What exactly is natural selection and precisely where does it occur?

Thanks Hoot, for opening this topic. Although my position closely parallels AZPaul's (so repeating what he's already said would be pointless), I would like to emphasize a couple of points. One of the first problems I see is there appears to be some misunderstanding of what is meant by "natural selection". We can argue over whether or not the term is either useful or descriptive (I personally think it is neither), but it is one that has been in regular usage for 150 years since Wallace and Darwin first coined it, so we're pretty much stuck with it. Their attempt to differentiate "natural" vs. "human-directed" evolution is what is leading to many of the difficulties (as others have sort of noted). Unless we are going to postulate that humans are not "natural", then whatever we do to a population is indistinguishable from what the environment does to a population.In essence, then, anything that affects the fitness of an individual organism is "natural" selection. The key word here is selection, not the adjective modifying it. Any filter that affects the survival, reproductive success, or reproductive rate of the individual members of a population is natural selection. Hence, when we talk about "sexual selection", or "kin selection" or whatever selection, we are referring to specific types of natural selection. The terminology can be confusing, I admit. However, when biologists/ecologists, and others of that ilk bandy those terms about, the whole edifice rests on the unstated understanding that in every case it is all natural selection.There are, of course, non-selective forces. As we discussed in a previous thread, genetic drift, etc, don't rely on the actions of natural selection. This is probably why we often hear the formulation "evolution is the change in allele frequency in a population over time". This formulation allows us to include epigenetic factors, drift, etc, in the definition of evolution. Our discussion in THIS thread, IMO, should be limited to natural selection, however. I'm not sure why you feel that if natural selection refers to the factors relative to the reproductive success of an individual organism there should be any confusion here. As PaulK pointed out, in some instances it may make sense to discuss the contribution of particular genes or gene complexes to the survival of its organismal "packaging". After all, it is the gene, not the organism, that is the unit of inheritance. On the other hand when we're talking about selection writ large, to me it doesn't make sense to discuss "gene selection" except in the larger context of selective pressures on the individual organism. After all, there are (to my knowledge), no environmental factors that can reach inside an organism's body to directly affect a single gene or suite of genes without ALSO affecting everything else about the organism.I think some of the problem arises from a misunderstanding of Dawkin's "selfish gene" concept - it's a problem he acknowledged, which is why he wrote The Extended Phenotype. Quoting:

quote:

---

I want to argue in favor of a particular way of looking at animals and plants, and a particular way of wondering why the do the things that they do...What I am advocating is a point of view, a way of looking at familiar facts and ideas, and a way of asking new questions about them. (Dawkins R, 1999, pg 1)

---

In short, Dawkins is saying here that the selfish gene concept is not a fact, but a way of looking at facts. I can think of a number of places where it is an exceptionally useful way of looking at facts - ethology, evolutionary developmental biology, etc, are all sciences well-served by the concept. However, I find it utterly useless for sciences such as ecology, conservation biology, population genetics, etc. In these latter, looking at the action of natural selection as it affects either organisms or whole species makes more sense. Therefore, probably from my own biases and experience, I find the use of the individual organism as the "target" of the selective filter to be the most relevant.

I bolded the part I have a quibble with. Although my points here may serve to confuse the issue more for those unfamiliar with it, I'm afraid I have to disagree with this characterization of natural selection. Setting aside for the moment the "genes'-eye-view" argument, selection is a filter that affects organisms based on their phenotype. I would argue that catastrophic natural disasters - either global or local - are not really subsumed under "natural selection". There has been a great deal of discussion over the years concerning the "whys and wherefors" of differential survival of specific taxa following large-scale extinctions. My personal opinion is that this survival owes more to luck than genetics. IOW, there is no true selective filter in operation. Meaning the frog whose stream dries up due to the beaver, or the continental fauna devastated by an asteroid are not being selected for or against. Their genotype/phenotype has absolutely no bearing on whether they survive or not (well, maybe in the case of the frog if there are individuals in the population more tolerant of a xeric environment, say).I don't know if you get the chance to read much popsci, but one interesting book on this subject is David Raup's Extinction: Bad Genes or Bad Luck? (WW Norton 1992). Although I strongly disagree with many of Raup's contentions (including especially the "Nemesis Hypothesis"), he makes a good case for genetics having diddly to do with survival of taxa following a mass extinction.Asteroids, in short, are not a selective filter acting on a genotype. They represent a field of bullets where the survival of any given taxa is due more to chance than selection. Yon butterfly may have the most perfect genotype/phenotype on the planet with all kinds of wonderful adaptations, but if a bird eats it before it reproduces, its genotype had nothing to do with it, and it is an evolutionary dead end.

