Questions for Sean Pitman

Jerry Smith · Feb 3, 2003

If even one of the double mutant E. coli had found any way whatsoever to use the surrounding lactose in their environment to some advantage, any advantage whatsoever, then natural selection would have aided that bacterium and its offspring so that over the course of time its offspring would replace those without this mutation. This never happened. Plain and simple, this never happened. What does this mean? It means that even if there are many different potential sequences with beneficial function, it takes a long time to find them if they are more than one or two point mutations away from what is already there.

This is where the rubber meets the road. Hall's experiment did not test to see how much the size of the "neutral gap" changed in the double mutant E. coli. To the best of my understanding, it did not even attempt to identify whether any new strains emerged that were more successful than the parent strain in a lactose enriched environment. Hall tested for a strain that would metabolize lactose - that is all.

Your interpretation seems to be that he showed not only the presence, but also the persistence of a limited evolutionary potential. Yet, he did not test for the persistence of such a limit. He did not test the "size" of the neutral gap that prevented the evolution of a new lactase in his double mutants. It could have been 2 mutations wide or 20. He did not test to see how much that gap changed over the course of his expirement. It could have been 15 mutations wide at the beginning, 3 in the middle and 11 at the end, for all we know. We do not have those results. What is to prevent evolution of other genes (with selective advantage for the organism, or just through drift), from bridging that "neutral gap"? What is the mechanism that prevents this?

Sean Pitman · Feb 3, 2003

Part 1 of 2 Reply

Originally posted by RufusAtticus
Again you make this mistake. Selection and mutation are not the only mechanisms that cause evolution. Genetic drift, migration, and the mating system are all important features. Mutations do not have to be guided by selection for evolution to occur. Drift can accomplish evolution quite nicely on its own.

 

If you are talking about neutral evolution (ie: neutral genetic drift), then you are correct. However, if you are talking about functional evolution (ie: the evolution of new/novel functions) then I think you are mistaken. Changing ratios of alleles in a particular population is not the same thing as the evolution of new functions in that population or gene pool. Functional evolution is the evolution of a new phenotypic function that was never in the ancestry of a particular organism before. Neutral evolution is a change in the genotype of a particular organism that results in no phenotypic change. Also, the concept of devolution is important. Devolution is a phenotypic change in an organism that arises as a result of a loss of genotypic information.

 

Gaps are not permanent, they are temporal. So what you view as a neutral gap today might not have existed 100,000 generations ago when the protein evolved into its current state.

Click to expand...

 

Anything "might" have happened in the distant past. Anything is statistically possible. The best that we can do today to determine what was likely to have happened in the past is to calculate the odds of particular events happening and how much time, on average, such events would require to take place. You can say that something "might" have happened or could possibly have taken place, but until you can give some specific examples of how these gaps are made as well as how brand new genetic functions can evolve via neutral drift, then, my friend, you are hoping on a star. Wishful thinking and "just so stories" are not science. They might sound good up front, but when put to the test, they often fall apart.

 

Furthermore, not all gaps will be crossed, but drift guarantees that some will.

Click to expand...

 

This is no big surprise. We both already agree that some neutral gaps have been crossed and documented in real time. Unfortunately these gaps have been very small gaps. If you know of some documented experiment in real time where larger gaps have been crossed, I would be very interested to hear about them. When I have presented this challenge to others, a common response was to tell me about nylonase evolution. 

The current example of nylonase evolution happened via a frame-shift mutation. Many think that this is much different from the evolution of other functions, such as lactase evolution, since lactase evolution was realized with a single nucleotide substitution.  Of course, this changed only one amino acid in the resulting protein. Obviously this is different from a frame-shift mutation that resulted in hundreds of changes in the resulting protein sequence. 

The fact of the matter is though, a single point mutation did cause the frame-shift and the resulting nylonase function was instantly realized. Given the genetic sequencing available to the particular colony of bacteria at that particular time, such a frame-shift point mutation was statistically likely to occur and to be selected as advantageous in a nylon rich environment. We simply do not know how long it took to get to this point.  It might have taken billions of years, or a few years of neutral evolution.  We don't know what the original ancestor sequence was for this bacterium.  We don't know if it was neutral or not.  What we do know is that the nylonase function was instantly realized with just one mutation. There were no neutral gaps crossed here that can be documented. 

However, if we did know the sequence of some ancestor bacterium that gave rise to this colony of nylonase positive bacteria, we could then calculate the average time needed to evolve this function again.  Likewise, knowing what we know now about the genetic sequences of this and other bacteria (as well as other living things), we can estimate how long, on average, it would take to evolve any particular function.

The problem is, wide gaps do exist between many functions and all other potentially beneficial functions.  And, these gaps in neutral function get wider and wider as the complexity of a given function increases. How then are these gaps crossed?

You suggest a very interesting hypothesis below:

 

Here is an example. A boy is in a boat on the east shore of a lake. To deicide whether he goes forward or not he tosses a coin. If it lands on heads he paddles one stroke forward. Otherwise he paddles one stroke back. If he is on the east shore and a tail comes up he goes home, similarly for heads on the west shore. There are only two places where he can stop his journey, the east and west shore. So what is the probability that he gets out on the west shore or the east shore? It is simply the 1 - (ratio of the distance he is from the shore to the total distance across the lake). (We can work through this in more detail if you need it.) If the boy does this every day of his life, then most of the days he will get out on the east shore. Some days however, he will drift across the lake and get out on the west shore. Try playing with this simulation to get an idea what drift can do. Here is a sample output; note how some drift across the gap and others do not.

Click to expand...

 

Again, changing the allelic frequency in a population is a far different thing from evolving a new/uniquely functional allele in that population in the first place. This algorithm is one that estimates that changing ratios of a neutral genotype in a given population. This algorithm has nothing to do with the crossing of neutral gaps of function between functional sequences. It only has to do with estimating what percentage of a population will have a particular sequence in their genetic codes depending upon a number of variables, such as the percentage of the population that first had this sequence, the size of the population, and the total number of generations that this sequence will be followed.

So, starting out with a neutral genetic sequence in 50% of the population, over time, that sequence will fluctuate, maybe even disappearing entirely from that population, or maybe even involving the entire population. If the sequence is part of a highly beneficial function within the population, that sequence will rise to and be maintained at nearly 100%. If a sequence happened to be detrimental in function, that sequence will drop out of the population at a rate directly proportional to the disadvantage conferred by that function. 

Now, in order to evolve a new function one of two things must happen. Either a genetic sequence must change via random mutation to become phenotypically beneficial, or the environment must change so that a sequence that is already there will become phenotypically beneficial. 

If the environment remains the same, then genetic sequences must mutate randomly in order to evolve a new function. This would be like your boy in the boat example, except for a few minor but significant changes. Instead of each paddle stroke representing a change in sequence frequency within a population, each paddle stroke would represent a mutational change in a single type of sequence regardless of how many individuals in a population had that sequence in their genomes (ie: sequence type vs. sequence frequency). What we are interested in here is the type of sequence, not in the frequency of that sequence in the population. Are we clear so far?  

http://naturalselection.0catch.com/Pictures%203/random%20drift%206.jpg

Now, with each stroke of the paddle, the boy can go east or west or south or north on the Pond of Neutral Sequences. He starts out on the South side of the pond where there just so happens to be a functional sequence. However, this sequence gets duplicated so that there is an extra sequence that can undergo random drift to see if any other functional sequences can be discovered on the pond. The boy sets out on his voyage. However, since he doesnt know that there are two other beneficially functional frequencies (for his needs in his environment) on the pond at the northeast and western edges of the pond, he paddles at random. Everywhere he goes, the sequences are all the same as far as beneficial function is concerned (ie: neutral). So, his boat travels randomly. By sheer good luck, he might end up at one of the beneficial functions rather quickly (red line), or, by sheer bad luck, he might not end up at either function for a very long time (black line). So, how long, on average, will it take him to realize either one of the beneficial functions? Certainly, because of his random course, the average time required to traverse the distance between the function that he started with and the new beneficial function, will be far greater than if he knew where he was going or at least had some sort of guidance. But, since he can only tell the difference between different functions, the various functionally neutral sequences that he comes upon give him absolutely no clue as to where to paddle next. So, he continues to paddle randomly around the lake until . . . until by blind luck he happens upon a new and functionally beneficial sequence.  On average, this take a long time and gets exponentially longer the farther apart the Islands of Beneficial Function are from each other.

Sean

Sean Pitman · Feb 3, 2003

Part 2 of 2 Reply

Actually it is statistically guaranteed that random chance alone will cause evolution. Unless a population is of infinite size, sampling during reproduction will cause fluctuations in the allele frequencies of a population, and hence evolution.

Actually, fluctuations in the frequency of an allele or genetic sequence in a population are not the evolution of a new genetic/allelic function in that population. For example, if you had a trillion bacteria and none of them were resistant to penicillin, they would all die in a penicillin environment . . . period. However, if this same population had just one bacterium that happened to evolve resistance just before the penicillin environment came along, the entire population would die except for this one bacterium. Now, the population would be 100% penicillin resistant. From one one-millionth to 100% just like that. If the selective advantage was not so great, but still high, one bacterium would give rise to offspring that would more gradually replace the offspring of the other bacteria in a steady state population until 100% would have the new advantageous function. Again, allelic frequency goes from one one-millionth of the population to 100% in fairly rapid progression. Is this evolution of new genetic function, or just the spread of a new functional sequence to the rest of the population? You see, as soon as a new beneficial function evolves, it will spread. The question is, how long, on average, will it take even one individual in a given steady state population of individuals to evolve any new beneficial function? If the beneficial sequence is already in at least one other member of a given population, then spreading that sequence around really isn't making anything new for the population . . . is it? The total functional capacity of a given gene pool is based on the absolute number of different functions it can provide, not on the percentage of individuals who carry these functions.

You do see the difference here dont you? You are talking allelic frequency while I am talking new allelic function in any one member of a population regardless of that population's size.

Drift. See the above example. You have dismissed random fluctuations but dont fully understand them.

Evidently you dont understand the difference between random fluctuations of functions in a population (that were already there), and the evolution of new functions that were never in a population before. You think your website illustrates a boy in a boat getting from one function to another, but it doesnt. What it illustrates is getting a certain genetic sequence from 10% to 100% of the population. This is a key point. Hopefully you understand it now.

Sean

 

End of Part 2 of 2 Reply

RufusAtticus · Feb 4, 2003

Sean,

Im going to try to keep this short because posts on this thread are getting way too long.

Evidently you dont understand the difference between random fluctuations of functions in a population (that were already there), and the evolution of new functions that were never in a population before. You think your website illustrates a boy in a boat getting from one function to another, but it doesnt. What it illustrates is getting a certain genetic sequence from 10% to 100% of the population. This is a key point. Hopefully you understand it now.

There is no difference. Every step in your neutral pathway is still an allele. That allele still behaved just like the one in simulation you played with. Neutral nucleotide polymorphisms do drift around in populations on one axis, the axis of frequency. From the populations perspective the only difference between a novel neutral allele and a neutral allele that is already present is the frequency. So the frequencies of alleles floating around in neutral function space are still governed by genetic drift. Substitution occurs when a rare allele replaces the most common allele. Drift will cause substitution at a rate equivalent to the mutation rate. (I can work out the theory if you need it.) Such processes can fix neutral alleles and provide pre-adaptation.

