similarity filtering seems to select "clones" of existing species

[amynang]

Hey,
similarity filtering behaves strangely, selecting groups of species that are really similar but dissimilar from other groups. I suspect the reason is that the algorithm is trying to minimize the sum of similarity not the average?

example

[amynang]

I think it is (at least partly) an artifact of the fact that, with an allometric model for interactions, each species eats everything that falls within a range (which is not realistic). If you add

N <- 5e+05 # the number of random values to replace
inds <- round ( runif(N, 1, length(L)) ) # draw random values from [1, length(L)]
L[inds] <- 0 # use the random values as indicies to L, for which to replace

after line 18 in the example, average of non-zero similarities is lower after filtering.