The spatial heterogeneity of soil respiration (Rs) is of great significance in accurately estimating the carbon budget in erosion areas. This study investigated the soil respiration (Rs), total nitrogen (TN), carbon-nitrogen ratio (C/N), soil organic carbon content (SOC), leaf area index (LAI), soil temperature (T10), and soil moisture (W) in 59 soil samples collected from Hetian Town in Fujian Province. Both classical statistics and geostatistics were used to analyze the spatial heterogeneity of soil respirations and other measured factors. The variability of Rs and other measured factors in the samples ranked from the largest to the smallest: LAI > SOC > TN > Rs > C/N > T10 > W. Rs was positively correlated with T10 (P<0.01) and with TN (P<0.05), but not significantly correlated with other factors (P>0.05); TN, SOC and T10 could be used to explain the spatial variation of soil respiration among the samples. The results from geo-statistical analysis showed that Rs was in a medium spatial autocorrelation, with 52.89% of spatial heterogeneity caused by structural factors and 47.11% of spatial heterogeneity resulted from random factor; the fractal dimensions of soil respiration and its interacted factors were ranked as: Rs > LAI > C/N > T10 > SOC > W > TN; the spatial distribution patterns of Rs were similar with those of TN and T10, but different from those of C/N, SOC or LAI. At the 95% confidence level and 90% estimation accuracy, the reasonable sampling number of Rs was 62.
Keywords: geostatistics; semi-variogram function; soil respiration; spatial heterogeneity; vegetation types.