#The following text are codes in R
library(vegan)
data(dune)
dune
data(dune.env)
dune.env
results
Hey Kai Liu,
to my Knowledge this is a gap in the vegan adonis function and it have no post hoc test. But you can try some Contrast Coding
https://www.researchgate.net/post/R_Script_for_adonis_with_more_levels_of_nestedness
http://thebiobucket.blogspot.de/2011/08/two-way-permanova-adonis-with-custom.html
http://www.ats.ucla.edu/stat/r/library/contrast_coding.htm
You can build a contrast matrix for all pairwise comparisons you're interested in, consider "H X" in place of the original "X" design matrix (H is the contrast matrix), and do it basically by hand. Or you can just do a pairwise t-test as described in the link hereafter.
http://www.r-bloggers.com/r-tutorial-series-one-way-anova-with-pairwise-comparisons/
If a procedures provides p-values you can always use the p.adjust() function. A safe option is the Holm procedure:
> p.vector p.adjust(p.vector, 'holm')
Hello Kai,
as far as I know, it's not possible to run pairwise comparisons in vegan using adonis().
I guess you could do it artificially by selecting each time different subsets of data with couples of treatments (e.g. SF+BF, SF+HF and so on). But maybe it requires some kind of adjustments (not sure about that).
Anyway, you can find several discussions on this issue. For example here:
http://www.mothur.org/forum/viewtopic.php?t=3897
or here:
https://stat.ethz.ch/pipermail/r-sig-ecology/2009-August/000731.html
With kind regards
Paolo
The email is sent by Dr. Paolo.
Thank you for all of the question followers.
Hey Kai
Just Click on the picture below the Link and everything should be fine
it is just a discussion on the topic
best Leon
Have you checked the function pairwise.perm.manova?
frome the package RVAideMemoire
I think it works
Hi there
My p-value from adonis was significant, but the pairwise.perm.manova did not show a significant p-value among my 6 groups comparisions...is it possible?
Hi Katia, it is definitely possible that your overall MANOVA is significant and none of your pairwise comparisons are significant. p-values from pairwise comparisons are generally adjusted by some p.adjust method to make them more conservative because of the increasing chance of making a type II error.
https://github.com/pmartinezarbizu/pairwiseAdonis
Cheers. Pedro
A similarity profile analysis can also work. You can see clustsig library.
- Badges
- Science method
- Similar topics
- Biological Science
- Ecology