Can we accurately predict the distribution of soil microorganism presence and relative abundance?

Soil microbes play a key role in shaping terrestrial ecosystems. It is therefore essential to understand what drives their distributions. While multivariate analyses have been used to characterise microbial communities and drivers of their spatial patterns, few studies focused on modelling the distribution of Operational Taxonomic Units (OTUs). Here, we evaluate the potential of species distribution models (SDMs), to predict the presence-absence and relative abundance distribution of bacteria, archaea, fungi and protist OTUs from the Swiss Alps. Advanced automated selection of abiotic covariates was used to circumvent the lack of knowledge on the ecology of each OTU. ‘Presence-absence’ SDMs were successfully applied to most OTUs, yielding better predictions than null models. ‘Relative-abundance’ SDMs were less successful, yet, they were able to correctly rank sites according to their relative abundance values. Archaea and bacteria SDMs displayed better predictive power than fungi and protist ones, indicating a closer link of the latter with the abiotic covariates used. Microorganism distributions were mostly related to edaphic covariates. In particular, pH was the most selected covariate across models. The study shows the potential of using SDM frameworks to predict the distribution of OTUs obtained from environmental DNA (eDNA) data. It underscores the importance of edaphic covariates and the need for further development of precise edaphic mapping and scenario modelling to enhance prediction of microorganism distributions in the future.