I am calculating Convex Hulls for high-dimensional trait data for a set of communities (Functional Richness) to understand how species pack and fill trait space. I was wondering if these calculations can be done using categorical trait data or are the analyses affected by not having a matrix of continuous values? Is there a way to circumvent this issue?
Dear Mr. Rodrigo,
You can use ImageJ (http://rsbweb.nih.gov/ij/download.html) with plugin 3D Convex Hull (http://rsbweb.nih.gov/ij/plugins/3d-convex-hull/index.html).
Best wishes,
Ion.
I also managed this using PCoA on Gower distances as suggested by http://villeger.sebastien.free.fr/ , it worked well.
@Francesco Maria Sabatini
dbFD in the FD package will calculate FRic as the convex hull volume, but only when the data is normal. With categorical data (e.g. Gower) it will measure the number of unique trait value combinations in a community (see ?dbFD).
George Roff as an addendum to that. dbFD will measure unique trait value combinations if ALL your data is nominal or ordinal, not if some is. so if you have a few continuous traits in your matrix, it should still calculate the convex hull volume. or at least that is what the documentation says.
- Badges
- Science method
- Similar topics
- Biological Science
- Ecology