Abstract: An experiment was conducted to assess whether large-scale rice-duck farming could improve the plant architecture of rice, which in turn influences the yield and yield components. Three treatments including large-scale rice-duck farming, normal rice-duck farming and direct sowing were established to investigate plant architecture at late tillering and fullheading stages and grain yield of rice. The result showed that, compared with the traditional rice-duck farming, largescale rice-duck farming decreased the actual grain yield and the number of effective ears. There were no significant differences among the various aspects of the plant architecture. However, compared with direct sowing, large-scale rice-duck farming could optimize the plant architecture. It decreased the stem basal width and increased the canopy range at the late tillering stage and fullheading stage, giving plants certain looseness. The opening angel, drooping angle and leaf length of the flag leaves and second leaves saw an increase and the third leaves and fourth leaves declined, making the upper leaves flatter and lower leaves more erect than the control treatment. The leaf areas of the flag leaf, second leaf and third leaf increased, enabling the upper part of the plant to contain more leaf area. Large-scale rice-duck farming enhanced the growth of rice roots, and increased the basal leaf height and SPAD value of different phyllotaxises, which could slow down the aging of leaves. Therefore, large-scale rice-duck farming optimized the rice plant architecture, which improved the photosynthetic efficiency of rice, increased dry matter accumulation, thus ensuring a high grain yield.

Key words: Olsen-P, long-term fertilization, critical Olsen-P value for crop yield., P balance