage' to a clade that has them by increasing cladogenesis or flexibility, I find it hard to view this as adaptive in the same way that resource ultilisation is (I have the same problem with many treatments of sex).

But as to the direct point, I think the dynamics of TEs demand separate consideration from mutation. They are mutagens of a sort, but they also 'infest' genomes. Their effect on within-population fitness can be pretty dramatic. And although in the mutational scenario I assumed that zero mutation would inhibit speciation, when we add TEs to the mix, you'd need zero mutations and zero TEs to do the same. Given mutation + TEs, I don't think you necessarily get more speciation than under mutation alone.

I guess the fundamental question is: is there any bias in the distribution of TEs that is due to their effect on cladogenesis or evolutionary flexibility and not purely due to their mode of transmission? It's certainly possible, but I don't know how one would distinguish the two.

anonymous said...

@ Allan Miller

In other words, you are not in complete accord with all of Brunet-Doolittle’s conjectures.

Frank said...

My short answer: it can't. Shannon's channel capacity reads
C = W * log(2) (P/N + 1), and it is impossible to transmit a signal (genome information in this case) at a rate greater than C. C is in bits per second, W is frequency (seconds^-1). As Signal to Noise (P/N) gets larger, the frequency W must increase, or C becomes smaller and smaller. At some point, however, W is limited by, e.g., instability of the DNA molecule itself, or diffusion of monomers.

Conclusion: there will always be error in the system. Thermo is hell, and not just in P. Chem class.