Changing ratios of alleles in a particular population is not the same thing as the evolution of new functions in that population or gene pool.

If your new function is the result of the evolution of a new gene, then it will change gene frequencies in the gene pool.

Neutral evolution is a change in the genotype of a particular organism that results in no phenotypic change.

No, neutral evolution is the evolution of a trait via drift. It has nothing to do with phenotypic change.

Also, the concept of devolution is important. Devolution is a phenotypic change in an organism that arises as a result of a loss of genotypic information.

LOL. There is no concept of devolution in biology. Change due to a knockout is still evolution. The fact that you think that there is such a concept means that you really need to go review the basics of evolution and genetics. I can recommend a few resources that my department uses to teach undergrads.

Unfortunately these gaps have been very small gaps. If you know of some documented experiment in real time where larger gaps have been crossed, I would be very interested to hear about them. When I have presented this challenge to others, a common response was to tell me about nylonase evolution.

Why must large gaps be crossed in real time? Models of molecular evolution involve exon shuffling and combining. For example, in Drosophila the jingwei gene is the product of retrotransposition of an AdH allele that captured upstream exons from a completely unrelated protein. You should be addressing the current state of the field of molecular evolution, not asking it to prove something that it doesnt even advocate.

Sean Pitman · Feb 4, 2003

Part 1 of 3 Reply

Originally Posted by RufusAtticus

The point is that your probability calculations are extremely flawed if you dont take domain shuffling into account. For example, you had previously ruled that it would be nearly impossibly to form the word "batter," but once you take into consideration recombination the impossibility disappears.

Evidently we have a very different understanding of how genetic recombination works. Genetic recombination is in fact capable of changing phenotypes very quickly over time, but it is clearly limited in how far it can go by itself. Genetic recombination can explain the differences that we see within many groups of creatures, such as within species groups and even some higher taxonomic groups (classification is fairly subjective). However, genetic recombination alone cannot explain the differences that exist between many groups of creatures.

Genetic recombination is found in creatures that use sex as a means of reproduction. Those creatures that do not use sex as a means of reproduction, such as bacteria, do not undergo genetic recombination events. Bacteria multiply by a means of simply cell division. In other words, all resulting offspring are clones of the parent except if there is some mutational event.

In those creatures that do multiply via sexual reproduction, genetic recombination is used to mix and match genetic sequences between chromosomal pairs. The pairs line up in a very precise way during this process and each pair trades sections of DNA with the other at precisely the same location. This tradeoff is random as far as which sequences will be traded, but it is not random as far as what portions of the sequences will be traded. The sequences are traded in specific units of information and the trade is perfectly balanced. In other words, no chromosome gets or gives more or less than it had before. Also, the order of information remains the same. A traded sequence that coded for eye color before the trade will still code for eye color after the trade . . . unless a mutation or error occurs in the process of recombination.

Often, a given site that codes for a particular trait, such as eye color, might have various options (alleles) that could occupy this position. These options might code for blue, green, brown etc . . . eye colors. However, they all code for the eye color trait. Some other allele coding for fuzzy fur, just will not fit in this particular locus. Why? Because the fuzzy fur code in this position creates a meaningless coding sequence that cannot be interpreted.

Genetic recombination does not recombine parts of alleles to make a new uniquely functional allele. Genetic recombination takes intact alleles (with the same function-type in a given position) and recombines them with other intact alleles in the same order that these alleles were in the first place.

Using English as an example, if the function of a particular position in a paragraph was to code for a "baseball bat", different alleles would be able to modify various aspects of the bat while still maintaining the bat function. There could be wooden, silver, or plastic baseball bats. Or, big, medium or small baseball bats. But, given the construct of the particular paragraph, mutating the function of the actual word bat into the word "batter" might not work so well. For example, if the paragraph was describing how to use a bat, the various qualities of base ball bats, the grips, the feel of the swing, etc. Substituting the word "batter" would not make any sense. Imagine saying, "Now, pick up the batter with both hands and choke up a little on the batter. Swing the batter around a few times to loosen up your swing. Take some time to get comfortable with your batter. Some prefer a metal better over a wooden batter . . ."

Anyway, you get the picture. Genetic recombination is very specific and trades intact units of information in a very balanced way. It allows for a great deal of variation, but this variation is not genetic evolution because no new allelic functions are produced. In other words, my brother and I look very much different from each other and even our parents, but these differences are not the result of any mutational event or genetic evolution. The differences are the result of genetic recombination differences, but genetic recombination has clear limits as to what phenotypes can be produced. For instance, using genetic recombination alone, a dog cannot be turned into a cat. Many different forms of dogs can be obtained via genetic recombination. In fact, this is what breeding is all about. However, no matter how strong selection pressures become, genetic recombination alone cannot turn a dog into a cat.

Many try and include changes that arise via genetic recombination into the theory of evolution, but really, there is no gene pool evolution going on with genetic recombination as far as the absolute number of functions and potential phenotypes available to that pool. Relying on genetic recombination alone, a gene pool can produce a huge number of different phenotypes, but not an infinite number (static/unchanged gene pool). Because of this limitation to the potential for change that genetic recombination can offer, the theory of evolution must rely on other sources for change that lie beyond genetic recombination (ie: random mutation combined with natural selection).

The fitness criteria say. The formation of "hodogt" or "catrater" might fail, but "batter" is successful. Therefore, you cant claim that evolution is impossible or improbable.

Your criteria of this analogy require that when a protein is formed it is recognized by the vocabulary. "Batter" is a valid English word. (And in fact, even if it was not part of the vocabulary before it would be instantly recognized as such by the rules of English grammar.) Thus when it was formed it was not formed in the neutral sea of your analogy. So thus you cant dismiss it by saying otherwise.

This is simply an amazing statement to me. Are you really suggesting that even if the word, "batter" were not part of the English vocabulary (ie: dictionary), that it would be instantly recognized as a functional word if it happened to be used, say, in a game of Scrabble? Let me tell you, that is not how language systems work. The words bat and batter mean very different things in the English dictionary. Consider the word "star". By your definition of how language systems work, adding "er" on to the word star would by definition make the word sequence of letters, "starer" understood by an English speaking person. So tell me, what does starer mean?

You see, symbols do not created their own meaning. Meaning is given to symbols by a language system or some other system of function such as the mechanical systems of a living cell. Symbols carry higher meaning than they, by themselves, bring to the table since they are part of a higher system of function.

Some people think that crystal formation is a type of spontaneous ordered complexity, but crystal evolution is not based on the evolution of symbols in a system of function where the symbols carry a higher meaning that is beyond each individual character. The information for crystal formation is entirely contained within each atom or molecule itself. It needs no higher or external source of information to know where it fits in this structure. Not so with symbols. Throw a bunch of letters out on the table, and they wont spontaneously arrange themselves in a meaningful order of higher function.

Oh, and by the way, you evidently did not understand my explanation as to why it was so easy to evolve "new" two and three letter words in the English language. You do know why dont you? One reason is that the definition of the words was not evolved, just the pre-defined sequence. Without previous definitions being assigned to these sequences, their evolution would have been meaningless. In other words, there was already a pre-established system of function waiting for the right sequence to come along.

And that is exactly why "batter" is instantly functional.

Yes, that is why. The system was already there, in place, before the particular sequence came along. Without the system in place before hand, the sequence itself carries no inherent meaning. Without a definition waiting for "batter" in the dictionary, the evolution of "batter" in any position of a sentence or paragraph would make no sense. It would be functionless. This is not to say that just because a particular sequence is defined that it will be beneficial in a given position. It might have function itself, but it might destroy or hinder the function of other parts of the larger system of function that it happens to mutate into.

So, the problem comes back around to, "What are the odds that the proper sequence will be there when it is needed?" It might have come around yesterday, but if it is not here today when I need it for my current environment, what good was it if I had it yesterday when I didnt need it?

Sorry, but that is the teleological straw-man of evolution. The appropriate question is "what are the odds that some sequences will be around when the environmental conditions will select for them?" Evolution isnt about the "proper" sequence getting selected for. It is about any sequence getting selected for. There are many examples in nature of selection producing adaptation, the nylon digesting enzymes are one that come to mind right now. You can calculate all your hand waving probabilities that you want. But the fact remains that we do observe evolution both morphologically and molecularly.

I am using the word "proper" in this context to mean "beneficial". What are the odds that a beneficial sequence will be there when it is needed? It might have been there before, several million years ago, but what good is that to me now? You see, if the sequence isnt selectively advantageous now, nature will not keep it around in hopes that some future environmental shift might make it useful. The evolution is nylonase is a fine example of this problem. What if this sequence evolution had taken place a hundred million years ago when there was no nylon or any other environment that gave it some beneficial function? Over the course of time it would have been mutated out of existence. No more nylonase. Then, suddenly, when it is needed, it is not there.

The fact that it was only one point mutation away from function when it was needed was indeed fortunate, but not necessarily helpful in demonstrating how neutral gaps can be crossed in a reasonable amount of time. We have no idea what the ancestral sequences of this gene were so their changes cannot be followed over time. We also dont know how many other sequences could produce a nylonase enzyme. Therefore, we dont know how simply such a function can be coded for (how small the ocean is between what is there and this island of new function).

So, the fact remains that we do observe molecular evolution and the changes in morphology that result, but the molecular evolution that we do observe in real time never crosses gaps in neutral function that are more than one or two mutations wide. Not very impressive.

Sean Pitman · Feb 4, 2003

Page 2 of 3 Reply

 

Originally posted by RufusAtticus

Also, human languages and codes evolve easily . . . no doubt about it. That is because humans have the powers of high intelligence and reason.

There is no reason behind language change, only error.

You need to study historical linguistics. Language change is random. It is an unconscious process whereby mistakes made in acquisition replace the previous paradigm. Drift and migration are important parts of it. So is selection, since all language changes are not equal.

But symbolic mutation and symbolic selection do control language change and diversification.

Language change is not a thinking process either.

Computer code can be evolved too. The computer science field of genetic programming is doing some phenomenal work. In fact I use genetic programming to identify rules that govern the classification of data. It is much more powerful than any thinking process.

Simply amazing. So, human languages and even computer software evolves the same way that genetic codes evolve. Why didnt I think of this before?

Come on now, youve got to be joking. Do you really believe this?

I have done a bit of reading about historical linguistics and have learned enough to know that human languages, to include writing as well as other forms of communication, do change and evolve new and unique functions over time. However, the changes in meaning or function never result from a change in the symbol itself. Changes in symbols never create their own meaning in human language evolution. Some societies may drop a letter or change a portion of a symbol or word from what it was before and yet still maintain its former understood meaning. Or, an entirely new idea may be attached to a word or symbol that used to have a different meaning entirely. All of these changes are, however, based on human need and agreement, not upon random mutation that evolves meaning for words outside of human intelligence and direction. Certainly, human language evolution is not based on any naturalistic processes like biological evolution is supposed to be based on.

So, your statement that, "The change from Latin to French was not by design, but by mistake" shows a lack of understanding about how human language evolution is different from genetic evolution.

