Option for post-hoc test after permanova for community data?

[Bob3]

Hi there,

Thank you for reading my post and hopefully someone can provide me with some help!

For my master's thesis, I have used microbial community identification with QIIME. I have 47 samples spread over 4 different treatments and 4 different time points (2 different years and 2 different seasons), so n=3, except for 1 treatment at the first time point. From QIIME I ended up with 1151 genera, where for at least one sample the abundance of a genus is >0. To be able to handle this data I have set a minimum abundance level of 1% for at least 1 sample, which left me with 37 genera. For this data I have used a central log ratio transformation and created a euclidian based distance matrix, and ran this through a permanova in R with the following model:

adonis(formula = X ~ Treatment * as.factor(Year) * as.factor(Season))

From this I get significant results from the factor treatment (p = 0.005) and the interaction between year and season (p = 0.001), as I expected. Now there is no post-hoc test for a permanova, but I am wondering if there is any kind of test that I could do that could tell me which genus/genera are driving this change?

Thank you in advance for your help!

Kind regards,
Bobby

 

[bugman]

You can use simper() in the vegan package. This will give you a decomposition of how much each genus is contributing to the similarity between groups.

 

[Bob3]

Thank you! I have been working with it and struggling to figure it out for a little while, but I've got it. Thank you for your advice!

 

[bugman]

Thanks for following up Bob3.

Often posts are left hanging without resolve. Glad to hear you got it sorted out!