Hello José: I send something related called permutational multivariate analysis of variance using distance matrices. It is part of the manual in ecological mutivariado Queco software, developed by the University of Cordova and CATIE, which has been used in other publications. Is in Spanish. Regards

Thank you very much to all of you, Eduard, Oswaldo and Gonzalo. I'll try to find some time for the Adonis software. Gonzalo, also, thanks for the tip regarding the PERMDISP (like in the Primer software). I'll try it!

All the best:

J.

I would suggest that for the next stage, you need to check each variable for stationarity in trend. If any of your tested variables exhibit not stationarity then your PERMANOVA results might be compromized.

You can also follow the stepwise protocols for data exploration established by Zuur et. al., 2010. 

"Zuur, A. F., Ieno, E. N., & Elphick, C. S. (2010). A protocol for data exploration to avoid common statistical problems. Methods in Ecology and Evolution, 1(1), 3-14."

In addition to the excellent answers already given, it may be that you don't need a multivariate test. If your PERMANOVA is significant, it may make sense to do univariate tests on the variables you think are responsible. If that's not a reasonable approach, I second the Adonis approach, and the vegan package in R offers a lot of options for investigating individual factors. Check out envfit().

https://github.com/pmartinezarbizu/pairwiseAdonis

Pedro

Hello José Antonio González-Oreja

I think this depond on your statistics results if there is a significant diference between your samplig / the p-value show a result less than alpha value 5%. A NMDS map can give you vision on your results and selecting the appropriate PERMANOVA applicated in this analysis.

Hope i ansewering your question,

Regards,

SIMPER I guess you already have an nmds now find out how much the species contribute to the difference.