Abstract: The modifiable areal unit problem has significant implications for ecological research that involves the use of spatial data. The objective of the present study was to discuss the scale effect and zoning effect by investigating and analyzing the spatial heterogeneity of vegetation communities in the transitional zone between oasis and deserts. Geostatistical techniques were used to evaluate the spatial characteristics of plant communities. The spatial structures of different plant populations were characterized using exponential model variogram parameters, including nugget (C0), range (A0) and sill (C0+C). Two methods were employed to determine the scale effect of spatial heterogeneity: 1) a constant plot area (10 m×10 m) and the sampling area expanded to analyze spatial heterogeneity of plant populations; 2) the plot area was changed for maintaining a constant sampling area to analyze spatial heterogeneity. In addition, the plots at the scale of 500 m×500 m were clustered into nested units of different shapes and directions (10 m×200 m, 20 m×100 m, 40 m×50 m, 50 m×40 m, 100 m×20 m and 200 m×10 m) and analyzed at a resolution of 10 m×10 m to examine the zoning effect. The results showed that the spatial heterogeneity of plant populations in the transitional zone between oasis and desert were influenced significantly by scale and zoning approach. Using sensitivity analysis, we were able to obtain a suitable sampling area and zoning approach to weaken the scale and zoning effects. The scale at which different parameters were sensitive was variable. For example, C0 and C0+C were relatively sensitive at small scales, A0 was affected significantly by changes at larger scales, and C0 and A0 were relatively sensitive to different zoning approaches. In order to gain more representative vegetation information of spatial heterogeneity, the sampling area should be more than 200 m×200 m for Nitraria sphaerocarpa populations, 100 m×100 m for Reaumuria soongorica populations, and a plot area of 20 m×20 m to 30 m×30 m should be feasible.