Jump to content

Talk:Genetic drift/Archive 1

Page contents not supported in other languages.
From Wikipedia, the free encyclopedia
Archive 1Archive 2Archive 3Archive 4

From Talk:Genetic drift

"refers to changes in the gene frequency of a given population owing to random forces (as opposed to Natural selection). It is primarily a "sampling phenomena;"

is jargon heavy and vague. "Gene frequency"? What kind of force is a random force? I think these edits have made the article less accessible to naive readers. 168...

I went ahead and restored the deleted material and original intro, but added a bit of what I thought the latest edits had to add. I'd be curious to know what problems they were meant to fix, because it's not obvious to me from the nature of the edits. 168...

Mostly, I thought it was unnecessarily wordy. I think some of the material on natural selection simply is unnecessary in this article and the link to other articles is sufficient. I also think the stuff on "luck" and the coin-tossing analogy were obfuscatory -- in a classroom lecture, where a prof can provide more context and more information, they certainly would be inappropriate -- but in an encyclopedia I think it is best to start by describing what things are, and save metaphors and analogies for later, and only if necessary; in this case I just didn't think they were necessary. I also tried to use language more in keeping with the current literature.
I just didn't think that sentences like:
Luck impacts the commonness or rarity of an allele, because an individual who possesses it and the trait that it confers is not guaranteed either to survive or to produce a particular number of offspring.
or
Drift and natural selection are effectively opposites, but they are assumed to occur in concert.
were very clear or illuminating. What was wrong with the shorter version? What crucial information did it leave out -- or how was it unclear? Slrubenstein


The first sentence addresses what I think will be a common preconception among readers who come to this topic naive, or who arrive familiar only with natural selection (the most typical scenario, I expect). That preconception is that "survival goes to the fittest." These readers won't ever have stopped to think about the fact that this maxim is about averages, and is not a hard and fast rule for individuals. If we don't get that preconception out of the way, I believe, many readers will be reading this article in a fog, wondering all the while how to square it with what they think they know. Although I expect it to be clear, the sentence actually attempts only half an illumination. The second half is made clear by the sentence that follows it: "Instead survival and numbers of offspring depend on chance factors, such as weather and being in the right place at the right time." To the extent one is able to parse the sentences, I would expect them at least to make plain what's being talked about when we invoke "luck" or "chance" or "randomness" in our explanation of drift. I launched into the topic the way I did to avoid any vague off-putting abstractions, which is what I think "randomness" ends up being absent a context. I think it's important not to put a reader off, because they may not have been assigned to read this article.

The second sentence you quote is simply an assertion. Do you know what it means to say two things are opposites? Do you know what it means to say things occur in concert? If you do, then you know what that sentence means. The rest of the paragraph in my view proves the assertion. Apparently you view it as an uninteresting or unuseful assertion. But besides the virtues of succinctness and (at least I thought) provocativeness, it has the use of generating a framework for what I think is a very important discussion: About the distinction between drift and selection and how they inter-relate. Evidently you think their distinction and relation is self evident from the definition of drift in the text and the definition of natural selection in the link out. I would say "self-evident" is far too strong. Yes, if the reader does link out and reads the NS article, and if he or she stops a few moments to think, then even a naive reader is liable to appreciate by his or herself what this paragraph explains. But I think that's expecting too much of the reader.

I think everybody knows what luck is, and it's certainly not a metaphor here. I think that everybody has tossed a coin, knows what needs to be known about it, and knows that they know all they are being expected to know about it in the context in which it was invoked. Obfuscatory? How so? To me, this stuff is the very gist of what drift is. "Gene frequency" and "random forces" and "sampling phenomenon" are more succinct, and they certainly represent a tradition in the explanation of drift and in academic explanations in general: The tradition, namely, of using terms that only become clear at the end of the lecture or at the end of the semester. They don't work if you haven't had drift explained to you before. 168...


I agree with SLR - I liked his/her version better, though it may have excluded information, because it was simply much more readable, and thus did a better job of explaining genetic drift. Graft

More readable? Hmmm. Did the unedited text bore you in places? (which?) Were the ideas hard to understand? (which?) Was the prose style thwarting? (which sentences?). Of course, there's no need to elaborate, but I'd be interested and it might be helpful to striking a compromise. 168...

Another problem with the earlier version, besides being less readible, is that it implies that drift is a process affecting individuals. But 168's point about the importance of luck is important and I added another sentence to what is now the second paragraph to make that clear. Slrubenstein

Implies it affects individuals? How so? 168...

