Abstract: By the methods of geostatistics,this paper studied the spatial variation of microbial biomass,microbial community structure and microbial activity in a creosote-contaminated soil.The microbial biomass was indicated by the total amount of 26 examined phospholipid fatty acids (totPLFA),the microbial community structure was characterized by the first two principal components (PC1 and PC2) of the PLFA patterns through subjecting the PLFAs to principal component analysis,and the total amount of CO₂-C respired (Cre) during incubation was used to describe the soil microbial activity.The results showed that most of the variables exhibited various degrees of spatial autocorrelation, and spherical models with nugget could fit their semivariograms. The estimated spatial dependence distances of the variables varied from 2.50 to 16.60 m.The PLFA PC1,totPLFA and Cre displayed relatively high spatial dependences,their relative structural variance,i.e.,the proportion of total variance that can be attributed to the spatial autocorrelation, being 82.3%，79.6% and 64.7%,respectively,while PLFA PC2 was not spatially autocorrelated.The kriged maps showed that there were several hot spots with relatively high microbial biomass and high microbial activity located in the study site.Gram- negative bacteria characterized by PLFAs 16:1ω7t,cy17:0,18:1ω7 and cy19:0 were the dominant community in these hot spots.The concentration and spatial distribution of polycyclic aromatic hydrocarbons as the main contaminants in the soil could be one of the important factors affecting the spatial variation of soil microbial properties.

Key words: Patch, Boundary, Sandy landscape, Mu Us sandy land