How are variables selected in tree-based learning?

a. Variables are automatically selected by the tree logic as the best variable

(and value to split it at) is identified at each step of the recursive partitioning.

b. Variables must be selected in a preprocessing step using a filter method.

c. A regularization penalty is factored in the impurity measure of each variable split during the induction process.

d. Cross-validation is used to select the variables used in a tree.