Precision breeding for drought tolerance using a Copeland method: Integrating receiver operating characteristic analysis and multi-criteria decision making to identify superior Persian walnut genotypes

The scarcity of drought-tolerant rootstocks threatens the sustainability of walnut (Juglans regia L.) production, given the crop's high-water requirements and sensitivity to drought. Selecting a drought-tolerant genotype is a critical challenge due to the broad impact of drought on various traits, which requires evaluating multiple, unequally weighted criteria, making it a complex multi-criteria decision-making (MCDM) problem. This study investigates the genetic and phenotypic diversity of 183 walnut genotypes, originating from seven regions of Iran with diverse climatic conditions, which were subjected to severe drought stress (DS) in a common garden approach. DS was applied by water withholding for a 30-day period, and various phenotypic, physiological, and biochemical traits were assessed at the end of the period. The results revealed that drought-tolerant genotypes exhibited higher tolerance indices, with less reduction in RWC, MSI, and photosynthetic pigments, along with decreased wilting and leaf shedding. Moreover, these genotypes demonstrated significantly higher antioxidant enzyme activity, osmotic regulation, and photosynthetic efficiency compared to sensitive ones. The receiver operating characteristic (ROC) curve-plotting algorithm, based on the criteria of area under the curve (AUC) and Youden index (YI), highlighted the superior discrimination ability of specific indices between susceptible and tolerant genotypes and determined the optimal thresholds (OT) for the measured traits. Among the characteristics, FV/FM (AUC: 0.9965, YI: 0.9669, OT: 0.6365), PIABS (AUC: 0.9961, YI: 0.9556, OT: 0.89), TChl (AUC: 0.9915, YI: 0.9158, OT: 2527), carotenoids (AUC: 0.9892, YI: 0.9139, OT: 708), and RWC (AUC: 0.9739, YI: 0.9026, OT: 62.7 screening key indices, respectively. To identify superior genotypes, a proposed framework was employed, utilizing the ROC approach to weight the measured traits, followed by the application of four prominent MCDM methods, including VIKOR, TOPSIS, SAW, and ELECTRE, integrated with a Copeland ranking aggregation technique to achieve a consensus-based final ranking of genotypes. Based on the aggregated Copeland results, the genotypes Ke16, KeSH9, and Ke18 were identified as the superior drought-tolerant genotypes. It can be concluded that the application of MCDM methods and the Copeland aggregation approach provides breeders with a powerful tool to select superior genotypes for developing drought-tolerant cultivars with high precision and efficiency.