Also, your notion of computer evolution is also way out there. If software could evolve, why do we need computer programmers? The problem with computer software evolution (via mutation and selection) is very much similar to the problem that blocks genetic evolution. If you change a zero or a one in computer code, this change does not always result in some change in computer function. The ones that do result in a change in computer function generally result in a non-beneficial change (ie: a program or a portion of a program stops working = devolution). Further changes to this mutated portion of software code will most likely be neutral (ie: It will remain non-functional. The comparison between two non-functional computer code sequences is a neutral comparison). How then would mutations to a particular software program, say, a photo editor, evolve via random mutation and natural selection into any other unique computer software function, such as a calculator program or a screen saver program, if the selection method was only phenotypic? In other words, based only on what changes you see on your screen, you must select between changes in zeros and ones (genotypes) that occur in your computers software codes. Im betting that using this selection method combined with random mutations to zeros and ones in your computers software, that very soon you will be left no software at all, much less improved software of various kinds.

As long as the number is greater than one, you cant claim that it is impossible to select between two islands.

I never said that it was "impossible". Nothing about the theory of evolution is statistically impossible. However, the statistics do say that the theory of evolution doesnt have enough time to create the various systems of independent function that we see all around us. You think 4 or 5 billion years is long enough? This is a drop in the bucket as far as the trillions upon trillions upon zillions of years needed to cross gaps in neutral function that average just a dozen or so steps across in an average sized sequence.

So, the number of islands may be greater than one, but if it takes zillions of years to find the other island on average, what good does that do you?

You say that, "Evolution doesnt require a time limit, so placing one on it is ridiculous." However, the theory of evolution does have a time limit of 4 billion years or so. If you think this is enough time for any significant evolution to take place . . . think again. This isnt enough time to cross even one gap in neutral function that averages just a few steps in width.

Changing environments really dont help the odds out too much as far as I can tell.

They just happen to show that your probability calculations are bust.

How so? Changing environments dont help if on average they move away from potentially beneficial sequences as far as they move toward other potentially beneficial sequences.

For example, lets say that I have a computer user who really wants computer software that helps him work with pictures. Any changes that he sees on his screen that tend toward photo editing software, he will select for these changes above any other changes. In fact, he dislikes calculator programs so much that he selects for anything that gets rid of calculator programs. Lets say that one day some coded sequences are really getting close to making a functional photo-editing program. But, just before this code evolves function, a different user comes on the scene who really likes calculator programs but cant stand photo-editing programs. All of a sudden, the photo-editing program starts to work. However, since there has been a change in the "selector" or "environment" this new function is no longer considered beneficial, and it will soon get mutated out of existence.

 

Sean Pitman · Feb 4, 2003

Page 3 of 3 Reply:

Originally posted by RufusAtticus

Sean,

Im going to try to keep this short because posts on this thread are getting way too long.

Sorry, but I want to give you your moneys worth if I take the time to respond. Often, if I do not cover most or all of a persons argument, they accuse me of sniping the most vital parts of their argument on purpose so as to avoid answering/dealing with them. So, if I do find a particular rebuttal interesting enough to take the time to respond to it, and if I do happen to have the time, I try to respond to all those points I actually find interesting.  Your arguments here happen to interest me today.  You don't have to read them or respond to them you know.  No one is twisting your arm here.

Evidently you dont understand the difference between random fluctuations of functions in a population (that were already there), and the evolution of new functions that were never in a population before. You think your website illustrates a boy in a boat getting from one function to another, but it doesnt. What it illustrates is getting a certain genetic sequence from 10% to 100% of the population. This is a key point. Hopefully you understand it now.

There is no difference. Every step in your neutral pathway is still an allele. That allele still behaved just like the one in simulation you played with. Neutral nucleotide polymorphisms do drift around in populations on one axis, the axis of frequency. From the populations perspective the only difference between a novel neutral allele and a neutral allele that is already present is the frequency. So the frequencies of alleles floating around in "neutral function space" are still governed by genetic drift. Substitution occurs when a rare allele replaces the most common allele. Drift will cause substitution at a rate equivalent to the mutation rate. (I can work out the theory if you need it.) Such processes can fix neutral alleles and provide pre-adaptation.

You just dont get it do you? We are not talking about the frequency of alleles in a population. We are talking about the absolute number of uniquely functional alleles in a population. We are concerned here with the evolution of FUNCTIONAL phenotypes. How do the changing ratios of various non-functional or even functional alleles bring any new function to a population/gene pool? If one individual has a given genetic sequence or all of them have this sequence, how does this explain the evolution of the function of this sequence in the first place?

You say that "there is no difference" between the frequency of expression of a particular function in a population and the arrival of that function in the population in the first place? Is that really what you think? Look at your graph that you posted. What does it mean if the allelic frequency falls to zero? It means that there is no one in that population that has access to that particular genetic sequence . . . right? So, how does that population get that sequence back again? How long on average will it take to get that sequence back again?

That is the question. How does changing the relative percent of an allele (neutral or functional) in a given population evolve anything new? Can you turn a lizard into a bird this way? If you dont have feathers to begin with, how does shifting around the non-feather coding sequences that you do have, get you feathers? However, if you evolve a feather coding sequence at just the right place in just one individual, this sequence can be quickly spread throughout the breeding population. The same is true of the lactase enzyme evolution experiments of Barry Hall. If you dont have the lactase coding sequence in any one member of your population, how long will it take you to get it to evolve in any one member of that population (not 10%, not 50%, not even 100% . . . just one individual)? Well, it depends on how many neutral steps must be taken to travel from what is there in the combined population to a lactase producing sequence in just one member of that population. That is the problem. That is the difference.

As an aside, your statements that, "Substitution occurs when a rare allele replaces the most common allele" and that "Drift will cause substitution at a rate equivalent to the mutation rate" is not exactly clear. If a given individual has a "rare" neutral allele/genetic sequence on one chromosome, how does this sequence increase in frequency in the population? During genetic recombination it is passed on to an average of 50% of the gametes. Depending upon its proximity to other beneficial sequences, our neutral sequence will be more closely linked or less closely linked to selectively advantageous genes. If it is more closely linked to these beneficial sequences, then our neutral sequence will increase in the population, thus substituting itself for more common neutral alleles in that population. If it is really close to one of these beneficial sequences, it will seem to become "fixed" in the population . . . for a time. However, random mutations will gradual change the neutral sequence in the population over time while leaving the beneficial sequence relatively untouched. Why? Because the neutral sequence does not have the stabilizing protection of natural selection. That is why selectively neutral sequences are thought by many to be of some benefit as "molecular clocks" to estimate times of ancestral divergence based on direct or indirectly calculated mutation rates. So you see, the substitution rates of "neutral alleles" in a population are not necessarily "equivalent" to nucleotide mutation rates even if they are closely linked to beneficial alleles.

Changing ratios of alleles in a particular population is not the same thing as the evolution of new functions in that population or gene pool.

If your new function is the result of the evolution of a new gene, then it will change gene frequencies in the gene pool.

Yes, once a gene with a beneficial function does in fact evolve in at least one member of a given population, the idea is that the offspring of this individual will gradually increase in relative numbers over time. This is the whole, "survival of the fittest idea", which does work. However, the changing frequency of genes or alleles in a population is not the main problem or the real issue at hand. The problem is how to get a particular function to be realized in at least one member of a given population using whatever naturalistic means are available.

Neutral evolution is a change in the genotype of a particular organism that results in no phenotypic change.

No, neutral evolution is the evolution of a trait via drift. It has nothing to do with phenotypic change.

Uh. Neutral evolution has a lot to do with phenotypic change, or at least the lack thereof. That is why it is called "neutral" evolution. The name "neutral" comes from the fact that many genetic changes do not result in any selective advantageous or harmful phenotypic change. Thus, these changes are selectively "neutral", being neither harmful nor beneficial to the organism. Now, a neutral change creates a new sequence genetic sequence non-the-less. This new sequence has "evolved" but it does nothing different from what came before. This neutral sequence may increase in the population and one day become "fixed" in the population . . . for a time.

Also, the concept of devolution is important. Devolution is a phenotypic change in an organism that arises as a result of a loss of genotypic information.

LOL. There is no concept of devolution in biology. Change due to a knockout is still evolution. The fact that you think that there is such a concept means that you really need to go review the basics of evolution and genetics. I can recommend a few resources that my department uses to teach undergrads.

The fact that there is no concept of "devolution" in popular biology doesnt mean that such a concept should not exist in biology. I realize that popular biology considers all genetic changes to be "evolution." However, in my opinion, destroying something is far different than creating something. In popular understanding, the word "evolution" no longer means, "any change", it means, "improved change". So, to make things more clear, the use of the word devolution might be helpful to distinguish the gain of genetic information from the loss of genetic information.

Some may be surprised at this, but many changes that are considered to be examples of evolution in action are actually caused by a loss of genetic function, not a gain of new or unique genetic functions. For example, the loss of eyes in cavefish or the loss of pigment production in albinos is not the result of a gain of some newly functional gene or group of genes, but occurs when mutations destroy the function of an existing gene or genes. These examples of "change" should really not be used as examples of evolution in action as they so commonly are since such changes cannot explain the theorized "change" bacteria into all that we see around us today. These are different processes entirely and should be given different names.

Sean

Jerry Smith · Feb 4, 2003

Bacteria multiply by a means of simply cell division. In other words, all resulting offspring are clones of the parent except if there is some mutational event.

Please don't build a whole argument and leave out plasmid transfer from your list of means by which genomes can be reshuffled!

I'll leave the rest for Rufus to reply to. He seems to be much more engaged with this discussion than I. I will suggest, however, that you take some time to research language evolution and genetic algorithms used in computer science. I believe you have some mistaken ideas about both.

RufusAtticus · Feb 5, 2003

Sean,

You are right. We do have different understanding about recombination and genetics in general. Thats because Im a geneticist and much more familiar with the subject matter then you are. I can recommend a couple of undergraduate textbooks to clear up all these misconceptions you have. It might be simpler, though, if you sit in on both a course in molecular genetics and a course in population genetics and evolution at your nearest research university. For your education Ive included a table about mutations and polymorphisms at the end of this post.

First off, bacteria do reproduce sexually. The most common form is conjugation. That is where a plasmid or even a chromosome is transferred from one bacterial cell to another through a tube called a pilus (Latin for hair). This has been known for sixty years. A specific plasmid, the F plasmid, is responsible for this. F+ bacteria form the pilus and act as the donor cell to an F- cell. The process transfers the F plasmid and any genes that happen to be on it to the recipient cell. However, in some cases the F plasmid recombines with a bacterial chromosome and produces whats called an Hfr cell. Such cells actually have the F plasmid integrated with the chromosome which allows chromosomal genes to be transferred during conjugation.

Recombination does happen in bacteria. One of the earliest studies on recombination was done using bacterial phages which can recombine during an infection. In fact, recombinations primary role in bacteria is to repair DNA damage. One essential protein in the process is RecA. This protein is responsible for the invasion of the single-stranded whisker into the double helix of the other DNA molecule. Studies have shown that colonies with RecA knock out mutations do not survive long because their effective mutation rate increases and they are unable to repair their chromosomes.

Just like replication, recombination is not perfect; mistakes do happen. Unequal crossing over is an important factor in shaping genomes. A simple example is regions of chromosomes known as minisatellites these regions are composed of repeats of ten to forty base pairs in length. Because of the uniformity, recombination machinery has difficulty aligning the DNA in these regions and makes many mistakes. Thus when crossover occurs, minisatellites can gain and lose repeats. This leads to a lot of polymorphism in populations with respect the number of repeats in a minisatellite. Each minisatellite can contain anywhere from tens to thousands of repeats of the sequence. That is the reason why they are used in DNA fingerprinting; as few as six loci are enough to distinguish one individual from anyone else in the world. Unequal crossing over not only affects minisatellites but it also affects coding genes, where it can cause gene duplications to occur. So you are clearly wrong to say that recombination exchanges intact genetic material in a balanced way.

