Hi Naima,
the datasets are in the shape of emp2:

nrows = 22014 #max no. of asvs
ncols = 2000  #no. of samples

all you need to do is append emp2$genus as a column to the matrix.

The matrices are presence/absence datasets (0 or 1s), so if you sum down a column you get the number of ASVs in the sample.

paData1 is generated with a Poisson distribution (lambda = 0.01 for the whole matrix). This always generates a negative sloped dataset.
paData2 is generated using your site structure, i.e. each site has its own distribution. 
So for every site (e.g. the first site is the first 81 samples) I chose a lambda from 0 to 0.01 and used it for those samples.
Site 2 had 128 samples and they were all generated from the same distribution, and so on. This always generates a positive sloped dataset.
I then applied DBD or EC to these two baseline datasets.

The matrices with DBD or EC are 100%, meaning all elements were affected by DBD or EC (i.e. the effect is very strong).