Hello Dragos,
Sounds like you already have your strategy planned out, but just in
case you are looking for an alternative I suggest you look into MRPP
(multi-response permutation procedures). Basically, MRPP is a
multivariate clustering program and the results of the analysis can then
be used to illustrate community structure and relationship to other
measured variables (i.e., as an overlay with joint plot on top of a NMS
ordination). You can then follow with an indicator species analysis to
determine which species (if any) represent groups of lakes.
I use the program PC-ORD, which was developed by plant community
ecologists to address exactly the types of question you are working with
and I think it would serve you well. (PS - I have no financial stake in
PC-ORD, just a happy user).
Good Luck,
Michael
Michael Cooperman, PhD
National Research Council - Research Fellow
in residence at NOAA-Fisheries, NE Fisheries Science Center - Maine
Field Station
Dragos Zaharescu wrote:
I thank Dr. Dan Mckenzie, Dr. Francisco Di Castro and Dr. Mike Hillstrom for
their suggestings. Also thanks Dr. Zuur for previous help. They were all useful
and helped making a more clear idea how to treat the binary plant data.
I also followed the steps for FSO calculation recommended by Dr. Rick Boyce in
FSO webpage. I liked it very much and helped it till one point.
My original question was simple: One matrix with 170 riparian plant species (columns) and 300 lakes (cases). All, binary variables (presence/absence). I want to know how lakes group together based on their similarities, make a categorical variable representing those groups, and then plot the plants on those groups (categories from cat. var) to see, for ex, plants (plant list) associated to a particular lakes group.
I turned back again at the very beginning with this analysis.
What I will try now is
1. performing UPGMA cluster analysis using Sorensen similarity index (from FSO webpage) to infer lakes groups and make a categorical variable (in R).
2. Use this cat. variable to associate plant species to those categories. Most
probably I will transform the raw variables using Hellinger transformation
(from Legendre and Gallager, 2001: Ecologically meaningful transformations for
ordination of species data) (in R) and use Fisher´s classification coefficients
from a discriminant function Analysis (in SPSS) to associate plants to lakes
categories.
I am quite new in R but try the best to learn quick. Following the steps from FSO webpage I find no problem in computing Sorensen matrix (must thank Dr. Boyce once again for leave it clear and simple) but no way I can draw UPGMA clustrer analysis on that matrix.
Is anyone aware of a routine that can combine both, Sorensen and UPGMA cluster.
Unfortunately, my patching did not work.
Many thanks to all that helped once again.
Dragos
Dragos Zaharescu
Animal Anatomy Laboratory
Faculty of Biological Sciences
Vigo University, apd. 137
36310, Vigo (Pontevedra), SPAIN
[email protected]
[email protected]
http://webs.uvigo.es/zdragos/
~ You should be the change you want to see in the world ~ Ghandi
________________________________
From: Richard L. Boyce <[email protected]>
To: [email protected]
Sent: Wednesday, August 19, 2009 2:26:01 PM
Subject: Re: [ECOLOG-L] Fuzzy Set Ordination for classifying plant species ??
It's a little hard to tell from you description what exactly is the
problem. I suggest you visit my FSO web page at
<http://www.nku.edu/~boycer/fso/> and see if the examples there are
helpful.
=0A=A0=0AHello,=0A=A0=0AI have a matrix with 180 plant species (variables, =
binary) and 270 rows (altitude waterbodies). There is also one categorical =
variable (4 categories representing similar lakes groups, which resulted fr=
om a prior analysis of the transposed matrix).=0AIs there any way to load p=
lant species on those categories in fuzzy-set-ordination analysis? Is there=
anyone here that has faced this question before or could provide a hint? A=
ll my efforts so far=A0=A0have lead to loading the lakes on the categories =
(fso in R); but I want to load the plant species.=0A=A0=0AI would greatly a=
ppreciate your help.=0A=A0=0ADragos Zaharescu=0AVigo University=0A=0A~ You =
should be the change you want to see in the world=A0~ Ghandi=0A=0A=0A