This readability thing is intriguing, but without elaboration it's too vague to be helpful. I suspect we just disagree on how techical and abstract to make the article. For me "readable" implies interesting and inviting to a naive and only tentatively committed reader. If one doesn't do some work to make the argot accessible, it's not "readable," in my book. I also believe that stuffing critical content into parentheses makes a text hard and less than enjoyable to read. Let's remember that a lot of people who want to learn about evolution do because they like animals and plants, not because they're into statistics or esoterica. 168...

I agree with the goals of both 168... and Slrubenstein and I think we can (and should) accomodate both approaches. At the start of the article there should be the intuitive definition and relevance for natural historians etc. and then we could have a section on the more statistical nature later on, this conforms to the wikipedia's concept of news style. I was planning to add some more technical material on drift relating to the underlying diffusion equations and the the fixation probabilities for a new mutations given various regimes of Ns, where N is pop size and s is the selection coefficient with a similar treatment to that usually given in elementary population genetics texts. However, I agree with the goal of having the intuitive and non-specialist definitions first (a la 168...), followed up with technical material later (because one persons 'esoterica' is another persons 'bread and butter ;-)). I think there's a place for both. -- Lexor 07:43 19 May 2003 (UTC)
It's great to learn that news style actually is a Wikipedia recommendation. That sums up what I'm striving for exactly. I have no problem whatsoever with technical material lower down in the article. People who read that far obviously are more curious about the subject--either b/c they've been made curious by the more accessible approach up top or b/c they arrived with an interest in that level of detail. To my taste, the article launches far too early into the point that drift is a sampling phenomenon. More or less nobody knows what a sampling phenomenon is, so just two beats into an article about drift we are forced to make people read an article about sampling, just so they aren't baffled when we invoke the term. A definition of sampling is not what readers will have come to this article for. There's no need, furthermore, to know that drift is a sampling phenomenon, if all you want to understand is how it works. The fact that it is a sampling effect qualifies as a suitable bit of enrichment for people with an interest or background in statistics, and as such it belongs farther down, not on top. 168...

Note to those who may not have been aware: "Phenomena" has a singular form. It's "phenomenon." I've had to correct that a couple times. 168...

Re: gene pool, does the term actually ever connote the frequency distribution of alleles or does it only refer to what alleles are simply present in the population, without regard to how common they are? 168... 17:31 19 May 2003 (UTC)

Frankly, I used gene pool out of deference to an earlier contributor (in fact, I thought it was you) -- I'd prefer changing it to "population" or something like that. I just think the effect is on something, and we need to specify what it is an effect on.

As for readibility, I have no general argument with your intentions -- I think we disagree as to what is readible to a lay audience. In any event, I am satisfied with the current opening. However, I wonder why someone keeps cutting the example of the car accident? Such examples are good ways to communicate an abstract principle to a lay audience, because they are concrete. Also, I think the car accident is a better example than disease, because there may indeed be an inherited susceptibility to many diseases. Slrubenstein

Cars aren't part of many species traditional natural environment (not to mention that alleles conferring speed and alertness play a role in crash immunity), so I found it a long mental excursion (costing probably ten times as many words as my coin-tossing allusion). Most recently I removed the example b/c it struck me as redundant and not fleshing out something that needed fleshing. Yes, I invoked the gene pool. I'd have to go back to check how I did it, but I'm prepared to think that was a mistake. In answer to my own question, I think allele frequency has to be a separate matter, b/c it wouldn't make sense plot the alleles for all genes on the same graph, which is what my imagined hybrid concept of "gene pool" as "allele frequency" would have to be. 168... 17:46 19 May 2003 (UTC)

I still think "gene pool" would be great to invoke early in some context, just not (now that I think about it) in a sentence that strives to be the most general statement of what drift is. To invoke it such a context is to say that drift only covers fixation (e.g. when an allele drifts from the gene pool all together) 168... 17:50 19 May 2003 (UTC)


I realize that other contributors have been trying to make the opening as clear as possible. But in the process, the article was becoming inaccurate. Accidental death may be a cause of drift, but it is not the basic cause. There is no way around it: drift is first and foremost a statistical phenomena. I decided to add a brief explanation of the Hardy Weinberg theory -- perhaps other can join me in writing the linked article. I also reintroduced 168's coin-toss analogy -- I rejected it in its earlier context, but I do understand 168's point and I hope that in this new context it accomplishes what s/he intended. I removed the stuff on sexual selection because it is not relevant to drift. The point is that there are a number of reasons why HW may not work out -- natural selection is one reason, sexual selection is another, and drift is a third. But they are all different processes. Slrubenstein

