Figure 2.

Microbial data are in the form of relative abundance, and some taxa are much more abundant than others, which are factors that may cause taxa to be spuriously correlated. Thus, we devised a null model to account for the bias that these data features introduce into our metrics. We repeated this process with each taxon as the ‘focal taxon’, which is A in this figure. For each of the 200 iterations, we randomized all taxon abundances besides the focal taxon. We then calculated correlations between the focal taxon and all other taxa. We recorded the median value of the 200 correlations calculated for each pair of taxa in the median correlation matrix.