…and this is a 
This talk has me a little concerned: it's proposing something rather radical, for which Arber is going to have to show me some unambiguous evidence to convince me, and I'm coming into it with a very skeptical mindset. Here's the relevant portion of his abstract:
The theory of molecular evolution that we also call "Molecular Darwinism" is based on the acquired knowledge on genetic variation. In genetic variation, products of evolution genes are involved as variation generators and/or as modulators of the rates of genetic variation. These evolution gene products act together with several non-genetic elements that can be assigned to intrinsic properties of matter, to environmental mutagens and to random encounter. We conclude that natural reality takes actively care of biological evolution. The evolution genes must have been fine-tuned for their functions by second-order selection, so that spontaneous genetic variation with different evolutionary qualities occurs at quite low rates. This ensures a relatively high genetic stability to individuals, as well as an evolutionary progress at the level of populations.
The presence of evolution genes points to a duality of the genome: while many genes act to the benefit of the individuals for the fulfillment of their lives, the evolution genes act to the benefit of an evolutionary development, for a slow, but steady expansion of life and biodiversity.
You see the problem, I hope. These hypothetical genes that do not necessarily directly affect the fitness of the individual are assumed to be promoted in lineages by a higher level of selection. This is not easily supported by evolutionary theory: there isn't a mechanism given for individuals to maintain a gene that will only help its many-times-great-grandchildren. It is inferring a kind of foresight to evolution that is doesn't have a mechanism…unless, perhaps, Arber is going to give use one. We'll see. This talk will start in about 15 minutes, and I'll update this post as he fills us in.
A simple history lesson: modern evolutionary biology is the convergence of work that began with Miescher (1874: nucleic acids) which led to molecular biology, Mendel (1876) which led to genetics, and Darwin (1859) that approached the problem at the level of organisms and species. The neo-Darwinian synthesis fused the genetic and Darwinian line, molecular genetics brought together genetics and biochemistry/molecular biology, and molecular evolution brings all three together—he seems to claim some kind of intellectual ownership of the last concept, which is what he calls molecular darwinism.
How do bacteria generate new variants? By transformation, conjugation, or transduction. All are mechanisms that transfer genes from an external source to the bacterium. Work in the 1940s demonstrated that DNA was the carrier of genetic information.
Arber gave a little summary of E. coli gene structure, which I suppose would be helpful to all the chemists here. He defines mutation as an alteration of the nucleotide sequence; in classical genetics, it's defined differently, as an altered phenotype that is transmitted to progeny.
Mutations are rarely favorable; often unfavorable, and very often silent or neutral. There is no good evidence for directedness of spontaneous mutations. Mutations do not appear in response to a need.
He argues that there are three elements to evolution: evolution is driven by genetic variation (mutation), directed by natural selection, and modulated by isolation as a mechanism for speciation. There are multiple mechanisms generating genetic variation: spontaneous DNA sequence alteration, DNA rearrangement or recombination, and DNA acquisition (horizontal gene transfer).
So far, this is all very unchallenging and basic, at least for someone with any background in genetics and cell biology. After sitting through one talk that completely lost me with a failure to explain the basic terms of the work, I can't complain, but I confess, I'm having trouble staying alert through all this.
Some genes can affect the rate of occurence of mutations — these are modulators of the frequency of genetic variation. He calls these evolution genes. He says neatral reality actively takes care of biological evolution, and that this is an expansion of the biological theory of evolution. This leads to an expansion of biological diversity, and, he argues, higher complexity.
I'm not very impressed. This is a combination of the commonplace and some odd interpretations. Of course there is variation in fidelity of replication that is influenced by genetic variation. Some of it is simply thermodynamically necessary: perfect fidelity is impossible to achieve, and greater fidelity has a metabolic cost, so some of that variation is utterly unsurprising. Some is; when we have organisms that have specializations to directly generate greater genetic variation — and sex is the first to come to my mind — we have a problem to explain. I don't see that Arber has proposed anything to explain the real problems.
At the same time, what Arber said here does not make him a friend to intelligent design creationism, or creationism of any kind, despite the claims of some unreliable creationist sources, a claim that Arber has directly rejected.
I'd have to say it was a nice enough overview, but didn't really propose anything novel, and definitely didn't demonstrate anything that can't be explained in the context of modern evolutionary theory.
Life Science
Comments
Posted by: Coturnix
|
July 2, 2009 6:00 AM
For this one, I will descend from the heavens and actually watch it in the hall itself. Be there in a minute...
Posted by: jsclary | July 2, 2009 6:02 AM
Ugh, the guy needs an editor or a grammar school nun rapping him on the knuckles. Also, what a horrible choice of terminology... "evolution genes"?
Posted by: Kobra | July 2, 2009 6:06 AM
Please not let it be bunk.
Posted by: GPNguyen | July 2, 2009 6:10 AM
Since when is "evolution" an adjective?
Posted by: cdx | July 2, 2009 6:11 AM
About 90% probability of bogosity, ime....
Posted by: GPNguyen | July 2, 2009 6:18 AM
Also, "takes actively care" ? The sloppy writing makes me think of a heightened heart rate, adrenaline nervously pumping through shaking hands, as if he couldn't wait to share his revolutionary revelation with the world, not the calm, calculating, thought process that usually goes into a paper. It is as if he couldn't be bothered to get somebody to edit his work, and it hints at unprofessionalism.
Posted by: Corneel | July 2, 2009 6:19 AM
I am curious about the non-evolution genes. They are the ones labeled "Do not mutate", right?
Posted by: Mozglubov | July 2, 2009 6:26 AM
I am glad other people are concerned about the grammar as well... An abstract put together poorly does not bode well for a paper or talk.
Posted by: mikespeir | July 2, 2009 6:30 AM
I think "Molecular Darwinism" is an unfortunate term, mainly because I don't particularly care for the term "Darwinism."
