The effect of habitats
Habitat type was the strongest factor (p < 0.002) in explaining community composition in the PERMANOVA analysis (18S – R2 = 0.12, COI – R2 = 0.18, and ITS – R2 = 0.08), with the exception of the ITS dataset. In NMDS, the seasonally flooded forests, igapós, and várzeas were more similar to each other than to campinas and terra-firmes, which were the most similar to each other (Fig. 5). Campinas had the highest mean number of OTUs and the highest number of OTUs considered indicators of this habitat for all datasets (Table 3). Regarding the number of exclusive OTUs, campinas had the highest number of OTUs in the COI datasets (Fig. 5E). Terra-firme was the habitat with the highest number of exclusive OTUs for the 18S and ITS datasets (Figs. 5B and 5H). In all habitats, the majority of indicator OTUs were saprotrophs, followed by a high proportion of OTUs that could not be classified by their functional group (Table S2). Campinas have a moderate proportion of phytopathogen indicator OTUs and terra-firme a moderate proportion of parasite indicator OTUs (Table S2).

Discussion

Our results highlight the importance of habitat type for fungal community composition in Amazonia and suggest that Amazonian fungi have different diversity patterns for habitat and locality variables, with the importance of each predictor varying between markers. By contrast, community turnover shows a consistent pattern, with habitat being a strong factor explaining community similarity between plots. This is likely to be because different areas can have similar species richness but different species composition due to historical, geographic, and environmental factors. For instance, in a study of leaf litter fungi in Central Amazonia, the abundance and richness of fungal morphospecies did not change between low and high rainfall periods, but there was a low proportion of shared morphospecies between periods56. Our results also showed a low proportion of shared OTUs when compared with a HTS study of micro-organisms in general in the same area10 (Fig. 5).
Soil texture did not explain fungal diversity, while chemical soil characteristics was of importance for COI and ITS soil communities, indicating a high diversity in less fertile soil (Table 1). Although it appears counter-intuitive, the habitat with lowest soil fertility was the one with highest fungal and other microbial diversity: the campinas10, 25. These results suggest that factors other than soil properties explain a habitat’s fungal diversity and community composition.
The soil diversity of the 18S dataset was negatively related with carbon, while the specifics of the other datasets were not related to carbon. This could be explained by taxonomic coverage of the 18S dataset, which included the Chytridiomycota and Mucoromycota, which are mostly saprobe groups57, 58. Saprobes decompose matter into various constituent components, making the nutrients available to other organisms. Saprobes are, in other words, important agents in carbon cycling59. Hence, a high fungal richness may lead to a faster carbon decomposition in soil as well as a faster carbon assimilation in the above-ground biomass. This is in agreement with Liu et al.60, who found that phylotype richness and phylogenetic diversity of black soil fungi responded negatively to total carbon content in China. Experiments controlling the variables and quantifying the above-ground biomass are necessary to further verify these observations.
Contrary to our expectations, pH had no effect on fungal richness. This finding was surprising, since soils with more neutral pH generally have a higher richness of micro-organisms11,61,62,63. Our soil samples were all acidic, with the pH varying between 3.5 and 5.14. Soil fungi studied by Liu et al.60 showed a similar pattern as reported in this study – a higher relative influence of soil carbon content than of soil pH. They also noted that fungi often have a wider tolerance to pH variation than other micro-organisms, suggesting that in soils with low pH variation such as presented here, the acidity impact should be less striking60. On the other hand, pH was important to explain community turnover for all datasets (Table 2). Furthermore, in tropical areas the relationship between fungal communities and soil pH is affected by the fungal trophic guilds64. It may indicate that in a highly diverse area, such as Amazonia, fungal diversity will not be impacted by pH variation but there will be a turnover of fungal species related with the pH range.