Differences in MOTU richness among treatments
Generalized Linear Mixed Models allowed identifying shifts in the richness of observed MOTUs likely due to unproportional growth of different taxonomic groups or to differential DNA degradation under different preservation conditions of each sample.
When we considered all the detected MOTUs, GLMM detected significant differences in MOTUs richness among treatments for all the markers considered (Bact02: χ₄ = 38.9, P < 0.001; Fung02: χ₄ = 18.2, P = 0.001; Euka02: χ₄ = 21.7,P < 0.001; Fig. 2). Compared to the control, contrasts showed small but significant changes in MOTUs richness under treatment 3 (Bact02: z = 2.54, P = 0.010; Fung02: z = -2.17, P = 0.029; Euka02: z = 2.65, P = 0.008), treatment 4 (Bact02: z = -2.93, P = 0.003; Fung02: z = -3.99, P < 0.001; Euka02: z = 3.92, P < 0.001), and treatment 5 (Bact02: z = -3.74; Fung02: z = -4.02; Euka02: z = 4.18; all P < 0.001). Treatment 2 caused a small but significant decrease in MOTUs richness for fungi (z = -2.42; P = 0.015), but not for bacteria and eukaryotes (P = 0.456, P = 0.283, respectively; for all contrasts: Table S1).
Nevertheless, when we repeated analyses by excluding rare MOTUs (i.e. with a frequency <1%), differences in richness were much smaller, and were significant only for bacteria and fungi (Bact02:χ₄ = 9.69, P = 0.045; Fung02: χ₄ =14.1, P = 0.006; Euka02: χ₄ =2.22, P = 0.693; Fig. 2). Compared to the control, MOTUs richness decreases for Bact02 under treatment 3 (z = -2.91; P = 0.003) and increases for Fung02 under treatments 4 and 5 (z = 2.77; P = 0.005; z = 1.75; P = 0.080; respectively), while no significant effect was detected for Euka02 under any of the treatments (all P > 0.170; for all contrasts: Table S1).
Habitat caused a significant effect in MOTUs richness only for Fung02 either before and after removing rare MOTUs (before: χ₁ = 11.8,P < 0.001; after: χ₁ = 20.5, P< 0.001).