Forecasting the soil microbiome at a continental scale

Abstract Despite the critical role of soil microorganisms in ecosystem processes, it is unclear how well we can anticipate changes in the Earth's microbiome before they occur. To test this, we integrated continental-scale, standardized soil genomic and environmental surveys to develop the first forecasts of soil microorganisms across biomes and time. We find that the abundance of fungi and bacteria can be predicted with 50% and 91% accuracy, respectively, for up to a year ahead of time, by simulating seasonal cycles in taxon abundances. Forecasts perform best for microbes grouped at broad taxonomic ranks or by ecological groups, such as decomposers and plant symbionts, and the presence of plant-associated seasonal cycles increases the predictability of microbial groups. We highlight strengths and weaknesses in our capacity to predict the future of microorganisms on Earth and propose future directions to improve longer-term projections of soil organisms and the processes they drive.