On the contrary, large numbers of "small furry mouse" taxa DIDN'T in fact survive the KT event. Whereas there is a lot of evidence that indicates various dinosaur taxa were in decline prior to the KT event, and the hypothesis that the asteroid finished them off seems pretty conclusive, there were too many vastly different taxa that survived the event to conclude that phenotype had any play in it. There is no identifiable thread that runs through the survivors - no particular phenotype gave any special advantages. This is one reason I like the "field of bullets" analogy. It derives from carnage of WWI battlefields. Picture 10,000 infantry charging across open fields in the face of entrenched machine guns. Individual ability, training, etc, has absolutely NOTHING to do with which ones of those infantry survive. It's pure luck of the draw. There's no selection; it really is random chance. The same goes for a bloody great rock falling on someone's head. Random events are NOT selective because the genotype/phenotype of the individual organism has no bearing on whether the organism survives or perishes - as PaulK noted, luck is not a component of natural selection. Right. The frog thing was probably unclear in my post. I was visualizing a sudden drying of the stream - and probably stretched the analogy further than it should have gone. Yep, them's the questions. Those are ALL examples of natural selection - the filter that "weeds out" the less successful organisms.My butterfly example was another poor analogy, I guess. What I was trying to illustrate is that regardless of all the wonderous adaptations and neat genes the now-eaten butterfly may possess, the random event of a bird managing to snack on it had nothing to do with them. It got eaten regardless of the adaptations. This probably wasn't a good analogy of a random event - unless it was simply the fact that the bird got lucky... As PaulK noted, I'm not sure but what you are confusing natural selection - the sum total of the biotic and abiotic factors that affect individual reproductive success - and the results of this selective filter (e.g., evolution). The two are not synonymous. A bloody great rock may eliminate enough taxa to provide the wide-open spaces necessary for ecological release, but it isn't selection in action. Hope this helps rather than obfuscates.

Of course not. Evolution (what you describe here), operates at the level of population, not individuals. Natural selection, on the other hand, is an individual selective filter. Evolution operates ONLY over generations. Selection operates during the individual's lifetime. Individuals have nothing to do with evolution, because individuals don't evolve. Additionally, your second statement is also incorrect: individuals ARE the victims of natural selection, which factors serve to either promote individuals who have some selective advantage in the current environment by increasing their chance of reproduction and hence passing on this advantage to their progeny, or select against individuals who are "less fit" in the particular environment. Although it isn't an "all or nothing" affair, and leaving aside the gene-centric or organism-centric viewpoints, this is the basic definition of natural selection.

I think you and crash are essentially saying the same thing in different ways. I'm pretty sure he acknowledges that genes determine phenotype to a large extent, and that they are the units of inheritance, and thus the part that is of evolutionary significance. On the other hand, individuals most certainly do "get selected". Without the ability to pass on their genes - which has everything to do with phenotype - genotype is immaterial. The filter of natural selection doesn't pick out a particular gene or suite of genes on which to have an effect. It selects the totality of the phenotype - i.e., the individual organism - to operate on. An organism might have several excellent genes, but if the overall phenotype is less fit than another set of genes (individual), then they may not get passed on to subsequent generations. This is one of the reasons I don't fully agree with Dawkins' concepts. Yes, in some ways (especially when you're talking generational timescales) to talk about how suites of genes get passed down. However, unless you can demonstrate any selection pressure that operates at the level of the gene in "real time", then I'm not sure the concept holds. You just stated it: successful phenotypes are what get passed down the generations. Why? Because natural selection operates on the phenotype, not the genotype. This is where people tend, I think, to misunderstand what Dawkins is talking about. He's using the "cooperative gene" and "selfish gene" concepts to show - not how, which is what natural selection is all about - but why.

