morphmet  

Re: permutation test for shape trajectory between sexes

morphmet
Thu, 12 Nov 2009 06:47:13 -0800



-------- Original Message --------
Subject:        Re: permutation test for shape trajectory between sexes
Date:   Wed, 11 Nov 2009 00:30:19 -0800 (PST)
From:   Dean Adams <dcad...@iastate.edu>
To:     morphmet@morphometrics.org
References:     <4af9a02c.6090...@morphometrics.org>



Stefan,

Yes, assessing the angle between vectors is the way to go.

A general approach for determining whether phenotypic trajectories
differ is to first run a factorial MANOVA, with population and sex as
main effects, along with a pop*sex interaction term.  A significant
interaction term implies that the sexual dimorphism vectors are not
consistent between populations.  However, be aware that for multivariate
data, a significant interaction term could mean a number of things. For
instance, it could be that the direction of sexual dimorphism is
consistent between populations, but the amount (magnitude) of sexual
dimorphism is greater in one population than it is in the other.
Alternatively, the two populations could have similar magnitudes of
sexual dimorphism, but be oriented in different directions. Finally,
both the magnitude and direction of sexual dimorphism could differ.  As
such you need to run several additional analyses to investigate these
possibilities.

We recently developed an analytical approach for assessing such patterns
in multivariate vectors and trajectories.  Our approach quantifies
attributes of phenotypic trajectories (magnitude, orientation, etc.) and
statistically assesses them using residual randomization. The paper
describing the general approach is:

Adams, D.C., and M.L. Collyer. 2009. A general framework for the
analysis of phenotypic trajectories in evolutionary studies.
/Evolution/. 63:1143-1154.

and earlier papers:

Collyer, M. L., and D. C. Adams. 2007. Analysis of two-state
multivariate phenotypic change in ecological studies. /Ecology/. 88:683-692.
Adams, D. C., and M.L. Collyer. 2007. Analysis of character divergence
along environmental gradients and other covariates. /Evolution/. 61:510-515.

All of these papers are available on my web page. Additionally, there is
some R code associated with the 2009 paper for implementing the approach.

Hope this helps.

Dean

Dr. Dean C. Adams
Associate Professor
Department of Ecology, Evolution, and Organismal Biology
Department of Statistics
Iowa State University
Ames, Iowa
50011

At 06:17 PM 11/10/2009, you wrote:


-------- Original Message --------
Subject: permutation test for shape trajectory between sexes
Date: Tue, 10 Nov 2009 05:39:12 -0800 (PST)
From: Stefan Schlager <stefan.schla...@uniklinik-freiburg.de>
To: morphmet@morphometrics.org
References: <4af5219d.1070...@morphometrics.org>

Dear all,

I am currently working on mandibles of two populations and I want to
prove,that there is no common trajectory explaining sexual dimorphism in
both populations.

Analyzing the PCs of the pooled data implied  the ladder- or at least
that there is no common separator between sexes along the axis of the PC
coordinate system.
I now wonder, if I could prove this hypothesis by a permutation test on
the angle of the residual vectors (in the PC coord. syst) between the
sexes means of the two populations.

What I thought was: if the trajectory was about the same, it means that
the residual vector between the sexes' means should point roughly into
the same "direction". -and if this angle is larger for the actual
populations compared to those of randomly assigned populations, this
would proof my hypothesis.

Am I completely misleaded, or could this work?

Many thanks
Stefan

--
Stefan Schlager M.A.
Medizinische Fakultät - Anthropologie
Hebelstr. 29
79104 Freiburg
Tel: +49(0)761/203-5522
Fax: +49(0)761/203-6898




--
Replies will be sent to the list.
For more information visit http://www.morphometrics.org <http://www.morphometrics.org/>

--
Replies will be sent to the list.
For more information visit http://www.morphometrics.org