Dear professors, Prof. Rohlf and Dr. Cardini thanks very much for your suggestions! Your sincerely:
Angélica María Cuevas Student of Bilogy Colombia National University ----- Mensaje original ---- De: morphmet <[EMAIL PROTECTED]> Para: morphmet <[email protected]> Enviado: martes, 27 de noviembre, 2007 4:46:36 Asunto: Re: outlines analysis Fourier coefficients Dear Angelica, there are certainly more appropriate ways for doing this, but a simple option for appreciating how much the number of harmonics affects the spatial relationships of your specimens could be to start with a high number of harmonics, compute pairwise Euclidean distances (ED) between all specimens based on EFA coefficients and then repeat this operation by progressively reducing the number of harmonics. Then, if you do a matrix correlation between the initial ED (largest number of harmonics) and those based on fewer harmonics, and you plot r onto the number of harmonics, you should be able to see when an increase in the number of harmonics does not make any appreciable difference. This is somewhat analogous to a scree plot to decide the number of PCs to use in an analysis and equivalent to what you can do for the same purpose in landmark based geometric morphometrics by plotting correlations between ED based on the first, say, 5, 10, 15 etc. PCs and Procrustes distances (PRD) onto the number of PCs. In your case, instead of PRDs you'll have ED based on the largest number of harmonics. For landmark based data, this approach is described in: Cardini A., Jansson A-U., Elton S., 2007 - Ecomorphology of vervet monkeys: a geometric morphometric approach to the study of clinal variation. Journal of Biogeography, 34: 1663-1678 A more detailed example can be found in: http://biocenosi.dipbsf.uninsubria.it/atit/PDF/Volume11(1)/11(1)_9.pdf EFA coefficients can be computed in Morpheus or NTSYSpc, for EDs you can use NTSYSpc or almost any other statistical software, and for matrix correlations NTSYSpc or Mantel (link in the SUNY morphometrics website). This does not answer your question about number of harmonics vs sample size. Beside Prof. Rohlf's suggestion, I suspect that you might be able to learn something about how (un-)stable your coefficients are by performing some kind of rarefaction analysis where you repeatedly compute parameters based on progressively smaller samples. Examples of rarefaction analyses are, to my knowledge, mostly from the literature on disparity analysis (Foote, to start, but also Zelditch, Stayton etc. for geometric morphometrics). If I remember well, David Polly also has an example of rarefaction analysis and in this case it is applied to matrices of variances and covariances. You can check in his website. Good luck. Cheers Andrea Dr. Andrea Cardini Lecturer in Animal Biology Museo di Paleobiologia e dell'Orto Botanico, Universitá di Modena e Reggio Emilia via Università 4, 41100, Modena, Italy tel: 0039 059 2056532; fax: 0039 059 2056535 Honorary Fellow Hull York Medical School The University of York, Heslington, York YO10 5DD, UK E-mail address: [EMAIL PROTECTED] <mailto:[EMAIL PROTECTED]>, [EMAIL PROTECTED] <mailto:[EMAIL PROTECTED]>, [EMAIL PROTECTED] <mailto:[EMAIL PROTECTED]> http://hyms.fme.googlepages.com/drandreacardini http://ads.ahds.ac.uk/catalogue/archive/cerco_lt_2007/overview.cfm#metadata -- Replies will be sent to the list. For more information visit http://www.morphometrics.org <http://www.morphometrics.org/> ------------------------------------------------------------------------ Comparte video en la ventana de tus mensajes (y también tus fotos de Flickr). Usa el nuevo Yahoo! Messenger versión Beta. Visita http://e1.beta.messenger.yahoo.com/ -- Replies will be sent to the list. For more information visit http://www.morphometrics.org
