[R] CCA - manual selection

[Jari Oksanen]

Fabricius Domingos <fabriciusm <at> gmail.com> writes:

> 
> Hello,
> 
> I am trying to obtain f-values for response (independent) variables from a
> CCA  performed in vegan package, to see which ones of them have
> significative influence in my dependent variables (like the manual selection
> in canoco), but I can't find any function (or package) that do such a thing.
> The dependents variables are species data, and the independents are
> ambiental data.
> 
> Than you.
> 
Dear Fabricius Domingos,

Use add1/drop1 for the vegan object with argument test = "permutation". For
adding constraints, you need a 'scope' of variables considered for adding to the
model. 

The following example shows how to do this with vegan package and one of its
standard data sets:

> library(vegan)
> data(mite)
> data(mite.env)
> mbig <- cca(mite ~ ., mite.env)
> m0 <- cca(mite ~ 1, mite.env)
> add1(m0, scope = formula(mbig), test="perm")
> m0 <- update(m0, . ~ . + Topo)
> add1(m0, scope = formula(mbig), test="perm")
> m0 <- update(m0, . ~ . + WatrCont)

You can also see if some of the included variables should be dropped:

> drop1(m0, test="perm")

The usage of add1.cca is explained on its help page (see ?add1.cca).

Cheers, Jari Oksanen