1 - ((s – min(s)) / (max(s)- min(s)))
Where “s ” corresponds to the suitability values obtained from the raster projections, and “min(s) ” and “max(s) ” correspond to the minimum and maximum suitability values from the model, respectively. Then, we calculated the kernel density distribution of marginality within the suitable conditions with the “density” function in R, which represents the relative amount of area of different marginality values within suitable conditions. The calculation of the density distribution of marginality was done for the present-day and LGM distribution models. Then, we calculated the change in the marginality distribution between LGM and present by subtracting the kernel density curves from both time periods. Finally, to identify clusters of taxa that showed similar changes in their marginality density distribution between LGM and present, the density curves were hierarchically clustered with the “dad” package (Boumaza et al. 2021) in R by calculating the Hellinger’s distance between density curve pairs.