Spatial heterogeneity of soil moisture in Stipa breviflora grasslands under different stocking rates

The effects of stocking rates on spatial heterogeneity of soil moisture in a Stipa breviflora desert steppe ecosystem were investigated. Both classical statistics and geo-statistical methods were adopted to generate descriptive statistics, semi-variogram analysis, and spatial interpolation methods for soil moisture in four types of sample plots. The results indicated: 1) The mean value of soil moisture in all different sample plots increased with increase in stocking rates, the coefficient of variation (CV) in each sample plot changed within the range of 7.29% to 9.79%, presenting weak variability. 2) The semi-variogram analysis indicated that the content of soil moisture in each sample plot was found to be in line with the exponential model, the structural variation ratio of non-grazing was 74.4%, its value was less than 75%, which corresponded to moderate variability. The C/(C0 + C) values of lightly, moderately and heavily grazed areas were 94.4%, 91.2%, and 96.0%, respectively; all values were larger than 75%, which corresponded to strong spatial auto-correlation. 3) The Kriging interpolation method was used to deal with interpolative analysis for unknown sampling points of each sample plot, and the results showed that the spatial variability of soil moisture for heavily grazed area was largest, while that for non-grazing was lowest. Our analyses showed that different stocking rates had different impacts on spatial patterns of soil moisture.