|
Author
|
Topic: Is "complexity" relative or absolute?
|
edmund
Member
Member # 206
|
posted 27. March 2002 03:17
Is "complex" as used in CSI an absolute or relative term?
It seems to me as though when we talk about an event being too complex for chance (CSI as opposed to mere SI, which is achievable by known processes), we must declare a certain set of "known processes" relative to which we assess the complexity of the event in question. That is, using our knowledge of a set of "known processes", we determine the probability for the event in question. If the event is wildly improbable, given our declared set of "known processes", then we know that the event was caused at least partly by some process which we have not included.
Consider, for example, coming upon a forest where every tree leans toward the west. If you were meteorologically naive, you might put in your set of "known processes" only the various rules by which plants grow, and conclude that, on average, trees should lean west no more often than they lean east or north or south. From this probability distribution, you would conclude that the west-leaning forest was extraordinarily unlikely, given the known processes under consideration: given this set of known processes, the forest is CSI.
You would not, however, necessarily be safe in concluding that this arrangement was designed. Suppose that you spent the night in this forest. In the daytime, the air was quite still, but you find that, during the night, a raging wind comes from the east. When you add this wind to your set of "known processes", altering your probability distribution, you find that the west-leaning forest is perfectly likely, and conclude that it does not contain CSI.
On the other hand, suppose that you checked the weather records for the area and found that the air was always calm. Suppose that you then went on to rule out any of the various geotropisms or phototropisms or magnetic influences which could possibly be influencing this forest. Under this set of "known processes", the west-leaning forest remains a statistical impossibility. If you managed to exhaust all natural, non-intelligent processes, you would conclude that this tilted forest was a product of intelligent design.
The way in which the word "complex" is sometimes used with reference to biological systems in the ID literature and fora such as this one troubles me. It is not the absolute complexity of the systems which matters. It is how unlikely the biological systems are *given the natural processes we know about*. Information only becomes complex information when we know that our "known processes" are statistically unlikely to produce that information.
From this point of view, Dembski's claim (in his Law of Conservation of Information) that natural processes cannot produce CSI runs the risk of becoming a tautology. If our declared set of "known processes" consists of all natural, non-intelligent processes, then of course they cannot produce CSI-- by definition! CSI is defined as a specified pattern which is highly unlikely to be produced by the set of "known processes". The only question that remains is: what is the probability distribution generated by "all natural non-intelligent causes"?
To put this in simple terms: what are natural causes likely to produce? In the biological context, these causes mean evolutionary processes, especially natural selection. It seems to me that we can only declare biological organisms CSI as opposed to mere SI if we can show conclusively that the patterns of these organisms are unlikely *given* evolutionary mechanisms. Which suggests that the important questions are the ones which evolutionists and creationists always dicker about:
1) What can natural evolutionary mechanisms produce, and what can they not produce?
2) How do we know that this is true?
I apologize if this has been discussed recently here-- I am a newbie. It is entirely possible that I have overlooked some obvious detail of the ID debate. If I have, I beg your indulgence. I am eager to see these ideas criticized within an inch of their life.
edmund
IP: Logged
|
|
Evan
Member
Member # 164
|
posted 27. March 2002 07:37
I believe that basically your reasoning is solid.
Inferring design is based on improbability. The only way we can know that something is improbable is to know enough about how it actually happened that we can in fact compute the probabilities. Assigned probabilities based on what we don't know does not make for a sold argument. It also leads to the problem that you mention that new information can show that what was once thought to be designed is not, because it is no longer considered improbable.
The major difficulty is trying to figure out how something would have happened if it happened by natural evolutionary processes, and then showing that to be improbable. In order to do this, one needs an hypothesis about [B]how[B] design happens so that one can know what evolutionary processes to look at so as to compare probabilities. In the thread "Evolution by Design", I have offered an hypothesis about how design might happen in an effort to confront this and other questions about design. With the help of the comments of ISCID member Charlie D, we have come to at least one possible testable hypothesis about one aspect of this design hypothesis.