Mod - I understand Dawkins' concept. I have also said that a "genes-eye-view" of evolution is both useful and, over the long term, makes sense. I thoroughly "get" the idea that we can look at evolution in terms of the success of genes to get themselves passed down the generations. On the other hand, we're not talking here about evolution per se, we're talking about natural selection - those factors that influence whether or not those "optimal" genes get transmittted. In that case, I submit that selection doesn't operate on genes, it operates on individual organisms, during their very own lifetimes. Although some selection pressures will operate over generations (it's how adaptation occurs in the first place, as you know), for the purposes of understanding those selection pressures we lump under the rubric of natural selection, the only thing that makes sense is to look at the effects of those factors on the individual organism. After all, it's the organism as a whole - good, bad, and indifferent genes - that has to survive to reproduce. It's not really a question of "sides". It's really more a question of looking at the facts in a different way. Both "sides" have their uses. Population geneticists would probably agree with Dawkins. On the other hand, I don't find this view particularly useful for understanding ecosystem-level interactions, especially extinctions - one of my particular concerns. Nor is it particularly useful for any other aspect of conservation biology or ecology that I can think of. Perhaps, as I mentioned, it's a question of personal biases and experiences. All I see it as is a nice sort of theoretical way of looking at the question. Right, I understand that. Again, however, you're not talking about natural selection as such, you're talking about evolution. Since the topic is "at what 'level' does natural selection operate, I think that the genes-eye-view is inappropriate. Perhaps I'm making a false distinction: evolution vs. one of the key mechanisms of evolution.

I understand that you define it that way. However, perhaps you could identify the "selection" element in a bloody great rock falling out of the sky? How does a particular adaptation in a given population of individual organisms - whether optimal for their current environment or merely "good enough" - allow those individuals to survive such an event? Catastrophic extinction, which is inherently a random process (the opposite of selection), should not be subsumed under the accepted rubric of natural selection - because, in essence, there is no "selection". You die, or you don't, on the luck of the draw. As PaulK mentioned, luck isn't a part of natural selection.I will say that ecologists and conservation biologists have long sought a solid, theoretical construct that answers the "why" question in extinction. Just like with bioinvasion, all of the theories proposed to date are weak, unfortunately. We very often understand "what", and can in many local cases figure out post facto why, but a general theory of extinction has proven elusive. I disagree. The bloody great rock was certainly a "natural" event. But was it "selective"? One possible source of confusion (although I don't know if that's the case here), is that we often bandy about the phrase "selectivity of a mass extinction". The problem stems from the use of selectivity in this context. We are NOT talking about Darwinian selection - e.g., an environmental filter. Rather, we are using an anthropomorphic analogy (for lack of a better term), to describe how it appears that a mass extinction randomly "selects" certain taxa for destruction while leaving others untouched. This one but not that one. Kind of like a mad god's finger. If we ever come up with a General Theory of Extinction, I think this confusion will disappear. Right, as far as it goes. However, remember that in a mass extinction event we are not talking about individual death based on an individual's lack of a particular set of characteristics. We're talking about entire taxa being obliterated. As we both (I think) agree that natural selection operates at the individual level, how can an event that affects entire lineages be subsumed under this? Individual adaptations have absolutely no bearing on survival here (which is what natural selection is all about, n'est-ce pas?). Those individual members of surviving taxa which are capable of survival/reproduction in the new environment will - as per Darwinian evolution based on differential reproductive success (e.g., natural selection) - "inherit the earth" (but see, for example, the "Dead Clade Walking" hypothesis). You're welcome. I only wish I could take credit for it. It's one of the standard models used in ecology to describe extinction patterns. Along with this model, we have the "fair game" model where natural selection works to eliminate the less fit (the standard Darwinian model); the "wanton" model where individuals survive based on purely random factors - and may in fact be less fit than those who DIDN'T survive (mostly used on local scale); and the "gambler's ruin" model (where taxa with fewer sub-elements are more likely to go extinct); and the "mass extinction" model where everything in a particular area gets obliterated. This latter is an unfortunate choice of terminology because both field of bullets and wanton models can also be used to describe the extinction pattern of what we normally call mass extinction. I prefer the term "first strike" for this one, just to avoid confusion. Unfortunately, I haven't been promoted to the terminology police squad.

Sorry, O' tentacled one. I nearly missed this reply. This is indeed the popular characterization of the result of the K-T event, at least in North America. And as a first approximation to get across the idea of the devastation caused by the K-T, or P-T events or any other of the "big five" events, it's really very good. It is also, like many scientific "sound bites", not entirely accurate when you look at the details. Just to name two taxa that "give the lie" to this hyperbole that are being sort-of discussed on another thread, look up the crocodillians (terrrestrial) and sharks (marine).

