Highly Correlated Variables in Random Forests

In a random forest model, if two variables exhibit a high correlation, how might this relationship affect the calculation of their feature importance?