Weighted vs Unweighted Unifrac interpretation

[Joseph Edwards]

I'm sorry if this question has already been posted before or if this is even the correct place to ask this.

I'm wondering if it is appropriate to use weighted vs unweighted unifrac as an analysis technique.  For instance in my experiment I want to test how different plant species may affect a microbiome.  Sending a weighted unifrac distance through permutational manova i find that plant species indeed describes a significant amount of the variation in the microbiome data.  But species is not significant with unweighted unifrac.  

Okay, I know unweighted is more sensitive to rare OTUs (which I was expecting for the species effect), but it doesn't indicate that microbiomes are effected by plant species.  Could I use this result to say that the species differentiation is mainly due to quantitative difference between OTUs instead of a presence/absence difference?  Basically, I'm wondering if it's appropriate to compare the results of weighted vs unweighted unifrac to answer biological questions.

Thanks,

Joe

[Greg Caporaso]

Hi Joe,

That is how I would explain that pattern. 

Greg