2.2 Sample site design
The study site consists of a relatively flat and even area of 600 m by 300 m, fenced since 1979 to prevent grazing by large animals (Li et al., 2015). In 1981, the area was separated into ten replicate blocks (60 m × 300 m each). Community aboveground biomass was surveyed in the middle of every month throughout the growing season (from May to September) each year by clipping all plants within a 1 m × 1 m quadrat that was randomly located within each block, over 1981–2011. After harvesting, all living vascular plants were sorted into species, and oven-dried at 65 ℃ to a constant weight. Species richness was calculated as the total number of species present in ten blocks in a year. All species were classified into five plant functional groups primarily on the basis of life forms (Bai et al., 2004): perennial rhizome grass (PR), perennial bunchgrasses (PB), perennial forbs (PF), shrubs and semi-shrubs (SS), and annuals and biennials (AB). Functional group biomass was determined as the biomass sum of all the species in each functional group. Then, the proportion of aboveground biomass of each functional group to the community total aboveground biomass was calculated and used for the analysis of biomass complementarity among different functional groups (Bai et al., 2004). Data on productivity and diversity were collected monthly between May and September over a period of 29 years from 1981 to 2011, excluding the missing data from the years 1995–1996 (Ma et al., 2010).