Recombination is the reason why gene families are found in clusters on chromosomes. It also has a cumulative effect so that the more genes in a gene family found together the more likely recombination will mess up and cause another duplication. One gene duplication occurs you can have divergene for novel functionality. One such example is are the embryonic and fetal beta hemoglobins that are expressed during development. The fetal hemoglobin is significant because it has a higher affinity for oxygen then adult hemoglobin which allows a fetus to absorb oxygen through the placenta. Duplication due to unequal crossing over was a key step in allow a novel function to evolve.

Additionally, chromosomal translocations and inversions can also cause rearrangement of genes. Transposition, the physical jumping of a segment of DNA from one position to another, is another important factor influencing genome evolution. In fact, the more genomes we sequence the more we come to realize that transpositions play a major role in exon shuffling and the creation of novel alleles. I suggest that you read up on the jingwei gene in Drosophila since it is a classic example of novel gene creation due to retrotransposition and foreign exon recruitment.

An allele does not code for eye color because it is at the eye-color locus. It codes for eye color because it produces a pigment and has regulatory regions around it that cause it to be expressed in eyes. Genes can move around a lot in a genome and still exhibit the same functionality. Ones under a lot of regulation are not as versatile unless their regulatory regions tag along too. In fact, if your comments about loci were accurate we couldnt make transgenic organisms. Yet we do, and they do work. The only thing tying a locus to a phenotype is evolutionary history. The reason why the same locus has brown and green alleles is that at one is descended from the other. Isnt it great how evolution works?

Given the abundance of adaptations the evidence for selection is pretty clear that it is highly likely that an allele will be present during a period that will select for it. Mutations are not one time events, they happen over and over. In fact that is one reason why there can be such a thing as mutation-selection balance. That is where the mutation rate of allele allows it to persist in a population even if it is being selected against. You have argued that evolution cannot cross neutral gaps more than a couple of nucleotides large. But the genetic data doesnt bear it out. Plenty of neutral substitutions occur; that is why there is such a debate in molecular evolution between neutralists and selectionists.

Neutral substitutions occur at a rate equivalent to the mutation rate for neutral genes. To clarify things an allele is neutral if its relative fitness is the same as other alleles. It can cause phenotypic change but if that change is inconsequential to reproductive success it will be neutral. If having hard ear wax makes no difference in my ability to reproduce over people with smooth ear wax then the both traits will be neutral. We know from coalescent theory that the probability of fixation for any copy of an allele at neutral locus in a diploid individual is 1/2N, where N is the number of individuals. This is because one copy has to win, i.e. become fixed, and each one is as likely to win as any other one. Thus the probability of winning is simply 1 over the number of contestants. This works out to the simple equation of 1/2N. Now the number of new neutral mutations arising every generation in the population is simply the number of genes times the neutral mutation rate. This works out to be 2N*u neutral mutations per generation, where u is the mutation rate. Thus the rate of substitution can be calculated by multiplying the number of neutral mutations by their probability of fixation, which yields 2N*u*(1/2N) = u substitutions per locus (or nucleotide) per generation. Given that point mutation rates are around 1e-8 to 1e-9 per nucleotide per generation, the substitution rate for humans works out to be 3 32 nucleotide substitutions per generation. With that significant of a rate, it is easy to see that drift alone can accomplish much in as little as a million years. Perhaps if youd reveal your calculations, I could understand how you calculated that something as simple as two steps couldnt be done in four billion years, despite what the classical neutral theory and the molecular data reveals.

Furthermore, you cannot talk about the evolution of a gene without talking about fluctuations in allele frequency. Every new allele that enters a population will have a frequency in that population and that frequency will change. A population will acquire a new trait if it becomes fixed for that trait. In other words, if that unique trait is nearly ubiquitous in the population, i.e. virtually no polymorphism, then it can be considered a diagnostic feature for that population. The evolution of novel functional proteins become ubiquitous in a population is due to the fixation of the alleles coding for them. Both selection and drift can cause this. Their relative importance will depend on the strength of selection and the effective population size. If the path way to a functional allele involves neutral stages then it is possible that by drift alone not only will they stay in the population, but it certain instances the neutral stages can also become ubiquitous in the population.

RufusAtticus · Feb 5, 2003

As for the language example, you must really be reaching if you want to contend that the meaning of batter is not immediately clear to any English speaker who understands what a bat is or the action one does with it. When the first person coined the term batter did he need to explain to someone else on the sports team that it meant someone who bats? No, because the suffix -er signals such a meaning. That is simply English grammar. Do you profess to be so clueless about English grammar that Id have to explain what an emailer is something that emails?

Yes, both language and software developed using genetic programming do evolve like DNA. All you need for evolution to occur is an imperfect replicator. Language replicates by a new generation acquiring it in childhood from previous ones. This is an imperfect system since a childs language acquisition devise is not error free. Children do make errors in replicating the grammar and phonemes of their parents. (You do realize that a language is much more than its vocabulary; dont you?) Many times they will also acquire different meanings, whether connotations or denotations, for previously existing words. This has been well documented by developmental linguists. These errors are language mutations. For the most part, language change behaves similar Ohtas nearly neutral model for genes. However, there is a lot of horizontal transfer and migration in language when compared to most molecular data, at least from higher organisms. Despite your protests, the change of Latin to French was the result of these linguistic forces and not the product of speakers deciding to change their language. In other words, certain errors in language acquisition increased in frequency in the population and caused the shift of grammar and pronunciation that distinguished Latin and French. That is why it is accurate to describe it as the result of mistakes instead of decisions.

Genetic programming is a software development technique that uses biological principals to solve computational problems. In it, populations of software solutions are randomly generated and selection is applied to the individuals in the population. It is simple enough to get software to mimic mutation by introducing error. Similarly, recombination is also easily introduced into the replication algorithm. In fact, computer scientists have found that the rate of crossover is much more important in finding solutions than the mutation rate. This is amazing similar to the results we get studying the molecular data. In my own work, Ive found that genetic programs do a much better job than other methods for finding certain computational solutions that lie in intractable parameter space.

You argue that devolution should be a biological term because evolution is popularly seen as improved change. But scientific terminology is not governed by popular usage but by scientific usage. The public has many misconceptions about science, and if science would go into disarray if its terminology were to be dictated by common misconception. There is no evolutionary difference between the loss and gain of a feature. Both are the result of the same genetic mechanisms, the same demographic mechanisms, and the same evolutionary mechanisms. The distinction you would like to make simply isnt there. So if youre going to talk science, you need to use its terminology correctly. Appealing to popular opinion among the masses is no way to function on a scientific level. It works well for entertainment, politics, and religion, but inferi congelabunt, before the opinion of people who brought you Big Brother 2 and Tammy Faye, gets to dictate how science should operate.

Sean Pitman · Feb 15, 2003

4th February 2003 at 10:26 PM RufusAtticus said this in Post #49 <http://www.christianforums.com/showthread.php?postid=633319>

Sean,

You are right. We do have different understanding about recombination and genetics in general. Thats because Im a geneticist and much more familiar with the subject matter then you are. I can recommend a couple of undergraduate textbooks to clear up all these misconceptions you have. It might be simpler, though, if you sit in on both a course in molecular genetics and a course in population genetics and evolution at your nearest research university. For your education Ive included a table about mutations and polymorphisms at the end of this post.

Oh, I am unworthy! How can you, a post grad student and all, even talk to someone with such "undergraduate misconceptions"?

First off, bacteria do reproduce sexually. The most common form is conjugation. That is where a plasmid or even a chromosome is transferred from one bacterial cell to another through a tube called a pilus (Latin for hair). This has been known for sixty years. A specific plasmid, the F plasmid, is responsible for this.

Click to expand...

Ah yes, the F+ plasmid carrying bacteria. This is often referred to as a type of sexual transmission of information, but it certainly has little resemblance to gamete production and genetic recombination utilized during meiosis in eukaryotes.

Recombination does happen in bacteria. One of the earliest studies on recombination was done using bacterial phages which can recombine during an infection. In fact, recombinations primary role in bacteria is to repair DNA damage. One essential protein in the process is RecA. This protein is responsible for the invasion of the single-stranded "whisker" into the double helix of the other DNA molecule. Studies have shown that colonies with RecA knock out mutations do not survive long because their effective mutation rate increases and they are unable to repair their chromosomes.

Click to expand...

Certainly recombination does happen in bacteria. However, bacteria do not undergo meiotic recombination where an almost infinite number of different gametes are produced. "Bacteria are formally haploid and thus don't go through a meiosis that will generate recombinants. However, they can conjugate." In other words, bacteria can trade genetic information and incorporate this information into their own genomes. This is a type of sexual "recombination" but it is not the primary means of reproduction that one thinks of when considering bacterial replication. Generally speaking, and I think you would agree, bacterial replication is clonal. Not so with eukaryotic replication which is, for all practical purposes, never clonal.

In any case, it doesn't matter as far as our discussion and my point is concerned. Genetic recombination, by itself, does not create new function for a gene pool, be it bacterial conjugation or human reproduction. In other words, the information itself was already there, preformed. Thus, changes that are based on recombination alone should not be used as examples of evolution in action since the gene pool itself is not evolving in such cases. For example, the main differences between me and my brother are not based on mutation, but on recombination. So, where is the evolution of anything new? Likewise, the changing ratios of light and dark colored moths doesn't make anything new, but simply ups the frequency of an allele in a population. Again, where is the evolution of something new as far as a new function in that population. And yet, this example is used all the time as "evolution in action". It is nothing more than natural selection changing allele frequencies, but no new functions are evolving here. These changes need no mutations. And yet, evolution cannot explain the various functions that we see around us without mutations. So, examples that do not invoke mutations should not be used as examples of evolution in action in my book because nothing new is really made unless mutations are involved.

http://opbs.okstate.edu/~melcher/MG/MGW3/MG3514.html

Just like replication, recombination is not perfect; mistakes do happen. Unequal crossing over is an important factor in shaping genomes. A simple example is regions of chromosomes known as "minisatellites" these regions are composed of repeats of ten to forty base pairs in length. Because of the uniformity, recombination machinery has difficulty aligning the DNA in these regions and makes many mistakes. Thus when crossover occurs, minisatellites can gain and lose repeats. This leads to a lot of polymorphism in populations with respect the number of repeats in a minisatellite. Each minisatellite can contain anywhere from tens to thousands of repeats of the sequence. That is the reason why they are used in DNA fingerprinting; as few as six loci are enough to distinguish one individual from anyone else in the world. Unequal crossing over not only affects minisatellites but it also affects coding genes, where it can cause gene duplications to occur. So you are clearly wrong to say that recombination exchanges intact genetic material in a balanced way.

Click to expand...

