Figure 1
Nonlinear response of differently abundant bacterial communities to deoxygenation. The entire 16S rRNA gene sequence data was divided to 22 subsets of decreasing OTU abundance (Refer the chapter Categories of abundant, moderate and rare OTUs). Each subset thus contained a part of the total OTUs of all bacterial communities (samples). The percentage of correctly clustered communities (y-axis) describes how well bacterial communities in each subset clustered based on the duration of deoxygenation (0, 3, 7, and 48 days) in the discriminant analysis, based on Bray- Curtis dissimilarity. The higher the percentage of correctly clustered communities, the better the bacterial communities clustered based on the duration of deoxygenation, i.e. the more sensitive they were to hypoxia.
