Dear Guiherme,

from what I can see you made some errors in your syntax. I assume you want to test the effect of soil on vegetation, so where vegetation is your dependent variable.

Then the syntax would be: adonis2(vegetation~soil), and not the other way around.

Then adonis2(vegation~soil + climate) would test the main effects of both a soil and climate independent variable, while (vegetation~soil*climate) would additionally test the interaction effect of both variables on vegetation.

I hope this helps, best wishes, Erik

Dear Guilherme,

in addition to the Erik´s answer, and regarding your questions #2 and #3, the output of adonis gives you the R2 associated to each of the tested variables, so you could know their relative importance in your analysis.

Check an older discussion in RG: https://www.researchgate.net/post/Can_you_report_PERMANOVA_estimated_components_of_variation_ECV_as_a_proportion_of_variation_explained_by_factors_terms

Best,

Álvaro

Dear Erik and Alvaro, thank you !

Yes, I made a syntax error, indeed. Anyway, you both got it =)

After reading a lot, now I understand the difference between adding group factors (vegetation ~ soil + climate) and multiplying them (vegetation ~ soil * climate). The second option is quite complex to explain in my case, and I will avoid it, since my data doesn´t allow me to go further in discussions. Though soil*vegation gave me the best fit, I really don't know how to interpret this correctly.

Best,

Guilherme

Dear Guilhaume,

you're welcome of course, I am glad it helped. Keep in mind that in the case of soil*climate you are actually testing the explained variation by three variables (soil, climate, and soil:climate) instead of two, so it is only logical that the R2 increases - this is why R2 should not be your only criterium in determining which model is best. Then I think it is a good idea to not only include it because it increases R2, as you indicate above.

Best wishes,

Erik

Yes indeed the formula's should be the other way around:

adonis (vegetation ~ soil + climate + soil*climate)

The soil*climate is the interaction factor. If this factor is significant, this means the effect of soil on vegetation depends on the levels of climate. If it's not significant this effect of soil is the same for all levels of climate.

Hope that helps.