Transplanting rice varieties grown in different seasons can lead to different yields due to different dry matter production. Early-season rice varieties transplanted in the late season can obtain high yields with short-growth duration and higher yields driven by higher dry matter production. To make clear the variations in dry matter production across seasons, four early-season rice varieties were chosen for late-season transplantation. The grain yield, dry matter accumulation, leaf photosynthetic, and leaf stomatal properties were studied. It was observed that the average yields of these four varieties in the late season were 33% greater, despite a reduced growth period of 13 days in comparison with the early season. Furthermore, there was a notable increase in both total and post-heading dry matter production during the late season. The leaf net photosynthetic rate, stomatal area, stomatal width, and stomatal length were higher in the late season. Despite no significant difference in stomatal density between seasons, strong positive linear relationships were observed between net photosynthetic rate and stomatal conductance, and between stomatal conductance and area. These relationships demonstrate that the increase of the stomatal width and length of the leaves in the late season leads to an increase in the stomatal area, thereby increasing the stomatal conductance and enhancing the photosynthesis of the leaves. Consequently, this leads to greater dry matter production and a higher yield compared to the early season. Therefore, when breeding new high-yielding and short-growing varieties, the large stomatal area can be used as a reference index.

Rice yield stability is a breeding goal, particularly for short-growth duration rice, but its underlying mechanisms remain unclear. In an attempt to identify the relationship between yield stability and source–sink characteristics in short-growth duration rice, a field experiment was conducted at three sites (Yueyang, Liuyang, and Hengyang) in 2021 and 2022. This study compared yield, yield components, source–sink characteristics, and their stability between two stable-yielding short-growth duration rice cultivars, Zhongzao 39 (Z-39) and Lingliangyou 268 (L-268), and two unstable-yielding short-growth duration rice cultivars, Zhongjiazao 17 (Z-17) and Zhuliangyou 819 (Z-819). The stability of agronomic parameters was represented by the coefficient of variation (CV). The respective CVs of yield in Z-17, Z-819, Z-39, and L-268 were 10.2%, 10.1%, 4.5%, and 5.7% in 2021 and 19.7%, 15.0%, 5.4%, and 6.5% in 2022. The respective CVs of grain weight were 6.3%, 5.7%, 3.4%, and 4.5% in Z-17, Z-819, Z-39, and L-268 in 2021, and 8.1%, 6.3%, 1.5%, and 0.8% in 2022. The mean source capacity per spikelet and pre-heading non-structural carbohydrate reserves per spikelet (NSC_pre) were 7%–43% and 7%–72% lower in Z-819 and Z-17 than in L-268 and Z-39 in 2021 and 2022. The mean quantum yield of photosystem II photochemistry of leaf, leaf area index, and specific leaf weight of L-268 and Z-39 were higher than those of Z-819 and Z-17 at the heading stage. This study suggests that high NSC_pre, caused by great leaf traits before heading, increases source capacity per spikelet and its stability, thereby increasing the stability of grain weight and yield. Increasing NSC_pre is critical for achieving grain weight and yield stability in short-growth duration rice.

Zero-tillage has become increasingly attractive in rice production in China. This study was conducted to determine the feasibility of two possible improved N management practices with fewer N applications in zero-tillage rice: (1) two split applications of urea at 75kgNha⁻¹ at mid-tillering and 45kgNha⁻¹ at panicle initiation (U_120–2), and (2) a single application of cross-linked polyacrylamide-coated urea (a slow-release fertilizer) at mid-tillering at a rate of 150kgNha⁻¹ (PCU_150–1). Three field experiments were conducted to compare grain yield and N-use efficiency among several N treatments: a zero-N control (CK), U_120–2, PCU_150–1, a single application of urea at mid-tillering at a rate of 150kgNha⁻¹ (U_150–1), and a commonly recommended N management practice for conventional tillage rice (three split applications of urea with 75kgNha⁻¹ as basal, 30kgNha⁻¹ at mid-tillering, and 45kgNha⁻¹ at panicle initiation) (U_150–3). Treatments with N application (U_120–2, PCU_150–1, U_150–1, and U_150–3) produced 1.08–3.16tha⁻¹ higher grain yields than CK. Grain yields under both U_120–2 and PCU_150–1 were comparable to that in U_150–3. Recovery efficiency of N (RE_N), agronomic N-use efficiency (AE_N) and partial factor productivity of applied N (PFP_N) were increased under U_120–2 and were similar under PCU_150–1 to those under U_150–3. U_150–1 showed lower grain yield, RE_N, AE_N, and PFP_N than U_150–3. These results suggest that U_150–3 can be replaced with U_120–2 to achieve both an increase in N-use efficiency and a reduction in number of N applications and or by PCU_150–1 to achieve a maximum reduction in number of N applications in zero-tillage rice production in China.