Alexandre, It seems to me that if you choose to divide by the standard deviation then it should be calculated as the difference between the predicted value of the point on the plane and the observed value. I'm not sure you would even want to standardize these values since it would eliminate differences in topographic variation between plots (i.e., plots with deep depressions and high mounds would end up having the same variance as those with more subtle microtopography). One approach would be to stratify plots with high microtopographic variation (i.e., deep pits) and low variation (more or less planar) and then standardize the values and test the strata individually.
Your hypothesis is that within a small area (1 ha), trees tend to avoid microtopographic depressions. What is the null hypothesis? Is it that the mean deviation from the plane is zero? The alternate hypothesis would then be the mean deviation is > 0. Cheers, Peter
