Talk:Fisher's geometric model

Figures needed

This article really needs visual aid. It's Fisher's geometric model after all. If free graphics are not available on the web, then can someone please create some?Trashbird1240 (talk) 16:38, 12 January 2012 (UTC)

I'd like to improve this article, but I don't have time right now, so I'm going to leave my notes and suggestions in case anyone else has the opportunity. Below is what I would suggest as a recipe for fixing the article:

1. rename the first section "Fisher's argument" and add a second section with a title something like "Subsequent Developments". Move the statement about Orr into the second section.
2. rewrite the first section to improve explanation and remove misleading implications
   1. The FGM actually is not an evolutionary model in the sense of specifying the dynamics or mechanism of evolutionary change. It is simply an argument to justify the idea that adaptation, especially multidimensional adaptation, takes place via infinitesimal adjustments. It is a kind of meta-optimality argument of the form "evolution must work like this, because this is the most sensible way for it to work." Kimura turned it into more of an evolutionary model by introducing the idea of the probability of fixation for new mutations.
   2. There should not be any suggestion in this article that Fisher was talking about evolution by new mutations, which was not part of his belief system (e.g., read his opening sentence in the next chapter, ch. 3). Fisher's argument nowhere refers to "mutations" (though immediately after presenting the argument, Fisher invokes it to reject the idea of large mutations having a role in evolution). In the Modern Synthesis view, evolution was a matter of shifting frequencies of multi-locus combinations of alleles already in the gene pool, not a matter of complete fixations of new mutations. It was Kimura who looked at this as an argument about new mutations. This is truly distorted in Orr 2005 (Genetica 123:3-13). Orr presents it as though Fisher really meant to have a mutation-fixation model, but got "confused" (p. 5), and Kimura really meant to have a multiple-step adaptive walk, but got confused about that, too, i.e., Orr presents a whiggish history in which previous scientists really meant to invent the Gillespie-Orr mutational landscape model, but got confused and fell short. I hope there is some review that correctly indicates Kimura's role. "Fisher geometric model" refers, or ought to refer, not to Kimura's innovations, but to Fisher's way of understanding the problem of adaptation as a movement by discrete steps in a continuous phenotypic space-- a simple space in which there is one optimum (a point) and the system is already close to it.
   3. replace the current microscope analogy with Fisher's actual text, which is shorter and more authentic: "If we imagine a derangement of the system by moving a little each of the lenses, either longitudinally or transversely, or by twisting through an angle, by altering the refractive index and transparency of the different components, or the curvature, or the polish of the interfaces, it is sufficiently obvious that any large derangement will have a very small probability of improving the adjustment, while in the case of alterations much less than the smallest of those intentionally effected by the maker or he operator, the chance of improvement should be almost exactly one half."
   4. explain the geometric model using a figure similar to that in Fig 1 of Orr, 2005 (Genetica 123: 3-13).
   5. possibly, include a graph identical to Fig 3 in Fisher, 1930, showing the multidimensional generalization of the argument. You can find it on google books here: http://books.google.com/books?id=sT4lIDk5no4C&printsec=frontcover#v=onepage&q=40&f=false
3. write a second section to cover subsequent developments. .
   1. Kimura revived Fisher's argument in his 1983 book, and also changed the argument into a quasi-model of fixation of new mutations based on their size of effect. Because he changed the argument fundamentally, he reached a different conclusion-- the distribution of effect sizes would have an intermediate maximum, rather than being the decreasing dist in Fig 3 of Fisher 1930.
   2. various later authors looked at Kimura's argument as a model of a one-step adaptive walk (e.g., Hartl & Taubes, 1996).
   3. Orr (also Welch & Waxman 2003) considered the distribution of effect sizes in a multi-step adaptive walk
   4. I'm not sure what is the best overall recent review of this material. The contemporary view is that the distribution of effects fixed during adaptation is not the same as Fisher imagined, but includes a mixture of smaller and larger changes. This should be an important part of section 2.

I'm sorry for not providing complete citations. I'll try to fill those in later. Dabs (talk) 14:49, 14 November 2013 (UTC)