You might want to read through the thread “Evolution by Design: a synthesis” elsewhere on this page and see if anything there relates to your question.
IP: Logged
|
|
Nelson Alonso
Member
Member # 52
|
posted 27. March 2002 11:13
Evan:
Inferring design is based on improbability. The only way we can know that something is improbable is to know enough about how it actually happened that we can in fact compute the probabilities. Assigned probabilities based on what we don't know does not make for a sold argument. It also leads to the problem that you mention that new information can show that what was once thought to be designed is not, because it is no longer considered improbable.
Nelson:
This is all irrelevant. If we know about the structure and function of the system in question we can assess probabilities, and hypothesize about the origin. Take for example, the very concept of sequence homology. It is itself based on a probability argument "without knowing exactly what happened". If SETI ever comes upon a sequence of numbers that is too improbable, even if they don't know exactly how it got to them, they would conclude intelligent design.
Of course future knowledge may overturn any research, but thats true for just about anything.
IP: Logged
|
|
Evan
Member
Member # 164
|
posted 27. March 2002 19:31
I wrote,
quote:
Inferring design is based on improbability. The only way we can know that something is improbable is to know enough about how it actually happened that we can in fact compute the probabilities. Assigned probabilities based on what we don't know does not make for a sold argument.
Nelson began his response by writing “This is all irrelevant.”
I don’t understand this response. My statement that “inferring design is based on improbability” is at the heart of Dembski’s explanatory filter. For instance in his opening post in the Causal Adequacy thread here at ISCID, Dembski writes,
quote:
For something to exhibit specified complexity it must exhibit a detachable pattern that maps onto an event of small probability (the "complexity" in "specified complexity" is a measure of probability).
In other places he explains more fully that design is inferred when we can eliminate law-and-chance as the cause, and to do so we have to show that a particular event is highly improbable.
So it does not seem irrelevant to me that I began my remarks by pointing this out.
My second point – “The only way we can know that something is improbable is to know enough about how it actually happened that we can in fact compute the probabilities.” – relates to a problem that has been brought up by several people here at ISCID: that Dembski’s ideas have yet to be applied to real, specific, biological events or structures in a realistic way.
So, this comment doesn’t seem irrelevant either.
Now Nelson goes on to say,
quote:
If we know about the structure and function of the system in question we can assess probabilities, and hypothesize about the origin.
On the surface, I don’t see how just knowing the structure and function of something could help us calculate the probability of its coming into existence. Probabilities need to be in respect to some hypothesis about the origin of something: the hypothesis about the origin of something has to come before the calculation of the probability, not afterwards, as implied in Nelson’s statement. I don’t think that just looking at how something is now, without regards to how it came to be, gives us the information we would need to calculate the probability or improbability of the object (and hence complexity, and hence whether it was designed or not.)
I would like to hear Nelson explain his thoughts on this, and perhaps give an example.
Here’s an example to think about. This weeks Science News has an article about how colobine monkeys in Asia have a digestive system with features somewhat like a cows which allows them to eat leaves as their primary diet. (Unfortunately this article is not online.) The researchers have traced this development to a gene duplication event that took place about 4 million years ago, and seems to have led to other changes that provided a benefit to these monkeys.
Now most people would ascribe this change to natural “micro-evolutionary” processes as opposed to a design. But how could we tell if that is true? – how could we calculate a probability that would empirically establish one way or another whether that was true?
I certainly don’t believe we could establish a probability just by examining the structure and function of the colobine’s digestive system (but I would be open to hearing how Nelson thinks that might be done.)
I would think that in order to establish a probability about this, we would have to know something about the probability of the gene duplication event happening in a individual monkey, the cumulative probability of it happening in population of monkeys in the particular environment, something about the overall ecology of the environment (other animals, other food sources), the odds of other mutations happening to enhance the effectiveness of the original duplication event, and so.
