Bodenhausen et al. 2023; Predicting fungal communities from soil properties. undefined

Inoculating soils with microbes presents an alternative to chemical fertilizers to increase agricultural sustainability. However, inoculation success varies from field to field and depends on the local microbial community. The local soil microbial communities typically remain unknown while the physico-chemical data of agricultural fields are often available or easy to obtain. In this study we specifically investigated whether it is possible to predict the composition of soil fungal communities based on physico-chemical soil data. We sampled 59 fields used for cereal production and assembled paired data of physico-chemical soil properties as well as profiles of soil fungal communities. Fungal communities were characterized using SMRT sequencing of the entire ribosomal internal transcribed spacer. We used redundancy analysis (RDA) to combine the physical and chemical soil measurements with the fungal community data and identified a set of 10 soil properties that explained fungal community composition. Soil properties with strongest impact on fungi included pH, potassium and sand. Finally, we evaluated the RDA model using leave-one-out validation. Prediction of community composition was successful for most soils, with Pearson correlation coefficients between observed and predicted communities >0.7 for more than half of the fields. We showed that prediction strength was negatively related to community evenness. Future research is now needed to test whether predictions of local soil microbial communities can increase the reliability of field inoculations with microbes.