Hi everyone, I am currently trying to analyse communities on natural and artificial reefs, and I decided to use Permanova (Vegan :: Adonis 2) in R to do so. However, being the first time I am running this statistical test, I want to be careful with what I do, and therefore, I am asking for your help. I collected my community data (species at each site), and I would like to test whether reef Age, Depth, and Material (my independent variables) affect biodiversity. However, I do not have the same number of replicates for each group within variables (e.g. Age - young(10), middle(7), and old(5)). How can I adjust my code to account for the unbalanced design of my data? Here is the code I am using (but which I think assumes balanced design).
adonis.1
Hi Manuel and Uwe,
thank you for your responses, they were very helpful to me. I do not feel confident enough yet to modify codes in R by myself, but probably I will give it a try. And Uwe, that is very good to know of Primer-e. If I had more time I would probably consider that too, but as it is for my undergraduate thesis, time is running quickly and deadlines are approaching. Definitely, I will consider it for future analysis. Thank you both.
Best regards,
Emanuele
And, if it helps, neither Permanova nor Anosim require balanced designs. A lack of balance will reduce the number of possible permutations, and as discussed in the Anderson paper there may be some (marginal) effects on the Permanova statistic, but in general for a design like yours you'd be fine just carrying on with the standard approach.