-------- Original Message -------- Subject: Re: predicting selection response Date: Wed, 23 Mar 2011 09:13:29 -0400 From: Chris Klingenberg <[email protected]> Reply-To: [email protected] Organization: University of Manchester To: [email protected] Dear Milos I think there is no simpler way to do what you are asking to do. To predict the response to selection is a rather ambitious thing to do, and so you just have to live with the complexity that goes with it. The mechanics of this sort of analyses is helped by the fact that they are implemented in MorphoJ. See: http://www.flywings.org.uk/MorphoJ_guide/frameset.htm?genetics/resp_selection.htm There is also a more recent paper that uses the same method with some slight modifications: Klingenberg, C. P., V. Debat, and D. A. Roff. 2010. Quantitative genetics of shape in cricket wings: developmental integration in a functional structure. Evolution 64:2935–2951. http://www.flywings.org.uk/PDF%20files/Evol2010.pdf The critical point is that you do need the G matrix. So you do need either breeding experiments or pedigree information from a natural population to estimate the G matrix. There is no way around having genetic information. Some people have used a scalar multiple P matrix as a substitute for the G matrix (G matrix = heritability * G matrix). That is assuming that the genetic and phenotypic covariances are proportional. If this were true, it would make everybody's life a lot easier. The problem is that analyses that have checked this assumption with easonably powerful tests have tended to reject the hypothesis of proportionality. If you cannot get genetic data, my advice would be to stick to the sort of things you can do with phenotypic data. I think there are a lot of interesting things about integration that can be done with phenotypic data, without pretending to do quantitative genetics. I hope this helps clarifying this issue. Best wishes, Chris On 3/23/2011 11:50 AM, morphmet wrote:
-------- Original Message -------- Subject: predicting selection response Date: Tue, 22 Mar 2011 12:07:37 +0100 From: Milos Blagojevic <[email protected]> To: <[email protected]> Hello to all Morphmet users, Again, only a question to clarify things up a bit... If, for example, my study aim was to model hypothesized shape change along some selection gradient, what approach should I use? The procedure outlined in Klingenberg and Leamy`s paper "Quantitative genetics of geometric shape in the mouse mandible" is helpful with respect to information about the way that GM landmark analysis can be linked to multivariate Qgenetic analyses, but it is too complex for me. For starters I would like an easier way to incorporate landmark data in multivariate breeder`s equation and G matrix. I am aware that shape alone must be reduced, e.g. by PCA but can PCA scores be used as direct measurements sensu Lande and Arnold`s classic paper "The measurement of selection on correlated characters"? Finally the simulated selection differentials should depict shape change of certain regions of the skull and then these model shapes used for testing how and if the overall skull morphological integration has changed. Of course, all data is strictly from natural populations and no breeding experiment can be designed, just simulation. Best regards, Milos Blagojevic Faculty of Science Institute for biology and ecology Kragujevac, Serbia
-- *************************************************************** Christian Peter Klingenberg Faculty of Life Sciences The University of Manchester Michael Smith Building Oxford Road Manchester M13 9PT United Kingdom Telephone: +44 161 275 3899 Fax: +44 161 275 5082 E-mail: [email protected] Web: http://www.flywings.org.uk Skype: chris_klingenberg ***************************************************************
