[R] Canonical Correspondence Analysis in R

Peter Perkins pperkins at ucsd.edu
Sat Feb 24 04:34:03 CET 2001

I think i mentioned this in private communication a while back, but in
case anyone else cares:

I have a version of canonical correspondence analysis (CCA) coded up in
R.  I based it on

   Ter Braak and Prentice (1988) Adv. Ecol. Res. 18:271-317,
   Ter Braak (1986) Ecol. 67(5):1167-1179, and
   Ter Braak (1995) Section 5.9 in Jongman, Ter Braak, and
      van Tongeren "Data Analysis in Community and Landscape

Actually, there are two versions:  one which follows Ter Braak's
iterative method (slower, but feasible for large problems), the other
which directly uses an eigenanalysis (fast, but uses too much memory for
large datasets).  Although the code does spit out those funny-looking
tri-plots, etc., I don't really consider it "finished", because i found
the references that i had access to very ambiguous in their description
of the method, and the desired outputs seem to have changed subtly over
the years.  In particular, biologists at my lab who used CANOCO got
certain kinds of (eigenvectors) output that were not defined or indeed
even referred to in the literature.  As luck would have it, the CANOCO
manual had mysteriously disappeared, so all my confusion may well be a
function of not having read the right documentation.  And i'm sure that
if i had taken the time to communicate with the founding fathers of CCA,
i could have sorted out what i was missing.

At any rate, these two pretty-well-documented functions are free to a
good home, for anyone who wants to work on/with them.

I have a question for those of you who've used CCA:  do you use it to
determine habitat relationships absolutely, or do you use it to
determine differences in habitat relationships?

- Peter
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at stat.math.ethz.ch

More information about the R-help mailing list