Breeding early-generation sorghum: integrating agronomic stress resilience, parent-progeny predictability, and the rhizosphere microbiome

Abstract

Selection strategies for sorghum [Sorghum bicolor (L.) Moench] adapted to low-input systems require understanding trait stability, stress response, and cross-generation predictability. This study evaluated F₂ and F₃ populations under conventional and low-input management to quantify trait responses, assess parent-progeny predictability, and identify resilient populations using a rank summation index (RSI). Twenty-nine F₂ populations were evaluated in summer 2024, and the selected eight were advanced to F₃ testing in 2025. Low-input management imposed stronger stress in summer 2025, resulting in 26 to 68% reductions in seeds per panicle and 11 to 61% in total seed weight. In contrast, 100-seed weight increased in eight of ten populations, indicating compensatory allocation. Parent-progeny regressions had the strongest cross-generation predictability for plant height (R² = 0.52 and 0.51 under conventional and low-input management, respectively), moderate predictability for 100-seed weight (R² = 0.32 and 0.44), and weak relationships for flowering time, seeds per panicle, and total seed weight. Rank summation index (RSI) analysis identified GB07118/BTx2921 (P1) and GR04104/RTx2939 (P8) as the most resilient populations, while Check 2 (Macia) had greater sensitivity to stress. These results support direct selection under low-input environments and multi-trait ranking to improve breeding for stress-resilient sorghum. Understanding rhizosphere microbial assembly in field conditions is critical for harnessing microbiome-associated traits in sorghum breeding, particularly given proposed links between arbuscular mycorrhizal fungi (AMF) and strigolactone signaling through the LOW GERMINATION STIMULANT 1 (LGS1) locus. A field study used phospholipid fatty acid analysis (PLFA) and neutral lipid fatty acid analysis (NLFA) profiling and root colonization assessment across V5 and reproductive stages in 131 sorghum (Sorghum bicolor) plants, spanning eight F2:3 breeding populations, two checks, and an LGS1-characterized genotype panel. Developmental stage was the dominant driver of rhizosphere community structure (PERMANOVA: R² = 0.063, p = 0.002), with the fungal-to-bacterial ratio and AMF NLFA abundance increasing from V5 to reproductive growth. Contrary to expectations under strigolactone-mediated AMF recruitment, lgs1 loss-of-function genotypes had higher AMF NLFA abundance than LGS1-functional genotypes, though this could not be isolated from genetic background. Arbuscular mycorrhizal fungi NLFA and root colonization percentage were uncorrelated overall but significantly negatively correlated within LGS1-functional genotypes alone (R = -0.57, p = 0.0021), indicating these were not interchangeable AMF metrics. The functional RTx430 NIL produced roughly three-fold greater grain yield and higher tissue potassium than its lgs1 counterpart (n = 2/genotype, exploratory). AMF colonization had weak, largely non-significant associations with plant performance. Developmental stage was the primary driver of sorghum rhizosphere assembly and genotype-dependent AMF-LGS1 relationships need confirmation in replicated trials.