-------- Original Message --------
Subject: Re: Procrustes form space
Date: Sat, 2 Jul 2011 07:17:24 -0400
From: Ondřej Mikula <[email protected]>
To: [email protected]
Now it is quite clear. There is nothing special neither in tpsRelw, nor
in my data. The discrepancy arose due to my misunderstanding of outputs
from the R package 'shapes'. But thank you for your comments, they
started my mind to find the right answer.
Ondrej Mikula
On 30 June 2011 17:33, morphmet <[email protected]
<mailto:[email protected]>> wrote:
-------- Original Message --------
Subject: RE: Procrustes form space
Date: Wed, 29 Jun 2011 19:04:40 -0400
From: F. James Rohlf <[email protected]
<mailto:[email protected]>>
Reply-To: [email protected] <mailto:[email protected]>
Organization: Stony Brook University
To: [email protected] <mailto:[email protected]>
Perhaps the documentation was not clear - the tpsRelw program does
use ln(centroid size). I will update the help file to make that clear.
----------------------
F. James Rohlf, John S. Toll Professor
Dept. Ecology and Evolution, Stony Brook University, NY 11794-5245
Please consider the environment before printing this email
-----Original Message-----
From: morphmet [mailto:morphmet_moderator@__morphometrics.org
<mailto:[email protected]>]
Sent: Wednesday, June 29, 2011 12:33 PM
To: morphmet
Subject: Procrustes form space
-------- Original Message --------
Subject: Procrustes form space
Date: Wed, 29 Jun 2011 06:56:40 -0400
From: Ondřej Mikula <[email protected]
<mailto:[email protected]>>
To: [email protected] <mailto:[email protected]>
Dear list,
I have one question more. I want to perform an analysis in
Procrustes form
space (size-shape space). It seems that in tpsRelw the option
"Procrustes
form space" leads to PCA on Procrustes shape coordinates
augmented by a
column of centroid sizes. However, Mitteroecker et al.
(2004) suggest to add column of logarithms of centroid sizes.
As far as I understood argument in Appendix of Mitteroecker et al.
(2004), log-csizes should be preferred because their variation
is comparable
to that of Procrustes shape coordinates. However, in my data set
log-csizes
have much lower variance than any of the Procrustes shape
coordinates. In
effect size differentiation appears to be totally unimportant,
although the
largest skull in my data set is about 30% larger than the
smallest one.
Could anyone clarify the issue?
Best wishes
Ondrej Mikula
Institute of Animal Physiology and Genetics Brno, Czech Republic