One could do that computationally but I would worry about the homology it might 
imply for such variable leaf shapes. You might try it and then check to see if, 
for example, a lobe on one leaf might be 20% of the linear distance around the 
outline but in another it might be 30% of the way around. In such a case the 
lobe on one leaf would effectively be treated as homologous to a location 
between lobes on another leaf. If so, does that make biological sense for your 
study? If simple leaves were also included the implied homology of a point 
along its outline to that of one of the lobed leaves might be pretty arbitrary. 
Would be better if one knew something about the development of these leaves 
(which I do not!) and used that knowledge. 

 

An alternative would be to use outline methods to group shapes for the purpose 
of say identification with little implication that groups need be biologically 
meaningful. Sorry to be rather negative but I find highly variable leaf shapes 
difficult to put in a simple standard framework. Perhaps others will have 
better suggestions.

 

____________________________________________

F. James Rohlf, Distinguished Professor, Emeritus. Ecology & Evolution

Research Professor, Anthropology

Stony Brook University

 

From: Tina Klenovšek <tina.klenov...@um.si> 
Sent: Wednesday, March 14, 2018 12:28 AM
To: f.james.rohlf <f.james.ro...@stonybrook.edu>
Cc: morphmet@morphometrics.org
Subject: RE: [MORPHMET] digitizing very variable leaves

 

Thanks to everyone who answered.

 

Dear Dr. Rohlf,

I assumed great variability would be a problem.

I was thinking the best way of digitizing would be to draw a curve on the 
outline of each leaf (simple and lobed leaves). Resample curves by length with 
approx. 40 landmarks. Then append curves to landmarks. Two landmarks, the leaf 
tip and base, would be fixed others could be defined as semi-landmarks. Is this 
sensible?

Alternatively, we could analyse simple and lobed leaves separately.

Thank you a lot for your help.

 

Tina

  

 

From: f.james.rohlf [mailto:f.james.ro...@stonybrook.edu] 
Sent: Wednesday, March 14, 2018 3:59 AM
To: Tina Klenovšek
Subject: Re: [MORPHMET] digitizing very variable leaves

 

An assumption of the usual GMM methods is that shape variation is "small". I 
think these leaves exceed this quite a bit! Another problem is landmarks. How 
to match leaves with and without lobes?

 

__________________

F. James Rohlf, Distinguished Prof. Emeritus 

Dept. Anthropology and Ecology & Evolution 

Stonybrook University

 

-------- Original message --------

From: 'Tina Klenovšek' via MORPHMET <morphmet@morphometrics.org 
<mailto:morphmet@morphometrics.org> > 

Date: 3/12/18 10:56 AM (GMT-10:00) 

To: morphmet@morphometrics.org <mailto:morphmet@morphometrics.org>  

Subject: [MORPHMET] digitizing very variable leaves 

 

Hello everyone,

 

we would like to digitize tree leaves that are very variable (from simple oval 
to strongly lobed on one tree). 

1.)    I am wondering if TpsDig can do some kind of automatic digitizing like 
the LeafAnalyser software: 
http://www.plant-image-analysis.org/software/leaf-gp, which evenly distributes 
a defined number of landmarks on the leaf outline... 

LeafAnalyser does not seem flexible or precise enough. Or I can’t use it 
properly. Any experience?

 

2.)    Is it possible/sensible to put objects that are so differently shaped 
(photos attached) into the same group or is it better to analyse simple and 
lobed leaves separately?

 

I apologize if similar questions have been already answered... 

 

Kind regards,

Tina

 

 

 

 

 




doc. dr. Tina Klenovšek, koordinatorica doktorskega 

študijskega programa Ekološke znanosti
Univerza v Mariboru | University of Maribor

Fakulteta za naravoslovje in matematiko
Faculty of Natural Sciences and Mathematics
Koroška cesta 160, 2000 Maribor, Slovenija

T: +386 41 808 366

E:  <mailto:ime.prii...@um.si> tina.klenov...@um.si,  
<http://www.fnm.uni-mb.si> www.fnm.um.si

 

 

-- 
MORPHMET may be accessed via its webpage at http://www.morphometrics.org
--- 
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 morphmet+unsubscr...@morphometrics.org.

Reply via email to