Soil fungal communities affect key aspects of ecosystem functioning, such as nutrient cycling and plant-plant interactions, and are thought to be affected by both associated tree communities and soil properties. Co-occurring functional groups of fungi (hereafter ‘fungal guilds’) play distinct ecological roles and are thus likely influenced in different ways by both biotic and abiotic factors. The relative importance of the biotic and abiotic drivers on community structure and diversity of co-occurring fungal guilds in lowland tropical forests remains unclear, since previous work has treated soil fungi as one homogenous functional group, ignoring key functional differences.
We analyzed soil properties and sequenced the ITS1 barcode region using the Illumina MiSeq platform to characterize fungal community composition and diversity in 70 soil cores collected across three 1-ha tropical rainforest sites in Panama. We also calculated composition, diversity, and abundance metrics for local tree neighborhoods surrounding each core. We predicted that the composition and species richness of fungal guilds that acquire C and nutrients from living plant hosts would be more strongly affected by tree variables, while the composition and richness of guilds that acquire resources from soil and non-living substrates would be more strongly associated with soil properties.
Fungal guilds showed strong differences in the strength of correlation with tree communities and soil properties. Plant pathogens were two times more strongly correlated with soil properties than with tree community composition (P ≤ 0.001). For saprophytes, the correlation with soil properties was only 25% higher than the correlation with tree composition (P ≤ 0.001). Contrastingly, ectomycorrhizal (ECM) fungal communities showed a correlation with tree communities that was three times stronger than with soil properties (P ≤ 0.002). Finally, the correlation between arbuscular mycorrhizal (AM) fungal communities and soil properties was 0.32 (P ≤ 0.001) and there was no correlation with tree composition (P = 0.7). Further, soil fertility was the best predictor of fungal species richness for all clades except for ECM fungi, which was best predicted by local tree basal area, and for AM fungi, which was not significantly explained by any single variable. Our results show how soil fungi from distinct functional guilds respond differently to the same environmental factors, depending on their life history traits. We surmise that soil properties act as an environmental filter for both trees and fungi, setting the stage for interactions between the two.