Abstract: Soil respiration changes not only in time but also in space obviously. Those temporal and spatial variations enhance uncertainties in soil respiration measurements. Understanding the spatial and temporal variation in soil respiration of a region is fundamental for an accurate estimation of soil respiration in the region. Geostatistics method is a spatial analysis method evolved from the traditional statistics analysis，which is not only can effectively reveal the spatial distribution variation characteristics of variables，but also can effectively explain the implications of spatial patterns on ecological processes and functions. With the wide application of geostatistics in ecological research，making conduct research on spatial heterogeneity of soil respiration and variance process on a larger-scale become possible. At the beginning of the century，the scientists abroad conducted research on spatial heterogeneity of soil respiration and its influencing factors using geostatistics method，while those researches in domestic began in recent years. In this paper，the soil respiration of three ecosystems was measured from November 2009 to February 2012 at Ebinur Lake Wetland Nature Reserve，Bortala Mongol Autonomous Prefecture，Xinjiang，China，and quantitative study carried out on the relationship between soil respiration and 0-50cm soil organic carbon，and then simulated soil respiration accomplished at 97 sites in Ebinur Lake Wetland Nature Reserve through model. And at last， the spatial heterogeneity was analyzed of soil respiration and its relation to other environmental factors using both traditional and geostatistics methods. The traditional statistical analysis（descriptive statistics）showed that the soil respiration rate was between 0.342 and 10.848 μmol·m^-2·s^-1，and the average was 1.109 μmol·m^-2·s^-1，the coefficients of variation were 167.37%，indicating that spatial variation of soil respiration in this area was strong. When semi-variance analysis of geostatistics was used，the results showed the spatial structure feature of soil respiration rate and soil organic matter at 0-50 cm layer could well be described by Gauss model．The C₀/（C₀+C）values were all range from 0 to 0.1% for soil respiration, soil organic carbon and soil moisture，indicating that spatial variation of them in this area were strong，and the spatial heterogeneities of them were resulted from structural factors. The ecological progresses influence soil respiration，soil organic carbon and soil moisture which change with time at the same scales. The spatial distribution map of soil respiration rates produced with the Kriging interpolation method shows that there is a high degree of spatial autocorrelation on the spatial distribution of soil respiration; and on the large scale and for a long time，the soil organic carbon content is the main factor to affect the spatial distribution of soil respiration in this study area. On the whole，soil respiration rates are high in the southeast and the northwest and low in the middle region of Ebinur Lake Wetland Nature Reserve，and as for the spatial traits of soil respiration rates, they are characterized by clear plaques distribution and continuous one.

Key words: geostatistics, Kriging interpolation method, Ebinur Lake