Posted by: Michael Kingsford Gray | July 2, 2009 6:31 AM
The phraseology rings loudly to me as inter-language translational problems, and little more.
Posted by: Shane | July 2, 2009 6:36 AM
Er... he's arguing on the one hand that there is selection for very very good DNA repair mechanisms (plausible), but that they need to be sloppy enough to let variation accumulate as the seedcorn for selection?
That strikes me as a bit squiffy. Firstly, there is no foresight in evolution. Secondly, is there any evidence that nucleic acid repair mechanisms need any help to be *less* efficient? Thirdly, is there any evidence that DNA repair mechanisms could, in theory, be *more* efficient?
What he would need to do would be to show that the population mutation rate is a limiting factor in evolution, but that really does not appear to be the case. NS does not work (that much) on the mutations as they arise in new individuals, but on mutations that happen to have accumulated in the population at large. You could probably decrease the new mutation rate by a large factor (theoretically), and the population ability to respond phenotypically to a new selective stress would probably not change that much (anyone simulated this?)
In other words, I think our NA repair is probably as good as it *can* get, and that the gap between "good" and "perfect" is not itself a phenotype that gets selected at all. At any level. Rather, selection will always prefer the best-possible genome repair mechanisms.
Could be wrong - spill dem beanz, PZ! :-)
Posted by: MadScientist | July 2, 2009 6:36 AM
My BS alarms are all set off by words like:
"The evolution genes must have been fine-tuned for their functions by second-order selection, so that spontaneous genetic variation with different evolutionary qualities occurs at quite low rates."
(1) Show me the genes! What are these 'police' genes being proposed and how are they postulated to work?
(2) the second half about spontaneous genetic variation etc particularly annoys me. If the copying mechanism were not extremely good, life simply could not exist because cells would not be replaced with good enough copies and the odds of reproduction would be very low. The proposition seems to be that police genes control the mechanism for copying and remove/destroy most mistakes and that these police genes have evolved to be highly specialized. Even if he is only talking about progeny rather than cellular reproduction the claims seem weird and the police genes seem unnecessary.
The style of writing reminds me more of managers (and theologians) than scientists. Is this the biologist's version of Cold Fusion? I would be extremely suspicious of his data (if he ever has any to present) and scrutinize it well.
Posted by: Coturnix
|
July 2, 2009 6:55 AM
Nobelists like to stray away from their specialty. Arber should stick to chemistry.
This was an undergrad talk, introducing the most basic basics of evolution, oversimplified, occasionally wrong.
What he is doing is "inventing" terms. But those terms already exist and have been used for decades to denote something different. "Acquired" has a meaning in biology - but not the one he is using. "evolution genes" have nothing to do with "genes for evolution". And, please do not use "Darwinism". And no, your work that you got your Nobel for is NOT central piece in the evolutionary theory no matter how hard you are trying to make the case for it, sorry.
Posted by: Martin_z | July 2, 2009 7:01 AM
Just a comment to you grammar pedants - Werner Arber is Swiss and his first language is German, I think - it's definitely not English.
If any of you can write German as well as he can write English, you have the right to criticize. Otherwise, stick to criticizing the science.
Posted by: Rorschach | July 2, 2009 7:11 AM
That's unfortunately not correct.
Last time I checked,German wasnt the lingua franca of international science,but English was.
So it can and will be expected from a scientist to be able to present his/her thoughts in English,as a general rule.
And I say that as a German.
Posted by: Martin_z | July 2, 2009 7:26 AM
I think the writer succeeded in presenting his thoughts perfectly well in English. He also, presumably, was able to give his talk in English.
My point is that a few trivial errors in grammar do not justify comments like (to quote just one example) "Ugh, the guy needs an editor or a grammar school nun rapping him on the knuckles."
Posted by: Confuseddave
|
July 2, 2009 7:41 AM
I think you're reading too much controversy into this. Or someone is deliberately trying to sex up a fairly straightforward suggestion by painting it a field-shattering.
From this abstract, I don't get the impression that this is "directed" evolution. I think what he's saying is that life has evolved to evolve. This is a concept that's occurred to me before - there must be restrictions on the mutability of species: if it's too high you will (in theory) start to impinge on the viability of the species as a whole, and of course without variation a species can't adapt and will die out if and when it's environment changes. There may also be mechanisms that limit genetic change that may be dangerous without limiting more productive mutation.
On the surface, all he's suggesting seems to be that species have evolved to allow a "goldilocks" scenario - not too much, not too little. While it's "higher order" in that it doesn't directly affect the viability of an individual with such a gene in any way, it will in the long term affect the viability of the species. In a gradually changing environement, there's no reason why this wouldn't confer a competitive advantage to a certain lineage within a population, gradually allowing them to displace the less inflexible, or the more disasterously inclined.
It strikes me as pretty sensible. It also strikes me as nigh impossible to prove. One possibility might be looking at variations between species which evolve rapidly and those which don't, but honestly, God knows how you'd filter that kind of search.
Posted by: MadScientist | July 2, 2009 7:42 AM
This is reminiscent of one talk I attended in the AGU Pacific conference in 1996. Someone was talking about "gravity waves" and obviously didn't have a clue what gravity waves were - he was viciously booed and barked out of the lecture room - a very rare thing even in science, but sometimes well deserved.
Posted by: Jim T | July 2, 2009 7:47 AM
I think this is interesting, if I understand it right. From the description he's saying that certain genes which influence mutation rates themselves get tuned by evolution processes independant of the individual animal. Since they're genes, it's not surprising they get tuned.
I've seen something similar in action in another field (I'm not a scientist, surprise, so please let rip).
For a giggle I was playing with genetic algorithms to copy a given picture, using the standard random mutations, combinations and fitness selection based on how closely the solution matched the target image.
