I am using the adonis function in the **vegan** package to determine differences in dissimilarities in a community (PCB congeners) between several different factors.

`## Adonis Model pcbtest3 <- adonis(pcbcong ~ BASIN + FISH_CLASS + REACH, data = pcbcov, method = "bray", permutations = 999) ## pcbcong=matrix of community dissimilarities `

`BASIN`

has 3 levels, `FISH_CLASS`

has two levels and `REACH`

has two levels. I would like to use this procedure to run a model where `BASIN`

is nested and directly test for community similarities between factors, `FISH_CLASS`

and `REACH`

. Does using `adonis()`

within vegan seem reasonable?

**Contents**hide

#### Best Answer

I am currently using this package as well for water quality data where I took samples longitudinally along a stream corridor. You have basin and reach, which are both interdependent (i.e. sites in the same basin are more likely to be the same, as well as those on the same reach). I would use reach and basin as blocking terms in the strata variable. You want to know if fish class is dissimilar, but you don't want the fact that they're in the same basin or reach confusing things:

PERM <- adonis(mydata ~ fish_class, permutations = 999, method = "bray", data = alldata, strata = basin + reach)

See if that works…. you can always run the fish class against a single blocking term and do it twice (reach first, then basin) to make sure the interdependency isn't confounding your results.

### Similar Posts:

- Solved – Nested design — adonis function in vegan R package
- Solved – Nested design — adonis function in vegan R package
- Solved – the reason behind the name “adonis” (for permutational MANOVA)
- Solved – Permutational MANOVA and Mahalanobis distances in R
- Solved – adonis in vegan: order of variables or use of strata