Extremely high temperatures (HT) caused by global warming pose serious threats to rice production. Potassium (K) is critical for plant stress tolerance, but its role in mitigating heat damage remains unclear. This study aimed to elucidate how high panicle K application affects mid-season rice HT tolerance in central China. A two-year field experiment grew two rice cultivars (heat-resistant Shanyou 63, SY63; heat-sensitive Liangyoupeijiu, LYPJ) under varying sowing dates and two K application levels (low K, LK, 50 kg K ha −1 ; high K, HK, 90 kg K ha −1 ) at the panicle initiation stage. Sowing date 1 (S1) and sowing date 2 (S2) increased the risk of heat stress exposure. Compared with late sowing (S3) under LK, early sowing reduced the yield in LYPJ by 41.3% (S1) and 51.3% (S2) in 2022, and by 35.4% (S2) in 2023, but did not affect the yield in SY63. Compared with LK in the same sowing date, HK increased yield by 44.7% (S1) and 61.5% (S2) in LYPJ in 2022, and by 30.6% (S2) in 2023, whereas it showed no significant effect on SY63 yield. Structural equation modeling analysis indicated that the yield loss could be primarily attributed to heat intensity at the panicle initiation and maturity stages. HK increased stomatal conductance and improved leaf water potential, thereby reducing canopy temperature by 1.2–1.3 °C at heading and 1.1–2.5 °C at maturity. Concurrently, HK enhanced carbohydrate supply and elevated enzyme activity for sugars utilization in anthers, collectively enhancing pollen viability and spikelet fertility. HK optimized source-sink traits via increasing leaf area index, specific leaf weight, spikelets per unit leaf area, post-anthesis translocation of stem dry matter (47.5%–48.9% in 2022 and 24.0% in 2023), and post-anthesis dry matter accumulation (33.0%–38.2% in 2022 and 19.0% in 2023). The study indicates that early sowing increases the risk of heat stress exposure for mid-season rice in central China, and the increase of panicle K application can mitigate yield loss by lessening canopy temperature and optimizing source-sink relationships.
Xie et al. (Wed,) studied this question.