When doing this, you have to choose a mutation frequency and a mutation amount - two values which I didn't particularly know how they would affect the final solution. After playing with it for a bit, I decided to give those values over to the solution as well - each solution had a mutation frequency and amount added to it, so that children created would be mutated by the father solution's values. These values would also be combined and mutated just like all the other parts of the solution, but they were ignored by the fitness function of the given solution.
The results were very interesting to see. The overall fitness curve, instead of being the normal smooth asymptote, had steps in it. Looking at the mutation rates you could see that the rates went up when the curve flattened out, stepping the fitness onto a higher curve, the mutation rates then flattened out again. In general, the rates were high to start, then lowered as they neared the ideal solution, but did this in definite waves.
The final solution was slightly better than what I found by hand, although for a long time it was below the best curve.
So, although the method had no direct manner of choosing values for its children and grand-children, it did have the appearance of doing so, all just as an emergent property of the model.
So I'm guessing that the 'evolution genes' would be the equivalent of the mutation rate values I used, and that he's saying these get tuned independantly of the overall population fitness, which is what I saw.
Am I reading too much into this, seeing correlations that aren't there? Is this actually interesting? (I found it so, but only from an 'oh cool' point of view).
Would be interesting to me to see if this really does happen in biology or not, or if you guys just think I've been stating the obvious.
Posted by: recovering catholic | July 2, 2009 7:48 AM
Confuseddave:
"It strikes me as pretty sensible. It also strikes me as nigh impossible to prove." (or, presumably, to even test)
Aren't these two statements self-contradictory from a scientific standpoint?
Posted by: Dan! | July 2, 2009 7:56 AM
I was reading this post with bated breath, since this is right up my alley. What a disappointment!
Presentations like this, that lure you in with a controversial title and then have next to no content, are usually given by people that think they're more important than they really are.
They try to be "paradigm-shifters," but all they really end up doing is supplying ID'ers with a steady stream of sound bytes to misrepresent.
Posted by: David Marjanović, OM | July 2, 2009 7:57 AM
It is well known that certain bacteria turn on less precise DNA repair enzymes (like DNA polymerase IV and V) under stress, which increases their mutation rate; this does not, however, increase the percentage of mutations that are beneficial – it increases the total number of mutations and thus the number of beneficial ones.
It's not. You see, English is a lot more like German than its spelling suggests. Evolution gene is a compound noun – a single word that happens to be spelled with a space in the middle. Compare teapot, which is traditionally not spelled with a space in the middle.
That looks to me like someone tried very hard to… not split an infinitive, even though there's no infinitive far and wide.
Nobody is paid to edit a scientist's work. Especially not their meeting abstracts.
(…Meeting abstracts is another such compound noun, BTW. It does not mean "abstracts that meet", it means "abstracts from a meeting".)
Possible.
True. Well, understandable English anyway – which he managed just fine.
Wow. I'd have loved to see that. :^)
Posted by: Mozglubov | July 2, 2009 8:02 AM
@Jim T,
Your interpretation sounds similar to what I thought he might be meaning (without straying too far into the danger zone of advocating a 'designer'), and I think it is interesting you have simulation results to back it up. However, I'm not sure why this is particularly earth-shattering... it seems like it would be a straight-forward outcome of having your gene-replicating proteins coded for the genome itself. If you code for better replicators, you get less mutation, which is at times desirable and other times it is not.
Anyway, I wasn't at the talk, and neither am I an evolutionary biologist... just an amateur fan of the theory.
Posted by: Jim T | July 2, 2009 8:07 AM
Mozglubov, I didn't think it was earth shattering either, just kinda cool. Hence I didn't get booked into a science show to present it.
Just found some of the old graphs, still quite cool :)
Posted by: Shane | July 2, 2009 8:24 AM
JimT #24,
This is interesting (testable is interesting to us scientific types, so you now are officially One Of Us ;-). Any chance of sticking your algorithm & graphs up somewhere so that we can have a play?
-Shane
Posted by: Naked Bunny with a Whip
|
July 2, 2009 8:31 AM
Sure they do, when the triggers are built into your DNA by the Ancients. Don't you read?
Posted by: Rosie Redfield | July 2, 2009 8:45 AM
Werner Arber isn't a chemist - he was investigating bacterial genetics long before most Pharyngula readers were born. His Nobel is for discovering the restriction-modification systems that gave restriction enzymes their name.
Posted by: franz dibbler | July 2, 2009 8:45 AM
Martin_z @ #14 (is that you Fussi?)
Arber is Swiss and I've heard him switch between English, French, German, Schweizerdeutsch and perhaps Italian in just a few minutes. He didn't use a computer (10 years ago, maybe now but I seriously doubt it) and dictated notes to a secretary who may have some questionable editing skills. Arber has been talking about "variety generators" and "genome plasticity" for at least 20 years. Sounds like he is coming up with new terminology.
Posted by: Jim T | July 2, 2009 8:52 AM
Shane, eek, now you're asking.
Yes, I've just put them here:
http://drop.io/gastuff/
but please be warned that it's not even slightly respectable. This is just something I was playing with and stopped when I saw something else shiny. - There's no write up, you probably won't get the code compiling very easily (and it's not in any sort of public quality). But it should be all there.
Have a look, if you're really interested I'd probably get a kick out of doing a highschool style method-results-conclusion type write up tonight.
Posted by: Ian B Gibson | July 2, 2009 9:01 AM
PZ says:
Then I'd be interested to know what you make of this:
A Switch from High-Fidelity to Error-Prone DNA Double-Strand Break Repair Underlies Stress-Induced Mutation
Rebecca G. Ponder Natalie C. Fonville and Susan M. Rosenberg
Molecular Cell, 2005
Posted by: Mozglubov | July 2, 2009 9:10 AM
@Jim T,
Oh, I wasn't criticizing you for not being earth-shattering... I hope you don't think that's what I meant. I thought your comment was quite good... my disdainful disregard was reserved solely for the talk.
