Canonical correspondence analysis

From: Mark Hostetler (
Date: Wed Jun 26 1996 - 19:54:57 MET DST

Hey Folks!

I am new to this list and to using ADE. I have a question about PCA on
Instrumental Variables (I am trying to do a canonical correspondence

Here is my question and data. I have 4 different sites where bird surveys
have been done (thus counts of individuals for each species) and at each
site, I have counts of tree canopy patches of different size ranges (each
size range equals a category with a count of the number of objects). I am
trying to use the variation of the counts of tree canopy patches at each
site to explain the variation in bird species counts.

For the bird species data: I have 4 rows (representing each site) and 46
columns (representing each species). = SPECIES

For the tree canopy patch data: I have 4 rows (representing each site) and
11 columns (representing each tree patch size category). = OBJECTS

Here is how I did the analysis:

I first did a COA on SPECIES
I then did a PCA on OBJECTS weighted (by row) by SPECIES.fcpl

I then did a Triplet Orthonormal Basis on OBJECTS.cnta
I then did a PCA on Instrumental Variables:
- the explanatory variable was OBJECTS.@ob (which by the way had 4 rows
and only 3 columns - I was expecting 11?)
- the dependent variable was SPECIES.fcta
- Output file name A

It seemed to run OK but I wanted the column coordinates for the species
(which I got in file A.ivco). But what I DID NOT GET was a file A.ivfa
(which I think should contain the canonical weights which I can use to plot
the gradients for each Patch category - thus the file should have 11 rows
and 2 columns).

I think I need to have both files to plot the normal two dimension graph
with the position of each species with respect to the gradient of each tree
patch category.

What am I doing wrong (or what am I misinterpreting)? I cannot find a file
to plot the gradient of each tree patch category. Thanks ahead of time for
any help.

P.O. BOX 118525 FAX: 352-392-3704

This archive was generated by hypermail 2b30 : Sat Feb 10 2001 - 10:21:20 MET