Talk:Natural selection/Archive 7

Fisher

In response to a request from Ted I have now got hold of a copy of Fisher’s ‘The Genetical Theory of Natural Selection (1930)’. The title suggests Fisher considers inheritance to be a component of Natural Selection; that he uses the same definition as Darwin. Here is a (slighlty abridged) extract:

Natural Selection is not Evolution. Yet ever since the two words have been in common use, the theory of Natural Selection has been employed as a convenient abbreviation for the theory of Evolution by means of Natural Selection, put forward by Darwin and Wallace. This has led to the unfortunate consequence that the theory of Natural Selection itself has scarcely ever, if ever, received separate consideration.

The overwhelming importance of evolution to the biological sciences partly explains why the theory of Natural Selection should be so fully identified with its role as an evolutionary agency, as to have suffered neglect as an independent principle worthy of scientific study.

For advocates of Natural Selection have not failed to point out, what was evidently the chief attraction of the theory to Darwin and Wallace, that it proposes to give an account of the means of modification in the organic world by reference only to the 'known' or independently demonstrable causes. The alternative theories of modification rely, avowedly, on hypothetical properties of living matter which are inferred from the facts of evolution themselves.

The present book, with all the limitations of a first attempt, is at least an attempt to consider the theory of Natural Selection on its own merits.

When the theory was first put forward the vaguest element in its composition was the principle of inheritance. No man of learning or experience could deny this principle, yet, at the time, no approach could be given to an exact account of its working. That an independent study of Natural Selection is now possible is principally due to the great advance which our generation has seen in the science of genetics.

The object of the present chapter is to combine certain ideas derivable from a consideration of the rates of death and reproduction of a population of organisms, with the concepts of a factorial scheme of inheritance so as to state the principle of Natural Selection in the form of rigorous mathematical theorem…

So Fisher uses Natural Selection in the Darwinian sense. I have updated my diagram.

— Axel147 08:51, 4 July 2006 (UTC)

Thanks for going to the trouble, Axel. I suspect that all the population geneticists up through at least the 60s or 70s would fit into the same place in your diagram: Wright, Comstock, Robertson, Hill, Cockerham, Dobzhansky, Kempthorne, Falconer, etc. Put together with Kim's comment in the previous section, it is clear to me that they all treated natural selection as a population concept (even with individuals as the unit of selection), where non-inheritable selection is irrelevant. It seems, however, that Darwin included evolution as part of his definition, which these people do not. Now, I guess I need to figure out why the more ecologically-oriented don't use it as a population concept. Ted^Talk/_{Contributions} 15:15, 4 July 2006 (UTC)

Well, population geneticists are mainly interested in population genetics, so this is not too surprising. Ecologically-oriented people are interested in all forces that affect populations, whether or not they act on heritable mechanisms. I think we're getting hung up here on the idea of a definition; as a mathematician, as I once was, definitions are important because of the deductions that follow from them, but most dictionary definitions aren't intended to be rigorous in this sense, but are only intended as providing characterising descriptions, and it doesn't matter much in this case if they are circular. However, evolutionary theory has come under attack because of the contrast between its claims for deductive rigour and explanatory power, and the contrasting looseness of definition of the core concepts. There are two key potential circularities: 1) The forces of selection, however you describe them, cannot know whether the traits selected for are heritable or not, so to include heritability in the definition entails defining natural selection by its effect. 2) Traits cannot be defined as favorable in any objective, meaningful sense except by their success in being selected for. It is possible to avoid these two circularities without difficulty when it is important to argue rigorously, and I think it is important to show that these circularities are not intrinsic to the edifice. Gleng 17:27, 4 July 2006 (UTC)

I don't accept the implied charge against the Darwinian definition. In terms of effect how is 'differential preservation of phenotypic characteristics' different from 'differential reproduction of individuals'. Surely 'differential reproduction' is also an 'effect'? — Axel147 17:58, 4 July 2006 (UTC)

In fact, neither Darwin, nor Fisher use natural selection in a single consistent sense; Fisher for instance does not include sexual selection as natural selection; now we would. Gleng 17:27, 4 July 2006 (UTC)

Well now I have produced substantial evidence of the positions of Darwin, Fisher, Dawkins and Endler in this talk forum, I challenge those who prefer the phenotypic defintion, in particular Gleng and Kim, to come up with a text which defines natural selection their way. As has already been noted Lande heavily qualifies his usage as 'phenotypic natural selection'. Authors who use 'differential reproduction' don't shed any light on the problem: that can be interpreted as either position in my diagram. — Axel147 09:37, 5 July 2006 (UTC)

Wallace 1871, "On Natural Selection" uses Natural selection as Survival of the Fittest (p302, "meaning simply that, on the whole, those who die are least fitted to maintain their existence" quotng Darwin "Origin" chap III.)

