​ Uncovering associations between DUS test traits and biochemical composition in safflower germplasm​​.

Safflower (Carthamus tinctorius L.), a globally valued medicinal and oilseed crop, exhibits geographically structured biochemical signatures critical for its nutraceutical quality. Our study reveals safflower nutrient blueprint through an integrated approach combining phylogeography, chemometrics, and machine learning. We identified: (1) Evidence suggestive of genetic bottlenecks​​ in Xinjiang germplasm driving biochemical homogenization; (2) ​​Geography-driven chemodiversity​​ where cationic mineral-amino acid complexes adapt accessions to regional soil geochemistry; (3) ​​Evolutionary tradeoffs​​ manifesting as systemic mineral-fatty acid antagonisms; and (4) ​​Machine learning-enabled trait prediction​​, with crude fiber content showing relatively higher predictability due to developmental hardwiring. We revealed that fiber deposition prioritizes morpho-developmental regulators, while calcium accumulation depends on amino acid-mediated transport. Our findings establish that geographical isolation conserves nutrient signatures through reduced gene flow, while metabolic constraints limit co-optimization of competing traits. Our work provides predictive frameworks for precision breeding of climate-resilient safflower with enhanced nutraceutical value.