-------- Original Message -------- Subject: Re: Phylogenetic size correction with sexual dimorphism Date: Fri, 11 Nov 2011 15:01:56 -0500 From: Dean Adams <[email protected]> To: [email protected] <[email protected]> Emma, Adding very short branches to the tips of your phylogeny for males and females in each species won't necessarily cause the phylogenetic VCV matrix to be singular, just as having polytomies won't necessarily cause this matrix to be singular. In fact, adding very short branches was one early way (albeit clunky) in which within-species variation was included in PGLS analyses; prior to Felsenstein 2008 and other more recent approaches (see e.g., Ruber and Adams 2001: J Evol. Biol.). So you might give this approach a try for your project. Best, Dean -- Dr. Dean C. Adams Associate Professor Department of Ecology, Evolution, and Organismal Biology Department of Statistics Iowa State University Ames, Iowa 50011 www.public.iastate.edu/~dcadams/ phone: 515-294-3834 On 11/11/2011 1:53 PM, morphmet wrote:
-------- Original Message -------- Subject: Phylogenetic size correction with sexual dimorphism Date: Fri, 11 Nov 2011 14:26:10 -0500 From: Emma Sherratt<[email protected]> To: [email protected] Dear fellow morphmetricians, I am stuck on the methods for phylogenetic size correction in my study looking at sexual shape dimorphism across a number of species. I have size and shape data for 50 species, averaged for males and for females in each species. The averages have been made from measuring several individuals per sex per species. So data structure is n = 100 (50 males, 50 females). I want to correct my shape data for size, taking into account phylogenetic non-independence, so that the phylogenetic size-corrected data in the original format, i.e. for each species an average for males and average for females. This is because I want to examine the sexual shape dimorphism in a PCA morphospace. Normally to correct for size in geo morphometric shape data, I would use a multivariate regression (a la Monteiro 1999 Sys Biol) and take the residuals for subsequent analyses. For correcting for evolutionary allometry, I would take the average shape and size for each species, and run a multivariate regression using either independent contrasts or make a phylogenetic VCV matrix from the tree. Where I run into a problem is that for each species, I have data for both sexes, and since I want to study the dimorphism between sexes, I cannot simply use an average for species during the regression. It doesn't work to have zero length branches for males and females in the phylogeny, since this leads to extra independent contrasts, or singular values in a VCV phylo matrix. Any ideas? Regards, Emma --- Emma Sherratt, PhD. Post-Doctoral Fellow in Organismic and Evolutionary Biology and Museum of Comparative Zoology Harvard University 26 Oxford St. Cambridge, MA 02138 [email protected]<mailto:[email protected]>
