2.2 Experimental design
From May 2000 to October 2001, an experiment was conducted to explore the responses of the 4 target plant populations to 8 levels of water additions, which were set according to the local annual mean precipitation during 1980–2000 (i.e. 350 mm), and they were 170 mm, 250 mm, 300 mm, 350 mm, 525 mm, 595 mm, 665 mm and 700 mm, respectively. This study aims to explore the effect of water amount on plants, not to mimic the local rainfall pattern. Thus, water was added daily to each treatment, and the amount of water was determined by dividing the total amount of rainfall in each pattern by the total growing days. Plants grew in pots with diameter 30 cm. There were three replications of each treatment for each species, ending with 19 pots in total. For L. chinensis , seeds were randomly sown in the pots in early May 2000, and four individuals with the similar size in each pot were kept after germination, and the rest were removed manually. For S. grandis , four ramets with similar size were transplanted from the nearby field to each pot in later May 2000 (similar sowing was done here in early May 2000, but the germination rate was too low). For A. frigida andP. acaulis , plants were dug out from the nearby field, and ramets were then separated into similar size. Four of them were transplanted into each pot in early May 2001. All the plants were grown under natural condition before applying treatments, which occurred from 10 June to 10 September 2001. At the end of the experiment, all the plants in each pot were washed out from their growing soils, and then separated into aboveground group (AG) and belowground group (BG), where soil surface is the baseline. For L. chinensis and S. grandis , AG included leaves and stems, while BG included roots and rhizomes. For A. frigida , AG included leaves, flowers and stems, while BG included roots. For P. acaulis , AG included leaves and stems, while BG included roots. All the biomasses were oven-dried at 65°C to constant weight and weighted.