Differences between the obtained communities
Redundancy analysis allowed us to measure the amount of variation explained by differences among habitats and by treatments. Overall, 33%, 24%, and 33% of variability were explained by differences in habitat for bacteria, fungi, and eukaryotes, respectively. The community differences among habitats were strongly significant for the three taxonomic groups (permutation test: all P ≤ 0.001). Differences among treatments were much weaker, and explained 9%, 2% and 2% of variation only for bacteria, fungi and eukaryotes, respectively. Nevertheless, differences were significant for bacteria (permutation test: P < 0.0001), but not for fungi and eukaryotes (both P = 1).
For bacteria, contrasts did not detect significant differences between “control” and treatment 2 or 3. Differences between “control” and treatment 4 and 5 were significant but explained a limited amount of variation (for both treatments, ≈3% of variation explained; P< 0.0001; Table 1). We thus used similarity percentage analysis to identify the MOTUs significantly contributing to these differences. Only one MOTU showed a significant contribution (P = 0.03 after FDR correction) to the differences between “control” and treatment 4; this MOTU showed a very limited frequency under treatment 4 (Fig. S1). After FDR correction, no MOTU showed a significant contribution to the differences between “control” and treatment 5. The MOTU highlighted by the similarity percentage analysis for treatment 4 also showed a very limited frequency under treatment 5, but these differences were not significant at the 5% threshold (corrected significance after FDR correction = 0.078).