Values for ds and bin.size.limit

[adityabandla]

Hi Ning,

Thanks for the great package. I went through your articles & the example code, however, I couldn't find how one can find optimal values for ds and bin.size.limit. Appreciate any advice/thoughts

On a related note, can categorical variables be used for identifying niche preferences i.e. used in the environment.txt file? We sampled across depth and as such depth integrates multiple niche dimensions, many of which we did not measure explicitly

Cheers, Adi

[adityabandla]

Figured it out myself, great explanations in the supplementary docs

[DaliangNing]

ds can be determined by phylogenetic signal analysis as Stegen et al 2013 (Fig.2 in doi:10.1038/ismej.2013.93) and Wang et al 2013 (Fig.1 in doi:10.1038/ismej.2013.30). From the previous publications and my test with simulated data, ds=0.2 seems a generally acceptable choice.

bin.size.limit is more tricky. In our paper (doi:10.1038/s41467-020-18560-z), I suggested two ways to determine the bin.size.limit: one is based on phylogenetic signal within bin (Supplementary Fig.4 in the paper; Step 7 and 8 in the example code); the other is based on stochasticity estimated by NST (https://github.com/DaliangNing/NST), and you may try several values (e.g. 12, 24, 48) of bin.size.limit and use the one making the relative importance of stochastic processes (dispersal + drift) closer to NST.

Current niche preference calculation needs numeric environmental factors. A categorical factor is not applicable.

[adityabandla]

Thanks, Ning! much appreciated