Posted by: a_ray_in_dilbert_space
|
July 2, 2009 9:25 AM
OK, just trying to wrap my limited, caffeine-starved physicist brain around this. Speaking hypothetically, let's say you had a mechanism whereby mutated genes could be repaired as long as the mutation was minor. If that mechanism responded to environmental stress (e.g. reduced caloric intake, heat, lack of water...) with decreased efficiency in repairing damage, that would increae the frequency of mutations. Granted, most of these would be disadvantageous and would die out. However, if the environment is telling us, "Look, your genetics just isn't cutting it for me anymore," an increased rate of mutation could be an adaptive response.
Would this be an example of what Arber is mooting?
Posted by: Tony Whitson | July 2, 2009 9:44 AM
Texas School Board member Cynthia Dunbar says she wants Arber's work taught in Texas K-12 science classes. See
http://curricublog.wordpress.com/2009/01/29/dunbar-scott-arber/
for audio & video of her exchange with Genie Scott.
Posted by: Jim T | July 2, 2009 9:52 AM
a_ray_in_dilbert_space - um, that's probably more accurate than what I was thinking.
Tony Whitson - crazy views of his other work aren't necessarily relevant on this presentation.
Posted by: Cyrock | July 2, 2009 10:22 AM
I guess this is a thought that often comes up to non-biologists. I thought about it many times.
From my ignorance, I don't understand PZ's argument that the cant find a mechanism to select for a gene that affects survival over multiple generations. A gene can have no effect on the survival of the first generation, but giving its descendants a tendency to evolve in a certain direction. This may be advantageous or disadvantageous. If the former, the "evolution gene" will stay in the population, if the later it will be (out)selected.
Not so hocus pocus...
Posted by: Cyrock | July 2, 2009 10:25 AM
I guess this is a thought that often comes up to non-biologists. I thought about it many times.
From my ignorance, I don't understand PZ's argument that the cant find a mechanism to select for a gene that affects survival over multiple generations. A gene can have no effect on the survival of the first generation, but giving its descendants a tendency to evolve in a certain direction. This may be advantageous or disadvantageous. If the former, the "evolution gene" will stay in the population, if the later it will be (out)selected.
Not so hocus pocus...
Posted by: Cyrock | July 2, 2009 10:29 AM
I guess this is a thought that often comes up to non-biologists. I thought about it many times.
From my ignorance, I don't understand PZ's argument that he cant find a mechanism to select for a gene that affects survival over multiple generations. A gene can have no effect on the survival of the first generation, but giving its descendants a tendency to evolve in a certain direction. This may be advantageous or disadvantageous. If the former, the "evolution gene" will stay in the population, if the later it will be (out)selected.
Not so hocus pocus...
Posted by: william e emba | July 2, 2009 10:29 AM
The question of evolvability of evolution itself is by now an old one. It is not an example of "directed" evolution. To put it simply, too much mutation and a genotype can't even exist, but too little mutation and your competitors will outevolve you over time. And even if they didn't, the environment is going to change on you anyway.
The bottom line is that living things control their own DNA repair rates. In bacteria, these mechanisms are easily broken, leading to rapid evolution under stress. If one of the mutations is able to counter the stress, it will survive. As a side effect, its DNA repair rate returns to baseline.
See, for example, some earlier work of Lenski, and the references therein: Sniegowski, Gerrish, Lenski "Evolution of high mutation rates in experimental populations of E. coli" Nature 387, 703-705 (12 June 1997)
Posted by: a_ray_in_dilbert_space
|
July 2, 2009 10:34 AM
Cyrock,
The question is how does evolution anticipate what the environment will demand in the future? And if the anticipation is wrong? Game over.
Evolution is fundamentally a feedback process--certain advantageous traits are amplified, while disadvantageous traits are attenuated. The thing is that the environment, and therefore the feedback is time dependent, and unpredictable on any timescale that coule be meaningful in the life of an individual.
Posted by: Cyrock | July 2, 2009 11:00 AM
I guess this is a thought that often comes up to non-biologists. I thought about it many times.
From my ignorance, I don't understand PZ's argument that he cant find a mechanism to select for a gene that affects survival over multiple generations. A gene can have no effect on the survival of the first generation, but giving its descendants a tendency to evolve in a certain direction. This may be advantageous or disadvantageous. If the former, the "evolution gene" will stay in the population, if the later it will be (out)selected.
Not so hocus pocus...
Posted by: Cyrock | July 2, 2009 11:02 AM
I apologize for the repeated comments, I don't comment often, so i didn't know the delay is considerable.
Posted by: Jay | July 2, 2009 11:25 AM
Can someone please explain some DNA functions over time? Do mutations of the originating parent accumulate through life and then get passed onto offspring at the time of reproduction? If so, is there a DNA_mutation(t) function that would give a rate of mutation, and could that rate of mutation increase in response to environmental changes?
I know thats a lot of dependent questions, but after reading the "stress induced evolution" stuff here, I'm really really curious.
Posted by: apthorp
|
July 2, 2009 11:31 AM
This sounds like the fictional force problem in physics. (see http://xkcd.com/123/ ) More generally it is a problem interaction terms that show up when you aggregate system elements in different ways that routinely causes squawks from reductionists (who are happy banging quarks together) and philosopical idealists (who think 'emergent properties' are real ontological entities and are the opposite of nominalists).
If I read correctly, the abstract says that there are system effects that encourage evolution, i.e. spreading into new species. This can be a simple effect where increasing the complexity of an ecosystem creates more viable slots, which diffusion will inexorably fill up. When you open a valve between the atmosphere and a vacuum it's common to think of a 'force' that pushes the air into the vacuum. We write the differential equation that way, and it works quite well. Even if what is 'really' going on is that the gas molecules are randomly bouncing off each other and the so-called pressure is nothing more than the fact that there are more that bounce into the vacuum than bounce out. Not much magic either way though.