We all know about unequal cross over. This is a type of mutational event. What I am saying is that even if everything went according to plan (no mutations or unequal cross overs of any kind), genetic recombination during meiosis would produce huge varieties in a population without any mutations of any kind being required. Now, since evolution of new information is dependent upon mutations, appealing to changes that can be explained using mutation free genetic recombination (ie: changing allelic frequencies/ratios in a population), does not help explain the evolution of new information. Your charts that demonstrate changing gene frequencies are a lame attempt to prove evolution by definition. Since evolution is defined by many to include changing frequencies of alleles in a population, you think that this solves all the problems, but it doesn't really solve anything as far as how to explain the evolution of new functions in a population. Another thing is, a loss of a function is not the same thing as the gain of a function, and yet often changes that are based on function loss are used as evidences of evolution in action. Well hey, I even believe in that kind of evolution. What I don't believe is that naturalistic mechanisms can explain the creation of many of the complex functions that I see in the natural world. I believe evolution can explain the creation of certain limited functions that are one or two point mutations away from sequences already available, such as the lactase function in certain colonies of E. coli and well as the nylonase function in other bacteria. But, beyond this, your arguments haven't explained much as far as anything I can understand.

I will say though that your arguments on nucDNA mutation rates were good. There seems to be much confusion with many as to exactly how many mutations occur between generations. I have been discussing many of your very good points in this area with a friend of mine (Dr. Zuccarelli - Ph.D. in biophysics from California Institute of Technology, Professor of biochemistry and microbiology). I have altered my paper on DNA mutation rates accordingly and am still in discussions with Dr. Z about this most interesting topic.

http://naturalselection.0catch.com/Files/DNA Mutation Rates.html

Recombination is the reason why gene families are found in clusters on chromosomes. It also has a cumulative effect so that the more genes in a gene family found together the more likely recombination will mess up and cause another duplication. One gene duplication occurs you can have divergene for novel functionality. One such example is are the embryonic and fetal beta hemoglobins that are expressed during development. The fetal hemoglobin is significant because it has a higher affinity for oxygen then adult hemoglobin which allows a fetus to absorb oxygen through the placenta. Duplication due to unequal crossing over was a key step in allow a novel function to evolve.

Click to expand...

Many such "examples" are used when trying to give examples of evolution in action. However, you never saw fetal beta hemoglobin evolve in any real time experiment, now did you? This is a "just so story". You think this is the way it must have happened, but no one has been able to demonstrate the evolution of such a comparable function in the laboratory. Gene duplication is a fine hypothesis, but it doesn't automatically overcome the neutral gap problem. Once a gene is duplicated it still has the same function that it had before it was duplicated. In order to evolve a new function, it must mutate. If the new function is more than one or two neutral mutations away, it may take a long time to get there since neutral changes cannot be guided by natural selection. It is fine to propose theories about how things might have happened, but this certainly isn't evidence that they did actually happen in this manner. Until you can explain the gap problem your duplication hypothesis is just as worthless as getting one gene to cross the gap. Now, you simply have more turtles on the same side of the highway . . . but no turtles on the other side yet. It is one thing to make more turtles on the same side of the highway, but it is another thing entirely to get the turtles across the highway that separates different functions.

Additionally, chromosomal translocations and inversions can also cause rearrangement of genes. Transposition, the physical jumping of a segment of DNA from one position to another, is another important factor influencing genome evolution. In fact, the more genomes we sequence the more we come to realize that transpositions play a major role in exon shuffling and the creation of novel alleles. I suggest that you read up on the jingwei gene in Drosophila since it is a classic example of novel gene creation due to retrotransposition and foreign exon recruitment.

Click to expand...

The jingwei gene does have the appearance of a chimeric gene and is interesting is that it lacks introns. It seems as though it is a functional gene, but its exact function is not yet known. It is thought that the gene was derived from a retrotransposed mRNA. All this is very interesting indeed, but I fail to see how this solves the problem. What function was produced by this chimeric gene? Was this function complicated? If so, what are the odds that this function would be produced by a given genome or gene pool. How long, on average, would it take to evolve this function in such a gene pool? The fact of the matter is that many chimeric genes exist and many can be demonstrated in real time. However, most of these give rise to diseases such as cancer etc. Getting one that is actually beneficial is not out of the question, but it is most likely rare and when it does occur, it most likely produces a relatively simple function.

You see, the questions still remain. And, for many functions, it appears as though naturalism simply cannot explain their existence. Too many variables are required before beneficial function is achieved, given a particular starting point. Certainly many function can be evolved in a reasonable amount of time, but these are severely limited in overall numbers. Blocks to further evolution are quickly realized in the form of neutral / nonfunctional gaps in function between what is there and what might be of benefit it it were to be realized.

 

Sean Pitman · Feb 15, 2003

Part 2 of 2 Response

An allele does not code for eye color because it is at the "eye-color" locus. It codes for eye color because it produces a pigment and has regulatory regions around it that cause it to be expressed in eyes. Genes can move around a lot in a genome and still exhibit the same functionality. Ones under a lot of regulation are not as versatile unless their regulatory regions tag along too. In fact, if your comments about loci were accurate we couldnt make transgenic organisms. Yet we do, and they do work. The only thing tying a locus to a phenotype is evolutionary history. The reason why the same locus has brown and green alleles is that at one is descended from the other. Isnt it great how evolution works?

The problem is that if the pigment produced by this allele happened to evolve in the cells that gave rise to the liver, it would be worthless. So, shuffling alleles around at random throughout a chromosome is not a good idea. They need to trade places between matching loci. If they did not line up this way during genetic recombination, there would be complete chaos.

Given the abundance of adaptations the evidence for selection is pretty clear that it is highly likely that an allele will be present during a period that will select for it. Mutations are not one time events, they happen over and over. In fact that is one reason why there can be such a thing as mutation-selection balance. That is where the mutation rate of allele allows it to persist in a population even if it is being selected against. You have argued that evolution cannot cross neutral gaps more than a couple of nucleotides large. But the genetic data doesnt bear it out. Plenty of neutral substitutions occur; that is why there is such a debate in molecular evolution between "neutralists" and "selectionists."

You still do not understand my point about neutral mutations. I believe in neutral evolution. It seems obvious and clear to me that many neutral mutations can and do occur over time. However, since the vast majority of potential sequences are neutral for a given population gene pool, getting from one functional to a different functional sequence might involve a fair amount of time sorting through a vast host of neutral sequences before a beneficial one was actually discovered. This sorting of neutral functions takes huge amounts of time with each neutral step that must be crossed. Just because a particular function might have evolved in a genome at a time when it was not needed, does not mean that it will evolve again when it is needed, even though, "mutations are not one time events." Mutations are not one time events, but particular sequences may be. A particular functional sequence may only come along once every trillion years in a particular gene pool. Now, the first time it came along, it might not have been needed (ie: blue pigment in the liver). What are the odds that blue pigment will be needed in the liver the next time this function comes along?

Neutral substitutions occur at a rate equivalent to the mutation rate for neutral genes. To clarify things an allele is neutral if its relative fitness is the same as other alleles. It can cause phenotypic change but if that change is inconsequential to reproductive success it will be neutral. If having hard ear wax makes no difference in my ability to reproduce over people with smooth ear wax then the both traits will be neutral. We know from coalescent theory that the probability of fixation for any copy of an allele at neutral locus in a diploid individual is 1/2N, where N is the number of individuals. This is because one copy has to win, i.e. become fixed, and each one is as likely to win as any other one. Thus the probability of winning is simply 1 over the number of contestants. This works out to the simple equation of 1/2N. Now the number of new neutral mutations arising every generation in the population is simply the number of genes times the neutral mutation rate. This works out to be 2N*u neutral mutations per generation, where u is the mutation rate. Thus the rate of substitution can be calculated by multiplying the number of neutral mutations by their probability of fixation, which yields 2N*u*(1/2N) = u substitutions per locus (or nucleotide) per generation. Given that point mutation rates are around 1e-8 to 1e-9 per nucleotide per generation, the substitution rate for humans works out to be 3 32 nucleotide substitutions per generation. With that significant of a rate, it is easy to see that drift alone can accomplish much in as little as a million years. Perhaps if youd reveal your calculations, I could understand how you calculated that something as simple as two steps couldnt be done in four billion years, despite what the classical neutral theory and the molecular data reveals.

The problem is that not all neutral mutations are equal. Some mutations might create a new function that was neutral and yet functional. Another mutation might create new junk that is not functional, but still neutral. For example, mutations involving previously non-functional DNA (ie: "Junk DNA") do not provide a new functional gene and yet they take no function away from the genome. Thus, they are neutral as well as non-functional. Neutral drift can thus cruise along for millions, billions, and trillions upon zillions of years, without ever coming across a particular beneficial function.

You ask me to show my math for my explanation of what is blocking lactase evolution in Hall's E.coli experiment. I thought my math was rather clear. In any case, since it is relevant to this scenario, I will post it again.

I propose that Halls experiment was a success only because the odds of blind chance "working" in such a case were statistically good. The odds of getting one, or even two necessary mutations "right" in a reasonable amount of time, even in a gene some 3,500 base pairs in length, is not unimaginable . . . since this gene is starting very close to the "goal" to begin with. As a hypothetical example, consider the following parameters:

· A stable population of one trillion (10e12) E. coli bacteria (equal to the volume of 10mL or an average hospital syringe)

· A bacterial generation time of 20 minutes (E. coli)

. An average bacterial genome of 4.1 million base pairs (E. coli)

· A mutation rate of 1 x 10e-4 mutations per base pair per generation. (Hall proposes that bacteria undergo "hypermutation" in starvation states. The average rate is much lower at about 1 x 10e-9 mutations per base pair per generation)

Given these parameters, how long would it take to evolve the first mutation? The odds that the correct position would be mutated in a E. coli bacterium would be 1 in 4.1 million (The math so far is based on there being 4.1 million base pairs in an E. coli bacterium's genome).

The odds that the correct base would get mutated at this location are 1/4 (Based on the fact that there are 4 based to choose from). The odds that both would happen are 1 in 16.4 million mutations (4.1 million times 4 = 16.4 million). With a stable population of one trillion and a mutation rate of about 410 mutation per genome per generation, the "correct" mutation will occur in about 25,000,000 bacteria in the first generation (Divide 410 by 16.4 million to get the odds of a correct mutation occurring in one bacterium and then multiply by the total number of bacteria in the population which is 1 trillion. The same sequence in used in the two, three and four mutation scenarios below).

But, what if the starting sequence was two mutations away from the desired lactase function? Would that make any difference? The odds of getting either one of the two needed mutations with the first mutation are 1 in 8.2 million mutations. The odds of getting the second mutation right are 1 in 16.4 million mutations. So, the odds of getting both mutations right are 1 in 134,480,000,000,000 (~1.34 trillion) mutations. With these odds, the correct mutation will occur in about 315 bacteria in the first generation.

If the starting sequence was three mutations away from the desired function, the odds of getting all three mutations right shoots up to 1 in 735,157,333,333,333,333,333 (~735 million trillion) mutations. With these odds, the correct mutation will occur in one bacterium in 1,792,682 generations. With a generation time of 20 minutes, that's a bit over 68 years.

Four mutations would take about 279,740,233 (~280 million) years.

These predictions also bear up in the laboratory. There are many bacteria that utilize galactose and glucose and yet they cannot utilize lactose. For example, practically all of the members of the bacterial type called Shigella cannot utilize lactose but they do utilize glucose. The same can be said for the many other bacteria such as Salmonella, Proteus, and Pseudomonas etc. All of these bacteria could in fact use a lactase gene if it became available to them. They would in fact be able to utilize the enzymatic activity of the galactosidase protein if they had access to it or to any other lactase enzyme. Why then do they not simply "evolve" any one of these possible lactase genes?

http://naturalselection.0catch.com/Files/Galactosidase Evolution.html

Furthermore, you cannot talk about the evolution of a gene without talking about fluctuations in allele frequency. Every new allele that enters a population will have a frequency in that population and that frequency will change. A population will acquire a new trait if it becomes fixed for that trait. In other words, if that unique trait is nearly ubiquitous in the population, i.e. virtually no polymorphism, then it can be considered a diagnostic feature for that population. The evolution of novel functional proteins become ubiquitous in a population is due to the fixation of the alleles coding for them. Both selection and drift can cause this. Their relative importance will depend on the strength of selection and the effective population size. If the path way to a functional allele involves neutral stages then it is possible that by drift alone not only will they stay in the population, but it certain instances the neutral stages can also become ubiquitous in the population.

