Abstract: Taking the Qianlianghu polder in Dongting Lake area as a case, this paper studied the soil microbial quantity, enzyme activity, and their canonical correlations of five typical land-use patterns, i.e., forestland, garden land (vegetable and orchard), nonirrigated cropland (cotton, maize, and sugar cane), irrigated cropland (rice), and wasteland in the area. Among the five landuse patterns, nonirrigated cropland had the highest quantity of soil bacteria and actinomycetes, while wasteland had the highest quantity of fungi. Bacterium was the main microbial group, occupying 44.42%-92.93% of the total, followed by fungus 4.89%-42.76% and actinomycete 1.71%-24.52%. The phosphatase, urease, proteinase and dehydrogenase activities in 0-50 cm soil layer were 0.01-0.07 mg·g^-1·d^-1,0.01-0.05 mg·g^-1·d^-1, 0.92-7.11 mg·kg^-1·d^-1, and 0.01-0.38 μl·g^-1·d^-1, respectively, with the lowest phosphatase, urease, and dehydrogenase activities in garden land, wasteland, and irrigated cropland, and the lowest proteinase activity in nonirrigated cropland. The quantities of actinomycetes and bacteria had positive correlations with the canonical variable of soil microbes, with the correlation coefficient being 0.174 and 0.003, respectively, but the quantity of fungi had a negative correlation with the variable -0.215. The activities of urease and dehydrogenase had positive correlations with the canonical variable of soil enzyme activities, with the correlation coefficient being 10.557 and 1.616, respectively, but the activities of phosphatase and proteinase had negative correlations with the variable, their correlation efficient being -17.275 and -0.041, respectively. Through canonical correlation analysis, coordination analysis, and cluster analysis, it could be possible to use the microbial quantity and enzyme activity in different soil layers for the diagnosis of soil health and the classification of site type.

Key words: Crops, Water deficit, Compensation effect