We classify 8 plant associations using 30 variables derived only from ALS-based DTM.

We identify variables which could be drivers for vegetation.

The reduced subset of 6 variables increases the accuracy from κ: 0.56 to κ: 0.72.

We achieve an accuracy of κ: 0.83 for 4 aggregated classes using only 3 variables.

Micro-topography can be an indicator of ecological processes in open landscapes.