Dear Igor,
Thanks for your interest in using Lory. Lory is purely a visualization
program, and does not do any statistical testing. It only performs the
operations necessary to produce visualizations of the relationships
between pairs of forms, using whatever configurations of landmarks are
furnished to the program as input. The program and documentation is
available at https://www.bio.fsu.edu/~dhoule/software.html.
More important though, is how to do similar analyses to those in the
Márquez et al. (2012) paper:
On the example data with /Drosophila/ species authors have computed
sample mean wing shape deformations for the two Drosophila species
(Fig. 4).
Fig. 4 in Márquez et al. (2012) is based on simulated deformations using
the Drosophila wing mean shape as the origin.
o How did they do that? Did they computed the group means in other
programs or R packages and then imported them in Lory or there is a
way to get them in Lory because I haven’t found an option to do that.
One of the ways, as I see it, is to compute the mean shapes in
geomorph and them import those means shapes in Lory to visualise those
means shapes.
Yes, you are correct. Forms that are functions of other forms must be
calculated outside of Lory, then furnished to Lory.
After one iteration of the Delaunay algorithm, without any selection
of evaluation nodes, Lory found 74 nodes for evaluation. Based on that
map of interspecific differences in local wing shape (Fig.5), on the
extracted Jacobian determinants (74) ANOVA with Bonferroni correction
for multiple comparison was performed which resulted with significant
(stars) and non-significant (crosses) differences.
o How did they computed that one shape with displayed interspecific
differences?
I believe that Fig. 5 was produced in Lory, by furnishing each
subcompartment's data to Lory separately.
o Did they extracted Jacobians determinants from Lory and then
performed ANOVA in some other program?
Lory does not output the Jacobian determinants directly. We first
calculated the determinants in Matlab, then analyzed the differences in
SAS.
o How did they know which node was significant or not? I suppose that
you have to extract the coordinates that reflect their position on the
map of interspecific differences or did I get it wrong?
Yes, you do local analyses on the sets of determinants that interest you.
How did they computed the mean representation of intraspecific
variance in Drosophila species? In Lory or in other program?
Again, this is based on simulated data using the Drosophila mean wing
shape. Means were calculated in another program, then furnished to Lory.
Figures 7. and 8. represent interspecific correlations and
interspecific differences and intraspecific variances. I suppose that
these figures were computed in Coriandis software** on Jacobian
determinants? I tried to import the data in Coriandis to explore if I
could compute and get those figures but I am having problems with
assigning group info and names to the imported non landmark data such
as Jacobian determinants – no way I could manage to do that even when
I tried to follow the footsteps in the manual that authors provided.
The left plots in Fig. 7 were produced in Lory, the right side plots
were not. Since the bubble plots do not represent shape directly, they
are not part of Lory. Eladio Márquez made the bubble plots, and Fig. 8.
Eladio Márquez wrote Lory, and performed most of the analyses in the
Márquez et al. (2012) paper. He would probably have more insight into
precisely how some of the Figures you mentioned were produced.
Unfortunately, this paper was published before journals started to force
us to furnish all the code necessary to produce the data. I probably
have all the programs archived, but these are not clearly documented. As
noted in the paper, analyses were performed in SAS, Matlab, Java, C++,
and Python software. I can dig them out and send them on if you wish.
* Márquez, EJ., Cabeen, R., Woods, RP., Houle, D. 2012. The
Measurement of Local Variation in Shape. Evol Biol, 39: 419 – 439.
** Márquez, EJ. & Knowles LL. 2007. Correlated evolution of
multivariate traits: detecting codivergence across multiple
dimensions. J. Evol. Biol., 20: 2334 - 2348.
--
You received this message because you are subscribed to the Google Groups
"Morphmet" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
To view this discussion on the web visit
https://groups.google.com/d/msgid/morphmet2/7c4bfdf0-79a4-50d4-a9a9-20b0b0d77041%40bio.fsu.edu.
