0

I have a data frame of phytocoenological relevés:

  • rows represent sites
  • columns represent species

I run RDA on it using vegan function rda(matrix ~ 1). Now I have the axes of the ordination summary(rda)$sites, but how to find out which of them are significant?

And my further question is, would these approaches differ if I use different ordination methods?

Thank you attention and anwers.

Kryštof Chytrý
  • 348
  • 1
  • 3
  • 15

1 Answers1

2

Personally I would say that there is no way. However, some people have suggested ways to find the number of non-significant axes. Some of these are discussed in a vegan github issue. My personal experience is that many of these ways work poorly: if I generate data with a defined number of dimensions + random error, methods fail to find that known number of axes. Many of these methods are easy-ish to implement (although we have not implemented them in vegan), though. See the discussion linked above.

Your title asks about PCoA, but your example uses RDA. However, similar methods apply to both (or fail with both like I think).

Jari Oksanen
  • 3,287
  • 1
  • 11
  • 15