2.1 Study site
A field-manipulative warming experiment was performed at the Haibei Grassland Ecological Monitoring Station (Xihai Town, Qinghai province, 100°51´E, 36°57´N, 3,140 m a.s.l.). The study site has been used as a winter grazing grassland since 1976, with moderate grazing intensity during the non-growing season. Based on meteorological records from 1995 to 2013, the mean annual precipitation is 408 mm, and the mean annual temperature is 1.3 oC (Chen et al. 2016). Relatively high air temperature and rainfall occur from mid-April to mid-October (growing season), while low rainfall (< 5% of annual) and low temperatures occur during the non-growing season. Additional long-term, detailed information about the study site can be found in (Chen et al. 2015b; Guo et al. 2018; Li et al. 2019) (Fig. S1). The dominant species are Stipa krylovii, Poa crymophila, Koeleria cristata, Medicago ruthenica, and Kobresia humilis (Fig. S1 and Table S1).
2.2 Experimental design
The study site (200 m × 400 m) was fenced off for three years before the initiation of experimental warming to exclude disturbance by herbivores. All mammalian herbivores were completely removed for the whole experiment duration. In August 2010, the entire study site was divided into six blocks with 10-m buffer zones between each edge of adjacent blocks. Each block was then divided into two plots (5 m × 10 m), one of which was randomly selected for experimental warming. In each of the experimental warming plots, one OTC (with a base area of 2.1 m2, six OTCs in total) was installed to achieve a passively warmed environment (Fig. S1). Protocols for the OTCs used in this study are described in detail for previous studies at this site (Chen et al. 2017a; Chen et al. 2017b), as well as at other sites (Dorji et al. 2013; Baruah et al. 2017).