Untargeted Metabolomic Analysis Reveals Three Critical Metabolites as Potential Biomarkers Related to Soybean Yield Under Different Sowing Times and Planting Densities in Maize‐Soybean Strip Intercropping Systems

The precise adjustment of soybean sowing dates and planting densities effectively alleviates shading stress in maize-soybean strip intercropping systems. Soybean exhibited significant morphological plasticity in traits such as main stem nodes, effective branches, and plant height across treatments, while the bottom pod height remained invariant, ensuring compatibility with mechanised harvesting. An optimised sowing-density configuration (low density "a" at the second sowing date "S2") improved pod-setting characteristics, reduced the incidence of shrunken pods, and achieved peak performance metrics (RYG = 54.38%; SPI = 130.87%). Integrated metabolomic profiling, combined with Mantel tests, weighted gene co-expression network analysis (WGCNA), and Random Forest algorithms, identified three yield-associated metabolites: phthalic acid (PA), vanillic acid (VA), and physcion 8-glucoside. Functionally, PA and VA activate complementary stress resilience pathways. PA reprograms metabolism and root physiology to optimise resource acquisition and stress resilience, while VA enhances branching architecture and stem strength to confer lodging resistance. These biomarkers, quantified in seeds at the yield-determining R6 stage, serve as molecular indicators of active physiological processes governing yield formation. They reflect the plant integrated stress response and resource allocation strategy, providing mechanistic insights and potential targets for metabolic engineering to optimise cereal-legume intercropping optimisation.