dbRDA score scaling when using dissimilarity matrix or non-Euclidian distance

[ibrila]

Hi,

I'm trying to run dbRDA in vegan with binary Jaccard distance or unweighted UniFrac distance as response matrix.

Running summary(dbrda_jacc) shows:

Scaling 2 for species and site scores

• Species are scaled proportional to eigenvalues
• Sites are unscaled: weighted dispersion equal on all dimensions
• General scaling constant of scores: 9.826441

I want to extract the scores, to plot ordinations manually using ggplot2. When using scores(dbrda_jacc)$sites I noticed that the scores are indeed as when using scores(dbrda_jacc, scaling=2). I'm wondering if this is the correct scaling to use, when the response is a distance matrix? Or is it more appropriate to use a different type of scaling when extracting scores?

Thanks