----- Forwarded message from al cardini -----
Date: Mon, 19 Nov 2012 08:00:15 -0500
From: al cardini
Reply-To: al cardini
Subject: Re: Semilandmarks
To: morphmet@morphometrics.org
Dear Kalpana,
I know at least a few papers (e.g., Ivan Perez et al., 2006; possibly,
Bastir et al., 2006-7; some articles by Dave Sheets et al.; Sanfilippo
et al., 2010, Experimental Eye Research 91 405e414) where people
compared different methods for sliding (including non-sliding at all)
and possibly different densities of semilandmarks. However, I am not
an expert on these techniques. Philipp Gunz and the other Viennese
people (broadly speaking) are those who pioneered the methods. They've
written several papers including a few introductory ones.
At a very introductory level, you can find something on this topic,
and several refs (including some of those by Philipp etc.), also in a
recent article definition I wrote on geometric morphometrics. It's
still a preprint but I am happy to send it to you, if you want it.
I must say that my general purely empirical experience is that sliding
(using one or the other method or non-sliding at all) hardly changes
size estimates and often does not even change hugely the shape
distances between the specimens. Results also, when I compared methods
for instance testing group differences, were generally similar (P
values and % of variance explained by groups) and often similar also
to results from the same landmarks (when there were enough) without
the semilandmarks.
However, even if I am more interested in the sensitivity (or
non-sensitivity in my simple cases) of results to using different
methods, there might still be some benefits in sliding (see Gunz et
al. various papers).
Be particularly careful with semilandmarks when you have small samples
and you're doing parametric stats. Again, in some of the papers I
mentioned, you'll find suggestions about what to do (resampling
methods using distances, maybe, or dimensionality reductions, which
might be tricky but sometimes hard to avoid).
Be also aware that sliding reduce the df in the shape coordinates. I
found suggestions in the literature that 1 dimension is lost for each
semilandmark being slid, but that's not exactely what I find in
practice. However, if you do a PCA and discard all the highest order
dimensions with zero variance, that will 'tell' you the real number of
df in your data and will avoid at least some of (but not all) the
problems with parametric tests (wrong number of df if the software
'does not see' the redundancy in shape coordinates).
Of course, whether changing parameters when using semilandmarks
(method to deal with semilandmarks but also their density) makes a
difference or not is most likely to be strongly specific to your data
and samples.
It would be nice to explore a little bit if you can find clues about
why you're getting differences.
I am sure you'll get better suggestions from the real experts.
Good luck.
Cheers
Andrea
On 19 November 2012 05:48, <morphmet_modera...@morphometrics.org> wrote:
>
> ----- Forwarded message from Kalpana Das -----
>
> Date: Sun, 18 Nov 2012 09:15:34 -0500
> From: Kalpana Das
> Reply-To: Kalpana Das
> Subject: Semilandmarks
> To: morphmet@morphometrics.org
>
> Hey all,
> I am currently doing shape analysis of frog's snout shape using sliding
> semilandmarks method as there are very few landmarks.
> As there are two different ways to slide semilandmarks ,I am a little
> confused which one to use.
> I used both of them and they are generating different results.I need some
> insights into this problem.
> It would be equally great if some of you have good literature on this topic
> and willing to share with me.
>
> Thank you
>
> Best,
> Kalpana
>
> --
> Kalpana
> Center for Ecological Sciences
> Indian Institute of Science
> Bangalore,India
> Email- kkalpanaa1...@gmail.com
> Mobile no-9620313751/8010255368
>
> "Not everything that counts can be counted, and not everything that can be
> counted counts."
>
>
>
>
>
> ----- End forwarded message -----
>
>
>
--
NO DOCX, XLSX AND PPTX, PLEASE!!!
COAUTHORS AND CONTRIBUTORS, PLEASE, FOR SHARING FILES, CONVERT THEM IN
DOC, XLS (UNLESS LONGER THAN 256 COLUMNS) AND PPT
Dr. Andrea Cardini
Researcher in Animal Biology
Dipartimento di Scienze Chimiche e Geologiche, Università di Modena e
Reggio Emilia, l.go S. Eufemia 19, 41121 Modena, Italy
Honorary Fellow
Functional Morphology and Evolution Unit, Hull York Medical School
University of Hull, Cottingham Road, Hull, HU6 7RX, UK
University of York, Heslington, York YO10 5DD, UK
Adjunct Associate Professor
Centre for Forensic Science , The University of Western Australia
35 Stirling Highway, Crawley WA 6009, Australia
E-mail address: alcard...@gmail.com, andrea.card...@unimore.it,
andrea.card...@hyms.ac.uk, andrea.card...@uwa.edu.au
Webpage: http://sites.google.com/site/hymsfme/drandreacardini
Datasets:
http://ads.ahds.ac.uk/catalogue/archive/cerco_lt_2007/overview.cfm#metadata
Editorial board for:
Zoomorphology:
http://www.springer.com/life+sciences/animal+sciences/journal/435
Journal of Zoological Systematics and Evolutionary Research:
http://www.wiley.com/bw/journal.asp?ref=0947-5745&site=1
Hystrix, the Italian Journal of Mammalogy:
http://www.italian-journal-of-mammalogy.it/
----- End forwarded message -----
