Objectives The application of nitrogen fertilizer is one of the key measures to increase seed yield of Gramineous forage. In this study, a field experiment was conducted in artificial Leymus chinensis (Trin.) Tzvel grassland to understand the effects of nitrogen fertilizer on seed yield and its components, and to provide theoretical and practical basis for management of artificial Leymus chinensis (Trin.) Tzvel grassland.

Methods The experiment was conducted from 2013 to 2015 with Leymus chinensis (Trin.) Tzvel ‘Zhongke No.2’ as materials. The experiment included five nitrogen level treatments (N 0, 60, 120, 180, 240 kg/hm²) with three randomized blocks. In each treatment, half of N fertilizer was applied at green stage (late March) and half at harvest stage (late July). Seed yield and yield components were measured at harvest stage.

Results The experimental years and N fertilizer had significant effects on seed yield. The yield increased gradually with experimental years. However, the seed yield was initially increased and then decreased in 2013 and 2014. In 2013, the optimal N rate was 104.9 kg/hm² with a highest seed yield of 395.2 kg/hm², but the yield was not significantly affected by N fertilizers. In 2014, the optimal N rate was 173.5 kg/hm² with a highest seed yield of 857.8 kg/hm², which was 56% higher than control treatment. In 2015, seed yield increased with N rate, with the highest yield of 1865 kg/hm² at N rate of 180 kg/hm², which was 206.6% higher than control treatment. The results of path analysis indicated that seed yield was significantly correlated with spike number ( r = 0.883, P < 0.01). Spike number had largest direct path coefficient to seed yield (0.717) among all the variables, while tiller number had the highest indirect effect on seed yield through spike number, indicating that spike number contributes most to seed yield. Spike number was not affected by N treatmetn in 2013, but had similar responding pattern to N treatment to seed yiled in 2014 and 2015. The spike number increased significantly in 2014 and 2015 by consecutive application of N fertilizer among the three years. Furthermore, the application of N fertilizer significantly increased spike length, florets/spikelet, grain number/spike and seed setting rate, but did not affected 1000-seed weight and spikelet/spike.

Conclusions The optimal amount of N fertilizer for seed yield was 104.9-180.0 kg/hm² and increased with planting year increasing (104.9 kg/hm² in 2013 and 180.0 kg/hm² in 2014 and 2015). The spike number was the key factor influencing yield, which was not affected by N fertilizer application of current year, but was influenced by the precipitation from August to October and the amount of N fertilizer after mature stage in the previous year. The negative direct coefficient for spike number to spike length, 1000-seed weight and florets/spikelet could be offset by adding N fertilizer.