1
$\begingroup$

I want to run Redundancy Analysis (RDA) with the abundance of different functional groups of plants (7 groups in total: shrubs, coniferous, forbs, etc.) as my response variables.

I know that, for community data, the Hellinger transformation is a really popular choice. However, the Hellinger transformation is recommanded partly because it doesn't give high weight to rare species by accounting for the presence of numerous zeros in the data matrix.

In my case, I aggregated my species under 7 functional groups. So there is no more thing like a "rare species" or a huge number of zeros in the matrix (even if there are still some).

I am wondering if I still should do a Hellinger transformation on my functional groups' abundances.

Or should I use another transformation, or directly use the raw data into my RDA?

Many thanks

$\endgroup$
1
  • 1
    $\begingroup$ RDA is often used to decorrelate explanatory variables from each other that are fed into a regression to "non-redunandantly" predict the exogenous variables, but what is your use case here? $\endgroup$
    – Galen
    Jul 20, 2021 at 4:16

1 Answer 1

1
$\begingroup$

I am no expert, but I don't think you need to transform your data. This video by Pierre Legendre clearly explains why the Hellinger transformation is used. It should be used on long ecological gradients (matrix with multiple zeros). In your case, I think that grouping your plants has reduced the ecological gradient.

Here is the video: https://www.youtube.com/watch?v=PzbjxIke7zs . At 57:47 he gives a great rapid summary of RDA and CCA and the whole video is really great as well.

$\endgroup$

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.