I have been scratchiong my head for a few days on this problem, which should be pretty simple. I really hope someone can point me in the right direction.

I have a matrix of data, 40 samples (dispatched in 6 groups of inequal size) and 50 species (454 pyrosequencing data if you want to know everything).

The data contains a lot of 0, so I applied an Hellinger transformation. Also I tested the normality, and some species are clearly not normaly distributed.

Now I want to find out which species vary between the groups.

I can use JMP to perform a Krustal-Wallis (non-parametric) test with a Dunn post test, but I can only do that for 1 species at a time, which is quite time consuming considering I have 50 species, and many more to come. Same problem with Instat : 1 species at a time. I tried using R with Rcmdr and FactomineR, but it wont let me perform a Krustal-Wallis test (the option is greyed out and I don't know why).

Anyone knows how I can perform a Krustal-Wallis test with a Dunn's post test on all the species at once ?

Thanks !!