SIMPER analysis was used to identify the contribution of
each OTU to similarity among different clusters [19] (Table 3).
In order to determine the distribution of OTU contributions,
the frequency value of each OTU contribution was plotted and
presented in terms of a threshold value (i.e. 0.5%) virtually
divided into two subgroups. This resulted in statistical support
by linear regression and confirmed the two different trends of
the curves (R2 ¼ 0.8917 and R2 ¼ 0.9191) (Fig. 3). The 17
OTUs belonging to the first curves were considered to be the
most important OTUs (i.e. single contributions between 0.9
and 0.5) with a cumulative contribution of 10.6%. The other
532 OTUs, ranging between 0.5% and 0.01%, had a cumulative
contribution of 89.4%. Analysis revealed that differences
observed among differing clusters could not be attributed to
the 17 most important OTUs, but were determined by the
bacterial community as a whole. Although the single contribution
of each rare OTUs was not high, the sum of all contributions
of rare representatives was likely to determine