I am a bit frustrated looking for a competent function doing variation partitioning (e.g. on linear model output)
it should be able to tick the following boxes
optionally:
There are a few functions ticking some of the boxes,
i.e. the varPart()-function from the modEvA package. The most obvious flaw imo here is that it can only deal with 2 or 3 explanatory variables. Secondly, the input is quite tedious, were you have to enter the numeric values of each R² for each combination of explanatory variables (although it is not to hard to write a workaround)
Then there is the varpart()-function from the vegan package, which only deals with distance matrices as far as I can tell. The variancePartition-package (from bioconductor) has a calcVarPart-function, which only seems to calculate the sum of 'pure' partition and overlaps (I think?)
I am confused that there is neither a tool for such a prevalent task nor an thread about it. Or am I missing something?