How do you figure? This is precisely the definition of natural selection Darwin and Wallace developed. NO! How many times do you have to be told? You even quoted us in this post on this subject! Evolution is the result of natural selection operating on individual members (or suites of genes if you like Mod's construction better) over generations in a population. Natural selection is (one of) the mechanism(s) of evolution. It is NOT evolution. Populations evolve. Individuals (or genes) are selected. There is no disagreement here. Good grief - you even state it here! It is the agency or mechanism of evolution! Why do you keep repeating the nonsense (and apologies, but that's what it is) about individuals evolving if subject to natural selection? And here again, for some reason, you lose it. I really and truly do not understand what element of the concept you're not grasping, Hoot. I really don't. Setting aside for the moment whatever it is you're trying to convey with the "active context" and "active sense", which, since they aren't terms used in science as far as I know, are inherently meaningless (you'll really need to expand/explain what you're trying to say), Darwin absolutely DID use natural selection as the explanation for biological evolution. In fact, it was his (and Wallace's) key insight into the mechanism of evolution - natural selection - that was so revolutionary. Up until then, various folks from Lamarck to Erasamus Darwin could see that evolution occurred - but nobody until Darwin came up with a how that fit all the facts.

Heh, this is what I've been saying all along - although obviously not as well. The last paragraph in your quote also sums up well - the genotype is indeed what gets replicated (outside of clonal species, of course). That is never in question. Throughout I've been talking about what gets selected - not what gets replicated. Perhaps that's where our miscommunication arises?

Mod,I'm going to skip the cultural evolution piece, because that's not what we've been talking about in this thread, and I really think that the topic will derail things too much. I have my own views of cultural evolution, but they would be better discussed on another thread. Suffice, for the moment, that we are primarily in agreement there - but I feel there is little relevance for natural selection. We are not in disagreement. However, the reason I keep coming back to the phenotype (the individual sum of genotype and environment), is that selection can operate on non-inheritable characteristics. Although these characteristics may have no bearing on evolution, they DO prevent or enhance the reproduction of the individual. Which is how genes get transmitted. Since the selection part of natural selection refers to the filter created by the totality of the biotic and abiotic factors affecting the organism in its particular environment, this would, perforce, have to include things that operate only at the individual level - regardless of genotype. For example, an organism that is damaged but not killed by a predator may have difficulty reproducing, no matter what its genotype may be. An epiphyte overburden may exceed the structural tolerance of a forest emergent, causing it to fall and bring down multiple neighbors - in spite of their genotype. The problem I have with the gene's-eye-view is that it ignores the stochastic events that may change the allele frequency of a population - by eliminating those individuals who carry them - it is an inherently deterministic view of natural selection.Now, before you go ballistic, please remember that I am NOT talking about evolution, only natural selection. Since we both agree that evolution occurs at the level of population, then I assume we have no disagreement that talking of the change in allele frequencies - the result of natural selection (among other things) - in a population is evolution. In this sense, it is absolutely imperative to use the gene's-eye-view, because that's what evolution is all about - adaptation of populations to changes in the environment (again, among other things). Yes, when discussing evolution.

Okay, Mod. I'm clearly not explaining myself well. That being the case, let me post a few quotes from people who are MUCH smarter and better at explaining things than I am.

quote:

---

What is exposed to natural selection is not the individual gene or the genotype, but rather the phenotype, the product of the interactions of all genes with each other and with the environment. (Ernst Mayr, Evolution and the Diversity of Life, Belknap Harvard, 1997, pg 318)

---

quote:

---

Natural selection acts on phenotypes, regardless of their genetic basis, and produces immediate phenotypical effects within a generation that can be measured without recourse to principles of heredity or evolution. In contrast, evolutionary response to selection, the genetic change that occurs from one generation to the next, does depend on genetic variation. (Lande R, Arnold SJ, 1983, "The measurement of selection on correlated characters", Evolution, 37:1210-1226)

---

quote:

---

Evolutionary biologists differ on whether or not the definition of selection should require that the classes differ genetically. Some authors...define selection as a consistent difference in fitness among phenotypes, acting within a single generation. Whether or not it alters the frequencies of phenotypes in the next generation depends on whether and how the phenotypical differences are inherited. The change in the population from one generation to another is termed the response to selection. Authors who advocate this phenotypical definition distinguish the response, which is solely a matter of inheritance, from differences in survival and reproduction, which constitute selection itself. (Futyma DJ, Evolutionary Biology, Sinauer, 1998, pg 349. Emphasis in original)