Again, changes in allele frequency are not the same thing as the evolution of new alleles or new functions to begin with. Every new allele that enters a population will most likely start within one individual. Once even one individual has a new function, the gene pool of functions evolves by one function. The entire population need not have this allele in order for it to be part of the gene pool for that population. This is what I'm interested in . . . The evolution of a new function in even one member of a population. After that, changing frequencies of that function is a given. That is clear and easy to understand. These frequencies are subject to both selection and drift as you suggest. Natural selection is a real and powerful force of nature. It can and does cause changing frequencies of alleles within a population depending on changing environments etc. However, this is not the question. The question is entirely based upon the arrival of an new function in the gene pool to begin with. Distribution of a new function and the evolution of a new function are entirely different things. Hopefully you understand this because this is the real issue as I see it.

Sean Pitman · Feb 16, 2003

Part 1 of 2 Reply

4th February 2003 at 10:26 PM RufusAtticus said this in Post #50

As for the language example, you must really be reaching if you want to contend that the meaning of batter is not immediately clear to any English speaker who understands what a bat is or the action one does with it. When the first person coined the term batter did he need to explain to someone else on the sports team that it meant someone who bats? No, because the suffix -er signals such a meaning. That is simply English grammar. Do you profess to be so clueless about English grammar that Id have to explain what an emailer is something that emails?

This is really amazing.  You must honestly not get it.  The whole point is that changing symbols around does not, by itself, create definitions.  The function of a word or seqeunce is attached to it by an outside agency.  Adding "er" to "bat" did not by itself create the idea of "batter".  If this is such a tough concept for you to understand, perhaps using a different example might help, but I'm not holding my breath.

We have already done this before, but since you seem not to understand, consider again the word "cat".  Mutating one letter to the word "hat" did not create the meaning for "hat".  Likewise, in language evolution, the mistakes in spelling did not themselves create new information or concepts in the langauge.  The concepts came first and were then attached to various letter seqeunces.  Changes to these letter sequences did not change their meaning. 

Of course, as you pointed out, certain sequences can go together to make other combined meaningful functions.  However, not every sequence will work with every other sequence.  You must consider the odds that a change will result in a new a beneficial function if you want to compare human language to genetics or computer code.  What are the odds that the letters "er" would mutate into a particular sentence and result in a beneficial change in sentence function? 

For example, consider the sentence, "He has a very good bat."  The "er" sequence could get mutated into the phrase so that it read, "He has a very gooder bat"  or "Heer has a very good bat" or "erhe has a very good bat" or "he heras a very good bat" or a large number of meaningless options before you would get the meaningful phrase, "He has a very good batter."  You see the problem don't you?  There are a lot more phrases that are meaningless or at best neutral as compared to those that are beneficial.  In order to mutate, at random, the one that would be beneficial involves a lot of mistakes comming first that would not be beneficial.  These mistakes take time.  The more of them there are that surround a potentially beneficial change, the longer, on average, it will be before a beneficial mutation is realized.  Get it?

Yes, both language and software developed using genetic programming do evolve like DNA. All you need for evolution to occur is an imperfect replicator.

Click to expand...

This is complete BS.  Just try and explain how you can evolve a software program using random mutations and a selection method that only recognizes functional change.  You can't even evolve a simple computer program this way.  Now, you may be able to refine a particular computer program this way within a small range of function.  But, you will never evolve even a simple software program this way.  Try it and see.  You will be greatly disappointed. 

For evolution to occur, you need a lot more than an imperfect replicator.  You need a selection mechanism to guide the changes.  Natural selection is suppose to do this for biological evolution.  Computer programmers do it for computer evolution, and humans do it for human language evolution.  The problem that natural selection has that humans do not have is that natural selection does not have a vision of the future and cannot predict what non-fuctional / neutral changes might be of some future benefit if they are maintained.  In otherwords, natural selection is limited in that it can only recognized changes in function, not in symbols.

Language replicates by a new generation acquiring it in childhood from previous ones. This is an imperfect system since a childs language acquisition devise is not error free. Children do make errors in replicating the grammar and phonemes of their parents. (You do realize that a language is much more than its vocabulary; dont you?)

Click to expand...

When a child makes a grammer mistake, it is meaningless unless someone else agrees with that child that this mistake has meaning.  For human langauge to evolve, those who wish to communicate must all agree that a given change has meaning.  The change, by itself, does NOT create its own meaning.  It is given its meaning by the agreement of those who use it to communicate a particular idea.  This agreement may change, but it is the agreement, not the random change of a symbol, that creates new functions for various symbols.

Many times they will also acquire different meanings, whether connotations or denotations, for previously existing words. This has been well documented by developmental linguists.

Click to expand...

Certainly this is true, but how did these words "acquire different meanings"?  Did they acquire their meanings purely by random mutation of their sequences?  Certainly not!  They acquired their new meanings because of a change in the agreement of those who use words on what various words would mean.  This is also how codes are developed.  Ideas are attached to certain symbols according to the common agreement of those who wish to use these symbols to communicate secretly amoung others who do not know what the agreement is.  The symbols themselves, however, did not have anything to do with their own function in this code.

These errors are language mutations. For the most part, language change behaves similar Ohtas nearly neutral model for genes. However, there is a lot of horizontal transfer and migration in language when compared to most molecular data, at least from higher organisms. Despite your protests, the change of Latin to French was the result of these linguistic forces and not the product of speakers deciding to change their language. In other words, certain errors in language acquisition increased in frequency in the population and caused the shift of grammar and pronunciation that distinguished Latin and French. That is why it is accurate to describe it as the result of mistakes instead of decisions.

Click to expand...

   Man are you reaching.  Language evolution is based on common agreement.  If a small group of people start using a particular word to mean something, and then other people hear them and also start using this word to mean the same thing, and then other people hear from these people, the frequency of the function of this word will increase in the population.  However, this is not the same method that is used in biological evolution.  When a person hears others using a particular word, they recognized the association and can make the connection between arbitrary symbol and meaning.  Once they do this, they can join that group in the understood use of that word. 

Genetic simply does not work like this because genes cannot make the connection between arbitrary sequence and function.  In other words, genetic sequences cannot simply "learn" new functions.  A gene cannot look at another gene and say to itself, "Gee, I wish I had that function."  and then get it.  No, a genome must get new functions without learning or intuition.  Random mutations must be beneficial now.  No new meaning will be given to them just because some other cell somewhere else happens to use this sequence to some benefit.  If I mutate a gene that produces insulin so that the insulin function is no longer recognized, the new amino acid sequence that is produced may not have any function defined by that cell.  Unlike a human, that cell cannot arbitrarily give that sequence a new function either.  The sequence must continue to mutate until it match something within that cell's established system of function.  Until this time, it will wander around it its functionless state.  Depending upon how close it was to some functional sequence when it started on its wandering path, it may take a short time, or a very very very long time before it happens across something that will be recognized as beneficial in this cell's system of function. 

Again, this is nothing like human language evolution.  If you think that it is, then really, I don't think that there is anything that can help you to understand the problems with biological evolution until you can understand the difference between it and human language evolution.

Genetic programming is a software development technique that uses biological principals to solve computational problems. In it, populations of software solutions are randomly generated and selection is applied to the individuals in the population. It is simple enough to get software to mimic mutation by introducing error. Similarly, recombination is also easily introduced into the replication algorithm. In fact, computer scientists have found that the rate of crossover is much more important in finding solutions than the mutation rate. This is amazing similar to the results we get studying the molecular data. In my own work, Ive found that genetic programs do a much better job than other methods for finding certain computational solutions that lie in intractable parameter space.

Click to expand...

Certain computational problems can be solved via random mutation and selection.  The same is true for biological systems.  However, such functions must be reasonable close (one or two or three random mutations) away from potential solutions or even computer evolution will stall.  This fact limits computer evolution of certain functions to include even simply software development.  Try using these computer evolution simulation models to evolve even a simple software program when the selection mechanism is based purely on functional change recognition.  For example, I am willing to bet that even a simple calculator or screen saver program could not be evolved by such a evolution program.

You argue that devolution should be a biological term because evolution is popularly seen as improved change. But scientific terminology is not governed by popular usage but by scientific usage. The public has many misconceptions about science, and if science would go into disarray if its terminology were to be dictated by common misconception.

Click to expand...

The problem is that even scientists use examples of loss of information and the phenotypic changes that result as cases of evolution in action.  This would be fine if such changes could explain "gew to you" evolution, but it cannot.  The subtraction of functions often results in survival/reproduction benefits, but these changes do not explain how new benefits come on the scene, only how functions can be removed.  Even among scientists, there should be some way of labeling these two very different types of processes with different words. 

 

Sean Pitman · Feb 16, 2003

 

Reply 2 of 2 (RufusAtticus post #50)

There is no evolutionary difference between the loss and gain of a feature. Both are the result of the same genetic mechanisms, the same demographic mechanisms, and the same evolutionary mechanisms. The distinction you would like to make simply isnt there. So if youre going to talk science, you need to use its terminology correctly.

This is yet another difference between you and me. Science needs to change its terminology when new ideas come along that simply do not fit any established definitions. The current idea is that all changes are equivalent. You just said so yourself. You think that there is no difference between a loss and a gain of a function. But, there is a huge difference. It is a very simple thing to destroy a function, but it is another thing entirely to create a function. Random mutations can easily remove functions because there are so many non-functional sequences that surround a given function. But, it is a a lot harder to randomly evolve a new function starting with a non-functional sequence. . . or even a functional sequence.

For example, consider the sentence, "This sentence has meaning." Using random mutations alone, I could quickly change this sentence into, "asdoi, oithe aoihhe ehyrmcna" or any number of other non-functional phrases.  Now, how quickly could you change, "asdoi, oithe aoihhe ehyrmcna" into something that has meaning using a selection mechanism that only recognizes function? I am betting that it would take a fairly long time.

After all, even a child knows that breaking humpty dumpty is a lot easier than putting him back together again.

 

Appealing to popular opinion among the masses is no way to function on a scientific level. It works well for entertainment, politics, and religion, but inferi congelabunt, before the opinion of people who brought you Big Brother 2 and Tammy Faye, gets to dictate how science should operate. [/B]

Then, when scientists talk to the "masses", they should use terms that the masses can understand. Currently, when they talk to the masses and try and explain evolution, they equate two different processes as being the same . . . evolution in action. This is misleading because the scientists should know that they are often presenting a process of loss of information and yet they do not tell the public this. They present such changes as the same as a gain in information. At other times they present changes that are the result of changes in allelic frequency as "evolution in action" and give such changes the same weight with the public as they would changes that actually create new functions. Really, they should distinguish between these different types of changes because the public really thinks that they are all the same.

