Predictive metabolomics of multiple Atacama plant species unveils a core set of generic metabolites for extreme climate resilience

Summary Current crop yield of the best ideotypes is stagnating and threatened by climate change. In this scenario, understanding wild plant adaptations in extreme ecosystems offers an opportunity to learn about new mechanisms for resilience. Previous studies have shown species specificity for metabolites involved in plant adaptation to harsh environments. Here, we combined multispecies ecological metabolomics and machine learning‐based generalized linear model predictions to link the metabolome to the plant environment in a set of 24 species belonging to 14 families growing along an altitudinal gradient in the Atacama Desert. Thirty‐nine common compounds predicted the plant environment with 79% accuracy, thus establishing the plant metabolome as an excellent integrative predictor of environmental fluctuations. These metabolites were independent of the species and validated both statistically and biologically using an independent dataset from a different sampling year. Thereafter, using multiblock predictive regressions, metabolites were linked to climatic and edaphic stressors such as freezing temperature, water deficit and high solar irradiance. These findings indicate that plants from different evolutionary trajectories use a generic metabolic toolkit to face extreme environments. These core metabolites, also present in agronomic species, provide a unique metabolic goldmine for improving crop performances under abiotic pressure.