morphmet <[EMAIL PROTECTED]> wrote:
I am examining variation in butterfly morphological
data, and decided to learn about geometric
morphometrics. In addition to reading about the
theory behind, I have been using tps. To understand
better what the different programs are doing, I have
tried to derive different variables from the outputs,
and used them in other statistical packages. I am
feeling more confident about what I have learned, yet
I have a couple of questions and I would greatly
appreciate your help with them:
* The variables used in the relative warp analysis
(tpsRelW) correspond to the procrustes distances? If
so, can these distances becalculated
from the aligned matrix that has the landmark
coordinates after doing the procrustes?
You have to used of Weghit Matrix (W) as your data (shape
variables) that is calculatable from tpsRelw, then put the W
into the standard statistical packages such as SPSS,
STATGRAPHICS, SIGMASTAT, NTSYSpc and so on (it refers to your
preference and there isnt any obligation), but you have to do
some corrections for inputing the W or the matrix of Partial
Warp Scores, e.g. vectorized or rectangulared that means you
have to bring all variables in a single line, for instance in
this way you could used of Editpad that is free downloadable
from web.
Please note that tpsRelw used of Supesimposition method to
alighning your data and it calculated automatically.
About RWA, it's like a analog of PCA but on your shape variables
and you could do it in common statistical packages that you
mentioned and carried out, in tpsRelw & tpsRegr gave you the
scores of shape varibles in your analysis (those scores that
have eigenvalue more that 1), after pushing the bottom of RWA,
look after the plots to see your individuals across the RW1 &
RW2 and looking their dispersion, after that you could input
your shape variables into tpsRegr to see the shape changs by TPS
grids in extreme positive and extreme negative positions at the
RW1.
tpsDig gave you the raw coordinate date (X,Y in 2D and X,Y,Z in
3D) then you would inputing your raw coordinate data into
tpsRelw to obtained the partial warps and then Weight matrix
that is your database for your future multivariate analyses.
* Does the plot of relative warps correspond to a PCA
biplot, that used as input the procrustes distances?
Where can I find the PCA scores for the landmarks? The
scores for the specimens corresponde to the relative
warp matrix?
Please note that there is a bit diffences between the number of
your Landmarks and the number of variables in this method, for
example if you have P landmarks you will have 2P-4 shape
variable that reduced in alignment (rotation, scaling,
translating and ...) so in this way you cant have any
interpreted on your landmarks if you are intersted on it you
have to used of tpsRegr to seeing the shape changs using tps
grids.
* I have tried all of the above, and used the
procrustes distances in other statistical packages.
In general the PCA results look similar BUT not
identical. So I am wandering whether this is a result
of my incipient knowledge of procrustes methods and
tps or perhaps that the programs use a different
algorithm.
I would like to thank you in advance for your kind
help. Have a Merry Christmas and a Happy New Year!
=====
Alexandra Herrera Martinez
Estudiante de Maestra
[EMAIL PROTECTED]
Laboratorio Dr. Richard Thomas
Departamento de Biologa
Universidad de Puerto Rico
San Juan,PR 00931-3360
P.O. Box 23360
U.S.A.
Hope it was usefull to you, anyway I'm a novice in this way and
may be the specialist in this field could help you more than me,
it was only the principals of geometric morphometrics that I
told you.
I'm working on Lep. too, please fill free to contact me if I
could helpfull to you!!!
Take care
Regards
Reza
_________________________________________________________
Do You Yahoo!?
Informacin de Estados Unidos y Amrica Latina, en Yahoo!
Noticias.
Vistanos en http://noticias.espanol.yahoo.com
--
Replies will be sent to the list.
For more information visit http://www.morphometrics.org
************************************************************************
Eng. Reza Zahiri,
M.Sc of Entomology,
Insect Taxonomy Research Department (ITRD),
Plant Pests and Diseases Research Institute (PPDRI),
Evin/Tabnak St.,
P.O.Box 19395,
Tehran 1454, Iran.
Tel: +98(21) 2403326.
************************************************************************
________________________________________________________________________
Do you Yahoo!?
Yahoo! Mail - Helps protect you from nasty viruses.
--
Replies will be sent to the list.
For more information visit http://www.morphometrics.org
