Hi all,
I have a dataset of ~ 50 species means for two traits that are correlated
with each other, and with a species mean of an index of environmental water
supply. I want to quantify how strongly trait 1 and trait 2 are correlated
when controlling for their confounding correlation with water supply.
Normally, I would use a partial correlation, but because water supply is
correlated with both traits, it seems like that would be inappropriate here
(based on Murray & Conner 2009, Ecology). To get around that, I would
usually fit a model of trait 1 ~ trait 2 + water supply and use the
independent effects analysis in the hier.part package to calculate the
effects of each predictor on trait 1. However- and this is the reason I'm
posting to R-sig-phylo- pgls in the caper package shows that there is a
phylogenetic signal (Pagel's lambda significantly greater than 0) in the
correlation between trait 1 and trait 2. I'm not sure how to account for
this phylogenetic effect in an independent effects analysis.
One possibility I thought of was to use pgls in caper to find the optimum
lambda value for the full model (trait 1 ~ trait 2 + env), fit the subset
of models (trait 1 ~ trait 2 and trait 1 ~ env) using the same lambda, and
compare the r2 values in the same way as hier.part would. Does that sound
like a reasonable way to proceed, or is there another approach anyone else
would recommend to control for both phylogenetic non-independence and the
confounding correlation with water supply to quantify the correlation
between trait 1 and trait 2?
Thanks a lot for your help!
Best,
Megan
[[alternative HTML version deleted]]
_______________________________________________
R-sig-phylo mailing list - R-sig-phylo@r-project.org
https://stat.ethz.ch/mailman/listinfo/r-sig-phylo
Searchable archive at http://www.mail-archive.com/r-sig-phylo@r-project.org/
