Vegetation coverage is an important quantitative indicator of community structure and it has a range of uses, such as in the calculation of land surface evapotranspiration for assessment of vegetation condition, and in the simulation of climate or ecological dynamics. Spatial heterogeneity is an important property of an ecosystem. Research on spatial heterogeneity has both theoretical and practical significance for understanding spatial scales of ecosystem functions and processes. Semivariance is defined as half the variance of the differences between values at paired locations, and it is a useful tool to analyze spatial heterogeneity. Analysis of the semivariance has been widely used in ecological research in recent years, but studies have mainly focused on spatial heterogeneity of soil properties, and rarely on vegetation coverage. Desert steppe has unique vegetation coverage features and spatial heterogeneity exists at all spatial scales. Investigating the spatial heterogeneity of desert steppe vegetation coverage has great significance for sampling and remote sensing. During June, July and August, vegetation on the desert steppe grows rapidly, which may lead to temporal dynamics of the spatial heterogeneity in vegetation coverage. Different months may have different spatial heterogeneity features. However, there has been little research on the temporal dynamics of this spatial heterogeneity in desert steppe vegetation.
In this study, we surveyed vegetation coverage in Inner Mongolia during June, July and August in a 30 m × 30 m grid, in which there were 900 1m×1m quadrats. Spatial and temporal features of the vegetation coverage were analyzed following the semivariance analysis method. The analysis was mainly done in the R statistical environment, using sp and geoR packages. Average vegetation coverage gradually increased from June to August and standard deviation had a range of 3.08-6.02. The results showed that an exponential model was a good description of the semivariance of the vegetation coverage, but the curve shape and semivariance parameters varied with the month. The range of vegetation coverage was the largest in June, reaching 100 m, and the smallest in August, only 15m, although it had the highest spatial variability. The semivariance in July had a nested hierarchy structure. The nugget of vegetation coverage increased from June to August, ranging from 4.28 to 7.29. The sill also increased from June to August, and the sill of August was significantly higher than those of June and July. The structural ratio of vegetation coverage in June, July and August ranged from 72% to 85%, showing a strong spatial autocorrelation. Anisotropy analysis revealed that only the semivariance in June had anisotropic characteristics, because its semi-variance at 135° direction was significantly lower than those of the other three directions (0°, 45° and 90°). The anisotropy ratios of vegetation coverage in July and August were close to 1, indicating isotropy.
The results of this study suggest that the spatial heterogeneity of desert steppe vegetation cover has temporal dynamics, and that the parameters of semivariogram models can vary by month. The temporal dynamics of desert steppe vegetation spatial heterogeneity should be considered when designing field sampling or mapping protocols.