Posted by: Blake Stacey | July 2, 2009 11:33 AM
So, Arber rediscovered the idea of the evolution of evolvability?
Posted by: Don Garty | July 2, 2009 11:40 AM
Werner Arber made no sense (as far as evolutionary biology is concerned) even twenty years ago. At eighty, he seems less lucid in Englisch than ever.
Posted by: Cyrock | July 2, 2009 11:41 AM
@apthorp
Good comment. Damn emerging properties, like women cant live with nor without them ;).
@a_ray_in_dilbert_space
Just like normal population selection. If it was bad, the population with that gene will reproduces less, if it was good, it ends ups spreading throughout the whole population of the species.
Posted by: Peter Ashby | July 2, 2009 11:54 AM
I agree with David Marjanović #22, the English only has to be understandable. I was once invited to sit in on a group meeting in a lab in Geneva. One of the students was talking, in English, essentially a practice for a meeting presentation (another one of those compounds). He made one mistake (out of the whole talk) and the Group Leader picked up on it and asked me as a native English speaker to confirm. I said that yes, it was technically a mistake, but it did not make it any less understandable.
Also English as the Lingua Franca for science is a relatively recent phenomenon. Physics and Engineering was written in German for a long time and the Journal currently known as Development back when it was JEEM (Journal of Embryology and Experimental Morphology) published papers in French with English Abstracts as recently as the Mid '80s. I quoted some of them in my thesis and yes I did read them, Academic French is not as hard as colloquial French. I did need a dictionary though.
There is still a lot of science that gets published in Russian, Chinese, Korean or Japanese. Us monoglot English speakers are not as informed as we think we are.
Posted by: a_ray_in_dilbert_space
|
July 2, 2009 12:12 PM
Cyrock,
Except that it's not like normal evolutionary pressure, because the feedback that reinforces the trait doesn't give any advantage (and in fact may penalize) the parent/grandparent...before there is any feedback.
I sort of view evolution as being like getting a pair of glasses. The optometrist is asking you to give immediate feedback--is this better or this? You're asking "Is this better or will the one you haven't seen be better?"
Posted by: Jim T | July 2, 2009 12:32 PM
a_ray_in_dilbert_space: See my overlong post above. There's some simulation evidence that second order selection of that type (not directly connected to the fitness function) can still benefit the population as a whole. The genotype actually appears to change in an interesting manner.
Posted by: a_ray_in_dilbert_space
|
July 2, 2009 12:45 PM
Jim T., In essence, your evolution model then has to become non-Markov, right? And you then have to look at how many generations a trait can persist until it's current handicap extinguishes its future benefit. That could get quite complicated. What is more, it seems to me that it would be an advantage in species that have large numbers of offspring, so would you look for such traits in the insect world?
Posted by: Brian | July 2, 2009 12:48 PM
Reading the abstract rung my bogometer, but I think it's become hypersensitized from reading Sokal's latest book. That this is an laureate scientist straying outside his domain strikes me as the likelier characterization.
Posted by: Jim T | July 2, 2009 12:58 PM
a_ray_in_dilbert_space: Ok, I'm only talking about the simulation I did for fun - my knowledge ends there. The best that the simulation can show is that second order selection is a viable theoretical possibility - I don't know how to test it in the real world.
But no, the model preserved its markov properties - manipulation of the genetic markers depended entirely on the current genome and nothing else.
In terms of existing traits, if you download the results and look at log3.ods, you can clearly see the mutation rate consistently increasing for significant numbers of generations then falling coherently. You can also see that it's closely correlated with the inverse of the mutations sizes, and that a spike in mutation sizes drastically increased the fitness of the population at around generation 7500.
That's a pretty clear indication of a trait persisting whilst it's useful and removing itself when it's no longer useful, all through the standard evolutionary model. Keep in mind that the mutation rate genes were NOT part of the fitness function and were free to mutate without pre-determined aim.
Posted by: a_ray_in_dilbert_space
|
July 2, 2009 1:16 PM
Jim T,
How did you determine the advantage/penalty that a mutation incurred? Presumably it was a function (nonlinear?) of mutation size. Moreover, how did you determine how it changed with time? And would it really be Markov--I mean, yes, if one is born with the trait, one either survives or dies, but one has to reproduce in the process, and depending on how successful one has been, one may reproduce a different number of offspring. Thus, you could have an extinction function associated with a trait. Or is this not relevant?
Posted by: Cyrock | July 2, 2009 1:21 PM
A simple example of how it MAY work:
(excuses for the oversimplifications)
A gene X that gives controls the standard deviation in the amount of vertebra in the column of a species, say a snake. With the analogy of a slider, it can be set to either min or max.
Say the amount of vertebra has influence in some factors. For our imaginary purposes lets assume: One more vertebra allows for higher speed. One less vertebra improves agility.
One population of snakes(A) has gene X and the other doesn't(B). If the environment starts to change, (B), with a fixed standard deviation for the amount of vertebra, will "try" to adapt to the changing environment, but after 20 generations of decreasing population, they extinguish. (A) will also try to adapt, and will reduce in population, but will have it easier to change the number of vertebra. Descendants of the individuals with the gene X "slider" set to max, will show more variations in the amount of vertebra, and will have more individuals adapted to the new environment.
Once the environment is stable again, having gene X set to max is useless, so the sons of these individuals wont reproduce, and the slider will go back to "min".
------------------------
There are second order effects in any branch of science, why would biology be any different.
Posted by: bunny | July 2, 2009 2:41 PM
I am not a biologist, so I lack a background in the specific biological mechanisms of evolution, but from my work in statistics, information theory and discrete entropy (primarily relating to natural language processing, but the principles hold for many statistical system,s population dynamics notwithstanding), I would contend that provided such mutation-frequency-controlling genes exist, evolution *does* provide organisms with a mechanism to control their own evolution.