Also, not all scientists are clear on the matter. Even you think that all of these changes are all the same thing. But, they are not. How you could possibly class a loss of function as equivalent to a gain in function is beyond me. You think that a tree limb breaking a window is the same as making the window. It is understandable how you have been brought to this opinion though. How can you help yourself when this is what you have been constantly fed throughout your education by those who should know better? You see, calling every change "evolution" is confusing . . . even for scientists.

Jerry Smith · Feb 16, 2003

Pitman, Rufus is handling this exchange quite well, but on this point:

The problem is that if the pigment produced by this allele happened to evolve in the cells that gave rise to the liver, it would be worthless.

I challenge you to to think "germ line cells", and re-read Rufus' point. Your rebuttal fails to address the point. When you move any mutation to somatic liver cells or liver stem cells - even one that would be improve liver function, it is still worthless. Please don't allow yourself to be misdirected.

And on this point:

For example, consider the word the sentence, "This sentence has meaning." Using random mutations alone, I could quickly change this sentence into, "asdoi, oithe aoihhe ehyrmcna".

But could you? Remember you are only allowed one step at a time, that there are 1,000,000 copies of the original form of that sentence in the environment, and that selection for meaning is operating. Now if you gave each speaker in the population two copies of that sentence, you could possibly change one copy of that sentence relatively quickly to make it into the non-sense phrase you describe, but if you were to attempt to do so, there is no guarantee that you would not find yourself staring at group of speakers with both "This sentence has meaning" and "This sentence has self-referential meaning," and no non-sense phrase a few thousand generations down the road. There is a perhaps a slightly greater probability that you could make copy 2 become non-sense, but there is certainly no guarantee, when there is selective pressure present...

Sean Pitman · Feb 16, 2003

Today at 05:26 AM Jerry Smith said this in Post #55

I challenge you to to think "germ line cells", and re-read Rufus' point. Your rebuttal fails to address the point. When you move any mutation to somatic liver cells or liver stem cells - even one that would be improve liver function, it is still worthless. Please don't allow yourself to be misdirected.

You just don't get it do you?  I am thinking "germ line cells" when I am hypothetically proposing the movement of alleles to other locations affecting other types of cells, like liver cells.  You do realize that every cell in the human body has a complete copy of the entire genome (save for some such as RBCs).  Yet, even though each liver cell has a complete copy of the entire genome, it only uses a certain portion of that genome that deals with liver cells.  Now, those cells that make the human iris also have a complete copy of the entire human genome.  And yet, like the liver cells, they only use that portion of the genome that deals with the iris cell function. 

Now, if an allele, say for brown pigment production, from an iris coding region of the genome was copied and pasted into a region that coded for liver cell function, in a germ cell , it might not have any beneficial function when the next generation started using this region to make liver cells.  This allele did have function in the iris coding region of the genome, but when paisted to the liver coding region, it may have no funciton at all.  Get it?

For example, consider the sentence, "This sentence has meaning." Using random mutations alone, I could quickly change this sentence into, "asdoi, oithe aoihhe ehyrmcna". Or, any other non-functional series of letters.

But could you? Remember you are only allowed one step at a time, that there are 1,000,000 copies of the original form of that sentence in the environment, and that selection for meaning is operating. Now if you gave each speaker in the population two copies of that sentence, you could possibly change one copy of that sentence relatively quickly to make it into the non-sense phrase you describe, but if you were to attempt to do so, there is no guarantee that you would not find yourself staring at group of speakers with both "This sentence has meaning" and "This sentence has self-referential meaning," and no non-sense phrase a few thousand generations down the road. There is a perhaps a slightly greater probability that you could make copy 2 become non-sense, but there is certainly no guarantee, when there is selective pressure present...

Click to expand...

You don't seem to understand that random mutations do not need the guidance of natural selection pressure to destroy function.  They can do this quite easily.  However, to create new function, some sort of selection pressure is needed.  Getting from a functional phrase to a non-functional phrase is easy because there are so many non-functional phrases as compared to functional phrases.  A few steps in any direction are all that are needed to create "non-function".  Thus, a completely blind process, based purely on random chance, can accomplish the "non-functional" function.  Neutral function is also in the same boat.  It takes no selection pressure to create new neutral sequences in a rapid manner.  The problem comes when one tries to create new beneficial functions in a random manner.  Because of their relative rarity, this can't be done without the help of some sort of selection pressure.

This is the same reason why a window can be broken by a tree limb, but not fixed by the tree limb.  There are a lot of ways to break a window, but a relatively few ways to make one.  A great many changes can destroy its function, but only a few changes can restore its function.

If sequences with new functions where as common as neutral and non-functional sequences, then selection pressure would not be needed.  Random chance alone would be able to provide all the new functions.  The fact of the matter is, if mutations were purely random, evolution would be impossible.  Why?  Because purely random changes result in homogenous ooze, not in new functions.

Natural selection is suppose to provide the guidance required for the theory of evolution to work.  The problem, as I see it, is that natural selection only recognizes functional changes (Unlike human selection in language evolution.  Human selection can recognize non-functional changes and then give them meaning arbitrarily).  This is a problem because most mutations are not functional, but neutral.  If a mutation affects a functional protein, it may change the amino acid sequence of this protein into another sequence that does the same function or another function with a neutral advantage for a given creature.  In such a situation, nature cannot distinguish between the two sequences.  Both would be selected equally.  Likewise, if a mutation affected a non-coding sequence in the genome that had no function, the new sequence would also have no function.  Thus, the change would be a neutral change . . . beyond the seletive pressures of natural selection. 

Natural selection is basically blind to such neutral changes.  Because of this, these changes are entirely random and thus tend toward homogeny.  They may happen upon some new function, but this event is statistically rare, with increasing rareness as the complexity of the function increases.  Because of the rareness of such beneficial seqeunces, it takes a whole lot of random wandering, on average, to find them.  This wandering takes time; a lot of time; even more time than the theory of evolution has time for.

That is a problem.

Sean

Jerry Smith · Feb 16, 2003

Today at 12:54 PM Sean Pitman said this in Post #56

You just don't get it do you?  I am thinking "germ line cells" when I am hypothetically proposing the movement of alleles to other locations affecting other types of cells, like liver cells.

Ok, now that you are focused on germ line cells, please (as I asked) re-read Rufus' point:

 An allele does not code for eye color because it is at the "eye-color" locus. It codes for eye color because it produces a pigment and has regulatory regions around it that cause it to be expressed in eyes. Genes can move around a lot in a genome and still exhibit the same functionality. Ones under a lot of regulation are not as versatile unless their regulatory regions tag along too. In fact, if your comments about loci were accurate we couldnt make transgenic organisms. Yet we do, and they do work. The only thing tying a locus to a phenotype is evolutionary history. The reason why the same locus has brown and green alleles is that at one is descended from the other. Isnt it great how evolution works?

Your prior response (beginning "The problem is that if the pigment produced by this allele happened to evolve in the cells that gave rise to the liver, it would be worthless...") would only have made sense if you were talking about mutations in differentiated stem cells that lead to somatic cells, not in germ-line cells. Otherwise it does not speak to Rufus' point, or offer any reason to dismiss his fine rebuttal.

You don't seem to understand that random mutations do not need the guidance of natural selection pressure to destroy function.

On the contrary, in a population with selective pressure, a random mutation cannot destroy the function of an allele easily because it will be selected against!!

Therefore...

Getting from a functional phrase to a non-functional phrase is easy because there are so many non-functional phrases as compared to functional phrases.

This is incorrect. Because each of these non-functional phrases will be subject to negative selective pressure compared to the functional phrase that must continue to exist in the population.

Because of this, these changes are entirely random and thus tend toward homogeny.  They may happen upon some new function, but this event is statistically rare, with increasing rareness as the complexity of the function increases.  Because of the rareness of such beneficial seqeunces, it takes a whole lot of random wandering, on average, to find them.  This wandering takes time; a lot of time; even more time than the theory of evolution has time for.

That is a problem.

Sean

Rufus has already demonstrated why the "problem" is not actually a problem. I will let him continue with that. I just wanted to point out the two problems with your last post that I remarked upon in my last post. Obviously, they still stand. 

Sean Pitman · Feb 16, 2003

Today at 10:06 AM Jerry Smith said this in Post #57

Your prior response (beginning "The problem is that if the pigment produced by this allele happened to evolve in the cells that gave rise to the liver, it would be worthless...") would only have made sense if you were talking about mutations in somatic cells, not in germ-line cells. Otherwise it does not speak to Rufus' point, or offer any reason to dismiss his fine rebuttal.

Again, you simply do not understand.  My statement makes perfect sense as it stands because I was talking about germ-line mutations here.  The evolution of an allele in "cells that gave rise to the liver" makes perfect sense in the context of germ line evolution because germ line cells do give rise to liver cells. 

You are simply trying to come up with something, anything, however lame, to try and contradict something, however insignificant.  You have done this over and over.  Trust me, it's not helping your cause . . .

. . .in a population with selective pressure, a random mutation cannot destroy the function of an allele easily because it will be selected against!!

Click to expand...

Again, you miss the point of my argument.  Obviously natural selection does maintain beneficial functions in a population, preventing their loss.  However, natural selection is not needed to evolve non-functional sequences.  That is the point.  The evolution of non-functional sequences is easy because there are so many of them.  Such evolution of non-function does not REQUIRE the pressures of natural selection.  It is like breaking a window.  Such a change in function does not require any selection mechanism because there are so many options that give the same nonfunctional function. 

This is not the case with the evolution of new function which does require some sort of selection mechanism to act.

Rufus Wrote - An allele does not code for eye color because it is at the "eye-color" locus. It codes for eye color because it produces a pigment and has regulatory regions around it that cause it to be expressed in eyes. Genes can move around a lot in a genome and still exhibit the same functionality.

Click to expand...

Genes can move around, but their new location might not be beneficial.  You do understand this... right?  If a gene coding for brown pigment happened to get copied from those genes used by the iris and moved to an area of the genome that was used by liver cells, then the liver cells would also produce brown pigment.  However, this pigment production by liver cells might not be beneficial to the organism.

Because of this problem, the shuffling of genes around the genome is not a good idea.  During meiotic recombination, genes really do need to line up very precisely or chaos will result.  Genes simply cannot wander around the genome on a regular basis without lethal concequences to the organism.  Even when they do rarely wander around the genome, most of the time their wandering is either neutral or harmful, just as is the case with point mutations.  Beneficial functions may rarely arise due to this wandering (like adding "er" to "bat" to make "batter" . . . might rarely be beneficial in certain situations).  However, the average time needed for such beneficial wanderings to arise is relatively huge as compared with the time needed to obtain neutral or detrimental functional changes.

Rufus has already demonstrated why the "problem" is not actually a problem. I will let him continue with that. I just wanted to point out the two problems with your last post that I remarked upon in my last post. Obviously, they still stand. 

Click to expand...

Rufus is almost as confused as you are, but at least he comes across a bit better than you do.  The arguments that you come up with are simply amazing.  It is difficult to beat Rufus in this regard, but I think that if anyone can do it, you can. 

Jerry Smith · Feb 16, 2003

Today at 01:45 PM Sean Pitman said this in Post #58

Again, you simply do not understand.  My statement makes perfect sense as it stands because I was talking about germ-line mutations here.  The evolution of an allele in "cells that gave rise to the liver" makes perfect sense in the context of germ line evolution because germ line cells do give rise to liver cells.

