Environmental variables
Precipitation, air temperature, and soil water content data for 2015-2019, years for which we have data of both grass and shrub ANPP, are shown in Fig. 1. The mean daily air temperature for 2015-2019 was 9°C, and the maximum and minimum temperatures were about 25°C (December 2018) and -9°C (June 2017), respectively (Fig.1A). Annual precipitation for the period of 2015-2019 ranged from 92 to 233 mm, with 2019 being the driest and 2017 the wettest year. Between March 29 and April 02, 2017, an extreme precipitation event of 100 mm was recorded (Fig. 1A). Precipitation in 2013 and 2014, years for which information of grass productivity is available, was 152.5 and 99.5 mm, respectively. Soil volumetric water content showed a marked seasonal pattern in the top soil layer (10 cm), ranging from 5% in summer to 15-20% in winter and early spring (Fig. 1B). Water addition events were identifiable at that depth, with transient increases in soil water content. At 40 cm depth, soil water content had a similar pattern to that of the upper layer, but water addition events were not visible at that depth (Fig. 1C). The extreme precipitation event in autumn 2017 resulted in a higher increase in water content throughout the soil profile compared to other years, but there was a delay in such increase in deeper soil layers. At 100 and 200 cm soil depth, water content was relatively constant, between 6 and 9% in all treatments except in 2017 (Fig. 1D, 1E). A soil water retention curve indicated that soil water potential at 40 cm depth was near 0 MPa when moisture was higher than 12% and was higher than -1 MPa when moisture was between 9 and 12% (Fig. S3).
Inorganic N and available P content were higher in the +N and +NW treatments than in the control (Table 2). The addition of nitrogen alone increased the inorganic N content about three times more than the co-addition of nutrients and water (p <0.001). Available P content increased about three times in +N and +NW treatments compared with the control (p <0.05). Total N and C and N/C ratio were not significantly different between treatments (Table 2). Soil pH was lower in the +N and +NW treatments than in the control, decreasing from 7.15 in the control to 6.40 and 6.48 in those treatments, respectively (Table 2).