Traits such as the instinct to pass knowledge to children, which to my knowledge is almost unanimously observed among cephalized populations, provide an advantage only to the offspring, and not to the parent. A modified mutation rate could significantly affect the survivability of a population based on the volatility of the environment; for instance, in a very specific climate, such as a rainforest, a lower mutation rate would allow an organism to fit much more precisely into a niche, allowing for higher overall population survivability. In more dynamic climates, where temperatures varied distinctly not only from season to season, but from generation to generation, higher mutation rates would allow the next generation to adapt more easily to a wandering niche. In the computer science world, this 'meta-mutation' would be equivalent to changing the parameters to a search algorithm to more easily and accurately navigate the search space, with the goal being an accurate and precise population centered around a niche. For some evidence of this, check out any of the thousands of papers on optimizing search algorithms, or pick up a copy of Russel & Norvig's "Artificial Intelligence: A Modern Approach" for an good overview of the topic.
Even without such a specific one-generation benefit to such modulation of the mutation rate, a cursory statistical analysis suggests this effect would still be observed. Assume that the conclusion of the previous paragraph is false, and that there is no benefit to modulating mutation rates for generation + 1. Then because the regulatory genes exist, over time genetic drift will cause separate populations to have significantly different rates of mutation, meaning that after a less frequent genetic event, ie. a bottleneck or global major climate change (say, asteroid collision, or ozone layer) and the evolutionary landscape changed significantly, the population with the more suited mutability rates would expand much more quickly into the new adaptive territories, and be notably more genetically viable than the other, preserving not only their traits but also their meta-traits, ie. the mutation rates.
I think Arber's thesis is true, but not so much a fundamental observation about the genome as an observation of a phenomenon that according to statistical principles, is obviously likely to emerge in any system with the basic principles of evolutionary theory.
Posted by: Jim T | July 2, 2009 2:55 PM
a_ray_in_dilbert_space: The fitness function was simply the sum of the distance in rgb colorspace between the solution pixels and their ideal value.
That was the first order selection.
The size and number of mutations that occurred were irrelevant in their own right, only whether their cumulative affect made the solution better or worse compared to the other solutions mattered.
The point was that the mutation rate values were NOT a part of determining the fitness of an individual, but were stored and mutated along with each individual. For the individuals concerned, the genes controlling the mutation did not affect them at all, only their children. The only way it affected the children was the chance that a mutations would occur, and the severity of those mutations.
Because the effect wasn't direct on the carrying individual, any correlations on those genes were a second order selection. The effect of the mutation was a tuning that produces a better solution progress than a static manual selection for the mutation rates (see analysis.ods).
The only things affecting the mutation rates in my simulation was the second order evolution pressures on the genes. It's quite possible that in real creatures these genes could also be affected by environmental factors as you described earlier, but that's not part of my simulation.
Posted by: Steve LaBonne | July 2, 2009 3:34 PM
This is an argument that's been going on for a long time (since before I was a grad student and that was A WHILE ago): is the mutation rate as low as it "can" be (taking into account both the irreducible physical minimum error rate and the metabolic cost of higher fidelity) or is there some higher-level selection going on to keep it from getting "too" low to generate "enough" diversity. There hasn't been anything new said about this for a long time (certainly not here by Arber) nor have the proponents of the latter view ever presented anything more than just this kind of hand-waving.
Posted by: Kagehi
|
July 2, 2009 3:34 PM
Yeah. It makes sense that the genetics of species may have mechanisms that survived "because" they could produce wider variation, under conditions that allowed more members to survive in general, and thus benefited from variations that could adapt them to new conditions. We even see this in some cases. And, yes, there is no more reason to "imagine" that the universe itself "fine tuned" these things into existence, like a police directing traffic, any more than its rational to assume that some invisible force told artificial life in the Avida platform, "Here is how to fake being an evolutionary failure, when the scientists are feeding you the same data stream over and over, to see if they want to kill you off." Those that one some level had the ability to recognize the stream and react to it negatively didn't die. Those than fed on it did. The result being, "evolved" forms that didn't react to it as food didn't die in later tests, even though, in "normal" conditions, they fed more efficiently. In other words, they evolved a taste bug, and figured out that "bitter" meant "poisonous". Big fracking deal.
Arber's stuff is the same sort of nonsense. Species work out when its "bad" to have wide variation, so, when those that couldn't produce it died, and those that produced it all the time, died, the ones that could, to some degree, turn it on and off, didn't.
The problem here is whether or not this is what the guy means, or if he is just really seriously dumb about how he picks his words, and doesn't have a clue that nearly all of them he used sound like borderline ID BS. Or worse, some of it "is" borderline ID BS, and he fails to grasp why its "not necessary".
Posted by: apthorp
|
July 2, 2009 3:36 PM
there isn't a mechanism given for individuals to maintain a gene that will only help its many-times-great-grandchildren. It is inferring a kind of foresight to evolution that is doesn't have a mechanism
Lets assume there is no foresight and active planning by chemicals.
Adding stress such that the survival criteria change (invasive species, climate change, meteor blowing up a continent) and the expectation is that the distribution of species will change, and new ones will appear from the pre-change population.
It seems pretty reasonable that those species that can evolve the fastest are most likely to dominate the new ecosystem.
It also seems reasonable that those with the most accessible options (what in physics would be called a phase space -- the good ones keep them on stun) will evolve the fastest.
So is there a mechanism to ignore a gene that has no present disadvantage? Or has been turned off by present advantage as in the dinosaur fingers we read about a few days ago. But it lurks around as 'junk' for a few hundred generations in a stable ecosystem and, like any other mutation wins the lottery and confers advantage by being turned on.
Of course if mutations must be completely new 'code' every time, this doesn't work.