 

In your mind, you may have been talking about germ line cells, but in your post, you said "cells that gave rise to the liver". Germ line cells do not give rise to liver cells. They give rise to gametes. Other differentiated stem cells give rise to the liver. Your post would make sense if these other differentiated stem cells had anything to do with evolution.

Now, in your mind, you may have still been talking about the different loci for liver cell proteins and retinal cell proteins. That is where you should have been responding to Rufus' explanation about how loci are only tied to pheontypic expression by evolutionary history, and not out of the necessity for a particular region of DNA to be expressed in a particular kind of cell. Instead, you ignored that and talked about mutations happening in cells that lead to liver cells. That is why I corrected you.

You are simply trying to come up with something, anything, however lame, to try and contradict something, however insignificant.  You have done this over and over.  Trust me, it's not helping your cause . . .

Given that you have not responded to Rufus' point, nor recognized or corrected the error that I pointed out, I can only conclude that your analysis above is skewed.

Again, you miss the point of my argument.  Obviously natural selection does maintain beneficial functions in a population, preventing their loss.  However, natural selection is not needed to evolve non-functional sequences.  That is the point.

Is this the point? Or is the next quote (below) the point? Because this part is not in error (though I do seem to miss the point, yes).

The evolution of non-functional sequences is easy because there are so many of them.

Now you miss my point. The evolution of non-functional sequences from functional ones is not easy because of natural selection. Selection prevents loss of function, except in cases where loss of function has a neutral or beneficial impact on reproductive success. My challenge was to your statement that you could easily evolve a non-sense statement from a meaningful one. My reply is - no more easily than you could evolve a meaningful statement from a meaningful one. Natural selection makes your job difficult, for as long as it operates.

Such evolution of non-function does not REQUIRE the pressures of natural selection.

But it is not IMMUNE to the pressures of natural selection. Do you understand now?

Genes can move around, but their new location might not be beneficial.  You do understand this... right?  If a gene coding for brown pigment happened to get copied from those genes used by the iris and moved to an area of the genome that was used by liver cells, then the liver cells would also produce brown pigment.  However, this pigment production by liver cells might not be beneficial to the organism.

Of course. Any mutation might not be beneficial to the organism. Evolution only requires that some are. There is strong empirical evidence that some mutations are beneficial to the organism, therefore there is no problem. 

Because of this problem, the shuffling of genes around the genome is not a good idea.  During meiotic recombination, genes really do need to line up very precisely or chaos will result.  Genes simply cannot wander around the genome on a regular basis without lethal concequences to the organism.  Even when they do rarely wander around the genome, most of the time their wandering is either neutral or harmful, just as is the case with point mutations.  Beneficial functions may rarely arise due to this wandering (like adding "er" to "bat" to make "batter" . . . might rarely be beneficial in certain situations).  However, the average time needed for such beneficial wanderings to arise is relatively huge as compared with the time needed to obtain neutral or detrimental functional changes.

Think about what you mean by "obtain" neutral or detrimental functional changes. If you mean that there is less time needed for some individual within the population to acquire a neutral or detrimental functional change, then you are correct but have no point. If you mean for a neutral or detrimental functional change to become fixed in the genome, then you are incorrect, because natural selection prevents this from happening over a long term. 

Rufus is almost as confused as you are,

LOL! Thank you..

Nicest thing anyone has said about me in a long time, even if it isn't true. 

but at least he comes across a bit better than you do.

Maybe because he is intimately familiar with the subject under discussion: population genetics. I have only a passing familiarity with it.

I just re-read part of the post I quoted above and I realized that you do think that germ line cells give rise to liver cells. I thought a definition would be helpful:

Germ Line:

The continuation of a set of genetic information from one generation to the next.

From

http://www.ornl.gov/TechResources/Human_Genome/glossary/glossary_g.html

RufusAtticus · Feb 21, 2003

Sean,

You in your post (#51) you said that Certainly recombination does happen in bacteria and yet in the post #45 you said bacteria, do not undergo genetic recombination events. So either you were grossly wrong before and are trying to slide around it, or you knew what you posted was wrong but did it anyway hoping that I would not call you on it. You are wrong to say that eukaryotic reproduction is almost never clonal because mitotic cellular division is certainly clonal. Plenty of eukaryotic species reproduce by both mitosis and meiosis, including multicellular ones. For example aphids will reproduce asexually for about ten generations and then have one generation of sexual reproduction before they go back to producing clones. You are seriously misrepresenting nature to claim that eukaryotes almost never reproduce clonally.

Changes due to recombination, especially unequal crossing over, do affect the gene pool of a population. Gene interactions are very important in establishing phenotypes of an organism. Thus a gene pool is not simply that allele A of gene B is found in x number of organisms. It is that allele A of gene B is found in x number of organisms with allele C of gene D, with allele E of gene F, etc. A gene pool is made up of the genomes of all organisms complete with genetic the associations and disequilibria. Thus recombination is very important in establishing it. Your characterization that recombination trades units based on functionality is false. It trades them based on sequence similarity. Alleles can be completely different functionality and still recombine. You only have to look as far as knock out mutations, which have no function yet can still recombine with other alleles.

You have claimed that recombination cannot add anything new to the gene pool of a population, but this is obviously wrong since recombination is a mutational event and mutations do add novelty to the gene pool. Most new genes resulting from duplication or exon shuffling that have been studied so far show elevated rates of molecular evolution over their parental genes (Long 2001, Curr. Opin. Gen. Dev. 11:673-680). The proportions of synonymous to nonsynonymous substitutions show that selection is acting on these loci. For example, the molecular evidence from the jingwei gene shows that selection began to operate on it from about the time of its origin and that selection is still operating on it. Such strong molecular evidence for selection wouldnt be present if these chimeric genes did not offer a new function to the gene pool. In plants, Long and colleagues have even found that the mitochondrial targeting function of cytochrome c1 was acquired via exon shuffling from GapC.

You keep claiming that there is an evolutionary distinction between loss and gain. The distinction that you want to make does not exist. Evolution happens by change and such involves both loss and gain. During evolution, species will both gain and lose features. For example, through evolution the ancestors of humans gained the ability to walk upright but lost the ability to walk on fours. Before that, tetrapods gained the ability to breath air but lost the ability to breath water. On the molecular level, the substitution of one nucleotide for another is loss of the old one and the gain of a new one.

You also do not understand the reason why you need to consider allele frequencies. You cannot separate out the evolution of a new function from the distribution of alleles because every step along the evolution of a new function is itself an allele and has a frequency. The only way to separate the two is to say that the protein must evolve in a non-stepwise manner. But then your calculations and arguments would not reflect what genes do in real life.

You have missed my point about the shuffling of pigment alleles, and I suspect that is because you dont understand what regulatory regions are. A gene is not a liver gene because of its location in the genome. A gene is a liver gene because it is expressed in the liver cells among other places. It gets expressed in liver cells because it has appropriate regulatory regions that are active in liver cells. Much of evolution involves changes in regulatory regions and not changes in the coding region of the gene itself. Selection produced recombination machinery that relies on sequence similarity to align chromosomes, not functional similarity since the machinery cannot judge the function of a region. Sequence similarity does have a secondary consequence of being faithful to functional similarity, but it is not always the case. Just look at gene families.

Now I will address your calculations which attempt to show that any function greater than three neutral mutations cannot evolve in a timely fashion. Your assumptions are flawed for what youd like them to prove. For instance you attempt to prove that for any one population of a trillion bacteria the expected time to reach a goal sequence four steps away is impossibly large for the average life span of a species. However, you forget that the actual population size of E. coli is millions or billions time larger then the population you were considering, thus even if every thing else you did is right, the expected time to evolution of the goal protein is at least six fold shorter then you conclude. However, that is not your only mistake. Your concentration on a single gene is also misleading. Yes there might be a low probability of this happening to any one gene, but considering the tens of thousands of genes existing in organisms it is surprising that a few genes would acquire the four or more mutation which they need to evolve a new function. Considering that even for a single gene there are multiple options for its functional future your argument is even less week. You also incorrectly assign the conditional probability of getting the goal nucleotide given that a mutation happens. It is one out of three not one out of four. Fourth, you neglect to realize that the situation you describe is stage structured. For example in the three mutation situation, you have four stages each one representing zero, one, two, or three goal mutations. The probabilities you calculate are only for going from zero to the final state in one generation. You neglect to calculate any of the other transition probabilities so it is impossible to scale up from one generation to another. I modeled the three mutation situation in MatLab and found that it takes only 30010 generations (~417 days) to get the first three-mutation individual, not 68 years as you calculate. That is even going a step further to consider backwards mutation and a lower mutation rate. Now here is the MatLab file I used to do this.

Code:

 % hall.m - Written by RufusAtticus
% Assumptions:
% 1. Effectively infinite population size (i.e. drift is negligible)
% 2. No selection or migration on different character states
% 3. Stable population size
% 4. Non overlapping generations
% 5. constant mutation rate
% 6. no recombination or indels
%
% All this is to say that the only force operating on the population is point mutation.

Ns = 1.0e12;  % population size
N0 = [Ns; 0; 0; 0]; % state vector of the number of individuals
gn = 1e-8; % prob goal->not goal
ng = gn/3;  % prob not goal->goal
gg = 1-gn;  % prob goal -> goal
nn = 1-ng;  % prob not goal-> not goal

% A is the transistion matrix from the state of the population at time t to
% the state at time t+1.
%
% A(i,j) = the element in the i-th column and j-th row.  It represents the
% probability of going from state i at time t to state j at time t+1.

A = [ nn*nn*nn,   gn*nn*nn,            gn*gn*nn,            gn*gn*gn;
      3*ng*nn*nn, gg*nn*nn+2*gn*ng*nn, 2*gg*gn*nn+gn*gn*ng, 3*gg*gn*gn;
      3*ng*ng*nn, 2*gg*ng*nn+gn*ng*ng, gg*gg*nn+2*gn*gg*ng, 3*gn*gg*gg;
      ng*ng*ng,   gg*ng*ng,            gg*gg*ng,            gg*gg*gg ];

% To get the state of the population one generation from now simply
% multiply the transisition matrix by the state vector: N(t+1) = A*N(t).
%
% To get the state of the population at time t simply use the following
% equation: N(t) = (A^t)*N(0), where N(0) is the initial state of the
% population

% This gives the state at time 30010, whose forth entry is > 1.0.
N = (A^30010)*N0;

I havent done the four mutation model yet because that would involve a 5x5 matrix with I didnt feel like coding, but you get the idea.

To you these examples and mechanisms of molecular evolution are not convincing because no one has observed them in the lab. Yet I do not see you applying this criterion to your own argument. Show me a real time experiment where a supernatural force was seen to guide a protein through a neutral gap. You have said that naturalism cannot explain the origin of new functional proteins. But this is simply not supported by your argument. All you have tried to show is that selection and drift cant explain it. Basically you have assumed that there are only two sides to the issue: nature as we currently understand it or the supernatural. Sure you can try to disprove the first, but that will not prove the second. There is a third option, nature as we dont currently understand it. Basically your entire argument is deus ex ignoratia dressed up in molecular biology clothing and wearing a crown of bad statistics.

Questions for Sean Pitman

Fish out of water

Active Member

Active Member

PopGen Grad Student

Active Member

Active Member

Active Member

Fish out of water

PopGen Grad Student

PopGen Grad Student

Active Member

Active Member

Active Member

Active Member

Fish out of water

Active Member

Fish out of water

Active Member

Fish out of water

PopGen Grad Student

Similar threads