"Accidental death may be a cause of drift, but it is not the basic cause. There is no way around it: drift is first and foremost a statistical phenomena." Huh? I think you've been reading too many statistics books. Statistics never killed anybody, nor did it ever produce progeny beyond the average for that person's genotype. Most readers will want to know what on earth you are talking about. Also, why are you spend three paragraphs up top telling people that drift doesn't happen? 168...
Slrubenstein is correct here, drift is primarily a statistical issue, it relates to the sampling error as described. Almost all texts on population genetics and evolutionary biology approach it from the perspective of Slrubenstein. Although I understand the need to describe the process intuitively as 168... has been trying to do, accuracy and precision must be primary. I am actually in the process of writing an article on the Hardy-Weinberg principle, so give me a few hours to install this, before writing a stub (actually I would describe this is as the "Hardy-Weinberg principle" rather than "Hardy-Weinberg theory", because it's really a principle which is part of the larger theory of population genetics). -- Lexor 19:24 19 May 2003 (UTC)

What, precisely, was imprecise or inaccurate? I certainly never asserted that time of death was the primary randomizer. I also cited fecundity/numbers of offspring. It was a significant oversight not to mention the selection of parent alleles as a randomizer (to the extent we want to include the important nuts and bolts), but it can't be right to describe it as the essense or the be-all and end all of drift. With respect to the Hardy-Weinberg princple, perhaps. But what about asexual organisms, which drift too? And what about fecundity? Does the HWP imply that deviations from average number of offspring don't matter? And anyway, I suppose drift occurs when a population doesn't satisfy the condition for the equilibrium. Does fecundity cease to matter in such a context? 168... 20:34 19 May 2003 (UTC)

I just found this on the Web regarding the Hardy-Weinberg equlibrium: "Assumptions include: random mating, equal sex ratios, large population size, no migration, and no difference in fertility or mortality of different genotypes." Obviously fertilities differ and deviate from the average on an individual basis. Ditto for mortality/lifespan. 168... 20:47 19 May 2003 (UTC)

It is true that deviations from HWE can be caused by the above factors, but deviation from HWE don't necessarily imply drift. For example, deviations from HWE are also caused by selection, but selection must be very strong in order to detect this deviation statistically. -- Lexor 01:11 20 May 2003 (UTC)

O.K.., but why would it be necessary to invoke the HWE to explain drift? And what about the HWE actually applies to the circumstances where drift occurs? Perhaps only the principle that the selection of parent alleles in the sex cells that form the zygote is a randomizer. But not the only one. 168...


If you were to go to the slide that defines "Genetic Drift" at baen.tamu.edu/users/cengler/biology/notes/chapt16.ppt you would read: "Genetic drift is a change in allele frequency due to random variation in fecundity and mortality in a population." This seems to be exactly the point I was trying to make in the text that provoked the charge of inaccuracy. Would the accuser please clarify? 168... 21:09 19 May 2003 (UTC)

168, I know that you have considerable knowledge of genetics, but here you simply do not seem to understand the principle. You say "Statistics never killed anybody," and you do not seem to get the point: there is no need at all for anyone to be killed, for drift to occur. It need not be a matter of variation in fecudnity either -- I have added a point in the article that the model couple can have (by luck) four children all blue-eyed. But I did add the point about fecundity and morbidity in the beginning. Lexor, please go over my changes.


168, I think part of the lack of clarity is in your use of "selection," as you yourself recognize, drift is different from selection (whether natural or sexual) -- it is not that disease "randomly selects." Perhaps you understand "sampling" too literally?Slrubenstein

Given the context, I can see it may be asking for trouble to use "selection" in any other sense than as shorthand for non-random selection ala natural selection, but "randomly selects" is unambiguous here, it seems to me. Rather than me taking sampling too literally, you make me think you aren't taking it literally enough. "Sample" is a verb with which statisticians often will pair a subject and an object: As in "The dealer samples cards from the deck at random in hands of five cards" (just to concoct an example). You could substitute the word "select" here, analogously to the way I did. When I did it, I was just trying not to use "sample" too many times in the same paragraph. I don't think it's liable to cause any problems. 168... 06:54 20 May 2003 (UTC)

