-------- Original Message --------
Subject: RE: How to eliminate the effect of sexual dimorphism?
Date: Mon, 7 Mar 2011 09:36:36 -0500
From: andrea cardini <[email protected]>
To: [email protected]
I suspect that for a comparison between populations, if I've got Tina's
question right, you will have to add those residuals back to a population
mean shape (e.g., the male one*) and thus get 'masculinized' females within
each populations. This will, I guess, require that the pattern of sex
differences is similar across species, which in a traditional ANOVA
framework one would test looking at the interaction term (sex by
population).
There might be an example of this in one (or more) of my papers: possibly
it's the 2010 book chapter on biogeography and blue monkeys (available in
my webpage as other pdfs).
This should be similar in principle to the 'size-correction' done by
MorphoJ and explained in detail also in the Green Book section on MANCOVAs
and allometry, if I remember well. The main difference is that here one has
two grouping variables (sex and populations) and there it's a grouping
variable plus a continuous covariate. The idea of controlling for a factor
(sex or size) after checking that its effect is similar across groups is
the same.
Steve Frost et al., again if I remember correctly, did a slightly more
sophisticated version of this approach in their 2003 paper on baboons,
where they simultaneously 'corrected' for sex and allometry. Steve, please,
correct me if I am wrong.
For 2D data, using freeware software and if one is fine with that kind of
parametric approach, one might be able to do almost everything in PAST
(size) / TPSRegr (shape) plus a bit of manual computation in an xls
spreadsheet to compute sex-corrected shapes. In TPSRegr one will have, I
think, to create dummy variables for both groups and their interaction and
then run a series of analyses following the reasoning of the MANCOVA model
(test of slopes and intercepts) well described in the help file. R also
will allow to do everything. To build the dummy variables, I am sure there
are guidelines on the web.
MorphoJ does not test for interactions as far as I know. Thus one has to do
that test in TPSRegr or another software that does MANOVAs. However, I
wonder whether one might be able to do the sex 'correction' using a dummy
covariate for sex (coded as Markus suggested) and a pooled-within subgroups
regression using populations as subgroups. This would be a shortcut that
avoids manual computations and it is a better implementation of the MANOVA
protocol than the simple 'masculinization' (or 'feminization', if one
prefers to do it the other way round) I suggested. If the pattern of sexual
dimorphism is really the same across populations, the difference between
the first and second way of correcting for sex should be trivial.
I am sure that more experienced morphometricians will make better
suggestions. There's quite a bit on this stuff in the recent and less
recent literature.
Cheers
Andrea
At 07:36 AM 3/7/2011 -0500, you wrote:
-------- Original Message --------
Subject: RE: How to eliminate the effect of sexual dimorphism?
Date: Mon, 7 Mar 2011 04:26:45 -0500
From: Markus Bastir <[email protected]>
To: <[email protected]>
Hello,
If you code the gender as dummy variable ( for males -1 and for females 1)
and use this variable as covariate then you can do a regression of shape on
this dummy and work with the residuals..
markus
Dr. Markus Bastir
Científico Titular
Museo Nacional de Ciencias Naturales (CSIC)
c / J.G. Abascal 2, 28006 Madrid, Spain
tel.: +34 91 566 8976
fax.: +34 91 566 8960
skype: mbastir
web: http://www.evan.at/Members/mbastir
-----Mensaje original-----
De: morphmet [mailto:[email protected]]
Enviado el: domingo, 06 de marzo de 2011 13:13
Para: morphmet
Asunto: RE: How to eliminate the effect of sexual dimorphism?
-------- Original Message --------
Subject: RE: How to eliminate the effect of sexual dimorphism?
Date: Sun, 6 Mar 2011 04:30:51 -0500
From: STRAND-VIDARSDOTTIR U. <[email protected]>
To: <[email protected]>, "morphmet"
<[email protected]>
This is something you can do in the EVAN tool box. I.e. quantify a
vector of shape variation and then remove it from future analyses.
Una
-----Original Message-----
From: morphmet [mailto:[email protected]]
Sent: Sat 3/5/2011 6:53 PM
To: morphmet
Subject: How to eliminate the effect of sexual dimorphism?
-------- Original Message --------
Subject: How to eliminate the effect of sexual dimorphism?
Date: Thu, 3 Mar 2011 04:03:49 -0500
From: tina klenovsek <[email protected]>
To: [email protected]
Dear All!
I hope you can help me!
The thing is that I would like to compare mandibles of five populations
of two different mammal species. The problem is I also have sexual
dimorphism both in shape and size and small samples. Therefore I was
thinking if I could somehow eliminate the part of variation that is the
effect of gender and in subsequent analysis of phylogeny use pooled sexes.
I know how to get the 'allometry-free' shape data and do the test for
common slopes. But what about sexual dimorphism? Is it possible?
Thank you in advance!
Tina Klenovsek
Faculty of natural sciences and mathematics
University of Maribor
Slovenia
Dr. Andrea Cardini
Researcher in Animal Biology
Dipartimento di Biologia, Universitá di Modena e Reggio Emilia, via Campi
213, 41100, Modena, Italy
tel: 0039 059 2055526 ; fax: 0039 059 2055548
Honorary Fellow
Functional Morphology and Evolution Unit, Hull York Medical School
University of Hull, Cottingham Road, Hull, HU6 7RX, UK
University of York, Heslington, York YO10 5DD, UK
Adjunct Associate Professor
Centre for Forensic Science , The University of Western Australia
35 Stirling Highway, Crawley WA 6009, Australia
E-mail address: [email protected], [email protected],
[email protected]
Webpage: http://sites.google.com/site/hymsfme/drandreacardini
Datasets:
http://ads.ahds.ac.uk/catalogue/archive/cerco_lt_2007/overview.cfm#metadata
