• 研究报告 •

### 内蒙古草原生物量和地下生产力空间格局及其关键影响因子

1. (1北京师范大学地表过程与资源生态国家重点实验室， 北京 100875； 2北京师范大学资源学院， 北京 100875)
• 出版日期:2016-01-10 发布日期:2016-01-10

### Geographic patterns and controlling factors of biomass and belowground net primary productivity of Inner Mongolia grassland.

ZHAO Ming-fei1,2, WANG Yu-hang1,2, ZUO Wan-yi2, KANG Mu-yi1,2*, JI Wen-yao1,2, DAI Cheng1,2

1. (1State Key Laboratory of Earth Surface Processes and Resource Ecology, Beijing Normal University, Beijing 100875, China; 2College of Resources Science and Technology, Beijing Normal University, Beijing 100875, China)
• Online:2016-01-10 Published:2016-01-10

Abstract: In order to know the spatial patterns of belowground net primary productivity (BNPP), aboveground biomass (AGB), belowground biomass (BGB) and belowground/aboveground biomass ratio (B/A), and the key environmental factors affecting those variables, BNPP, ABG, BGB and soil samples were obtained from 33 sites within a transet (>1500 km) in the Inner Mongolia grasslands, especially BNPP were collected by ingrowthbag method and BGB by soil core method. For vegetation survey in each site, we selected 4 subplots (a size of 1 m×1 m) within one 10 m×10 m plot to record the properties of each plant species involving height, coverage and abundance. In laboratory, we analyzed 5 soil chemical and physical characteristics. We also got 19 climate indicators and calculated 4 species diversity indices including richness index, Shannon index, Simpson index and Pielou index. We used linear regression analysis to investigate the spatial patterns of the dependent variables, and classification and regression trees (CART) to screen the key envirnmental factors. Linear regression models show that from the southwest to the northeast, there are obvious increasing trends in the response variables apart from B/A. Especially, the AGB and BGB data from the southwest exhibit signifant linear relationships. CART models of BNPP, AGB, BGB and B/A explain most of variations of the predictors, which represent 58.3%, 53.3%, 78.8% and 53.8% total sum of squares, respectively. We detected the possible key factors affecting those dependent variables by CARTs (namely, soil bulk density and Pielou index to BNPP, maximum temperature of warmest month to AGB, annual precipitation to BGB, and eleviation to B/A). We used BNPP, AGB, BGB and B/A respectively with the key factors identified by CART to establish GAM. Explained deviation rates of all models are over 50%, which indicates that the GAM represents most of the variation of the dependent variable and to a certain extent, verifing the accuracy of CART. The relationship between BNPP and soil bulk density is a piecewise function. AGB and warmmest and highest temperature show a nonlinear relationship. BGB is positively correlated with average annual rainfall integrally. B/A and elevation have a close but complicated relationship, greatly influenced by the extreme value points on both ends around.