From metabolomes to phenotypes: predictive metabolomics across scales.

Abstract

Metabolomics has transformed plant biology by delivering high-resolution insights into the biochemical networks that govern growth, adaptation, and environmental responsiveness. The plant metabolome, encompassing both primary and specialised metabolites, acts as a dynamic interface between genotype and phenotype, capturing the interplay of intrinsic genetic regulation with extrinsic environmental cues. Significant technological progress has enabled increasingly comprehensive metabolic profiling, accelerating the discovery of novel natural products, the mapping of metabolic pathways, and the mechanistic dissection of stress responses. Beyond individual organisms, metabolomics is now being deployed at ecosystem scales: ecometabolomics and meta-metabolomics reveal the metabolic signatures of plant-microbe interactions, soil processes, and environmental gradients. Coupled with machine learning, predictive metabolomics further expands this scope by linking metabolic fingerprints to complex phenotypes such as yield and stress resistance, offering actionable opportunities for crop improvement and agroecosystem management. Nonetheless, analytical costs, sampling constraints, and limited data interoperability continue to restrict broader implementation. This review examines the opportunities and limitations of metabolomics across biological scales, from subcellular compartments to agroecosystems, and highlights its potential to connect molecular processes with ecosystem-level outcomes. Integrating metabolomics with imaging, volatilomics, and soil metabolomics will be essential for developing more sustainable and resilient agricultural systems under accelerating global change.