Climate & biomass & phenology & FG composition.csv

From 1980 to 2014, annual biomass production of the plant community was monitored using a harvesting method. For our study, we defined annual biomass production as the maximum aboveground biomass observed in August or September (Liu et al. 2018a). Annual biomass production was further separated into grass, forb, and sedge functional groups in the following years: 1980–1985, 1989, 1998–2000, and 2006–2014. Seasonal biomass dynamics of the plant community were monitored by clipping aboveground biomass once or twice each month from May to September during the periods of 1980–1985, 1989, 2002–2004, 2006–2010 and 2012–2014; the seasonal biomass dynamics of different plant functional groups were further monitored during 1980–1983 and 2007–2010. After harvesting, live plant samples were oven-dried at 65 °C until they reached a constant weight.

Two sampling methods were used from 1980 to 2014 to monitor plant biomass (Liu et al. 2018a). Before 2005, five to ten 50 × 50 cm quadrats were randomly clipped during each harvest within a permanent 250 × 230 m area. Starting in 2005, a new strategy of systematic sampling was adopted. An area of 150 × 150 m was divided into 25 permanent squares, and the five squares on the diagonal were chosen. Each chosen square was further divided into 25 blocks that were each 6 × 6 m. Five 25 × 25 cm replicates were randomly harvested from one of the 25 blocks in five chosen squares.
Seasonal biomass dynamics were simulated using linear, exponential, monomolecular, and logistic functions (Paine et al. 2012). We found that a three-parameter logistic function appropriately described the aboveground biomass dynamics across growing seasons (Fig. S2): AGB = L/(1+exp(-k*(x-x0)))
where AGB is the aboveground biomass and x is the Julian day. The parameters L, k, and x0 represent the annual maximum aboveground biomass, the intrinsic rate of plant growth, and the timing of maximum growth, respectively (Table S1). Fitted results from this method were validated with annual aboveground biomass production data (r2 = 0.84, P < 0.001). We then calculated the growth rate for each day by using differential coefficients from this fitted equation (Fig. S3). Finally, we defined spring, summer, and autumn biomass production as the sum of daily growth rate from April to May, from June to July, and from August to September, respectively (Zhang et al. 2013a).

To explore how changes in plant phenology and growth rate influenced seasonal biomass production over time, we used the fast-growing phase concept (Gregorczyk 1991). The mid-season ‘fast-growth phase’ was identified by the seasonal dynamics of growth rate (Fig. S3). Specifically, the start and end of the fast-growing phase were defined as the days of maximum increase and maximum decrease in growth rate, which also correspond to the days at which aboveground biomass reaches 21% and 79% of the annual maximum biomass, respectively. The length of the fast-growing phase was calculated as the number of days between the start and end of the fast-growing phase.