Site description and experimental design
The study area is a typical alpine steppe, which is situated in the northeastern Qinghai - Tibetan Plateau (QTP), China (37°18′N, 100°15′E). The study site exhibits a plateau continental climate. The average annual temperature is ~0.1°C. The mean annual precipitation is ~390 mm, most of which occurs from June to August. The vegetation is mainly dominated by grasses, such as Stipa purpureaGrisebach, Leymus secalinus (Georgi) Tzvel, and Poa crymophila Keng (Xiao et al., 2020). The growing season is from May to October, and peaks in August. At the beginning of the experiments, the chemical properties of the soil (0–30 cm depth) were as follows: total N, 2.5 g kg−1; NO3-N, 11.5 mg kg−1; NH4+-N, 5.1 mg kg−1. The topsoil was mainly composed of chestnut soil with a pH of 8.3.
The experimental platform was established in 2013 (Figure 1). The experiments consisted of six different treatments (NP: ambient nitrogen with ambient precipitation; NP−: ambient nitrogen with 50% reduced precipitation; NP+: ambient nitrogen with 50% increased precipitation; N+P: nitrogen addition with ambient precipitation; N+P−: nitrogen addition with 50% reduced precipitation; N+P+: nitrogen addition with 50% increased precipitation). These 30 plots (2.7 m × 3.3 m each) were randomly established in a 5×6 block design and were each separated by a buffer zone (2 m wide). Concave sunlight-pervious polyvinyl chloride (PVC) boards without slots were placed above the 50% reduced precipitation treatments to intercept rainfall. The collected water (50% ambient precipitation) was immediately transferred to the 50% increased precipitation treatments after the rain (sprinkling evenly). To avoid light differences between the treatment plots, the same PVC boards with slots were also installed on the ambient precipitation and 50% precipitation increment treatment plots. N fertilizer (NH4NO3: 10 g m−2yr−1) was dissolved in 1 L water and was evenly applied two times to the N supply plots (June and July every year). Identical amounts of water without N fertilizer were sprayed evenly on the ambient treatment plots. A previous study indicated that the N saturation level was 8 g m−2 yr−1 in this study area (Peng et al., 2017). Therefore, the current N input level (10 g m−2 yr−1) should be sufficient to simulate N saturation of the alpine grasslands.