At some point, those wishing to apply Dembski’s design ideas to biological reality will need, I think, to find ways (direct or indirect) to calculate such probabilities. It would seem useful to start with things that we assume aren’t designed, and offer empirical ways to establish probabilities for those events. If such procedures could gain some credibility by being empirically grounded, then applying the same procedures to things that we think are designed would have some chance of being accepted as valid.
IP: Logged
|
|
Nelson Alonso
Member
Member # 52
|
posted 27. March 2002 19:49
Nelson:
My "irrelevant remark" was referring to this point:
quote:
The only way we can know that something is improbable is to know enough about how it actually happened that we can in fact compute the probabilities.
Evan:
My second point – “The only way we can know that something is improbable is to know enough about how it actually happened that we can in fact compute the probabilities.” – relates to a problem that has been brought up by several people here at ISCID: that Dembski’s ideas have yet to be applied to real, specific, biological events or structures in a realistic way.
So, this comment doesn’t seem irrelevant either.
Nelson:
No you said that we can not know if something is improbable unless we know how it "actually happened". For many of the systems discussed, we will never know what "actually happened", so it is irrelevant. We can still, based on the structure and function of the system, estimate probabilities. Dembski did do this for real world examples, the bacterial flagellum and T-Urf are examples that come to mind.
Nelson:
If we know about the structure and function of the system in question we can assess probabilities, and hypothesize about the origin.
Evan:
On the surface, I don’t see how just knowing the structure and function of something could help us calculate the probability of its coming into existence. Probabilities need to be in respect to some hypothesis about the origin of something: the hypothesis about the origin of something has to come before the calculation of the probability, not afterwards, as implied in Nelson’s statement. I don’t think that just looking at how something is now, without regards to how it came to be, gives us the information we would need to calculate the probability or improbability of the object (and hence complexity, and hence whether it was designed or not.)
Nelson:
There is no reason why we cannot do any of the things you say above we cannot do. But at any rate, Dembski, in his book No Free Lunch, calculates the probability of a natural selection process producing the flagellum. This cannot be done unless you know the structure and function of the system. The structure is relevant when it comes to the "parts" of the system. The function is relevant when it comes to the specification.
Not having a link to your example, I cannot specifically tell you why structure/function is important there as well. But I can use another system, one that brought up in Behe's thread, the F-ATP synthase. Many people would try to explain the origin of this system through gene duplication of the alpha or beta subunits. But knowing that this is an 8-part system, and knowing the system's function, I would say that the system is irreducibly complex. We can now apply Dembski's Porig X Plocal X Pconfig calculation to determine whether chance events like natural selection of duplication events would likely form this molecular machine.
[ 27 March 2002, 19:53: Message edited by: Nelson Alonso ]
IP: Logged
|
|
Evan
Member
Member # 164
|
posted 27. March 2002 21:06
Nelson, in explaining why he said my previous post raised irrelevant points, says, “No you said that we can not know if something is improbable unless we know how it "actually happened". For many of the systems discussed, we will never know what "actually happened", so it is irrelevant.”
I agree that my wording is wrong (and I thing I’ve been more careful about stating this in the Evolution and Design thread.) What I mean to say is that we need to be able to calculate probabilities about ways that something might have happened if it had happened through natural means. I believe Dembski calls these “chance hypotheses” when he says that the design inference needs to eliminate all realistic ways that law-and-chance might have caused the event in question.
Please understand that I understand the realistic limitations of this demand. All I am saying is that it would be useful to try to show estimates of probabilities for things that we know (or are fairly sure we know) evolved naturally and were not designed. Some kind of useful and usable set of notions, based on some hypothesis of how the action of design differs from evolution, needs to be developed so that Dembski’s explanatory filter can be actualy applied rather than being just a theoretical framework.
Nelson also writes,
quote:
There is no reason why we cannot do any of the things you say above we cannot do [i.e., estimate realistic probabilities]. But at any rate, Dembski, in his book No Free Lunch, calculates the probability of a natural selection process producing the flagellum.