But if reuse/re-purpose is available and several cycles of stress have been survived one would expect most extant species to have an adaptability reserve inherited from their ancestors who survived because of it. At critical points, their over specialized and optimized co-inhabitants suffered a terminal disadvantage.
Look at it another way, if the overall gene system wasn't resilient to actually occurring disruptions it wouldn't be here.
Now suppose you observed the property and attributed it to the "evolution gene". The implication of a single gene is the "fictitious force". It may be a desirable property of the whole system of genes.
And it may be completely wrong. But is an example of a mechanism that has no need to attribute foresight or external external intervention.
Posted by: apthorp
|
July 2, 2009 3:53 PM
bunny at #55 -- yes. But a few quibbles.
It is probably possible to distinguish between instinct = reactive behavior and knowledge = symbolic forward prediction (perhaps based on memory).
And there doesn't need to be "a gene". It's not a simple system which means there are strong system effects both within the organism and with the external ecosystem.
Steve at #57 -- what kind of test/new information could test this sort of thing?
Posted by: Intelligent Designer | July 2, 2009 4:32 PM
Jim T @ 19
In other words, your wrote a convergence algorithm. It doesn't mean or prove anything.
Shane @ 11
Jim,
This is why your "genetic algorithm" is irrelavent. However, if Shane could be proved wrong it would imply that evolution is the result of intelligent design. Kind of like a developing software.
Posted by: Jim T | July 2, 2009 5:08 PM
Intelligent Designer: Nice try but no dice. Actually no, pretty lousy try. When modelling the parameters of a theory produce the results that the theory predicts, then the theory is demonstrated to be sound (not proven, or shown in the real world, but demonstrated). Whether the search space or optimal solution is known beforehand is irrelevant. The system models the ideas of evolution and produces the expected results.
You can't discard galactic collision simulations or CFD simulations just because they're setup with the intent of running the theory.
But the interesting bit of the model is that there was no pre-determined result in the inclusion of the mutation values. I included them without knowing if they'd help, hinder or be neutral to the process, and without guiding them in any way.
Yet these slider values were expertly manipulated by the process with an accuracy that would suggest that someone was sliding the controls with intent, yet there was no intent at all attached to those values.
Whilst I did suspect that the values would start out large and finish small, I did not expect the inversion features, the periodic nature of the changes, or the clear steps that are visible in the fitness graphs.
Based on what I've seen of intelligent design advocates, if you had seen those kinds of values recorded in a natural system you would be screaming "That's clear evidence of intention and therefore design." You be saying "Show me the mechanism which can detect when fitness increase is slowing down and kick start it into higher gear again without some designer at the wheel - there can be no such mechanism."
This system shows that such manipulations are indeed possible without intent - as an emergent and unexpected feature of the system.
Posted by: Intelligent Designer | July 2, 2009 5:47 PM
Posted by: Heraclides | July 2, 2009 7:01 PM
Food for thought: could this be a case of someone who has been around for a long time trying to "remind" the new crowd something old that they think isn't discussed or thought about enough.
Posted by: bunny | July 2, 2009 7:04 PM
Intelligent Designer @ 63:
While it is true that his simulation does represent a convergence algorithm, as do the vast majority of search algorithms which search continuous domains, the convergence was not being disputed nor even really looked at in the analysis. I'm not quite sure what your argument against the experiment is, but the evidence provided by the experiment is the analysis of the relative rates at which the algorithm produces a convergence over time, not its definitive convergence, and this is why it is both relevant and very interesting.
Jim T:
I am very interested in this line of experimentation. Have you tried comparing your results with some of the observed population variance statistics?
I'm not much for the number crunching, but I sure like coding, and will probably try to expand upon this line of simulation. Right now I'm looking at trying to simulate language development in populations.
Posted by: Steve LaBonne | July 2, 2009 8:53 PM
Ay, there's the rub. It's very hard to think of definitive tests so on this subject you mostly get philosophizing. It's been along time but I believe there were some papers purporting to show that the efficiency of error correction and DNA repair are pretty close to some kind of theoretical maximum. Sorry, this stuff is way to much part of a past life for me to be able to cite chapter and verse.
Posted by: DarkDad | July 2, 2009 9:00 PM
The references listed above (#30 and #38) are where my mind went when I read the abstract. In the spontaneous mutation world, the question of the "right" mutation rate for a given organism would come up over beer very often, given that both anti-mutator and mutator alleles could be selected for pretty much all DNA replication and repair genes.
Population studies have been done back into the sixties on the effects of selection on populations (sorry, don't have the references handy) and have shown that the mutation rate of non-selected genes rise in concert with the selection. In Sue Rosenberg's ( and Pat Foster's) work it became apparent that very slow growth during selection, coupled with decreased fidelity could allow mutational escape from pretty severe selections.
I don't think that the results in bacteria can be easily translated to eukaryotes (our lab tried with Saccharomyces and couldn't get a definitive answer to the problem) much less larger species where the mutation event is temporally separated from the selection. (mind you, some of the rodent studies on maternal effects I remeber reading during my thesis writing seemed to suggest stable inter-generational effects on DNA repair by low-level irradiation of the maternal parent. Weird and not replicated to my knowledge.)
I'd have to come down on the "too far out of his field" side, given what PZ has written about the talk.
Posted by: rachel Wells | July 2, 2009 9:38 PM
By modulators, I think he means DNA polymerase, RNA polymerase- the DNA replicators and gene transcribers respectively. That is about all i got from this guys talk.
Other than that i don't really know what this guy is talking about. I'm not sure what is new about his hypothesis. I also don't have a clue what he means by 'neutral reality' what!? Does he mean time?
Wish I was there to see Peter Agre, I'm doing a presentation on aquaporins next week!
Posted by: raven | July 2, 2009 10:18 PM
I think I know what he means and it is old and mainstream science.
The mutation rate of bacteria seems to be itself under selection and selectable. Too many mutations are not good, too few are not good either.