This is clearly circular as a definition, but fine as a description. The circularity can be removed by defining natural selection as the natural processes of selection, as against the artificial processes used by farmers, in which case "Natural selection comprises the natural processes by which some individual organisms survive and reproduce while others do not."

The theory of natural selection states that, if there is heritable variation in a population, then when a component of selection is selection for a heritable trait, that trait will become enriched in the population, and adaptive evolution will take place. The confusion is between whether you use Natural Selection as a name for the theory, which clearly is a common shorthand usage, or strictly for the processes of selection. If you use it for the processes of selection, and seek to explain some effect by this, then it is essential to avoid defining natural selection by that effect, for then your argument will be vacuous. However, this article is not an academic article, and it's most important to be clear; I'd rather avoid well known traps Gleng 16:12, 28 June 2006 (UTC)

How is this mechanically different from defining natural selection as "Natural selection comprises the natural processes by which some phenotypic characteristics are preserved while others are not." I don't see that you can dismiss one defintion as a 'theory' rather than a set of processes. In the first case the hypothesis is that reproduction is nonrandom (i.e. biased according to properties of the individual). In the second case the hypothesis is that preservation of phenotypic characteristics is nonrandom. Both can be cast as theories. I don't see the difference. — Axel147 11:43, 5 July 2006 (UTC)

Axel, I will retract myself from this article once again, I am just not willing to discuss this issue anymore. Please see Futuyma's evolution for modern thinking about this. -- Kim van der Linde ^{at venus} 12:07, 5 July 2006 (UTC)

Kim, I have withdrawn my complaint! You are right that Futuyma uses the phenotypic defintion and you've quoted him to that effect and you are right that the debate about heritability is pretty much exhausted. Thirdly you are right that your original opening line is better. Whether one says natural selection 'is nonrandom differential reproduction' or 'the existence of consistent fitness differences' the idea is that reproductive success is correlated with traits. This seems to have been lost from the currect opening line. — Axel147

If we can get away from the perpetual discussion about definitions, I am very willing to continue to improve this article, for which I have enough ideas. I am just not willing to continue with this extended definition discussion that has been exhausted as far as I am concerned. Both definitions are in the article, and I am sure I can find at least two or three different ones more to add if I wanted. I rather would like to focus at the more interesting stuff, which is about how the process of natural selection works in detail, and how it helps in the nice changes we can observe in nature. -- Kim van der Linde ^{at venus} 17:29, 5 July 2006 (UTC)

OK,:How is this mechanically different from defining natural selection as "Natural selection comprises the natural processes by which some phenotypic characteristics are preserved while others are not." This is fine as a description or characterisation of natural selection. To put the creationist argument bluntly, it is in part that the theory of natural selection is dogma not science, comprising unscientific (unfalsifiable) assertions. There is some validity in this because if we define anything by its effect, then we lose the power for it to explain that effect - and if we define natural selection in the above way, the statement that some traits are enriched in a population as a result of natural selection becomes tautological, unfalsifiable, and vacuous; it is simply an assertion of something true by definition, but which might be trivially true if the definition does not in fact apply to anything. If natural selection is defined in an open and simple way however, the same statement becomes a bold prediction and is falsifiable, and is true only when some sources of variation are heritable and when there is some component of selection for heritable traits. Gleng 16:11, 5 July 2006 (UTC)

Well, you seem to be starting with the explanation you wish to make, then defining natural selection in a way which aids that. The debate (which I have given up on from now) is terminological. The theory taken as a whole doesn't somehow lose its predictive power or become unfalsifiable just because of our choice of words. But I do see the trap you are trying to avoid. My personal view is that by using a different definition from Darwin we risk falling into another. Anyway I will accept the phenotypic defintion from now on. — Axel147 17:12, 5 July 2006 (UTC)

