Integrating marker-assisted identification and multi-environment trait stability models to select superior rice restorer lines for hybrid breeding

The efficient identification of stable and agronomically superior restorer lines is critical for hybrid rice breeding, particularly under multi-environment testing where genotype × environment interaction confounds the selection decision. The present study aimed to integrate marker-assisted fertility restoration screening for Rf3 and Rf4 and multi-environment-, multi-trait-based selection models to identify elite rice restorer lines suitable for hybrid breeding. A panel of 240 rice restorer lines was screened using molecular markers for Rf3 and Rf4 , which revealed that 85.41% lines carried the Rf4 allele, 11.25% lines possessed both Rf3 and Rf4 , and 3.34% lines carried only the Rf3 allele. Then, these restorers are evaluated in multi-location for grain yield and associated agronomic traits. The WAASBY-based ranking identified superior genotypes combining high yield and stability, and genotypes selected under a 10% selection intensity were also present in quadrant IV of the Y × WAASB biplot, indicating above-average yield and high stability. Coincidence analysis across all stability and multi-trait selection models identified eight restorer lines (RR130, RR121, RR72, RR140, RR196, RR79, RR23, and RR233) as common restorers. These restorers showed mid-early flowering duration, high spikelet fertility, favorable panicle exertion, and superior seed yield per plant. The integrated selection strategy adopted in this study provides a practical basis for identifying elite restorer (R-line) parents for the development of high-yielding and widely adapted hybrid rice cultivars. To our knowledge, no prior research has concurrently utilized MGIDI, MTSI, MTMPS, and FAI BLUP for the identification of rice restorer lines in multi-environment trials.