Tree.shared versus dist.shared


I asked this question before, but without too many details. Maybe I was not clear enough. I am analyzing 40 samples with two treatments (20 samples for each treatment). I want to find out if the communities between the treatments have significantly different structures. I started with tree.shared using the Yue & Clayton theta similarity coefficient followed by a parsimony test. The results indicate that the structure is (highly) significantly different between my two treatments. Then I ran dist.shared with the same calculator followed by an amova test, which indicate that the structure is not significantly different between the two treatments (p=0.39).

There might be something that I’m missing, because I really don’t understand these results… How can they be so different?

Thank you in advance for your help!


Well the tests test different null hypotheses. You might check out my ISMEJ paper from a few years ago that compares these different tests and what they’re trying to do.