...But I can use another system, one that brought up in Behe's thread, the F-ATP synthase. Many people would try to explain the origin of this system through gene duplication of the alpha or beta subunits. But knowing that this is an 8-part system, and knowing the system's function, I would say that the system is irreducibly complex. We can now apply Dembski's Porig X Plocal X Pconfig calculation to determine whether chance events like natural selection of duplication events would likely form this molecular machine.
I have not read No Free Lunch, but what I understand from reading other descriptions of this is that Dembski calculated the probability of all the components of the flagellum coming together simultaneously by chance using simple combinatorial mathematics. This does produce a lower bound on the probability, I think, but it does not take into account any possible historical scenario by which smaller steps might have led to the flagellum.
I understand that no one has proposed a hypothesis about such a series of steps, and that, given that this is an issue that goes back clear to the origin of life, it may prove to be very difficult to propose anything other than pure chance at this time. Again, this is why it would be useful to start with more recent events and try to develop the tools to calculate improbability in order to infer design.
Part of the problem with Dembski’s account in No Free Lunch, as I understand it from others, is that Dembski does there what Nelson does in this example: he assumes irreducible complexity (and assumes Behe’s claim that IC things cannot have occurred in steps), concludes therefore that simultaneous chance combination is the way the flagellum originated, and then calculates a number that shows the flagellum was designed. But since the original assumption of irreducible complexity implied this, nothing new has been obtained via the calculation.
Nelson does the same thing here when he says “I would say that the system is irreducibly complex. We can now apply Dembski's Porig X Plocal X Pconfig calculation ...” If we begin by accepting Behe’s claim that irreducibly complex system can’t evolve as an assumption, and use that to produce an hypothesis and its associated probability, then of course we can prove that the system is designed (because if something can’t have happened by law-and-chance, design is the only other option.)
Let me try to make clear the larger point I am trying to make in this and the Evolution and Design thread. At this point, design theory lacks specifics that tie it to the real biological world. In the thread Evolution and Design I offered an hypothesis about how design might work so as to provide a concrete base upon which to explore where the concept of design might lead. Irrespective of whether one agrees with the assumptions I make there, I would hope people would see that moving towards some type of specificity about what design is (as opposed to what evolution isn’t) needs to be done.
The same principle applies here. I don’t think it furthers design theory to merely argue by assertion that evolution could not have produced certain things. It furthers design theory a bit to provide theoretical models for inferring design (i.e., notions of complexity as improbability, notions of specification, and so on). But those ideas eventually have to become grounded in real biology. If design theory is going to be based on showing the improbability of law-and-chance producing something, and convince the world of those claims, then it will need to show how such calculations can be done for both events which are evolved and events which are designed. Circular arguments which just disguise the assumption of improbability are not going to do the job.
IP: Logged
|
|
edmund
Member
Member # 206
|
posted 28. March 2002 02:02
I think that some important points have been brought up in this thread; in particular, I agree with Evan that ID needs to get down to brass tacks and calculate real-world probabilities.
However, I think the major points I was trying to make haven't quite been addressed, at least not as fully as I would like! My first point was that ruling out natural processes requires a very good understanding of those natural processes. Much of the material I've seen on ID consists only of logic and theory. Logic and theory are very necessary, but no amount of theory will allow us to rule out natural evolutionary processes if we don't know what those processes can do. In particular, I think the use of the word "complex" in CSI leads some people to conclude that biological systems, because they have a high absolute complexity, are automatically beyond the constructive ability of natural processes. This is an erroneous conclusion because improbability is properly measured relative to the known mechanisms at hand.
My second point, which I think is the really important one, is: what can evolution do, and how do we know it can do it? In particular, I was hoping to spark a discussion of the empirical evidence on this topic. There has been a lot of discussion about "irreducible complexity", but to my knowledge it has not been shown rigorously that evolution cannot build systems which appear IC-- in fact, some evolutionists have proposed seemingly reasonable evolutionary paths to IC. Right now we seem lost in this subjective "gray area" where people argue that "it does/doesn't seem likely to me." Those arguments don't have much force.
