First of all, great initiative peerj.

Q: much needed and a great improvisation of the CCA, RDAs with OMI in this context. Wondering why OMI is using mean as the decision instead of 95% quantile. This is assuming that one does a complete representation of the communities in their observation, however, if this is violated neither will mean work nor quantile. But if one has done a complete representative sampling, especially with plants where most times same species occurs largely in contrastingly high densities and low densities in the same area, using mean will reduce the differences between species rather than a 95% quantile. Making the inter-species differences smaller would result in the arrows in the OMI ordination plot much closer. Consider 3 species that occur in densities of 0-30, 0-60 and 0-90, respectively. Means of this - 15,30,45 and 95% quantile might be 28, 56 and 88, respectively. In the former case, arrows would span between 0 degree to 45 degree and in the latter 20 degree to 90 degree.

Apologies if its already been addressed elsewhere.