Can you use type III errors in Adonis2()?

[Rob-murphys]

Firstly thanks for an obviously amazing package overall!

I am wanting to run a ANCOVA on my distance matrix to control for genome quality and am thus looking if I can use adonis2() for that ppurpose with the following:

adonis2(GCF_profiles ~ metadata$Host + genome_metrics$N50, permutations = 10000, by = "terms", method = "bray")

However I am given to understand that when conducting an ANCOVA we should be concerned with Type III errors and controlling for type I errors can give the wrong results. Normally one would do with with the Anova() function, like so on the iris dataset:

fit2=aov(Petal.Length~Species+Sepal.Length,df)
Anova(fit2, type="III")

Is there a way to change the error structure in adonis2 as Anova obviously does not accept adonis2 objects as output.

[gavinsimpson]

Won't by = "margin" give you what you want?