---

I'm one of those who advocate the distinction between selection - which is against phenotypes - and the evolutionary response to selection - which is based on genotypes. I hope that makes my position a bit clearer. Edited by Quetzal, : Kuz I kant spel

I thought you said Mod was "right" and the rest of us were full of it (or words to that effect). Fickle, aren't you? Before I go any further, I want to emphasize that neither Mod nor I are "right" in the sense that the other perspective is "wrong". As Percy pointed out, these are just two different ways of looking at the facts. Indeed, both perspectives are "right". It all depends on what you're trying to measure (or describe). For me, conceptually, isolating selection as in the phenotypical definition of selection is more useful when discussing what happens in a given generation. However, when describing the effects of selection as it pertains to evolution, the gene's-eye-view is the more relevant - because it describes the result of selection over evolutionary timescales. This is, in essence (and Mod can correct me if I mis-state), the only difference between the two. It's really a fairly esoteric disagreement. No. The population is the fundamental element of evolution. It makes no sense to talk about evolving individuals, or for that matter the arbitrary groupings we call species, genera, etc, except in the abstract. However, it is quite useful to talk about gene lineages evolving. In this, Mod and I are in 100% agreement, and the genecentric view provides a clearer picture of what is happening. The argument on this thread is about selection, not evolution. In what sense? I'm not sure I understand the question. Please clarify what it is you are asking. Traits DO evolve. Single characteristics, suites of genes, etc, all evolve in response to long-term selection pressures. I'm not sure where this question comes from. That is the definition of evolutionary response to selection. I even quoted it above. As long as the differences are inheritable, then phenotypes, genotypes, single genes, polygenes, etc, all can evolve. And by the way, I absolutely loathe the term "microevolution". It's all evolution. The arbitrary distinction serves no useful function outside of paleontology, in my opinion.I think for some reason you are still conflating "natural selection" with "evolution". As just about everybody participating in this thread has stated about a million times, they are two distinct, separate concepts. Natural selection is one of the engines that drives evolution. It is NOT evolution. It is a mechanism of evolution. The motor is NOT the car. Let's let ol' Douglas Futuyma (yeah, I know I mis-spelled his name in the citation above, it's too late to go back and re-edit that post. Sue me.) state it (from further down the page I cited previously):

quote:

---

For our purposes, we will define natural selection as any consistent difference in fitness (i.e., survival and reproduction) among phenotypically different biological entities. The entities may be individual genes (which must have some phenotypically variable property if the differe consistently in fintess), groups of genes, individual organisms, populations, or taxa such as species. (Although we have adopted a phenotypic definition, we will almost always discuss the fitness of phenotypes that are inherited to at least some degree, because selection has no evolutionary effect unless there is inheritance.) (Futuyma 1998, op cit. Emphasis in original).

---

What? I think you're still confusing selection with response to selection. Natural selection is basically just a seive that "sorts" individual organisms based on some trait or characteristic (or suites of traits or characteristics). This sorting may affect the individual's survival, reproduction, reproductive rate, etc. In other words, the trait or whatever impacts the individual's fitness in its current environment. That's all.The response to this sorting, whether we look at it from a populational standpoint OR from a genecentric standpoint, is evolution. Speciation is one form of evolution - one of the directions evolution can take. Yes. No. The seive is always operating. The "sorting machine" of natural selection is always "on". The key thing to remember is that in reference to evolution, it is only when the phenotypical effects of NS are inheritable that NS drives evolution. This latter point, by the way, is the basis for the "loophole" that allows NS to promote traits that are actually less fit that Mod mentioned. No. "Selected for" (i.e., selected because) is precisely what the seive does. It sorts organisms based on (a) particular trait(s). There may be selection of some other trait that the organism also posesses - but simply because that other trait also passed through the seive at the same time - not because the seive was "set up" to filter that other trait. Natural selection ISN'T the "occasion of differential reproductive sucess". That's the response to selection (sort of - although that's a rather poor way of describing it). What do you mean, "absurd"? Chiro said precisely the same thing I said. You can't agree with me and disagree with Chiro, since we merely used different words to say the same thing. Maybe you need to re-read Chiro's post in light of what I've said? Again, it seems to boil down to "selection" vs. "response to selection", n'est-ce pas? Understand the difference?

