----- Forwarded message from Soledad Esteban -----

Date: Sat, 24 Nov 2012 03:05:55 -0500
From: Soledad Esteban
Reply-To: Soledad Esteban
Subject: Re: Do you need to know size for Procrustes superimposition?
To: [email protected]

Dear Patrick:

You do not need to know the real size of the individuals to do a Procrustes superimposition, it just would compute the centroid size from the relative picture size and scale all the individuals to the same centroid size. With this you remove differences due to differences in scaling when taking the pictures, but if there are some differences in shape related to differences in size (allometry) they would be present in your sample after superimpose them despite your specimens are fitted to the same size. The only way of get ride of allometric scaling is knowing the real size of each specimen (thus centroid size really represents the size of each specimen). In biology allometry many times is quite an important factor, and usually it is better to get the real size information whenever it is possible and check for the importance of allometry on them. Indeed, depending on your study reviewers may ask you to control for allometric factors, i.e.: are the shape differences you get due to differences in size? And if yes, you may want to remove the allometric effect working with the residuals of a regression between shape and centroid size (or log centroid size). Thus, the real size is not important for GPA, but it may be for your biological conclusions.
Therefore it is up to you wether you take the risk and ignore the size factor which you may will need later, but no need of it just for GPA.

I hope this helps!

Best regards
Sole

Soledad De Esteban Trivigno
Area de Paleobiología
Institut Català de Paleontologia
Edifici ICP, Campus de la UAB
08193 Cerdanyola del Vallès
Barcelona. Spain
www.icp.cat


-----Mensaje original-----
De: [email protected]
A: [email protected]
Fecha: 11/24/12 07:38
Asunto: Do you need to know size for Procrustes superimposition?


----- Forwarded message from Patrick Kennedy -----

Date: Wed, 21 Nov 2012 21:19:26 -0500
From: Patrick Kennedy
Reply-To: Patrick Kennedy
Subject: Do you need to know size for Procrustes superimposition?
To: "[email protected]"

 
 
Hi,

I'm completely new to morphometrics, so sorry if this question seems a bit basic/obvious to those in the know!

I have a roughly 400 individuals, which are divided into nine different groups, and I want to find out whether these different groups have different mean shapes. I'm going to use MANOVA here. But to get to that point, I'll need, of course, to do Procrustes superimposition. Each group was photographed at a different time, so the scale differs each time. I have a reference to scale each group to, but it would be a long and tedious process to go through each individual setting the scale. Is this necessary? Could I just run the Procrustes superimposition as it is with individuals photographed at different distances? 

In other words (and sorry if this seems a stupid question), do I need to tell the software the actual/relative sizes of the specimens in order to do the superimposition? I can't really see why you would need to (though I am a humble undergraduate just beginning here, so I thought I would check).

Many thanks to anyone who can advise. 

Patrick 


 
 


----- End forwarded message -----





----- End forwarded message -----



Reply via email to