1
$\begingroup$

I am currently working on data representing the abundance of microorganisms in a categorical way, like 0 = no organisms; 1 = 1-5 organisms; 2 = 6-10 and so on (5 being the highest number). And i am not quite sure which dissimilarity index is best to use with vegdist for that kind of data, to then do Pcoa etc. So i wanted to ask if someone has any experience and could recommend an index?

Thanks in advance!

For reference, that what the data is looking like:

$\endgroup$
2
  • $\begingroup$ Is there a reason you are categorizing them this way in the first place? $\endgroup$ Nov 15, 2023 at 23:57
  • $\begingroup$ Yes, i get the data from wastewater treatment plants and they categorize it that way, unfortunately. $\endgroup$
    – Fabi B
    Nov 16, 2023 at 8:41

2 Answers 2

2
$\begingroup$

It looks like you have an ordinal variable. This is sometimes used for plant species but also some microorganisms. I'd first inform R that those columns are an ordinal factor:

factor(myData$free_bacteria, ..., ordered=TRUE)

Then, I'd use some form of model-based unconstrained ordination method to come up with an ordination diagram, which would give you the same result as PCoA.

This is because none of the dissimilarity coefficients available in vegan::vegdist can handle ordinal variables.

On the other hand, gllvm can to that by using gllvm(data, ..., family="ordinal").

The package ecoCopula also has an ordinal family.

$\endgroup$
0
$\begingroup$

Your scaling is like any class scale that are in common use. For instance, vegetation people in Germany use analogical Braun-Blanquet scale. You can use any dissimilarity index of your choice.

$\endgroup$

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.