Wow. This thread has definitely moved on since the last time I was able to participate. I don't want to go back and rehash old ground, but I would like to make a couple of final points and/or clarifications, if it's not too late. Right. I'm not sure why you brought this up, since it's not a point of disagreement. Further up-thread, I mentioned to Hoot that I felt that the term "natural selection" was a poor one, precisely for the reason you mentioned - all selection is natural. I also mentioned to him that the terms "natural" and "artificial" have been in constant use for ~150 years, so it's unlikely that we're going to see the last of them.As to the bolded part, I don't disagree - normally when we are discussing NS, we're discussing it in terms of evolution. Perhaps it's only because we interpret the purpose of this particular thread differently, but I was under the impression that we were trying to discuss NS more or less in isolation - not evolution. After all, the original question was: "What exactly is NS and precisely where does it occur?". Evidently we've been discussing two different things - and that may be why you're not understanding my position. I understand the genecentric viewpoint well. It is nearly purely a way of looking at multi-generational selection response. It is not a good way of discussing NS itself - except in the context of evolution. I have stated multiple times on this thread alone that in the context of the adaptive response of taxa to NS (i.e., evolution), the gene is far and away the best (and I'd go so far as to say the only) way to cogently discuss the question.Let me go a bit further, however, and make two points on "practicality". Hopefully you'll be a bit clearer on my perspective. My discipline is focused on discerning, describing and evaluating the interrelationships of whole organisms to one another and to their environment. As such, we are concerned with phenotype, and outside of the context of population dynamics (i.e., response), in many cases it doesn't even make sense to talk about genetics. In fact, we most often don't even consider the individual organism - we're concerned with populations, communities, or entire ecosystems (c.f., the subdisciplines of macroecology, landscape ecology, etc). One of the things that goes along with this is trying to understand selection pressures in isolation from the organisms being affected. It doesn't matter whether these pressures are a bloody great rock or fluctuations in a local microclimate. Therefore, it is more "natural" for me to think about selection (not evolution) in terms of the effects on individuals during their lifetimes, than the effects on populations over evolutionary or generational timescales - and to describe NS in that context.The above brings me to a second, subsidiary point that may have some relevance for this thread. You have stated repeatedly that you feel the genecentric view is "easier" to use to explain NS to those without a biology background. I very much disagree. People (erm, non-biologists anyway) don't walk outside, look around their backyards, and see genes. They see individual plants and animals. I have led enough natural history ecotours, and given enough presentations to non-scientists, to have learned that the best way of providing simple illustrations of complex concepts (from the effects of ecosystem degradation to symbiosis to adaptation and evolution), is by couching the explanation in terms that people can intuitively grasp. In other words, discussing NS in terms of organisms. The genecentric view merely introduces an additional level of abstraction that can get confusing (as can be seen on this thread). If the purpose of threads such as this one is to describe NS to folks who don't have a bio background, then I submit that the individual organism is the best way to do it.If I was talking to scientists, then I would use whichever viewpoint best illustrated whatever point I was trying to make, at whatever level was appropriate. And if I was describing evolution, then I would use (and have used) the genecentric position - because that's what evolution is all about.Hope this clarifies.

Hi Hoot,If you still have questions on anything else in my post 175 let me know. Since this thread has moved on quite a bit, I don't want to waste what little time is left. I would like to try and clarify one point, however. Like any metaphor, we've got to take it with a grain of salt, and be careful of how far we push it. Here is a very simple illustration/explanation for NS that I have used in the past, that may explain what I was trying to get across. Remember, please, that I used the seive analogy in response to your question: They used to sell a toddler's toy (I think it was one of those Fisher-Price things), which consisted of a clear plastic cylinder divided vertically into separate compartments by plastic plates with different size holes. You put a bunch of different colored and different sized plastic marbles in the top, shook the thing, and the marbles cascaded down through the toy. The marbles got sorted by size, with the biggest staying at the top, and the littlest tumbling through the various holes and ending up at the bottom. The rest are sorted by size to the different levels. Let's say that all the littlest marbles are also red. The toy "selected" small, red marbles, in other words. However, it didn't select for red color, it selected for size. There is selection of red marbles, but only because all the red marbles are also small. Thus the individual selection filter doesn't "obviate the selected for" aspect. It actually explains the difference between selection for a particular trait and selection of un-related traits.