Exploring the potential of soil metabolic and microbial composition in predicting ecosystem functions across biomes and land use types.

Soils support a wide range of ecosystem functions and services, including climate regulation, nutrient cycling and carbon sequestration. Most of these functions are strongly impacted by a large diversity of microorganisms hosted in soil (e.g., bacteria, fungi) which are increasingly threatened by human-induced global change factors such as climate warming or land-use change. A deep understanding of how microbial communities function is thus crucial to evaluate how they influence ecosystem services but also how anthropogenic perturbation may affects soil quality and the delivery of these services. While great efforts have been made to evaluate the relationships between microbial diversity and ecosystem functions, much less attention has been paid to the metabolomic profiling of soil microbial communities. However, recent advances in mass spectrometry and big data processing now allow us to measure hundreds of known and unknown metabolite features constituting the soil metabolome, which can mirror the key biological processes occurring below-ground, and present an important opportunity to better understand the microbial characteristics and metabolic pathways driving soil ecosystem functions.In this study, soil metabolic profiles and microbial communities were explored on 25 European soils from different biomes and land use types alongside soil physical and chemical measurements in order 1) to characterise soil metabolomes across a large range of soil types, 2) to investigate the links between soil microbial communities and associated metabolic profiles, and 3) to evaluate the potential of soil metabolomics to predict ecosystem functions such as soil gas exchange.Soil metabolic profiles were screened using UHPLC-LTQ-Orbitrap mass spectrometry (LC-MS) and showed a strong gradient across sites alongside bacterial and fungal community shifts characterised using metabarcoding. The ability of soil metabolic profiles and microbial communities to predict soil ecosystem functions was evaluated through machine learning models across biomes and the interconnection of a core set of metabolic features and microbial genus was further investigated to deepen our understanding of the potential mechanisms and microbial communities involved.