Skip to content

Testing the difference between fractions in RDA

2 messages · Qi Li, Gavin Simpson

4 days later
#
On Mon, 2010-07-05 at 11:59 +0800, Qi Li wrote:
If you mean the section from p2620-2622 in that paper, then no, that
bootstrap procedure is not implemented in vegan. The variance
partitioning part is (in function varpart) but it itself does not do any
testing of or generate confidence intervals on the R^2 components
returned. Looking at the paper it shouldn't be too hard to implement
given all the work that Jari has put into providing various methods for
cca (rda) objects but it is not something I currently have time to work
on.

If you fancy a stab at coding this up, be careful with computing the
coefficient matrices B_X and B_W as you probably don't want to be
computing the inverse of X'X directly. If you do give this a try but
want some help, try the vegan forum on R-Forge

https://r-forge.r-project.org/forum/forum.php?forum_id=194&group_id=68

where we can give further advice etc.

HTH

G