That is exactly the question Vesna and I had after reading a few papers using FD indices as proxies of function. Other than a few papers on plants (mostly experimental and with biomass production as the function studied) we couldn’t find the answer. Then is when we started a side project to gather the data necessary to answer our question. The problem is that there are many FD indexes (see previous post here) and many functions to test. So after realising that this was not a side project anymore we gathered data on as many functions as we can find (a.k.a. calling past collaborators and assaulting ecologists for data in the darker corridors of the conferences we attended) and developed a serious analysis.
The answer? Well, I wouldn’t use FD metrics to predict a single function (i.e. pollination to a single plant species or predation of a single aphid species) because their predictive power are usually below 0.5. However, it worth nothing that non-experimental single functions are usually better described by a few traits than by classic taxonomic diversity indices. It seems it depends on the function studied, but functional identity and to some degree weighted functional diversity metrics can improve our mechanistic understanding of the BEF effects, and this is pretty cool.
Plus, this is the first paper with two lead authors and lots of GitHub commits I did, and it worked really well!
Enjoy the paper here.