Yes, it's true that I'd prefer a definition that makes this key argument easier to make, and yes I agree that the power of the theory isn't affected one way or another; as for using a different definition from Darwin - well Darwin said things in different ways at different times too, but I think the main sense in which he used natural selection is as opposed to artificial selection - the emphasis was that the processes were natural, undirected, with no intelligent designer overseeing the process; he took God's role away. Variation and heredity are the mechanisms by which natural selection can result in evolution, and are the key parts of the theory of evolution, but the shock to the system is that this needs no GodGleng 10:00, 6 July 2006 (UTC)

If we are serious about this

If we are serious about coming up with a good lead paragraph, I think that two things have to happen:

1. We have to realize that we are not really arguing about the core principle of Natural Selection. I believe we are in agreement, even if we may not want to admit it. What we differ in is the approach. Population geneticists tend to use a "bottom-up" approach, which involves working on the parameter space of the problems, and, in some ways, leaving the genotype->phenotype connection to the estimation of those parameters. Ecological geneticists take a "top-down" approach, which involves working extensively in the phenotype space of the problems, with obvious concerns about understanding the phenotype->genotype connection. For the most part, we can leave out Darwin altogether, other than for historical interest. Darwin's ideas of Natural Selection were accomplished without a correct understanding of inheritance (pre-Mendelian).
2. The condescending attitudes have to stop. Both approaches are used today by modern, albeit by different, researchers. They are both valid approaches, and they are both useful. Because the core principle of Natural Selection is the same, we should be able to find that core principle, then talk about how the various approaches are different.

The alternative is continuing to argue continuously and pointlessly. Ted^Talk/_{Contributions} 12:28, 6 July 2006 (UTC)

I'm sorry if I've seemd to be condescending, certainly don't feel that; i was just trying to explain the relative advantages of one position as I saw it; I agree that there is nothing to separate us in the core position, and probably nothing to separate us at all except in differences in how best to write this article clearly. At least we're keeping these discussions here not in the article. I've done my bit and am happy to withdraw nowGleng 16:08, 6 July 2006 (UTC)

I think the condescending remark is directed at my debate with Kim! Kim is right in that there does seem to be a slight shift in opinion over time. Just a word on my personal view as to why we have 2 definitons. I don't think the change is caused by greater understanding of inheritance, or other mechanisms of evolution. Nor do I think it's caused the introduction of the word 'evolution' into the English language. I attribute the change to historical misunderstanding of the original meanings of 'natural selection' and 'survival of the fittest', and a shift to a definition which is more friendly towards quantitative analysis, one which also lends itslef to making the kind of explanation Gleng was talking about above. Maybe Futyama plays a part by influencing people with his preference. — Axel147 16:51, 6 July 2006 (UTC)

After some thought, I think I pass on trying this, and I will follow the steps of Gleng in this case. -- Kim van der Linde ^{at venus} 19:33, 6 July 2006 (UTC)

Small point overview paragraph

The sentence in the overview is 'some traits are determined by just a single gene, but most are affected by many different genes.' But is this correct? Marcosantezana once said in this talk 'the fitness-wise most important phenotypic variation is either genetic but not heritable in the narrow sense or is not genetic at all'. Kim has expressed similar views in separating selection from response. — Axel147 13:59, 6 July 2006 (UTC)

The purpose of this sentence was to dismiss the idea that there is a one to one relationship (in general) between a gene (or an allele) and a single function (or trait). There isn't a "gene" for say "big muscles". but a huge number of genes, variation in which will have some effect on muscle growth, and in probably every case the variations will have other consequences too. So the sentence was a warning that the link between genetic vand phenotypic variation is not simple. It could be argued that there are no traits determined by a single gene, but this comes down to how you define a trait and what you mean by determined by. For example, in c elegans, mate seeking behaviour is said to be determined by a single gene because ablation of that gene eliminates the behaviour; however many other genes influence that gene so... If the point your getting at is that many phenotypic traits that affect fitness are influenced strongly by environmental and epigenetic factors and have no heritable component, then yes, absolutely right. Gleng 12:19, 7 July 2006 (UTC)