So I'd like to discuss the empirical evidence. I'll even make my challenge a little inflammatory: living organisms have a lot of characteristics which are classic "fingerprints" of natural selection. Unless ID theorists can show conclusively that non-intelligent processes CAN'T construct systems equivalent to those in living organisms, I think natural selection gets the benefit of the doubt. So what hard, irrefutable evidence do we have of the limits of natural evolutionary processes?
edmund
[ 28 March 2002, 17:31: Message edited by: edmund ]
IP: Logged
|
|
Paul A. Nelson
Member
Member # 26
|
posted 28. March 2002 07:53
Hi Edmund,
You raise interesting questions, but it is impossible to answer them at the level of abstraction you are using. "Brass tacks" requires, well, some actual brass tacks. Please post particular biological examples, with supporting documentation. Only then can the discussion proceed meaningfully.
IP: Logged
|
|
Mike Gene
Member
Member # 149
|
posted 28. March 2002 08:38
Edmund: So I'd like to discuss the empirical evidence. I'll even make my challenge a little inflammatory: living organisms have a lot of characteristics which are classic "fingerprints" of natural selection. Unless IDC's can show conclusively that non-intelligent processes CAN'T construct systems equivalent to those in living organisms, I think natural selection gets the benefit of the doubt. So what hard, irrefutable evidence do we have of the limits of natural evolutionary processes?
1. What are the fingerprints of natural selection?
2. That living organisms have a lot of fingerprints of natural selection merely means that natural selection has been involved in evolution. Yet being involved and being the "designer" is not the same thing.
3. Thus, in your desire to discuss the empirical evidence, I second Paul Nelson. Pick the most impressive example where it can be shown/demonstrated that random mutations and natural selection were the constructors. Don't confuse common descent/similarity with random mutation/natural selection.
IP: Logged
|
|
Drosera
Member
Member # 139
|
posted 28. March 2002 11:04
quote:
1. What are the fingerprints of natural selection?
Well, there are a large number of things that follow directly from the basic theory: reduced diversity in regions of the chromosome undergoing selection in a population; novel features come from the modification of older features (and this can result in designs that have aspects that appear "dumb" to designers with foresight like us); the "designs" of organisms will have the ultimate "purpose" of increasing reproduction of the genes of the organism (unlike human designs, which basically always have the purpose of performing a task for something else, usually the human designer).
These are just off the top of my head; there's a lot more stuff in molecular genetics that one could investigate regarding specific patterns of diversity along the chromosome, etc.
quote:
2. That living organisms have a lot of fingerprints of natural selection merely means that natural selection has been involved in evolution.
So I guess we agree that there are fingerprints of natural selection and that they are common? Why question #1 then?
quote:
Yet being involved and being the "designer" is not the same thing.
Right, and the fact that laser measurements of continental drift and direct observation of sea-floor spreading (which is all we can observe of plate tectonics on human time-scales) matches up with a large pile of historical evidence...doesn't mean that these slow processes resulted in the breakup of the continents.
Sure, it could be that aliens pushed africa and south america halfway apart and that the rest was accomplished by natural mechanisms.
C'mon, we're talking about an explanatory theory, the probable truth of which can be established without having the time machine you appear to require. If there's lots of evidence for the capabilities and actual action of natural selection, and none for other proposed causes like mysterious unnamed (and unobserved -- far more unobserved than natural selection, for instance) causes like intelligent designers billions of years ago, then the extraneous proposed explanation (intelligent design) is warrantless and eliminated as superfluous by Occam's razor.
quote:
3. Thus, in your desire to discuss the empirical evidence, I second Paul Nelson. Pick the most impressive example where it can be shown/demonstrated that random mutations and natural selection were the constructors. Don't confuse common descent/similarity with random mutation/natural selection.
