Leandro's argument that absence or duplication of a landmark might so
fundamentally change the shape of an object that quantative analysis
should be unnecessary must be correct under certain circumstances and
for some purposes, but it is not always true. A structure may be located
along the edge of an object, and its location can be used to help define
the shape of that object. Absence or duplication of that structure may
or may not affect the shape of the edge of the larger object of which
the structure is a component, and one may be interested in the object's
shape irrespective of the presence, absence, or duplication of that
structure.
For example, we digitize a series of landmarks around threespine
stickleback fish using the lateral view. We are often interested in
shape variation within and among populations in relation to
environmental variables or in shape change through time. Some
individuals in our system lack the pelvis, and one of our landmarks is
located on the pelvis. Presence or absence of pelvic structures is
easily recognizable and interesting in other contexts, but it creates
problems for comparison of the shape between populations or different
generations with and without pelvic reduction.
-- Mike Bell
Michael A. Bell, Professor
Department of Ecology and Evolution
Stony Brook University
Stony Brook, NY 11794-5245, USA
Phone:1-631-632-8574. Fax: 1-631-689-6682
I have an opinion about the question presented by Dr. Greiner.
Morphometric methods are best used when variation is continuous around a
mean shape. The kind of shape differences described by Dr. Greiner
(absence or duplication of features) are clearly discontinuous and
therefore do not need a quantitative description of variation. The over
digitizing procedure described would certainly place the mean shapes so
far apart in shape space that their comparison would not require a
statistical test.
Leandro
Em Qui 17 Ago 2006 08:46, morphmet escreveu:
My question for the morphometrics community is how to handle and absent
and duplicated features in your landmark data. I am asking about true
biological variation here, and not just missing data. Because these are
real specimens, you cannot easily disregard some just because they don’t
easily fit into our morphometric methods – or at least you shouldn’t
disregard them for that reason. But the question remains, what do we do?
One suggestion I have had for the absent data is to over digitize the
nearest landmark locations, essentially digitizing each point twice. I
am not sure that this would work logically, but it’s the only solution I
have heard of. I suppose, if you knew that there would be duplicated
features you could create your digitizing scheme with the over
digitizing in place until you reached the duplicated features. I’m still
not sure how well that would work.
So, I send this out to the collective brains of the morphometric
community. What do we do in this situation?
Thanks for you input.
*/Thomas M. Greiner, Ph.D./*
Assistant Professor of Anatomy
University of Wisconsin – La Crosse
Department of Health Professions
4054 Health Science Center
1725 State Street
La Crosse, WI 54601
Phone: (608) 785-8476
Email: [EMAIL PROTECTED]
--
##################################################################
Leandro R. Monteiro
Laboratorio de Ciencias Ambientais - CBB
Universidade Estadual do Norte Fluminense
Alberto Lamego, 2000, Campos dos Goytacazes, RJ,
cep 28013-600, Brasil
E-mail: [EMAIL PROTECTED]
Tel: (55-22) 27261471
FAX: (55-22) 27261472
home page: http://morfolab.infobio.net
##################################################################
--
Replies will be sent to the list.
For more information visit http://www.morphometrics.org
--
Replies will be sent to the list.
For more information visit http://www.morphometrics.org