The proof. Select spontaneous mutations. Which are rare, on the order of 10exp-7 or 8. A percentage of those will be "mutator strains". You inadvertanly select for mutator strains which then produce the selected for mutations. These mutator mutations are in the genes you would think, ones involved with DNA metabolism and replication.
You can also select for anti mutator phenotypes.
IIRC, Lenski picked up some mutators in his citrate metabolism evolution experiments.
Posted by: raven | July 2, 2009 10:32 PM
More than most want to know about mutators. They are common and date back in discovery to the mid 20th century.
Cynthia Dunbar wants Arber taught in Texas schools because she thinks he is a creationist. He isn't. She also doesn't have the brains or will to even understand what his work is all about.
Genetics, Vol. 164, 843-854, July 2003, Copyright © 2003
The Evolution of Mutator Genes in Bacterial Populations: The Roles of Environmental Change and Timing
Mark M. Tanakaa, Carl T. Bergstromb, and Bruce R. Levina
a Department of Biology, Emory University, Atlanta, Georgia 30322
b Department of Zoology, University of Washington, Seattle, Washington 98195
Corresponding author: Mark M. Tanaka, University of New South Wales, Sydney NSW 2052, Australia., m.tanaka@unsw.edu.au (E-mail)
Communicating editor: S. P. OTTO
Recent studies have found high frequencies of bacteria with increased genomic rates of mutation in both clinical and laboratory populations. These observations may seem surprising in light of earlier experimental and theoretical studies. Mutator genes (genes that elevate the genomic mutation rate) are likely to induce deleterious mutations and thus suffer an indirect selective disadvantage; at the same time, bacteria carrying them can increase in frequency only by generating beneficial mutations at other loci. When clones carrying mutator genes are rare, however, these beneficial mutations are far more likely to arise in members of the much larger nonmutator population. How then can mutators become prevalent? To address this question, we develop a model of the population dynamics of bacteria confronted with ever-changing environments. Using analytical and simulation procedures, we explore the process by which initially rare mutator alleles can rise in frequency. We demonstrate that subsequent to a shift in environmental conditions, there will be relatively long periods of time during which the mutator subpopulation can produce a beneficial mutation before the ancestral subpopulations are eliminated. If the beneficial mutation arises early enough, the overall frequency of mutators will climb to a point higher than when the process began. The probability of producing a subsequent beneficial mutation will then also increase. In this manner, mutators can increase in frequency over successive selective sweeps. We discuss the implications and predictions of these theoretical results in relation to antibiotic resistance and the evolution of mutation rates.
--------------------------------------------------------------------------------
Posted by: Jim T | July 3, 2009 2:17 AM
bunny @65
No, I've not done anything to compare my results with population statistics, I'm not even sure what would be a relevant comparisson.
Also, I'm far from unique in doing this - polyworld used the same technique of including mutation rates in the genome.
Posted by: Jim T | July 3, 2009 2:29 AM
Intelligent Designer @63
Well, from my non-scientist point of view, I'd claim that evolution *is* a convergence algorithm. But that the point of convergence is constantly shifting and determined by the entire state of the system, not just a fixed goal.
For demonstration purposes, a fixed goal works fine. It would be easy to setup a shifting goal based on the entire state of the system, but it would be difficult to do so and be able to pick out any meaningful states from observing the system.
... and as bunny says, that's not even the point of this discussion.
You can't just say you disagree with one aspect of the system and then discount anything else. With that, you'd never accept any simulation run on a computer, or laboratory experiment performed with intent under controlled conditions.
You're not convincing me that you understand the discussion or are anything other than an obvious troll.
Posted by: David Marjanović, OM | July 3, 2009 2:11 PM
This is precisely what I was talking about at the beginning of comment 22. Note how it does not increase the proportion of beneficial mutations; it increases the total mutation rate.
But how can it be selected for or against when it has no influence on the first generation whatsoever? What advantage or disadvantage does the first generation get from having such an allele? If it doesn't get any, there's no natural selection.
Yes, and in bacteria this is what happens. Just to rub it in, though: this does not change the proportion of beneficial : neutral : harmful mutations.
Yes, except it's complicated by the fact that both min and max are themselves subject to mutation.
Both true, and both completely irrelevant. What matters is that they provide an advantage to the parent's genes.
Read some Dawkins to understand your utter worthlessness and irrelevance :o)
Posted by: Michael Ralston | July 3, 2009 3:50 PM
Yes there is, over the course of multiple generations. This is trivial - if an allele has no effect on immediate reproductive success, but has an effect on long-term reproductive success, then the organisms who possess that allele will have more descendants at the point where the long-term effect shows up.
For instance, imagine a recessive allele that suddenly appears (ie, by mutation) on one organism of a population, and imagine that possessing two copies of that allele provide significant reproductive advantage, while possessing only one provides no advantage or disadvantage.
It is obvious and trivial that the organism with the recessive allele will, on average, have more descendants after a sufficient number of generations than organisms without the recessive allele, all else being equal.
And yet, the recessive allele provides no first-generation advantage to the organisms that possess it.
Posted by: Pantalaimon | July 6, 2009 7:49 PM
„Wie solche bereits recht komplexen Strukturen zusammenkommen können, bleibt für mich ein Geheimnis. Die Möglichkeit der Existenz eines Schöpfers, Gottes, ist für mich eine befriedigende Lösung des Problems.“ (Arber, 1994)
"How such already complex structures [cells] can come togehter remains a mystery for me. The possibility of the existance of a creator, God, is a satisfactory solution to the problem."
This sentence is often quoted by german IDiots. Arber himself seems not to be part of the movement, so i guess this was quotemined. Given PZs protocoll of his talk, maybe the quote isn't so far fetched.
Posted by: Pantalaimon | July 6, 2009 7:52 PM
....God, is a satisfactory solution to the problem for me."
Sorry for double post but that had to be corrected.