Well, here's a stack of them I've seen brought up in recent online ID debates:
- nylon degradation in bacteria
- Sdic, a new gene evolved recently in Drosophila
- several examples similar to Sdic known from the Drosophila genome (which is very well-studied)
- numerous and varied cases of resistance to drugs and pesticides, an impressive one being the evolution of a 3-enzymes-required pathway for the degradation of pentachlorophenol
- arms races between e.g. poisonous predators and prey...certain fish-eating snails are a good example here I believe
...I am quite sure you can turn all of these up in a PubMed search (so please search there before accusing me of not posting references). One good place to start that mentions many of the above cases is this article:
quote:
Curr Opin Genet Dev 2001 Dec;11(6):673-80
Evolution of novel genes.
Long M.
Much progress in understanding the evolution of new genes has been accomplished in the past few years. Molecular mechanisms such as illegitimate recombination and LINE element mediated 3' transduction underlying exon shuffling, a major process for generating new genes, are better understood. The identification of young genes in invertebrates and vertebrates has revealed a significant role of adaptive evolution acting on initially rudimentary gene structures created as if by evolutionary tinkers. New genes in humans and our primate relatives add a new component to the understanding of genetic divergence between humans and non-humans.
Drosera
IP: Logged
|
|
|
|
edmund
Member
Member # 206
|
posted 28. March 2002 12:37
quote:
3. Thus, in your desire to discuss the empirical evidence, I second Paul Nelson. Pick the most impressive example where it can be shown/demonstrated that random mutations and natural selection were the constructors. Don't confuse common descent/similarity with random mutation/natural selection.
Actually, I was more interested in examples where it can be shown or demonstrated that natural evolutionary processes CAN'T build a particular system. Those are the examples which draw the limits of evolution (and, by extension, ID). I was hoping that someone on this BB would be able to provide such examples.
I would make two points here:
1) When I said that living organisms bear the "fingerprints" of natural selection, I meant essentially what Drosera said: quote:
the "designs" of organisms will have the ultimate "purpose" of increasing reproduction of the genes of the organism
Natural selection can generate only functions with this specification. I don't know of any complex structure or function in the biological world which does not have this specification. This would seem to be a strong, though indirect, piece of evidence for NS-- the fact that biological systems fit precisely the functional profile of products of natural selection suggests that they *are* products of natural selection, even if the mechanisms by which they were evolved remain murky.
2) That said, I think it would be useful to gather and discuss empirical tests of the efficacy of natural selection. The ones I have been most intrigued by are examples of evolving computer programs (unlike biological evolution, we can manipulate and observe these!). I'm thinking mainly of T.S. Ray's "Tierra" program, which seems to be able to produce unnervingly complex systems. I'm also thinking of the "genetic computer" of a few years ago, which was able to solve the notorious "travelling salesman" problem by a process which approximated natural selection.
Both of these are cases where a process known to be without intelligence (save for the intelligence which designed the environment in the first place!) solved problems which were difficult or impossible for a human analyst. Tom Ray remarks that his evolved programs took advantage of computer-programming strategies he didn't even know about, and the "travelling salesman" question is known to be NP-complete.
In short, stepwise selection processes without any immediate intelligent guidance *seem*, in these cases, to be capable of doing things which otherwise appeared to require a human or superhuman intelligence. Since arguments about CSI and IC are basically logic-based rather than dependent on biological details, can we take these findings as evidence that natural selection can solve intractable biological problems as well? If so, why is this a reasonable inference, and what implications does it have for evolutionary theory? If not, why not?
Some citations I have for Tierra are below. These are just what I happen to have around, and a Web search would probably show up others. I confess that it's been a couple of years since I've looked at the "genetic computer" literature, so feel free to correct me if my understanding of these issues is incorrect. Also feel free to add examples of what known selection processes have been able or unable to achieve; I envisioned this topic as sort of a collective effort toward solving these questions. Again, I was especially hoping that some of the ID theorists out there would provide empirical evidence for the *limits* of natural selection, since those examples would be the most exciting.
Ray, T. S. 1991. ``Is it alive, or is it GA?'' Proceedings of the 1991 International Conference on Genetic Algorithms, Eds. Belew, R. K., and L. B. Booker, San Mateo, CA: Morgan Kaufmann, 527--534.