What I was trying to get at is is it possible to have a trait that is influenced by no genes at all? On second thoughts this seems impossible. (Even if a trait is taught or happens by some accident the capacity to be taught or have the accident must somehow influenced by genes.) — Axel147 15:30, 7 July 2006 (UTC)

This is very well possible, but also think about maternal effects, leaning, dominance, epistatsis, GxE, etc. -- Kim van der Linde ^{at venus} 15:43, 7 July 2006 (UTC)

You're quite right Axel, and I can see that there is a looseness here. A trait doesn't really mean anything useful unless you think of it strictly as a phenotypic feature that varies between individuals in a population. So the important question is not whether all traits are determined by genes, obviously they all are, but how much of the inter-individual variation in traits is genetic, and for some traits all the relevant variation may be non-genetic. Any suggestions?Gleng 15:49, 7 July 2006 (UTC)

The first of Mendel's postulates is that traits are unitary. One of the traits he studied was dwarfism. The genius of his postulate is that there is quite a bit of variability of height in pea plants, both genetic and environmental, but it is obvious when the plant is dwarf or normal. Albinism (tyrosinase-, if you want to limit it to specify a single gene) is a unitary trait. However, not all tyrosinase- albinos have exactly the same skin or hair color, but you can quickly identify the albinism genotype (even between tyrosinase+ and tyrosinase- if you are good at it). Mendel was going beyond the phenotype to look at the genotypes. In real and meaningful ways, every gene is affected by lots of other genes as well as the environment. How you treat those differences is dependent on your approach to genetics. Ted^Talk/_{Contributions} 16:59, 7 July 2006 (UTC)

first sentence again

I don't mean to be a pain in the ass but I feel pretty strongly that the word "environment" or "population" belongs in the first sentence. Can anyone come up with an elegant way to accomplish this? I promise, if all of you disagree with me, I will drop it. Slrubenstein | Talk 17:02, 7 July 2006 (UTC)

Natural selection can be caused by for example mate choice, which is not environment. Furthermore, it acts on individuals, not populations, and the statistical methods that require populations to estimate parameters is not equivalent to natural selection. If we want to be precise, we even have to get away from individuals in the first sentece, because there are other units of selection as well. -- Kim van der Linde ^{at venus} 17:26, 7 July 2006 (UTC)

"Natural selection encompasses processes by which individual organisms reproduce at greater or lesser rates compared to others."

Of course, this doesn't work if you believe you can have natural selection with a single individual.Ted^Talk/_{Contributions} 21:47, 7 July 2006 (UTC)

This is the statistical method to detect natural selection, not natural selection itself. -- Kim van der Linde ^{at venus} 21:55, 7 July 2006 (UTC)

We aren't talking about estimation or testing here. Just the existence of natural selection. No statistics is required. I know of no published sources that allow for natural selection with one individual. The basic equations of population genetics reduce down to H-W when there is only one individual -- no selection. I'd love to see any sources you might have the say otherwise. Ted^Talk/_{Contributions} 00:44, 8 July 2006 (UTC)

Lets try this. If I have a single object of any kind, can I select that or not, or do I have in that situation no choice if I want to have it or not? If I like it, I select it in the store and buy it, but when I do not like it, I leave it and do not buy it. NatSel works the same way. The reference in that case is not other individuals, but feeling, or something like that. Suppose we have a population that is going extinct because they can not manage in the changed environment, aka, they are selected against. This works fine till we have two individuals, and NatSel can select the one of the two that is least adapted first. Then the last individual dies. Suddenly, the selective forces that brought the species to near extinction do not select anymore, but it is now something compeltly different? A species does not go extinct by chance, but by maladaptation to for example the environment.

In general, biologists deal with fitness, and those are easier to deal with as you can make the relative. However, nature deals with absolute fitness (reproductive success), and when fitness (of a population, genotype, whetever) is larger than one, the numbers will increase. When it is smaller than one, it will decrease.

