Abstract: The sample average is usually used to estimate the holistic average in random sampling of an ecological survey. However, determination of sample size plays an important role in calculating the sample average. Based on the statistical theory, a method of determining the necessary sampling size in random sampling in the Horqin Sand Land was obtained, and it was tested with some data sets, such as the contents of soil moisture, soil organic matter, nitrogen, phosphorus and potassium, composition of plant species, and vegetation height, density and coverage. The necessary sample sizes for these ecological indicators were obtained. As the values of cv (coefficient of variation) of these indicators were lower than the assumed values, the results are as follows: (1) The necessary sample sizes for the contents of soil organic matter, total nitrogen, available nitrogen, total phosphorus, available phosphorus and potassium in cropland ranged from 7 to 15; (2) The necessary sample sizes for the height, density, coverage and species composition of vegetation over fixed dunes and sandy grasslands were 11, 8, 28 and 15 respectively; (3) The necessary sample sizes for soil moisture content of fixed and semifixed dunes in depth 0-30 cm were 32 and 12 respectively. These results are of certain guiding and referring value for rationally selecting the sample sizes in the Horqin Sand Land and similar regions in the future.

Key words: sample size, sand land ecosystem, coefficient of variation, sample variance, Horqin Sand Land