Abstract: Xilin River Basin is the most representative area of typical steppe in Inner Mongolia. Characterizing the dynamics of aboveground biomass and its relationship with environmental factors can provide basic data for the research of ecosystem carbon cycling and a theoretical frame for rational use and management of grassland. Xilin River Basin locates in semiarid region. With filed empirical data, the methods of correlation analysis, classification and regression tree, nonlinear regression analysis were used to quantitatively analyze the relationship between aboveground biomass and its key driving factors in the upstream and downstream of the basin at different times (around June 20, July 5, July 20, and August 6 of 2016). The results showed that: (1) Aboveground biomass changed most significantly in July, with a peak on August 6. The aboveground biomass of the upper, middle and lower basin on August 6 were 79.64, 74.87, 69.34 g·m^-2 respectively. Biomass showed a decreasing tendency from the southeast to the northwest. (2) Key factors affecting aboveground biomass at different times in the upstream of the basin were precipitation in April, soil water content on July 5, soil water content on July 20, and Simpson index on August 6. The Simpson index had a negative effect on aboveground biomass, while all other factors had positive effects on aboveground biomass. Corresponding key factors accounted for 77%, 72%, 79%, and 65% of the variation of aboveground biomass, respectively. (3) Longitude, precipitation in June, precipitation in July, and precipitation in July were the key factors affecting aboveground biomass of the downstream on June 20, July 5, July 20, and August 6, respectively. All the factors had a significant positive correlation with aboveground biomass. Corresponding key factors accounted for 88%, 75%, 85%, and 82% of the variation of aboveground biomass, respectively. Our results provide a reference for the analysis of the temporal and spatial variation of plant biomass at a large scale and provide data for in depth research and long-term monitoring of grasslands.

Key words: climate change, shading, productivity., plant type, ear development, planting density