Why do standard decision trees usually not require feature scaling?

A. Splits depend on order and thresholds rather than distance magnitude B. Trees accept only categorical variables C. Scaling destroys class labels D. All tree features must have variance one