[R-sig-eco] How to choose between RDA and CCA?

Gavin Simpson gavin.simpson at ucl.ac.uk
Thu Aug 4 15:51:08 CEST 2011


On Wed, 2011-08-03 at 21:54 -0300, Diogo B. Provete wrote:
> Dear List.
> In their great book, Bocard et al. (2010) mention that the function
> vegan::anova.cca() can be used to test the significance of each axis of a
> CCA or RDA, and that based on this test it is possible to choose the best
> analysis between the two, if I undertood it clearly. But I'm not sure how to
> interpret the output of this function. I run both a CCA and RDA using the
> formula for data input in both funcions and the first axis of both analysis
> were non-significant.
> What should I do? Or alternatively, If the first axis of CCA were
> significant and that of RDA weren't, should I choose CCA instead of RDA?
> 
> Thank you for your attention!

One might sensibly ask therefore, why do you want either of them?

Without knowing anything of your data, the constraints do not appear to
explain significant amounts of variance in the species/response matrix.

Usually the choice between RDA and CCA boils down to the types of
responses expected in the species matrix. RDA traditionally used for
linear or linear-ish responses, CCA for unimodal ones. Of course things
are never this simple, and CCA has a linear side to its personality and
hence ends up doing well for many data sets regardless of the gradient
lengths. And Legendre and Gallagher have shown that we can apply a
transformation to data that have non-linear response that allows them to
be analysed via PCA/RDA so that one doesn't incur the problems that
CA/CCA has with deviant site/species and site/species/environment
combinations.

A pragmatic person might say, fit both and see which explains the
largest proportion of the variance in the response. But do consider RDA
on a transformed version of the data and not just vanilla RDA.

HTH

G

-- 
%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%
 Dr. Gavin Simpson             [t] +44 (0)20 7679 0522
 ECRC, UCL Geography,          [f] +44 (0)20 7679 0565
 Pearson Building,             [e] gavin.simpsonATNOSPAMucl.ac.uk
 Gower Street, London          [w] http://www.ucl.ac.uk/~ucfagls/
 UK. WC1E 6BT.                 [w] http://www.freshwaters.org.uk
%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%



More information about the R-sig-ecology mailing list