partition residuals in a constrained ordination

Hei,

is there an approved way how to partition variance into explained vs unexplained at the level of sites?

After performing a constrained ordination (CCA or dbRDA) I want to check if explained variance varies in a systematic way along given gradient, in a univariate analysis, like plotting the absolute residuals of a regression analysis along a given gradient.

I've tried using rstandard in vegan, but am not sure this allows to partition like in a lm (TSS = RSS + ESS).

thanks for suggestions

Robert