4
$\begingroup$

Does any one know why Permutational Multivariate Analysis of Variance Using Distance Matrices (PERMANOVA) is also called adonis?

See https://www.r-bloggers.com/adonis-permanova-assumptions or http://cc.oulu.fi/~jarioksa/softhelp/vegan/html/adonis.html

$\endgroup$
4
  • 2
    $\begingroup$ Isn't that the name of the R function? $\endgroup$
    – T.E.G.
    Dec 1, 2017 at 4:05
  • 3
    $\begingroup$ Well I suspect that it is the package authors' preference without any statistically relevant reason, just like the zelig() function from Zelig package. Is there any other reference to this name beyond the R context? $\endgroup$
    – T.E.G.
    Dec 1, 2017 at 4:55
  • $\begingroup$ bAckronym Density On Namespace Is Saturated? $\endgroup$
    – smci
    Dec 1, 2017 at 9:33
  • 3
    $\begingroup$ Terminology questions are on-topic here. Also, there is a very good accepted answer already. I made another edit and vote to re-open. $\endgroup$
    – amoeba
    Dec 2, 2017 at 14:22

2 Answers 2

5
$\begingroup$

I am the culprit and this is my true confession: it was Hank's idea (Martin Henry H. Stevens). We already had a function called anosim (Analysis of Similarities, of Primer strain) in vegan, and even that was based on dissimilarities, because standard R uses dissimilarities instead of similarities. A regular name for this "permutational ANOVA" of dissimilarities would have been "analysis of dissimilarities" or anodis – but we already had anosim that also uses dissimilarities. So Hank came up with this pun, and I found it cool, so the function was called adonis and the name stuck. As a further twist, there is a company called ANOVA (!) that makes kitchen tools for sous vide cooking, and they seem to have a recipe for ANOVA cooking of adonis veal... https://recipes.anovaculinary.com/recipe/cotelette-de-veau-marine-adonis

$\endgroup$
10
$\begingroup$

Actually it isn't -- the statistical method is not called adonis.

The authors of the vegan software package for R (http://vegan.r-forge.r-project.org) implemented the statistical procedure you refer to in an R function called adonis, but that is just a name for the function, not a name for the method. The name does not appear to be in use anywhere in statistics other than the vegan package. Yet the vegan package is quite popular, so you will see the vegan function name will crop up quite often in online discussions of permutation anova.

The package authors are free to call the function anything they like. Their implementation may have differences or additional capabilities compared to previous implementations of permutation anova, so choosing a new name avoids confusion. "adonis" has a nice ring to it. It sounds a bit like "anova" and includes a number of letters from "distance", so why not? Perhaps the authors also enjoy making a reference to Adonis from Greek mythology, a heroic figure associated with youth and beauty.

My guess is that the authors started with function names anosim for analysis of similarities and anodis for analysis of dissimilarities. Then they realized they could interchange two letters in anodis and make a Greek God. Good for them!

You could write to the vegan authors and ask them whether the function name "adonis" has any other significance, but does it matter?

$\endgroup$
4
  • 4
    $\begingroup$ +1 adonis() is named for analysis of dissimilarities and I presume Hank (the original author) or Jari thought the "anodis" to "adonis" pun funny. PERMANOVA wasn't always called that; originally it was non-parametric MANOVA or permutational MANOVA, & as the idea is quite general but implementational details have changed, via further work of the originator, Marti Anderson and colleagues, as I recall, there are a number of other methods that also do the same thing; dbrda() for example also in vegan. PERMANOVA is also the name of Marti's software, so "adonis" avoids that conflict too. $\endgroup$ Dec 1, 2017 at 16:41
  • 1
    $\begingroup$ @GavinSimpson and Gordon, I have voted to re-open this thread (after doing another edit of the question) and as you should both have enough rep to vote I suggest you do so if you want. $\endgroup$
    – amoeba
    Dec 2, 2017 at 15:40
  • $\begingroup$ @amoeba Thanks, but I don't have enough points to vote. $\endgroup$ Dec 2, 2017 at 20:20
  • $\begingroup$ Indeed it seems that one needs 3k reputation points for this. I have now upvoted two of your older answers so you should gain the voting privilege momentarily. $\endgroup$
    – amoeba
    Dec 2, 2017 at 21:22

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.