Metabolomic analysis of a core collection of Brassica rapa and Brassica oleracea unveils unexpected chemical diversity with potential applications in chemical ecology and breeding

Specialized metabolites play key roles in ecological interactions and stress responses, yet their diversification remains poorly understood in many crop species. In the Brassica genus, most metabolomic studies have focused on a limited number of compound classes, thereby underestimating the breadth of chemical variation and its evolutionary significance. Here, we use untargeted metabolomics combined with genomic resources to explore how metabolic diversity, enlarged through CuCl2 inductive treatments, can be systematically analysed to uncover novel biochemical features and generate testable hypotheses on biochemical innovation across a diverse panel of divergent Brassica species, as well as within species. Using a diverse panel of 10 B. oleracea and 10 B. rapa accessions, we constructed a curated metabolomic dataset integrating root and shoot metabolomes from plants grown under optimal conditions and exposed to CuCl₂-induced stress responses, followed by manual compound annotation. Available genomic data were used to support mechanistic interpretations of the observed variation among accessions. This approach enabled the exploration of a wide array of Brassica metabolites, notably by incorporating underexplored classes such as megastigmanes, phenolamides, and tetra/pentahexosylated acylated flavonoids. We uncovered pronounced inter- and intraspecific metabolic signatures, revealing distinct evolutionary trajectories between the two species. Strikingly, we detected distinct classes of blumenol derivatives across Brassica species, a plant genus considered non-mycorrhizal. This finding extends the known occurrence of these compounds and raises new questions about their biological roles. In addition, we link variation in glucosinolate chain-length chemotypes to specific mutations and structural variants in MAM genes, illustrating how metabolomic patterns can guide mechanistic genomic investigations. Together, these results show how a curated metabolomic dataset can simultaneously serve as a reference framework and as a driver for hypothesis-oriented research. By connecting metabolic variation to genomic features, our study provides a basis for functional investigations and offers new opportunities to exploit specialized metabolic traits in breeding programs aimed at improving stress resilience and ecological performance in Brassica crops.