Hey everybody! Does anyone have a clear idea about whether use or not use Wisconsin double standardization on fatty acids data? I'm doing ANOSIM analysis on fatty acids data and some values are way higher than others. As far as I read this may render some biases or even make fail ANOSIM in recognize differences if any. Also, that the Bray-Curtis dissimilarity distance sometimes needs the data to be transformed in this way in order to improve performance of calculations.
If anyone could help solving this doubts or suggest any literature will be of great help!
Cheers!
G
Dead Frederic,
Thank you very much for your comments and suggestion!
All the besst!
Gaston
Hello-
My collaborators and I are working on a means (manuscript) to describe what these sorts of transformations change when interpreting fatty acid data in an ecological context. At the moment, our analyses suggest that there is no real advantage to using the sequential standardization, known as Wisconsin-double, over simply standardizing by sample total.
We also find that when using non-parametric tests like ANOSIM, the differences caused by square-root or log(x+1) transformations are negligible. It is clear that these transformations place more emphasis on fatty acids with lower abundances, but that is not always something that is necessary nor desired.
My suggestion would be to analyze fatty acids as a proportion of the total mass of fatty acids of interest, including ordination using nMDS if testing with ANOSIM. You can then repeat the analysis but transforming the data prior to calculating your resemblance metric. Then you can compare the two nMDS plots and ANOSIM results to see if your general conclusions change.
Cheers!
Austin
- Badges
- Science method
- Similar topics
- Biological Science
- Ecology