-------- Original Message --------
Subject: Re: Symmetry in MorphoJ
Date: Tue, 26 Jul 2011 13:50:21 -0400
From: Ruth Flatscher <[email protected]>
To: [email protected]
Hallo Jacco,
I know Procrustes ANOVA mainly as a tool to assess the variety in your
sample or to quantify measurement errors.
If you want to know, for example, how exact your positioning of
landmarks is, you do several repeats of digitizing on the same
individual and mark them in morphoJ with the same classifier. Or if you
have a number of identically built structures on one individual, like we
botanists often do, you may want to find out how much it varies within
one individual. In this case, you measure for example several flowers
of the same plant, and give them the same classifier. Likewise, if there
is matching symmetry (i.e. if your organism has structures that are
built after the same principle, but as mirror images of each other, like
insect wings, or hands/ feet in humans) you may want to know whether
there is a systematic difference between left and right sides.
Therefore, to answer part of your question 2, I think that the
application purpose of Procrustes ANOVA is simply quite different from a
DFA.
1. I think the term "individual" makes no sense if you have only a
single landmark configuration per individual.
If you have several measurements per individual (either repeats or from
identical structures), a significant output means that there is a
significant difference between them. in case you've digitized repeats,
you should probably think about the quality of your landmarks and
whether it's really possible to place them accurately and unambigously.
If you used different copies of a structure, e.g. flowers or leaves,
within-individual variation is very high and you should consider
measuring several copies and using an individual average. You should
also compare the magnitude of variation within individuals to that
between individuals to find out whether your planned analyses make any
sense at all.
2. When significant, the term "side" tells you whether there is a
systematic difference between left and right sides of a structure with
object symmetry or between the left and right "copy" in case of matching
symmetry. this means that yopu have directional asymmetry, i. e. all
left sides differ from all right sides in a certain way. If you know
your studied organism well, you might perhaps have an idea whether this
could have a plausible biological reason, and you might have discovered
an interesting thing which might deserve further investigation.
(I must say I did not understand part of your question - what you meant
by "all individuals grouped together"? Could you maybe explain that again?)
3. The graphical output shows you the directional asymmetry, i. e. the
nature/direction and magnitude of systematic differences between left
and right sides or left and right "copies" of a structure. It does NOT
show fluctuating asymmetry effects, which occur randomly. If your sample
size is big enough, they will in fact balance each other so that no
considerable overall difference between left and right sides will result
(and I guess the graphics won't show you a great deal in this case).
Hope this helps. ...
With best regards from Austria,
Ruth
-------------------------------------------------------------------
Mag. Ruth Flatscher
Department of Systematic and Evolutionary Botany
Research group Biogeography
University of Vienna
and
Department of Systematics, Palynology and Geobotany
University of Innsbruck
Austria
On 26 July 2011 17:06, morphmet <[email protected]
<mailto:[email protected]>> wrote:
-------- Original Message --------
Subject: Symmetry in MorphoJ
Date: Tue, 26 Jul 2011 06:01:09 -0400
From: J.C. van Rijssel <j.c.van.rijssel@biology.__leidenuniv.nl
<mailto:[email protected]>>
To: <[email protected] <mailto:[email protected]>>
Dear morphometricians,
Regarding the procrustes Anova in MorphoJ, I have a few questions.
1. What does a significant outcome of the term "individual" mean? Does
it mean that there is individual variation between samples in
asymmetry?
2. What does a significant outcome of the term "side" mean? Is there
then no symmetry between sides (e.g. left and right)? So are all
individuals grouped together? Can we compare this with a paired t-test?
And if left and right are grouped together, what is than the difference
with a DFA and a procrustes ANOVA?
3. The graphical output of the procrustes Anova shows the one side
minus
the other side (left-right). Is there with this analysis accounted for
variation of asymmetry? So in other words, is there accounted for the
fact that sometimes distances between landmarks are larger at one side,
and sometimes at the other side which would equilibrate differences in
asymmetry?
Thanks in advance,
/Jacco C. van Rijssel, PhD student/
Section Integrative Zoology
Institute of Biology Leiden
Leiden University
P.O.Box 9505, 2300 RA LEIDEN
Room number 6.5.14a
The Netherlands
Tel: +31(0)71 5274503
Mobile: +31(0)6 41453003
==============================__===========================
Visiting address: Sylvius Laboratory, Sylviusweg 72, 2333 BE Leiden
Website:
http://www.science.leidenuniv.__nl/index.php/ibl/van_rijssel
<http://www.science.leidenuniv.nl/index.php/ibl/van_rijssel>
==============================__===========================