Most often, NatSel is a name for statistical differences in reproductive success among genes, organisms, piopulations or species. This is what most genetisists, statesticians etc use as a working definition. Suppose now that a species is limited to a single genotype, and all phenotypes are exactly the same, as if they were produced in a factory, there are no fitness differences, and as such, there could be no natural selection. However, if they are maladapted to the environment, they do not survive, and the number decrease until the species goes extinct. Is that not natural selection? And if not, what would you call that. -- Kim van der Linde ^{at venus} 01:50, 8 July 2006 (UTC)

I'm neutral on this point. I would like it to be natural selection. If the process could cause differential reproduction, if there were fitness differences and if there were more than one in a population. Even if these conditions are not met I agree the process/mechanism is still the same in the sense the reproductive success is determined by environment (which may or may not include other indiviuals of the same species). But this is at odds with the 'existence of consistent fitness differences'. We would have say something like 'Natural selection is the set of processes that determine an individual's number of offspring according to its traits.' — Axel147 10:33, 8 July 2006 (UTC)

You misunderstood my request. I didn't ask for an explanation. I already understand what you are saying. What I would like to see if any publication that uses selection in that way. It is common to use Malthusian parameters to define selection; it is OK to use phenotypic differences to define selection, even when you are unsure if the trait is heritable. My problem is that I think this use of the term selection trivializes the process of natural selection. Ted^Talk/_{Contributions} 00:56, 9 July 2006 (UTC)

I will have to find some older references from many years back when I did not yet add everything to endnote. This is a topic that I have encountered mainly amomg scientists that work with endangered species, as the traditional differential reproduction definitions do not help you out on small populations of endangered species that are genetically homogeneous and are subject to selection. It will take a week or two before I can get around to that, there are pressing deadlines that I have to get (NSF etc). -- Kim van der Linde ^{at venus} 01:46, 9 July 2006 (UTC)

Thanks. I'd be very suprised if you found any. Even the intuitive biologists in the area (such as Soule) don't describe things in that way. When the population goes to extinction, eventually, selection becomes less of an issue and random drift becomes the dominant factor, the realm of us theoreticians. Unless, of course, you use this expansive definition of selection. Ted^Talk/_{Contributions} 04:10, 9 July 2006 (UTC)

Humm, wierd idea that NatSel disappears when the population become smaller when it near extinction due to maladaptation. Talking about a contradictio in terminus. -- Kim van der Linde ^{at venus} 04:24, 9 July 2006 (UTC)

My apologies. I didn't explain it clearly. The common rule of thumb (see Ewens book, for example; or Kimura and Ohta) is that selection can be effectively ignored if s<1/Ne. When N goes toward 1, the dominant power is random drift. Selection doesn't disappear, it just takes a back seat to other forces. Ted^Talk/_{Contributions} 04:46, 9 July 2006 (UTC)

What you stateticians are saying is that maladaptation does not play a role anymopre when the population becomes to small. In other words, a population can not go extinct because of maladaptation. So, theoretically, a fully maladapted species can live forever because NatSel is defined away when it get closer to extinction. Where is the biology? -- Kim van der Linde ^{at venus} 04:53, 9 July 2006 (UTC)

Natural selection has a role in evolution, but there is little sense in forcing everything to be due to natural selection. Ted^Talk/_{Contributions} 05:24, 9 July 2006 (UTC)

I would be the first to stress that other mechanisms, such as drift, geneflow, founder effects and chance play an important role in evolution. No, I am just talking about the key here, maladaptation and the role of nature in selection in that leading to extinction. -- Kim van der Linde ^{at venus} 12:58, 9 July 2006 (UTC)

No, what you are trying to do is to redefine selection so as to lose all real genetic meaning. Growth parameters can not be so easily trivialized. I'd suggest you look at the role of absolute fitness (which include growth parameters) in any set of selection equations. Of course, that is what "we statisticians" would do, so I don't suppose it has any validity with you. Ted^Talk/_{Contributions} 14:51, 9 July 2006 (UTC)

Indent reduction. Exactly, it is absolute fitness that counts. And absolute fitness can be estimated for a single population, single genotype, without comparing with another group. -- Kim van der Linde ^{at venus} 15:16, 9 July 2006 (UTC)

I suggest you actually look at the equations. In discussions of selection, absolute fitness cancels out. Depending on how you define parameters, the denominator is the mean fitness. This discussion has degenerated into pushing truthiness instead of being a search for understanding. Is there anything productive in this anymore? If there is, I don't see it. Ted^Talk/_{Contributions} 18:52, 9 July 2006 (UTC)