I concede that invoking and linking to "sexual selection" was a bad idea, and that formally it was probably "inaccurate", in that sexual selection is a technical term that means non-random sexual selection, even if the name itself doesn't imply such specificity. Nevertheless, as far as I can tell, it's appropriate to talk about variations in fertility (they are even invoked in some discussions you can find on the Web regarding deviations from Hardy-Weinberg) and appropriate to invoke pairing with unusually fertile mates by chance. Also, while sexual selection may not be random, my guess would be that this selection would be considered random with respect to many or perhaps most of the alleles the selected individuals are carrying. I don't know though. I don't represent myself as a geneticist or even somebody whose taken a college course in it. I've just picked up a bit by osmosis after working in the biosciences and in my outside reading. 168...


168, this:
Chance affects the commonness or rarity of an allele, because no trait guarantees survival or a given number of offspring. Instead these outcomes depend on factors like the weather and being in the wrong place at the wrong time. In other words, even when individuals face the same odds, they will differ in their success.
just reveals a real misunderstanding of what drift is. No matter what the weather, or what place one is in, it is possible for a couple to have four blue-eyed children even if the parents are Bb and Bb (in other words, the frequency in gen. 1 is 50%-50%, in generation 2 100%-0%) -- that is an example of drift, and it matters in small populations -- it is a statistical phenomena. Slrubenstein
SLR, when you go out to measure what alleles exist at what frequency in the population, the results don't depend only on who was born, but also who is still alive. I have been speaking in the language of who, where, what, why and how (as in, Colonel Mustard in the living room with the knife). When you say "statistical effect" (which I don't disagree, with by the way) you are speaking in the language of abstract ideas. It's true that no one need die for drift to occur, assuming infinite resources, but not only is this an absurd scenario (since all species die and at ages that vary among individuals), but mortality needs mentioning if it's a primary factor in drift, which my recent reading seems to confirm. 168... 22:15 19 May 2003 (UTC)
Here's an independent explanation of genetic drift on the Web, http://www.talkorigins.org/faqs/genetic-drift.html, which seems to have been written by a bioinformaticist and quotes some textbooks at length. I believe Talkorigins has a good reputation for accuracy.168...
Oh no, not talk.origins... that was one of my addictions before I discovered Wikipediholism... uh, not that I'm a Wikipediholic or anything. Just do me a favour and don't look too far back through the archives, if they still keep those there. ;) -- John Owens 07:10 20 May 2003 (UTC)

I for one thought SL's intro was far more clear (at least for a biologist - technical words like "fecundity" should be replaced by "birth rate" when they are in the intro) but much of 168's body text was better. I was about to edit the article accordingly and add added emphasis the drift affects small populations far more than large ones and also add in more explanation on how populations get to be small to begin with (bottleneck and founder), but the article was reverted. --mav

I'm confused. Was there an intro in which the word "fecundity" appeared? Which intro did you like better? 168... 02:14 20 May 2003 (UTC)
? Do you read a page before you revert it? This is the first sentence in SL's version: "Genetic drift refers to changes in the allele frequency of a given population owing to random variation in fecundity and morbidity (as opposed to Natural selection)." Notice the word "fecundity." --mav

I see. Your post confused me. You like SLR's intro, but you don't like his use of fecundity. It seemed to me you were saying I had used the word, which I knew I hadn't. No, I didn't read the version that _immediately_ preceded my reversion, b/c I saw it as needing wholesale revision. We had a couple people explicitly for news style, plus the Wikipedia institutional recommendation, and SLR was rocketing away in the opposite direction, ignoring this line of discussion on the Talk page. 168... 03:46 20 May 2003 (UTC)

Mav, I agree. Perhaps you could post the text of the proposed changes here, so we don't get caught in multiple edit conflicts? -- Lexor 01:11 20 May 2003 (UTC)
I'll wait to edit the article as soon as the revert war between SL and 168 stops. I'm not too keen on having my work caught in the cross fire of the edit war and I also have plenty of other things to do while this settles down. --mav

I added this article to Wikipedia:Brilliant prose. Nice going wikipedians! Kingturtle 02:35 20 May 2003 (UTC)

Which version? There is an edit war going on rigtht now. --mav

I saw this when it went up, plus you can look at the date (and weep, if that's how you feel). Not that I'll let it go to my head, but it's clearly referring to my last version. 168... 03:46 20 May 2003 (UTC)