Ray, T. S. 1991. ``An approach to the synthesis of life.'' Artificial Life II, Santa Fe Institute Studies in the Sciences of Complexity, vol. XI, Eds. C. Langton, C. Taylor, J. D. Farmer, & S. Rasmussen, Redwood City, CA: Addison-Wesley, 371--408.
[ 28 March 2002, 17:29: Message edited by: edmund ]
IP: Logged
|
|
charlie d.
Member
Member # 159
|
posted 28. March 2002 13:08
Add to Drosera's biological examples the evolution of "anti-freeze" proteins from disparate, independent precursor proteins with unrelated original functions in several species of arctic fish (e.g. from trypsinogen in arctic cod, lectin in herrings, etc).
Comprehensive reviews can be found here and here (access may be restricted); this is also a neat little paper.
[ 28 March 2002, 13:10: Message edited by: charlie d. ]
IP: Logged
|
|
Janitor@MIT
Member
Member # 125
|
posted 28. March 2002 13:22
To answer the question at the top, to the best of my understanding, complexity, specification, and information are all relative terms. An argument from ignorance naturally follows. (Which is not the same as an argument to ignorance.) My personal ignorance could be greatly alleviated:
“I think that some important points have been brought up in this thread; in particular, I agree with Evan that ID needs to get down to brass tacks and calculate real-world probabilities.”—Edmund
For a theory of design, calculating the “real-world probabilities” is a comparatively trivial exercise in heuristics. It’s not a “chance hypothesis.”
The theory in question, however, invokes dual “chance hypotheses.” Please supply the relevant probability calculations.
Thank you in advance.
IP: Logged
|
|
Drosera
Member
Member # 139
|
posted 28. March 2002 13:27
Thanks Charlie d., I forgot about that one.
The first paper you mentioned makes an interesting point about gene amplification (meaning, gene duplications being *selected*, an important point that is often forgotten):
Annu. Rev. Physiol. 2001. 63:359-390.
ANTIFREEZE PROTEINS OF TELEOST FISHES
Garth L Fletcher1, Choy L Hew2, and Peter L Davies3
quote:
AFP Gene Amplification
[AFP=anti-freeze protein]
While AFP diversity provides one of the best examples of natural selection in action, one of the hallmarks of this rapid response to environmental cooling is gene amplification (136). In those fishes (and insects) that show significant thermal hysteresis activity, and for which an AFP gene probe is available, the AFPs are invariably encoded by large gene families. Estimates range from 10 AFP genes in yellowtail flounder (74), 30–50 copies in winter flounder (73, 137), 80 copies in the wolffish (138), to 150 copies in a population of ocean pout from Newfoundland waters (75). Where extensive sequencing has been done at the protein, cDNA, and/or genomic levels in fish or insects, there is evidence for multiple isoforms, many of which differ by only a few conservative amino acid replacements. It is unlikely that these very similar isoforms have specialized functions because they are typically not well conserved from species to species (compare, for example, serum AFPs in the yellowtail flounder and winter flounder, or the wolffish and ocean pout). These small differences might be the result of genetic drift. There are, however, some isoforms that display greater divergence (50% identity) and even tissue specificity, such as the flounder skin AFP isoforms (mentioned above), which is an indicator of functional divergence. The skin isoforms in flounder are themselves encoded as a multigene family that is as extensive as the liver-specific serum isoforms (3). Another indicator of the rapidity of this evolutionary event is the tandem amplification of certain genes that accounts for the dominance of a specific isoform or sets of isoforms such as the HPLC-6 and -8 isoforms of type I AFP in flounder serum (73). These abundant isoforms are encoded by 1 kb genes in 7–8 kb tandem-direct repeats that may be kept relatively homogeneous by gene conversion and/or unequal crossing over (139). Gene amplification has been seen in many artificial systems where there is intense selective pressure to overcome an environmental insult, such as the treatment of mosquitoes with insecticide (140) and the poisoning of cells with methotrexate (141).
IP: Logged
|
|
|
Topic Closed
Printer-friendly view of this topic