No, the discussion has gone from biology to statistics, and in that shift, relevant biology has been lost. Nature selects against maladapted individuals. So, even without variation in fitness, nature selects. What is curious is that that by some operational definitions of NatSel (like the differential reproduction one), that is not NatSel anymore. Natural selection in the first place is a biological concept, the concept that nature does select on individuals, and tends to eliminate maladapted individuals. Nature does not make a comparison, it acts on single individuals. and if a population as a whole is maladapted, it will select that population away. Differential reproduction is a usefull operational definition to address some questions about NatSel, but it is not the same as NatSel. -- Kim van der Linde ^{at venus} 19:01, 9 July 2006 (UTC)

Nope. No statistics. There is genetics. There are models. There is mathematics (there is a difference). There is science. When you take out genetics from natural selection, all you have left is Intelligent Design. As I said, any kind of intelligent discussion seems to have stopped several exchanges ago. Ted^Talk/_{Contributions} 21:02, 9 July 2006 (UTC)

I have no abjection to this last paragraph by KimvdLinde. to respond to her earliest objections to my request, isn't it true that while selection may be sexual or act on individual organisms, it nevertheless acts on organisms or pairs within a particular environment, or part of a particular population? Slrubenstein | Talk 10:22, 8 July 2006 (UTC)

Sure, they are in an environment, but if the environment is really nice to them and does not really result in selection relative to the effect the sexual selection has on the survival of individuals, it is difficult to say that it is.-- Kim van der Linde ^{at venus} 01:46, 9 July 2006 (UTC)

---

The more I read the first sentence, the worse it seems. What do we mean by "processes"? How pollen is distributed? Sperm mobility? I read this, and I immediately see biological processes. While those are differences are useful in natural selection, they are not the process of natural selection. Does Intelligent Design fit this first sentence (who says God isn't a process)? I won't even get started on encompass.

Unfortunately, it looks like a sentence created by a committee, "full of sound and fury, signifying nothing." Ted^Talk/_{Contributions} 21:58, 7 July 2006 (UTC)

as I indicated, it is not my preferred sentence. -- Kim van der Linde ^{at venus} 22:07, 7 July 2006 (UTC)

as I also indicated, not my preferred sentence. I think it is better to emphasise the link between individual repoductive successs and traits. — Axel147 10:33, 8 July 2006 (UTC)

This was the version as of June 22. My impression was that this was a consensus version among Glen, Ted, Kim, Samsara, and Axel: "Natural selection is the process by which individual organisms with favorable traits are more likely to survive and reproduce." We can certainly just go back to it. The reason I started mucking with it was three-fold: first, NS is a differential process, so it isn't enough to say that some organisms survive and reproduce - they do so more than others; others surive and reproduce less; this is one major reason why I think it is important to specify that we are talking about individual members of a population. Second, while I agree 100% with Axel that traits should be included, the term "favorable" seemed tautological to me. We know that a trait is favorable because nature selected for it; sometimes we then need to figure out what made it favorable. True, sometimes it is obvious but once we assume it is always obvious we are on the road towards teleological arguments about evolution. Third, and a related point, is that traits are favorable only vis a vis something else which is why I wanted to include reference to the environment in which these organisms live. Can we address my three concerns and still have a consensus? Slrubenstein | Talk 10:43, 8 July 2006 (UTC)

I would prefer to go back to that for the time being. When my deathlines (sterfstreepje in my natuve language) are past tense, I am pretty sure I can work out a better, albeit longer version that is easier to understand, and actually even more inclusive (currently, individuals should actually be replace by unit's of selection or entities or something like that. The longer I think about this, the more I am convinced it is impossible to get down to a simple straighforward definition that is at the same time comprehensible for lay people. -- Kim van der Linde ^{at venus} 01:46, 9 July 2006 (UTC)

My contributions suggest I was not here from 19 June to 27 June. Nonetheless, it is good enough for now. I'm sure we can come up with a decent definition if we work on a core definition. The problem will come if editors insist on adding exceptions and contingencies. The core definition must give a sense of natural selection and its importance to modern biology. In the same way, when I teach Mendelian genetics, I state the 5 postulates (or, 2-3 Laws, if that's the way you count them). We spend some time talking about each, and when they know them, I then hit them with the fact that there are exceptions to each and every one of them. I hope we are able to do this with natural selection. Ted^Talk/_{Contributions} 04:10, 9 July 2006 (UTC)